RETHINKING FOREIGN POLICY

ANALYSIS

Stephen G. Walker, Akan Malici, and Mark Schafer present a definitive, social-
psychological approach to integrating theories of foreign policy analysis and
international relations—addressing the agent-centered, micro-political study of
decisions by leaders and the structure-oriented, macro-political study of state
interactions as a complex adaptive system. The links between the internal world of
beliefs and the external world of events provide the strategic setting in which states
collide and leaders decide.
The first part of this ground-breaking book establishes the theoretical frame-
work of neobehavioral IR, setting the stage for the remainder of the work to apply
the framework to pressing issues in world politics. Through these applications stu-
dents can see how a game-theoretic logic can combine with the operational code
research program to innovatively combine levels of analysis. The authors employ
binary role theory to demonstrate that relying only on a state-systemic level or an
individual-decision making level of analysis leads to an incomplete picture of how
leaders steer their ships of state through the hazards of international crises to estab-
lish stable relations of cooperation or conflict.

Stephen G. Walker is Emeritus Professor of Political Science at Arizona State

University. He has edited Role Theory and Foreign Policy Analysis and co-edited
Beliefs and Political Leadership in World Politics.
Akan Malici is Assistant Professor of Political Science at Furman University. He is
the author of When Leaders Learn and When They Don’t and The Search for a Common
European Foreign and Security Policy.
Mark Schafer is Professor of Political Science at Louisiana State University. He
has co-edited Beliefs and Leadership in World Politics and co-authored Groupthink
versus High-Quality Decision Making.
RETHINKING FOREIGN
POLICY ANALYSIS
States, Leaders, and the
Microfoundations of Behavioral
International Relations

Edited by Stephen G. Walker,

Akan Malici, and Mark Schafer
First published 2011
by Routledge
270 Madison Avenue, New York, NY 10016
Simultaneously published in the UK
by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
Routledge is an imprint of the Taylor & Francis Group, an informa business
This edition published in the Taylor & Francis e-Library, 2011.
To purchase your own copy of this or any of Taylor & Francis or Routledge’s
collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.

© 2011 Taylor & Francis

The right of Stephen G. Walker, Akan Malici, and Mark Schafer to be
identified as the authors of the editorial material, and of the authors
for their individual chapters, has been asserted by them in accordance
with sections 77 and 78 of the Copyright, Designs and Patents Act 1988.
All rights reserved. No part of this book may be reprinted or
reproduced or utilized in any form or by any electronic,
mechanical, or other means, now known or hereafter
invented, including photocopying and recording, or in any
information storage or retrieval system, without permission in
writing from the publishers.
Trademark Notice: Product or corporate names may be
trademarks or registered trademarks, and are used only for
identification and explanation without intent to infringe.
Library of Congress Cataloging in Publication Data
Rethinking foreign policy analysis: states, leaders, and the
microfoundations of behavioral international relations/
[edited by] Stephen G. Walker, Akan Malici, and Mark Schafer.
p. cm.
Includes bibliographical references and index.
1. International relations—Decision making. 2. International relations—
Psychological aspects. I. Walker, Stephen G., 1942-II. Malici, Akan,
1974-III. Schafer, Mark, 1962–
JZ1253.R48 2011
327.101'4—dc22
2010028853

ISBN 0-203-83486-0 Master e-book ISBN

ISBN 13: 978–0–415–88697–0 (hbk)

ISBN 13: 978–0–415–88698–7 (pbk)
ISBN 13: 978–0-203–83486–2 (ebk)
CONTENTS

List of Figures viii

List of Tables x
Preface xi
List of Contributors xiii

PART I
Foreign Policy Analysis 1

1 Foreign Policy Analysis and Behavioral International Relations 3

Stephen G. Walker

2 Macropolitics and Foreign Policy Decisions: The Billiard Ball

Model of IR 21
Stephen G. Walker

3 Micropolitics and Foreign Policy Decisions: The Behavioral

Model of IR 44
Stephen G. Walker

4 Quantum Politics and Operational Code Analysis: Theories

and Methods 62
Stephen G. Walker
vi Contents

PART II
Foreign Policy Decision Making 81

5 The United States and Rogue Leaders: Understanding the

Conflicts 83
Akan Malici

6 Deceptive Bargaining and Nuclear Ambitions: Prospect

Theory and North Korea’s Decision to Go Nuclear 97
Kai He and Huiyun Feng

7 Small Group Dynamics: The Psychological Characteristics of

Leaders and the Quality of Group Decision Making 112
Mark Schafer, Jonathon Nunley, and Scott Crichlow

8 Alliances and Their Microfoundations: France and Britain in

the 9/11 Era 130
Akan Malici

PART III
Foreign Policy Learning 151

9 Learning to Resist or Resisting to Learn? The Operational

Codes of Fidel Castro and Kim Il Sung 153
Akan Malici

10 Stability and Change in Belief Systems: The Operational Code

of George W. Bush from Governor to Second Term President 169
Jonathan Renshon

11 Experiential Learning by U.S. Presidents: Domestic and

International Influences in the Post-Cold War World 189
Samuel B. Robison

12 Cognitive Responses by U.S. Presidents to Foreign Policy

Crises: Belief Changes in Response to Positive and Negative
Experiences 205
B. Gregory Marfleet and Hannah Simpson
 Contents vii

PART IV
Foreign Policy Dynamics 221

13 Dueling with Dictators: Explaining the Strategic Interaction

Patterns of U.S. Presidents and Rogue Leaders 223
Stephen G. Walker and Mark Schafer

14 Binary Role Theory: Reducing Uncertainty and Managing

Complexity in Foreign Policy Analysis 245
Stephen G. Walker

15 The Integration of Foreign Policy Analysis and International

Relations 267
Stephen G. Walker

Appendix: Formal Models of Symbolic and Strategic Interaction 283

Stephen G. Walker

Bibliography 290
Index 318
FIGURES

1.1 Two Systems in World Politics. 12

2.1 Behavioral and Billiard Ball Models of International Relations. 22
2.2 A Static Political Field Diagram of International Relations. 30
2.3 A Dynamic Political Field of Escalation (–) and De-escalation (+). 31
2.4 Connections between the World in Their Minds and the World of
Events. 33
2.5 Extensions of Power Logic from Dyads to Triads. 36
2.6 Mapping Two-Player Games in a Political Field Diagram. 39
2.7 Two-Player Games for the ABC Triad. 40
3.1 A Hierarchical Power Landscape of the Grand Power Game in
the IR Field. 46
3.2 Ego’s Role Enactment Predicted by Exchange Theory and Role
Theory. 50
3.3 Contents of the Revised Holsti Operational Code Typology. 56
3.4 Constructing a Political Belief System Typology. 59
3.5 Drawing Inferences about Preferences. 60
4.1 Basic Research Designs for Understanding Foreign Policy Analysis. 63
4.2 Hierarchical Landscape Features in the Worlds of Beliefs and Events. 64
4.3 A Strategic Interaction Episode during the Vietnam Conflict. 67
4.4 Barack Obama’s VICS Indices and Subjective Games. 74
4.5 Obama’s Operational Code Beliefs and Types of Leadership Styles. 77
5.1 Kim Jong Il’s and Fidel Castro’s Subjective Game. 92
5.2 Ahmadinejad’s and al-Asad’s Subjective Game. 94
6.1 North Korea’s Strategies, Payoffs, and Outcomes. 106
8.1 Inter-State Comparisons of Beliefs for France and Britain. 141
 List of Figures ix

8.2 Subjective Games of France and Britain in Operation Enduring

Freedom. 143
8.3 Subjective Games of France and Britain in Operation Iraqi
Freedom. 146
9.1 Real GNP/GDP Growth Rates, 1985–1994. 157
12.1 Probability of Belief Change. 216
13.1 Levels of Decision and Strategic Interaction in the Normal
Form of a 2 × 2 Game. 226
13.2 Internal and External Focal Points for Inferring Strategic
Preferences. 229
13.3 Local Attacker–Defender Objective Game in Persian Gulf and
Kosovo Conflicts. 232
13.4 Global Defender–Attacker Objective Game in Persian Gulf and
Kosovo Conflicts. 233
14.1 Static and Dynamic Matrix Analyses of Role Location Processes. 250
14.2 A Dynamic Political Field Diagram of Role Location. 253
14.3 Rules of Symbolic and Strategic Interaction in Binary Role Theory. 254
14.4 Role Location and Strategic Interaction Processes for a Role Dyad. 255
14.5 General Rules for Power Triads with Agent Strategies for Each
Segment. 263
15.1 Pivot Effects Models for Strategic Triads among Agents ABCD. 273
A.1 Grand Strategies and Inferred Preferences from Power and
Interest Distributions. 286
A.2 Binary Role Theory Locations and Rules for Grand Strategies. 287
TABLES

4.1 Barack Obama’s Operational Code Profile. 76

5.1 The Beliefs of Rogue Leaders. 88
7.1 Summary of Directional Hypotheses. 117
7.2 Bivariate Hypothesis Tests. 119
7.3 Multivariate Models Explaining Group Structural Faults. 122
7.4 Multivariate Models Explaining Decision Processing Faults. 123
7.5 Multivariate Models Explaining National Interests. 124
7.6 Multivariate Models Explaining Level of Conflict. 125
9.1 Operational Codes of Castro and Kim Compared to Norming Group. 162
9.2 Fidel Castro’s and Kim Il Sung’s Cold War and Post-Cold War
Operational Codes. 164
10.1 The Operational Code of George W. Bush. 180
10.2 Extended Analysis of Phase 3. 184
11.1 Factors Influencing Presidential Learning. 198
12.1 Binary Logistic Regression Results for Post-Crisis Belief Change. 214
12.2 Cross-tabulations of Belief Change with Learning Indicators. 217
13.1 Three-Level ANOVA Results for the Transparency and
Receptivity Hypotheses. 238
13.2 Predictive Powers of Objective and Subjective Games. 240
13.3 Predictive Powers of Subjective and Communications Games. 241
13.4 Levels of Experiential Learning by Leaders. 242
13.5 Direction of Learning by Leaders. 243
A.1 Intersections of Subjective Games for Appeasement (+) and
Bully (–) Role Sets. 284
A.2 Entropy Matrix for Intersections of Subjective Games Specified
by Binary Role Theory. 289
PREFACE

This book addresses simultaneously the connections between (a) beliefs and behav-
ior in human affairs, (b) the subfield of Foreign Policy Analysis and the field of
International Relations in the study of world politics, and (c) agents and systems in
the natural and social sciences. The general assumption informing the intersection
of these phenomena is that each of these problems resembles the others in impor-
tant ways. If so, then it may be possible to increase our understanding of each one
by juxtaposing them with one another and considering each of them as instances of
the same phenomena at successive levels of abstraction.
This argument informs the organization and presentation of the contents of
this volume under the general rubric of “rethinking foreign policy analysis.” It
accounts as well for the sometimes uneasy coexistence in various chapters between
such concepts as entropy, complexity, and hierarchical energy landscapes from the
physical and biological sciences and such concepts as social power, rationality, and
beliefs from the information and social sciences. We shall leave it for readers of the
following pages to decide whether our effort to bring these perspectives together
in a conversation about world politics is worth it. We simply present here a brief
narrative of how this confluence of intellectual crosscurrents occurred.
Over the past several years a political scientist ran daily with a biological physicist
in the mountainous forests of northern Arizona and received an informal seminar
in the history of physics. These conversations prompted the political scientist to do
some reading in basic physics and its historical evolution from a macroscopic study
of large objects to a microscopic study of small particles, accompanied by a shift in
focus from objects to the forces that connect them as fields. The biological physi-
cist also often referred to his ongoing collaboration with colleagues at Los Alamos
Laboratories in neighboring New Mexico, which he noted is just down the road
from the Santa Fe Institute whose mission spans the natural and social sciences in
xii Preface

the study of complex adaptive systems. This hint led the political scientist to do
some more reading outside of his area of world politics and renew an interest in
general systems theory, which had informed the behavioral revolution’s impact on
the field of International Relations almost a half century ago.
Three additional events acted as catalysts to bring together the contents in this
book as a unifying theory of world politics. The first event was a pair of round-
tables, “Behavioral IR as a Subfield of International Relations,” organized by Alex
Mintz at the annual meetings of the International Studies Association in 2006 and
2009. The second event was a panel organized by Akan Malici at the 2009 ISA
Meeting, “Advances in the Study of Foreign Policy,” which formed the nucleus
for Parts II and III of this volume. The third event was a workshop, “Integrating
Foreign Policy Analysis and International Relations through Role Theory,” con-
vened by Cameron Thies and Marijke Breuning at the 2010 ISA meeting.
Our participation in various capacities at these functions became occasions to
put into print what may otherwise have remained just ruminations about the orga-
nization of knowledge and intellectual progress in Foreign Policy Analysis and
International Relations. While our colleagues in political science at these events
were instrumental in galvanizing the effort that has made this book happen, they
are in no way responsible for any errors by the editors and contributors. The same
absolution extends as well to the biological physicist whose conversations with one
of us were perhaps a necessary (if not sufficient) initial condition to account for its
existence.
The editors would like to thank the authors in this volume for their contribu-
tions, three anonymous referees for providing valuable feedback to us, plus editor
Michael Kerns and editorial assistant Mary Altman at Routledge for their assistance
and support. We are grateful as well for the institutional support of the School of
Politics and Global Studies at Arizona State University and the Departments of
Political Science at Furman University and Louisiana State University. We also
thank our families for putting up with us as we assembled the book in addition
to attending to our roles as family members. We should also like to acknowledge
permission from the following journals to use and reprint copyrighted material:
Journal of Peace Research (Chapter 5), Psicologia Politica (Chapter 5), Journal of Conflict
Resolution (Chapters 8 and 10), and Political Psychology (Chapter 9). Finally, we want
especially to thank biological physicist Robert D. Young for his tutorial contribu-
tions, which shaped the final form into which this book has evolved.

Stephen G. Walker Akan Malici Mark Schafer

Flagstaff, AZ Greenville, SC Baton Rouge, LA
CONTRIBUTORS

Scott Crichlow, West Virginia University, has co-authored Groupthink vs. High-
Quality Decision Making (2010). His recent research appears in Journal of Conflict
Resolution, International Studies Quarterly, and Political Psychology.

Huiyun Feng, Utah State University, has authored Chinese Strategic Culture and
Foreign Policy Decision-making (2007). Her recent research appears in Security
Studies and The Pacific Review.

Kai He, Utah State University and on leave as a postdoctoral fellow in the Wilson
School at Princeton University, is the author of Institutional Balancing in the Asia
Pacific (2009). His recent research appears in Security Studies and the European
Journal of International Relations.

Akan Malici, Furman University, is the author of When Leaders Learn and When
They Don’t (2008) and The Search for a Common Foreign and Security Policy (2008).
His recent research appears in Journal of Conflict Resolution and Journal of Peace
Research.

B. Gregory Marfleet, Carleton College, is a game theorist and computational

modeler. His research appears in Political Psychology and Foreign Policy
Analysis.

Jonathon Nunley, Louisiana State University, is a student in the Department

of Political Science where he specializes in automated content analysis using
Profiler Plus software.
xiv Contributors

Jonathan Renshon is a Harvard University Ph.D. student and the author of Why
Leaders Choose War (2006). His recent research appears in Journal of Conflict
Resolution and Political Psychology.

Samuel B. Robison recently received his Ph.D. from the Department of Political
Science at Louisiana State University. His research appears in Foreign Policy
Analysis and International Studies Perspectives.

Mark Schafer, Louisiana State University, has co-edited Beliefs and Leadership
in World Politics (2006) and co-authored Groupthink vs. High-Quality Decision
Making (2009). His recent research appears in Journal of Conflict Resolution and
International Studies Quarterly.

Hannah Simpson is a student in the Harvard Law School.

Stephen G. Walker, Arizona State University, has edited Role Theory and Foreign
Policy Analysis (1987) and co-edited Beliefs and Leadership in World Politics (2006).
His recent research appears in Journal of Conflict Resolution and International
Studies Quarterly.
PART I

Foreign Policy Analysis

1
FOREIGN POLICY ANALYSIS AND
BEHAVIORAL INTERNATIONAL
RELATIONS
Stephen G. Walker

Introduction
In the twenty-first century it is hard for political leaders to steer the ship of state
without both an internal compass to define the state’s destination and a map to
mark the locations of others and relevant geopolitical features of the environment
(Walker and Malici forthcoming). Foreign policy mistakes and fiascos can occur
when states collide as leaders decide how to navigate the treacherous waters of
world politics (Tuchman 1984; Neustadt and May 1986). This problem becomes
acute during international crises—turning points in world politics—when at least
two states collide and perhaps threaten their mutual existence. In a world of
nuclear weapons, a collision may also threaten their neighbors and even the entire
planet.
The collision between the United States and the USSR in the October, 1962
Cuban missile crisis signaled almost 50 years ago that the possibility of a regional
or global conflagration is real in the nuclear age. While the end of the cold war
brought the era of superpower confrontation between the United States and
the Soviet Union to a close, the capacity for a cataclysmic collision is greater
today with the proliferation of nuclear weapons to regional powers in East and
South Asia and the Middle East. American President John Kennedy and Russian
Premier Nikita Khrushchev were able to steer their respective states away from
a nuclear disaster in 1962; however, it is also possible for America, Russia, and
China to confront one another again with the capacity to do more damage than
was possible with previous generations of nuclear weapons.
The new leaders of these large states and their smaller, regional allies may not
be so fortunate as Kennedy and Khrushchev in being able to steer their respective
ships of state successfully through such crises with adversaries. There are also new
4 Stephen G. Walker

challenges posed in today’s world by the less catastrophic (but no less significant)
threats from severe economic dislocations, terrorist attacks with weapons of mass
destruction, and ecological hazards associated with the processes of globalization,
cultural rifts, and natural disasters. In order to prevent or manage them, states and
their leaders involved in these events need to be able to diagnose the actions of
others and make choices that lead to beneficial outcomes both for themselves and
their neighbors.
The central problem in the twenty-first century is the difficulty of balancing the
geopolitical reality that we are all neighbors, due to the proximity generated by
military and economic technologies, against the psychosocial reality that we are
still strangers in many respects who do not understand one another (Polk 1997;
Weiner 2007; Suskind 2009). The proximity created by globalization enhances
the importance of answering the following fundamental questions that students of
world politics share. What are the causes of conflict, cooperation, domination, and
submission outcomes among the actors in world politics? Which foreign policy
decisions in the form of strategies, tactics, and moves by these actors lead to these
outcomes as international relations? How are these relationships transformed or
maintained by the decisions of states and leaders in world politics? These questions
are puzzles that both policymakers and scholars must solve in order to manage or
understand world politics, respectively, as a series of discrete foreign policy deci-
sions by actors or as a set of international relations among them (Nye 1990, 2008;
George 1993; Gelb 2010). We address both of these kinds of puzzles in this book.
On the one hand, we confront particular, real-world puzzles facing states and
their leaders in the twenty-first century, e.g., the difficulty of other leaders in
understanding decisions by “rogue” leaders and adopting policies that are likely
to lead to peace and avoid war with them. We also investigate how leaders can
organize the decision-making process inside states and among allies so as to make
high-quality decisions and avoid sub-optimum outcomes in managing conflicts.
We examine as well whether and how leaders of states learn to adapt their deci-
sions to changing circumstances and actions by others in the political universe.
On the other hand, we argue that general, theoretical solutions to these puzzles
are relevant to scholars, as they attempt to understand such historical cases as
instances of more general puzzles that are endemic to world politics. This latter
task is the central focus of our book.
Our approach to the general puzzles of conflict, cooperation, domination, and
submission in world politics is to examine them both from the perspective offered
by the actors and from the system in which they act, i.e., as a set of microscopic
foreign policy decisions by actors and as a set of macroscopic social relations among
them. This dual approach is unified theoretically by the assumption that actors are systems,
too. Leaders are individuals-as-actors with cognitive, emotional, and motivational
subsystems that constitute an interior system of psychological relations compara-
ble in complexity to the exterior system of states-as-actors and the social relations
in which they are members (Wolfers 1962; Houghton 2007; Wendt 1999). The
 Foreign Policy Analysis and Behavioral IR 5

psychological processes of decision making and learning by leaders are emergent

properties of an interior communication and control network, which interfaces
with the social processes of cooperation and conflict that are the emergent prop-
erties of an exterior communication and control network. Both of these networks
can be understood as complex adaptive systems of interaction with inter-related
emergent properties (Axelrod and Cohen 1999; Mitchell 2009; see also Deutsch
1966, 1968; Burton 1969).
These features of our approach identify it as a social-psychological analysis of
world politics, which employs a general systems theory to unify the understand-
ing of the actors, actions, and relations that constitute foreign policy and inter-
national relations phenomena as puzzles to be solved. The dual focus on agents
and their relations in world politics bridges the subfield of Foreign Policy Analysis
and the field of International Relations, thereby inviting comparisons with other
approaches to the study of each of these phenomena. Our position with respect
to such opportunities throughout this volume is that we invite comparisons that
foster relations of coordination and cooperation rather than competition and con-
flict among these approaches (Walker 2003; Laudan 1977). All of the contributing
authors attempt to link their efforts in various ways to other schools of thought in
international relations theory and foreign policy analysis.
The general systems theory that informs and unifies our social-psychological
approach is role theory. It has a long history in the disciplines of psychology and
sociology as a psychosocial model of social life, which links actors occupying roles
with other actors in counter-roles that collectively define a social system of sym-
bolic and strategic interaction (McCall and Simmons 1978; Biddle 1979; Stryker
and Statham 1985; Mead 1947; Goffman 1959). Together these disciplines also
constitute the origins of the transdisciplines of social and political psychology
in which our general approach is located (Sarbin and Allen 1968; Stryker and
Statham 1985; Tetlock 1998; Sears et al. 2003).
The relevance of role theory for understanding world politics has been rec-
ognized since at least the 1930s (K. Holsti 1970; Walker 1987c; Thies 2010;
Harnisch 2010). The application of role theory in the concluding chapters of this
volume to synthesize and summarize the analyses of decision making and learning
presented in earlier chapters identifies the important processes and relations that
link psychological processes and social processes in world politics. These links are
specified with insights from mathematical game theory models of symbolic and
strategic interaction provided by the research program in operational code analy-
sis, which examines the links between the political belief systems of leaders, the
foreign policy decisions of states, and the international outcomes of cooperation,
conflict, domination, and submission in world politics (Walker 2003; Schafer and
Walker 2006a; Walker and Schafer 2010).
The general argument in this book is that the smaller, “world in their minds” of
individuals as actors and the larger, “world of events” that states as actors inhabit
are connected in interesting and important ways (Vertzberger 1990; Wendt 1999;
6 Stephen G. Walker

Lake and Powell 1999a). Seeing those connections is partly a conceptual matter
of having a unifying theory to anticipate the connections and partly an empirical
matter of having the proper instruments to observe them. Our goal is to demon-
strate the theoretical and empirical patterns that connect these two worlds. This
effort engages a larger set of arguments about “behavioral IR,” which involves
the relationship between the subfield of Foreign Policy Analysis and the field of
International Relations (Mintz 2007). In turn, the issues raised in this argument
about the organization of knowledge in the domain of world politics touches
on larger questions about the acquisition and organization of knowledge in the
natural and social sciences within the framework of complex systems analysis
(Mitchell 2009; Axelrod 1984; Jervis 1997). Therefore, we have framed our argu-
ment about the worlds of beliefs and behaviors in this book within the context of
these two larger conversations.

Foreign Policy Analysis and Behavioral International Relations

Alex Mintz has advanced “behavioral IR” as an important social-psychological
approach to Foreign Policy Analysis (FPA) and International Relations (IR). This
approach identifies and solves some basic problems unique to the former as a
subfield of the latter. Foreign Policy Analysis as an agent-centered, micropolitical
study of decisions by leaders is usually subordinated to IR as a structure-oriented,
macropolitical study of interactions in regional or global international systems. It
is possible and even desirable to focus on the interactions of states as actors to ana-
lyze large-scale, long-term, historical trends and shifts in world politics. However,
it is also appropriate to focus on individuals and small groups as actors within states
and analyze the small-scale, short-term behaviors that produce patterns of con-
tinuity and change in larger political systems. It is particularly the case in foreign
policy making, as so many major decisions affecting global politics are made by a
small number of individuals.
The application of a behavioral approach to individuals as actors in world poli-
tics embodies a set of concepts, methods, and heuristics as a research program. We
identify and present applications of these characteristics of “behavioral IR” within
the context of the operational code research program in foreign policy analysis.
Operational code analysis as a neobehavioral approach to politics focuses on the
explanation of foreign policy decisions and their consequences by reference to two
political worlds: (a) the external world of events, generated by the presence, power,
and behavior of other actors; (b) the internal world of beliefs, generated by the
cognitive, emotional, and motivational processes of leaders. A leader’s operational
code or belief system connects these two worlds by representing the external world
of events as philosophical beliefs about the nature of the political universe and by
prescribing strategies, tactics, and moves based on instrumental beliefs for making
decisions about the exercise of power vis-à-vis other actors in the political universe
(Leites 1951; George 1969).
 Foreign Policy Analysis and Behavioral IR 7

We identify operational code analysis as a neobehavioral approach, because it

combines features of two older research programs or traditions in the study of
foreign policy and international relations. It employs both the concepts of ratio-
nality and power and the concepts of beliefs, emotions, and motivations. These
concepts and the theoretical perspectives that they engender are associated with
rational choice and behavioral research programs, respectively, which have tradi-
tionally been viewed as rivals in the IR field and its subfield of FPA (Oye 1986;
Downs 1989; Geva and Mintz 1997; Lake and Powell 1999a; Bueno de Mesquita
and McDermott 2004). The former employs more structure-oriented theories of
International Relations while the latter focuses primarily on agent-based theories
of Foreign Policy Analysis.
Arguably, the rise of both rational choice and behavioral approaches to world
politics have their roots in realist international relations theory (Bueno de Mesquita
1981; Vasquez 1983); however, their evolution as separate research programs has
been the dominant pattern until very recent times when a relaxation in tensions
between the rivals began to occur (Bueno de Mesquita and McDermott 2004;
Levy 2003: Herrmann 2003). If this détente evolves into an entente of collabora-
tive engagement across research programs, the result may well be a “new” (neo)
behavioral approach in which the concept of rationality is informed, defined, and
modeled by the concepts of beliefs, emotions, and motivations in applications
to the study of international relations and foreign policy. The evolution of the
research program in operational code analysis is an example of such a new alliance
between these old rivals (Walker 2003).

Old Approaches to Foreign Policy Analysis

Traditional approaches to the study of foreign policy divide primarily between
focusing on “states-as-actors,” whose decision making is governed by constraints
and incentives associated with their membership in the international system, and
“individuals-as-actors” who make decisions under the influence of a broad range
of additional constraints and incentives located within states and even within indi-
viduals (Wolfers 1962; Hudson 2005). In this book we approach foreign policy
analysis primarily at the level of the strategic dyad and focus primarily on individuals
as actors (Richardson 1987; Walker 2002). Our assumption is that dyads in world
politics are “where the action is” no matter whether the approach to the field of
International Relations is from the “outside-in,” state-systemic level of analysis or
from the “inside-out,” individual-decision-maker level of analysis (Walker 1993).
Analysts often prefer to choose only one or the other focus, which has led to a
bifurcation between the study of foreign policy and international relations phe-
nomena. Foreign Policy Analysis is conceptualized as a subfield embedded within
the field of International Relations; however, it is often treated as separate from
the larger field, which leaves a gap between the two that is not explicitly bridged
with a common set of theories, methods, or metrics for analysis. Lip service may
be paid to the need for multi-level explanations, but the more common responses
8 Stephen G. Walker

to this gap are to: (a) ignore it; (b) pit the two levels of explanation against one
another as rivals; or (c) separate the combatants on the basis of a division of labor
between process-oriented explanations of foreign policy and outcome-oriented
explanations of international relations.

Behavioral IR: Fad, Fantasy, or (sub)Field?

To paraphrase part of the title of an essay by James Rosenau (1968) many years
ago about the viability of establishing a field of comparative foreign policy, is the
behavioral approach to IR a “fad, fantasy, or (sub)field?” Rosenau’s Question
echoes a cautionary tale from the distant past when the behavioral revolution
in political science initially spread to the study of International Relations. This
development was arrested in its application to Foreign Policy Analysis when
scholars in that era faltered in their efforts to find cost-effective solutions to the
problems of accessing necessary data about individual leaders, decisions, and
behavior (Walker 2000c, 2007a). Some caution is warranted about claiming too
little or too much on behalf of “behavioral IR” as an approach to foreign policy
analysis even though significant progress has accelerated over the last decade or so
in solving these problems with automated measurement strategies applied “at-a-
distance” to texts (Mintz 2007; Schafer 2000).
Therefore, it is worth considering explicitly whether this approach is a fad, a
figment of imagination, or a (sub)field. It seems appropriate to look first at what
a “field” is in order to assess whether behavioral IR is a subfield, something
more, or merely a fad or fantasy. Field comes from an old English noun (feld) that
referred originally to:
1a. An open, level expanse of land. b. A meadow. c. A cultivated expanse of
land. d. A portion of land or a geological formation containing a specified
natural resource (e.g., oil or minerals). 2. A battleground. 3. A background
area, as on a flag. 4. Sports. a. An area in which an athletic event takes
place. b. All the contestants in an event. 5a. An area of human activity. b.
Profession, employment, or business. c. A setting of practical activity out-
side an office, school, or laboratory. 6. Phys. A region of space characterized
by a physical property, such as gravitational force, having a determinable
value at every point in the region. 7. Comp.Sci. A defined area of a storage
medium, such as a set of bit locations, used to record a type of information
consistently. Syns: field, bailiwick, domain, province, realm, sphere, territory n.
(American Heritage Dictionary, 3rd edn, New York, 1994: 313.
Italics and bold are in source)
These various meanings of “field” as a noun divide into static and dynamic con-
notations with the first five meanings implying a spatial dimension and the last
two meanings evoking a dynamic dimension. Is behavioral IR a field in any
of these senses? A subfield? Or should it be defined as something else? Our
 Foreign Policy Analysis and Behavioral IR 9

position in this book is that behavioral IR as an approach to Foreign Policy

Analysis is both something more and something less than a field or subfield. It may
have some of the attributes that define one or more of the meanings of “field,”
e.g., it may be a battleground or an arena of competition and contestation, but
they are not its core defining characteristics. It is more precise and profitable
to consider it as a research program that coexists with other research programs
within the field of International Relations or within one of its various subfields,
e.g., Foreign Policy Analysis, Security Studies, Political Economy, Diplomatic
Studies, or International Organization.
If behavioral IR is conceptualized as a research program, then what is a research
program and how does behavioral IR fit this designation? The construct of a
research program is contested in the philosophy of science; however, a simple
definition is that it is a community of scientists who share assumptions about
both what to study—a subfield—and how to study it—an approach to the study
of a field or subfield. As a research program behavioral IR is located in a subfield,
e.g. Foreign Policy Analysis, and demarcated within that subfield primarily by
theoretical constructs and secondarily by methodological commitments. Defining
“how” in a research program is primarily conceptual rather than empirical; how-
ever, methodological assumptions form the core of a research program in addition
to its theories. These assumptions become positive heuristics about what to study
and how, while their corresponding exclusions are negative heuristics about what
and how not to study (Lakatos 1970; Elman and Elman 2003).
There may be both theoretical and methodological diversity within a research
program at two levels of organization. Theories may coexist and even compete
or cooperate across different research programs and also within the same research
program (Elman and Elman 2003). Examples of such relationships are the patterns
of rivalry and rapprochement between psychological theories in the behavioral
research program and decision-making theories in the rational choice research
program within the field of IR (Geva and Mintz 1997; Bueno de Mesquita and
McDermott 2004). Similar relationships exist between different psychologi-
cal theories of cognition and affect within the behavioral IR research program
(Walker 2002; Tetlock 1998; Tetlock and Levi 1982).
Philosophers of science differ over which pattern—revolution, rivalry, or coop-
eration—characterizes scientific progress for research programs within and across
subfields or fields of inquiry (Kuhn 1962; Lakatos 1970; Laudan 1977; Elman and
Elman 2003). Perhaps the most cogent and congenial analysis is a pluralist account
of complex interdependence in the form of short-term alliances (cooperation as
well as some rivalry and occasional revolution) across theories, which may extend
in the longer term across research programs, evolving endlessly over time as the
processes of reaching commensurability (common empirical metrics), entailment
(common theoretical ideas), or consilience (disparate facts in accord with one
another) pose and solve theoretical and empirical puzzles while leaving others to
be investigated with the same processes (Laudan 1977; James 2007).
10 Stephen G. Walker

Metaphorically, therefore, research programs occupy “space” in a field as does

a thicket in a meadow, a natural resource in a geological formation, a contes-
tant or combatant in an athletic event or battlefield, a physical force in a region
of space, or bits of information located in a defined area of a storage medium.
Examples from other disciplines with behavioral approaches, e.g., behavioral
finance, behavioral marketing, and behavioral economics, are research programs
within subfields of Economics. These possibilities suggest the need to identify a
set of markers for defining and locating behavioral IR as a set of research programs
in the various subfields of International Relations.
Mintz (2007: 157–158) identifies six characteristics that serve as markers for a
research program in behavioral IR. They are its assumptions, actors, concepts,
levels of analysis, methods, and questions of interest, which are addressed at some
length in the next three chapters of this book. Here the focus is on the larger
gestalt or configuration of these characteristics that captures the patterns emerging
from their interaction. The first pattern is an emphasis on the scientific method as
a methodological principle to govern the empirical practice of behavioral IR. The
second pattern is an emphasis on the rationality theorem as a theoretical principle
to unite the conceptual thinking of behavioral IR. These principles are contro-
versial taken separately and sometimes even pitted against one another (Ricci
1984; Green and Shapiro 1994; Geva and Mintz 1997). Their alliance or amalga-
mation pattern rather than a pattern of rivalry or enmity may also be construed as
a sign of scientific progress (Walker 2003; Laudan 1977). The synthesis of behav-
ioralism and rationalism with the employment of these two organizational prin-
ciples is called in this book a neobehavioral approach to the study of Foreign Policy
Analysis. These principles are also the touchstones for synthesizing macro-level,
international relations phenomena and micro-level, foreign policy phenomena
under a common theoretical and methodological system of principles.

Foreign Policy Analysis and the Scientific Method

Although our focus of inquiry is politics, the intellectual spirit guiding this volume
is the application of the scientific method of “observation, reason, and experi-
ment” initially associated with the study of physics (Feynman 1995: 24). The sci-
entific method was developed as a means for understanding physical phenomena.
While understanding physical phenomena (such as the stars or the planets in the
galaxies) is not the same task as studying social-psychological phenomena (such
as the movie stars in Hollywood or the political psychology of decision making
by elites), there are some important parallels in conducting scientific modes of
inquiry into physical and political phenomena.
One is that “understanding” can mean the same thing to physicists and political
scientists, as reflected by this example from Nobel Physicist Richard Feynman’s
introductory lectures for physics majors at Caltech, in which he compares a game
played by the gods metaphorically to a game played by humans:
 Foreign Policy Analysis and Behavioral IR 11

What do we mean by “understanding” something? We can imagine that

this complicated array of moving things which constitutes “the world” is
something like a great chess game being played by the gods, and we are
observers of the game. We do not know what the rules of the game are;
all we are allowed to do is watch the playing. Of course, if we watch long
enough, we may eventually catch on to a few of the rules. The rules of the
game are what we mean by fundamental physics.
(Feynman 1995: 24. Italics his)

Feynman’s use of a chess metaphor from human social phenomena to illustrate

what and how physicists study physical phenomena is rather ironical, because the
major criticism of the behavioral approach to politics is that emulating the physical
sciences in trying to understand politics (including world politics) with the scien-
tific method is inappropriate on at least three grounds: (1) it is simply a categorical
error to study humans with the same (scientific) method used in the physical sci-
ences because they are two quite different kinds of phenomena; (2) human behavior
is much more complex than physical behavior and cannot be observed (mea-
sured) with the same precision; (3) the goal of the physical sciences is prediction,
which is too ambitious for the study of human behavior (Ricci 1984; Hoffmann
1960; Alker 1966; Ashley 1984; Bueno de Mesquita 1981; Little 1991: 222–237;
Monroe 2005).
Feynman shows that these criticisms are actually far-fetched, as he does not
subscribe to the characterization of physics contained in them in continuing to
use the chess metaphor:

Even if we knew every rule, however, we might not be able to know why
a particular move is made in the game, merely because it is too complicated
and our minds are limited. If you play chess you must know that it is easy
to learn all the rules, and yet it is often very hard to select the best move or
to understand why a player moves as he does. So it is in nature, only much
more so; but we may at least be able to find all the rules.
(Feynman 1995: 24)

These words debunk the stereotype of physicists who implement the scientific
method and reach highly specific predictions either about the future or about par-
ticular changes in nature, based on principles (rules) that “explain” everything.
Feynman next tackles questions raised by skeptics of all scientific knowledge
who ask, “How do we know (that we know), and how much can we know?”
His answers to these two questions vary. Feynmann (1995: 25) is rather modest
regarding how do we know (that we know) the rules, identifying three ways: (1)
isolating simple situations (only a few chess pieces for study); (2) checking obser-
vations against more specific rules derived from general rules (how a specific chess
piece moves); (3) rough approximation (protecting the king without specifying
12 Stephen G. Walker

how). Feynman is more ambitious regarding the aim of physics, in which “the
aim of physics is to see complete nature as different aspects of one set of phenom-
ena. That is the problem in basic theoretical physics today—to find the laws behind
experiment; to amalgamate these classes” (Feynman 1995: 26. Italics his).
Should this goal of amalgamation also be the aspiration of a social science like
the study of international politics and foreign policy? The answer is probably the
same for politics and physics. In Feynman’s (1995: 24–25) words:

Aside from not knowing all of the rules, what we really can explain in terms
of those rules is very limited, because almost all situations are so enormously
complicated that we cannot follow the plays of the game using the rules,
much less tell what is going to happen next. We must, therefore, limit our-
selves to the more basic question of the rules of the game. If we know the
rules, we consider that we “understand” the world.

We argue in this book that in the political universe the answer to Feynman’s
Question is that there are two basic elements that constitute the field of world
politics, in which “field” is conceptualized dynamically the way it is in physics and
computer science, respectively, as cognitive beliefs or behavioral actions occupy-
ing space as systems in a region.
Rationality and power are the respective conceptualizations of cognition and
behavior whose processes are systematically linked and offer theoretical explana-
tions of patterns of continuity and change over time in the political universe. One
way to picture these dynamic fields is in Figure 1.1, which portrays their relation-
ships over time between two systems: the “world of beliefs” and the “world of

TWO WORLDS OF POLITICS

ACTORS ACTIONS
World of Beliefs World of Events
Beliefs Intentions Signals Indices

THEOREMS

Rationality

Power

OBSERVATIONS

Words/Deeds

Time “Language is the medium” Time

FIGURE 1.1 Two Systems in World Politics

 Foreign Policy Analysis and Behavioral IR 13

events.” Three basic entropy propositions link the two processes of power and
rationality that define the fields in Figure 1.1: (1) The World of Beliefs Proposition
says that there is a set of different beliefs in the political universe, whose elements
define the possible “states of mind” (beliefs about the ways of exercising rational-
ity) and constitute the entropy or randomness of the world in the minds of actors;
(2) The World of Events Proposition says that there is a set of possible political actions
in the political universe, whose elements define the different “states of behav-
ior” (ways of exercising power) and constitute the entropy or randomness of the
world of events; (3) The Two Worlds of Interaction Proposition says that interactions
between the world of beliefs and the world of events reduce the entropy (possible
states) in each world by reducing the possible states of mind and behavior and
thereby explaining stability and change in each world. As the arrows in Figure 1.1
indicate, the two worlds of interaction proposition has two theorems, a rationality
theorem and a power theorem corresponding to the two directions of causality in
the proposition, which are discussed below.
The conceptualization of foreign policy decisions as choices by leaders during
strategic interaction episodes between states leads us to the formulation of the
rationality theorem in order to state what constitutes rationality. What exactly is
the rationality theorem? It specifies that rational decisions are purposeful, mindful,
and empathetic in varying degrees. Simon (1985: 294) defines “rational” more
generally as denoting “behavior that is appropriate to specified goals in the context
of a given situation” and distinguishes two forms of rationality, global substantive
rationality and procedural bounded rationality, which constitute two versions of
the rationality theorem:

Substantive Rationality. If the characteristics of the choosing organism are

ignored, and we consider only those constraints that arise from the external
situation, then we may speak of substantive or objective rationality, that is,
behavior that can be adjudged objectively to be optimally adapted to the
situation.
(Simon 1985: 294)
Bounded Rationality. If we take into account the limitations of knowledge
and computing power of the choosing organism, then we may find it inca-
pable of making objective optimal choices. If, however, it uses methods
of choice that are as effective as its decision-making and problem-solving
means permit, we may speak of procedural or bounded rationality, that is,
behavior that is adaptive within the constraints imposed both by the external
situation and by the capacities of the decision maker.
(Simon 1985: 294. Italics his)

Simon (1985) also recognizes the possibility of radical irrationality, i.e., behavior
that is based solely on emotional impulses unmediated significantly by thought
14 Stephen G. Walker

processes or external constraints, but he argues that this total violation of the
rationality theorem is relatively rare. We shall examine this assumption more
closely in Chapter 3.
The exercise of power can take many forms during strategic interaction epi-
sodes between states, as each state attempts to shape the behavior of the other by
means of foreign policy choices to exercise either positive or negative sanctions
in dealing with their interaction with one another. This answer to the “What is
power?” question aligns our approach broadly with the realist school of interna-
tional relations theory, in which “International politics, like all politics, is a strug-
gle for power” (Morgenthau 1978: 29). The struggle for power is a systemic-level
process, which does not directly address the question of “What is rationality?” As
noted above, the answer to this question is found by focusing on agent-level pro-
cesses, which aligns our approach as well with the decision-making approach to
international politics and foreign policy (Snyder et al. 1962; Vertzberger 1990).
Therefore, our exact location is at the nexus between the two approaches
with a focus on the agent and micro-structural levels represented by the strategic
dyad, i.e., strategic decisions between pairs of actors informed by definitions of
power and interests contained in the beliefs of their leaders (Wendt 1999; Walker
2004a). This position invokes two versions of the power theorem based on dif-
ferent conceptualizations of power (see Baldwin 2002):

Elements of Power. This conceptualization of power emphasizes the distribu-

tion of resources that are necessary for one actor to exercise control over
another actor in a strategic dyad. In order for A to control B, military weap-
ons, wealth, population, technology, organization, ideology, and other fac-
tors may be required (Hartmann 1983; Morgenthau 1978; see also French
and Raven 1959). One actor is judged more or less powerful than another
actor, depending on the relative resource endowment of each one and the
extent to which these elements of power are mobilized for use in a given
situation at a given point in time.
Exercises of Power. This conceptualization of power emphasizes the use of
words or deeds by Actor A to control the behavior of Actor B, which
employs one or more elements of power to achieve the desired effect. We
distinguish six basic forms of behavior (words or deeds) that constitute dif-
ferent ways to exercise power: reward, promise, appeal/support, resist/
oppose, threaten, punish. Others have identified additional forms that are
extensions of these types (Bachrach and Baratz 1962; Lukes 1974; see also
Nye 1990).

In each version of the power theorem Actor A engages in an effort to get Actor
B to do what it would not otherwise do (French and Raven 1959; Dahl 1957;
McClelland 1966; Baldwin 1989). The basis of this effort distinguishes the exer-
cise of one form of power from another. Ideas as elements of moral authority
 Foreign Policy Analysis and Behavioral IR 15

or expertise are the basis for the exercise of appeal/support and oppose/resist
behavior as forms of power; present or future material sanctions as elements of
power are the basis for exercising rewards (positive sanctions) and punishments
(negative sanctions) or promises (future positive sanctions) and threats (future
negative sanctions).
A focus on strategic decisions between pairs of actors, informed by the defini-
tions of power and interests contained in the beliefs of leaders, makes politics a
social process (Wendt 1999; Lake and Powell 1999b). The relationships and pro-
cesses that occur within this dyad are characterized by the exercise of power in the
world of events and rationality in the world of beliefs. Our theoretical approach
in understanding the rules that govern the interaction between these two worlds
is to identify one as the system of interest and the other as the environment in which
the former operates. If both worlds are systems isolated from one another, they
can have possible states of maximum entropy in their respective configurations of
beliefs and behavior.
However, the two worlds of interaction proposition stipulates that they are not
isolated systems and that the actual configuration in one world affects the actual
configuration in the other world, which reduces the entropy of each world. That
is, the “state of mind” in the world of beliefs influences the “state of behavior” in
the world of events, and vice versa. Specifying one of these worlds as the system of
interest makes its state of entropy the explanandum, or effect to be explained; the
other world becomes the environment, making its state of entropy the explanans,
or cause of the state of entropy in the system of interest. The concepts of power
and rationality describe and specify the processes that affect the level of entropy
across these respective worlds. Their respective configurations of behavior and
beliefs become congruent, i.e., map on to one another, as their respective systems
interact over time (Walker 1977; George and Bennett 2005).
If behavioral patterns in the exercise of power as the system of interest become
congruent with cognitive patterns of beliefs about the exercise of power, the cog-
nitive environment in the world of beliefs acts as information to shape the level of
entropy in the world of events. More simply, beliefs about the exercise of power
have steering effects (continuity or variation) on behavior as the exercise of power.
If the system of interest is specified as the world of beliefs and the environment
as the world of events, then the roles of explanandum and explanans are reversed.
Then configurations of behavior have learning effects (re-enforcement or altera-
tion) on configurations of beliefs. How is this exchange of effects across the fields
of rationality and power accomplished between these two systems?
In order for the processes of rationality and power to occur, there must be social
contact between the actors in the strategic dyad. The members must be aware of
each other in order for steering effects and learning effects to occur (Searle 1995).
Each member must also have the capacities of rational thought (purposefulness,
mindfulness, and empathy) in order to diagnose and calculate their own actions
based on beliefs and to be able to alter those beliefs to mirror the information
16 Stephen G. Walker

coming from the other member of the dyad (Searle 1995; Edelman 1992; Walker
and Schafer 2010). These capacities generate forces that can vary along with the
awareness level across dyads and between members of a dyad; these variations can
account for variations in steering and learning effects as the members of a dyad
interact over time.
The medium for making social contact between the members of a strategic
dyad is language. Words and deeds reported as “text-actions” in the public state-
ments of leaders and in the reports of journalists contain attributions to self and
others that reflect beliefs of their leaders and the strategic interactions between
them. These documents contain observations from which it is possible to infer
the exercise of rationality and the exercise of power. Scholars access these obser-
vations “at a distance” in much the same way that astronomers access observa-
tions of the sun and other stars with instruments that measure “at a distance”
what data are available. Whereas physical scientists can measure the spectrum of
light emanating from stars and make inferences about heat and other processes
that create it, political scientists can measure the spectrum of speech emanating
from governments and draw inferences about thought and other processes that
constructed them (Hawking with Mlodinow 2005; Schafer 2000; Schafer and
Walker 2006c).
In this book we take the concepts of rationality and power and build a model of
aspects of the political universe encompassed by these concepts. The model for-
malizes the systems of thought and action regarding the exercise of rationality and
power that are represented in texts of belief inventories and event histories. We
stipulate a set of rules that relate properties of beliefs and behavior to observations
that we can make about political belief systems and the exercise of power between
members of strategic dyads. Our ultimate goals are to describe large sets of beliefs
and behaviors based on this model and make associations between them based on
successive observations of the worlds of beliefs and events over time. In order to
make these observations logical and coherent, we shall link them with exhaustive
and mutually exclusive rules that allow us to distinguish between random and
congruent combinations of beliefs and behavior.
If this effort is successful, we will reduce the level of entropy in each of
these worlds and thereby explain patterns of continuity and variation between
them as well as change within them over time from initial states of thought
and action to final states or equilibria in the political universe. This endeavor
is worthwhile because existing models of foreign policy decisions are
underspecified. They do not reduce the entropy of actions and thoughts in the
political universe to the point where there is very little left to explain. In statisti-
cal terms, today’s sophisticated rational choice models account for approximately
half of the variance in political behavior (Bueno de Mesquita et al. 2003; Vasquez
1983).
The rejoinder to this observation, that the remaining unexplained variance is
due to idiosyncratic factors, is only partly satisfactory at best. While the chemistry
 Foreign Policy Analysis and Behavioral IR 17

of politics in the form of different combinations of regimes, cultures, and other

elements of political systems than those represented by power and rationality
may be relevant, it may also be the case that one or the other of these two more
fundamental properties of the physics of politics are underspecified. If we are
successful in explaining the exercise of power in world politics with a robust
behavioral model based on richer and more rigorous conceptualizations and mea-
surements of rationality and power, then we can claim to make scientific prog-
ress in the study of International Relations and Foreign Policy Analysis without
adding more elements (Forward 1971; Elman and Elman 2003).

Models of World Politics

The billiard ball model of IR in Chapter 2 of this book is specified by six realist
propositions regarding power and interests, which assume the substantive ratio-
nality version of the rationality theorem. If the distributions of power and interests
are specified as constraints that arise from macrofoundations (distributions of power
and interests) in the external situation, then states-as-actors will rank preferences
according to the six propositions of this realist theory and make logical choices
that are consistent with the rules of play and the two-sided information condition
specified as rational decision making and modeled by ordinal game theory. This
billiard ball model of IR with states-as-actors assumes that the beliefs, motiva-
tions, and emotions of the individuals-as-actors who are the leaders of those states
are not relevant for explaining those decisions and predicting their outcomes
(Wolfers 1962).
The behavioral model of IR in Chapter 3 contests this assumption by employ-
ing the bounded rationality version of the rationality theorem together with the
same six realist propositions regarding the exercise of power and modeling them
as well with ordinal game theory. Leaders in this account make foreign policy
decisions about the exercise of positive and negative sanctions based on micro-
foundations, including the beliefs of individuals-as-actors regarding the exercise of
power by others, the utility of different forms of power for realizing fundamental
political interests as outcomes, and the distribution of power between self and
others (George 1969, 1979).
It is important to acknowledge that employing even a bounded version of the
rationality theorem in the construction of a subjective game of strategic interaction
is somewhat controversial. Herbert Simon (1985: 293), the father of the bounded
rationality concept and one of only two political scientists (in addition to Elinor
Ostrom) to win a Nobel Prize for his research, warns that rationality specified as a
general principle in an abstract model such as game theory, “unless accompanied
by extensive empirical research to identify the correct auxiliary assumptions, has
little power to make valid predictions about political phenomena.” He goes on to
argue that these auxiliary assumptions usually act as boundaries to rationality and
involve information about actors’ “representation of the world in which they live,
18 Stephen G. Walker

what they attend to in that world, and what beliefs they have about its nature”
(Simon 1985: 300). Regarding game theory’s utility, Simon concludes that:

Perhaps the major contribution of game theory to political science has been
to demonstrate how rare and unusual the situations are where a game has
a stable equilibrium consistent with the principle of objectively rational
choice. Under these circumstances, the task of determining how people
actually do behave in situations having game-like characteristics must be
turned over to empirical research: research that seeks to determine what
values people actually act on, and how they form their expectations and
beliefs.
(Simon 1985: 300)

At first glance, it would seem that Simon dismisses game theory as underspecified
and therefore not useful for understanding decision making. Others have rel-
egated it to a normative role for making good decisions (Stein 2007). However,
some political psychologists and rational choice theorists now see the possibility
of a synergistic relationship between psychological and rational choice approaches
to the study of strategic interaction (Lupia et al. 2000; Walker 2002; Herrmann
2003; Levy 2003; Bueno de Mesquita and McDermott 2004).
It is accurate to say that classical game theory models make relatively heroic
assumptions about how rational, i.e., how purposeful, mindful, and empathetic
humans are in making decisions. In this book one of the main tasks is to show that
it is possible to use a game theory matrix to represent successively more complex
levels of rationality: (a) purposeful decision makers represented by exchange theory
and role theory who are primarily “ends-oriented” with a utility-maximization
version of the bounded rationality theorem; (b) mindful decision makers who
are “ends-means-oriented” but not other-oriented with prospect theory’s version
of the bounded rationality theorem; and (c) empathetic decision makers who are
“ends-means-other-oriented” with game theory’s version of the bounded rational-
ity theorem as a subjective game (Maoz 1990; Maoz and Mor 2002; Walker and
Schafer 2010).
As we shall demonstrate in the next two chapters, crossing psychological and
rational choice models of foreign policy analysis generates a neobehavioral model
of IR, in which theories from the older behavioral research program and the
newer rational choice program form a theoretical alliance (Herrmann 2003; Levy
2003; Bueno de Mesquita and McDermott 2004; Walker 2002, 2003, 2007a;
Walker and Schafer 2010). This alliance resembles previous developments in
economics where classical rational choice models are giving way to a mix of
psychological-behavioral and rational choice models to explain the dynamics of
economic decision making, including: (1) decisions by groups inside firms; (2)
the strategic interactions between firms to set the prices of goods and services;
(3) interactions between investors and the stock market regarding the buying
 Foreign Policy Analysis and Behavioral IR 19

and selling of stocks; and (4) interactions between investors in the stock market
and the government as the latter makes economic policy decisions (Mintz 2007;
Camerer 2003; Thaler 1994).
The emergence of similar developments in Foreign Policy Analysis and
International Relations may represent a potential renaissance of the behavioral
revolution in political science, i.e., a “new” behavioral movement that combines
the positive features of political psychology and rational choice research programs.
In the next two chapters we address both the promise and the problems associ-
ated with the application of such a neobehavioral approach to solving problems
of foreign policy and international relations theory.

Overview of the Book

We turn now to the task of understanding the basic rules of political phenomena in
Part I of our book with the behavioral approach and the scientific method applied
to Foreign Policy Analysis. Chapter 2 examines the theoretical and behavioral
links between the macropolitics of International Relations and the micropolitics
of Foreign Policy Analysis by comparing and contrasting their relationships with
macroscopic and microscopic relationships in other disciplines, such as macro and
micro economics and especially macroscopic (classical) and microscopic (modern)
physics. A discussion of the rationality theorem and the analysis of how leaders
decide and learn in Chapters 2 and 3 unifies macroscopic and microscopic units of
analysis (systems of states and individuals) and thereby the field of International
Relations with the subfield of Foreign Policy Analysis. The discussion in Chapter
4 of methodological problems and research designs with examples shows inter-
ested readers how to carry out as well as understand this kind of neobehavioral
foreign policy analysis.
Part II contains extended applications of the potential research designs intro-
duced above to Foreign Policy Decision Making, which are intended to show
how the tools of game theory, content analysis, and statistics discussed in Part I
can be applied to single-case studies, structured-focused comparisons, and more
complex extensions of these research designs for doing foreign policy analysis
across time and space. Theoretically, these chapters emphasize different levels of
complexity in the application of the rationality theorem, namely, that decision
makers are purposeful, mindful, and empathetic. The operational codes or belief
systems of leaders are linked with foreign policy decisions or their antecedents
under different conditions and at different levels of aggregation across individuals,
groups, and states.
The purposefulness (ends-orientation) of leaders is the focus of Chapter 5, which
applies the rationality theorem and the methods of behavioral IR to an analysis
of some “difficult” cases, namely, rogue states in Latin America, Asia, and the
Middle East. The mindfulness (means-orientation) of the North Korean leaders
is the focus of Chapter 6’s analysis of North Korea’s risk-taking in deciding to
20 Stephen G. Walker

pursue the acquisition of nuclear weapons. Solutions to some analytical problems

of aggregation facing students who want to apply the rationality theorem to larger
units than individuals are illustrated in two subsequent chapters. Chapter 7 deals
with the phenomenon of single-group dynamics and the quality of decision mak-
ing within states. The empathy (other-orientations) of leaders in relations between
allies is addressed in Chapter 8’s examination of decision making by France and
Britain in the 9/11 era.
Part III also presents extended empirical applications of research designs dis-
cussed earlier in applying formal models, content analysis, and statistics to foreign
policy analysis; however, the primary focus shifts from decision-making processes
to Foreign Policy Learning. The cognitive learning processes associated with the
operational codes or belief systems of leaders are the principal mechanisms ana-
lyzed in these chapters. In Chapter 9 there is an analysis of learning by cold
warriors Fidel Castro and Kim Il Sung, which returns to a theme raised earlier
in Chapter 5 about how to deal with leaders who may refuse to “learn.” The pat-
terns of stability and change in the beliefs of U.S. President George W. Bush over
time are the focus of Chapter 10, which also discusses extensively the concept of
learning and how to detect it. The sources of experiential learning for three U.S.
presidents is the focus of Chapter 11, an examination of different possible sources
to account for the presence of changes in the belief systems of Presidents George
H. W. Bush (Bush 41), William Jefferson Clinton, and George W. Bush (Bush
43). Leaders learning lessons (changing beliefs) as a result of their experiences in
foreign policy crises is the focus of Chapter 12’s analysis of changes in the formal
models of strategic rationality represented by these beliefs. The last two chapters
also expand exponentially the number of observations under analysis, as they illus-
trate the possibilities for large-scale, empirical comparisons of linkages between
the external world of foreign policy behavior (events) and the internal world of
operational codes (beliefs).
Part IV brings together decision-making and learning processes of strategic
and symbolic interaction as Foreign Policy Dynamics, employing the methods of
content analysis and formal modeling on this more complex research problem.
Chapter 13 is a two-sided strategic interaction analysis of Presidents Bush (41)
and Clinton and rogue leaders Saddam Hussein and Slobodan Milosevic as they
attempted to manage crises over Kuwait and Kosovo. Chapter 14 synthesizes
symbolic and strategic interaction models of learning and steering processes with
a binary role theory for understanding the dynamics of foreign policy analysis. It
also expands and extends the application of role theory from two-sided (dyadic) to
three-sided (triadic) strategic interaction problems. Chapter 15 ends the volume
with a discussion of how binary role theory acts as a unifying theory to integrate
the agent-centered subfield of Foreign Policy Analysis and the systems-oriented
field of International Relations.
2
MACROPOLITICS AND
FOREIGN POLICY DECISIONS
The Billiard Ball Model of IR

Stephen G. Walker

Introduction
Foreign Policy Analysis (FPA) has long been recognized as a subfield of International
Relations (IR) within the discipline of Political Science. The relationship between
the study of foreign policy decision making by states and the study of the inter-
national system is similar in important respects to the study of economic decision
making by firms and the study of the larger economic system. Firms are members
of an economic system of firms just as states are members of a political system of
states. Similarly, firms may make their decisions primarily on the basis of constraints
and incentives governing their membership in an economic system or under the
influence of additional constraints and incentives located within the firm and the
individuals that constitute them (Wolfers 1962; Thaler 1994; Camerer 2003).
Approaches in various subfields of microeconomics, e.g., finance or market-
ing, which emphasize the importance of these additional internal constraints
within the firm, constitute instances of “behavioral economics” as an approach
to economic analysis. Approaches in various subfields of International Relations,
e.g., foreign policy analysis, which also emphasize the importance of “individuals-
as-actors” within the state, can similarly become the core of “behavioral IR” as
an approach to the micropolitics of foreign policy analysis (Wolfers 1962; Snyder
et al. 1954; Mintz 2007). The case for this approach to the field of International
Relations rests on the same arguments advanced on its behalf in an earlier era of
political science and in other social sciences. Approaches that either limit their
scope to the macroscopic level of analysis or rest on untested assumptions about
empirical phenomena at the microscopic level of analysis yield explanations that
are lacking, or at least limited, in validity.
22 Stephen G. Walker

Macro and Micro Approaches to Foreign Policy Analysis

The behavioral approach in political science initially arose in reaction to the limita-
tions of legalistic-institutional explanations of political phenomena, which failed
to explain political behavior with sufficient precision in the absence of informa-
tion about the attitudes and beliefs of individuals and groups within institutions
and societies (Eulau 1963; Ricci 1984). The behavioral revolution in political
science in the first half of the twentieth century spread to the field of International
Relations; however, it suffered a case of arrested development in the face of formi-
dable challenges for access to attitudinal and behavioral data about foreign policy
elites insulated from external scrutiny by scholars (Walker 2007a). Rational choice
models of foreign policy decision making were instead assumed to be valid and ini-
tially met the test of outcome validity during the cold war era (Jervis 1994; Hudson
2005; Walker and Schafer 2006). However, their power to describe and explain
the process of decision making was lacking, and with the end of the cold war the
ability of rational choice models to meet the tests of either outcome or process
validity became more limited (Jervis 1994).
At the same time the development of more sophisticated and cost-effective
means of accessing the psychological and behavioral characteristics of foreign
policy elites has made “behavioral IR” a better bet as an approach to micro-level
phenomena (Schafer 2000; Mintz 2007; Walker 2000, 2007a; Hudson 2005).
These developments and the increasing number of important anomalies “from
the traditional analytic, rational, expected utility model of choice” has led one
prominent scholar to propose that, “Behavioral IR increases the explanatory
power of the field of IR by providing it with more realistic psychological founda-
tions” (Mintz 2007: 157–158). He goes on (159–161) to nominate behavioral IR
as a new subfield of International Relations with several identifying characteris-
tics: (1) assumptions; (2) actors; (3) concepts; (4) levels of analysis; (5) methods;
and (6) questions of interest, all of which are summarized in Figure 2.1.

Characteristics Individual-as-Actor State-as-Actor

Model Behavioral IR Billiard Ball IR

Assumptions Individual differences matter States are homogeneous

Actors Individuals singly or in groups States singly or in groups

Concepts Thick Rationality Thin Rationality

Levels Situation-dependent System-dependent

Methods Measure Micro-Agency Measure Macro-Structure

Questions Actor-specific Decisions System-oriented Collisions

FIGURE 2.1 Behavioral and Billiard Ball Models of International Relations

 The Billiard Ball Model of IR 23

Mintz argues on behalf of the behavioral IR model, in which relatively het-

erogeneous individuals, singly or in groups, “decide” in a quasi-rational manner
to make actor-specific choices that can be observed in a variety of contexts with
different methods, depending on the research questions of interest. He argues
against the billiard ball IR model, which assumes states are relatively homogene-
ous actors who act rationally according to system-dependent, structural criteria.
Within those systemic constraints states “collide” like the homogeneous balls on
a billiard table (Wolfers 1962).
The distinction between decisions and collisions is at the core of the differences
between the “individuals-as-actors” and the “states-as-actors” approaches to IR,
because they rest on different conceptions of rationality. The billiard ball model
assumes a “thin” model of rationality in which leaders respond to reward and pun-
ishment contingencies specified by the structural balances of power and interests
among the states in the international system (Tetlock 1998). This assumption is
associated with realist international relations theory, which assumes in its classical
formulation that, “International politics, like all politics, is a struggle for power”
(Morgenthau 1978: 13). The classical realist assumption about power leads via
a simple rationality assumption (that actors are goal-oriented) to the neorealist
proposition that states “seek” (increase, maintain, or demonstrate) power in order
to survive in much the same way that classical economic models assume firms
“seek” (increase, maintain, or demonstrate) profit and respond to free market
conditions in order to survive (Morgenthau 1978; Waltz 1979; Keohane 1983).
Behavioral IR theorists question the assumption and accompanying proposi-
tions associated with realism, arguing instead that micro-level decisions are gov-
erned by a “thick” model of rationality in which leaders respond to additional
constraints and incentives (Snyder et al. 1954; Vertzberger 1990; Little 1991).
The primary effect of this behavioralist claim is to argue that a more robust theory
of agency is necessary in order to explain both foreign policy decisions by lead-
ers and the outcomes of collisions by states. A secondary effect of this claim is
to qualify the rationality theorem connecting power and behavior (Green and
Shapiro 1994; Geva and Mintz 1997; Vertzberger 1990; Downs 1989).
Rational decision making is a more complex task in behavioral theory, hence
the use of thick rationality as a description of foreign policy decision making. In
addition to selecting goals and choosing among alternatives under the conditions
of different distributions of power and interests in the international system, leaders
also confront constraints and interests defined by societal distributions of power
and interests inside the state (Wriggins 1969; Tsebelis 1990; Putnam 1988; Bueno
de Mesquita et al. 2003). Moreover, the capacity of leaders to recognize and assess
these constraints and incentives is limited by their own cognitive and motivational
biases (Jervis 1976; Stein 1988). These influences and the complexity of the politi-
cal context for making decisions may lead to suboptimal choices when evaluated
by invariant rational choice criteria. The decision-making process is better char-
acterized as quasi-rational under these conditions (Mintz 2007; Lau 2003).
24 Stephen G. Walker

The challenge created by the need for a thicker version of the rationality
theorem is how to construct such a model of foreign policy decision making.
General IR theory with states as actors has evolved so that realism’s billiard ball
model is supplemented by neoliberal institutionalist and social constructivist mod-
els, which emphasize that states as actors make rational choices conditioned by
the constraints of institutional rules, roles, and other social norms as well as power
distributions (Keohane 1984; Baldwin 1993; Lapid and Kratochwil 1996; Wendt
1999). These constraints are often specified as shared at the level of the state or
the international system rather than attributed directly to individuals as actors,
however, omitting the microfoundations of foreign policy decision making by
leaders and small groups.
While models of rationality for both state and individual actors have added
constraints to become “thicker,” these constraints differ in scope and duration.
For states as actors the democratic peace construct of republican liberalism, the
strategic culture construct derived from cultural realism, and the national identity
construct associated with social constructivism are examples of larger, more stable
constraints (Doyle 1986; Johnston 1995; Wendt 1999). For individuals as actors
acting singly or jointly, there are smaller, less stable constraints on rationality
such as the consistency-seeking and risk-framing cognitive processes of leaders
or the concurrence-seeking and risky-shift social processes of groups (George
1969, 1979; Kahneman and Tversky 1979; Janis 1982; Simmel 1979; Schafer and
Crichlow 2010).
The differences in spatial size and temporal stability across individuals and states
as actors may reflect the kinds of ontological and theoretical gaps existing between
macro-oriented and micro-centered branches of other disciplines, such as the
gap in studying energy between the classical physics of Newton and the modern
physics of Einstein. Although Newtonian physics has a rich history of accomplish-
ments in modeling the movements of large objects in the physical universe, it
offers a “thin” account of important processes at the microscopic level of physics.
For example, objects simply “collide” or “repel” in the world of classical physics
as a function of their relative positions and certain principles that summarize their
interactions without a satisfactory account of how their microfoundations gener-
ate these patterns of behavior. Einstein’s contribution was to construct a “thick”
model of these processes in terms of atoms and their constituent subatomic parts,
which accounted for both the movements and collisions of macroscopic objects
like billiard balls and also for their microscopic origins in atomic structures and
subatomic processes (Feynman 1995; Adler 2002; Brush 1988).
How useful is this analogy with physics and what is its relevance for making
the case for a “behavioral IR” model of foreign policy analysis? First, the analogy
points to the potential importance of distinguishing a temporal dimension (short/
long) as well as a spatial dimension (micro/macro) in foreign policy analysis. The
unit of analysis in physics is specified partly by its size and partly by its rate of
change, properties that are aspects of the inherently problematic nature of study-
 The Billiard Ball Model of IR 25

ing microscopic phenomena. The smaller atomic and subatomic particles are in
constant motion in the physical universe, which means that they are impossible
to measure precisely, as their constant movement changes their position so that
micro-level measurements are continuously outdated and therefore approximate
(Feynman 1995).
As atoms combine into molecules and compounds, they take on a qualita-
tively new identity as macroscopic objects and lose the probabilistic nature that
defines a microscopic view of the physical world. The study of larger and slower
objects narrows the gap between observations and temporal accuracy for measur-
ing macro-level phenomena (Feynman 1995). Some scientists in biology and the
physical sciences also distinguish an intermediate, “mesoscopic” level of organ-
ized behavior between the microscopic and macroscopic levels of organization,
which may have its own organizing principles or rules of behavior (Laughlin et al.
2000). Analogs in the study of politics might be organizations such as bureaucra-
cies or cultural constructs such as clans, which are larger and more complex than
groups or individuals but smaller and less complex than states or nations.
Second, these temporal and spatial relationships in physics suggest it is impor-
tant that foreign policy analysts calibrate measurements of actor properties and
behavioral properties in the political universe with common or at least propor-
tionate metrics for both space and time. If the primary actors are individuals, it
may not only be difficult to access them spatially as insulated elites; it may also be
difficult to keep up temporally with their constantly changing states of mind and
make accurate observations of the relationships between thoughts, feelings, and
motivations and their behaviors over time.
The application of this analogy from physics to world politics is to distinguish
between the thin accounts of the exercise of rationality and power offered by
macroscopic politics for collisions among states-as-actors in the international sys-
tem versus the thick accounts offered by microscopic politics for the decisions by
individuals-as-actors that explain those collisions and their origins. While both
states and individuals are agents within a larger system of international politics,
the behavior of states-as-actors is explained directly with general IR theories by
reference to structural properties or macrofoundations of the international system,
such as the distributions of power and interests, the rules of shared institutions,
and the roles defined by shared cultural norms or identities.
In contrast, the behavior of individuals-as-actors is explained directly with
foreign policy theories by reference to agent properties or microfoundations of
the international system, such as the belief systems and risk orientations of indi-
viduals acting singly, or to the properties of concurrence-seeking and risky-shift
patterns within groups when individuals act together, resulting in international
interactions and systemic outcomes. Ideally, one would want a common metric
and source of observation for such microscopic observations of both the beliefs
and behaviors of leaders, which would contain small changes in behavior over
small periods of time.
26 Stephen G. Walker

The most relevant candidate for the source of “real-world” observational data
(as opposed to simulated or experimental laboratory observations) may well be
texts that divide into two categories: (1) public or private statements by leaders
that record as primary sources relevant attributes of individuals and groups as
they make decisions; (2) accounts of events reported in secondary sources such as
newspapers and wire services or extracted from primary sources such as govern-
ment archives.
A key difference between these sources is whether they are spoken or written
as primary sources by officials or as secondary sources by journalists and histo-
rians. Both primary and secondary sources may report either actor attributes or
behavior, e.g., a news account may quote a public official and a public official
may announce an event. However, the difference is whether the official associates
himself or herself with the attribute or behavior directly as the speaker or author
of the text. Then the text represents “the world in their minds” while an unattrib-
uted account reported by an observer is the “world of events” (Hermann 1980;
Vertzberger 1990; Schafer 2000; McClelland 1966; Hermann 1971; Schrodt et
al. 1994).
It is relatively rare for the two worlds to overlap entirely either as statements
by leaders or as narratives of events by observers. The likely methods of iden-
tifying observations of interest from these sources are various forms of content
analysis developed by social scientists and automated by computer techniques to
retrieve, parse, and code texts with dictionaries (Young 2001; Smith 1992; Schafer
and Walker 2006c; King 1989; Hudson et al. 2008). The metrics available from
texts that measure both individual properties and social behavior are the various
parts of speech that can indicate attributes of a leader’s personality and a state’s
behavior.
Nouns and pronouns, adjectives and adverbs, conjunctions and verbs in a lead-
er’s statements indicate emotional, motivational, and cognitive attributes of the
speaker. When combined into larger indices, they can reflect larger constructs
of “personality,” such as the belief systems, motivational profiles, and affective
orientations of leaders. Making common or proportionate aggregations spatially
across individuals and temporally over time can represent as well the collective
orientations of single groups, coalitions, and even the state-as-actor (Smith 1992;
Post 2003; Hermann 2001).
Does a theoretical gap exist between lower and higher levels of spatial and
temporal aggregation in the study of International Relations, in which different
explanations apply under different scope conditions at different levels of aggrega-
tion (Waltz 1979; Elman 1996; Lake and Powell 1999b)? This question raises some
perennial issues in the philosophy of science about whether explanations at higher
levels of aggregation must be reduced, i.e., applied to lower levels, or whether it
is possible merely to explain a phenomenon at one level by reference to another
phenomenon at the same level. For example, can macro-level social phenomena
be explained by micro-level individual phenomena (Little 1991, 1998)?
 The Billiard Ball Model of IR 27

To illustrate, neorealist IR theorist Kenneth Waltz (1979) holds the position

that his structural macrotheory of international politics cannot be reduced to
microprocesses below the level of the state system, which has made him a con-
troversial figure in IR theory (Ruggie 1983; Ashley 1984; Elman 1996). Foreign
policy theorist Valerie Hudson instead contends that, “All that occurs between
nations and across nations is grounded in human decision makers acting singly or in
groups” (Hudson 2005: 1. Italics Hudson’s). She argues that individuals are the
ontological basis for social action; it is also not sufficient to explain social actions
without incorporating human agency. “Social science is unlike the physical sci-
ences in that what is analyzed possesses agency. Description of an act of agency,
or assertion that natural law was operative in a particular case of the use of agency,
cannot fully satisfy, for we know that agency means the agent could have acted
otherwise” (Hudson 2005: 4).
Having staked out this position, Hudson is willing to acknowledge that the
feasibility of meeting this explanatory goal is difficult and is also content to have
general IR theorists provide shorthand accounts of agency in their explanations
of social phenomena at higher levels of aggregation—so long as some foreign
policy theorists fill in the gap (Hudson 2005: 5). This microfoundations thesis
does not subscribe to the strict methodological individualist thesis, requiring that
the meaning of social concepts must be reducible to individual concepts and con-
versely, that social explanations must be derived from individual phenomena.

[A] microfoundational account of the determinants of individual action

should refer to social relations, structures, etc. The latter are grounded in
facts about individuals, but the thesis does not require that the explanation
supply the details of such a grounding. Thus, the microfoundations thesis is
not committed to anything like the individualist thesis about the meaning
of social concepts.
(Little 1991: 200–201)

An alternative position regarding this issue is the supervenience possibility, which

admits macro-level regularities must “supervene”—come from and rely ontologi-
cally for their existence—on micro-level phenomena; however, their existence
may be realized causally in multiple ways via different micro-level processes (Little
1991: 190–195; Wendt 1999: 155–157). Therefore, it is possible that macro-level
phenomena need not be reduced to micro-level conditions in a causal explanation
and, as Waltz contends above, can stand alone as macro-level explanations. The
argument here is that no useful explanatory value is gained by the effort to excavate
the basis on which these regularities exist in order to explain them and that some
macro-level phenomena are relatively autonomous and cannot be clearly reduced
to micro-level antecedents.
The supervenience possibility can lead to valid and powerful explanations
without microfoundtions in some domains of scientific inquiry; however, two
28 Stephen G. Walker

requirements for waiving the task of linking macro-level patterns to microfoun-

dations need to be met. The regularities must be robust patterns, and the nature
of the causal connection must also be clear between the elements in the macro-
level causal field. These requirements are met rarely in the social sciences where
macro-level explanations either are often relatively weak tendency statements
and/or rest on underlying mechanisms such as rationality in order to be cogent as
causal relationships (Little 1991: 195–199).
Since the core focus in this book is on foreign policy decisions during sym-
bolic and strategic interaction episodes in world politics, the rationality theorem
in some form is a requirement in order to form a cogent explanation for these
decisions. More generally,
Once we accept the point that macroexplanations require microfounda-
tions, we must next ask what types of individual-level processes we should
look for. And here there are two broad families of answers: rational choice
models and social-psychology models. The first approach attempts to
explain a given social process as the aggregate result of large numbers of
individuals pursuing individually rational strategies. The second approach
attempts to explain the social phenomenon as the complex outcome of a
variety of motives, rational and nonrational, that propel individual action.
(Little 1991: 198)
Waltz (1979) and other macro-level IR theorists in varying degrees have tended to
pursue the first approach, a rational choice path offered by the substantive rational-
ity theorem that emphasizes external constraints on foreign policy choices (Simon
1985; Elman and Elman 2003). We shall pursue the second approach, a social-psy-
chological path offered by the bounded rationality theorem in which individual
differences in the beliefs, emotions, and motives of individual leaders as well as
external constraints and the actions of others make a difference in their foreign
policy decisions (Walker 2002). Both versions of the rationality theorem have the
same logic of recognizing successive levels of rationality (purposefulness, mindful-
ness, and empathy); however, these focal points of rational choice vary between
macrofoundations in the external world of events in the case of substantive ration-
ality and microfoundations in the internal world of beliefs in the case of bounded
rationality.
The application of each version of the rationality theorem to the task of
explaining foreign policy as the exercise of power in strategic interaction episodes
is illustrated below by the discussion of the billiard ball model of IR in this chapter
and by the discussion of the behavioral model of IR in the next chapter of this
book. How exactly do observations taken at the microfoundations of politics in a
subfield such as Foreign Policy Analysis describe and explain important phenom-
ena at higher levels of aggregation in a field such as International Relations? This
question is the central challenge facing behavioral IR as an approach to foreign
policy analysis. It is also the focus of this volume as various contributors address
 The Billiard Ball Model of IR 29

different aspects of this question. This enterprise overlaps, but is not completely
identical, with the question of whether behavioral IR is a worthy endeavor. But
if taking a behavioral approach to the microfoundations of foreign policy analysis
can answer this question, then the results would support the case on behalf of
the bounded rationality theorem as the organizing principle for conducting for-
eign policy analysis and the case made by Mintz (2007) and Hudson (2005) for a
behavioral IR research program focusing on individuals and groups in the subfield
of Foreign Policy Analysis.

The Billiard Ball Model

The discussion of the billiard ball model in this chapter and the behavioral model of
foreign policy analysis in the next chapter follow closely the metaphorical inspira-
tion provided by the macro-micro distinction noted earlier in this chapter between
classical (Newton’s) and modern (Einstein’s) physics. Also influential is the visual
imagery derived from the dictionary meanings of “field” in Chapter 1, as they apply
to the study of politics and the subfield of Foreign Policy Analysis within the field
of International Relations. The metaphorical equivalent of “energy” in physics is
the exercise of “power” in politics between individuals, groups and larger entities,
and observed behaviorally as various forms of Cooperation (positive sanctions)
and Conflict (negative sanctions), respectively, which have positive (CO+) and
negative (CF–) valences. Individuals-as-actors are the smaller, less stable, “social
atoms” of politics while the larger, more stable “liquids and solids” of politics are
the larger, more stable organizations, such as political parties, other political institu-
tions, states-as-actors, and supranational or international organizations.
Politics as a social process requires at least two individuals-as-actors in order to
occur, i.e., for there to be interaction between them, so that a dyad and its com-
ponents constitute both the micro and macro levels of political analysis. If the dyad
extends to constitute even larger units, e.g., sets of individuals, groups, organiza-
tions, or states-as-actors, then it is possible in principle to decompose these larger
aggregations of individuals as a matrix of intersecting pairs of individuals to reach
a focus on a particular pair of individuals-as-actors. The forms of power (positive
or negative sanctions) exchanged between a pair of actors are represented by the
language of politics as Words (W) or Deeds (D), which are depicted in the spatial
field diagram in Figure 2.2 as representing the “field” of IR.
Macroscopic politics studies larger configurations and shifts in political
“energy” between states-as-actors, e.g., the shifts between mutual Conflict (–,–)
and mutual Cooperation (+,+) plus the transitions (–,+ and +,–) between these
two configurations. The basic energy states between pairs of actors in politics
in the field of International Relations (IR) are the symmetrical states of mutual
cooperation or mutual conflict, i.e., peace (+,+) or war (–,–) plus asymmetrical
transition states (+,– or –,+) between them; these valences represent the forms of
power exercised between two actors (Ego, Alter) as either positive (+) or negative
30 Stephen G. Walker

D+ W+ ← Alter → W– D–

D+

Peace (CO+, CO+) Transition (CO+, CF–)

W+
↑
Ego
↓
W–
Transition (CF–, CO+) War (CF–, CF–)

D–

CO = Cooperation (+) and CF = Conflict (–); W = Words and D = Deeds

FIGURE 2.2 A Static Political Field Diagram of International Relations

(–) sanctions (Dahl 1957; Baldwin 1989). Foreign policy analysis is the study of
these power processes from the perspective of at least one of the actors in a dyad
(Richardson 1987).
The vertical axis in Figure 2.2 represents the continuum of power exer-
cised by one actor (Ego) while the horizontal axis represents the continuum of
power exercised by the other actor (Alter) in a political dyad. Their joint
locations in this field are defined by the exercise of power as political “energy”
toward one another, identified by plotting their respective Cooperation (CO+)
or Conflict (CF–) behavior toward one another into one of the four quadrants in
the field diagram generated from an origin of mutual non-action at the intersec-
tion of each axis. This behavior is measured by their Words or Deeds as either
positive (+) or negative (–) valences, which are attributed to each actor in texts,
i.e., either from the statements by each actor that represent the “world in their
minds” or the second-hand accounts by observers that represent the “world of
events.”
Microscopic politics studies smaller configurations and shifts in political energy,
instances of escalation (e–) or de-escalation (d+), which are represented as smaller,
less stable, second-order derivatives in Figure 2.3. They may take different forms,
depending on the direction of the shift from the Status Quo along the axes of a
political field anchored by the same forms of positive sanctions (Cooperation) and
negative sanctions (Conflict) depicted earlier as Words and Deeds earlier in Figure
2.2. In this more dynamic version of the political field, the shifts are within quad-
rants as well as across quadrants. However, the logic is the same, as it is possible to
 The Billiard Ball Model of IR 31

D+ W+ ← Alter → W– D–

D+ (d+,d+) (d+,e–)

Status
W+ Quo

(e–,d+) (e–,e–)
↑
Ego

W–

D–

e– = Escalation and d+ = De-escalation

FIGURE 2.3 A Dynamic Political Field of Escalation (–) and De-escalation (+)

aggregate escalatory and de-escalatory shifts in ascending spatial order across the
valenced Word and Deed categories within or across quadrants.
For example, a de-escalatory shift upward within the upper-left quadrant by
Ego and a de-escalatory shift leftward within the same quadrant by Alter are rep-
resented by (d+, d+) while symmetrical escalatory shifts downward and to the
right, respectively, are represented by (e–, e–). Patterns of asymmetrical shifts (one
member of the dyad shifts in an escalatory (e–) direction while the other shifts in
a de-escalatory (d+) direction) are represented as (d+, e–) or (e–, d+). Extending
the degree of escalation or de-escalation by at least one actor (Ego or Alter) in
the upper-left quadrant can also change their respective locations to one of the
other quadrants.
The two basic analytical tasks in politics, including the FPA subfield within
the political field of IR, are (1) to describe the different basic energy states between
political actors and (2) to explain how and why they take on one of these basic
forms and shift between them. To paraphrase and extend Morgenthau’s (1978)
conceptualization of politics: the study of international politics, like all politics, is the
analysis of the struggle for power as the manifestation of the basic energy states of politics.
In the parlance of the philosophy of science, the explananda (the phenomena to
be explained) in politics are the various forms of the struggle for power between
political actors (Walker 1987a: 72).
The basic tasks of Foreign Policy Analysis as a subfield are to account for how
and why members of a dyad locate and re-locate themselves within the political
32 Stephen G. Walker

field of International Relations. A billiard ball approach focuses on the macro-

foundations of this puzzle while a behavioral approach focuses on the puzzle’s
microfoundations, i.e., the analysis of the smaller patterns of foreign policy escala-
tion and de-escalation, in order to account for the larger patterns of international
peace, war, domination, or submission. The microscopic relations generated by
individuals as social atoms plus the relations among their beliefs, emotions, and
motivations as subatomic parts constitute microfoundations of the more dynamic
patterns of escalation (e–) and de-escalation (d+) in Figure 2.3 and the more stable
patterns of Conflict (CF–) and Cooperation (CO+) in Figure 2.2.
What causes or conditions can a microfoundations approach specify in order
to construct what philosophers of science call explanans or covering-law generali-
zations, rules which can account for variations in the patterns of peace and war?
With the exception of a simple stimulus-response model of behavior, the cover-
ing laws of human decision making generally assume some form of rational choice
by the individual. That is, the decision is at least purposeful (goal-oriented). At
the same time the assumption of a focus on power politics is that a decision also
either elicits or is a response to the exercise of power by other(s). So a satisfactory
explanation of behavior needs to incorporate both intrinsic and external focal
points of rationality as causes in stating the proposition that explains the behavior
(Chwe 2001). In effect, the behavior is either the product or the precursor of a
collision and a decision.
Generally, the more complex the external stimuli and the behavioral response,
the more likely that internal decision-making processes play an increasingly
important role in explaining the behavior. For example, everyone knows that
it is more difficult to drive a car in heavy traffic than in light traffic. The stimuli
from other cars and other driving conditions make driving the car more difficult,
prompting more calculations by the driver, although each decision by the driver
can be isolated and identified or aggregated into larger sequences of behavior and
treated as an integrated whole. That is, “driving the car” can be conceptualized
alternatively as reduced to a series of small “movements of the steering wheel” or
aggregated to the larger construct of “keeping the car on the road” (see Hewitt
1988).
At a still higher level of aggregation and with more information about the
driver’s internal focal point or goal, “driving the car” can be conceptualized in
its temporal entirety as “going to grandma’s house.” Extending the definition of
simple rationality further from purposeful (goal-oriented) behavior to include
mindful (means-oriented) comparisons of alternatives, it is possible to explain the
driver’s choice among alternative available routes to grandma’s house as an opti-
mum one. This model of instrumental rationality can be extended even more to
become a strategic model of rationality that is empathetic (other-oriented), so that
the optimum (quickest) route to grandma’s house is not necessarily the shortest
distance. It is instead the route with the greatest speed limit enforced by the police
and the lightest traffic as determined by the choices of other drivers. The levels
 The Billiard Ball Model of IR 33

of rationality in this example extend from simple choices of discrete behaviors

through choosing the most efficient tactical sequences of behaviors to strategic
calculations about behaviors leading to optimal outcomes based on the contingent
behaviors of others in the decision-making environment.
Moving from peace to war can be at least as complex as getting to grandma’s
house. Leaders of states are likely to be empathetic regarding the choices of Alter
to exercise power, mindful of the alternative sequences of words and deeds in
Ego’s exercise of power leading to war or peace, and purposeful about Ego’s and
Alter’s goals as desired outcomes of escalatory or de-escalatory decisions lead-
ing to these outcomes. Ideally, therefore, a good explanation of leaders’ foreign
policy choices should include as causes the internal focal points that constitute the
world in their minds and relate them in the form of an explanans or covering law
to the world of events as the explanandum or phenomenon to be explained.
The macro world of actions by states and the micro world in the minds of
leaders are represented and connected in Figure 2.4 via the language world of
texts (Habermas 1979, 1984, cited in White 1987). Habermas considers the lan-
guage world of texts to be relatively autonomous from the other two worlds
with each world having its own objective, subjective, or communicative norm
of rationality, respectively, and the world of language coordinating reflexively
relations between the other worlds (White 1987: 114–127; see also Rapoport
1960). This function of language is entailed in the symbolic exercise of power in
policy debates and diplomatic negotiations versus the physical use of power in the

(3) Action Network

Alter
+ –
+ +,+ +,–
Ego
– –,+ –,–

(2) Language Network

+ –
E a b

A c d
(1) Neural Network Trilogy

Cognition

E, A Motivation

Emotion

E = Ego and A = Alter

FIGURE 2.4 Connections between the World in Their Minds and the World of
Events
34 Stephen G. Walker

exchange of rewards and punishments, which is represented later in Chapter 3

by the distinction between Word games versus Deed games (see also Jervis 1978;
Franck and Weisband 1971; Beer and De Landtsheer 2004).
The function of the world of language in Figure 2.4 is merely to represent the
other two worlds of events and beliefs and provide a common metric for observ-
ing the use of power in them. Each of these worlds is constituted by a network of
relationships between Ego and Alter. The world of events consists of the networks
of actions attributed to Ego and Alter, represented by the verbs in the accounts
(event histories) compiled by chroniclers such as journalists and historians. The
mental world of leaders consists of the neural networks of emotion, motivation,
and cognition with “subatomic” attributions to Ego (Self) and Alter (Other), rep-
resented primarily by the verbs in the statements (belief histories) of leaders.
The identities of Ego and Alter in both kinds of texts depend upon who is the
author of the text that is the source of the attributions. Both the patterns or regu-
larities that constitute possible “laws” or “rules” connecting the behavior of states
in the world of events with the thoughts of leaders in the world of beliefs can be
mathematically represented as positive (+) or negative (–) attributions attributed
to Ego or Alter as bits of information from texts in the world of language that mir-
rors the other two worlds in Figure 2.4 (Shannon and Weaver 1964; McClelland
1968, 1972; Adler 2002: 168–172).
Comparisons of the two kinds of texts (event histories and belief inventories)
can reveal connections and disconnections between the mental world of leaders
and the behavioral world of events, which together may describe and explain the
foreign policy decisions of each actor (Ego and Alter). Repeated observations of
these indices can make analyses dynamic representations of sequences of decisions,
behaviors, and outcomes manifested by the exercise of various forms of power
by members of an [Ego, Alter] dyad. It is possible as well to extend the analysis
to triads and, indirectly, to even larger configurations of actors. This extension
from dyads to triads generates what physicists call a more complex, “three-body”
problem, as opposed to the simpler, “two-body” problem represented by a dyad
(Goldstein et al. 2002).
In classical physics the two-body problem was solved mathematically with
relatively simple, linear equations, which yielded a deterministic theoretical solu-
tion that fit (within the limitations of measurement errors and local disturbances)
the empirical observations of such large bodies as two planets (Goldstein et al.
2002). When physicists extended the scope of their analysis to include three (or
more) objects as a larger, more complex system, however, their theoretical solu-
tions became non- linear and less deterministic. Statistical models replaced math-
ematical models, which are less deterministic because of the mutual interaction
possibilities among all three objects even if the interactions exist only between
pairs of objects (Barrow-Green 1997; Goldstein 2002).
The same complications appeared again as the focus of physics shifted down-
ward to subsystems within large objects. The large number of components
 The Billiard Ball Model of IR 35

(compounds, molecules, or atoms) that constituted the large objects resulted in

systems too complex in their interaction possibilities to model with deterministic
mathematical solutions. Not until the emergence of the field of quantum physics
would atomic particles appear initially to interact in a manner again consistent
with deterministic mathematical models. However, there is a remaining undeter-
mined component of quantum mechanics, the actual state of the particle at the
time of observation, as it is in constant motion and can only be approximated with
existing instrumentation and modeled spatially as a dynamic “state.” One of the
most famous examples of such indeterminacy is the phenomenon of duality in the
analysis of light as particles or waves; electrons, neutrons, and protons also exhibit
duality at subatomic levels of observation (Feynman 1995).
There appear to be some parallel problems in studying politics as the focus has
changed in the field of International Relations from larger, “two-body” problems
to n-body problems constituted by systems of states and their smaller subsystems
(Kaplan 1957; Deutsch and Singer 1964; Waltz 1979; Schweller 1998; Jervis and
Snyder 1991; Snyder and Jervis 1993; Jervis 1997). The disaggregation of states
into groups and individuals as actors and the aggregation from dyads to triads
are analytical choices that create “three-body” problems within the subfield of
Foreign Policy Analysis. For example, two-player dyadic games between indi-
viduals or states have deterministic solutions while n-player games have statistical
solutions at best or rely on updating information revealed as the players make
moves within the analytical framework of game theory (Brams 1994: 182–206).
These generic solutions of n-player game theory are the ones adopted by
physicists as well to deal with the indeterminacies of incomplete information
about the interaction possibilities among more than two bodies (Feynman 1995;
Goldstein et al. 2002). There are additional limits to their application in politics
that do not apply in physics. As Brams (1994: 213–214) points out, “[I]t would
take some effort to develop an efficient algorithm for making nonmyopic calcu-
lations in larger matrix games. But do people really think through the manifold
choices in larger games and impeccably choreograph their moves in them? Or,
plagued by incomplete information, are they more likely to follow simple rules
of thumb?”
The approach in Figure 2.5 to the “three-body” problem, therefore, takes a
middle stance by aggregating to triads that are interconnected, but it stops short
of considering directly more than three bodies at a time from the larger interac-
tion matrix. This approach is based on three assumptions: (1) it is possible to solve
the two-body (dyadic) problem in politics; (2) the three-body (triadic) problem
is both common and important in politics, e.g., a two-party dispute with a third
party as a mediator or a balancer; (3) larger versions of the three-body (nth-actor)
problem extending to four or more actors are likely to be solved in politics by
creating organizations with institutional rules that bind the collective interactions
of the members or by decomposing them into a series of dyads or triads linked by
one or more common members.
36 Stephen G. Walker

A
A

B C
A

B C D B C

Three-Body Problem Four-Body Problem

D

A > B = C > D > E = F >G

A – + – + – + E F
B + – + – + –
C – + – + – +
D + – + – + –
E – + – + – +
F + – + – + –
G – + – + – + G
Power Matrix Other Extensions

(>, =, <) Greater than, equal to, less than Power Relations.
(+) = Positive Sanctions; (–) = Negative Sanctions.

FIGURE 2.5 Extensions of Power Logic from Dyads to Triads

The actors (A through G) within the power matrix in Figure 2.5 are arrayed as
columns in descending order from most powerful to equal to least powerful and
connected as dyads by the cells in this matrix. Then they are arrayed and con-
nected three-at-a-time in triangles and positioned vertically in descending order
of power from A to G as larger configurations (three-body, four-body, and other
extensions) from the two-body cells in the power matrix. These combinations
are aggregations of two-body problems, which can be considered separately or in
conjunction with other two-body problems.
For example, the three-body ABC problem in Figure 2.5 is the aggregation
of three two-body problems: AB, BC, and AC where A, B, and C are political
actors. Information about the exercise of positive (+) and negative (–) sanctions
and the complementary distribution-of-power relationships (greater than, equal
to, less than) for each member of these two-body problems appear in Figure 2.5’s
power matrix as follows:
A|B = (–, > | +, <), i.e., (A’s exercise of power toward B is negative, and
A is more powerful than B; B’s exercise of power toward A is positive, and
B is less powerful than A).
B|C = (–, = | +, =), i.e., B’s exercise of power toward C is negative, and B
is equal in power to C; C’s exercise of power toward B is positive, and C is
equal in power to B.
 The Billiard Ball Model of IR 37

A|C = (+, > | –, <), i.e., A’s exercise of power toward C is positive, and A
is more powerful than C; C’s exercise of power toward A is negative, and C
is less powerful than A. According to the rule of transitivity, the distribution
of power between A and C is derived as follows: if A > B and B = C, then
A > C and C < A.
These statements simply specify how the members of each dyad are “coupled,”
or joined logically by formal relationships between the properties identified in the
power matrix. A foreign policy theory, a set of propositions based on these proper-
ties of power, is necessary in order to infer how they matter systematically in the
explanation of decisions to exercise power and generate outcomes at higher levels
of aggregation (Phillips 1974; Blalock 1969).

The Macrofoundations of Foreign Policy Decisions

In the remainder of this chapter the primary focus is a theory of the macrofounda-
tions of foreign policy decisions with states-as-actors. Constructing a macroscopic,
billiard ball model of IR as a realist theory regarding the exercise of power per-
forms the two tasks of describing and explaining collisions between pairs of states
and the symmetrical outcomes of peace (CO+, CO+) or war (CF–, CF–) plus the
asymmetrical states of domination or submission (CO+, CF–) and (CF–, CO+)
in Figure 2.2. It addresses the two-body problem of strategic dyads by specifying
the initial conditions regarding the use of positive or negative sanctions and the
distribution of power between any two actors (A|B, B|C, or C|A in our triadic
example) and conceptualizes the outcomes of interaction between them as the
intersection of their choices of positive and negative sanctions defining one of
the following possible outcomes: (1) settlement or mutual cooperation (+,+), (2)
deadlock or mutual conflict (–,–), and either (3) domination by the first over the
second (–,+) or (4) submission of the first to the second, (+,–), or vice versa.
The following six power politics propositions in this realist theory couple the
possession and exercise of power between two actors via a strategic rationality
theorem to specify the conditions leading to the four possible outcomes of the
“struggle for power” (Morgenthau 1978) in which the two actors exercise power
in opposition (–,–) or in concert (+,+) vis-à-vis one another, or one actor domi-
nates, i.e., seeks power over the other (–,+) or submits to the power of the other
(+,–). This rational choice theorem expands slightly the purposeful assumption
that political actors simply seek power in order to survive, to include as well the
mindful assumption that political actors explicitly choose between the alternatives
of positive and negative sanctions in the exercise of power. It thereby formalizes a
relatively “thin” model of rationality as the explanation for foreign policy choices,
as it specifies only a single constraint and a single incentive.
Depending on the asymmetrical or symmetrical distribution of power (<,
=, or >) between them, each actor in the dyad seeks power within the con-
straint specified by the distribution of power and the incentive specified by the
38 Stephen G. Walker

distribution of national interests (vital/secondary) between the actors (Wolfers

1962; Hartmann 1983; see also Walt 1987). The addition of a national interests
condition into the equations formed by the six power politics propositions below
provides the incentives to choose positive versus negative sanctions in the exercise
of power. The national interests at stake for each actor determine whether each
one acts according to one of the first three or one of the last three propositions.
If an actor’s interests are considered secondary (not worth fighting for), the first
three propositions apply; if the interests are vital (worth fighting for), the last three
propositions apply (Hartmann 1983).
As both members of a dyad follow the rationality theorem specified by the dis-
tributions of power and interests in the exercise of positive or negative sanctions,
they generate the systemic (dyadic) outcomes of settlement, deadlock, domina-
tion, or submission for each actor and the dyad. Applied to an actor’s respective
choices of positive or negative sanctions in the exercise of power and the genera-
tion of the outcomes of settlement, deadlock, domination, or submission, the six
propositions in the theory couple these two properties logically as follows for any
Actor (A) toward any Actor (B).

I. Power Politics Propositions re: Secondary Interests

Proposition 1. If Actor A’s exercise and distribution of power regarding Actor

B is (+, <), then Actor A prefers settlement > deadlock > submission >
domination as outcomes of their interaction.
Proposition 2. If Actor A’s exercise and distribution of power regarding Actor
B is (+, =), then Actor A prefers settlement > deadlock > domination >
submission as outcomes of their interaction.
Proposition 3. If Actor A’s exercise and distribution of power regarding
another Actor B is (+, >),then Actor A prefers settlement > domination >
deadlock > submission as outcomes of their interaction.

II. Power Politics Propositions re: Vital National Interests

Proposition 4. If Actor A’s exercise and distribution of power regarding

another Actor B is (–, <), then Actor A prefers domination > settlement >
submission > deadlock as outcomes of their interaction.
Proposition 5. If Actor A’s exercise and distribution of power regarding
another Actor B is (–, =), then Actor A prefers domination > settlement >
deadlock > submission as outcomes of their interaction.
Proposition 6. If Actor A’s exercise and distribution of power regarding
another Actor B is (–, >), then Actor A prefers domination > deadlock >
settlement > submission.

Realist scholars determine vital and secondary interests in different ways. Wolfers
(1962) distinguishes between possession (vital) and milieu (secondary) goals or
 The Billiard Ball Model of IR 39

interests. Others use the alliance portfolios of actors to determine vital interests,
as alliances specify commitments to fight for certain goals, and secondary interests
become the remaining goals of foreign policy decisions (Bueno de Mesquita 1981;
Bueno de Mesquita and Lalman 1992). Some distinguish between revisionist and
status quo goals in an attempt to detect which goals are more likely to be incentives
to use negative sanctions rather than positive sanctions (Schweller 1998; Maoz and
Mor 2002).
To sum up, the four theoretical outcomes from the exercise of power in this
realist billiard ball model of IR are (+,+), (–,–), (+,–), (–,+), which specify solu-
tions to the two-body problem by explaining the possible relationships between
states-as-actors A and B. The propositions in the theory explain these outcomes
by reference to the rationality theorem as specified by the macroscopic distribu-
tions of power and interests between the two actors. The propositions in this
theory specify as well the intersection of the ranked preferences of Ego (A) and
Alter (B) regarding their respective locations in the field of power politics.
Adding this information about preferences in Figure 2.6 transforms the field
diagram into a “stack” of game theory matrices with the four quadrants becoming
the four cells of a 2 × 2 game in which each player has two choices: Cooperate
(CO) or Conflict (CF). It permits inferences about: (a) the logical decisions for the
two players to make in the context of this information about preferences; (b) the
logical outcomes that are equilibrium solutions provided by game theory to this

D+ W+ Alter W– D–

D+

Peace (CO+, CO+) Transition (CO+, CF–)

A|B = (2,4) A|B = (1,1)
B|C = (3,4) B|C = (1,2)
W+ A|C = (4,3) A|C = (1,4)

Ego

W– Transition (CF–, CO+) War (CF–, CF–)

A|B = (4,2) A|B = (3,3)
B|C = (4,1) B|C = (2,3)
A|C = (3,2) A|C = (2,1)

D–

CO = Cooperation (+) and CF = Conflict (–); W = Words and D = Deeds.

Outcomes are ranked from 4 (highest) to 1 (lowest).

FIGURE 2.6 Mapping Two-Player Games in a Political Field Diagram

40 Stephen G. Walker

two-body problem. Each of the game matrices now specify the bilateral relation-
ships in the ABC triad from Figure 2.5 as a game of strategic interaction.
With this information analysis of the ABC triad can proceed as a three-body
problem, e.g., A|B and A|C linked by A, or the various dyads within this triad
can be disaggregated into pairs of two-body problems or analyzed singly as three,
separate, two-body problems {A|B, B|C, A|C}. Further extensions of these
analytical possibilities include expanding triadic analysis from one triad to other
triads sharing a common member and even a common dyad. For example, the
ABC triad in Figure 2.5 is linked to both the ABD and ACD triads by a common
member (A), and the ABC triad is linked as well to the ABD triad by a com-
mon dyad (A|B) and to the ACD triad by a common dyad (A|C). The game
matrices for these dyads in the ABC triad and their respective solutions appear in
Figure 2.7.
It is useful initially to consider each of these matrices separately as an instance
of a two-body problem, i.e., two actors A|B, B|C, and A|C, which together
also constitute a three-body problem. Each of these games has several properties,
including solutions to their respective two-body problems. The actors are players
who have two choices: Cooperate (CO) or Conflict (CF) with one another. The
intersections of these choices form the cells in the game matrix and specify the
possible outcomes of the game as mutual cooperation (CO, CO), mutual conflict
(CF, CF), domination and submission by each player (CF, CO and CO, CF).
Entered into these cells are the preferences of each player for these four out-
comes, ranked by convention from (4) highest to (1) lowest, with the row player’s
ranking entered to the left and the column player’s ranking entered to the right
of the comma in each cell. The solutions for each game are indicated as possible
outcomes to the game, given the information about preference rankings in the
cells and different assumptions about the level of rationality for each player as
purposeful, mindful, or empathetic.
If each player is merely purposeful (goal-oriented), then a player will select
CO or CF, depending on which row or column contains his/her highest-ranked

Player B Player C Player C

CO CF CO CF CO CF

CO 2,4 1,1 CO 3,4** 1,2 CO 4,3** 1,4

Player A Player B Player A

CF 4,2 3,3* CF 4,1 2,3* CF 3,2 2,1

*Nash equilibria are asterisked; **Pareto-superior outcomes are double-asterisked. Brams

nonmyopic equlibria are in bold. Minimax solutions and Dominant strategies are underlined.

FIGURE 2.7 Two-Player Games for the ABC Triad

 The Billiard Ball Model of IR 41

preference. The outcomes of the games in Figure 2.7 under the assumption of
purposeful choices by each player are (4,2) for the A|B game (4,1) for the B|C
game and (1,4) for the A|C game. The row player gets its highest (4) prefer-
ence in the first two games while the column player gets its highest preference in
the third game. These “winners” fare better than their counterpart “losers” in
these games who get either their lowest (1) or next-lowest (2) preference as an
outcome.
In order for a player to be mindful (means-oriented) as well as purposeful, s/he
must rank all of the outcomes as well as identify the highest-ranked outcome.
Both the winners and the losers in these games are likely to learn from this exer-
cise that the outcomes of future games depend not only on their goals expressed as
ranked preferences but also on the other player’s preference rankings for the four
outcomes. This realization seen intuitively by smart players or learned previously
by experienced players is likely to make them more sophisticated in their levels
of rationality even in an initial play of a game. That is, assuming that both players
initially know only their own rankings and not the rankings of the other player,
they can exhibit a mindful (means-oriented) level of rationality. The maximin
(maximize gains/minimize losses) solutions underlined in the cells of each game
are based on each player’s knowledge of his/her own ranked preferences and
calculating a choice between CO or CF that avoids the player’s lowest-ranked
(1) outcome no matter what the other player’s choice and ranked preferences are.
These outcomes are (4,2) for A|B, (2,3) for B|C, and (3,2) for A|C.
In some cases a player may have a dominant strategy, which introduces a fur-
ther “mindfulness” calculation regarding the choices between Cooperation (CO)
and Conflict (CF). Knowing only his own preference rankings, s/he can adopt
a dominant strategy of “always choose CO” or “always choose CF.” In Games
A|B and B|C, the row players have a dominant strategy of “always choose CF.”
This strategy is based on a comparison of the answers to two alternative “what if”
statements: What does the row player choose if the column player chooses CO?
What does the row player choose if the column player chooses CF?
If the answer is the same to both questions, then the row player has a “domi-
nant” strategy. If the answer differs, then the row player has a “contingent” strat-
egy. In the two A|B and B|C games in Figure 2.7, the row player’s answer is the
same for both questions: “choose CF.” However, the answers are different for the
column players in these two games, and the answers are also different for both the
row player and the column player in Game C|A. This pattern of responses means
that there is a dominant strategy for Player A in Game A|C and for Player B in
Game B|C; the other players in the three games have contingent strategies.
Acting mindfully, therefore, the two row players should choose CF as their
dominant strategy in Games A|B and B|C. The calculation of the answer to the
two questions in each case is as follows. If Player B in Game A|B chooses CO,
then Player A should choose CF because the outcome for A will be (4) instead of
(2); if Player B in Game A|B chooses CF, then Player A should also choose CF
42 Stephen G. Walker

because the outcome for A will be (3) instead of (1). If Player C in Game B|C
chooses CO, then Player B should choose CF because the outcome for B will be
(4) instead of (3); if Player C in Game B|C chooses CF, then Player B should also
choose CF because the outcome for B will be (3) instead of (1). The mindfulness
of the remaining players with contingent strategies is limited to minimax strate-
gies, because their choices between CO and CF would otherwise depend on the
preference rankings and choices of the other players in their respective games.
Finally, in order for a player to be empathetic (other-oriented) as well as mind-
ful (means-oriented), the “two-sided information” condition is desirable. Ideally,
both players should know each other’s rankings as well as their own rankings of
the different outcomes. If this condition exists, then it is possible for both players
to exhibit the highest level of rationality and choose between CO and CF so that
one of the equilibrium outcomes in Figure 2.7—Nash, Pareto, Brams—is the
outcome for each game. However, all three outcomes are not necessarily separate
properties of each game. Game A|B only has a Nash and a Brams equilibrium.
Game B|C has all three equilibria while Game C|A lacks a Nash equilibrium.
The Pareto and Brams equilibria overlap in Games B|C and A|C.
The likelihood of each one being the actual equilibrium depends on assump-
tions about the rules of play for each game and the formal definitions of these
three equilibria. If the rules of play assume under the two-sided information
condition that each player must select his/her choice simultaneously in a single
play of the game, then the logical equilibrium is the Nash equilibrium. A Nash
equilibrium is an outcome from which neither player has an incentive to defect,
because the immediate (myopic) consequence of defection is a worse outcome for
the defector. In Game A|B the (3,3) outcome is a Nash equilibrium: if A defects
from CF to CO, the consequence is a lower-ranked (1) outcome; if B defects, the
consequence is a lower-ranked (2) outcome. The same logic applies to the (2,3)
outcome in Game B|C: if A defects, the outcome is (1) for A and if B defects, the
outcome is (1) for B. In Game A|C there is no Nash equilibrium because from
any of the cells at least one player can improve his/her outcome by defecting.
In Game B|C and in Game A|C, mutual cooperation (CO,CO) is a Pareto-
optimum equilibrium, defined as an outcome when both players get a higher-
ranked preference from this outcome than from the Nash equilibrium for each
game. This Pareto-superior outcome is less stable than a Nash equilibrium,
because at least one player can still gain a higher-ranked preference by defection
from this cell. The same instability is true of the Brams nonmyopic equilibrium
for each game: the column player can gain by defecting from (4,2) to (3,3) in
Game A|B; the row player can gain by defecting from (3,4) to (4,1) in Game
B|C; the column player can gain by defecting from (4,3) to (1,4) in Game A|C.
The Brams nonmyopic equilibria for these games are stable only when the rules
of the game allow for alternating moves by each player and repeated plays of the
game, so that the victim can punish the defector by moving to a Nash equilibrium
or deter defection by threatening to do so. In order for the threat to be credible,
 The Billiard Ball Model of IR 43

alternating moves and repeated play must be rules of the game so that the threat
can be carried out if deterrence fails (Brams 1994).
Collectively, the three games in Figure 2.7 formally model solutions to three,
separate, two-body problems. The game matrices and logical rules of inference
based on information in the cells about the preference rankings for each player
lead to these solutions. The solutions vary, depending on different assumptions
about the rules of play for the game and increasingly thicker notions of rationality
regarding purposefulness, mindfulness, and empathy along with the two-sided
information assumption. Ordinal game theory is the mathematical theory that
models these deductions from realist theory. The six realist propositions ranking
preferences for the outcomes of mutual-cooperation (settlement), mutual-conflict
(deadlock), conflict-cooperation (domination), and cooperation-conflict (sub-
mission) specify applications of the rationality theorem, based on the relationships
between the distributions of power and interests in a billiard ball model of IR.

Conclusion
The analysis in this chapter has demonstrated that the substantive rationality ver-
sion of the rationality theorem is sufficient to specify formally a billiard ball model
of IR and theoretical solutions to the separate two-body problems represented by
the A|B, B|C, and A|C dyads, in which A, B, and C are states-as-actors. These
three states are also linked together as a three-body problem, as each state has a
dyadic relationship with the other two states. It is then possible to ask whether the
power distribution among them is sufficiently malleable so that B and C can pool
their power and balance against A or whether C’s mutual cooperation (CO, CO)
equilibria with both A and B position C to mediate the mutual conflict dynamics
between A and B. We shall explore these kinds of possibilities as we move forward
with our analysis of strategic interactions between dyads and among triads in this
book.
However, we shall first address the microfoundations of foreign policy analysis
and the application of the bounded rationality theorem to patterns of escalation
and de-escalation in foreign policy decisions over time. In this task we shall con-
tinue to use the analytical tools introduced in this chapter, the six propositions
from realist IR theory, and rules of play and assumptions from game theory,
which together specify a billiard ball model of foreign policy decisions. This per-
spective is a relatively static, macroscopic view of foreign policy decisions, mod-
eled as large-scale shifts within a political field characterized by locations in the
quadrants of war, peace, domination, and submission. It will be re-engineered in
the next chapter as a more dynamic, microscopic analysis of small-scale move-
ments within each quadrant of the political field, specified by a behavioral model
of foreign policy decisions.
3
MICROPOLITICS AND FOREIGN
POLICY DECISIONS
The Behavioral Model of IR

Stephen G. Walker

Introduction
In this chapter there is a shift from states to individuals as actors and a focus
on their belief systems, emotions, motivations, and risk orientations. This focus
involves a theoretical move from the substantive rationality theorem to the
bounded rationality theorem in the specification of the realist and game theory
models introduced in Chapter 2, which analyze the exercise of power by leaders
in making foreign policy decisions. The analytical approach in the present chap-
ter of describing and explaining foreign policy decisions is to model the internal
“world in their minds” in order to understand the “external world of events.”
The empirical approach to measuring foreign policy decisions and their origins
in the minds of decision makers is the microscopic content analysis of the event
histories of observers and the belief inventories of leaders.
The explananda (puzzles to be explained) under investigation are foreign policy
decisions reported in behavioral chronologies extracted from newspapers and
other sources (event histories) as variations or changes in big behavioral con-
figurations, identified previously in Chapter 2 as either Peace or War and scaled
as the intersection of Cooperation (CO+) and Conflict (CF–) behavior. These
phenomena are disaggregated now spatially and temporally in this chapter via
content analysis into smaller variations and shifts, described as de-escalation (d+)
and escalation (e–), and scaled as bits of information in the form of words and
deeds along a continuum of Cooperation and Conflict.
The underlying rules for these patterns are extracted from statements by lead-
ers (belief inventories), which contain attributions to self and others as anteced-
ent conditions and indicate the influence of a leader’s cognitive, motivational,
and emotional systems on these patterns of foreign policy behavior. The power
 The Behavioral Model of IR 45

politics propositions and the bounded rationality theorem couple these antecedent
conditions (belief systems) and behavioral phenomena (behavioral events) into
explanans—law-like generalizations or rules—about beliefs and intentions consis-
tent with the realist theory of power and interests in Chapter 2, thereby providing
theoretical unity and continuity between descriptions and explanations of foreign
policy decisions at the macro and micro levels analysis (Feynman 1995; Lake and
Powell 1999a).

Simple Models of Rationality

In this chapter the main argument is that politics as human behavior is generated
from relatively few elements of Words and Deeds, conceptualized theoretically
as social acts of communication in the form of the exercise of power between
individuals in a social setting. A fictional example of a First Encounter situation,
“a world without shared ideas” (Wendt 1994: 328–336), illustrates the relatively
few elements in this single set of phenomena. In Daniel Defoe’s novel, Robinson
Crusoe, the protagonist is shipwrecked and marooned on a desert island. Crusoe
believes that he is alone (except for occasional visits by cannibals from another
island) on his island until he liberates a prisoner of the cannibals, whom he names
“Friday” after the day on which they meet. Friday is a black man whom Crusoe
assigns a role of “manservant,” or “assistant,” as they organize their life together
on the island until a ship rescues them (Shinagel 1994).
The few important elements of politics emerge on the island in Defoe’s story
when Crusoe rescues Friday and their interactions result in different roles for them
in their social setting. Crusoe exercises control over Friday both through Deeds
and Words, respectively, by displaying a command over superior Western technol-
ogy and converting Friday to Christianity before they are rescued. Some cultural
interpretations of this novel label it a metaphor for the power dynamics of British
imperialism as practiced in the seventeenth century (Shinagel 1994). The political
questions that this tale both prompts and answers are, “What is political power and
how is it exercised?”
The simplest formulation of these questions is perhaps the one associated with
Lenin’s Question: “Kto-kovo?” or “Who (controls) whom?” (Leites 1953: 27–
29). The answer to Lenin’s Question in the language text of Robinson Crusoe is that
“Crusoe (controls) Friday” by exercising power in the form of Words (invoking
Christian norms) and Deeds (displaying Western technology). The results conform
to the general conceptualization of power as a control relationship, in which A
(Crusoe) gets B (Friday) to do something that the latter would otherwise not do
and thereby constructs a definition of the political situation (Dahl 1957; Wendt
1994: 329–332).
Control relationships can take several forms, which form a continuum of posi-
tive and negative sanctions (Baldwin 1989). The continuum resembles meta-
phorically what physicists who study microscopic physical behavior would call
46 Stephen G. Walker

a “hierarchical energy landscape,” in which certain changes in behavior occur

in concave “valleys” with hierarchical boundaries (barriers) arrayed along a con-
tinuum of change, making it more difficult to move from one behavioral state to
another when the beginning and ending locations are separated by these bound-
aries. Within these valleys behavioral change is more incremental and stable,
but it is possible for behaviors to transcend their boundaries. The former pat-
terns are labeled Alpha changes and the latter patterns are labeled Beta changes
(Frauenfelder et al. 2009).
How does this concept of “hierarchical energy landscape” apply metaphorically
to identify the various locations along a continuum of social power (control rela-
tionships) as a “hierarchical power landscape?” It is possible to depict this model
by adapting the political field diagram of large-scale changes between Peace and
War from Chapter 2, as shown in Figure 3.1. In this more elaborate figure, various
power games are located in the valleys defined by the barriers represented by the
different boxes drawn within and across the quadrants. Ego and Alter play these
games as either Word (Agreement or Disagreement) games or Deed (Engagement
or Estrangement) games, depending on whether the exercise of power by each
player takes the form of words or deeds. It is also possible to consider the entire
field (all four quadrants) as one Grand Power Game, in which the players (Ego
and Alter) choose both Words and Deeds in the exercise of power in the form of
Barrier Barrier
D+ W+ Alter W– D–

D+ Symmetric Asymmetric D+
Engagement Transition
(+,+) (+,–)

Barrier PEACE TRANSITION Barrier

(CO+, CO+) (CF–, CO+)

W+ W+

Agreement Transition
(+,+) (–,+)
Ego Debate
(+,–) (–,–)
Transition Disagreement

W– W–

Barrier TRANSITION WAR Barrier

(CO+, CF–) (CF–, CF–)
(–,+) (–,–)
Asymmetric Symmetric
D– Transition Estrangement D–

D+ W+ W– D–
Barrier Barrier

CO = Cooperation (+) and CF = Conflict (–); W = Words and D = Deeds

FIGURE 3.1 A Hierarchical Power Landscape of the Grand Power Game in the IR Field
 The Behavioral Model of IR 47

positive (CO+) and negative (CF–) sanctions. All of these possibilities are consid-
ered as possible explananda (phenomena to be explained) below.
A brief tour of Figure 3.1 at this point will suffice to highlight the impor-
tant locations and relationships among them in this hierarchical power landscape.
Plotting the joint location for Ego and Alter requires answering the “Who (con-
trols) whom?” question posed by Lenin regarding the form of power exercised by
each actor. If they exercise positive sanctions as Deeds toward one another, then
they are exchanging Rewards and located in the Engagement valley of mutual
Cooperation (+, +) in the upper-left corner of the field diagram. If they are exer-
cising negative sanctions as Deeds toward one another, then they are exchanging
Punishments and are located in the Estrangement valley of mutual Conflict (–,
–) in the lower-right corner of the field diagram. If one of the actors in a Deeds
game is exercising power in the form of Rewards and the other actor is exercising
power in the form of Punishments, then they are located in one of the asymmetric
Transition valleys in the other two corners of the field diagram. In each case the
exercise of power takes the form of Deeds rather than Words.
If the actors are both exchanging Words, then they are located in one of the
cells of the Debate valley around the origin of the field diagram. In this valley the
exercise of power as Words takes the form of exchanging Promises, Declarations
of Appeals and Support, Declarations of Opposition and Resistance, and Threats.
This Debate valley is also a location of diplomatic transition through which a
movement from one of the four corners of the diagram to another corner may
occur. Movement from one corner to another may also occur directly without
Debate, or movement may originate initially from the Debate location as two
actors (such as Crusoe and Friday) establish contact initially with one another.
Both “movement” (change) and “location” (continuity) are defined in terms of
the exchange of positive and negative sanctions along a power dynamics scale
on the perimeters of the field diagram: D(+3 Reward), W(+2 Promise), W(+1
Appeal/ Support), W(–1 Oppose/Resist), W(–2 Threaten), D(–3 Punish). The
coordinates and connecting arrows between these locations collectively “map”
the parameters of the hierarchical power landscape in the International Relations
field and thereby offer alternative schematic accounts of the origins and evolution
of world politics conceptualized as the interplay of power relationships between
and among actors.
For instance, “first contact” does not always begin with Word games (debates)
between actors and instead promptly erupts in War. As Anatol Rapoport (1960)
illustrates in Fights, Games, and Debates by reference to an incident in Mark
Twain’s Tom Sawyer, the youthful, working class hero of Twain’s story comes
upon a rich dandy about his own age on the street and, feeling an instant mutual
dislike, they begin to fight until the defeated boy retreats. He later concludes that,
“In principle, if we knew how these reactions occur, we could predict the whole
progress of the fight from the way it started without reference to the goals of the
opponents or to the potentialities of the outcomes” (Rapoport 1960: 2).
48 Stephen G. Walker

In this fight model of human conflict, therefore, the management of conflict

is the anticipation and reaction by each actor to the actions (deeds) of the other,
and conflict resolution (movement from war to peace) is the termination of the
conflict through the transition state of surrender by one of them (Walker 2004b:
279). This kind of episodic and intermittent pattern of Estrangement is also com-
mon between societies of animals without language and between human societies
who have a low power differential and do not have a common language (Polk
1997; Diamond 1992).
The counterparts to these Estrangement patterns are also found in similar kinds
of Engagement patterns reported in anthropology, literature, and history. For
example, “love (lust?) at first sight” between wild animals, fictional characters in
novels, or teenagers in real life may elicit gestures of interest without words, such
as eye contact, facial expressions, and other forms of flirtation. If these initiatives
by one actor toward another actor are reciprocated in kind, then courting rituals
(such as dancing, holding hands, or promenading) may ensue and mating rituals
may follow. In historical accounts of early tribes who come into contact with one
another, exchange relations of mutual cooperation may begin by one tribe leav-
ing items for trade at a conspicuous site to be found and reciprocated by another
tribe, leading eventually to barter via hand signals and eventually learning one
another’s language. In some cases, one tribe’s captives in previous conflicts with
another tribe may learn the languages of both tribes, leading to the establishment
of diplomatic relations and Word games (diplomacy) that establish and keep the
peace between them (Polk 1997).

Exchange Theory and Role Theory

These primitive forms of social exchange relations between individuals and peo-
ples illustrate why the symmetrical relationships of Peace (+,+) and War (–,–) are
characterized ceteris paribus as more stable than the asymmetrical relationships of
Transition (–,+ and +,–) in Figure 3.1. For the symmetrical, stimulus-response
models in the Deed games of conflict and cooperation, these relationships are
based on assumptions of reciprocity (the exchange of similar exercises of power)
and a relatively low power differential between actors (the roughly equal ability
of each actor to control the other). Simple versions of social exchange theory and
role theory explain variations in these relationships with approval and aggres-
sion propositions and with balance and congruence propositions, respectively, in
which the responding actor takes its cues for behavior from the behavior exhib-
ited by the initiating actor. The rationality of the respondent in these models is
primarily purposeful (goal- oriented) rather than mindful (means-oriented) or
empathetic (strategic) regarding the “potentialities” for different outcomes.
The mechanics of these simple rationalist solutions, based on a synthesis of
behavioral exchange theory and cognitive role theory models, to the “two-body
problem” of cooperation and conflict are depicted as a political field diagram in
 The Behavioral Model of IR 49

Figure 3.2 from the point of view of Ego as the respondent and Alter as the initia-
tor. Depending on the cue from Alter on the horizontal axis and Ego’s role concep-
tion on the vertical axis, Ego’s behavioral role enactment will vary, according to the
applicability of the following propositions from exchange theory and role theory
(Walker 1987b: 85; paraphrased from Homans 1961/1974: 15–50), which pre-
dict the positive or negative valence of the respondent’s behavior (role enactment)
based on the valence of its own disposition (role conception) and the correspond-
ing valence of its expectation regarding the initiator’s behavior (cue).
Proposition 1. In choosing between alternative actions, an actor will choose
that one for which (as perceived by the actor at the time) the value (V)
of the result, multiplied by the probability (p) of getting the result, is the
greater (rationality proposition).
Proposition 2. The more often in the recent past an actor has received a
particular reward, the less valuable any further unit of that reward becomes
(deprivation-satiation proposition).
Proposition 3. When an actor’s actions does not receive an expected reward
or receives unexpected punishment, aggressive behavior becomes more
likely (aggression proposition).
Proposition 4. When an actor’s action receives an expected reward, espe-
cially a greater reward than expected, or does not receive an expected pun-
ishment, the actor becomes more likely to perform approving behavior
(approval proposition).
These propositions are based on axioms derived from elementary economics and
Skinnerian propositions from animal behavior. Kunkel and Nagasawa claim that
the Homans model provides only one explanation of how behaviors are established,
namely, “that behavior must occur before it can be re-enforced; hence a new behav-
ior must ‘exist’ before it can be ‘learned,’” and they contend that there is “much
evidence ... (e.g., Bandura 1969) ... that most human learning proceeds through
modeling rather than direct experience, as Homans’s model implies” (Kunkel 1977:
445–446; Kunkel and Nagasawa 1973: 531–532). A fifth proposition added to the
Homans model takes account of these criticisms, incorporating the notion of a rep-
ertoire of learned-modeling responses, which is consistent with the assumptions of
role theory that an actor’s behavioral response can be learned vicariously via mod-
eling by others and will be balanced and congruent with the behavioral stimulus
(Kunkel and Nagasawa 1973: 535; see also Walker 1987b: 85–89).
Proposition 5. If an activity is learned through modeling, then the actor is
likely to repeat it in similar circumstances—provided the model’s behavior
is reinforced (balance/congruence proposition).
In the upper-left quadrant of Figure 3.2, the positive (+) valence of Alter’s cue is
consistent with the positive (+) valence of Ego’s role conception to produce the
prediction from the balance proposition (Prop. 5 above) that Ego’s role enactment
50 Stephen G. Walker

POS (+) ← Alter’s Cue → NEG (–)

P
O
S Ego’s Role Enactment (+) Ego’s Role Enactment (–)
(+) (balance proposition) (aggression proposition)

↑
Ego’s Role
Conception

↓
(–) Ego’s Role Enactment (+) Ego’s Role Enactment (–)
N (approval proposition) (balance proposition)
E
G

Source: Adapted from Walker (1987b: 88).

FIGURE 3.2 Ego’s Role Enactment Predicted by Exchange Theory and Role Theory

behavior will also have a positive (cooperative) valence. In the lower-left quadrant,
there is an imbalance between the positive (+) valence of Alter’s cue and the negative
valence of Ego’s role conception; the prediction from exchange theory’s approval
proposition (Prop. 4) is that Ego’s behavior will have a positive (+) valence.
Role theory’s explanation of the mechanism that generates this response from
Ego is that Alter’s positive cue produces an “altercasting” effect, i.e., that Alter’s
positive cue re-casts Ego’s response by evoking an alternative (positive) role
from Ego’s role repertoire with a valence that corresponds to the expectations
expressed by the positive cue from Alter (Walker 1987b; Wendt 1999; Malici
2008). Altercasting works in reverse in the upper-right quadrant, as the positive
valence of Ego’s role conception is not balanced with the negative valence of
Alter’s cue, leading to the prediction from the aggression proposition (Prop. 3)
that the Ego’s role enactment behavior will have a negative valence. The same
conflict behavior for Ego is predicted in the lower-right quadrant by role theory’s
balance proposition, as Alter’s cue and Ego’s role conception have the same nega-
tive valence (Walker 1987b: 88–89).
Exchange theory’s rationality proposition (Prop. 1) and deprivation-satiation
proposition (Prop. 2) are not necessary in this account to explain the direction
(valence) of Ego’s role enactment behavior; however, they can predict theoreti-
cally the congruence (degree of match between Alter’s stimulus and Ego’s response).
Such predictions require information regarding which rewards Ego values more
(Prop.1) and the relative strength of the cues regarding different roles emanating
from Alter (Prop. 2) in a situation characterized by role competition or role conflict,
 The Behavioral Model of IR 51

respectively, when Ego is stimulated by cues from more than one Alter to meet
competing expectations simultaneously or to choose between conflicting expec-
tations (Walker 1987b: 86–87, 1992). Similarly, the relative strength of Ego’s
response in the upper-right quadrant can be predicted theoretically by Dollard et
al.’s (1939) frustration- aggression proposition, which specifies that there is a con-
gruent relationship between the strength of frustration experienced by Ego and
the strength of Ego’s aggressive response. These propositions and situations lead
exchange theory in the direction of a higher degree of rationality (mindfulness),
in which the decision maker differentiates among goals and the different means
associated with them.

The Case of Woodrow Wilson

Before turning to a full consideration of more complex models of rationality, let
us consider briefly whether the decisions of political leaders can be explained ade-
quately without even the simple model of rationality associated with the exchange
theory and role theory model in Figure 3.2. In this model a decision maker (Ego)
responds positively or negatively, depending on positive or negative cues pro-
vided by the stimulus from Alter’s behavior. Ego’s response is also governed by
emotions and motivations aroused from the reward or punishment effects associ-
ated with Alter’s stimulus, as specified by the propositions in Homans’s exchange
theory. There is a relatively “thin” account of a search for alternatives and a cal-
culation of costs and benefits associated with them—what Rapoport (1960) calls
their “potentialities”—making this model at best a rather simple model of rational
choice. Do political leaders engage in such simple decision-making behavior,
which borders on what Simon (1985) calls “radical irrationality,” based primarily
on impulses (feelings and motivations) unmediated significantly by thought pro-
cesses such as the search for alternatives and the calculation of costs and benefits
regarding goals?
A candidate case of radical irrationality is President Woodrow Wilson’s con-
troversial decision to oppose amendments to the Treaty of Versailles by the U.S.
Senate at the end of World War I. Most analysts agree that his actions contributed
heavily to an outcome that was contrary to his ostensible goal of ratification.
The historical record indicates that Wilson chose to oppose any amendments
or reservations attached to the Versailles Treaty, even ones that would ensure
its ratification without altering the terms of the treaty in other than a cosmetic
fashion (Bailey 1958; George and George 1956; Freud and Bullitt 1967). The
puzzle is “why a distinguished scholar of executive-congressional relations and an
experienced president of the United States would take actions likely to reduce the
prospects for ratification—when other, more reasonable alternatives were avail-
able” (Walker 1995: 697–698).
The simple explanation offered by exchange theory is that Wilson returned from
Paris with the Versailles Treaty in hand, feeling positive about his accomplishments
52 Stephen G. Walker

in negotiating the Treaty and expecting to be welcomed as a victorious peace-

maker. Instead, he encountered hostile cues from influential members of the U.S.
Senate (especially Republican Senator Henry Cabot Lodge) who were opposed to
U.S. membership in the League of Nations organization embodied in the Treaty.
This combination of Ego’s positive (+) role conception and Alter’s negative (–) cues
placed Wilson’s location in the upper-right quadrant of Figure 3.2, and exchange
theory’s aggression proposition predicts his behavioral response to altercasting will
be to enact a negative (–) role. Wilson’s negative response in opposing any amend-
ments to the Treaty confirmed this prediction, and a fight (in Rapoport’s sense)
ensued, ending with Wilson’s defeat both physically and politically. He suffered a
stroke during a whistle-stop campaign to rally public support for the Treaty and the
U.S. Senate failed to ratify the Treaty (George and George 1956).
Although this account conforms to the relatively simple model of human
behavior offered by exchange theory and role theory, it does not explain impor-
tant features of Wilson’s decision making. When he returned to Washington,
why did Wilson expect to be greeted with positive cues and assume that victori-
ous peacemaker was his appropriate role identity? Why didn’t he search explicitly
for alternatives to simply opposing any amendments and choose the one (cosmetic
amendments) that provided the goal of ratification at an acceptable cost (Appleton
1968)? Several “thicker” accounts of Wilson’s decision-making processes offer
answers to these questions. A medical explanation offered by Wilson’s apologists
is that he was a sick man with diminished decision-making capacities by the time
he returned from Paris. Weakened by previous strokes, a bout with influenza in
Europe, and the laborious peace negotiations in Paris, he was not in a frame of
mind to act rationally (Weinstein et al. 1978/79; Weinstein 1981).
However, a rival psychoanalytical explanation advanced much earlier by
George and George (1956) emphasized Wilson’s emotional predisposition to
brook no opposition when others attempted to thwart his goals. Domination by
his father during childhood and power struggles with colleagues during his tenure
as the President of Princeton University predated his decision making regard-
ing the Versailles Treaty and exhibited similar patterns of stubborn intransigence
on Wilson’s part over issues in which he was emotionally engaged (George and
George 1981/82; see also Tucker 1977). A psychiatric explanation attempted a
synthesis of these contending explanations. Post (1983) argued that as an individ-
ual’s health declines in the face of advancing age and with the onset of physical ail-
ments, his emotional and motivational predispositions are likely to have a greater
impact on his behavior (see also Post 1980).
Finally, an analysis of the motivational foundations of Wilson’s operational
code (political belief system) attempted to test the implications of these con-
tending explanations by comparing the motivational imagery in Wilson’s public
statements during the debate over ratifying the Versailles Treaty (1919) with two
of his earlier decision-making episodes regarding U.S. military intervention in
the Mexican Revolution (1913) and entry into World War I (1917). The results
 The Behavioral Model of IR 53

indicated that Wilson’s emotional “state of mind” did not differ significantly for
the Versailles case in comparison with the other two cases (Walker 1995).
The actual distributions of power, achievement, and affiliation imagery in
Wilson’s public statements conformed generally to the expected distributions pre-
dicted by the George and George explanation in all three cases. Need for power
imagery was much greater overall than need for achievement imagery or need
for affiliation imagery and differed significantly from the predictions by Tucker
regarding the need for achievement and by Weinstein et al. regarding differ-
ences between Wilson’s decision making over the Versailles Treaty compared to
earlier decision-making episodes (Walker 1995: 708–712). There was also some
evidence to support Tucker’s claims of significant interaction between Wilson’s
needs for power and glory (achievement), which motivated his decision making
during the overt conflict phase of the Versailles Treaty debate though not in the
other two cases (Walker 1995: 709–711).
The general implication is that Wilson’s emotional predispositions did not
differ significantly in the Versailles Treaty episode compared to the other two
episodes, supporting the psychoanalytical explanation over the medical explana-
tion for Wilson’s decisions and rendering a psychiatric explanation combining
Wilson’s mental and physical health moot in some respects. However, Post’s
(1983) synthesis does offer new evidence of Wilson’s relationship with an adoring
mother, which expands and supports both the George and George account of the
origins of Wilson’s need for power in his relationship with a domineering father
and Tucker’s claim that a need for achievement motivated Wilson along with a
need for power (Post 1983; Tucker 1977).
These analyses of Woodrow Wilson’s foreign policy decisions as well as the
analyses offered by exchange theory and role theory are “first approximations,”
because they encompass a large, undifferentiated behavioral space and extend
over a relatively long period of time. Their validity requires only a relatively loose
fit among the concepts of cues, role conceptions, and role enactment in Figure
3.2 and Wilson’s emotional predispositions and motivations, and his three deci-
sions to “intervene” in Mexico, “enter” into World War I, and “oppose” any
Versailles Treaty amendments. Intervention encompassed behaviors extending
at least over several months or years, as did entry and opposition (Walker 1995).
They represent “big” decisions to shift from “peace” to “war” with Mexico and
Germany, and from “negotiation” to “fight” with the U.S. Senate. They do not
provide a detailed descriptive account of the words and deeds that represent those
role enactments, and there is no detailed explanation of how Alter’s cues and
Ego’s role conceptions interacted to provide a detailed script for these behaviors.

Complex Models of Rationality

The contending explanations for Wilson’s decision-making behavior suggest that it
is desirable to thicken our formulation of the rationality theorem, in order to make
54 Stephen G. Walker

progress and take account of the kinds of phenomena revealed by our discussion of
this case. It is possible to pursue models and measures of more sophisticated versions
of either exchange theory or role theory (e.g., Blau 1964; Emerson 1972a, 1972b;
Stryker and Statham 1985; Walker 1992; Wendt 1999); however, the focus turns
now to theoretical models associated with operational code analysis. The lower,
microscopic levels of operational code analysis interface eventually with these other
mesoscopic possibilities, which operate at higher theoretical and empirical levels of
spatial and temporal aggregation (Walker 1987a, 2004a, 2007a).
Case studies of Woodrow Wilson’s decision making suggest that it is possible for
operational code analysis to address the omissions in these conceptual and empiri-
cal approximations by providing an analytical framework and a methodology for
addressing them (Walker 1995; Walker and Schafer 2007). The rationality theorem
that drives operational code analysis is explicitly mindful and empathetic as well as
purposeful; the analysis of a leader’s operational code directly focuses and expands
on Lenin’s Question of “Who (controls) whom?” with a series of theoretical propo-
sitions about the strategic exercise of alternative forms of power by Ego and Alter
in the political universe (George 1969). As demonstrated in the operational code
analysis of the various explanations for Wilson’s decision to “fight” the U.S. Senate,
it also has the flexibility and comprehensive reach to encompass the emotional and
motivational foundations of the leader’s belief system (Walker 1983, 1995; see also
Leites 1951, 1953 and Walker and Schafer 2007).
Nathan Leites coined the term, “operational code,” to refer narrowly to “the
conceptions of political strategy” in Bolshevik ideology, which he extracted
from the texts of public statements by Lenin, Stalin, and other Bolshevik leaders
and related to cultural themes that he found in literary texts by Russian authors
(Leites 1951, 1953). He located the core of Bolshevik strategic thinking in Lenin’s
cryptic “Kto-kovo?” question, i.e., “Who (will destroy, control, utilize) whom?”
which Leites found to be associated with an unusually high need for power and
an emotional predisposition to fear annihilation. He found the immediate origins
of these features rooted in Lenin’s personality, also influenced by his socialization
experiences as the prototypical Bolshevik revolutionary, features shared in vary-
ing degrees by Stalin and Khrushchev (Leites 1951, 1953, 1964). A generalization
of the findings from the Leites studies of the Bolsheviks leads to the following
propositions of operational code theory (Walker 1983, 2003).

Proposition 1. As a result of childhood and early adult socialization experi-

ences in family and society, an individual acquires the dominant motives in
his personality prior to adopting a political belief system.
Proposition 2. An individual tends to adopt a political belief system that is
compatible with his/her constellation of the needs for power, affiliation,
and achievement.
Proposition 3. Although an individual’s belief system may develop a consis-
tency that is independent from random fluctuations in immediate personal
 The Behavioral Model of IR 55

needs, the activation of these beliefs by environmental stimuli may arouse

personal needs embedded in the belief system as the individual uses the vari-
ous elements of his belief system to interpret a decision-making situation.
Proposition 4. Once aroused, these motives may contribute to the cognitive
rigidity of an individual’s beliefs and account for the intensity of cognitive
dissonance and behavioral intransigence in the face of new information or
other stimuli from the environment.
Proposition 5. Conversely, vivid stimuli or changes in context, respectively,
may lead to behavioral change in the form of social learning or structural
adaptation and even cognitive change in the form of experiential learning
(changes in beliefs).

The evolution of the operational code analysis research program has not lost the
cognitive, emotional, and motivational focal points of the original Leites studies
of the Bolshevik operational code; however, it has developed a more economical
way of amalgamating these features within the parsimonious analytical and meth-
odological framework shown in Figure 3.3 (Walker 1983, 1990). This framework
emphasizes the cognitive contents of the individual’s belief system, shown in
the four quadrants of the typology of operational codes in this figure. The mo-
tivational foundations of these belief systems are the horizontal and vertical axes
emanating from the “nuclear self” as the development of needs for achievement,
affiliation, and power (Kohut 1971, 1977; Winter 1973, 2003; Walker 1983;
Walker and Falkowski 1984). The relationship between these cognitive and mo-
tivational contents of the individual’s personality dispose a decision maker to-
ward different preference rankings regarding the political outcomes of settlement,
deadlock, domination, and submission, depending upon which of the “states of
mind” defining the social self are aroused by the occasion for decision.
The four types (A,B,C,DEF) in this figure are ideal types located in the three-
dimensional space formed by the dimensions of power, affiliation, and achieve-
ment, which Figure 3.3 has “flattened” into a two-dimensional field diagram of
the “social (political) self” by moving the location of the “nuclear self” to the
new origin dividing the four quadrants (the intersection of the horizontal axis
and the new vertical axis represented by the dotted line). While this two-dimen-
sional field diagram does not capture the three-dimensional spatial quality of the
personality, it does allow easier representations of possible “states of mind” of the
“social self,” i.e., the “self-in-situation” represented by the texts of philosophical
and instrumental beliefs about the exercise of power in different political situa-
tions located in the four quadrants of the typology.
These texts summarize the main Philosophical and Instrumental beliefs identi-
fied by George (1969, 1979) and Holsti (1977), which constitute a leader’s opera-
tional code. George’s (1969) review of the Leites (1951, 1953) studies suggested
that a leader’s Philosophical beliefs included beliefs about the nature of the politi-
cal universe (P-1), the prospects for realizing fundamental political values (P-2),
56 Stephen G. Walker

High nAff
(Essential Alikeness)
TYPE A TYPE C

Conflict is temporary, caused by human misunder- Conflict is temporary; it is possible to restructure the
standing and miscommunication. A "conflict spiral," state system to reflect the latent harmony of interests.
based upon misperception and impulsive responses, is The source of conflict is the anarchical state system,
the major danger of war. Opponents are often which permits a variety of causes to produce war.
influenced by nonrational conditions, but tend to Opponents vary in nature, goals, and responses to
respond in kind to conciliation and firmness. Optimism conciliation and firmness. One should be pessimistic
is warranted, based upon a leader's ability and about goals unless the state system is changed,
willingness to shape historical development. The future because predictability and control over historical
is relatively predictable, and control over it is possible. development is low under anarchy. Establish optimal
Establish goals within a framework that empha- goals vigorously within a comprehensive
sizes shared interests. Pursue broadly interna- framework. Pursue shared goals, but control risks
tional goals incrementally with flexible strategies by limiting means rather than ends. Act quickly
that control risks by avoiding escalation and acting when conciliation opportunities arise and delay
quickly when conciliation opportunities arise. escalatory actions whenever possible; other
Emphasize resources that establish a climate for resources than military capabilities are useful.
negotiation and compromise and avoid the early
use of force.

Preference Order: Settle>Deadlock>Dominate>Submit Preference Order: Settle>Dominate>Deadlock>Submit

High
Nuclear Social nAch
Self Self (Ideals)
Preference Order: Dominate>Settle>Deadlock>Submit Preference Order: Dominate>Deadlock>Settle>Submit

Conflict is permanent, caused by human nature (D); Conflict is temporary, caused by warlike states;
nationalism (E), or international anarchy (F). Power miscalculation and appeasement are the major causes
disequilibria are major dangers of war. Opponents may of war. Opponents are rational and deterrable.
vary, and responses to conciliation or firmness are Optimism is warranted regarding realization of goals.
uncertain. Optimism declines over the long run and in The political future is relatively predictable, and control
the short run depends upon the quality of leadership over historical development is possible. One should
and a power equilibrium. Predictability is limited, as is seek optimal goals vigorously within a comprehen-
control over historical development. Seek limited sive framework. Control risks by limiting means
goals flexibly with moderate means. Use military rather than ends. Any tactic and resource may be
force if the opponent and circumstances require it, appropriate, including the use of force when it
but only as a final resource. offers prospects for large gains with limited risk.

TYPE DEF TYPE B

(Ambition)
High nPow
*Instrumental beliefs are in bold, and philosophical beliefs are not.

FIGURE 3.3 Contents of the Revised Holsti Operational Code Typology

the predictability of the political future (P-3), control of historical development

(P-4), and the role of chance in the political universe (P-5); coupled with these
beliefs were Instrumental beliefs regarding the strategic approach to political goals
(I-1), the tactical pursuit of these goals (I-2), the management of risks (I-3), the
role of timing (I-4), and the utility of different means for exercising political
power (I-5). Holsti (1977) identified combinations of different philosophical and
instrumental beliefs that formed six types of belief systems (A through F), which a
subsequent re-analysis revised into the four types in Figure 3.3 (Walker 1983).
The five propositions extracted earlier from the original Leites studies of the
Bolsheviks suggest that an individual is born with a nuclear self of genetic predis-
positions toward developing needs for power, affiliation, and achievement, which
are shaped by early-childhood socialization experiences into the structure of a
 The Behavioral Model of IR 57

social self containing particular configurations of these motivations and accompa-

nying beliefs about the political universe (Kohut 1971, 1977, 1984; Walker 1983;
Kowert and Hermann 1997; Kowert 2002). These beliefs are aroused intermit-
tently and influenced as “states of mind” by the stimuli of political life, mainly
the exercise of power between Self and Other in various social settings as political
situations.
An individual may have a relatively undifferentiated and static operational code,
dominated by beliefs and motivations that are relatively simple, making just one of
the types in Figure 3.3 the “default” setting for his/her political personality. Other
individuals may have a more complex and dynamic operational code, character-
ized by the development of a more differentiated, multi-faceted set of beliefs and
motivations in which a single leader may experience all of the alternative “states
of mind” represented by each of the four types in Figure 3.3. Depending on which
type of individual is under analysis, either Proposition 4 or Proposition 5 above may
better describe the dynamics of a decision maker’s operational code in different situ-
ations (structural adaptation) and explain the absence (Prop. 4) or presence (Prop. 5)
of learning.
The strategic beliefs that leaders attribute to Self (I-1) and Other (P-1) in their
public statements regarding the exercise of power in the political universe include
affective tags representing positive (+) or negative (–) emotions. The motivational
images of power, achievement, and affiliation associated with those beliefs form the
theoretical foundations for inferring the behavioral intentions (preference rankings)
attributed to Self and Others (Walker 1983; Walker and Schafer 2010). A dual focus
on Self and Other makes the operational code analysis of a leader’s belief system
compatible theoretically with role theory’s focus on Ego and Alter and with game
theory’s formal models of strategic interaction between two such players (Walker
2004b; Walker and Schafer 2007; Brams 1994). The accompanying methodological
strategy is to construct indices for a leader’s operational code beliefs by first retriev-
ing and then classifying the exercise of different forms of power attributed to Self
and Other in a leader’s statements, which can be done either by hand or by machine
with the Verbs In Context System (VICS) of content analysis.
Together the use of more rigorous game theory models and more detailed em-
pirical observations obtained by VICS moves operational code analysis past the
earlier conceptual and empirical approximations provided by exchange theory and
role theory. The game theory models specify sequences of behavior (scripts) for
role enactment by Self and Other while the VICS observations supply data to test
whether and how such scripts can be inferred from beliefs extracted from a leader’s
public statements. These features allow for an expansion in scope from fights to
games and debates in the study of conflict behavior and include their counter-
parts as well in the study of cooperation behavior (Rapoport 1960; Walker 2004a,
2004b, 2007).
The logic of the VICS categories for identifying a leader’s beliefs regarding
the exercise of power is also compatible with the logic of the categories for
58 Stephen G. Walker

identifying the sequences and frequencies of foreign policy behavior by a leader’s

state in event histories. This interface provides the opportunity to explain the
patterns of state interaction in the event histories by reference to similar patterns
in the belief inventories of leaders who either chose directly or were responsible
indirectly for the actions in the event histories. The connection between these
patterns is provided by the bounded rationality theorem, which specifies a state’s
actions reflect the degree of purposefulness (goal orientation), mindfulness (means
orientation), and strategic empathy (other orientation) in the operational code
beliefs of its leaders. The bounded rationality theorem specifies that the beliefs
of leaders and the actions of states are congruent, i.e., the belief systems of leaders
match the actions of their states.

Integrating Formal Models of Beliefs and Actions

Identifying and comparing these patterns of beliefs and actions is a complex
undertaking and represents an “ill-structured” problem until or unless it is trans-
formed into a “well-structured” problem for solution (Voss 1998; Newell and
Simon 1972; see also Steinbrunner 1974). Voss (1998: 9–12) identifies “ill-struc-
tured” problems as having the following features. Goals are indefinite (no single
outcome is correct), and the problem does not have a well-defined initial state,
constraints, and/or means to reach a goal. Constraints are also generated during
the solving process, which are not in the problem statement and become criteria
for evaluating a solution.
If the “ill-structured” problem of interest is the explanation of foreign pol-
icy decision-making and learning processes, can an analysis of beliefs detect the
latent structure and solution(s) for understanding these processes? Logically, it
requires circumscribing the task environment (social and physical) and problem
space (mental structure) in which the problem occurs; identifying goals, initial
states, constraints, and means to reach goals that constitute a representation of the
problem and its alternative solutions; specifying the operators and stop-rules that
allow movement from the initial state representing the problem to a final state
representing the solution to the problem (Voss 1998: 13–17).
Operational code analysis profiles a decision maker’s belief system via the use
of a common metric for accomplishing these tasks. The common metric employs
attributions regarding the use of positive and negative sanctions (the exercise of
power) by self (S) and others (O) found in the texts (speeches and interviews) of
decision makers. These attributions are represented as entries in cells a, b, c, and d
in the center of Figure 3.4. These attributions are used to construct VICS indices
of key beliefs in Figure 3.4 regarding friendly (+) or hostile (–) strategic orienta-
tions of Self and Other plus the locus of control (<, =, >) relationship between
Self and Other within the decision maker’s political universe.
The combinations of these indices in Figure 3.4 for Self and Other constructed
from the attributions reduce the entropy (possible random states) represented by all
 The Behavioral Model of IR 59

S (+, <) S (+, <) S (+, =) S (+, >) S (+, >)

O (+, >) O (–, >) O (+, =) O (–, <) O (+, <)

+ –
S (+, =) S a b ∑S S (–, =)
O (–, =) O c d ∑O O (+, =)

S (–, <) S (–, <) S (–, =) S (–, >) S (–, >)

O (–, >) O (+, >) O (–, =) O (+, <) O (–, <)

S = Self; O = Other.

FIGURE 3.4 Constructing a Political Belief System Typology

possible combinations of beliefs to a typology of belief systems regarding the exer-

cise of power consistent with the realist theory of power politics from Chapter
2. The VICS indices of Friendliness/Hostility are calculated as percentage differ-
ences between positive (+) and negative (–) attributions, i.e., [(a – b)/a + b)] for
Self (S) and [(c – d)/c + d)] for Other (O). The indices for Historical Control
are calculated as percentage differences between the total attributions for Self and
Other: [(a + b)/(a + b + c + d)] – [(c + d)/(a + b + c + d)] for Self and [(c + d)/(a
+ b + c +d)] – [(a + b)/(a + b + c + d)] for Other (Walker and Schafer 2010;
see also Schafer and Walker 2006c). The different combinations of key beliefs in
Figure 3.4 define the parameters of the problem space in which the processing
of the problem is assumed to take place, that is, “(1) all the possible states of the
problem, including the initial state and goal(s), (2) the operators that allow a person
to move from one state to another, and (3) the constraints of the problem” (Voss
1998: 10. Italics his; Newell and Simon 1972).
The combinations of problem representations and solutions in Figure 3.4 are
modeled in Figure 3.5 with the use of formal computation algorithms specified
in operational code analysis by an alliance of realist power politics theory, cogni-
tive consistency theory, and sequential game theory (Marfleet and Walker 2006;
Walker and Schafer 2007; Brams 1994) and measured with the Verbs in Context
System of content analysis. The use of weaker methods of decomposition (break-
ing the problem into sub-problems) and conversion (transforming the problem
into another problem for better analysis) are intermediate steps in employing
stronger methods from game theory and content analysis (Voss 1998).
The seven subjective games in Figure 3.5 reduce the entropy (possible random
states) represented by all possible states of the problem (78 structurally distinct 2
× 2 games) to a manageable set of subjective games specified by the realist theory
of power politics from Chapter 2. These games represent initial, intermediate,
and final states as steps in the solution to the problem of foreign policy choice.
The solution for each game is purposeful (goal-oriented) as the entries in each cell
represent the rank order of preferences for each player regarding the outcomes of
60 Stephen G. Walker

O (+, >) O (+, =)

CO CF NO-CONFLICT CO CF
CO 4,4* 2,3 GAMES (+, +) CO 4,4* 1,2
S(+, <) S(+, =)
CF 1,1 3,2* CF 2,1 3,3*

O(–, >) O(+, >) O(+, =)

CO CF MIXED CO CF MIXED CO CF
CO 4,2 2,4 MOTIVE CO 3,4 2,3 MOTIVE CO 3,4, 1.2
S(+, <) GAMES S(–, <) GAMES S(–, =)
CF 1,1 3,3* (+, –) CF 4,1 1,2 (–, +) CF 4,1, 2,3*

O(–, >) O(–, =)

CO CF CO CF
CO 3,2 2,4* CONFLICT CO 3,3 1,4
S(–, <) GAMES S(–, =)
CF 4,1 1,3 (–, –) CF 4,1 2,2*

ASYMMETRICAL ← POWER → SYMMETRICAL

S = Self; O = Other. Nash myopic equilibria are asterisked. Brams

non-myopic equilibria are in bold. Dominant strategies are underlined.

FIGURE 3.5 Drawing Inferences about Preferences

settlement (CO, CO), deadlock (CF, CF), domination or submission (CF, CO)
or (CO, CF) generated by the choices of Cooperation (CO) or Conflict (CF) by
Self and Other (S, O). They represent as well each player’s rank order of prefer-
ences, respectively, highest (4) to lowest (1) for the multiple goals of settlement,
deadlock, domination, and submission.
Solutions to these games derived are also mindful, specifying constraints (along
with goals) to represent the problem and act as criteria for evaluating alternative
solutions to the problem. These constraints are the beliefs from Figure 3.4 about the
symmetrical (=) or asymmetrical (< or >) power relationships between the players
and their cooperative (+) or conflictual (–) orientations. The means by which a
decision maker moves from one problem state to another are operators or stop-rules
from game theory that specify the choices to “move” or “stay” at a given state (cell),
based on the ends-means method of backward induction from the final state (equi-
librium) to the initial state of play. Players alternate choices of “move” or “stay”
until the player with the next move chooses “stay” and the other player does not
choose “move” or until play returns to the initial state (Brams 1994, 2002).

Conclusion
To sum up, operational code analysis represents the problem of foreign policy
analysis as a sequential game and predicts multiple solutions, including Nash
 The Behavioral Model of IR 61

myopic equilibria and Brams nonmyopic equilibria. Changes in problem repre-

sentations are accounted for directly by changes in the key beliefs (experiential
learning) about power relationships and strategic orientations, which change pref-
erence rankings for the goals of settlement, deadlock, domination, or submission.
The origins of these constraints are not directly accounted for in the problem
space defined by these beliefs; however, they can be accounted for indirectly by
observing changes in the task environment, such as changes in the identities of
the decision makers (personality differences), the actual power relationships and
stakes between the players (structural adaptation), and the choices of each player
(social learning) as they move through the possible states of the problem (Levy
1994; Hermann 2003; Walker 2007a).
The retrieval of belief systems and subjective games from actionable beliefs is
possible by applying the VICS content analysis methods to the public and private
statements of decision makers (Young 2001). Algorithms for studying repeated
play by individuals plus analyzing groups and organizations with intersecting sub-
jective games are also available (Brams 1994; Marfleet and Walker 2006). We
shall explore the steps in making these applications in the next chapter with the
aid of some examples from the operational codes of Henry Kissinger and Barack
Obama.
4
QUANTUM POLITICS AND
OPERATIONAL CODE ANALYSIS
Theories and Methods

Stephen G. Walker

Introduction
In order to demonstrate how to do operational code analysis and apply it to solve
the problem of explaining continuity and change in foreign policy decisions,
it is desirable first to represent the problem and the general solution as a whole
and then to decompose and convert it into a set of smaller, more manageable
problems and solutions as a series of steps with the aid of some examples from
operational code analyses of Henry Kissinger and Barack Obama. These steps are
demarcated as the following research questions, which are addressed generally in
this chapter and are the subject of more specific research designs in Part II Foreign
Policy Decision Making and Part III Foreign Policy Learning of this book.
1. What are important patterns of continuity and change at different levels
(strategies, tactics, and moves) of foreign policy decisions?
2. What are the important elements (key beliefs) and relationships in a leader’s
operational code (political belief system)?
3. What are the relationships between a leader’s key beliefs about strategic
orientations and the intentions (motivations) and actions (choice and shift
propensities) that s/he attributes to self and others?
4. What is the relationship between a leader’s intentions and actions at various
levels of decision (moves, tactics, strategies)?
5. How does a leader “learn” (change beliefs) and what does it mean for a leader
to learn in different ways and at different levels of learning?

Models of Preferences and Moves

In order to answer these research questions, we shall employ four basic types of
research designs to identify and understand the basic phenomena of foreign policy
 Quantum Politics and Operational Code Analysis 63

decision making. These designs are located in Figure 4.1 within a research field
diagram defined by Space and Time (Galtung 1967; see also Cook and Campbell
1979; King et al. 1994; George and Bennett 2005). The simplest research design is
the single-case study—a snapshot of a single leader’s beliefs and behavior at a sin-
gle point in time—what Feynman (1995: 25) calls “the move of a (single) bishop
on the chess board.” A more complex research design is the structured-focused
comparison of at least two cases compared across space—snapshots of the beliefs
and behavior of two leaders at a single point in time—what Feynman (1995: 25)
describes as “a few pieces at work … in one corner of the (chess) board.”
By extension these two research designs can be extended temporally as well
as spatially into two more research designs in Figure 4.1. One is the repeated-
measures research design in which the beliefs or behavior of at least one case are
observed over time. The other is the quasi-experimental panel research design in
which the beliefs and behavior of at least one case are observed in an alternating
Space

Space

*Case #2
*Case #1
*Case #1

Time Time
Single-Case Study Structured-Focused Comparison
Space

Space

* * * * * * Case #2 * * * Beliefs (Case #1)

* * * Behavior (Case #1)

* * * * * * Case #1

Time Time
Repeated-Measures Series Quasi-Experimental Panel

FIGURE 4.1 Basic Research Designs for Understanding Foreign Policy Analysis
64 Stephen G. Walker

temporal sequence over time. These two research designs refine Feynman’s
method of “rough approximation” to incorporate increasingly rigorous controls
for the effects of space and time on testing ideas about the basic rules of politics.
The general pattern of answers to these research questions is represented in
Figure 4.2 as a pair of interrelated political field diagrams: (1) a hierarchical behav-

d+ ← Alter → e–
Alter Alter Alter Alter TIP Rules of Inference
D E D E D E D E
D “4,3”|← 1,4 D “4,3” →|1,4 D 4,3 → “1,4” D 4,3 |← “1,4” (1) If (+,<), then Set>Ded>Sub>Dom
__ __ __ (2) If (+,=), then Set>Ded>Dom>Sub
↓
Ego __ ↑ ↑ ↓ Ego ↑ ↓ ↓
__ ↑ (3) If (+,>), then Set>Dom>Ded>Sub

E 2,1 → 3,2 E 2,1 ← 3,2 E “2,1”|← 3,2 E 2,1 → 3,2

d+ Ego: Stay Alter: Cycle Ego: Move Alter: Stay
↑
Ego
↓
e– Alter Alter Alter Alter (4) If (–,<), then Dom>Set>Sub>Ded
D E D E D E D E (5) If (–,=), then Dom>Set>Ded>Sub
D 4,3→| 1,4 D 4,3 |← 1,4 D 4,3 |←1,4 D 4,3→|1,4
__ __ (6) If (–,>), then Dom>Ded>Set>Sub

Ego ↑ ↓ ↓
__ ↑ ↓
Ego __ ↑ ↑ ↓
E “2,1”|← 3,2 E “2,1”→ 3,2 E 2,1 → “3,2”E 2,1 ←“3,2”
Ego: Move Alter: Move Ego: Stay Alter: Move

World of Events World of Beliefs

P1/I1
(+)
Alter
Appease Reward Deter
Exploit
DED DDD D E DEE
DDE
TOM Rules of Play (+,<) (+,=) (+,=)
D 4 1 (+,>)
(1) Play starts at an “initialstate.” Ego
Alter E 2 3 Alter
(2) Either player can choose either to D E D E
“move” or “stay” from the “initial
state.” The player who moves is P1. D 4 2 D 4 1
Ego Ego
(3) Player 2 (P2) can respondby choosing E 1 3 E 3 2
“move” or “stay.”

(4) Alternating responses continue until P4 P4

(+)
a Player with the next choice chooses (–)
“stay,” terminating the game at a final
state (underlined final outcome).
Alter
(5) A player will not move from an Alter D E
initial state if it leads to a less-
D E D 2 1
preferred final state or cycles
play to the initial state as the D 3 2 Ego
final outcome. Ego E 4 3
E 4 1
(6) If one player chooses “stay” and
the other player chooses “move”
from the “initial state,” then the Alter
choice of “move” overrides “stay.” Bluff Compel Punish Bully
EED EDD D E EEE EDE
(–,<) (–,=) D 3 1 (–,=) (–,>)
Ego
E 4 2

(–)
P1/I1

FIGURE 4.2 Hierarchical Landscape Features in the Worlds of Beliefs and Events.
 Quantum Politics and Operational Code Analysis 65

ioral landscape of the “world of events” and (2) a hierarchical cognitive land-
scape of the “world of beliefs.” As the big arrows in this figure indicate, the
two worlds are connected by two general models of motivated-cognition (TIP)
and intentional-behavior (TOM). Inside each field diagram is the set of pos-
sible states (entropy) for each one of these worlds. The two models manage the
entropy of each world by reducing the number of important and likely states
within each world and connecting them via pathways specified by the Theory
of Inferences about Preferences (TIP) and the Theory of Moves (TOM), which
identify the relationships between changes in particular states within and across
the two worlds. Each one of these worlds represents a “system” of inter-related
elements that can be decomposed into a set of “one-body,” “two-body,” …
“n-body” problems and solved at different levels of approximation.
The hierarchical landscapes in each field diagram represent Ego’s location,
respectively, as “states of play” in the world of events and as “states of mind” in
the world of beliefs. TIP’s six propositions locate Ego’s preferences for the out-
comes of settlement (D,D), deadlock (E,E), domination (E,D), and submission
(D,E) in the various zones created by the bisection of each quadrant in the world
of beliefs. For example, TIP’s Proposition (2) specifies that if Ego’s Strategic
Orientation is (+) and Locus of Historical Control is (=), then Ego’s preferences
are Settle (4) > Deadlock (3) > Dominate (2) > Submit (1). The location of the
game matrix reflecting this example of Ego’s ranked preferences is on the vertical
axis in the upper half of the world of beliefs diagram and associated with Reward
(DDD) and Deter (DEE) tactics. If Alter’s Strategic Orientation and Historical
Control coordinates are (–,=), then Proposition (5) specifies that Alter’s ranked
preferences are Dominate (4), Settle (3), Deadlock (2), and Submit (1) associated
with Compel (EDD) and Punish (EEE) tactics and located on the vertical axis in
the lower half of the world of beliefs diagram.
Together Propositions (2) and (5) specify that Ego’s subjective game is located
in one of the four quadrants of the dynamic field diagram for the world of
events in Figure 4.2. Ego’s actual location in the world of events depends on
the “state of play”, i.e., which cell of the game is the “initial state” and which
player has the next move in the game. These possibilities are mapped in the four
quadrants of the world of events field diagram. For example, if the “initial state”
of this game is (D,D), then its location is in the (d+,d+) quadrant of the field
diagram. Depending on which player has the next move, the method of backward
induction specifies that Ego will choose “stay” and predicts that Alter will choose
“cycle.” The equilibrium (final outcome) for this game with (4,3) as an “ini-
tial state” is either the Brams nonmyopic equilibrium (4,3) or the Nash myopic
equilibrium (3,2). Which outcome is the predicted final state actually depends
on whether or not the two players are playing the same game, which carries
the analysis to another level of complexity. Here we are interested primarily in
explaining and predicting Ego’s choices to “stay” or “move” from an “initial
state” of play.
66 Stephen G. Walker

For illustrative purposes if we assume simply that Ego’s subjective game and
Alter’s subjective game are identical, i.e., they both know and agree on the rank-
ings for each other’s preferences, then the prediction from the reasoning associated
with the method of backward induction (thinking ahead four moves) is that (4,3)
will be the final state. If Ego has the next move, reasoning clockwise from a final
state of (4,3) to the initial state of (4,3) indicates that Alter will not move [|←] from
(1,4) to (4,3), nor will Ego move from (3,2) to (1,4). Alter will move [→] from
(2,1) to (3,2), but Ego will not move [↓] from (4,3) and, therefore, will choose
“stay.” Reasoning counter-clockwise from the initial state of (4,3) if Alter has the
next move, backward induction reveals that if Alter chooses “move” from (4,3) to
(2,4), it is rational for the two players to “cycle” back to the initial state. Therefore,
there is no incentive for Alter to choose “move” if s/he believes that Ego will
move from (1,4) to (3,2), which may depend on whether Ego’s deterrent threat to
do so is credible in pre-play communication between the players (Brams 1994).
This general example of the connections between the world of beliefs and
the world of events, forged by the rules of TIP and TOM associated with game
theory, are more complex rules of rationality than the earlier, goal-oriented rules
of rationality in the simple models of exchange theory and role theory. It may be
that decision makers do not know or care about another player’s preferences and
instead choose between Escalation (E) and De-escalation (D) based only on their
own preference rankings for the outcomes of settlement, deadlock, domination,
and submission. For example, they may think more myopically, reasoning back-
ward from only one or two moves ahead to avoid their worst outcome. Or they
may be relatively risk-averse or risk-acceptant regarding their worst outcome,
depending on whether the initial state locates them in the domain of gains (one
of their two highest-ranked outcomes) or the domain of losses (one of their two
lowest-ranked outcomes). All of these possibilities are investigated below with the
aid of examples from U.S. Secretary of State Henry Kissinger’s operational code
and decision making during the Vietnam war.

The Operational Code of Henry Kissinger

When Dr. Kissinger came to Washington in 1969 to become President Richard
Nixon’s Assistant for National Security Affairs and later served as Secretary of
State, he brought a well-known set of beliefs regarding the conduct of interna-
tional relations. An analysis of his operational code reveals that he viewed par-
ticipants in world politics as one of four types of leaders: conquerors, prophets,
revolutionaries, or statesmen, who respectively tended to use force, charisma,
ideology, or negotiation as the basis for exercising political power (Walker 1977;
see also Swede and Tetlock 1986). They also differed according to their preferred
solutions to the general game of power politics in which competing claims for
the allocation of values have one of four outcomes from the perspective of an
individual leader (A):
 Quantum Politics and Operational Code Analysis 67

All participants gain something (relative gain outcome); all participants lose
something (relative loss outcome); A gains and the others lose (absolute gain
outcome); A loses and the others gain (absolute loss outcome). ... According
to Kissinger’s operational code, conquerors, prophets, and revolutionaries
are leadership types who prefer an absolute gain outcome and thereby risk
an absolute loss outcome, since human mastery of historical development
is rather modest. The statesman eschews an absolute gain solution and pre-
fers a relative gain solution; however, he will pursue a relative loss policy to
avoid an absolute loss solution.
(Walker 1977: 157. Italics his)

If A is a statesman and the other player is one of the other types of leaders, then
their game will be the game in the world of events diagram in Figure 4.2, i.e.,
absolute gain = E,D domination outcome, absolute loss = D,E submission out-
come, relative gain = D,D settlement outcome, relative loss = E,E deadlock out-
come for A. If Ego and Alter are equal in power, these outcomes are ranked for
Ego (self player) according to TIP’s Rule (2) and for Alter (other player) accord-
ing to TIP’s Rule (5). Did Dr. Kissinger play this subjective game in moving the
Vietnam conflict from war to peace?
The example in Figure 4.3 shows the tactics and key strategic moves by each
player during the first strategic interaction episode in a brief event history of the
Vietnam conflict between 1969 and 1973 (Walker 1977, 2004b). Beginning
in January 1969, the initial United States (USA) tactic in this chronology was
Deter (DEE) while the North Vietnam (NVN) tactic was Bluff (EED) in the first
four moves in this episode leading to an Initial Outcome (IO) of U.S. domina-
tion (E,D). The analysis of this chronology demarcates strategic episodes as four
consecutive moves by two players leading to an Initial Outcome, which may also
be the Final Outcome (FO) if the player with the next move chooses “stay” rather
than “move” or returns to the initial state (D,E) in the game. In this initial episode,
USA chooses “move” to (D,D), which is the move predicted by TOM for USA
from (2,1) to (4,3) in Kissinger’s subjective game of statesman-revolutionary. The
player with the next move (NVN) chooses “stay,” making (D,D) as the Final

Game #1: 00 E, E Deadlock

01 U-D 0169 Propose 2-track approach & mutual troop withdrawal.
02 V-E 0269 Launch fairly large-scale offensive.
03 U-E 0369 Begin secret bombing of Cambodia.
IO: 04 V-D 0469 State readiness for discussions.
FO: 05 U-D 0569 Call for mutual withdrawal/announce 25000 troops leave.
06 V-D 0669 NVN units withdraw SVN/break down to small units.

Behavioral elements for moves 01 through 05 in the Vietnam chronology where Cols. 1 and 2 = Move
#, Col. 3 = Actor (U = USA, V = NVN), Col. 4 = Escalatory (E) or De-escalatory (D) Move, Cols. 5–8 =
Date (Month/Year), and Text = Event Description. IO = Initial Outcome; FO = Final Outcome.
Adapted from Walker (2004b: Appendix 1).

FIGURE 4.3 A Strategic Interaction Episode during the Vietnam Conflict

68 Stephen G. Walker

Outcome of the strategic interaction episode according to TOM’s rules of play (see
Rule 6), and this play of the game ends at a (4,3) nonmyopic equilibrium.
The results of the entire sequential analysis of moves (D or E) by the United
States and North Vietnam in this event history between 1969 and 1973 indi-
cate that the tactical and strategic predictions for the United States in Kissinger’s
statesman-revolutionary game are generally supported by actual American tactical
and strategic decisions in this chronology. Kissinger pursued tactics of reciproc-
ity 71% (20 out 28 opportunities) and the predictions of his key strategic moves
toward the final outcome in the 11 strategic interaction episodes were correct
75% (6 out of 8 opportunities) in this chronology of 55 moves (see Walker 1977,
2004b). While game theory offers a robust explanation for Kissinger’s decisions in
this event history, the pattern of North Vietnam’s decisions does not correspond
to the predictions of the statesman-revolutionary game. The NVN government
pursued statesman tactics of reciprocity 54% (15 out of 28) of the time while
engaging almost as frequently in a variety of other tactics.
This degree of variation across tactics suggests that NVN may have been rela-
tively uncertain about American intentions, i.e., the ranked preferences of USA
for the outcomes of settlement, deadlock, domination, and submission, in com-
parison with Kissinger’s relative certainty that NVN was a revolutionary state. In a
relatively low information environment, it is more likely that Hanoi weighted its
own preference rankings in making decisions as follows: (a) either NVN followed
a goal-oriented maximin rationality principle of purposefulness by making risk-
averse choices that insured it would avoid its own worst outcome; (b) or NVN
followed the means-oriented rationality principle of prospect theory, making risk-
acceptant choices in a strategic interaction episode when the “initial state” of play
was in the domain of losses and risk-averse choices when it was in the domain of
gains (Kahneman and Tversky 1979).
Prospect theory and game theory are theories about two different kinds of
rationality. While both theories assume that decision makers make rational (ends-
means) choices, their other assumptions vary and may lead to different decisions.
Prospect theory’s rule of rationality is based on the one-sided information assump-
tion that the decision maker only knows or cares about his/her own preferences
(Kahneman and Tversky 1979). Game theory’s rule of rationality is based on the
two-sided information assumption that the decision maker knows and cares about
both his/her own preferences and the preferences of at least one other player in a
game (Little 1991). Depending on the decision maker’s domain location and the
relative uncertainty associated with the one-sided information assumption, the
more bounded, rational decision rule of prospect theory is to be risk-acceptant in
the domain of losses and risk-averse in the domain of gains regarding the choice
of decisions.
In contrast, game theory’s rationality rule is less bounded, due to the two-
sided information assumption, so that the rational decision rule of game theory is
invariant with respect to the decision maker’s location (called the “initial state” in
 Quantum Politics and Operational Code Analysis 69

game theory). No matter what the location, the certainty associated with the two-
sided information assumption leads to the same choice of decisions. However,
the emphasis in TOM on the location of the “initial state” as the starting point
for the game provides a link to prospect theory, which also emphasizes location
(domain of gains or domain of losses) as an “initial state” in the definition of the
decision-making situation.
If the information about Other’s preferences is ignored or unavailable, forc-
ing Self to decide between CO (D) and CF (E) from a cell designated as the
starting point, then the features of this game include: (a) the assumption of one-
sided information (Self only knows Self’s ranked preferences for the different out-
comes); and (b) the assumption that once Self chooses either “stay” at the initial
state or “move” from the initial state, Other may have an incentive and certainly
has the opportunity to choose “move” or “stay” in response to Self’s decision.
These conditions are consistent with the assumption of prospect theory that there
is an element of risk (relative uncertainty) involved in choosing between alterna-
tives. That is, the consequences of a choice may be either good (better) or bad
(worse) than the reference point or initial state for making a decision.
One such starting point employed in prospect theory is the status quo, which
is equivalent to the “initial state” specified as a cell in each of the game theory
matrices for the world of events diagram in Figure 4.2. The possible alternative
preference rankings for Ego are specified in the world of beliefs diagram in the
same figure. A comparison shows that the best “fit” between NVN’s predicted
and actual tactics (a choice between “move” or “stay”) fits the rules of prospect
theory better than the predictions of the maximin rule or the previous predictions
reported for Kissinger’s game theory. These results are summarized below.
If the “initial state” is defined as one of four possibilities [Ego, Alter = D,D,
E,E, E,D,D,E] with Ego (NVN) having the next move in the Vietnam event
chronology, then Ego’s next move (E or D) fits the maximin rule’s prediction
58% (16 out of 27) of the time. The fit is only slightly better for prospect theory’s
predictions and depends on assumptions about NVN’s preference rankings Self
(Ego). Assuming NVN preference rankings associated with Appease tactics or
Bluff tactics, respectively, the predictions are correct 59% (16 out of 27) and 56%
(15 out of 27) times. In contrast, they are correct only 41% (11 out of 27) of the
time, assuming either the tactics and preference rankings of the statesman or revo-
lutionary roles associated with Kissinger’s subjective game. The predictions for
the preference rankings associated with Exploitation tactics are also correct only
44% (12 out of 27) of the time (see Walker 2004b).
These results illustrate two important lessons. First, it is valuable to test different
levels of the rationality theorem in analyzing the fit between beliefs and decisions.
Different leaders may exhibit different levels of instrumental or strategic rational-
ity. Second, it is problematic to assume that one leader’s beliefs about another
leader are correct. Independent measurements of each leader’s beliefs should be
the norm. The results of the data analysis above suggest the hypotheses that (a)
70 Stephen G. Walker

the NVN leaders’ beliefs were relatively uncertain or divided regarding U.S.
preferences, and (b) they believed themselves to be relatively weak compared to
the USA regarding Lenin’s Question of “Who (controls) whom?” The next sec-
tion of this chapter addresses the task of explicitly measuring the beliefs of leaders
with the Verbs in Context System of content analysis.

Measuring the Beliefs of Leaders

The focus on continuity and change in foreign policy beliefs and behavior in this
chapter has several important parallels with the focus of quantum (subatomic)
physics, the study of atomic mechanics v. the classical mechanics of larger physical
bodies. They include a focus on the stability of small-scale objects (subatomic-size
particles of energy and individuals) versus large-scale objects (planet-size particles
of matter and nation-states) and small-scale changes (the dynamics of energy par-
ticles and the dynamics of thought) that nonetheless constitute large-scale changes
(the dynamics of colliding and repelling by objects and the dynamics of coopera-
tion and conflict behavior by states).
There are also theoretical and methodological parallels in studying micro and
macro levels of the mechanics for these two classes of phenomena in politics and
physics. Different covering laws govern the behavior of large-scale and small-scale
particles in physics and politics. However, there is a unifying theory in both dis-
ciplines: the theory of relativity in physics and the theory of rationality in politics,
which need to be qualified methodologically when they are applied at the micro
v. macro levels of analysis.
The theory of rationality in politics needs to be thickened from a “thin”
account acting as a “first approximation” of rational choice models when the
level of analysis shifts from a focus on macro-configurations of behavior (war and
peace) to the micro-dynamics of small-scale behaviors (words and deeds) that
maintain and change these larger states of conflict and cooperation behavior. The
principles of substantive rationality that focus on the external context (domestic
and international conditions) of state behavior need to be supplemented with the
principles of bounded rationality that focus on the internal source (beliefs, emo-
tions, and motivations) of individual behavior.
The parallel shift in focus in physics from classical to atomic mechanics was
accompanied by a crisis in observation, in which the laws of classical mechanics
governing the observation of large-scale aggregations of behavior were found
to be approximations that needed to be qualified radically by the laws of atomic
(quantum) mechanics that explained the behavior of small energy particles such as
electrons and photons. A similar crisis in observation accompanies the shift from
states to individuals in politics. The first approximations of the laws of power
politics theory in Chapter 2 of this book need to be qualified radically when
they are juxtaposed with the laws that seem to govern the behavior of individu-
als as decision makers in this chapter. The most striking parallel in the two crises of
 Quantum Politics and Operational Code Analysis 71

observation is the realization that the exact prediction of behavior at the micro level of analy-
sis is impossible in both physics and politics.
As Feynman (1995: 135) states this point regarding physics,
We would like to emphasize a very important difference between classical
and quantum mechanics. ... We do not know how to predict what would
happen in a given circumstance, and we believe now that it is impossible, that
the only thing that can be predicted is the probability of different events. It
must be recognized that this is a retrenchment in the earlier ideal of studying
nature. It may be a backward step, but no one has seen a way to avoid it.
The indeterminate nature of the behavior of very small physical particles is con-
firmed ultimately by the results of both thought experiments and laboratory
experiments, which conclude that the act of observation alters the behavior of
energy particles and their components (Feynman 1995: 115–138).
This conclusion led to the formulation of the uncertainty principle in physics by
Heisenberg regarding the observation of the properties of very small particles, which
states that if you know either the momentum or the position of an object, you can-
not know the exact value of the other because intrinsic uncertainties in position and
momentum at any instant must preclude an exact observation of either one. “The
more general statement of this principle is that one cannot design equipment in any
way to determine which of two alternatives is taken, without, at the same time,
destroying the pattern of interference” (Feynman 1995: 136). The phrase, “which
of two alternatives is taken,” refers to the fact that at very small levels of observation
energy particles are in constant motion in the sense that they are “spread out” and
not arbitrarily localized; by extension so are the larger forms of objects (solids, liq-
uids, gases) that energy can take, although the “spreading” is much less, and in fact,
relatively undetectable. This constant motion is an essential part of the explanation
for the indeterminate observation of small (atomic and subatomic) particles.
The observation of the small-scale constituents of behaviors and beliefs in
politics may also be indeterminate, and the methodological solution may also
be the same. The minutiae of human behavior and thought can change rapidly
from moment to moment and day to day, due either to the free will of human
beings and their capacities for deception, or to the acts of observation and learning
(Goffman 1959, 1969; Landsberger 1958; Herrmann et al. 1999). As in the case of
physics, the solution may well be to think of their observation and prediction in
terms of probabilities. In physics, the probability of such an event is estimated by
the sum of the probabilities of the alternative paths for the event: “If an experi-
ment is performed which is capable of determining whether one or another alter-
native is actually taken, the probability of the event is the sum of the probabilities
for each alternative” Feynman (1995: 134).
This approach is also the basic solution offered by the VICS method of content
analysis for measuring beliefs and behavior, expressed generally as P = P1 + P2,
i.e., the probability of an event P is the sum of the probabilities for each alterna-
72 Stephen G. Walker

tive P1 and P2. VICS indices are based on the probabilities of different beliefs,
which are constructed as percentages from the statements of leaders and hypoth-
esized to influence the decisions of states observed in the event histories of foreign
policy behavior. Normally, the hypothesized patterns of influence are analyzed
from observations retrieved and coded from sources compiled by journalists and
historians of public statements (beliefs attributed to leaders) and public events
(behavior attributed to states).
The VICS categories of observation are the six basic forms of the exercise of
power, defining its entropy, which corresponds to the relevant content of the
transitive verbs in a text: reward, promise, appeal/support, oppose/resist, threaten,
punish. All are forms of control (who-whom) relationships in which the subject
of the verb (who) gets the object of the verb (whom) to “do” something that
the latter would not otherwise do (Dahl 1957; McClelland and Hoggard 1969;
Baldwin 1989). Not all transitive verbs in a text contain this information and are
discarded in the construction of the VICS indices. Those verbs that do qualify
for retrieval and coding from a text can be summed or weighted into indices by
subject (Self), object (Other) or topic (Issue) for an operational code analysis of
the speaker’s political belief system, as illustrated by the following examples from
a speech by U.S. Senator Barack Obama on May 26, 2005 before he became
President of the United States (Obama 2005).

The Operational Code of Barack Obama

The following analysis is a snapshot of Senator Obama’s “state of mind” on May
26, 2005 at the time he made this speech. He may not have literally “believed”
everything he said at the time and may not act on any of these beliefs as President.
He may have also “changed his mind,” i.e., learned new beliefs and discarded old
ones since this time. Ideally, the observation of beliefs should be preceded and
followed by observations of behavior the way that surveys of individuals compare
public opinion before and after key events to detect whether preceding events
affect the attitudes of potential voters and whether those attitudes affect subse-
quent preferences for action by the government or voting behavior by the public
(see Herrmann et al. 1999). In contrast, the Obama speech represents a single-case
study design in which the cognitive contents of the Obama speech are collected
at a single point in time. The retrieval and recording of the observations from this
speech initially look like the coding decisions marked in bold in the text below
from Obama’s speech and also tallied in bold at the end of each paragraph.
Mr. President, the second set of questions I would like to raise concerns the
U.S.-Russian relationship. Where is this relationship heading? Will Russia
be an adversary? A partner? Or something in between? I don’t ask these
questions simply because I am a nice guy and I want to get along 18S(+1)
with the Russians. I ask these questions because they directly impact our
progress towards securing and destroying stockpiles of nuclear weapons and
 Quantum Politics and Operational Code Analysis 73

materials. In the last few years, we have seen some disturbing trends in
Russia—the rapid deterioration of democracy and the rule of law; bizarre
and troubling statements 19O(–1) from President Putin about the fall of
the Soviet Union; the abuses 20O(–3) in Chechnya; and Russian med-
dling 21O(–3) in the former Soviet Union from the Baltics to Ukraine to
Georgia. The Russians must understand that their actions on some of these
issues are completely unacceptable. At the same time, I believe that we
have to do a better job of working 22S(+1) with the Russians to make sure
that they are moving 23O(+1) in the right direction. This starts by being
thoughtful and consistent about what we say and what we do. Tone is
important here. Some of the statements by our own officials have been con-
fusing, contradictory, and problematic. At times, I have been left scratching
my head about what exactly is our policy and how Administration state-
ments square with this policy. S: 002 000; O: 001 102
Another issue is the level of sustained engagement with Russia. I am glad
that the President and Secretary of State have made 24S(+1) a number of
trips to Russia. But, as these trips are but a few days every year or so, this is
only one aspect of the relationship. An additional part, which has suffered in
recent years, is our foreign assistance programs to Russia and the rest of the
former Soviet Union. These programs are an essential way for the United
States to maintain 25S(+1) our engagement with Russia. They aren’t just
giveaways; they are programs that advance U.S. interests by strengthening
26S(+1) democracy and civil society, enhancing 27S(+1) economic devel-
opment, and dealing 28S(+1)with international health issues—in addition
to curbing 29S(–1) the nonproliferation threat. At a time when these pro-
grams are desperately needed, their budgets have been cut dramatically. At
a time when we should be doing more to engage and shape the future of
Russia, we seem to be doing the exact opposite. The non- proliferation
threat does not exist in a vacuum. The issues I just mentioned—along with
other important issues such as our own strategic nuclear arsenal—must also
be considered as we move forward. S: 005 100; O: 000 000

Coding these two paragraphs involved four steps: (1) underlining all of the transi-
tive verbs in each paragraph; (2) assigning an “S” for Self if the subject of the verb
is the speaker or someone with whom the speaker identifies and “O” for Other if
the subject is someone with whom the speaker does not identify; (3) assigning a
positive (+) valence for cooperation content or a negative (–) valence for conflict
content, or assign neither one if the verb is not codable clearly as one or the other;
(4) assigning a scale value of “3” if the content of the verb is a present or past deed
and either a “2” if the verb is in the future tense or a “1” if the verb is a word in
the present or past tense. It is possible to disaggregate both Self and Other into dif-
ferentiated topical categories, as well as to disaggregate the verbs into Issue Areas,
such as the issue of U.S.–Russian relations in the excerpt above, plus the issues of
74 Stephen G. Walker

Russian Nuclear Weapons and Nuclear Weapons Proliferation in the remainder

of this speech. A more detailed discussion of these coding rules is in Schafer and
Walker (2006; see also Walker and Schafer 2010).
The tallies for the relative frequencies of the relevant transitive verbs are the
basis for constructing the indices of Obama’s three key beliefs about Self’s Strategic
Orientation (S), Other’s Strategic Orientation (O), and the Historical Control
Difference (HCD) between Self and Others in the speech. These observations and
indices along with inferences from TIP and TOM are used to construct Obama’s
subjective games for the three Issue Areas in the entire speech in Figure 4.4. They
show that Self (Obama) has a consistently cooperative (+) Strategic Orientation
while Other (Russia in two of the Issue Areas and a generalized Other in the
Nuclear Proliferation Issue Area) have consistently conflictual Strategic Orientations.
Obama’s Historical Control Difference Index is greater than his Strategic Orientation

Entire Obama Speech Tallies

Self (n = 27) Other (n = 15)

+3 +2 +1 –1 –2 –3 +3 +2 +1 –1 –2 –3
Rew Pro App/Sup Opp/Res Thr Pun Rew Pro App/Sup Opp/Res Thr Pun
Freq. 1 1 20 5 0 0 0 0 3 4 2 6
Wtd. 3 2 20 –5 0 0 0 0 3 –4 –4 –18
% .04 .04 .74 .18 .00 .00 .00 .00 .20 .27 .13 .40

Who-Whom Matrix Subjective Game

+ – + – ∑ D E
S a b 22 5 27 D 4,3 1,4 S: (+,=)
O c d 3 12 15 E 2,1 3,2* O: (–,=)

Self’s Strategic Orientation Index = (a – b)/(a + b) = +.64

Other’s Strategic Orientation Index = (c – d)/(c + d) = –.60
Self’s Historical Control = (a + b)/(a + b + c + d) = .64
Other’s Historical Control = (c + d)/(a + b + c + d) = .36
Historical Control Difference (Self – Other) = .64–.36 = +.28
Historical Control Difference (Other – Self) = .36–.64 = –.28
Nonmyopic Equilibrium (NME) bold; Myopic Equilibrium (ME) asterisked.

Soviet Nuclear Weapons U.S.–Russian Relations Nuclear Weapons Proliferation

Rew Pro App Opp Thr Pun Rew Pro App Opp Thr Pun Rew Pro App Opp Thr Pun
S 1 1 5 2 0 0 0 0 7 1 0 0 0 0 8 2 0 0
O 0 0 1 1 2 4 0 0 1 1 0 2 0 0 1 2 0 0

Self (S) = +.56; Other (O) = –.75 Self (S) = +.75; Other (O) = –.50 Self (S) = +.60; Other (O) = –.33
HCD (S) = +.06; HCDiff (O) = –.06 HCD (S) = +.34; HCDiff (O) = –.34 HCD (S) = +.66;HCD (O) = –.66
NME bold; ME asterisked NME bold; ME asterisked NME bold; ME asterisked
D E D E D E
D 4,3 1,4 D 4,3 1,4 D 4,3 1,4
E 2,1 3,2* E 2.1 3.2* E 3,2 2,1
S: (+,=) S: (+,=) S: (+,>)
O: (–,=) O: (–,=) O: (–,<)

FIGURE 4.4 Barack Obama’s VICS Indices and Subjective Games

 Quantum Politics and Operational Code Analysis 75

Index in the third Issue Area of Nuclear Weapons Proliferation and less than his
Strategic Orientation Index in the other two Issue Areas.
The combinations of Strategic Orientation and Historical Control scores are
the same for Other in two Issue Areas while the Historical Control index is lower
than the Strategic Orientation Index in the Nuclear Weapons Proliferation Issue
Area. The reversal of differences in the Historical Control Index for Nuclear
Weapons Proliferation indicates a power asymmetry between Self and Other and
a different subjective game in this Issue Area than in the other two Issue Areas
that focus on Russian-American relations and Russian nuclear weapons. Obama’s
Strategic Orientation of cooperation makes a Nonmyopic Equilibrium (NME) of
settlement (D,D = 4,3) possible in all three subjective games. However, Other’s
Strategic Orientation of conflict in all three games makes deadlock (E,E) possible
as a Myopic Equilibrium (ME) in the two symmetrical power games and requires
Obama to at least threaten to move to deadlock in the asymmetrical power
game as well as in the two symmetrical power games, in order for settlement
(D,D = 4,3) to be a stable equilibrium in repeated plays of these games.
A complete profile of Obama’s operational code belief system is in Table 4.1.
The VICS indices for the answers to George’s (1969) ten questions about a lead-
er’s philosophical and instrumental beliefs are shown along with the formula for
calculating each index. A discussion of the logic for each index is in Schafer and
Walker (2006). Here it is necessary only to note that they are all based on the
probabilities expressed as percentages for exercising one of the six forms of power
by Self and Other in the text of a leader’s statement and tallied in Figure 4.4. The
interpretation of the VICS indices in this speech is in terms of the range of the
values for each index, which range between .00 and 1.0 with the exception of P-
1, P-2, I-1, and I-2, which range from –1.0 to +1.0. The lexicon of anchors for
each range is .00 Very Low, .25 Low, .50 Medium, .75 High, 1.0 High for the
former indices and –1.0 Extremely, –.75 Very, –.50 Definitely, –.25 Somewhat,
0.0 Mixed, +.25 Somewhat, +.50 Definitely, +.75 Very, +1.0 Extremely for the
latter indices (see Walker, Schafer, and Young 2003).
Applying this vocabulary to the scores for Obama’s speech generates the fol-
lowing operational code profile (Walker et al. 2003):

Based on his philosophical beliefs in this speech, Barack Obama will have a
propensity to diagnose the political universe as definitely hostile, be definitely
pessimistic regarding the realization of political values, view the predictabil-
ity of the political future as low, believe that he has a high level of historical
control, and attribute a high role to chance.
Based on his instrumental beliefs in his speech, Barack Obama will have
a propensity to choose a very cooperative strategy in the political universe and
implement it with somewhat cooperative tactics. He will be very acceptant in his
orientation toward risk and will manage risk by being somewhat flexible in
his propensity to shift between cooperation and conflict and low in flexibility
76 Stephen G. Walker

TABLE 4.1 Barack Obama’s Operational Code Profile.

Philosophical Beliefs
P-1. Nature of Political Universe (friendly +1.0/hostile –1.0)
Index =(%Positive Other) minus (%Negative Other)
–.60 [.00 + .00 + .20] – [.27 + .13 + .40] = –.60

P-2. Realization of Political Values (optimist+1.0/pessimist –1.0)

Index = (Mean Intensity of Other ÷ 3)
–.51 [0+0+3–26]/15 = –1.53; 1.53/3 = –.51

P-3. Predictability of Future (uncertain .00/certain 1.0)

Index = (Md – .166) ÷ .834
.29 [.40 – .166] ÷ .834 = .234; .234/.834 = .29

P-4. Historical Control (none .00 to total +1.0)

a. Self Index = (Self Total ÷ Total N)
.64 P4a = [27/42] = .64; P4d = .64 – .36 = +.28
b. Other Index = (Other Total ÷ Total N)
.36 P4b = (15/42) = .36; P4d = .36 – .64 = –.28

P-5. Role of Chance (low .00 /high+1.0)

Index = 1 – (Political Future)(Self Historical Control)
.81 1 – [(.29)(.64)]= 1 – .19 = .81

Instrumental Beliefs
I-1. Approach to Goals (Strategy)(cooperate +1.0/conflict –1.0)
Index =(%Positive Self) minus (%Negative Self)
+.64 [.04 + .04 + .74] – [.18 + .00 +.00] = +.64

I-2. Pursuit of Goals (Tactics) (cooperate +1.0/conflict –1.0)

Index = (Mean Intensity of Self ÷ 3)
+.25 [3+2+20–5]/27 = +.74/3 = +.25

I-3. Risk Orientation (risk averse .00)/risk acceptant +1.0)

Index = (Md – .166) ÷ .834
.69 [.74 –.166] = .574; .574 ÷ .834 = .69

I-4. Timing of Action (low flexibility .00/high flexibility 1.0)

a. Coop v. Conf Index = 1 – Abs. Value [%Coop Self – %Conf Self]
.36 1 – |[.82 – .18]| = 1– |.64| = .36
b. Word v. Deed Index = 1 – Abs. Value [%Word Self – %Deed Self]
–.28 1 – |[.04 + .74 + .18 + .00 – .04 –.00]| = 1– |.86| = .14

I-5. Utility of Means {infrequent exercise .00/frequent exercise 1.0}

Index = %Self for each category of the exercise of power
a. Rew = .04 b.Pro = .04 c.App/Supp = .74 d.Opp/Res = .18 e.Thr = .00 f.Pun =.00
 Quantum Politics and Operational Code Analysis 77

regarding his propensity to shift between words and deeds in executing his
strategy and tactics. He has a very low propensity to choose either rewards or
punishments, a very low propensity either to choose promises or threats, a high
propensity to choose appeal and support statements, and a low propensity to
choose oppose and resist statements regarding the issues in his speech.
If this speech by Obama is representative of his general leadership style, then it is
one in which Self has a propensity for pragmatic cooperation characterized by Deter
(DEE) and Reward (DDD) tactics in a political universe in which Other has a
propensity for pragmatic conflict characterized by Compel (EDD) and Punish (EEE)
tactics. This leadership style extends across two of the three Issue Areas (Russian
Nuclear Weapons and U.S.–Russian Relations) in this speech with the excep-
tion being the Nuclear Weapons Proliferation Issue. In this Issue Area, Self’s
leadership style is one of dogmatic cooperation with a tendency toward Exploitation
(DDE) tactics against Others with a leadership style of erratic conflict and Bluff
(EED) tactics.
Both Obama’s general leadership style and his leadership style for the Nuclear
Proliferation Issue are mapped on the operational code field diagram in Figure 4.5

Cooperation
P1/I1(+1.0)
(+,=) Settlement (+,=)
Pragmatic Pragmatic
Cooperation Cooperation
Reward Deter
(DDD) (DEE)

S (+.64, +.28) S (+.60, +.66)

* *
(+,<) (+,>)
Erratic Dogmatic
Cooperation Cooperation
Appease Exploit
(–1.0) (DED) (DDE) (+1.0)
P4d P4d
Submission * Domination
Low Control O (–.33, –.66) High Control
(–,>)
(–,<) Dogmatic
Erratic * Conflict
Conflict O (–.60, –28) Bully
Bluff (EDE)
(EED) (–,=) (–,=)
Pragmatic Pragmatic
Conflict Conflict
Compel Punish
(EDD) (EEE)
Deadlock
P1/I1 (–1.0)
Conflict

FIGURE 4.5 Obama’s Operational Code Beliefs and Types of Leadership Styles
78 Stephen G. Walker

with the coordinates provided by his (S: I-1, P4d) and (O: P-1, P4d) indices from
Table 4.1. With these locations and the information in Figure 4.2, it is possible
to infer from Obama’s leadership style what his subjective game is and his likely
choices of “move” or “stay” from an “initial state” in more than one game. It is
possible as well with the same methods to map the operational codes of two or
more different leaders, compare their leadership styles and the definitions of their
strategic situations in their respective subjective games, and then make predictions
about what choices (move or stay) they will make and/or expect other players to
choose from an “initial state” of play.

Conclusion
In Part I of this book we have offered a theoretical perspective that amalgamates
the theoretical study of the fields of International Relations and Foreign Policy
Analysis under a common principle of rationality. The substantive rationality
theorem in the study of IR and the bounded rationality theorem in the study
of FPA are based on the different levels of analysis that characterize these two
fields of politics. International Relations as a field of macropolitics is the study of
power relations in the international system and the states that constitute its subsys-
tems. Foreign Policy Analysis as a field of micropolitics is the study of particular
states, dyads, and triads within the international system and the individuals and
small groups that constitute the leaders of these states. The level of analysis that
spans the two fields of study is the strategic dyad, a two-body problem that can
be contracted to a one-body problem and expanded to a three-body or n-body
problem.
Metaphorically, we have applied important concepts from theoretical physics
to guide our application of rationality as a unifying theoretical principle across IR
and FPA and to argue that the micro level of analysis takes theoretical precedence
over the macro level in the study of politics as well as physics. This demonstration
has proceeded historically by first tracing the evolution of theoretical physics from
the classical mechanics of large-scale particles of matter to the atomic mechanics of
small-scale particles of energy and the quantum mechanics of subatomic particles.
Then we have drawn parallels with the theoretical study of world politics from
the classical billiard ball mechanics of power politics between large-scale states to
the neobehavioral mechanics of strategic interaction between dyads and triads and
the psychological dynamics of decision making by leaders and small groups.
We have assumed that the objective in both endeavors is the one identified by
physicist Richard Feynman for an understanding of the macrofoundations and
microfoundations of basic physics: “the aim is to see complete nature as different
aspects of one set of phenomena. That is the problem in basic theoretical phys-
ics today—to find the laws behind experiment; to amalgamate these classes” (Feynman
1995: 26; italics his). Similarly, the aim in understanding basic theoretical politics
is to see it in its various manifestations as different aspects of one set of phenom-
 Quantum Politics and Operational Code Analysis 79

ena—to identify the rules that govern IR and FPA and amalgamate these two
classes of phenomena.
Our approach to the study of politics with this goal in mind has been to define
this task first as an “ill-structured” problem, one represented in our discussion
as characterized by entropy, i.e., a universe of possible random states of behav-
ior in the world of events and possible random states of mind in the world of
beliefs. Then we amalgamated these two worlds, transforming them into a “well-
structured” problem by reducing their entropy to a manageable number of states
predicted and unified by appropriate applications of the rationality theorem.
Newton’s efforts in physics at the macro level of analysis resulted in the for-
mulation of various physical laws governing the behavior of such large-scale
material particles as planets. Einstein and others, such as Planck, Heisenberg, and
Schrödinger, later subsumed micro and macro phenomena under the same laws.
One example is Einstein’s famous equation E = mc², where E is energy, m is
mass, and c is the constant speed of light, from his theory of relativity for explain-
ing the conservation of energy by such small-scale energy particles as electrons
or neutrons. This formula represented the unification of the explanations for the
behavior of matter and energy, based on the realization that matter is just a form
of energy (Feynman 1995).
Similarly, the argument in this book is that the laws of state behavior from clas-
sical IR theory governing the exercise of power between Ego and Alter as large-
scale political objects can also be subsumed under a theory of relativity: E|A = S
(E|A) |O (E|A), i.e., the macro-level behavior of large-scale objects (Ego and
Alter) can be explained by an operational code analysis of the micro-level behavior
of small-scale leaders (Self and Other) within each state. The E|A game, defined
and explained in space by a macro-theoretical, realist theory of the distribution of
power between two states in a strategic dyad, is solved over time by the intersec-
tion of the subjective games [S(E|A)|O(E|A)] of Self and Other as the leaders
of the two states. As in Einstein’s synthesis of Newton’s classical laws of physics,
there is an element of relativity in our synthesis of the classical laws of power
politics symbolized by Lenin’s Question: Kto-kovo (Who (controls) whom)? The
definition of the game and the answer to Lenin is “relative,” depending on which
subjective game is the vantage point for making predictions.
The billiard ball model of IR in macro (state-oriented) theories of Foreign
Policy Analysis explains political behavior in International Relations by reference
to an absolute principle of substantive rationality grounded in the determinate
properties of the distribution of power between states. The behavioral model
of IR in micro (leader-oriented) theories of Foreign Policy Analysis explains
political behavior in International Relations by reference to a relative principle
of bounded rationality grounded in the ultimately indeterminate properties of
leaders’ beliefs about the distribution of power between states. A final parallel
between the study of physics and politics is the importance of a central concept
to illuminate and unify our understanding of phenomena at the macro and micro
80 Stephen G. Walker

levels of each discipline. In both macro and micro analyses of politics, the concept
of power is conceptualized as the exercise of positive and negative sanctions in the
representation and solution of one-body, two-body, and n-body problems. The
concept of energy plays the same dual role in macro and micro studies of physics
(Feynman 1995: 69–86).
Operational code analysis as a neobehavioral approach builds on the methods
and substantive concerns of the behavioral revolution in political science; how-
ever, it also amalgamates new elements as well from the rational choice tradition
and applies them to the microfoundations of foreign policy analysis through the
use of game theory and the concepts of power and rationality (Walker 2003,
2007a). In the remainder of this book exchange theory, role theory, prospect
theory, and game theory are employed in conjunction with the rationality theo-
rem as a constant to model patterns of continuity and change in the foreign policy
choices and strategic interactions of leaders and small groups of decision makers.
The modeling and measuring methods employed echo in several respects older
applications by founders of the behavioral revolution in the study of world politics
to the structure-oriented, macrofoundations of states-as-actors in International
Relations, including the use of models of communication from information
theory and systems theory (Kaplan 1957; Rosecrance 1963; McClelland 1966;
Wright 1942; Rummel 1975–1981). However, the focus in this book is on the
application of these methods to the agent-centered, microfoundations of indi-
viduals-as-actors in Foreign Policy Analysis. It extends an effort begun by other
behavioral IR scholars who focused their efforts on decision making under stress
in international crises and conflicts (Holsti et al. 1964; Hermann 1969; Holsti
1972; Zinnes 1976), organizational and bureaucratic decision-making processes
(Snyder et al. 1954; Allison 1969); elite perceptions and other psychological pro-
cesses (Steinbrunner 1974; Jervis 1976).
The empirical analyses in the following chapters represent applications of single-
case, structured-focused comparison, repeated measures, and quasi-experimental
research designs with the three basic methodological strategies recommended by
Feynman: isolating simple situations, deriving specific rules, and employing rough
approximations to understand the general rules that explain the exercise of power
in world politics. These tests are not laboratory experiments under controlled
conditions, but they are analyses of samples collected in the natural world of
politics from relevant time periods regarding illustrative issues of foreign policy.
They are presented as a series of refinements and improvements to demonstrate
Feynman’s (1995: 25. Italics his) method of “rough approximation ... [a] way to tell
whether our ideas are right [that] is relatively crude but probably the most power-
ful of them all.” This methodological philosophy acknowledges the impossibility
of making exact predictions of individual events as a goal and embraces instead
the more modest goal of understanding some basic rules of politics as they apply
approximately to the empirical dynamics of foreign policy analysis.
PART II

Foreign Policy Decision

Making
5
THE UNITED STATES AND ROGUE
LEADERS
Understanding the Conflicts*

Akan Malici

Introduction
When the cold war came to an end almost two decades ago, scholars contem-
plated that we might soon miss it (Mearsheimer 1990). The reason for such a
counterintuitive feeling is simple: with the move from bipolarity to unipolarity,
security threats no longer emanate from the rivalry of two superpowers playing
according to well-established rules, but from defiant and often relatively small
rogue states. Rogue states are said (or partly known) to sponsor or practice inter-
national terrorism and to engage in the acquisition and proliferation of weapons
of mass destruction (Tanter 1998). Their leaders are said to be genuinely belliger-
ent and hostile, and sometimes they are even described as crazy (Krauthammer
2005; Thornhill and Ward 2002).
Scholars, security analysts, and observers of international politics have two
fundamental options in thinking about these allegedly defiant leaders. The first
option is to join the conventional wisdom. Adopting this option is attractive
because of its apparent plausibility. However, there is also a danger associated
with this option. Judgments are often made on a purely descriptive basis without
sufficient effort towards critically asking why rogue leaders behave the way they
do. The simple descriptive analysis stands in the way of developing a deeper and
more appropriate understanding of these leaders. Merely asserting that they are
crazy or irrational is too simple and indeed wrong.
The second option is to leave the door open to the conventional wisdom while
simultaneously attempting to understand the crucial why question, i.e., why rogue
leaders behave the way they do. Understanding a leader means penetrating inside
his psychology. In an article well known to peace and conflict scholars, politi-
cal psychologist Philip Tetlock problematized the foreign policy decision-making
84 Akan Malici

process, asking, “What do we [psychologists] have to contribute?” (Tetlock 1986).

The obvious answer is “a lot,” and it follows from the recognition that politics is
an inherently social-psychological affair. A close attention and investigation of lead-
ers’ psychology, their subjective beliefs and perceptions, is, therefore an absolute
necessity.
This task is what I intend to do in this chapter: problematize the psychology of
four leaders who came to be seen as rogues, Kim Jong Il of North Korea, Fidel
Castro of Cuba, Mahmoud Ahmadinejad of Iran, and Bashar al-Asad of Syria. I
should emphasize that this chapter is not to be understood as a current policy dis-
cussion. Fidel Castro is no longer the official ruler of Cuba; Kim Jong Il is report-
edly ailing and it is questionable how much longer he will be the ruler of North
Korea’s affairs. My focus in examining these leaders is a larger one: Is there really
something like a “typical” rogue leader?
In answering this research question my overall goal is also to develop what the
preeminent peace researcher Ralph White (1984: 160) labeled “realistic empa-
thy” towards rogue states. As I seek to empathize with rogue leaders, I share his
argument that realistic empathy stands as “the great corrective for all forms of
war- provoking misperceptions.” Despite an abundance of (often ill-informed)
commentary on these leaders, little systematic effort has been undertaken to
investigate rogue leaders. By addressing this gap, I aspire to contribute to the
identification of alternative paths that may lead to a peaceful resolution of future
crises and conflicts.
I begin with a description of why rogue states are seen as a threat by the United
States. Next, I make a case for examining rogue leaders in more detail, especially
their belief systems. Then I examine their beliefs and subsequently discuss gen-
eral strategies that may prove effective in dealing with rogue leaders. I close this
chapter by highlighting a few critical insights about the U.S. relationship with
rogue states.

The Challenges of Rogues

Rogue states are considered to be a security threat mainly for two reasons. First,
they allegedly engage in the sponsorship of international terrorism. Each year
the U.S. State Department publishes a list of “State Sponsors of Terrorism.”
According to the State Department, “state sponsors of terrorism provide criti-
cal support to non-state terrorist groups. Without state sponsors, terrorist groups
would have greater difficulty obtaining the funds, weapons, materials, and secure
areas they require to plan and conduct operations.”1 Second, rogue states are
said to engage in the production or proliferation of weapons of mass destruction
(WMD). The fear here is not only that rogue states would deploy these weapons,
but also that they could be passed on to terrorist groups.
North Korea was put on the State Department’s list in 1988. After two decades
it was removed because it had not engaged in any further evident sponsorship since
 The United States and Rogue Leaders 85

the bombing of a Korea Airlines flight in 1987. However, very serious concerns
remain about Pyongyang’s nuclear ambitions. Beginning in the winter of 2002–
2003 the world witnessed the North Korean nuclear revelation. In December,
North Korea disabled UN monitoring cameras at the Yongbyon nuclear facility,
dismissed nuclear inspectors, and restarted its nuclear fuel processing laboratory.
In January, North Korea withdrew from the Nuclear Non-Proliferation treaty
(NPT), and in February it restarted its nuclear reactor. Two years later Pyongyang
announced that it had nuclear weapons. Since then, relations have remained very
tense. Although North Korea demolished part of its Yongbyon nuclear reactor in
2008, more recently is has once again restarted its nuclear program and engaged
in several missile tests arousing worldwide concern.
Cuba has been featured on the State Sponsors of Terrorism list since 1982. The
State Department does acknowledge that the Havana regime no longer actively
supports armed struggle in Latin America and other parts of the world. It also
acknowledges that there is no evidence of terrorist-related money laundering or
terrorist financing activities. However, Cuba remains on the list because of allega-
tions that the Havana regime continues providing safe haven to several terrorists
such as Basque Homeland and Freedom (ETA), the Revolutionary Armed Forces
of Columbia (FARC), and the Colombian National Liberation Army (ELN).
Cuba possesses no WMD capabilities and there is also no evidence for its pursuit
of them. Of all the present rogue states, Cuba is likely the one posing the least
threat.
Iran has been on the terrorism list since 1984, and according to the State
Department, it is “the most active sponsor of terrorism.” Iran’s alleged actions
are considered to be particularly destabilizing. According to the State Department
(2004), “Iran’s involvement in the planning and financial support of terror-
ist attacks throughout the Middle East Europe, and Central Asia had a direct
impact on international efforts to promote peace, threatened economic stability
in the Gulf, and undermined the growth of democracy.” Iran’s nuclear ambitions
have also led to international concern for some years and reached a new level
beginning in August 2002. An investigation by the International Atomic Energy
Agency (IAEA) revealed pressing questions about Iran’s nuclear program, includ-
ing the discovery of an undeclared uranium enrichment program. Since then,
the regime in Tehran has been intransigent, arguing that it has a right to peaceful
nuclear technology.
Syria has been on the State Department’s list since its inception in 1979. In
2003 the State Department noted, “Syria provided political and material support
to Hizballah and allowed Iran to use Syrian territory as a transit point for assistance
to Hizballah.” Syria is also accused of providing external bases to the leadership
of HAMAS, Palestine Islamic Jihad (PIJ), the Popular Front for the Liberation of
Palestine (PLFP), and the Popular Front for the Liberation of Palestine-General
Command (PFLP-GC). Regarding WMD, Washington views the country as
having the most advanced chemical weapons capabilities in the Middle East
86 Akan Malici

and that it is “without question among those states most aggressively seeking to
acquire or develop WMD and the means of delivery” (quoted in Kerr 2003).
However, in the case of neither Iran nor Syria is there concrete evidence that a
nuclear weapons capability is indeed being pursued (Leverett 2005: 13).

Understanding Rogues
Scholars working in the foreign policy decision-making tradition have long
argued that in order to understand the foreign policy behaviors of leaders one
must concentrate on their “psychological milieu” and their “attitudinal prism”
(Sprout and Sprout 1956; Brecher et al. 1969). These psychological areas of
human existence are of enormous importance because it is here where cognitive
distortions, motivational biases, and subjective beliefs for subsequent action are
situated (Tetlock 1998). These biases and beliefs lend subjective legitimacy to a
leader and his foreign policy actions. It is important to recognize and take them
into account in the analytical effort regardless of how illegitimate these subjec-
tive beliefs appear to an outside observer. In order to understand the cognitive
processes of leaders and how and why these processes compel them to certain
actions, it is important to engage in “realistic empathy” (White 1984, 1991).
White (1991: 292) defines empathy as:

[S]imply understanding how a situation looks like to another person (or

group). It does not necessarily imply sympathy, or tolerance, or liking, or
agreement with the person—but simply understanding. In many contexts
the word “understanding” can be substituted for empathy, but empathy
implies especially a focus on the other’s situation—trying to look out at his
situation through his eyes rather than at him as an individual.

Thus, the task in thinking about rogue leaders is not to proceed deductively as is
commonly done. More specifically, one should avoid drawing absolute and firm
conclusions about a leader’s personality dispositions on the basis of his behavior.
The result of such reasoning tends to condemn the situation into deterministic
hopelessness. If a leader acts aggressively because he is genuinely hostile, then
short of containing and isolating such a leader, nothing can be done to manage
the threat more effectively. However, such an analytic strategy is unproductive.
It only reifies the conflict and raises the danger of further escalation. The recent
experiences of the United States with rogue states offer convincing testimony
to this assertion.
For the sake of attempting more peaceful or at least more stable international
relations, it might be more productive to proceed more cautiously when thinking
about rogue leaders. One must of course acknowledge the aggressive behavior
of rogue leaders. However, at the same time one should not assume a priori
that this behavior results necessarily from a genuine hostile predisposition of the
 The United States and Rogue Leaders 87

rogue leader. Instead, one must examine a variety of potential factors that may
lead to aggressive behavior. This analytical strategy maximizes realistic empathy
and avoids what psychologists call the “fundamental attribution error,” which is
to assume that the behavior of others (especially enemies) is the product of their
personal disposition while our own behavior is the product of our circumstances
(Tetlock 1998: 877).

The Beliefs of Rogues

The examination of an opponent’s beliefs and worldviews and the potential
bases for them are important steps towards understanding crises and conflicts
with rogue states and subsequently towards devising an appropriate strategy.
In order to assess the beliefs of the alleged rogue leaders discussed here, I use
operational code analysis and the Verbs in Context System (VICS) as elaborated
in Chapter 4 of this volume.2 For the analyst to make substantive inferences, it
is also instructive to compare alleged rogue leaders to a norming group. This
norming group consists of a collection of 164 speech acts given by thirty different
and diverse world leaders—leaders of poor and rich states and weak and strong
states.3 This step provides a broader perspective, which is informative because it
permits us to develop a better sense of how the leaders examined here—leaders
that are portrayed as “rogues”—compare to the “average leader.” The results of
this analysis are in Table 5.1.
Because of the similarity of the belief systems of Fidel Castro and Kim Jong Il
on the one hand and Mahmoud Ahmadinejad and Bashar al-Asad on the other,
we shall analyze separately each of these pairs of leaders. The results from our
statistical analysis in Table 5.1 suggest that there is nothing average about Fidel
Castro and Kim Jong Il. Indeed, they are far more conflictual than the norming
group on a number of measures. As their philosophical beliefs indicate, both have
a negative assessment of the international environment and their place within
it. Moreover, their instrumental beliefs indicate conflictual dispositions towards
other actors in this environment.
However, a closer analysis reveals some deeper and interesting insights. A
person’s hostile images of the political environment (P-1) and the perception
that the fate of these goals is in the hands of others (P-4) are key manifestations of
the frustration-aggression syndrome.4 Pessimism about the realization of political
goals (P-2), in addition to the first two characteristics, can be another sign.
Psychologists have long argued that frustration is a main source for a person being
conflictual and aggressive. At the early stages of research into this syndrome, it was
argued that “the existence of frustration always leads to some form of aggression”
(Dollard 1939: 338). Subsequently, the linearity of this assertion was revisited,
and scholars concluded that other behaviors than aggression are possible as well
(Miller et al. 1941). However, if frustration is paired with an inclination towards
hostile behavior (I-1, I-2), as it is in the case for both Fidel Castro and Kim Jong
TABLE 5.1 The Beliefs of Rogue Leaders

Philosophical and Instrumental Fidel Castro Kim Jong Il Ahmadinejad al-Asad Norming Group
Beliefs (n = 8) (n = 14) (n = 13) (n = 13) (n = 164)

P-1. Nature of Political Universe 0.074** 0.163* 0.076** 0.137** 0.301

(Conflict/Cooperation)
P-2. Realization of Political Values –0.057** 0.079 –0.029** 0.026** 0.147
(Pessimism/Optimism)
P-3 Political Future 0.142 0.133 0.105 0.088** 0.134
(Unpredictable/Predictable)
P-4. Historical Development 0.150* 0.148** 0.181 0.223 0.224
(Low Control/High Control)
P-5. Role of Chance 0.978 0.986* 0.980 0.980 0.968
(Small Role/Large Role)
I-1. Strategic Approach to Goals
(Conflict/Cooperation) –0.084** –0.006** 0.418 0.471 0.401
I-2. Intensity of Tactics
(Conflict/Cooperation) –0.140** –0.084** 0.146 0.232 0.178
I-3. Risk Orientation
(Averse/Acceptant) 0.299 0.224 0.342 0.244 0.332
I-4. Timing of Action
a. Conflict/Cooperation 0.486 0.583 0.423 0.518 0.503
b. Words/Deeds 0.557 0.647** 0.416 0.576 0.464
I-5. Utility of Means
a. Reward 0.133 0.146 0.117 0.200 0.157
b. Promise 0.096 0.052 0.036 0.025 0.075
c. Appeal/Support 0.229** 0.297** 0.556 0.510 0.468
d. Oppose/Resist 0.056* 0.172 0.131 0.153 0.154
e. Threaten 0.277** 0.078** 0.069** 0.024 0.034
f. Punish 0.210* 0.255** 0.091 0.088 0.112

Significant differences between leader and norming group indices are asterisked at the following levels: *p < .10, **p < .05 (two-tailed test). For a detailed account of data
sources and samples see Malici and Malici (2005b) and Malici and Buckner (2008).
 The United States and Rogue Leaders 89

Il, the potential for aggression is more acute. In short, our analysis so far suggests
an important insight, namely that Fidel Castro and Kim Jong Il may be conflictual
not because they simply have a genuinely hostile personality, but because they
are frustrated.
In contrast, a different analysis applies to Ahmadinejad and al-Asad. The belief
systems of these leaders show no indication of frustration, and overall, the analysis
shows that there are fewer differences between them and the norming group.
Both Ahmadinejad and al-Asad perceive the nature of the political universe (P-
1) to be more conflictual than the average leader. They are also very pessimistic
about the realization of his political goals (P-2). In addition, when compared
to the average leader, al-Asad views the political future as unpredictable (P-3).
However, in contrast to Fidel Castro and Kim Jong Il, neither Ahmadinejad nor
al-Asad views the fate of their goals to be in the hands of others (P-4). Regarding
their instrumental beliefs, our analysis does show that Ahmadinejad is more likely
to use Threats (I-5e) than the average leader. However, contrary to the conven-
tional wisdom about Ahmadinejad and al-Asad, they are not especially inclined
strategically (I-1) nor tactically (I-2) towards aggressive and hostile behavior.

Empathizing with Rogues

My intention is to empathize with rogue leaders. Therefore, I shall now problem-
atize their threat perceptions further. All four leaders clearly perceive the United
States to be conflictual. The results of the belief system analyses have implications
for possible strategies of the United States towards rogue leaders. I will return to
these possibilities below, as I discuss alternative strategies that the United States
may employ in dealing with the leaders in Table 5.1, following a brief overview
of how each rogue state views its conflict with the United States.
The Yankee colossus in the North has long posed a major threat to Cuba’s
survival from the perspective of the Cuban regime. When Fidel Castro came to
power in 1959, there is good evidence that he sought an amenable relationship
with the United States. However, he also pursued economic independence from
the United States. Leaders in Washington saw this goal as an affront against U.S.
businesses in Cuba and viewed U.S. economic interests as threatened. It was then
quickly decided that Fidel Castro’s regime must be overthrown through covert
military operations. This plan resulted in the Bay of Pigs fiasco in 1961. According
to critics of U.S. foreign policy, it was then only reasonable for Cuba to seek an
alliance with the Soviet Union as the other superpower. Thereafter, the United
States imposed severe economic and political sanctions and even an embargo on
Cuba and continued to harbor plans for the overthrow of the regime.
The North Korea regime also considers the United States to be a major threat
to its survival. It was the United States that created an artificial separation of the
peninsula along the 38th parallel in 1948. The subsequent Korean War led to
a direct clash between North Korea and the United States and relations since
90 Akan Malici

the armistice in 1953 have remained conflictual. North Korea’s strategic situa-
tion worsened considerably when it lost the Soviet Union as a political and eco-
nomic ally in the late 1980s, and the traditional loyalties of China have also been
weakened considerably. Pyongyang now faces by itself its archenemies South
Korea and Japan, and both are politically (and militarily) supported by U.S. forces.
Indeed, from the perspective of Pyongyang the threat takes on explicit forms in
the yearly joint military maneuvers between the United States and South Korea.
The Iranian leadership’s hostile perceptions of the United States date back to
the 1950s. The United States feared that Iran would ally with the Soviet Union
and also restrict American and British control of Middle Eastern oil. The CIA
orchestrated a coup that led to the removal of democratically elected Prime
Minister Mohammed Mossadegh in 1953 and the establishment of the Shah.
In 1979, the Islamic Revolution ended the Shah’s repressive rule and fostered
Ayatollah Khomeni as the new leader. Viewing regional interests jeopardized
again, exacerbated by the hostage crisis with Iran, the United States imposed
sanctions. The resulting diplomatic isolation of Iran and its compounding eco-
nomic problems made the country so vulnerable that Saddam Hussein seized the
opportunity to strike militarily at what appeared to be an easy target. During the
following eight years of war, the United States tried to weaken Tehran further
by providing arms and economic resources to pro-Western Iranians opposed to
Khomeni (Farhang 1993).
American fears of a “communist outpost” also existed with regards to Syria dur-
ing the cold war. In 1957, the United States attempted a coup in Damascus. When
it failed, the U.S. leadership seriously contemplated direct military action against
Syria. However, Washington ultimately refrained because of the lack of support
from Saudi Arabia and Iraq, which it considered necessary to avoid a regional
upheaval (Lesch 1996: 134–137). Since 1979, the United States has imposed sanc-
tions on Syria which were reinforced in 2003 with the Syria Accountability Act,
a package of sanctions that calls for the adoption of new political and economic
punitive measures in order to coerce the Damascus regime into a more compla-
cent disposition towards the United States (Leverett 2005: 17).
How the United States perceives rogue states is well known. This brief over-
view has given a glimpse of how the situation looks like from their perspective.
Generally, those policymakers and international security analysts who apply the
rogue label to these states rarely make an effort to imagine how international
politics are perceived from the perspectives of rogue leaders, or consider how
these perceptions are part of an interactive dynamic crisis in which the United
States may be implicated as deeply as the regimes in Havana, Pyongyang, Tehran,
or Damascus. No doubt over the years Cuba, North Korea, Iran, and Syria have
engaged in actions that are justifiably perceived as threatening by the United
States. At the same time, it is important to realize the mutuality of the crisis.
Doing so with realistic empathy is very important, for it increases the possibility
of peaceful conflict resolution.
 The United States and Rogue Leaders 91

Strategies for the United States

So far, I have engaged in what may be called psychological empathy. I have exam-
ined the beliefs of four alleged rogue leaders and I have also asked about possible
reasons for their beliefs. Apart from this psychological empathy, White (1991:
293) also emphasizes the importance of what may be referred to as behavioral
empathy by reference to a chess game. The most fundamental question here is: “If
I do this, he does that, and I do such-and-such, what would he do then?” Political
action, in other words, must be strategic because “each actor’s ability to further
its ends depends on how other actors behave, and therefore each actor must take
the actions of others into account” (Lake and Powell 1999b: 3). Psychological and
behavioral empathy together may help create better-informed strategic consider-
ations of the United States as it faces rogue leaders.
The strategy for addressing questions about behavioral empathy is to utilize the
Theory of Inferences about Preferences (TIP), in which three key beliefs of the
operational code construct are relevant. The first key belief is a leader’s image of
the political universe (P-1). What are the motivations of other actors? Another
key belief is the belief in the ability to control historical development (P-4a) ver-
sus the degree of control attributed to other actors (P-4b). The final key belief
concerns the strategy a leader considers to be effective and appropriate for achiev-
ing political goals (I-1).
The signs in the following TIP propositions indicate whether a particular
leader’s I-1 and P-1 indices are above (+) or below (–) the mean for a norming
group of world leaders while the signs for the P-4 indices indicate whether
they are greater than (>), less than (<), or within (=) one standard deviation of
the mean for the norming group.5 The six TIP propositions specify preference
rankings associated with strategies in several well-known 2 × 2 ordinal games
(Snyder and Diesing 1977).
(1) Appeasement Strategy: If (I-1, P-4a) or (P-1, P-4b) is (+, <), then Settle >
Deadlock > Submit > Dominate.
(2) Assurance Strategy: If (I-1, P-4a) or (P-1, P-4b) is (+, =), then Settle >
Deadlock > Dominate > Submit.
(3) Stag Hunt Strategy: If (I-1, P-4a) or (P-1, P-4b) is (+, >), then Settle >
Dominate > Deadlock > Submit.
(4) Chicken Strategy: If (I-1, P-4a) or (P-1, P-4b) is (–, <), then Dominate >
Settle > Submit > Deadlock.
(5) Prisoner’s Dilemma Strategy: If (I-1, P-4a) or (P-1, P-4b) is (–, =), then
Dominate > Settle > Deadlock > Submit.
(6) Bully Strategy: If (I-1, P-4a) or (P-1, P-4b) is (–, >), then Dominate >
Deadlock > Settle > Submit.
The intersecting preferences for Self (I-1, P-4a) and Other (P-1, P-4b) identify
formal models of “subjective games” in a leader’s operational code belief system
92 Akan Malici

United States
CO CF
CO 3,3 1,4

North Korea/Cuba

CF 4,1 2,2*

*Nash equilibrium is asterisked and non-myopic

equilibria are underlined.

FIGURE 5.1 Kim Jong Il’s and Fidel Castro’s Subjective Game

with the political outcomes of domination, submission, settlement, or deadlock

in world politics (Walker 1977; Leng and Walker 1982; Maoz1990; Snyder and
Diesing 1977; Schafer and Walker 2006).6
From the North Korean and Cuban perspective the relationship between them
and the United States is represented through a classic prisoners’ dilemma game in
Figure 5.1.7 This subjective game has mutual settlement (3,3) and mutual dead-
lock (2,2) as a nonmyopic equilibrium while mutual deadlock is also a Nash
equilibrium. In this game the dominant strategy for each side is a conflict strategy.
Historically, for the United States this strategy has meant the pursuit of contain-
ment and isolation strategies. Because these rogue states are likely to reciprocate
through the conflictual means at their disposal, perpetuation of the deadlock or,
worse, an escalation is to be expected.
In the interest of a peaceful resolution of the ensuing situation it is worthwhile
to problematize the strategic situation further in the context of Fidel Castro’s and
Kim Jong Il’s belief system. I noted above that these two leaders are frustrated,
a psychological condition constituted through a hostile worldview, a pessimistic
outlook, and a perceived lack of control. The strategy of containment has a very
fundamental effect on the frustration level of a leader. Examining this effect is
important in developing a better informed, more context-sensitive and, therefore,
more effective strategy towards rogue regimes (Malici 2007: 107).
The strategy of containment aims at dominating the opponent through means
such as coalition-building against the target state and employing political and eco-
nomic sanctions. Such a strategy, therefore, contributes to an increase in the target
leader’s hostile and pessimistic perception of the political environment. Because
containment is a strategy with domination as the highest-ranked outcome, the
target leader in a prisoner’s dilemma game may perceive a significantly lower level
of control—especially when a disproportionately strong state aims at isolating a
relatively small state. In the end, the U.S. strategy of containment and isolation
has had the effect of increasing the target leader’s level of frustration and along
with it his propensity to act aggressively towards the outside world, an outcome
no one desires (Malici 2007: 107–108).
 The United States and Rogue Leaders 93

I suggest a strategy that appears counterintuitive, namely a strategy of engage-

ment. The strategy of engagement is a contingent-cooperation strategy and the
practicing actor aims at a settlement with the opponent primarily through means
of diplomacy and incentives. In contrast to a strategy of containment, it contrib-
utes to a decrease in the target leader’s hostile and pessimistic perception of the
international environment. Furthermore, because the strategy is aimed at mutual
settlement, the target leader will also likely perceive an increased level of control
over ensuing events in international relations. Therefore, the strategy of engage-
ment has the effect of decreasing a leader’s level of frustration and perhaps the
propensity to act aggressively along with it (Malici 2007: 107–108).
Thus, I suggest that in this kind of situation the United States move to (1,4) in
the game in Figure 5.1. This move is counterintuitive because I seem to suggest
that the United States should engage an enemy state. My recommendation is based
on a belief system analysis and the result that these leaders are frustrated rather than
simply hostile. This distinction is crucial for making this suggestion. My suggestion
also seems counterintuitive because (1,4) is a state of Cuban and North Korean
domination over the United States. The strategic interaction between the two
states does, however, not stop here. In the event that the rogue state chooses “stay”
at (1,4), the next step for the United States would be to use threat power in a replay
of the game with (1,4) as an initial state (Brams 1994: 68–69). Because the United
States is disproportionately more powerful than either Cuba or North Korea, it
can threaten credibly in pre-play communication to move back to (2,2) unless they
move to mutual settlement (3,3).
Turning to U.S. relations with Iran and Syria, I find also here that a game-
theoretic illustration together with the belief system analyses of Ahmadinejad and
al-Asad provide interesting insights towards a peaceful resolution of the situation.
The subjective games of these leaders have a Nash equilibrium solution of (3,2)
and a nonmyopic equilibrium of (4,3). However, this latter equilibrium solu-
tion is difficult for the rogue state to reach unilaterally. With an initial state of
(3,2) deadlock there is no incentive for either Ahmadinejad or al-Asad to choose
“move” because a move to (1,4) leaves the United States no incentive to move
any further from this state. In contrast, a United States move to (2,1) provides an
opportunity for Iran or Syria to move to the nonmyopic equilibrium of (4,3).
Therefore, there is an incentive for the United States to move first and thereby
to induce the nonmyopic equilibrium solution. As Brams (1994: 28) explains,
“If it is rational for one player to move and the other player not to move from
the initial state, then the player who moves takes precedence: its move over-
rides the player who stays, so the outcome will be induced by the player who
moves.” However, this movement towards the nonmyopic equilibrium requires
the United States to correct its misperception of Iranian and Syrian preferences.
If they define U.S. relations with these states as a prisoner’s dilemma game, they
may assess the risk of moving from deadlock as too high. It takes an analysis of
their subjective game in Figure 5.2 to show that Ahmadinejad and al-Asad already
94 Akan Malici

United States

CO CF
CO 4,3 1,4

Iran/Syria

CF 2,1 3,2*

*Nash equilibrium is asterisked and

non-myopic equilibrium is underlined.

FIGURE 5.2 Ahmadinejad’s and al-Asad’s Subjective Game

subscribe to an Assurance Strategy, i.e., a strategy of conditional cooperation.

This game-theoretic analysis leads to the conclusion that in this kind of situa-
tion the U.S. leadership faces two alternative strategic options. The first option
is continuation of the deadlock by maintaining a non-cooperation strategy. The
second option is to take advantage of a window of opportunity when it presents
itself. The present analysis shows that such opportunities do occur. Achieving this
mutually beneficial outcome would require the United States to exercise realistic
empathy and depart from short-sighted strategies based on miscalculations about
the actual intentions of alleged rogue states.

Conclusion
In this chapter I have problematized and examined rogue leaders and their inten-
tions. For many observers terms such as “rogue leaders” or “rogue states” are
problematic. They are “failing to provide either a clear picture of who and what
constitutes a rogue state, or, perhaps more importantly, its ramifications on U.S.
policy action” (O’Reilly 2007: 296). This warning is to be taken seriously indeed.
The use of this terminology suggests that there is a sufficient degree of “sameness”
among rogues and this sameness, in turn, suggests a more or less common policy
prescription.
My analysis shows that such assumptions are fundamentally false. Aside from
the United States perceiving rogue leaders as a security threat, these leaders can
and do show significant differences. In short, there is no single typical rogue
leader. This conclusion carries important policy implications. Instead of pursu-
ing a general approach towards rogue states, the United States ought to consider
a more context-sensitive approach. As I problematized the psychology of rogue
leaders, one of my goals was to understand them better. However, this objective
has nothing to do with being soft. It has to do with safeguarding the national
security of the United States in effective ways. Matters of national security are
always confronted with the same question, namely how to ensure stability and
 The United States and Rogue Leaders 95

peace. The answers should not be guided merely by the intuition that rogue lead-
ers are genuinely hostile and must be treated accordingly. Intuition can be a poor
advisor, and it is my contention that it is indeed in the case of rogue leaders. The
answers to questions about national security are sometimes counterintuitive, and
yet, they encourage deeper thinking about methods of peaceful conflict resolu-
tion with rogue states.
I also want to highlight another critical conclusion. Conventional accounts
of U.S. crises and conflicts with rogue states often frame the situation as one
in which the sources of contentions originate overwhelmingly from them. It is
almost a given that U.S. decision makers, at least publicly, feel innocent of any
previous actions against countries such as Cuba, North Korea, Iran, or Syria, and
they are quick to discard any accusations as “paranoid nonsense or blatant, delib-
erately distorted propaganda” (White 1991: 295). Such feelings, as White (1991:
294) points out are “often based partly on sheer ignorance of history and partly on
rationalizing whatever our own country has done.”
It cannot be denied that rogue states have been a menace in the international
system and that they indeed have posed threats to U.S. interests. However, my
analysis has also shown that rogue leaders perceive U.S. actions towards their
countries as highly hostile and threatening. I have shown as well that these per-
ceptions are not baseless. Given the interactive nature of all foreign policy making,
any useful effort to comprehend the ongoing conflict has to include an analysis of
the way in which rogue regimes perceive the U.S. foreign policy towards their
countries. A more candid discussion than we had so far is necessary. It is the task
of scholars to carry this task forward into the public and political realm.

Notes
* Sections of this chapter were published earlier in the Journal of Peace Research, Vol. 45,
No. 6 (November 2008): 738–800 and Psicologia Politica, No. 31 (November 2005):
7–22. These sections are reproduced here by permission.
1 All references, unless otherwise noted, from the State Department can be found at
http://www.state.gov/.
2 The samples of coding materials were drawn from Lexis Nexis Academic Universe and
the Foreign Broadcast and Information Services. Public statements had to be approximately
1,500 words long to be coded. Where this number was not reached, I combined two
chronologically successive public statements into one. The samples for Fidel Castro
and Kim Jong Il contain speech acts from 1995–2003. The sample for Ahmadinejad
contains speech acts from 2005–2007, and the sample for al-Asad contains speech acts
from 2000–2007.
3 The data for the norming group were obtained from Professor Mark Schafer,
Department of Political Science, 240 Stubbs Hall, Louisiana State University, Baton
Rouge, LA 70803–5433. For more details on the norming group see Malici and Malici
(2005b) and Schafer and Walker (2006).
4 For discussions of the frustration-aggression syndrome see Berkowitz (1969, 1978,
1989); Schafer et al. (2006).
5 The mean values for the norming group are (P-1 = +.30, SD = .29; I-1 = +.40, SD =
.43; P-4 = .22, SD = .13).
96 Akan Malici

6 Maoz (1990: 423) gives a comprehensive definition of subjective games. He writes:

“The notion of subjective games is predicated on the way each actor perceives and
evaluates its environment. A subjective game consists of national perceptions of: (a)
the other players which are relevant for the analysis of a given strategic configuration,
(b) the alternatives available to oneself as well as the alternatives available to other
players, (c) the preference ordering of the outcomes that result from intersecting the
alternatives of all relevant players by the focal player, (d) the attribution of preference
orderings of outcomes to other players, (e) the attribution of choice principles to other
players, and (f) the selection of a strategy.”
7 A prisoner’s dilemma game assumes a roughly equal distribution of power between
the players, which would not be the case between the United States and these two
rogue states today. However, their alignment with the Soviet Union made the power
distribution symmetrical during the cold war. The acquisition of nuclear weapons by
North Korea lessened significantly the power asymmetry between Washington and
Pyongyang in the post-cold war period, making Cuba more vulnerable than North
Korea following the end of the cold war. However, the subjective game in the public
statements from both Havana and North Korea does not reflect the actual power
distribution between them and the United States. Their historical control scores (P-4)
are less than one standard deviation from the average world leader’s score, which is the
threshold for inferring significant power asymmetry.
6
DECEPTIVE BARGAINING AND
NUCLEAR AMBITIONS
Prospect Theory and North Korea’s
Decision to Go Nuclear

Kai He and Huiyun Feng

Introduction
If there is one place in the world where the cold war has not ended, it is the Korean
Peninsula. An armistice terminated the Korean War between North Korea and
the United States in 1953. Technically, North Korea is still at war with the United
States and South Korea since no peace treaty has been signed between the former
enemies. The end of the cold war did not bring peace to the peninsula. Instead, the
North Korean nuclear crises starting in the early 1990s have threatened regional
security for more than a decade. North Korea’s nuclear tests in 2006 and 2009 fur-
ther intensified antagonisms and hostilities between North Korea and the United
States. The stability of the Korean Peninsula in particular and Northeast Asia in
general is at stake.
Since the end of the cold war there have been two North Korean nuclear crises
in the Peninsula.1 The first nuclear crisis occurred in the early 1990s when North
Korea failed to implement the inspection agreements with the International
Atomic Energy Agency (IAEA) and threatened to withdraw from the Nuclear
Nonproliferation Treaty. While the United States considered military strikes
to stop North Korea’s nuclear program, the crisis ended peacefully after former
U.S. President Jimmy Carter visited North Korea. The United States signed the
Agreed Framework in Geneva, in which Pyongyang pledged to freeze its nuclear
weapons program in exchange for international aid to build two light water reac-
tors (see Sigel 1998; Snyder 1999; Harrison 2002).
The second North Korea nuclear crisis happened almost ten years after the first.
The interaction process in the second crisis was similar to the first one. North
Korea was accused of having a secret, highly enriched uranium (HEU) nuclear
program in 2002, which can produce nuclear weapons. It led the United States,
98 Kai He and Huiyun Feng

Japan, and South Korea to stop their oil supplies to North Korea. Then North
Korea started to resume its nuclear plutonium program in Yongbyon and later
conducted its first nuclear test in 2006. After the nuclear test North Korea went
back to the negotiation table and signed an agreement in the 2007 Six Party Talks
to dismantle its nuclear programs in return for financial aid and diplomatic con-
cessions from the United States.
The military tensions and crises in the Korean Peninsula are rooted in North
Korea’s decision to go nuclear in the late 1980s and the early 1990s. While most
research focuses on explaining why North Korea decided to go nuclear, little research
examines how North Korea pursued a nuclear-weapons program. North Korea did
not go nuclear over night. From the first North Korean nuclear crisis to North
Korea’s nuclear test in 2006, it took more than a decade for North Korea to possess
nuclear weapons. It is important to explain the motivation behind North Korea’s
nuclear ambition. However, it is equally important to examine the microfounda-
tions of North Korea’s foreign policy behavior, i.e., how North Korea’s leaders
decided to go nuclear. Understanding the “how” question helps us better explain
North Korea’s past policy and predict North Korea’s future behavior.
In this chapter we employ a prospect theory-sequential game model to exam-
ine North Korea’s strategic decisions to “go nuclear” from the late 1980s to the
1994 nuclear crisis. We suggest a “deceptive bargaining” model to explain North
Korean leaders’ foreign policy decisions in different domains of action at the end
of the cold war. By using operational code indices and a subjective game matrix,
we identify the strategies of North Korea’s leader Kim Il Sung, before and dur-
ing the first nuclear crisis. We argue that Kim Il Sung’s decision to go nuclear
in the late 1980s is neither crazy behavior nor a rational action. Instead, Kim Il
Sung made a risky decision in the early 1990s to go nuclear publicly in a domain
of losses after making the same decision privately in a domain of gains during his
attempt to conciliate with the United States and South Korea in the late 1980s.
After the IAEA started to accuse North Korea of violating IAEA obligations
by engaging in reprocessing plutonium in late 1992, Kim Il Sung was caught in a
deadlock situation in which Kim had no incentive to change his provocative and
conflictual behavior. When former U.S. President Carter’s visit reframed Kim
in a domain of gains, Kim then took the risk-averse decision to reach a nuclear
dismantlement for oil deal with the United States. However, North Korean lead-
ers did not really give up nuclear weapons. Instead, North Korea continued to
conduct a secret nuclear weapons program, which was less risky behavior in a
domain of gains than a public program.
We argue that North Korea’s nuclear weapons policy, therefore, is a function
of their leaders’ location within a domain of action. North Korea is more likely
to conduct provocative nuclear policy when its leaders are framed in a domain
of losses, i.e., when facing either internal or external challenges for the regime’s
security. However, North Korea is also still likely to harbor nuclear ambitions
when its leaders are framed in a domain of gains, i.e., when being confident
 North Korea’s Decision to Go Nuclear 99

about both domestic and international situations. We suggest that the interna-
tional community, especially the United States, should change its expectations
and strategies to deal with North Korea. Rather than insisting on nuclear disar-
mament, the United States should unofficially recognize North Korea’s nuclear
power status, treat North Korea as a “normal country,” encourage it to engage in
international society, and constrain its behavior toward others by using interna-
tional nonproliferation treaties and norms.

Three Existing Theories—North Korea Goes Nuclear

There are three major arguments in explaining North Korea’s nuclear foreign
policies: “coercive bargaining,” “U.S. threats,” and “national identity.” Based
on realist power transition theory and prospect theory, Victor Cha (2002: 48–78)
suggests a “coercive bargaining” model to explain North Korea’s erratic nuclear
policy. Cha argues that North Korea is a militarily weak but threatening state.
Because of the economic and military gaps between North Korea and South
Korea on the one hand and the United States on the other, North Korean lead-
ers framed themselves in a desperate position, i.e., the domain of losses according
to prospect theory. Therefore, North Korean leaders had the incentive to take
more risky actions, i.e., to provoke the United States and South Korea, in order
to change the status quo to North Korea’s advantage.
Although Cha’s research focuses mainly on North Korea’s nuclear policy under
Kim Jong Il during the second North Korean nuclear crisis in the early twenty-
first century, the coercive bargaining argument is also applicable to the first North
Korean nuclear crisis under Kim Il Sung. Cha suggests that North Korea con-
ducted a “coercive bargaining” strategy by using limited provocations to seek
compromises from the United States and South Korea, asserting in 2003 (Cha
and Kang 2003: 34) that “depending on how desperate Kim Jong-il assesses the
situation, one can imagine other such [provocative] incidents in the future … it
might conduct a Pakistani-type nuclear detonation.”
Cha’s “coercive bargaining” argument is indeed insightful in explaining North
Korea’s on-again and off-again negotiating behavior. However, it is misleading to
argue that North Korea’s nuclear test is merely for the purpose of “coercive bar-
gaining”. It is true that North Korea used its nuclear weapons program as leverage
and forced the United States to go back to the negotiating table in 1994 and 2007
respectively. However, North Korea did not achieve any compromise from the
United States until it agreed to take the first step to disable its nuclear facilities at
Yongbyon. Why North Korea agreed to give up its nuclear weapons program is
not clear from Cha’s “coercive bargaining” theory.
According to the “coercive bargaining” theory, North Korean leaders must
be desperate enough about the current situation, i.e., located in a domain of
losses, before conducting the first nuclear test in 2006. Although the explosion
was smaller than expected, the nuclear test was still viewed as a successful one
100 Kai He and Huiyun Feng

and North Korea also claimed itself as a nuclear power after the test. However,
about six months later North Korea agreed to give up its entire nuclear program
in exchange for energy assistance in the form of heavy fuel oil from the United
States and other countries concerned. Using the entire nuclear weapons program
to bargain for some heavy fuel oils is by no means a good deal for North Korea.
One possible explanation is that North Korea did not genuinely dismantle its
nuclear program. However, this possibility is well beyond the explanatory scope
of “coercive bargaining” theory.
David Kang (2003: 301–324) suggests a “U.S. threat” argument to explain
North Korea’s nuclear intentions and behavior. Kang argues that U.S. deterrence
policy has been clear and unambiguous, and it has contributed to relative peace
and stability in the Korean Peninsula for more than fifty years. After the cold war
the transformation of the distribution of power in the system shifted the balance
of power against North Korea. Besides the sharp economic and military gaps,
North Korea also lost two cold war patrons, the Soviet Union and China. The
former collapsed and Russia, the successor, was no longer capable and willing to
provide economic aid and security protection to the North Korean regime. The
latter betrayed a traditional friendship in the eyes of North Korea, because China
established diplomatic relations with South Korea in 1992.
Kang argues, therefore, that North Korea is by no means a threat to start an
unprovoked war. On the contrary, North Korea has legitimate security con-
cerns because of its disadvantageous strategic position versus South Korea and
the United States. Regarding North Korea’s nuclear and missile programs, Kang
suggests that “it was only when the balance began to turn against the North that
it began to pursue a nuclear weapons program. Both the weapons program and
the bellicose nature of its rhetoric are attempts to continue to deter the U.S. from
taking any preemptive moves against it” (Cha and Kang 2003: 45).
Kang is right to point out the security rationale of nuclear weapons to North
Korea. The dramatic change in the power distribution in the international system
definitely shaped North Korea’s security and foreign policy. Nuclear weapons
as the most efficient deterrent in current history are the natural choice of any
state facing imminent security threats. Although facing economic difficulties and
natural disasters in 1964, China conducted its nuclear test to deter threats from
both the Soviet Union and the United States. One of the key considerations
behind Indian and Pakistani nuclear tests in the late 1990s also stems from security
concerns. Therefore, North Korea has a similar legitimate reason to develop its
nuclear weapons program after the cold war.
However, Kang’s U.S.-threat argument is weak in explaining North Korea’s
inconsistent behavior in pursuing nuclear weapons. If North Korea’s nuclear pro-
gram only aims to deter the imminent threat from the United States, North Korea
should develop the program consistently and rapidly as the only way to effectively
deter U.S. threats. From the first nuclear crisis to the second one, however, North
Korea did not insist on its nuclear weapons program, at least publicly. Instead, the
 North Korea’s Decision to Go Nuclear 101

North seemed always willing to use its nuclear program to bargain for financial aid
and energy assistance. Although North Korea also demanded a security guarantee
from the United States, it apparently knows that the United States will not sign
such a treaty and that a paper treaty cannot provide real security to North Korea.
Therefore, it is not clear why North Korea was willing to trade its nuclear weap-
ons program if it is really threatened by the United States after the cold war.
One possible explanation is that the U.S. threat is not imminent to North
Korea after the cold war. It is true that China-North Korea relations were at odds
in the early 1990s. However, strategically China never abandoned North Korea.
China is not only the largest food and energy supplier of North Korea; it also
maintains its security alliance with North Korea. Technically, China replaced the
Soviet Union and provided security protection to the North after the cold war.
Therefore, although North Korea might perceive threats from the United States
and the South because of its disadvantageous military position, China’s strate-
gic balancing definitely alleviated the urgency of the North’s security concerns.
Kang’s U.S.-threat argument overemphasizes the importance of the international
environment, especially U.S. threats, while ignoring other domestic and even
psychological reasons for North Korea’s decision of going nuclear.
Using a comparative foreign policy approach, Hymans introduces a construc-
tivist understanding of Kim’s nuclear intentions. According to Hymans, nonra-
tional factors, including emotions, play a crucial role in North Korea’s pursuit of a
nuclear weapons program. Specifically, Hymans (2008: 263) suggests that a leader’s
“national identity conception” (NIC) forms a “crucial basis for revolutionary for-
eign policy decision(s),” i.e., why a state decides to go nuclear or not. NIC refers
to a leader’s “basic sense of what the nation naturally stands for and how high it
naturally stands in comparison to others in the international arena.” When a leader’s
NIC is oppositional nationalist, going nuclear seems to be nothing less than the
natural choice because an oppositional nationalist identity combines identity-driven
fear and pride. Nuclear weapons can not only help overcome the fear but also bring
pride to oppositional nationalist leaders. Comparing Atal Behari Vajpayee of India
and Kim Jong Il of North Korea, Hymans (2008: 264) suggests that both leaders are
oppositional nationalists and their decisions to go nuclear are not a product of cool
calculation, but are psychological, i.e., driven by identity and emotion.
Hymans’s psychological and constructivist approach is useful for understand-
ing the influence of leadership on North Korea’s nuclear decisions. To some
extent, Hymans’s argument is complementary to Cha and Kang’s power-driven,
materialist explanations of North Korea’s nuclear behavior. However, Hymans’s
identity-based argument faces two problems. First, if both Kim Jong Il and his
father Kim Il Sung are oppositional nationalists in nature, why did North Korea
not conduct nuclear tests during the cold war or even soon after the cold war?
The identity-based argument may help us understand the psychological roots of
North Korea’s nuclear intentions, but it fails to explain the timing and dynamics
of North Korea’s nuclear policies.
102 Kai He and Huiyun Feng

Second, the relationship between oppositional nationalism and the external

environment is not clear. Hymans tries to measure different levels of opposi-
tional nationalism and argues (2008: 271) that “oppositional nationalism is typi-
cally at the root of decisions to go nuclear.” However, he also recognizes that
“the negative trends in its [North Korea’s] international position since the 1980s
may have caused the DPRK leadership to desire the bomb even more fervently
than before” (Hymans 2008: 271). If both external environment and leadership
identity contribute to North Korea’s decision to go nuclear, how can one iden-
tify which one plays the major role? In other words, the interaction between the
external environment and identity is ambiguous in Hymans’s argument.

Prospect Theory and North Korea’s Deceptive Bargaining

We introduce a prospect-theory-based, “deceptive bargaining” model to comple-
ment the three existing arguments for explaining North Korea’s nuclear behavior.
Prospect theory is a behavioral theory of decision making under conditions of risk
introduced by Daniel Kahneman and Amos Tversky (1979). In laboratory experi-
ments they find that how people interpret their situations as domains of gains or
losses influences how they behave in terms of risk-aversion or risk- acceptance.
They argue that people tend to evaluate choices with respect to a reference point.
People choose risk-averse behavior in a domain of gains but risk-acceptant behav-
ior in a domain of losses.2 Simply put, if people are in an advantageous situation,
or a domain of gains, they are more likely to behave cautiously (risk-aversion).
However, when people are in a disadvantageous situation, where all the choices
and possible outcomes are getting worse, they are more likely to choose risky
behavior (risk-acceptance) to avoid their losses (McDermott 2004: 294).
The major debate over the North Korea nuclear proliferation issue focuses
on how to assess North Korea’s nuclear intentions. Kang’s U.S.-threat argument
suggests that North Korea’s intention of going nuclear is rooted in its legitimate
security concerns due to a hostile U.S. policy toward North Korea after the cold
war. Hymans’s identity-based, oppositional nationalist theory basically argues that
North Korean leaders are ambitious in nature regarding nuclear weapons because
of the embedded emotions of fear and pride associated with their oppositional
nationalism identity. The debate over North Korea’s nuclear intentions is under-
standable politically but may be relatively fruitless, because intentions are not
constant and may change overtime. In addition, intentions may or may not be
consistent with actual behavior. We suggest that what should really concern the
international community about North Korea is not what North Korean leaders
want but what North Korean leaders really do.
Cha’s “coercive bargaining” model stems from prospect theory, and he does
not assume the ultimate intentions behind North Korea’s nuclear policy. Cha
focuses instead on explaining North Korea’s behavior through examining the
prospects for decision. We follow in the footsteps of Cha’s “coercive bargaining”
 North Korea’s Decision to Go Nuclear 103

model and introduce a new prospect theory model—“deceptive bargaining”—

to explain North Korea’s nuclear behavior without making assumptions about
North Korea’s nuclear intentions, i.e., why they want nuclear weapons. While
Cha’s “coercive bargaining” argument may account for North Korea’s provoca-
tive behavior in general, our “deceptive bargaining” model only applies to North
Korea’s behavior regarding nuclear issues.
Borrowing insights from both Kang’s and Hymans’s arguments, therefore, we
simply assume that nuclear weapons are desirable for North Korea because of the
interaction between an instigating cause (severe international environment after
the cold war) and a standing cause (long-term ambitions of North Korea’s leaders)
that together create the necessary and sufficient conditions to seek nuclear weap-
ons.3 We do not debate why North Korea wants nuclear weapons, but simply try
to explain how North Korea behaves in order to get nuclear weapons under these
conditions. How to go nuclear is a behavioral and empirical question and not an
intentional question regarding North Korean leaders.
North Korea can go nuclear publicly and provocatively without considering
any consequences, or North Korea can go nuclear secretly or deceptively in order
to minimize the potential risks associated with its nuclear weapons program. The
public and provocative way to pursue nuclear weapons reflects a high propensity
to accept the risks of undesirable consequences, because the international com-
munity may impose sanctions and rival countries may take preemptive actions to
militarily stop the infant nuclear program. The deceptive way of going nuclear is
also risky, but the degree of risk-acceptance is much lower than the public one.
North Korea’s nuclear program was first revealed in the early 1990s when IAEA
inspectors found the discrepancy between North Korea’s promise and the real
development of its nuclear program. From 1994 to 2006, North Korea expe-
rienced two nuclear crises with the United States. North Korea provoked the
United States by withdrawing from NPT and starting its nuclear programs while
the United States reportedly prepared military strikes against North Korea. Both
crises ended peacefully after North Korea agreed to stop its nuclear programs in
exchange for US financial and energy aid. In May 2009, North Korea conducted
its second nuclear tests. The future of this latest nuclear crisis is still unclear.
The puzzle is why North Korea initially provoked the United States on the
nuclear issue and later agreed to quit the nuclear weapons program. Cha’s “coer-
cive bargaining” model correctly describes North Korea’s behavior in negotiations
during the crises. However, from the trajectory of North Korea’s nuclear weapons
program, we can see that North Korea never really gave up its nuclear weapons
program. Therefore, North Korea did not only use its nuclear program to bargain
for oil and food, but also, more importantly, to gain time. Negotiation and bargain-
ing are only deceptive tactics by North Korea to buy time in order to continue its
nuclear weapons program in secret. This “deceptive bargaining” behavior is less
risky than the public and provocative way of going nuclear in two respects.
First, bargaining for food and oil with its nuclear weapons program gives the
104 Kai He and Huiyun Feng

international community, especially the United States, a false hope that North
Korea will give up its nuclear program if rewards are forthcoming. The United
States and other countries concerned will be reluctant to conduct preemptive mil-
itary actions because no one wants to spill blood if money can solve the problem.
Second, by signaling its willingness to negotiate and bargain, North Korea places
the United States and other countries in an immoral situation if these countries
insist on military actions. Based on the risks of North Korea’s different nuclear
policy options, we identify a public and provocative nuclear policy as risk-taking
behavior and a “deceptive bargaining” nuclear policy as risk-avoiding behavior.
In order to analyze whether North Korea made risk-acceptant or risk-averse
nuclear policies, we also need to set a reference point to specify different domains
of action. We use the status quo to measure the location of action domains as
reference points. North Korean leaders’ perceptions regarding the regime’s secu-
rity is a benchmark of the status quo. However, because of the secretive nature
of the regime and the closed society in North Korea, it is relatively difficult to
obtain reliable data to examine North Korean leaders’ perceptions and decision-
making processes. This is why some scholars criticize Cha’s coercive bargaining
argument—one of the applications of prospect theory—as based on speculations
rather than facts (Kang 2003).
In our research we employ operational code analysis and a subjective game
model to determine whether North Korean leaders framed the situations as either
in the domain of losses or the domain of gains. Relying on an operational code
analysis of North Korean leaders’ public speeches, we can identify changes in Kim
Il Sung’s belief system at the end of the cold war. Through combining operational
code analysis and the Theory of Moves (TOM) developed by Brams (1994), we
can identify North Korean leaders’ “initial state” (the status quo) in a subjective
game with its rivals. We employ a prospect theory-sequential game model to
identify Kim Il Sung’s reference points before and during the first nuclear cri-
sis in the early 1990s and formulate the following testable hypotheses with our
“deceptive bargaining” nuclear proliferation model: When North Korean leaders
are framed within a domain of losses, they are more likely to choose a risk-accep-
tant nuclear policy, i.e., to pursue nuclear weapons publicly. Conversely, when
North Korean leaders are framed within a domain of gains, they are more likely
to choose a risk-averse nuclear policy, i.e., to use “deceptive bargaining” tactics
to pursue nuclear weapons secretly.

Deceptive Bargaining and Kim Il Sung’s Nuclear Ambitions

The sudden collapse of the Soviet Union caused both economic and security
challenges to North Korea in the early 1990s. Although North Korea emphasized
the juche (self-reliance) ideology of Kim Il Sung, its economy largely relied on
subsidized fuel and other economic aid from the Soviet Union and China in the
1980s. For example, over 50 percent of North Korea’s foreign trade was with
 North Korea’s Decision to Go Nuclear 105

the Soviet Union in the 1980s (French 2007: 98; also see Cumings 2004). In
the late 1980s, due to its own economic crisis the Soviet Union under Mikhail
Gorbachev started to reassess its economic aid policy toward client states, includ-
ing North Korea. In September 1990, the Soviet Union signed an accord with
North Korea to settle two-way trade in hard currency and at world prices and
discontinued barter trade between the two countries (Beal 2005). After the col-
lapse of the Soviet Union, nearly all Soviet subsidies were withdrawn from North
Korea and its economy plummeted dramatically.
North Korea also lost its security protection and political allies due to the end
of the cold war. In the late 1980s, the Soviet Union started to engage diplomati-
cally with South Korea. In September 1990, the Soviet Union established dip-
lomatic relations with South Korea. China also followed a course of establishing
formal relations with South Korea in 1992. Soon after the Soviet Union collapsed
in 1991, Russia cut its military aid to North Korea. Although China kept its
“friendship” with North Korea, China was facing its own political challenges and
international isolation after the Tiananmen Square incident. North Korea became
a socialist orphan in the early 1990s in that it was abandoned or ignored by its
patrons in the international system.
An operational code analysis of Kim Il Sung’s belief systems from 1987 to 1990
shows that compared to the average world leader, Kim believed that the political
future (P-3) was significantly less predictable (p < .05) and his control over the
historical situation (P-4) was significantly lower (p < .001) than the average world
leader. Given the challenging strategic situation faced by North Korea soon after
the cold war, it is not a surprise that Kim Il Sung’s belief systems reflected his low
confidence in controlling international events even though he did not believe that
the nature of the political universe was significantly less friendly. The aggregated
scores for Kim’s key beliefs specify the preference orderings in Proposition 4 for
Self in the Theory of Inferences about Preferences (TIP), which ranks domina-
tion over settlement over submission over deadlock for self.4
Applying a prospect theory-sequential game model, we can map Kim Il Sung’s
strategy during the 1987–1990 period within a one-sided ordinal game matrix in
Figure 6.1 adapted from the Malici (2008) study. The strategy of Kim Il Sung in
this game matrix regarding what to do is Cooperation (CO) and the payoff for
Kim is either (2) or (3), depending on how its rivals move. Because of the dra-
matic structural changes in the international system, Kim realized the disadvanta-
geous and dangerous situation for the regime’s security. The low self-confidence
regarding control over international events drove Kim to seek conciliation with
its traditional rivals, the United States, South Korea, and Japan for its own security
from 1987 to 1990.
In 1987, Kim proposed troop reductions to the United States and South Korea.
In January 1988, Kim Il Sung proposed a North-South Conference. In January
1989, Kim issued invitations to South Korean officials. In 1990, North Korea
conducted the first round of prime ministerial talks with South Korea in Seoul. In
106 Kai He and Huiyun Feng

US/SK US/SK US/SK

CO CF CO CF CO CF
CO 3 2 CO Settle Submit CO 3 1

NK NK NK

CF 4 1 CF Dominate Deadlock CF 4 2

Period 1 1987–1990 North Korean Outcomes Period 2 1991–1994

P–3 = .082 (p ≤ .05) P–3 (Ave. Ldr) = .148 P–3 = .098
P–4 = .079 (p ≤ .001) P–4 (Ave. Ldr) = .212 P–4 = .147 (p ≤ .10)
I–1 = +.128 I–1 (Ave. Ldr) = .334 I–1 = +.251

Payoffs are ranked (4) highest to (1) lowest. NorthKorea’s public strategies
for each time period are underlined. Source: Malici (2008a: 109–111, 122–123).

FIGURE 6.1 North Korea’s Strategies, Payoffs, and Outcomes

1991, North Korea abandoned its previous resistance and agreed to join the United
Nations simultaneously with South Korea. In December 1991, the North and
the South signed an “Agreement on Reconciliation, Non-aggression, Exchanges,
and Cooperation.” Responding to the U.S. worldwide withdrawal of tactical
nuclear weapons, the two sides also signed a “North-South Joint Declaration on
Denuclearization of the Korean Peninsula” in January 1992.
However, it should be noted that Kim’s cooperative gestures did not initially
receive warm responses from the United States and South Korea. For example,
Kim’s troop-reduction proposal was rejected by the United States and South
Korea in 1987. After Kim proposed official talks with South Korea in 1989,
the United States deployed nuclear weapons in Japan in September, which was
apparently seen as a security threat to North Korea. From Kim’s subjective game
we see that Kim’s initial payoff regarding the pursuit of a Cooperation strategy is
submission (2). Kim’s subjective payoff shifted from (2) to (3) from 1987 to 1990,
because of slowly warming reactions from the United States and South Korea to
Kim’s cooperative initiatives.
The initial payoff of (2) located Kim in a domain of losses while the later
payoff of (3) placed Kim in a domain of gains, in which the domain of losses is
defined by the two lower-ranked outcomes of (2) and (1) while the domain of
gains is defined by the two higher-ranked outcomes of (3) and (4). Therefore,
Kim’s reference point shifted from a domain of losses to a domain of gains in the
1987–1990 period. The deceptive bargaining model predicts that Kim should
develop nuclear weapons provocatively in a domain of losses and secretively in
a domain of gains. Kim’s nuclear policy during this period basically vindicates
these predictions of the deceptive bargaining model. As reported widely, Kim
 North Korea’s Decision to Go Nuclear 107

Il Sung decided to start a nuclear weapons program around the mid-1980s dur-
ing the cold war. In 1987, Kim was located in a domain of losses (2) because of
the Conflict reactions of the United States and South Korea to his Cooperative
diplomatic strategy. Kim’s decision to go nuclear is a rational but risky behavior
from a domain of losses.
Just as the deceptive bargaining model suggests, Kim secretly developed nuclear
weapons when he was framed in a domain of gains (3). Because of the subsequent
shifts in Kim’s reference point from losses to gains in the late 1980s at the end
of the cold war, Kim Il Sung chose to continue secretly refueling and extracting
plutonium in the Yongbyon reactor. It is also an understandable, risk-avoiding
approach simply because developing nuclear weapons is still a taboo in world pol-
itics. Since North Korea joined the NPT in 1985, a public violation of the NPT
might have further isolated it from the international community. In addition, the
United States might have taken a preemptive strike to stop the nuclear program if
North Korea’s nuclear ambitions were revealed. A vivid example was Israel’s air
strike against the Iraqi Osirak nuclear reactor in 1981. While the domain of losses
in 1987 drove Kim initially to develop its nuclear programs, the taboo against
them and the prospect of future cooperation with the United States and South
Korea coincided to constrain the provocative form of Kim’s nuclear ambition.
Kim’s international situation was shaped as a domain of gains after the United
States and South Korea began to engage North Korea in cooperation.
Besides the breakthrough in North-South relations and international conciliation
between North Korea and the United States, Kim Il Sung’s strong domestic control
also acted to frame his location in a domain of gains in the late 1980s. First, his past
revolutionary experience gave Kim Il Sung unchallenged authority and legitimacy
in North Korea. Kim Il Sung promoted “Great Leader’s Absolutism” to worship
himself, and he became a living god in North Korea (Becker 2005: 70–71; Cumings
2005). Second, North Korea’s security concerns were largely caused by the loss of
security protection rather than the increase of military threats after the cold war.
As indicated above, the collapse of the Soviet Union changed the strategic land-
scape in Asia and North Korea lost the military and economic patron it could reply
on. However, the United States did not seize the opportunity to impose immedi-
ate military threats on North Korea. Because of domestic economic constraints
and a vanishing Soviet threat, the Bush Administration instead started to consider
withdrawing troops from Asia, including South Korea. In September 1991, Bush
announced a worldwide withdrawal of U.S. tactical nuclear weapons.
Although Kim Il Sung might have believed himself to be vulnerable and insecure
after the cold war, he did not face an imminent threat from the United States. His
location in the subjective game matrix for 1987–1990 in Figure 6.1 was not his
worst location (1). Therefore, Kim Il Sung’s diplomatic strategy of Cooperation
and military strategy of secret nuclear armament were strategic decisions in the
late 1980s that avoided his worst outcome (deadlock) with the possibility of war
between North Korea and its enemies. However, this relatively tranquil period
108 Kai He and Huiyun Feng

between North Korea and South Korea/United States at the beginning of the
1990s would not last long.
Because of the denuclearization declaration signed between North Korea and
South Korea in 1991, North Korea also signed the safeguards agreement with the
IAEA in 1992. Under the agreement the IAEA started a series of inspections of
North Korea’s nuclear facilities. In February 1993, the IAEA uncovered evidence
that North Korea reprocessed plutonium on three occasions in 1989, 1990, and
1991 (Becker 2005). In addition, the United States also presented satellite photos
to further reveal North Korea’s secret nuclear facilities in Yongbyon. North Korea
was convicted of cheating by the IAEA evidence and U.S. satellite photos.
The IAEA further asked to inspect two unreported facilities suspected of hold-
ing nuclear waste. North Korea refused the request of the IAEA and declared them
to be “military sites” and “off limits” (Bechtol 2007: 14). The IAEA brought this
issue to the UN Security Council in April 1993, and the UN Security Council
soon passed a resolution asking North Korea to allow IAEA inspections under
the NPT and the IAEA safeguards agreement. However, North Korea responded
with fury by threatening that it would withdraw from the NPT. North Korea’s
brinkmanship triggered the first nuclear crisis in the Korean Peninsula.
The operational code analysis of Kim Il Sung conducted by Malici (2008: 121)
shows that Kim’s operational code in this period prescribes a dominant strategy of
conflict consistent with TIP’s Proposition 5, which ranks domination over settle-
ment over deadlock over submission. Analyzed within a one-sided ordinal game
matrix, we see in Figure 6.1 Kim’s Conflict (CF) strategy when North Korea’s
nuclear program was uncovered by the IAEA and later threatened by the United
States between 1991 and 1994. Kim’s payoffs could be (2) or (4), depending on
how its rivals behave. When both sides choose Conflict, then Kim’s payoff is (2);
when Kim’s rivals (the United States) choose cooperation, Kim’s payoff is (4).
The one-sided game matrix suggests that the payoff of (2) defines a domain of
losses and the payoff of (4) indicates a domain of gains.
Because of the strong conflict reaction from the United States to Kim’s secret
nuclear program revealed by the IAEA, we conclude that Kim framed himself in a
domain of losses (2) when the nuclear crisis started in early 1992. The United States
declared as its firm position that North Korea’s nuclear program should be stopped
by all means, including economic sanctions and military strikes. North Korea resisted
with a Conflict strategy and threatened that any economic sanction would be taken
as an act of war and a war would turn Seoul “into a sea of fire” (McAllister 1994;
Smith 2006). When both sides continued playing hardball and the crisis escalated
to a breaking point, a dramatic turn was brought by former U.S. President Carter’s
private visit to North Korea in June 1994. Kim Il Sung reciprocated this concilia-
tory move by promising Carter that North Korea would go back to the negotiating
table, allow the IAEA inspections, and also freeze its nuclear program.
Soon after Carter’s visit, North Korea resumed bilateral talks with South Korea
and also opened a new round of talks with the United States on the nuclear issue in
 North Korea’s Decision to Go Nuclear 109

Geneva. Although the sudden death of Kim Il Sung in July interrupted the sched-
uled North-South summit, North Korea finally signed the Agreed Framework
with the United States in October 1994. According to the Agreed Framework,
North Korea pledged to freeze its nuclear program and in return the United States
promised to provide two light-water reactors as well as 500,000 tons of heavy oil
annually until the light-water reactors were established in 2000.
This sequence of events vindicated Kim’s risk-acceptant, Conflict strategy, pre-
dicted by prospect theory, eliciting a conciliatory move from the United States
that moved him from a domain of losses (2) to a domain of gains (4) in Figure 6.1.
From this new reference point in a domain of gains, prospect theory predicts a
risk-averse strategy by Kim of staying at (4) domination in talks or moving to (3)
leading to the Agreed Framework in 1994. According to Cha’s “coercive bargain-
ing” theory, it seems that North Korea successfully used nuclear brinkmanship to
get what it wanted, i.e., an energy supply and economic aid. Kang’s U.S. threat
argument suggests that the U.S. concession is the major reason for the peaceful
settlement of the crisis, which might be partly true. However, based on North
Korean defectors’ reports and the CIA’s evidence, North Korea never really gave
up its nuclear weapons program. It appears that North Korea never trusted the
promises of the United States, and the U.S. compromise was also not the major
reason for North Korea’s policy change. Cha’s coercive bargaining theory is cor-
rect in describing the negotiating behavior of North Korea, but it is inadequate in
explaining what North Korea is really bargaining to achieve.
North Korea did not bargain merely for aid and an energy supply; it used the
bargaining tactic to buy more time for its nuclear weapons program. Based on
our “deceptive bargaining” model, we suggest that North Korea did not intend to
give up its nuclear weapons program in the first place. It is why North Korea only
agreed to “can and store their spent fuel, but refuse(d) to give up their nuclear
components entirely” in the Agreed Framework (Snyder 1999). Facing inter-
national pressures, including the UN resolution, North Korea had two options.
One was to continue its nuclear program despite U.S. threats and possible military
conflicts. The other was to hold back temporarily and secretly continue its nuclear
program later. The first option appeared much more risky than the second one
since the United States had prepared an evacuation of U.S. citizens from Seoul as
well as an air strike plan, according to U.S. decision makers during the crisis such
as Ashton Carter and William Perry (2002).
We conclude that Kim Il Sung chose the second option to buy time for North
Korea. Technically, North Korea still needed more time to weaponize its nuclear
program. In the early 1990s, North Korea may have had enough plutonium to
conduct a nuclear test, but how to weaponize its nuclear capability with a deliv-
ery system was still a technical hurdle for North Korean scientists. Without mis-
siles that could deliver nuclear bombs, North Korea’s deterrent capability against
the United States would be less credible. More importantly for our argument,
Kim Il Sung framed himself within a domain of gains by the end of the crisis. It
110 Kai He and Huiyun Feng

led him to choose a risk-averse, nuclear policy through the deceptive bargain-
ing tactic of moving to (3) settlement of the crisis with the United States while
secretly continuing a nuclear weapons program—perhaps as a hedge against the
risk of submission (1) to the United States in the future if Washington did not
honor the terms of the settlement. As mentioned above, Kim’s domestic control
was unchallenged in the early 1990s. According to Hwang Jang-yop, the high-
est-ranked defector from the North, Kim Il Sung had a heated debate with Kim
Jong Il, his son, over how to deal with the 1994 nuclear crisis. While Kim Jong Il
was hawkish and even preferred preemptive attacks on South Korea, Kim Il Sung
rebuffed his son and insisted on returning to the negotiation table with the United
States (Becker 2005: 168–169).
The success of Kim Il Sung in rejecting his hawkish son’s idea shows that the
elder Kim still strongly controlled the regime. Moreover, Carter’s private visit
also offered Kim Il Sung psychological, face-saving satisfaction. North Korea’s
international embarrassment, due to its accused cheating and violations of the
NPT and IAEA agreements, was overshadowed by the spotlight of Carter’s visit.
As the prospect theory-sequential game model suggests, Kim’s payoff changed to
(4) when his rivals chose Cooperation while his choice remained Conflict. It was
reported that Clinton was initially uncomfortable about Carter’s visit, which was
broadcast via CNN, because he felt he was being upstaged by the former presi-
dent (Becker 2005: 168). On the contrary, Kim Il Sung was likely very happy to
meet Carter since it was an unofficial compromise of the United States. Carter’s
visit was seen as a U.S. compromise in Kim’s eyes, and it also changed Kim’s ref-
erence point from losses to gains.

Conclusion
Our deceptive bargaining model suggests that Kim is more likely to secretly
develop a nuclear program in a domain of gain. In fact, Kim employed this decep-
tive bargaining strategy to sign the Agreed Framework with the United States
and continue his secret nuclear program. Challenging the three existing theories,
“coercive bargaining,” “U.S. threat,” and “oppositional nationalism,” we have
introduced a prospect-theory-based, “deceptive bargaining” argument to explain
North Korea’s seemingly irrational and erratic nuclear policies during the first
nuclear crisis in the early 1990s. We have suggested that North Korea is more
likely to choose high risk-taking behavior in public, i.e., to conduct a provoca-
tive nuclear policy, when the leaders framed themselves within disadvantageous
and desperate international and domestic situations. When North Korean leaders
framed themselves in an advantageous situation, i.e., when they can confidently
control their domestic situation and cope with international challenges, they are
more likely to conduct a risk-averse policy to hide their nuclear ambitions and
secretly develop their nuclear weapons programs.
Although North Korea pledged to dismantle its nuclear program and destroy
 North Korea’s Decision to Go Nuclear 111

its nuclear facilities in Yongbyon, we suggest that it is just another deceptive

bargaining tactic of North Korea in order to achieve its nuclear ambitions. Since
the verification of denuclearization cannot be successful without North Korea’s
full cooperation, there is no reason for North Korea truly to give up its nuclear
weapons program. Instead, through deceptive bargaining tactics North Korea can
not only get economic benefits and energy aid from the United States, but also
hedge its bets through secret nuclear programs.
In May 2009, North Korea once again conducted an underground nuclear test
as well as multiple missile tests. According to the deceptive bargaining model,
North Korea’s provocations are risk-taking behaviors, which are more likely cho-
sen by North Korean leaders when their decision-making situation is in a domain
of losses. The widespread rumor of Kim Jong Il’s health problems in late 2008
revealed an intense domestic struggle with a focus on the succession issue after
Kim Jong Il. North Korea’s provocations through a second nuclear test and mul-
tiple missile tests in May 2009, therefore, is a risk-acceptant pattern of behavior
not only for attracting attention and seeking compromises from the United States,
but also, more importantly, for consolidating domestic support for the regime in
the context of a forthcoming power transition after Kim Jong Il in North Korea.
We suggest that the United States should abandon the illusion that North Korea
would give up its nuclear program voluntarily for economic aid and benefits,
because nuclear weapons are the only efficient deterrent for North Korea against
the United States in an anarchic international system. The United States should
instead unofficially recognize that North Korea is a nuclear power like India and
Pakistan. The major task for the United States now is not to disarm a nuclear
North Korea, but rather to prevent nuclear proliferation from North Korea to
other states and non-state actors. Therefore, the United States should treat North
Korea as a normal country and encourage it to engage in international society.
The more integrated into the international society, the less likely is North Korea
to violate nonproliferation norms and rules. In other words, if the United States
treats North Korea as a rogue, it will likely behave like a rogue; if the United
States treats North Korea as a “normal” state, this altercasting strategy may lead
North Korea to behave in a civilized way (Malici 2008: 30).

Notes
1 North Korea conducted its second nuclear test in May 2009, which triggered a third
nuclear crisis. The final outcome of this new crisis is still unclear as this chapter is being
written.
2 For applications of prospect theory in International Relations, see Farnham (1994),
Levy (1997), McDermott (1998), Mercer (2005), and Taliaferro (2004a, 2004b).
3 For a discussion of the distinction between “standing conditions and instigating
conditions within a causal field,” see Little (1991: 26).
4 The operational code analyses of Kim Il-sung beliefs and subjective games were
conducted initially by Akan Malici (2008: 109–127). The VICS indices and two-tailed
significance tests in Figure 6.1 are taken from Malici’s study.
7
SMALL GROUP DYNAMICS
The Psychological Characteristics of Leaders
and the Quality of Group Decision Making

Mark Schafer, Jonathon Nunley, and Scott Crichlow

Introduction
Central to the notion of the microfoundations of behavioral IR is that individuals
matter in the conduct of foreign policy making. Further, if individuals matter, it
must also be the case that they differ from one another; if individuals behaved the
same, then which individual occupied the leadership position would not matter.
One thing that makes individuals different from each other is their psychology,
i.e., their mental and behavioral characteristics. While it is obvious that individuals
are different from each other; what is less clear is whether or not those differences
matter in foreign policy making. This empirical question lies at the heart of this
chapter: Do the psychological characteristics of U.S. leaders affect the conduct of
foreign policy in their respective administrations?
We shall turn to the psychological characteristics under investigation here
shortly, but let us first specify our dependent variables of interest pertaining to
the conduct of foreign policy. We conceptualize foreign policy in two broad
ways. The first is the process of making foreign policy. This aspect pertains to the
style, structure, and procedures used by the president and his advisory system to
arrive at decisions. The locus classicus in this area is Irving Janis’s (1972) concept of
groupthink, which demonstrated that the quality of the decision-making process
had a significant effect on the outcome of a case (see also George 1980; Haney
1997; ‘t Hart 1990; Hoyt 1997; Hybel 1993; Preston 2001; Schafer and Crichlow
1996, 2000, 2010). We speculate generally that the psychology of a leader may
affect the quality of the decision-making process in an administration. It is easy
to conceive that the quality of the decision making in an administration may be
affected by such things as the leader’s openness to information, control orienta-
tion, trust level, task orientation, and others.
The second broad dependent variable here pertaining to the conduct of for-
eign policy is how a case turns out. In this chapter we look at outcomes across
 Leader Characteristics and Small Group Dynamics 113

two different dimensions: (1) whether or not a decision helped or hindered the
state’s national interests; and (2) whether a decision had the effect of escalating
or de-escalating conflict. Individual psychological characteristics such as conflict
orientation, distrust, need for power, and task orientation may affect one or both
of these qualitative outcomes. We shall discuss the operationalizations of these
variables in the methods section below.
This book focuses extensively on the operational code as a psychological con-
struct, and this chapter is no exception. While the operational code is primarily a
cognitive construct—focusing on the beliefs of leaders—our theoretical frame in
this chapter includes as well concepts that come from other parts of the psycho-
logical spectrum. Toward that end, we include in our analysis variables from the
research program called Leadership Trait Analysis (LTA) (Hermann 1999). David
Winter (2003) argues that cognitions, traits, and motives are central components
of personality.1 A broad, if imperfect, differentiation of these concepts, as opera-
tionalized here, is that cognitions as beliefs represent the more conscious side of
personality, while traits and motives represent the more unconscious side of per-
sonality (Greenstein 1969; Schafer 1999; Schafer and Crichlow 2010).
Interestingly, though both research programs have been around for some time,
there is virtually no empirical work that investigates the relationships between
variables in both programs. This chapter will present some data that addresses this
task. Keeping with the primary focus on the operational code in this book, our
analytical focus will be on what kinds of traits from the LTA program are associ-
ated with different types of operational codes. For example, if someone scores
high (cooperative) on I-1 (or other indexes) in the operational code, what kinds
of traits is that person likely to have?
Including variables from both research programs will not only help us to under-
stand the interplay of different components of personality, but it will also enable
us to more fully model our primary research questions: How do the psychological
characteristics of leaders affect the quality of decision making within their respec-
tive administrations, and how do they affect the outcomes of decisions in those
administrations?

Methods and Variables

For this project we have selected 34 foreign policy cases involving the United
States from the Ford administration through the George W. Bush administration.
The case is the unit of analysis in our research design. Each case includes different
categories of variables. The dependent variables include outcome ratings for each
case and ratings for the quality of the decision-making process. The independent
variables include some taken from the Operational Code Analysis program and
some from the Leadership Trait Analysis program.
We ascertained values for rating our two outcome variables by asking a panel of
foreign policy experts to respond to the following questions and scaling the answers:
114 Mark Schafer, Jonathon Nunley, and Scott Crichlow

• National Interests. During the days and weeks immediately following the foreign
policy decision in question, do you believe that the actor’s national interests
were significantly advanced, somewhat advanced, unaffected, somewhat hin-
dered, or significantly hindered as a result of the decision? (Scale range 1 to 5,
with 1 being significantly hindered and 5 being significantly advanced.)
• Level of International Conflict. During the days and weeks immediately follow-
ing the foreign policy decision in question, do you believe that the level of
international conflict increased significantly, increased somewhat, remained
about the same, decreased somewhat, or decreased significantly as a result of
the decision? (Scale range 1 to 5, with 1 being significantly decreased and 5
being significantly increased.)
We also asked the respondents to assess their own confidence in rating each case
as low, moderate, or high. If a respondent rated their confidence in a case as low,
meaning the individual was not very familiar with the case, we did not include
that respondent’s ratings in the calculation of the outcome scores for that case. To
determine the score for each outcome variable for each case, we computed the
average rating from all the experts who rated their own confidence in the case as
medium or high. The pool included both male and female scholars from a range of
academic levels who had diverse methodological and ideological backgrounds. In
all, 28 experts provided ratings for the cases.
We began rating decision-making quality by specifying operational definitions
for a number of indicators of decision-making quality based on the literature in this
field. We then gathered extensive secondary literature pertaining to the decision-
making process for each case in the dataset. We included such things as memoirs,
biographies, and accounts by journalists and historians. The objective was to get as
much information as possible for each case that would let us look post-hoc inside
the decision-making process. We then read through these materials and coded each
of our operationalized variables for each case, one at a time. Building on Janis’s
work (1972, 1982), as well as some of our own earlier work (Schafer and Crichlow
1996, 2002), we divided the individual indicators of decision-making quality into
two broad categories: group structures and decision processing.
• Group structures pertain to factors in the administration that are generally
already in place before the case begins. These are the organizational structures
and previously established procedural norms that have been put in place by
the leader and his or her advisors. They include such things as how insulated
the group is, how methodical the procedures are, whether the group has a
general tolerance for disagreement, how experienced the team is and how
well they work together.
• Decision processing pertains to the process that goes on during the decision
episode itself. Most of these factors pertain to the processing of information
in the case, such as thoroughly searching for information, carefully consider-
ing objectives and options, stereotyping the out-group or the situation, or
 Leader Characteristics and Small Group Dynamics 115

creating pressures for uniformity. We compute additive indexes for each of

these two areas by summing the number of negative occurrences of each
individual indicator within each broad area; higher scores on these indexes
indicate more decision-making faults or problems.2

There are at least two challenges faced by researchers who wish to assess the psy-
chological characteristics of leaders. The first is that, in order to run statistical mod-
els, we need some way to measure psychological characteristics. We are interested in
knowing relative levels of the characteristics, such as which leaders are higher than
average on distrust or lower than average in their conflict orientation. The second
problem is that we rarely if ever have direct access to leaders such that we might
assess their psychological characteristics in person.
Methods in modern political psychology resolve these challenges by using
psychological assessment “at-a-distance” (Hermann 1980, 1984; Renshon 2009;
Schafer and Walker 2006c; Schafer 2000; Winter et al. 1991). The key assump-
tion with this approach is that the verbal behavior of an individual—what he says
and how he says it—can tell us much about that individual’s psychology. The
connection between verbal behavior and psychological dispositions is neither
new nor original; for example, it is a very old idiom that the optimist is one who
states that the glass is half full. Modern political psychologists have simply signifi-
cantly expanded the notion by carefully specifying operational definitions for a
range of psychological characteristics and then developing large dictionaries of
words and phrases that fit the operational definitions (Walker et al. 1998; Schafer
and Walker 2006c; Guttieri et al. 1995; Hermann 1999, 2003; Winter 2005).
Content analysis of the subject’s verbal material thus allows us to assess different
components of the subject’s psychology.
Though space does not permit us to provide complete operational definitions
of all of the psychological variables in this project, a couple of examples will
help make this process clearer. One of the psychological characteristics in LTA
is conceptual complexity. Hermann defines this as the propensity “to differentiate
things and people in one’s environment” (Hermann 1999: 10). Those scoring low
in complexity tend to see the world in simple, black-and-white, either-or terms.
Those scoring high in complexity tend to see the world in much more differenti-
ated, shades-of-gray terms. Hermann operationalized this variable by developing
two large dictionaries of words and phrases, one that includes low complexity
markers, such as always, never, and definitely, and one that includes high complex-
ity markers, such as perhaps, maybe, and possibly. The subject’s verbal material is
then content analyzed for these high and low complexity markers; the subject’s
complexity score is simply calculated as a ratio between the two types of markers.
The power of today’s desktop computers allows us to digitalize this process and
use computer software for the content analyses.3
Another example comes from the operational code and its content analysis pro-
gram called the Verbs in Context System (VICS) (Walker et al. 1998; Schafer and
116 Mark Schafer, Jonathon Nunley, and Scott Crichlow

Walker 2006c). The first Philosophical index (P-1) is called the nature of the political
universe and is interested in how the subject sees other political actors in terms of
friendliness or hostility. VICS begins with an analysis of verbs in the subject’s verbal
material and codes those verbs as either positive/cooperative—such as support, give,
and help—or negative/conflictual—such as attack, take, threaten. Of course, every
subject uses some of these verbs when referring to self, but for P-1 we are interested
only in the subject’s views of the rest of the political universe outside of self. For
this reason we use only verb phrases where the subject of the sentence is someone
other than self or self’s ingroup. To determine the subject’s P-1 score, we simply
compute a ratio of the positive/cooperative verbs to the negative/conflictual verbs
in the subject’s verbal material when she or he is talking about others. The result in
both of these examples is that we end up with interval-level measurements of the
psychological characteristics in which we are interested.
This study includes nine psychological characteristics, five from LTA and four
from the operational code construct. We chose these nine based on hypothesized
relationships with the dependent variables. The five from LTA, along with their
brief conceptual definitions are need for power, which is about having an impact on
others and enhancing one’s power and influence; distrust “involves a general feel-
ing of doubt, uneasiness, misgiving and wariness about others—an inclination to
suspect the motives and actions of others” (Hermann 1999: 31); task focus is a trait
with a continuum having two poles: one end is a focus on task while the other
end is a focus on relationship building; conceptual complexity (discussed above); and
belief in ability to control events, which is the extent to which the leader sees him
or herself in control of situations. The four characteristics from the operational
code are: (P-1) nature of the political universe; (I-1) strategic approach to goals; (P-4)
control over historical development; and (P-5) role of chance. Each of these variables is
described in greater depth elsewhere in Chapter 4 of this volume.
We took measurements of each president’s psychological characteristics in con-
junction with each case. In order to avoid the endogeneity problem,4 we took
these “snap-shots” in the 90 days prior to the onset of each case. We gathered all
spontaneous verbal material for each president during that 90-day window and
used those materials, in conjunction with the content-analysis software program
Profiler Plus, to ascertain each president’s score on each of the nine psychological
characteristics for that specific case.

Hypotheses
With nine different independent variables and four different dependent variables,
space limitations prohibit us from extensive theoretical discussions pertaining to all of
our hypotheses. We do not have hypotheses for every combination of independent
and dependent variables; not all of our psychological traits should be predictors of
each of our dependent variables. Our study nonetheless includes 34 different hypoth-
eses. Table 7.1 provides the directional summary of each of these. Here we provide
 Leader Characteristics and Small Group Dynamics 117

TABLE 7.1 Summary of Directional Hypotheses

Dependent Variables
Independent Variables Group Structural Decision Processing National Level of
Faults Faults Interests Conflict

Need for Power – – + +

Distrust + + – +
Task Focus – – + none
Conceptual Complexity – – + –
Belief in Ability to Control + + – +
P-1 Nature of Pol. Universe – – + –
I-1 Approach to Goals – – + –
P-4 Control + + – +
P-5 Role of Chance + + – none

a succinct discussion of them, using each psychological characteristic as the basis for
discussing its hypothesized relationship with the dependent variables.
• Need for Power. This variable is about successfully getting things done in politics.
This means several things in terms of our dependent variables. We expect that
leaders high in need for power will have more effective decision-making proce-
dures, thus having fewer group-structural faults and fewer decision-processing faults.
We also expect that these leaders will produce better outcomes in terms of
national interests. High need for power is also associated with forcefulness, and there-
fore we expect these leaders to have outcomes that increase the level of conflict.
• Distrust. Leaders who have the psychological characteristic of distrust may be
reticent to seek or believe information coming from outside sources. They
also may not adequately rely on others, delegate tasks to others, or trust the
information provided by others within the advisory system. All of this is
likely to contribute to poorer decision making in the form of more group-
structural faults and more decision-processing faults. And, as a result, we expect
outcomes to be poorer in terms of national interests. Not surprisingly, we
expect that those scoring higher on distrust are likely to have higher levels of
conflict associated with their decisions.
• Task Focus. We expect that those with a psychological orientation on tasks
will create and make use of the kinds of decision-making procedures that
have been shown to be effective in the past (Haney 1997; Janis 1972, 1982;
Kowert 2002; Schafer and Crichlow 1996, 2002). Therefore, we hypothesize
that those with higher task focus will have fewer group-structural and decision-
processing faults. We also expect that their focus on the task at hand will enable
them to more successfully accomplish national interests. We do not have any
theoretical expectations regarding task focus and level of conflict.
• Conceptual Complexity. People higher in complexity are thought of as infor-
mation seekers and open to alternative explanations and possibilities. Because
118 Mark Schafer, Jonathon Nunley, and Scott Crichlow

information seeking is an important component of quality decision making,

we expect that leaders high in conceptual complexity will have fewer group-struc-
tural and decision-processing faults. We also expect that they will have better
outcomes in terms of national interests, as they will be open to considering
more alternatives, and thus may be more likely to find the best one pos-
sible. Further, based upon work by Margaret Hermann (1980), and tied to
the notion that high complexity individuals will consider a larger range of
options, we expect they will have lower conflict levels.
• Belief in Ability to Control Events. Those who have a psychological predis-
position to control things are less likely to be open to several factors that
contribute to good-quality decision making, such as input from a variety of
sources, conflicting information, and competing points of view. This leads
us to expect that those high in this characteristic will have higher levels of
group-structural and decision-processing faults. This in turn is likely to produce
poorer outcomes in terms of national interests. In addition, as part of being a
controller means wanting things one’s own way, we expect these individuals
to have higher levels of conflict associated with their foreign policy decisions.
• (P-1) Nature of the Political Universe. Those scoring higher on this characteris-
tic see the world in more positive terms. We believe this propensity will lead
these leaders to being more open in the decision-making process and thus
having fewer group-structural and decision-processing faults. Similarly, a coopera-
tive view of the world is also likely to lead these leaders toward lower levels
of conflict. Those with a more positive view of the political universe may see
more options that can facilitate national interests.
• (I-1) Strategic Approach to Goals. Here the focus is on self’s orientation toward
conflict and cooperation, but the logic is similar to that for P-1: those with a
more cooperative orientation should have fewer group-structural and decision-
processing faults. In addition, as with P-1, we expect those with a more coop-
erative strategic approach to goals to have better outcomes in terms of national
interests and lower levels of conflict.
• (P-4) Control over Historical Development. This is another variable that deals
with the subject’s control orientation, this one coming from the operational
code; the logic for our hypotheses here is the same as for those associated
with Belief in Ability to Control Events above. We expect those scoring
high on this characteristic to have higher levels of group-structural and decision-
processing faults, poorer national-interest outcomes, and higher levels of conflict.
• (P-5) Role of Chance. We expect that those with a more fatalistic view on life
are less likely to try to set up careful decision-making processes; after all, if
chance is what matters then internal decision-making structures are less likely
to affect situations for the better. Therefore, we hypothesize that leaders scor-
ing high on this characteristic will have higher levels of group-structural and
decision-processing faults. We also expect that leaders scoring high on this will
leave too many things to chance and not work effectively to advance national
 Leader Characteristics and Small Group Dynamics 119

interests. We do not have a directional hypothesis with this characteristic for

the Conflict variable.

Analysis
We first present simple tests of each of our 34 hypotheses in the form of OLS bivar-
iate regression models. These test the effect of one psychological characteristic on
one dependent variable at a time, thus giving us 34 different hypothesis tests. Later
in this section we present more fully specified multivariate models. The results of
the bivariate tests are presented in Table 7.2.
We begin our discussion of these results by considering each of the different
psychological characteristics in turn, and later turn to a brief discussion of the
broader pattern of results. Need for power, as hypothesized, is a good predictor of
better decision-making procedures. The model is significant regarding structures,
meaning that higher need for power in the president predicts fewer group-structural
faults in his decision-making apparatus.5 This motive also has the expected direc-
tion for information-processing faults, though the latter is not significant. Need for
power is also not significant for either of the two outcome variables.
Distrust is significant in the hypothesized direction for all four dependent vari-
ables. Those leaders who are more distrusting will have higher numbers of group-
structural and decision-processing faults. They are also more likely to have outcomes
that result in poorer national interests and higher levels of conflict. Indeed, distrust is by
far the best predictor variable among our psychological traits. When the leader is
high in distrust, it appears to start a syndrome of problems: decision-making is short-
circuited, conflict options become more likely, and national interests are hindered.
Those leaders who are high on task focus are likely to have better decision-mak-
ing procedures as seen by their lower scores on group-structural faults. The same
directional pattern is there regarding decision-processing faults, but that model is not

TABLE 7.2 Bivariate Hypothesis Tests*

Dependent Variables
Independent Variables Group Structural Decision Processing National Level of
Faults Faults Interests Conflict

Need for Power –0.27 (0.06) –.13 (.24) .10 (.29) –.02 (.45)
Distrust .46 (.006) .50 (.002) .29 (.10) .38 (.03)
Task Focus –.23 (.10) –.15 (.20) .27 (.06) none
Conceptual Complexity –.02 (.45) .23 (.09) –.16 (.19) .36 (.02)
Belief in Ability to Control .03 (.43) –.24 (.09) –.08 (.33) –.12 (.25)
P-1 Nature of Pol. Universe 14 (.22) –.19 (.14) .10 (.29) –.06 (.37)
I-1 Approach to Goals –.15 (.20) –.02 (.46) .16 (.18) –.04 (.42)
P-4 Control –.23 (.09) –.02 (.45) .02 (.46) .02 (.46)
P-5 Role of Chance .32 (.03) .12 (.25) –.27 (.06) none

* Cell values are Pearson’s R correlation coefficients with one-tailed p values in parentheses.
120 Mark Schafer, Jonathon Nunley, and Scott Crichlow

significant. Also as hypothesized, higher task focus results in better outcomes in

terms of national interests. Having a high task focus appears to be a good thing for
foreign policy decision making.
Conceptual complexity and belief in ability to control events did not produce any
models that supported our hypotheses. In fact, some models are in the opposite
direction of the one anticipated. Though conceptual complexity is conceptually asso-
ciated with higher levels of information processing, here it did not correlate with
better decision-making procedures, and it actually correlated with higher levels of
decision-processing faults. And, while we expected higher levels of conceptual complex-
ity to correlate with lower levels of conflict, the bivariate test showed just the oppo-
site, higher complexity correlated with more conflict. Finally, we expected that
high scores on belief in ability to control events would lead to poorer decision-making
processes, but our model with decision-processing faults as the dependent variable
actually goes the other direction: higher scores correlate with fewer faults.
The data for the operational code indicators are mixed. When the president has
more cooperative views on P-1, nature of the political universe, his administration is
likely to have fewer decision-processing faults, though that relationship only approaches
statistical significance. A similar pattern is seen for I-1, strategic approach to goals. Both
of the decision-making variables are in the expected direction—more cooperative
I-1 scores correlate with fewer decision-making faults—but in both cases the mod-
els only approach significance. I-1 also correlates, though again it only approaches
significance—with outcomes in terms of national interests: more cooperative scores
on I-1 for the president means national-interest outcomes may be advanced.
Our second variable pertaining to the leader’s control orientation is P-4 control
over historical development. And, as with belief in ability to control events above, this
variable does not provide any support for our hypotheses. In fact, only one of the
four models is significant, group-structural faults, and that model is in the opposite
direction of the one anticipated: higher control orientations result in fewer group-
structural faults.
Our final psychological characteristic, P-5 role of chance, provides support for
two of the hypotheses. Those leaders high in role of chance are less likely to have
effective decision-making structures: higher scores on P-5 predict more group-
structural faults. In addition, such fatalistic propensities seem to have a negative
effect on outcomes: when the President scores high on P-5, national-interest out-
comes are more likely to be hindered than advanced.
In looking over the significant models in Table 7.2, some patterns emerge.
First, the LTA indicators as a set correlated more often with the dependent vari-
ables than did the operational code indicators. This would seem to indicate that
psychological characteristics at the more unconscious level have more of an effect
on decision-making processes and outcomes than do those that are more cogni-
tive in nature. Of the LTA indicators, distrust and task focus are the best predictors
of our dependent variables. Those who are distrusting are likely to have poorer
decision-making procedures and outcomes marked by higher levels of conflict
 Leader Characteristics and Small Group Dynamics 121

and poorer results in terms of national interests. Those with a high task focus are
likely to have better group structures in their decision-making process and better
outcomes in terms of national interests. Of the operational code indicators, P-5 role
of chance is the best predictor. Those who view the role of chance as high are likely to
have more group-structural faults and poor outcomes in terms of national interests.
Second, psychological variables are more likely to affect the process of making
decisions than the outcomes of those decisions. In other words, more of these
bivariate models were significant for the two process variables, group structures
and decision processing, than they were for the outcome variables regarding national
interests and level of conflict. This pattern makes some sense because as we get further
down the causal chain, more and more factors come into play that can affect out-
comes. While many different factors can affect the process of decision making, it
is still a process that is internal to the state. The same is not the case for outcomes,
which depend at least in part on what other countries do. Presidential psychology
affects both of these dependent variables, but it has less of an effect when other
major factors come into play, such as the words and deeds of an opposing state.
Similarly, of the two decision-processing dependent variables, it is not surprising
that psychological characteristics have a greater effect on group structures than on deci-
sion processing. Group structures are set up before the onset of a decision episode; it is
in this area that a leader has the most time, influence, and input—things that may
greatly affect structures. Once again, as we move further down the causal chain the
leader has less direct control and influence: there are time constraints, limitations
on available information, behavioral vagaries of area experts, intelligence officers,
and other advisors and bureaucrats, and other factors outside of the influence of the
president. A president’s psychology still matters during decision processing, just less so
than when he is setting up the group structures in his administration.6

Multivariate Models
For many reasons it is appropriate to consider multivariate models involving dif-
ferent combinations of our psychological characteristics to explain our depend-
ent variables. Political psychology is always complicated and always multi-faceted.
While it would be impossible to capture all psychological elements in a multivari-
ate analysis, it is possible with our data to run models that include multiple facets.
We proceed in this section by considering models pertaining to each different
dependent variable. There are many correlations among our psychological vari-
ables (more on that below), meaning that statistically speaking the most efficient
model is unlikely to include all of them. Following good social science practice,
for each dependent variable we also include below a “fully specified” model that
includes all of the characteristics for which we have directional hypotheses. It is also
not practical to include all different combinations of independent variables that we
tested. Instead, we present below (in addition to the full model) one that we term
the “best” model. Generally speaking, this one is the most efficient of the models
122 Mark Schafer, Jonathon Nunley, and Scott Crichlow

we tested, as marked by the adjusted R2 and a set of independent variables that are
significant or approach significance. In addition, for each dependent variable we
discuss some of the other patterns we saw when running our multivariate tests.
Our first dependent variable is group-structural faults. The results in Table 7.3
show that both distrust and need for power are significant in the full model. Higher
levels of distrust result in more group-structural faults, while higher need for power
in the President results in fewer faults; both of these findings are in the direction
anticipated by our hypotheses. In addition, P-5 role of chance approaches statistical
significance in the expected direction. The only other variable that approaches
significance is P-1 nature of the political universe, but its sign is in the opposite direc-
tion of the one anticipated.
The best model explaining group-structural faults includes three of our psycholog-
ical characteristics: distrust and P-5 role of chance are significant in the hypothesized
direction, and need for power also has the correct sign, though it only approaches
significance. The model has a better adjusted R2 than the full model, and explains
32% of the variance in group-structural faults. This is a substantively significant find-
ing: knowing a leader’s scores on these three psychological characteristics alone
explains nearly 1/3 of the variance in problems associated with the structure of an
administration’s decision-making procedures.
Turning now to decision processing, the two models for this dependent variable
are in Table 7.4. Only distrust is significant in the fully-specified model. Once
again, we see the effect that a higher level of distrust by the President results in
more decision-processing faults. The two psychological characteristics that tap into
control orientation, belief in ability to control events and P-4 control over historical devel-
opment, both approach statistical significance but not in the anticipated direction.

TABLE 7.3 Multivariate Models Explaining Group Structural Faults

“Full” Model “Best” Model

Independent Variables B SE Beta p B SE Beta p

Need for Power –21.83 16.42 –0.31 0.10 –14.04 11.26 –0.20 0.11
Distrust 2667.53 1014.27 0.49 0.01 2389.08 826.29 0.44 0.00
Task Focus –1.63 10.95 –0.03 0.44
Conceptual 1.02 16.70 0.01 0.48
Complexity
Belief in Ability to –1.35 15.31 –0.02 0.47
Control
P-1 Nature of Pol. 8.03 6.54 0.35 0.12
Universe
I-1 Approach to –0.43 4.90 –0.02 0.47
Goals
P-4 Control 0.39 15.83 0.01 0.49
P-5 Role of Chance 72.27 61.60 0.38 0.13 41.10 29.75 0.22 0.09
R2 = .39, adjR2 = .16 R2 = .32, adjR2 = .26
 Leader Characteristics and Small Group Dynamics 123

TABLE 7.4 Multivariate Models Explaining Decision Processing Faults

“Full” Model “Best” Model

Independent Variables B SE Beta p B SE Beta p

Need for Power 2.24 13.31 0.04 0.43

Distrust 2084.68 822.46 0.49 0.01 2154.74 656.60 0.50 0.00
Task Focus –3.82 8.88 –0.08 0.34
Conceptual 7.36 13.54 0.12 0.30
Complexity
Belief in Ability to –12.85 12.42 –0.26 0.16
Control
P-1 Nature of Pol. –0.63 5.31 –0.04 0.45
Universe
I-1 Approach to –2.39 3.97 –0.14 0.28
Goals
P-4 Control –11.53 12.84 –0.27 0.19
P-5 Role of Chance –16.75 49.95 –0.11 0.37
R2 = .34, adjR2 = .09 R2 = .25, adjR2 = .23

It is a bit unusual, but in this case the best model that explains decision-process-
ing faults includes only one variable, distrust. In none of the other multivariate
models we tested for this dependent variable did any of the other psychological
characteristics reach significance when distrust was in the model, and no model
that excluded distrust explained any significant amount of variance. In addition,
no additions of one or more of the other variables resulted in a better adjusted R2
than the one with distrust alone. In this case a single variable explains over 25% of
the variance in decision-processing faults, another substantively significant finding.
Some of the other variables had consistent signs in the hypothesized direction
and sometimes approached statistical significance; these include need for power, task
focus, P-1 nature of the political universe, I-1 strategic approach to goals, and P-5 role of
chance. However, in no case did any of these variables reach significance or make
any contribution to explained variance.
The models explaining outcomes in terms of national interests are presented in
Table 7.5. Several of the psychological variables are significant in the full model
in the hypothesized direction, all but one of them from the operational code.
Having a more positive view on politics and political others (P-1 nature of the
political universe) results in worse national-interest outcomes, which is contrary to
our hypothesized relationship. When the locus of control (P-4 control over historical
development) is oriented more toward self than others, national interests tend to turn
out more poorly. And when the President’s belief that the role of chance is higher,
outcomes tend to be poorer for national interests. Both of these results conform to
our hypotheses.
Need for power, distrust, and belief in ability to control events all have the hypothe-
sized signs, but these relationships with national interests only approach our level of
124 Mark Schafer, Jonathon Nunley, and Scott Crichlow

TABLE 7.5 Multivariate Models Explaining National Interests

“Full” Model “Best” Model

Independent Variables B SE Beta p B SE Beta p

Need for Power 6.55 5.49 0.29 0.12

Distrust –309.33 339.03 –0.18 0.19 –366.80 310.30 –0.21 0.12
Task Focus 2.06 3.66 0.11 0.29 3.39 3.29 0.18 0.16
Conceptual –8.91 5.58 –0.34 0.06
Complexity
Belief in Ability to –4.40 5.12 –0.22 0.20
Control
P-1 Nature of Pol. –3.34 2.19 –0.46 0.07
Universe
I-1 Approach to 1.04 1.64 0.15 0.27
Goals
P-4 Control –8.95 5.29 –0.51 0.05
P-5 Role of Chance –42.90 20.59 –0.71 0.02 –14.06 10.11 –0.23 0.09
2 2
R = .34, adjR = .09 R = .17, adjR2 = .09
2

statistical significance. Conceptual complexity is significant, but not in the hypoth-

esized positive direction. This result suggests that those leaders with lower lev-
els of conceptual complexity produce better outcomes in terms of national interests.
However, our hypothesis was the opposite, and we have a hard time developing a
theoretical justification that says simpler thinkers are more likely to produce bet-
ter outcomes. Our one-tailed directional hypothesis test suggests that we simply
reject the finding because it has the incorrect sign.
The most efficient model explaining national-interest outcomes includes three
variables: distrust, task focus, and P-5 role of chance. Only the latter is significant in
the model with the other two variables approaching significance. Although this
model is more efficient because it uses only three independent variables, this time
the more efficient model does not improve upon the full model’s adjusted R2.
Even if there is some noise in the full model, in this case it may be preferred over
the more efficient model. The full model explains 34% of the variance in national-
interest outcomes, though some of that is artificially elevated because of the large
number of independent variables in the model.
Some other variables were consistent in the multivariate models we tested regard-
ing national interests, though they did not reach levels of statistical significance. For
instance, higher scores on I-1 strategic approach to goals tended to result in more posi-
tive national-interest outcomes. The sign for conceptual complexity remained negative,
opposite from what we anticipated, in all of the multivariate models we tested for
this dependent variable. Both of the variables that tap into control orientation also
had negative signs across the multivariate models, indicating that higher control
orientation is less productive for national interests, which fit our hypotheses but only
 Leader Characteristics and Small Group Dynamics 125

TABLE 7.6 Multivariate Models Explaining Level of Conflict

“Full” Model “Best” Model

Independent Variables B SE Beta p B SE Beta p

Need for Power 0.10 3.55 0.01 0.49

Distrust 470.28 226.50 0.37 0.02 494.52 211.74 0.39 0.01
Conceptual 7.99 3.98 0.42 0.03
Complexity
Belief in Ability to 0.46 3.40 0.03 0.45
Control
P-1 Nature of Pol. 0.58 1.16 0.11 0.31
I-1 Approach to Goals –1.24 1.13 –0.25 0.14 –0.41 0.84 –0.08 0.31
P-4 Control –1.04 2.61 –0.08 0.35
R2 = .29, adjR2 = .10 R2 = .15, adjR2 = .10

approaching statistical significance. Of the two, the one associated with the opera-
tional code construct performed slightly better than the one from LTA.
Our final models in Table 7.6 are for the dependent variable level of conflict.
In the full model,7 distrust is once again significant in the expected direction:
higher levels of distrust predict high levels of conflict. Strategic approach to goals (I-
1) approaches significance in the hypothesized direction—more cooperative I-1
scores correlate with lower levels of conflict. Conceptual complexity is significant,
but not in the hypothesized direction. None of the other variables approach sig-
nificance in the full model. The most efficient model achieves the same adjusted
R2 using only two variables, distrust and (I-1) strategic approach to goals. Distrust is
highly significant in the model and, while I-1 is not significant, including it in the
model increases explained variance. It contributes not only to the overall signifi-
cance of the model, but also to the effect of distrust in the model.
Generally speaking, distrust is the key predictor of level of conflict among our cases.
Of the two key conflict-cooperation beliefs indicators associated with the opera-
tional code, I-1 generally performed better than P-1, which makes sense given that
it taps into the leader’s own strategic orientation. However, neither of these two
operational code variables was significant in our models, and only I-1 approached
significance in one of them. The two control-orientation variables were not signifi-
cant and one (P-4) had a sign opposite from the positive one we expected.
Reviewing these multivariate models overall, some patterns emerge. First, dis-
trust continues to dominate the models, and in some cases is so dominant that it
alone is the only psychological variable that matters. Second, as with the bivari-
ate models discussed above, the psychological characteristics generally performed
better for predicting and explaining the two decision-making dependent variables
rather than the two outcome dependent variables. Again, this makes sense given
that the outcome variables are further down the causal chain allowing for more
extraneous variables to intervene and have an effect. Finally, the key variable
126 Mark Schafer, Jonathon Nunley, and Scott Crichlow

here, distrust, is from the LTA research program; however, the operational code
variables have an impact in several of the models with (P-5) role of chance and (I-1)
strategic approach to goals having more of an impact than the others.
Some key puzzles remain. For instance, why did conceptual complexity and the
two control-orientation indicators perform poorly in our models? What is the
relationship between some of the psychological characteristics? Are the traits
from LTA associated with our four operational code indicators? While our data
will not let us solve these puzzles once and for all, it is possible to explore some
possibilities. Namely, we can look at some of the bivariate correlations between
the different psychological characteristics, which we undertake in the following
section.

Psychological Characteristics: Some Correlations

We begin here with what is perhaps the most obvious correlation to investigate, the
two psychological characteristics that purport to tap into the leader’s control orienta-
tion, belief in ability to control events and (P-4) control over historical development. Neither
performed well in our models, but how well do they correlate with each other? The
expectation is that they should be positively and significantly correlated. In fact, they
are not. The bivariate correlation is negative and significant (R = –.388,8 p = .0239).
While they are not identical in terms of their conceptual definitions, and certainly
not in their operationalizations, it seems problematic to us that they are negatively
correlated. And yet, it also seems to us that the president’s control orientation, con-
ceptually speaking, ought to affect the administration’s decision-making process and
perhaps have an effect on the outcomes of decisions. It is beyond the scope of this
chapter to try to sort the underlying issues here, and perhaps the problem is idi-
osyncratic to our sample. But this area is worthy of additional research and perhaps a
change in the operationalization of one or both variables may be in order.
What do we make of the performance of conceptual complexity in our models?
Unfortunately, the bivariate correlations among the psychological characteristics
do not help us much here. Conceptual complexity is positively correlated with (P-4)
control over historical development and negatively correlated with belief in ability to con-
trol events, but we have already problematized those two variables leaving us with
no real insight. Complexity is also positively correlated with (I-1) strategic approach
to goals (R = .378, p = .027), which means that those with higher conceptual com-
plexity tend to have a more cooperative strategic approach to goals. This relationship
fits with our broad directional expectations, but it does not give us much insight
on why conceptual complexity was never significant in our models in the expected
direction and often had a sign that was opposite the one we expected.
It is possible that the performance of this variable is related to the particular
sample in our dataset. For instance, President George W. Bush scores second
highest in our sample on conceptual complexity, and yet his administration was at
the top of the list in terms of average number of group-structural faults and decision-
 Leader Characteristics and Small Group Dynamics 127

processing faults. Given that five of our cases are from the G. W. Bush administra-
tion, this kind of outlier pattern seems certain to have affected the results.
So why did George W. Bush score so high on conceptual complexity? Few politi-
cal observers would characterize him as a more complex thinker than others in
our sample such as Carter, Clinton, or G. H. W. Bush. Perhaps the operationali-
zation of complexity, which we discussed above, sometimes taps into something a
bit different than originally intended. It seems possible to us that a leader who uses
some of the “high” complexity words, such as maybe, possibly, or perhaps, may be
showing high levels of equivocation, perhaps indicating such things as confusion,
hesitation, or a sense of being overwhelmed. It is a complicated world out there.
Some leaders might recognize that and yearn for more input and information;
others might simply feel uncertainty or bewilderment. This is another area that
may be ripe for additional research.
Are there certain psychological traits (from LTA) associated with various indica-
tors of the operational code? Yes. In fact, need for power correlates with all four of the
operational code indicators we investigated here. Higher scores on need for power cor-
related with more cooperative views on the nature of the political universe (P-1), a more
cooperative strategic approach to goals (I-1), higher scores on control over historical devel-
opment (P-4), and lower scores on role of chance (P-5). There may be an inclination
to associate need for power with forcefulness and combativeness, but our data suggest
otherwise: a high need for power correlates with more cooperative orientations for the
two conflict-cooperation indicators associated with the operational code construct.
Along with some of our other research elsewhere (Schafer and Crichlow 2010),
this association leads us to place more emphasis on the role of need for power and its
connection to effectiveness. People with a high need for power seem to be those who
get things done, regardless of what it takes, and in international politics that may fre-
quently mean being cooperative. Finally, only one other psychological characteristic
from LTA correlates with our operational code indicators in this sample of leaders.
As noted above, conceptual complexity correlates positively with (I-1) strategic approach to
goals and also correlates positively with (P-4) control over historical development.
Let us look now at the most prominent variable among our psychological char-
acteristics. As discussed above, distrust has a significant effect virtually everywhere
we tested it. Its effect was so powerful that it frequently swamped the effect of
other variables in our multivariate models. And yet, distrust correlates significantly
with only one other characteristic, which is task focus. The correlation is negative,
meaning that higher levels of distrust are associated with less task focus (R = –.322,
p = .063). Clearly it is not the case that distrust kept other variables from being sig-
nificant in our models because of multicollinearity (except perhaps for task focus).
Instead, it appears that it is a powerful predictor variable: high levels of distrust in
a leader are likely to lead to more faults in group structures, more faults in decision
processing, poorer outcomes in terms of national interests, and higher levels of conflict.
Other psychological characteristics matter also, either on their own or in some
multivariate models with distrust, but none have the effect that distrust does.
128 Mark Schafer, Jonathon Nunley, and Scott Crichlow

Conclusion
We began by asking if leaders—and by extension their psychological characteris-
tics—have an effect on the conduct of foreign policy in their administration. The
answer to that question, based upon the data presented here, is yes. We have found
that psychological characteristics from both the unconscious (traits and motives)
and the conscious (beliefs) sides of personality affect foreign policy making. So far
as we know, this study is the first one that looks at such a broad range of psycho-
logical variables. Not surprisingly, the psychology of the leader has a stronger effect
on the structures and processes of making foreign policy than it does on the out-
comes associated with foreign policy decisions. As we have argued, the latter are
further down the causal chain and are certain to be affected by many other outside
variables.
Nonetheless, we find that even outcomes are affected by the psychology of
the president. Of the different psychological characteristics we investigated in
this chapter, distrust is by far the most important, and had an effect on virtually
all four of our dependent variables. However, other psychological characteristics
mattered as well. Higher scores on need for power and task focus are associated with
better group structures and better national-interest outcomes, while higher scores on
(P-1) nature of the political universe (P-5) role of chance correlated with better group
structures and worse national-interest outcomes. A more cooperative (I-1) strategic
approach to goals is moderately related to lower levels of conflict while a higher con-
trol orientation, as marked by (P-4) control over historical development, seems to have
a harmful effect on national interests.
Some of our psychological variables did not generate the results we hypoth-
esized, specifically conceptual complexity and the two control-orientation variables
belief in ability to control events and P-4 control over historical development. These anom-
alies do not necessarily mean that these concepts do not have an effect on foreign
policy making. It may be that results are idiosyncratic to the sample; it may be
that the underlying operationalizations need some adjustment; it may be that the
conceptualization of the variable leaves room for bi-polar effects, such as those we
have speculated about regarding George W. Bush’s complexity score.
It is important to note some cautions and limitations regarding our findings.
Perhaps the most important is our limited sample. Although we have 34 cases in
the sample, we have only six leaders. Our research design accounts for possible
changes in beliefs and traits of the leaders over time, making the appropriate unit
of analysis the case, but the fact remains that we have a small pool of subjects. In
addition, the subjects are all former U.S. presidents, and we caution about gener-
alizing outside of this population. For example, though small, we did find that the
psychological characteristics of the president had an effect on outcomes. Given
the relative influence of the United States—and therefore its president—during
this time period, it is not hard to imagine that personality effects on outcomes
could be smaller in a broader sample of world leaders.
 Leader Characteristics and Small Group Dynamics 129

We have also considered only a limited number of psychological characteristics

in this study. There are other traits from LTA and other beliefs from the opera-
tional code construct, plus there are many other psychological constructs outside
of these two research programs that might matter. We chose the variables in this
study because they seemed to have the best possible theoretical connections to
our dependent variables among those available to us. Indeed, as noted, this study
includes a larger number and a broader range of psychological characteristics than
past studies. Current methods place some limitations on what other motives,
traits, and cognitions might be considered, but it is certainly a worthwhile area
for future work.
These cautions and limitations notwithstanding, our research contributes to
understanding the importance of the microfoundations of foreign policy making.
We do not argue that psychology is all that matters; indeed, some of our other
research shows that at least some situational factors also affect both the quality of
decision making and the outcomes of decisions, most notably the power differ-
ential between actors in a case (Schafer and Crichlow 2010). But for those who
argue that states are unitary actors, that power and interests are the key explana-
tory variables, and that the psychology of leaders does not matter, the findings
here paint a very different picture. The leader’s psychology matters, most notably
in affecting the quality of the decision-making process, but also in affecting out-
comes of that process. While there is room for much more research in these areas,
for now we can say with confidence that models leaving psychological variables
out of the analysis of foreign policy decision making are leaving out an important
explanatory source of variance.

Notes
1 The LTA construct includes one motive, need for power, among its variables.
2 For additional discussion of these variables, see Schafer and Crichlow (2010).
3 The software program used for the present research is Profiler Plus. See http://
socialscience.net/Default.aspx for more information on this program.
4 The endogeneity problem comes about when the dependent variable may be having
an effect on the purported independent variable, thus making it impossible to deter-
mine the correct direction of causation in the regression model.
5 Because all of our hypotheses are directional, we report one-tailed tests of significance
in our tables. Our threshold for assessing significance is when p<.10. We also note
those models that “approach” significance when .10>p<.20. The small sample size
justifies the use of these p values, though in all tables we report actual p-value estimates
so that readers may make their own interpretation.
6 This finding is consistent with George’s (1980) argument that a leader’s personality
influences decisions indirectly by affecting the structure of the advisory system.
7 Recall that we do not have directional hypotheses for two of our psychological char-
acteristics for level of conflict (task focus and P-5 role of chance), so the full model only
includes seven independent variables.
8 All correlation coefficients reported here are Pearson’s R.
9 Since these correlations are not formal hypothesis tests, and in some cases we do not
have directional expectations, here we report two-tailed tests of significance.
8
ALLIANCES AND THEIR
MICROFOUNDATIONS
France and Britain in the 9/11 Era*

Akan Malici

Introduction
“It was the best of times, it was the worst of times,” begins Charles Dickens’s
A Tale of Two Cities, set in Paris and London. Given the spat between French
President Jacques Chirac and British Prime Minister Tony Blair over the U.S.
decision to invade Iraq, the decision-making elite in both countries could be
excused for calling to mind the second half of Dickens’s introduction as they
reflected on their neighborly ties that were stretched over the issue of Iraq. On
the eve before military action of allied forces in Iraq and after a meeting between
Jacques Chirac and Tony Blair, the former remarked: “We talked about Iraq,
and our approach is not the same,” while the latter confirmed, “There are differ-
ences,” regarding the forthcoming U.S.-led operation, dubbed Iraqi Freedom.1
Disagreement between France and Britain is nothing new, though not an
iron law. After the tragic terrorist attacks of 11 September 2001, U.S. President
George W. Bush split the international community into “us” and “them” (Bush
2001). Along with the United States both France and Britain saw themselves as
major players within the “coalition against terrorism,” as forces of the “us,” and
considered themselves to have the same approaches and goals towards the blurry
defined “them” in Operation Enduring Freedom. This time Jacques Chirac and
Tony Blair declared that they were in “complete agreement.”2
What makes the relationship between France and Great Britain interesting and
important is that they are both key members of three multilateral organizations,
namely the North Atlantic Treaty Organization (NATO), the Security Council
of the United Nations (UN), and the European Union (EU). Concord or discord
between France and Great Britain within these organizations is an important
factor for the future viability, credibility, and effectiveness of NATO, the UN,
and the EU regarding questions of international and regional security.
 Alliances and Their Microfoundations 131

The research questions guiding this investigation on the most general level are:
Why did France and Britain cooperate in Operation Enduring Freedom after
11 September 2001, and why did they fail to achieve cooperation in Operation
Iraqi Freedom starting 19 March 2003? The main assumption here is that the
mechanisms for cooperative or non-cooperative behavior between France and
Great Britain—and by extension the mechanisms for the viability of the afore-
mentioned international organizations—are located in the belief systems of the
decision-making elite of these countries.
An examination of these questions is important. The events leading to the end
of the cold war created a theoretical vacuum. Conventional international relations
theories—and here most prominently realism—found themselves unable to pro-
vide satisfactory explanations regarding these events (Wohlforth 1998). Scholars
have argued that the reason for this failure is that realists, and for that matter all
structural theories, do not theorize preferences to a sufficient degree, “which
makes their analysis inherently incomplete” (Frieden 1999: 50). Indeed, in much
of realist theorizing actors are treated “like units” and their preferences and causal
beliefs are taken for granted, as they are defined only by reference to the exter-
nal environment (Waltz 1979: 79–101; Goldstein and Keohane 1993: 4). Such
approaches are characterized by “situational determinism” (Wendt 1987: 342).
They cannot explain why two states, facing the same environment, would adopt
diverging strategies as was the case with France and Great Britain in Operation
Iraqi Freedom.
“By taking preferences as given,” as Robert Jervis (1988: 324–325) summa-
rizes, “we beg what may be the most important question on how they were
formed.” Analysis is facilitated, “but at the cost of drawing attention away from
areas that may contain much of the explanatory ‘action’ we are interested in.”
The tragic events of 11 September 2001 also suggest this necessity (Brown 2002;
Tow 2003). Perhaps now more than ever scholars ought to realize that an under-
standing of international politics is impoverished if we do not sufficiently consider
the impact of ideational variables, such as the perceptions and beliefs of national
decision-making elites.
In order to advance this argument, I take the literature on group behavior as
a starting point. This literature first originated in the disciplines of sociology and
social psychology. It came then to be considered relevant by international rela-
tions scholars for analyses of state behavior and the viability of larger groups, such
as international alliances, to which these states belong. One of the most promi-
nent propositions within this literature is that common perceptions of external
threat lead to cooperation among states belonging to the same group. Although
most of the literature was directed toward questions of cooperation and discord
among members of an alliance, there is no reason to exclude other groupings,
such as international institutions, operating within the international system from
the applicability of the underlying logic of the argument. I will begin by briefly
reviewing the literature advancing this proposition. In subsequent sections I
132 Akan Malici

compare and contrast the beliefs of the French and British leadership through a
Multivariate Analysis of Variance (MANOVA), and then compare the strategic
orientations of these leaders with the aid of a theory of inferences about prefer-
ences. The conclusion draws attention to the importance of microfoundations
in examining a state’s foreign policy behavior and the viability of international
organizations.

The External Threat—Cooperation Proposition

Group behavior among sovereign states, between alliance members, or within
international institutions is one of the most fundamental processes of international
politics and security. The emphasis on sovereignty is important because it indi-
cates that the viability of coalitions is conditioned by the choices of individual
member states with far-reaching consequences. The choices “that bring … states
together and drive them apart,” as Stephen Walt (1987: 1) rightly asserts, “will
affect the security of individual states by determining both how large a threat they
face and how much help they can expect.” At the same time these choices also
determine the viability and effectiveness of coalitions in the international system
and its evolution.
If alliances and international institutions are indeed keys to an understanding of
international politics, then it becomes important to understand what brings states
together and drives them apart. A prominent proposition in the field of group
dynamics and conflict processes is the idea that the threat from external conflict is
highly conducive to group solidarity and cooperation towards a commonly defined
goal (Thompson and Rapkin 1981: 615; Stein 1976). Conversely, as common
perceptions of threat decrease, group cohesion and cooperation is expected to
give way to “more normal patterns of interaction” among the members of the
group (Thompson and Rapkin 1981: 615).
The origins of the hypothesis asserting a positive relationship between an
increase of threat perception and group cohesion can be found in the socio-
logical and social-psychological literature (Sherif and Sherif 1953; Coser 1956;
Sherif et al. 1961; Dahrendorf 1959, 1968). Half a century ago the sociologist
George Simmel (1955) argued that external conflict leads to increased levels of
cooperation because the exigencies of conflict require political conformity. Ralf
Dahrendorf elevated the proposition to the level of a general law arguing that
“It appears to be a general law that human groups react to external pressure by
increased internal coherence” (cited in LeVine and Campbell 1972: 31).
A few years later the issue spilled into the discipline of social psychology where
Hamblin (1958: 68) asked why cooperation between members of a larger group
increases during times of crisis. He suggests a similar answer to that of Simmel
arguing that “In every crisis situation … [there] is a likely solution to the crisis
problem—a solution that requires the cooperation of all or most of the members of
the groups involved.” Cooperation, therefore, becomes an instrumental function
 Alliances and Their Microfoundations 133

originating from the sharing of common threat and common suffering (Fritz and
Williams 1957: 48). Lott and Lott (1965: 266) put it the following way:
Attraction among individuals will be found to increase when their common
threat stems from an external source (i.e., is not a function of their own lack
of skill), when there exists the possibility that cooperative behavior may
reduce or eliminate the threat, and when single individuals cannot escape
from either the group or the threat.
Much of the evidence was brought forth in experimental studies.3 In an
early study Lanzetta et al. (1954) found that subjects under the threat of potential
punishment are more sociable, cooperative and friendly towards each other while
working on a specific task than are groups working under no threat. Schachter
(1959) substitutes the cooperation variable with a measure of an individual’s need
for affiliation. His experimental study led him to the conclusion that high-anxiety
subjects (those with high levels of threat perception) exhibit greater affiliative ten-
dencies. Schachter suggests that this may be because affiliation reduces the feeling
of solitary exposure to the perceived threat—a suggestion confirmed by Weller
(1963). In a subsequent field experiment Mulder and Stemerding (1963) tested the
hypothesis that individuals seek the company of others when they feel threatened.
Not only did they find support for this hypothesis, they also confirmed the hypoth-
esis that external threats lead individuals to lean on a stronger leader.
In subsequent years the hypothesis asserting a positive relationship between an
increase in threat perception and group cohesion also became anchored in the
theoretical literature on international politics (Kaplan 1957; Wolfers 1962; Liska
1962; Boulding 1962; Scott 1967; Rothstein 1968; Burgess and Moore 1973;
Holsti et al. 1973; Stein 1976; Sullivan 1976). Most prominently the application
occurred in the literature on alliance politics. The earliest statement indicating the
intellectual inheritance from social psychology was made by George Liska (1962:
100–129), who argued that the existence and perception of an external threat is
necessary for alliance cohesion, i.e., for members of the alliance to cooperate with
each other.
Subsequent tests of the hypothesis were performed by Holsti (1965, 1966,
1969), who used content analysis as a method to determine whether Chinese and
Soviet perceptions of the United States as the “external threat” were similar dur-
ing periods of intense East-West conflict. The hypothesis is supported in a follow-
up study by Hopmann (1967). In Unity and Disintegration in International Alliances
(Holsti et al. 1973), Holsti and his collaborators extended the analysis to an exami-
nation of cooperation among NATO member states. They also amended the
theoretical construct itself, arguing that while common threat perceptions are a
prerequisite, subsequent cooperation also requires a common attitude or prescrip-
tion towards the source of the threat. This constitutes an important amendment
as it identifies a missing link in the theoretical construct for the initial formulation
of the hypothesis.
134 Akan Malici

Political behavior is a manifestation of an actor’s preferences (Frieden 1999).

The latter is composed, however, not only of an actor’s diagnostic perceptions,
but also of prescriptive inferences within the same cognitive process. It is theoreti-
cally possible that two actors share the same perceptions of any security external-
ity,4 but nevertheless believe that distinct means of conflict resolution are to be
employed. Diagnostic perceptions and prescriptive inferences together define an
actor’s preferences and his perceived preferences of others regarding the political
outcomes of domination, submission, settlement, or deadlock in world politics
(Walker et al. 2001; see also Snyder and Diesing 1977; Leng and Walker 1982).
Stephen Walt (1987) added to the literature by formulating his balance of threat
theory. In a rejection of classical balance of power theory inherent in structural
realism (Waltz 1979), Walt argues that the stimulus governing state behavior is not
so much power but rather the level of perceived threat. Although widely lauded
as an important and progressive amendment to the realist literature, Walt’s work
remains inherently unsatisfactory. Any systematic measure of the causal variable
of threat perceptions and any specification of how threat perceptions translate into
strategic orientations is absent (Schweller 1994; Vasquez 1997).
The reformulation of classical balance of power theory into balance of threat
theory has some value nevertheless. The value consists mainly of Walt’s contri-
bution of ideational variables such as perceptions to the mainstream discourse
of international relations theory. Variables identified by Walt were subsequently
confirmed by newly emerging neoclassical realists who “posit an explicit role for
leaders’ preexisting belief systems, images of adversaries, and cognitive biases in
the process of … military planning and foreign policy decisionmaking” as they
assume that much of world politics would be “inexplicable without reference to
the perceptions of central decisionmakers” (Taliaferro 2000/01: 141).
The relevance of asking questions about the psychology of decision-making
elites is re-enforced by the end of the cold war and even more so by the terrorist
attacks of 11 September 2001 a decade later. These events created “nonroutine
situations that require more than merely the application of standard operating
procedures and decision rules.” In such situations, “decisions are made at the
pinnacle of the governmental hierarchy,” most notably by the leader (Holsti
1976: 30). However, especially for democratic states, such as France and Great
Britain, it should be acknowledged that rarely is the leader (the traditional focus
of psychological investigations) the only referent for state action, and that a
state leader rarely makes a decision in isolation without at least some input from
other administrative members in the process (Schafer 2000: 513). The foreign
policy decision-making process thereby evolves into a two-level game in which
coalescing dynamics occur simultaneously at the intra- and inter-state level
(Putnam 1988; Lake and Powell 1999a).
The research questions to be asked at the intra-state level are fourfold: (1) Did
the diagnostic and prescriptive beliefs of key figures within the French and British
decision-making elite (i.e., the President and Foreign Minister in France, and the
 Alliances and Their Microfoundations 135

Prime Minister and Foreign Secretary in Great Britain), respectively, converge in

the period immediately following the terrorist attacks of 11 September 2001? (2)
Did the diagnostic and prescriptive beliefs within each state’s decision-making elite
converge in the period preceding the war in Iraq (i.e., before 19 March 2003)?
(3) Did the French President and Foreign Minister have the same or similar stra-
tegic orientation in each of the two crises? (4) Did the British Prime Minister and
Foreign Secretary have the same or similar strategic orientation in each of the two
crises? The answers to these questions should indicate the degree of cohesion in
foreign policy orientations among individuals that are considered to be pivotal in
the foreign policy decision-making process within each state.
Questions at the inter-state level require to an aggregation of the two key deci-
sion makers within a state. This aggregation “presupposes that … international
politics can be studied fruitfully by assuming that actors [within the state] interact,
that this interaction effectively aggregates these actors into states, and that these
states, in turn, interact with each other.” States, as actors, are thereby “social
aggregates” (Lake and Powell 1999b: 14). Such a move is legitimate as long as
there is no split within the decision-making elite, that is, if both the leader and
Foreign Minister share a “collective view” regarding the security externality. If
such is not the case, a determination of the “prevailing view” within the decision-
making elite becomes necessary.
Questions at the inter-state level are analogous to those at the intra-state level:
(1) Did the diagnostic and prescriptive beliefs of the French and British leadership
converge or diverge in the period after 11 September 2001? (2) Did the diagnostic
and prescriptive beliefs of each state converge or diverge in the period preceding
the war in Iraq? (3) What was France’s strategic orientation in each of the two
crises? (4) What was Great Britain’s strategic orientation in each of the two crises?
The answers to these questions should indicate the degree of cohesion between
France and Great Britain. All answers taken together test both the diagnostic and
prescriptive elements of the external threat—internal cohesion proposition, albeit
on different levels of analysis (i.e., within each state and between each state).

Method and Design

In order to measure French and British diagnostic perceptions, prescriptive
inferences, and strategic orientations in operations Enduring Freedom and Iraqi
Freedom respectively, this study employs operational code analysis (Leites 1951,
1953; George 1969, 1979; Holsti 1977; Walker 1977, 1983, 1990). Cognitive
perceptions and prescriptions and the resulting strategic orientations are at the
center of operational code analysis. The research program’s conceptualization
of the “self in situation” with the “self” having diagnostic propensities about
a security externality and prescriptive propensities towards this situation makes
it an appropriate tool for the purpose of finding answers to the research ques-
tions above (George 1979; Walker and Schafer 2003). In order to determine
136 Akan Malici

a leadership’s diagnostic and prescriptive propensities the researcher is required

to systematically examine the speech acts of the leaders under consideration as
described in Chapter 4 of this volume.
The result of these procedures is a set of coded verb constructions manifested in
the speaker’s rhetoric that demonstrate beliefs about self and others in the political
universe in terms of cooperative and conflictual attributions (Schafer 2000: 520).
With the aid of a simple theory of inferences about preferences by Walker and
Schafer (2003), it is possible to test the general hypothesis that a leader’s opera-
tional code systematically influences his general strategic orientation regarding the
possible outcomes of domination, submission, settlement, or deadlock (Snyder
and Diesing 1977). Each proposition in the theory states the preference ordering
for Self (I-1, P-4a) or his perceived preference ranking of the Other (P-1, P-4b).
The signs in the propositions below indicate whether a particular leader’s I-1 and
P-1 indices are above (+) or below (–) the mean (I-1 = +.33; P-1 = +.25) for
a norming group of world leaders. The symbols (> and <) represent the VICS
scores of the P-4a and P-4b indices above (>) or below (<) the mean (P-4 = .21)
for the norming group (Walker and Schafer 2003).5

Prop. 1. If (I-1, P-4a) or (P-1, P-4b) is (+, <), then Settle>Deadlock>Dominate>

Submit.
Prop. 2. If (I-1, P-4a) or (P-1, P-4b) is (+, >), then Settle>Dominate>Deadlock>
Submit.
Prop. 3. If (I-1, P-4a) or (P-1, P-4b) is (–, <), then Dominate>Settle>Deadlock>
Submit.
Prop. 4. If (I-1, P-4a) or (P-1, P-4b) is (–, >), then Dominate>Deadlock>Settle>
Submit.

It is by means of comparing the diagnostic and prescriptive beliefs and the strate-
gic orientations of the leaders and foreign ministers on the intra- and inter-state
levels of analysis in the two time periods under consideration that I intend to
answer the research questions above. A comparison of each chief executive’s and
foreign minister’s operational code beliefs at the intra-state level of analysis will
allow me to determine either a “common view,” or alternatively, a “prevailing
view” within each state. On the inter-state level of analysis, the main hypothesis
is that if diagnostic and prescriptive beliefs are related to the cooperative behavior
of France and Britain regarding the war on terror after 11 September 2001, then
their operational codes should be marked by a high level of congruence.6 That is,
both national leaderships should exert a similar strategic orientation to the outside
world. Conversely, the discord of France and Britain regarding the war in Iraq
should be reflected in a divergence of operational codes and by a corresponding
divergence of strategic orientations.
For each state two pools of samples were constructed. The first pool contains
statements from the period 11 September 2001–9 September 2002. The second
 Alliances and Their Microfoundations 137

pool contains statements from 10 September–19 March 2003—the onset of the

war against Iraq. The vast majority of statements pertain directly to the situation in
Afghanistan and Iraq. The dividing date for the two pools marks the publication
of the new U.S. national security strategy.7 The new security strategy elevated
preemptive strikes to a legitimate action in the conduct of international politics.
After U.S. President George Bush branded Iraq along with Iran and North Korea
as constituting a lethal “axis of evil” some months earlier, it became clear that the
amorphous war against international terrorism would be expanded into a wider
struggle, which would include the disarming of Iraq (Cox 2003: 527; Dunne
2003: 267; Hollis 2003). I will now turn to an analysis of French and British
operational codes in the two time periods under consideration.

French and British Intra-State Comparisons

In order to find out if the beliefs of the French leadership, namely President
Chirac and Foreign Minister Vedrine, converged in the post-11 September 2001
period and in the period immediately preceding the war in Iraq, I conducted a
two-factor MANOVA, where leader (President or Foreign Minister) and time
period (post-11 September 2001 or post-10 September 2002) were the indepen-
dent factors and the operational code indices were dependent variables.8
The results from the MANOVA show that there are several significant main
effects for time period, a few significant main effects for leader, and no significant
interaction effects. The findings indicate that, independent of the leader’s particular
role identity as president or foreign minister, changes in the international security
environment were substantial enough to cause changes in the leader’s perception
of the international security environment and the best tactics to deal with this
environment. When compared to the post-11 September period, the French
leadership perceived the international security environment to be significantly
more cooperative (P-1 = .155 vs. .426), and were significantly more optimistic
about their chances for realizing political goals (P-2 = –.006 vs. .198) in the
period preceding the war in Iraq.
In the realm of instrumental beliefs, French leaders were significantly more
cooperative in their strategic approach to goals (I-1 = .276 vs. .539), were
significantly more cooperative in their intensity of tactics (I-2 = .051 vs. .217),
were significantly less likely to shift between cooperative and conflictual tactics
(I-4a = .724 vs. .461) and between words and deeds (I-4b = .592 vs. .259),
and were significantly less likely to ascribe utility to Punish tactics as a means of
achieving political goals (I-5f = .175 vs. .058) in the period preceding the war in
Iraq. The MANOVA results also indicate a few significant differences between the
prescriptive operational code beliefs of President Chirac and the French Foreign
Minister. Independent of the international security environment, the French
President had a significantly more cooperative strategic approach to goals (I-1 = .516
vs. .298), was significantly less likely to shift between cooperative and conflictual
138 Akan Malici

tactics (I-4a = .484 vs. .702), and was significantly less likely to ascribe utility
to Oppose tactics as a means of achieving political goals (I-5d = .144 vs. .133).
The results from the French intra-state analysis suggest that the French leader-
ship perceived significant changes in the international security environment from
the post-11 September period to the period immediately preceding the war in
Iraq. These perceived changes in the level of threat had a significant impact on
French operational code beliefs, such that leaders believed the political universe to
be significantly more cooperative in the period preceding the Iraq War relative to
the post-11 September period. The results also indicate that the French leadership
was united in their diagnosis of the international security environment but there
were a few differences between President Chirac and his Foreign Minister in their
beliefs about the best tactics to achieve political goals. These differences in instru-
mental beliefs suggest that President Chirac preferred relatively more cooperative
strategies and tactics than did his Foreign Minister.
As in the French case, I conducted a two-factor MANOVA in order to find
out if the beliefs of the British leadership, namely Prime Minister Tony Blair
and Foreign Secretary Jack Straw, converged in the post-11 September 2001
period and in the period immediately preceding the war in Iraq. Again leader
and time period were the independent factors and the operational code indices
were dependent variables. The results from the MANOVA show that there were
no significant main effects for time period, one significant main effect for leader,
and a few significant interaction effects. The findings indicate that independent
of the international security environment, Prime Minister Blair believed he had
significantly more control of historical development (P-4 = .263 vs. .181) than
did Foreign Secretary Straw. The philosophical index measuring the Role of
Chance (P-5) in international politics was also significant, but cannot be inter-
preted directly as it is part of an interaction effect discussed below.
Two interaction effects indicate that there were significant differences in two
of the diagnostic propensities (P-3 and P-5) between Prime Minister Blair and
Foreign Secretary Straw, making it appropriate to conduct post hoc analyses to
determine the magnitude of these differences. The post hoc analyses reveal sig-
nificant differences between Blair and Straw from the post-11 September period
to the period preceding the war in Iraq in terms of their beliefs about the pre-
dictability of the political future. The results indicate that Prime Minister Blair
believed the political future became significantly more predictable in the period
preceding the war in Iraq (P-3 = .069 vs. .113, p = .019). At the same time
Foreign Secretary Straw believed the political future became significantly less
predictable from period 1 to period 2 (P-3 = .105 vs. .052, p = .007). The same
trend is present in the leaders’ beliefs about the role of chance in international
politics (P-5). While Blair attributed a significantly lower role to chance in the
period preceding the war in Iraq 1 (P-5 = .984 vs. .968, p = .042), Straw believed
that chance played a significantly higher role in the same time period (P-5 = .979
vs. .992, p = .087).
 Alliances and Their Microfoundations 139

The results of the British intra-state analysis present a different picture than the
French intra-state analysis. It seems that British perceptions of the international
security threat remained relatively static from the post-11 September period to
the period preceding the war in Iraq. This contrasts sharply with the French case,
as French leaders’ perceptions of threat decreased substantially in the period pre-
ceding the war in Iraq. Like the French, there were some differences between
the beliefs of Prime Minister Blair and Foreign Secretary Straw. However, in
the British case, this difference was diagnostic rather than prescriptive in nature,
suggesting a minor inconsistency in the perception of the international security
environment on the part of the British leadership. The two significant interac-
tion effects indicate that the differences in British leaders’ operational codes were
exacerbated somewhat by the international environment. Overall, however, the
beliefs of Prime Minister Blair and Foreign Secretary Straw were marked by a
high level of congruence. As we shall see below, both exerted the same strategic
orientation towards the external world.
In sum, although there were some differences within each state, they were
the exception rather than the rule. We shall see below that these differences did
not cause a difference regarding the strategic orientations of the different deci-
sion makers; each state exerted a “common view” of the pertaining “security
externality.” I will now turn to a comparison of French and British beliefs on the
inter-state level of analysis.

French and British Inter-State Comparisons

The preceding results from the intra-state MANOVA analyses of the French
and British leadership suggest that there may have been differences in the two
regimes’ diagnosis of the international environment in the period preceding the
war in Iraq. Only a direct comparison of the two states, however, will allow me
to discern if there were significant differences between French and British threat
perceptions in the post-11 September period and the period preceding the Iraq
War. This comparison also allows me to discern whether the French and British
leadership differed in their prescriptive beliefs about how to deal with the threats
that they faced. To answer these questions, I again conducted a MANOVA using
time period (post-11 September 2001 or post-9 September 2002) and leadership
(French or British) as the independent factors and the operational code indices as
the dependent variables.
The results of the MANOVA show that there were no significant main effects
for time period, a number of significant main effects for leadership, and several
significant interaction effects.9 The findings reveal that national identity played an
important role in determining threat perceptions and prescriptive beliefs about
how to deal with the security environment, independent of whether the leaders
faced the post-11 September security environment or the pre-Iraq War security
environment. In the realm of philosophical beliefs, the French leadership believed
140 Akan Malici

that the political universe was significantly more predictable (P-3 = .133 vs. .085),
and perceived a significantly lower role of chance in international affairs (P-5 =
.965 vs. .981) when compared to the British leadership. In terms of instrumental
beliefs, the French leadership was significantly more acceptant of risk (I-3 = .263
vs. .151), and was significantly more likely to ascribe utility to Appeal tactics
(I-5c = .531 vs. .399) as a means of achieving political goals when compared to
the British leadership.
The MANOVA results also indicate significant interaction effects between
time period and leadership for several beliefs (P-1, P-2, I-1, I-2, I-4a, I-4b, I-5f).
The interaction effects are illustrated in Figure 8.1. It is appropriate to conduct
post hoc analyses to determine the extent of these differences. The post hoc
analyses reveal that while the French leadership’s diagnostic and prescriptive
beliefs changed significantly—becoming much more cooperative—the British
leadership’s beliefs remained static. In the realm of philosophical beliefs the
French leadership’s diagnosis of the security environment was significantly more
cooperative in the period preceding the war in Iraq (P-1 = .155 vs. .426, p =
.006); there was no significant difference in the British leadership’s diagnosis of
the security environment between the two time periods (P-1 = .139 vs. .104, p =
.704). Likewise, whereas the French leadership was significantly more optimistic
about their chances for realizing political values in the pre-Iraq War period (P-
2 = –.006 vs. .198, p = .004), there was no significant change in the British
leadership’s beliefs about their chances for realizing political values (P-2 = .038
vs. .006, p = .627).
Whereas the French leadership adopted significantly more cooperative strategies
(I-1 = .276 vs. .539, p = .019) in the realm of instrumental beliefs, there was no
significant difference in the British leadership’s strategic approach to goals (I-1 =
.291 vs. .207, p = .436). Similarly, as the French leadership became significantly
more cooperative in terms of intensity of tactics (I-2 = .051 vs. .217, p = .024),
the British leadership showed no significant change in their intensity of tactics (I-
2 = .113 vs. .047, p = .352). In terms of shift propensities, the French leadership
was significantly less likely to shift between cooperation and conflict tactics (I-4a
= .724 vs. .461, p = .015) and between words and deeds (I-4b = .592 vs. .259,
p = .017) in the pre-Iraq War period; there were no significant changes in the
British leadership’s shift propensities (I-4a = .709 vs. .770, p = .553; I-4b = .600
vs. .607, p = .956). Finally, the post hoc analyses reveal one important difference
between the two regimes’ utility of means. While the French were significantly
less likely to ascribe utility to Punish tactics as a means of achieving political goals
(I-5f = .175 vs. .058, p = .010) in the period preceding the war in Iraq, there was
no significant difference in the British regime’s utility of means beliefs (I-5f = .148
vs. .180, p = .452).
Taken together, these results suggest that while French and British perceptions
of threat and the best strategies to counter these threats converged in the post-11
September period, they diverged significantly in the period preceding the war in
 Alliances and Their Microfoundations 141

P–1 (Nature of Political Universe) P–2 (Realization of Political Values)

0.5 0.25
0.4 0.426 0.2 0.198
0.15
0.3
0.155 0.1
0.2 0.038
0.05
0.1 0.139 0 0.006
0.068 –0.006
0 –0.05
Period 1 Period 2 Period 1 Period 2

I–1 (Strategic Approach to Goals) I–2 (Intensity of Tactics)

0.6 0.25
0.539 0.217
0.5 0.2
0.4
0.15
0.3 0.291 0.113
0.276 0.207 0.1
0.2
0.1 0.05 0.047
0.051
0 0
Period 1 Period 2 Period 1 Period 2

I–4a (Timing of Action – Confl./Coop.) I–4b (Timing of Action – Words/Deeds)

0.8 0.65
0.724 0.77 0.6 0.607
0.75 0.6
0.7 0.55 0.592
0.709 0.5
0.65
0.6 0.45
0.55 0.4
0.35
0.5 0.3
0.45 0.461 0.259
0.25
0.4 0.2
Period 1 Period 2 Period 1 Period 2

I–5f Punish (Utility of Means – Punish)

0.18 0.18
0.175
0.16 Key
0.14
0.148 France
0.12
0.1 Britain
0.08
0.06 0.058
0.04
Period 1 Period 2

FIGURE 8.1 Inter-State Comparisons of Beliefs for France and Britain

Iraq, such that French beliefs became significantly more cooperative while British
beliefs remained conflictual. But are these statistical convergences and divergences
substantively meaningful? That is, do they explain the general strategic orientations
of France and Britain in the two time periods under consideration? This question
will be addressed in the next section.
142 Akan Malici

Discussion
The intra-state analyses for France and Britain for Operation Enduring Freedom
following 9/11 suggest that there were some differences between the French
President and his Foreign Minister and between the British Prime Minister and
his Foreign Secretary. However, these were not sufficiently significant to cause a
split within either the French or the British decision-making elite. A comparison
of each decision maker’s strategic orientation, inferred from the simple theory of
inferences about preferences described above, reveals that the general strategic
orientation within each government was very similar.
The differences that did exist were not substantive. President Chirac’s VICS
scores specify the strategic preference ranking in Proposition 4 for Self (I-1 =
.30, P-4a = .27) and Proposition 3 for Other (P-1 = .15, P-4b = .15) in the
period after 11 September 2001. Foreign Minister Vedrine’s scores specify the
strategic preference ranking in Proposition 3 for Self (I-1 = .25, P-4a = .17) and
Proposition 4 for Other (P-1 = .16, P-4b = .25) in the same period. Thus, both
ranked domination as their most preferred outcome and submission as their least
preferred outcome. Chirac was slightly more confrontational than Vedrine as his
alternate strategy would be to pursue a deadlock. Vedrine preferred a settlement
instead. Both Chirac and Vedrine agreed that Other is driven by a confronta-
tional strategy. Overall President Chirac and Foreign Minister Vedrine shared a
sufficiently similar strategic orientation, which allows pooling the scores for both
decision makers and calculating “France’s operational code.”10
A similar scenario applies to the British case. Like President Chirac, Prime
Minister Blair’s VICS scores also specify the strategic preference ranking in
Proposition 4 for Self (I-1 = .25, P-4a = .24) and Proposition 3 for Other (P-1
= .14, P-4b = .18). Foreign Secretary Straw’s VICS scores, like those of Hubert
Vedrine, specify the strategic preference ranking in Proposition 3 for Self (I-1
= .33, P-4a = .20) and Proposition 4 for Other (P-1 = .14, P-4b = .22). Thus,
as in the case of France, both British decision makers favored domination as the
ultimate strategy and both agreed that submission was their least preferred choice.
Analogous to the French case, the Prime Minister appears to have been somewhat
more confrontational than the Foreign Secretary. Furthermore, Blair and Straw
also believed that Other was motivated by a confrontational strategy. Here again,
the similarity in the overall strategy allows me to pool the scores for Blair and
Straw and to calculate “Britain’s operational code.”11
The operational code of France specifies the strategic preference ranking in
Proposition 4 for Self (I-1 = .28; P-4a = .22) and Proposition 3 for Other (P-1 =
.15, P-4b = .20). Also Great Britain’s operational code specifies the strategic pref-
erence ranking in Proposition 4 for Self (I-1 = .29; P-4a = .22) and Proposition
3 for Other (P-1 = .14, P-4b = .20). Thus, both governments ranked their pref-
erences domination over deadlock over settlement over submission while they
perceived Afghanistan to rank domination over settlement over deadlock over
 Alliances and Their Microfoundations 143

submission. A 2 × 2 game matrix in Figure 8.2 gives a pictorial illustration of these

strategic orientations (Rapoport and Guyer 1966). The outcome (equilibrium)
depends upon each player’s strategic calculations within its “subjective game”
(Maoz 1990: 423; Maoz and Mor 2002; Morrow 1994; Snyder and Diesing
1977).12 The games of both France and Britain are conflictual. They have a domi-
nant strategy of CF for Self and an anticipated Nash equilibrium of (3,2) deadlock
(Maoz 1990: 203–204, 381).
Both France and Britain were taken by surprise in the 11 September attacks.
Based on a prediction of strategic rationality, we would expect both France and
Great Britain to choose a conflict strategy against Al Qaeda and the Taliban regime
in Afghanistan. The convergence of diagnostic and prescriptive beliefs of French
and British leaders in accordance with the threat/cohesion proposition, also leads
us to expect cooperative coalescing dynamics between France and Britain within
the EU, NATO, and the UN. Coalescing dynamics within NATO, the UN,
and the EU were smooth indeed. In an extraordinary meeting on 21 September
2001, the European Council declared itself in “total solidarity with the United
States” and recognized the legitimacy of any military response that was to follow
(Delpech 2002: 13, 2l; Peers 2003: 227). France and Britain were in complete
agreement, and the Council concluded with a call for “the broadest possible
coalition against terrorism” (Gordon 2001/2002: 95).
A similar scenario unfolded within NATO. Traditionally France was “reluctant
to see NATO’s role enhanced, or the scope of its missions expanded.” However,
this time France “did not hesitate to support the invocation of NATO’s article 5
mutual defense guarantee” (Gordon 2002; Hoffmann 2003b: 1030). Great Britain
proved itself in its traditional role of “first friend and ally” (Penska and Mason
2003: 271). One day after the attack NATO invoked Article 5 of the Alliance
Charter for the first time in its history (Gordon 2001/2002: 92; Daley 2001).

Afghanistan Other Afghanistan

CO CF CO CF CO CF
CO 2,3 1,4 CO Settle Submit CO 2,3 1,4

France Self Britain

CF 4,1 3,2* CF Dominate Deadlock CF 4,1 3,2*

France’s Strategy France/Britain Outcomes Britain’s Strategy

*Dominant strategies are underlined and the Nash equilibrium is asterisked. It is a state
from which neither player would have an incentive to depart unilaterally
because it would do worse, or atleast not better, if it did (Nash1951).

FIGURE 8.2 Subjective Games of France and Britain in Operation Enduring Freedom
144 Akan Malici

Member states considered the attack on the United States as an attack against all
and, therefore, undertaking retaliatory measures as legitimate.
Also one day later, France proposed a resolution at the UN Security Council
(UNSC) condemning the attacks and having them declared a “threat to interna-
tional peace and security” under the Charter’s Chapter VII regarding the use of
force. On the same day, the UNSC unanimously adopted Resolution 1368 and
articulated its determination to “combat by all means threats to international peace
and security caused by terrorist attacks.” Two weeks later the Council adopted
Resolution 1373, reaffirming its original stance (Katselli and Shah 2003: 245).
These two resolutions, in effect, legitimized any necessary military action in the
ensuing war in Afghanistan (de Jonge Qudraat 2003). The diplomatic concord
between France and Britain within these multilateral organizations was reflected
in their military cooperation in the ensuing operation Enduring Freedom.13
Although the forthcoming military operations were not initially organized under
the auspices of either NATO, the UN, or EU, but were rather led by the United
States, these multilateral organizations were confirmed to be legitimate and viable
fora in world politics.
Both France and Great Britain cooperated by engaging all three branches of
their armed forces. The French air force was engaged with six Mirage-2000
fighter bombers, two Mirage IV and one C-160 Gabriel reconnaissance aircraft,
two KC-135 aerial tanker aircrafts as well as several types of transport aircrafts.
Great Britain contributed VC-10 and Tristar aerial tankers, C-130 transport and
Nimrod and Canberra PR 9 reconnaissance aircrafts and E-3 AWACS aircrafts.
France’s naval engagement was reflected in the emission of its only carrier battle
group to support combat operations in the North Arabian Sea. This group con-
sisted of the aircraft carrier Charles de Gaulle, three frigates, one tanker, and one
submarine. Great Britain’s naval contribution consisted in its emission of the air-
craft carrier Illustrious, one amphibious ship, two frigates, three submarines, and
seven auxiliary ships. On the ground, France contributed an infantry regiment and
Engineer forces to operate alongside U.S. and British special operations person-
nel. By mid-December, France and Britain had deployed around 5,000 military
personnel each to Afghanistan and surrounding areas. Franco-British cooperation
in Operation Enduring Freedom is especially illustrative in the French ground
troop participation in the International Security and Assistance Force (ISAF)
which, until early 2002, stood under British command.
In sum, in accordance with the threat/cohesion proposition, harmonious
French and British diagnostic and prescriptive beliefs and strategic orientations
were reflected in smooth coalescing dynamics within the fora of the EU, NATO,
and the UN and resulted in a conflictual posture towards the external enemy.
This would not be the case in the forthcoming Operation Iraqi Freedom. In
September 2002, the United States as the leader of the fight against terrorism
extended its war aims to Iraq with Operation Iraqi Freedom. The justification for
a military invasion was apparent evidence of the Iraqi regime intending to acquire
 Alliances and Their Microfoundations 145

stockpiles of weapons of mass destruction (WMD) and an apparent link between

Al Qaeda and Saddam Hussein. However, conclusive evidence, apart from
inferential extrapolations from the Iraqi dictator’s previous attempts to acquire
nuclear weapons and apart from unsubstantiated speculations about the connec-
tion between Al Qaeda and the regime in Iraq, remained absent (Anderson 2003;
Goh 2003: 87; Cortright et al. 2003).
The ambiguity of the situation triggered what turned out to be one of the
most damaging crises in the history of the EU, NATO, and the UN. As the
MANOVA analyses above suggest, the diagnostic and prescriptive beliefs of the
French and British leadership diverged significantly as time progressed and as
plans got underway to extend military operations to Iraq. The French President
and his Foreign Minister were unified in their disinclination towards the employ-
ment of military means in Iraq. Unlike in the preceding period, they did not even
have a disagreement about a possible alternate strategy. Taken together, the VICS
scores specify the strategic preference ranking in Proposition 2 for France (I-1 =
.54, P-4a = .28) and Proposition 1 for Other (P-1 = .43, P-4b = .14). France’s
strategy was one in favor of settlement over domination over deadlock over sub-
mission. French leaders also perceived that settlement was a realistic outcome,
as they also perceived Iraq to move towards this outcome by ranking settlement
over deadlock over domination over submission.
The British leadership, in contrast to the French, displayed a conflictual
strategic orientation. Prime Minister Blair’s VICS scores for Self (I-1 = .26, P-4a
= .28) and Other (P-1 = .15, P-4b = .14) specify the strategic preference ranking
in Proposition 4. Foreign Secretary Straw’s orientation is also confrontational.
His VICS scores for Self (I-1 = 16, P-4a = .16) and Other (P-1 = .06; P-4b =
.26) specify the strategic preference ranking in Proposition 3. As in the previous
period, Blair appears to have been somewhat more confrontational than Straw,
as he preferred a deadlock as an alternate outcome while the Foreign Secretary
preferred settlement as an alternative to domination. Both perceived Iraq to have
confrontational intentions. The general resemblance of Blair and Straw’s strategic
orientation again allows for the pooling of their VICS scores. Great Britain’s
operational code specifies the strategic preference ranking in Proposition 4 for
Self (I-1 = .20; P-4a = .22) and Proposition 3 for Other (P-1 = .10, P-4b = .20).
Thus, in stark contrast to the strategic orientation of France, Great Britain was in
favor of domination over deadlock over settlement over submission. Also in stark
contrast to France, Great Britain did not perceive the regime in Iraq to have a
cooperative outcome as their first preference.
The 2 × 2 game matrix in Figure 8.3 gives a pictorial illustration of the Franco-
British divergence regarding Operation Iraqi Freedom (Majeski 2004; Rapoport
and Guyer 1966). The French game is cooperative while the British game is
conflictual (Maoz 1990: 381). Based on predictions of strategic rationality, we
would expect France to choose a cooperative strategy calculating that Iraq will
do the same. The prediction for Great Britain, however, is that it would engage
146 Akan Malici

Iraq Other Iraq

CO CF CO CF CO CF
CO 4,4* 1,2 CO Settle Submit CO 2,3 1,4

France Self Britain

CF 3,1 2,3* CF Dominate Deadlock CF 4,1 3,2*

France’s Strategy France/Britain Outcomes Britain’s Strategy

*Dominant strategies are underlined and the Nash equilibrium is asterisked. It is a state
from which neither player would have an incentive to depart unilaterally
because it would do worse, or atleast not better, if it did (Nash1951).

FIGURE 8.3 Subjective Games of France and Britain in Operation Iraqi Freedom

in a conflict strategy reasoning that Iraq would also choose a conflict strategy.
In contrast to the previous period, the divergence of diagnostic and prescriptive
beliefs of French and British leaders and foreign ministers do not lead us to expect
cooperative coalescing dynamics within any multilateral organization.
Indeed, the ensuing crisis between France and Great Britain over Iraq, in
effect, pitted “old” Europeans against “new Europeans” and “Venutians” against
“Martians” (Kagan 2003; Cox 2003: 528; Duke 2003; Glennon 2003; Marfleet
and Miller 2005; Rumsfeld 2003). France, in favor of a continuation of diplo-
macy vis-à-vis the regime in Iraq, was considered to be part of the old Europe.
Military operations were to be limited to the initial target of Afghanistan (Gordon
2001/2002: 94). Great Britain, aligned with the United States and ultimately in
favor of a confrontational foreign policy, was considered part of the new Europe
(Lewis 2002). Coalitions that had previously formed through consultations and
exchanges in the fora of multilateral organizations came to be replaced by a “coali-
tion of the willing” (Hoffmann 2003b: 1033). As a consequence, the crisis over
Iraq provoked a division within the EU and NATO, and effectively devalued the
UN as an organization for the legitimization of force in world politics (Hoffmann
2003a).
The disagreement between France and Britain quickly translated into a
polarization of the EU. Britain led the efforts to create a coalition of European
countries in support of a preemptive war against Iraq. At the same time, France
tried to develop a common EU position that would set a separate course (Daalder
2003: 156). France argued strongly that any decision on how to disarm Iraq
should remain that of the UN (Howorth 2003: 248). While the EU figured in the
French strategy as a possible source of reinforcing political backing for its stance,
the diplomatic campaign was conducted in the forum of the UN (Brenner 2003:
203).
 Alliances and Their Microfoundations 147

The controversy was also reflected in NATO. The elusiveness of reaching

unanimity here was illustrated by the opposing positions France and Great Britain
took regarding the Turkish request for military aid in case of an Iraqi attack
(Daalder 2003: 147, 155). Regarding the ensuing episode of NATO consulta-
tions, which led to no agreement, Henry Kissinger (2003) went so far as to con-
clude that differences over Iraq have “produced the gravest crisis in the Atlantic
Alliance since its creation five decades ago.” In the end “NATO was sidelined”
(Hoffmann 2003b: 1034).
In the UN, France requested a second explicit mandate beyond Security
Council Resolution 1441 (Duke 2003: 6). France judged war to be unnecessary
and something that should be avoided because the regime in Iraq was not consid-
ered to “pose a … danger to either the stability in the Gulf or the security of the
West” (Brenner 2003: 202). By January the collision course of Paris and London
in the UNSC had become evident: “France had moved toward an absolutist posi-
tion against the use of military force, climaxed by President Chirac’s declaration
in March that France was against a new resolution endorsing the use of force
under any circumstances” (Brenner 2003: 203; Chirac 2003).
In the meantime, Britain took an absolutist position in the opposite direction,
declaring alongside the United States a readiness to take military action against
Iraq (Bowers et al. 2003). Military operations commenced on 20 March 2003.
Like Operation Enduring Freedom, Great Britain cooperated by engaging all
three branches of its armed forces, employing even a more extended arsenal than
in the previous period. France remained absent. Such an outcome is what the
threat-cohesion proposition would lead us to expect. A disappearance of a com-
mon diagnosis of the international environment, common prescriptions towards
it, and the resulting distinct strategic orientations ultimately resulted in the ero-
sion of the viability of the EU, NATO, and the UN as France and Great Britain
pursued fundamentally opposed strategies.

Conclusion
What conclusions can be drawn from the foregoing analyses? First, while there
were some general differences in beliefs within the French leadership on the one
hand, and within the British leadership on the other, neither regime showed
major disagreements between the chief executive and its foreign minister. Indeed,
although differences in diagnostic or prescriptive beliefs within both regimes were
revealed, ultimately, their strategic orientations proved to be similar. The key
difference between the two intra-state comparisons was that while the French
perceived a stark difference in the level of threat in the international security
environment from the post-11 September 2001 period to the period preceding
the war in Iraq, British perceptions remained static. The British diagnosed the
situation in Iraq to be parallel to the situation they had encountered after the 11
September terrorist bombings and the ensuing war in Afghanistan.
148 Akan Malici

Second, after 11 September 2001, both France and Britain were united in the
belief that the international security environment was highly threatening. These
analogous diagnoses led to similar prescriptive beliefs about the best means to
attain political goals in this high threat environment. These beliefs and strategic
orientations did not solely exist in the leaders’ heads. The similarity of French and
British diagnostic and prescriptive beliefs, in accordance with the threat/cohesion
proposition, manifested itself in the smooth coalescing dynamics of France and
Britain within the EU, NATO, and the UN. The similarity of beliefs translated
into identical strategic orientations towards the external world.
French and British unity in the war on terror, however, disintegrated when the
United States made it clear that Iraq would now be considered part of the terrorist
world, and as such, must be conquered, even if this meant striking preemptively.
Whereas British diagnostic and prescriptive propensities remained very conflic-
tual, the perceptions of the French leadership stood in complete contrast. The
French seemed to think that Afghanistan and Iraq were two different ballgames,
each deserving its own “game plan.” The implications of the threat/cohesion
proposition would lead us to expect an erosion of the viability of coalitions formed
within multilateral organizations. The differences in diagnostic and prescriptive
interpretations of the period preceding the war in Iraq led to distinct strategic ori-
entations and contrasting patterns of behavior. Undeniably, the British expended
a substantial amount of its national resources to combat the Iraqi foe while the
French committed none of its resources, as their diagnosis suggested there was no
imminent danger at all.
The reasons for diverging national threat perceptions may be manifold and
cannot be answered conclusively. Scholars have argued that Britons (along with
Americans) are from Mars while their French counterparts are from Venus (Kagan
2003). Different assessments of threat and different appraisals of power would
thus originate from the cultural matrix of a society. Scholars have also argued that
individual idiosyncrasies of decision makers, such as power or affiliation motives,
lead to perceptual biases (Peterson et al. 1994). These, in turn, can be perpetuated
by other needs, such as cognitive consistency. Still other explanatory candidates
are individual modes of information processing in which decision makers commit
fundamental attribution errors or use analogies when they are not appropriate
(Tetlock 1998). Investigating which of these factors, or a combination of them,
can be held accountable for diversions in national threat perceptions lies beyond
the scope for this chapter and should be addressed in future research. The more
modest goal in this chapter was to determine whether France and Britain had
different perceptions and to relate them to their strategic behavior in Operation
Enduring Freedom and in Operation Iraqi Freedom.
What makes Franco-British relations in the reviewed periods interesting and
important is that they belong to the same multilateral organizations. Ikenberry
(2002: 53) laments that most recent times have brought a “general depreciation
of international rules, treaties, and security partnerships.” Rules, treaties, and
 Alliances and Their Microfoundations 149

partnerships have their roots in shared perceptions of both interests and the threats
to them. This certainly was the case for the United Nations Security Council,
NATO, and the EU. These multilateral organizations are “shared mental models”
(Denzau and North 1994) and, as such, despite being objective physical facts, they
are “cognitive entities that do not exist apart from actors’ ideas about how the
world works” (Wendt 1992: 399). The foregoing analyses of French and British
decision makers’ diagnostic and prescriptive beliefs and their respective strate-
gic orientations demonstrates that an examination of the viability of multilateral
organizations, whether these are alliances or international institutions, cannot or
should not neglect the specification of their microfoundations.

Notes
* Reprinted with permission and with some alterations from the Journal of Conflict
Resolution, Vol. 49, No. 1 (February 2005), pp. 90–119.
1 Press Conference of Tony Blair and Jacques Chirac in Le Touquet, France, 4 February
2003.
2 http://www.cnn.com (2001) Blair, Chirac “United on Terror” (November 29).
3 For a comprehensive review of this literature see Stein (1976).
4 The term “security externality” is borrowed from Gowa (1994). The term as used here
refers to the existence of stimuli in the international environment that may or may not
disrupt the security of state actors.
5 The data for the norming group contains scores for a collection of 164 speech acts
given by thirty different and diverse world leaders—leaders of poor and rich states
and weak and strong states. It was obtained from Professor Mark Schafer, Department
of Political Science, 240 Stubbs Hall, Louisiana State University, Baton Rouge, LA
70803-5433. The simple theory of preferences here was later expanded from four
to six propositions (see Marfleet and Walker 2006; Walker and Schafer 2007; and
Chapter 13 in this volume).
6 This aggregation, which has also been referred to as “cognitions of a state,” does not
interfere with the VICS method. No significant variation in the unit of analysis, the
utterance, is to be expected, since the object of analysis remains official statements of
the ruling elite (Barnet 1972; Bobrow, Chan, and Kringen 1979; Herrmann 1985;
Schafer 2000).
7 In sampling statements of the French and British decision-making elite, I followed
Walker et al. (1998: 182) who set the following criteria for foreign policy speeches:
“(1) the subject and object are international in scope; (2) the focus of interaction is a
political issue; (3) the words and deeds are cooperative or conflictual.” The sample
of speech acts were drawn from the Lexis Nexis AcademicUniverse and French and
British governmental web pages. Each speech act had to be at least 1,500 words long
in order to be subjected to the coding procedures. Due to the relatively short time
periods examined, sampling randomly from a large pool was not possible. The final
pool contained a total of 32 speech acts—16 speech acts each for France and Britain.
8 The French Foreign Minister in the first period under consideration was Hubert
Vedrine. In the second period Vedrine was replaced by Jacques de Villepine. Thus,
what is compared in the MANOVA analysis is not the personal identity of an individual
but rather the role identity of the Foreign Minister (Wendt 1999).
9 Although time period had a significant main effect on P-1, P-2, and I-4b, these results
are incorporated and interpreted in the findings for interaction effects. The same is true
for P-1, I-1, I-4a, and I-4b main effects regarding the leadership factor.
150 Akan Malici

10 Although Chirac and Vedrine had slightly different preference orderings, their
individual subjective games reveal that the favored strategies of both leaders lead to
the same Nash equilibrium. This allows me to aggregate their individual games to an
aggregated French game.
11 Like in the case of France, Blair and Straw had slightly different preference orderings.
However, here too, their individual subjective games lead to the same Nash
equilibrium.
12 For the definition of subjective games see Chapter 5 in this volume.
13 For an exact overview of French and British military participation in operation Enduring
Freedom see United States Department of Defense, Report on Allied Contributions to
the Common Defense (June 2002).
PART III

Foreign Policy Learning

9
LEARNING TO RESIST OR
RESISTING TO LEARN?
The Operational Codes of Fidel Castro and
Kim Il Sung*

Akan Malici

Introduction
The end of the cold war brought with it the demise and transformation of the
communist bloc led by the former Soviet Union. Indeed, after August 1991 only
a handful of communist states remained in the international system and even
fewer maintained their confrontational relationship with the United States. Two
such states, Cuba and North Korea, had depended heavily on support in the form
of economic aid and political backing from the Soviet empire during the cold
war. With the breakdown of the Soviet Union, this support system quickly dried
up: the economies of Cuba and North Korea declined significantly, and the lead-
ers of these states became politically isolated.
Despite these developments, why did Cuba and North Korea continue to walk
on the communist path? Was it simply a cold war artifact of geopolitics, due to
Cuba’s location next to the Yankee colossus and the presence of U.S. troops on
the Korean Peninsula? The way the foregoing question is framed is admittedly
somewhat of an exaggeration, as it suggests rigid and strict continuity in Cuban
and North Korean foreign policy behavior. Some change towards moderation
and (selective) opening did occur in both states. And yet, the magnitude of the
events that led to the end of the cold war, as well as the fundamentally changed
nature of international politics on both the structural and the dyadic level of
analysis, make the nature of these changes appear rather minor.
Structural approaches call our attention to incentives in the international envi-
ronment for particular types of policies. It is posited that “governments respond
in a rational manner to the reward and punishment contingencies of the interna-
tional environment” (Tetlock 1991: 24). The rational actor premises underlying
structural approaches lead us to expect cooperative foreign policy behavior, or
154 Akan Malici

policies of accommodation in general when cooperation is indeed prudent. It

is the very small degree of change in Cuban and North Korean foreign policy
since the end of the cold war—an event that had a very pronounced impact on
these states—that leads us to question the validity of analyses that focus primar-
ily on external conditions when explaining Cuban and North Korean foreign
policy behavior. This suspicion is further strengthened given that an explanation
of foreign policies of small states such as Cuba and North Korea should be an easy
case for approaches that focus on the international environment as an explanans
(Elman 1995).1
Questioning the validity of structural explanations is, of course, not new to
international relations scholarship. Scholars have convincingly argued that struc-
tural explanations do not sufficiently explain the Soviet dissolution (Breslauer
1987, 1992; Breslauer and Tetlock 1991; Evangelista 1991; Lebow 1994). Instead,
it is often argued that the “new thinking” policies were the result of changes made
within the Soviet leadership, led by Mikhail Gorbachev, who has been described
as an “uncommitted thinker and motivated learner” (Stein 1994). Reversing this
logic, I offer a “learning” explanation of the Cuban and North Korean non-
transitions (Deutsch 1966; Jervis 1976; Levy 1994). That is, I contend that sub-
stantive and lasting change in Cuba and North Korea did not occur because of the
absence of learning by the leaders of those states. By learning, I mean “a change
in beliefs (or degree of confidence in one’s beliefs) or the development of new
beliefs, skills, or procedures as a result of the observation and interpretation of
experience” (Levy 1994: 283). Thus, the main assumption of this study is that
the mechanisms of continuity are situated in the beliefs of the Cuban and North
Korean leadership.
The great transformations of the late 1980s and early 1990s, so I contend, pro-
vided plenty of opportunity for the observation and interpretation of experience:
the international structure, Cuba and North Korea’s position within it, and the
dyadic relationships between the Soviet Union vis-à-vis Cuba and North Korea
all changed dramatically. Thus, it is very puzzling that Cuban and North Korean
foreign policy did not undergo sizeable changes following this catalytic period.
Following from this and given my basic assumptions, my primary research questions
are: “Did Fidel Castro and Kim Il Sung learn from their cold war experiences, and
if so, what did they learn?”
There is a large demand to apply theories used within the field of international
relations to the cases of Cuba and North Korea (e.g., Ahn 1997; Kim 1998). The
most frequently articulated plea is for individual level theorizing (e.g., Gonzales
1996; Gonzalez and Ronfeldt 1986; Planas 1993; Smith 1988), and common
sense explains why this is so. While democratic countries are characterized by sys-
tems of checks and balances and multiple veto points, hindering the personalized
rule of one individual, authoritarian regimes often lack these features (Tsebelis
2002). Leaders in authoritarian regimes are better able to pursue their subjective
will, as they are often considered to be what Hermann and Hermann (1989) call
 The Operational Codes of Fidel Castro and Kim Il Sung 155

“predominant leaders.” Yet, ironically, to the best of my knowledge, no system-

atic effort has been undertaken to theorize about the leaders of Cuba and North
Korea—a gap that I intend to fill.
Answering my primary research questions also allows me to shed some light
on an empirical, theoretically uninformed debate found in the literature on Cuba
and North Korea. Traditionally, scholars considered Castro and Kim to be hard-
line communists who rejected liberal ideas and policies on the domestic as well
as foreign policy level. With regard to foreign policy, Castro and Kim were char-
acterized as having rather hostile beliefs and attitudes toward the (overwhelm-
ingly capitalistic) outside world (on Cuba see Domínguez 1989: 29–33; Erisman
1985; on North Korea see Buzo 1999; Eberstadt and Ellings 2001; Kim 1998).
Throughout the cold war decades Cuba and North Korea pursued hard-line
policies of confrontation and were particularly antagonistic towards the United
States as the imperialist force of the capitalist world but also against most other
“capitalist” states.
This consensus regarding the beliefs and attitudes of Castro and Kim, how-
ever, largely disappeared upon the cold war’s end. According to some revisionist
interpretations of Cuban and North Korean foreign policy behavior in the post-
cold war era, the two states appear to have departed somewhat from policies of
“revolution” towards more moderate policies of pragmatism that seem to better
serve the national interest. Scholars now also tend to offer two conflicting inter-
pretations of these leaders.

Two Schools of Thought

One group of scholars, whom I refer to as the “status quo school,” argued that the
beliefs of Castro and Kim have not been affected and, therefore, remain autono-
mous from external events such as the end of the cold war and the Soviet col-
lapse. Put simply, they argue that Castro’s and Kim’s beliefs have not changed
and that any post-cold war behavior is due to external factors (on Castro see
Centeno 1997; Domínguez 1993, 1997; Mesa-Lago 1993a, 1993b; Ritter 1994;
Pickel 1998; Suchlicki 2000; on Castro and Kim see Mazarr 1991; on Kim see
Ho-yol 1992; Moon-young 1994; Youn 1999). On the other hand, a second
group of scholars, whom I call the “reformist school,” argued that the events
leading to the end of the cold war had a significant impact on Castro and Kim;
they argue that Castro and Kim have adjusted their beliefs in response to changes
in the international system, and furthermore, that post-cold war behavior is due
to altered beliefs (on Castro see Erisman 2000; Fernández 1994; Rich-Kaplowitz
1993; Smith 1992; on Kim see Kim 1994; Oh 1990; Sang-Woo 1991, 1992). In
short, these two schools disagree over whether the two leaders have exhibited
significant learning.
The debate and disagreement between the status quo and reformist schools poses
an interesting research puzzle that recurs within the literature on foreign policy
156 Akan Malici

decision making and political psychology (e.g., Breslauer and Tetlock 1991; Levy
1994; Stein 1994; Tetlock 1998): Do individuals learn from process and experi-
ence? If so, what do they learn? Furthermore, is learning related to foreign policy
behavior? In order to answer these research questions, I employ operational code
analysis (George 1969, 1979; Holsti 1977; Walker 1977, 1983, 1990).
Operational code analysis allows me to determine Castro’s and Kim’s philo-
sophical beliefs about the nature of the political universe—whether it is coopera-
tive or conflictual, and the nature of political conflict—as well as their instrumental
beliefs about the most effective strategies for accomplishing political goals. When
applied over time, operational code analysis makes it possible to detect changes in
a decision maker’s belief system as a result of learning (Walker et al. 2001; see also
Bennett 1999). Operational code analysis thus permits me to offer answers regard-
ing my primary research puzzle and thereby to function as arbitrator between the
status quo and reformist schools. While the explanandum of this study is beliefs,
there also appears to be a relationship between beliefs and behavior (George 1979;
Holsti 1976; Tetlock 1991, 1998). I shall formulate my hypothesis so that it is pos-
sible to falsify either of the interpretations given by the reformist and status quo
schools regarding the origins of Cuban and North Korean post-cold war behav-
ior. Doing so also allows me to address an auxiliary puzzle, namely whether in the
present cases beliefs are linked to behavior or whether these are epiphenomenal
vis-à-vis behavioral changes.

The Cold War: Soviet Withdrawal from Cuba and North Korea
In structural terms, the cold war era was marked by bipolarity—a condition in
which states are polarized into two opposing coalitions of about equal power
(Wagner 1993; Waltz 1979). However, the economic and political problems of
the Soviet Union as the leader of one of these coalitions led to the independence
of Eastern Europe and the dissolution of the Warsaw Pact. The ultimate outcome
was the end of the cold war accompanied by a change in the distribution of
power among states and a structural transformation from bipolarity to unipolarity
(Mastanduno 1997). Because units within the international system adapt to new
structural configurations, the end of the cold war also affected the particular
situations of individual states, which in turn affected dyadic relationships between
states (Lebow 1994). Thus, the disintegration of the Soviet Union as one of the
systemic poles changed the structure of the international system, but it also led to
situational changes within the particular dyadic relationships between the Soviet
Union (later Russia) vis-à-vis Cuba and North Korea.
For the vast part of the cold war era both Cuba and North Korea, as “fortresses
of communism,” were able to rely on the Soviet Union for economic and military
support in the form of aid and subsidized trade. This relationship extended into the
Gorbachev era, peaked there, but was then terminated by Gorbachev and other
reform-minded allies as they recognized and acknowledged the “errors” of their
 The Operational Codes of Fidel Castro and Kim Il Sung 157

past policies (Erisman 2000: 110–111).2 Relations that had formerly been with
other communist partners began being replaced by new relations with Western
states, especially the United States (Blacker 1993: 188; Blasier 1993: 68–70).
Subsequent trade losses at the end of the cold war would have been less damaging
to the economies of North Korea and Cuba had they been able to compensate
for the loss by initiating new partnerships. Yet they could not. During the time
period under consideration, North Korean total trade was virtually cut in half, and
Cuban total trade declined by almost 75 percent leaving both countries in severe
economic recessions (Mesa-Lago 1993b: 138; Erisman 2000: 148–149).
The magnitude of this ensuing economic recession becomes even clearer when
considering North Korea’s and Cuba’s GNP/GDP and the evolution of foreign
debt in the years surrounding these events.3 As shown in Figure 9.1, before 1990,
real GNP growth rates for North Korea were modest yet steadily positive, oscil-
lating between 2.1 and 3.3 percent. The abrupt downfall came in 1990, when
the Soviet Union undertook concrete measures that curtailed the North Korean
economy, which bottomed out in 1992 with a negative growth rate of 7.6 per-
cent. Though recovering somewhat in the following years, the North Korean
economy was unable to reach positive growth rates. Similar observations hold
for Cuba’s real GDP growth rates. Although we are missing data between 1986
and 1988, it appears as no coincidence that the real GDP growth rate, while still
positive in 1989, dropped into the negative in 1990—the same year that the
Soviet-Cuban economic agreement, which provided Cuba with much less favor-
able terms of trade than it had been accustomed to receiving, was renegotiated.
By 1993, real GDP reached rock bottom with a negative growth rate of 14.9
percent. Thereafter it recovered somewhat, yielding a rate of 0.7 percent in 1994
(Hernández-Catá 2000).
The magnitude of Cuba’s and North Korea’s economic recession is further-
more illustrated when considering the evolution of their foreign debt. Whereas
in 1985 North Korea’s foreign debt made for “only” 19.2 percent of its GNP,

5
0
% Change

–5
–10
–15
–20
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
Year
GNP Growth North Korea GDP Growth Cuba
Sources: Oh and Hassig (2000: 42), Hernández-Catá, (2000: 30),
United Nations (1997:47).

FIGURE 9.1 Real GNP/GDP Growth Rates, 1985–1994

158 Akan Malici

by 1991 North Korean foreign debt was 40.5 percent of GNP, and even reached
50.3 percent in 1994. Again, a similar pattern developed in Cuba. Whereas in
1985 Cuba’s foreign debt was 12.1 percent of its GDP, by 1991 foreign debt
accounted for as much as 35.6 percent of GDP, and this figure reached 59 percent
in 1993. In both cases the percentage of foreign debt to GNP/GDP increased
significantly after the Soviet Union pulled out (on Cuba see Erisman 2000: 122;
on North Korea see Oh and Hassig 2000). Parallel to these developments Fidel
Castro and Kim Il Sung had to witness how their former communist allies in
Eastern Europe were—sometimes violently—pushed out of their offices.
Did Castro and Kim consequently learn from these events, and if so, what did
they learn? These questions are derived from and motivated by my basic assump-
tions: (1) that the late 1980s and early 1990s was a catalytic period for these coun-
tries, and (2) that the mechanisms of continuity and change are situated in the
beliefs of Cuban and North Korean leadership.4 Although there is disagreement
over whether Castro and Kim learned, what scholars do seem to agree on is the
nature of Cuban and North Korean foreign policy behavior in the cold war era
as well as in the immediate post-cold war years, which is marked by a mixture
of change and continuity.5 On the one hand we observed some moderation and
pragmatism of Cuban and North Korean foreign policy behavior in the post-cold
war years. Instances are Cuba’s decreased participation in socialist revolutionary
movements, its withdrawal of armed forces in foreign countries, and, in the eco-
nomic realm, the legalization of a small-scale private sector, joint ventures, and
foreign ownership (Pickel 1998: 77). Similarly, North Korea conceded and signed
a 25-article Agreement on Reconciliation, Non-aggression, and Exchanges and
Cooperation (the “Basic Agreement”) with South Korea, signed the “Agreed
Framework” with the United States, and, in the economic realm, created a free
economic trading zone (FETZ) (Lee 2001: 60; Cotton 1998).
At the same time Cuba and North Korea demonstrated a tendency to resume
practices of the cold war era. Instances here are Cuba’s violent crackdown of
political dissidents, thereby violating international human rights, a continuously
harsh denunciation of United States’ foreign policy, or the rejection of United
States’ envoys (Fernández 1994: 57, 60). North Korea, for its part, demonstrated
continuity as it became delinquent in meeting its IAEA obligations, threatening
to withdraw from the Nuclear Nonproliferation Treaty (NPT), and engaging in
the reprocessing of plutonium (Manning 1998: 145; Snyder 1997).
Cuba’s and North Korea’s departure from cold war policies thus appeared
unstable at best. Accordingly, the “status quo” group of scholars questions the
sincerity of any changes that did occur in the post-cold war period. We should also
bear in mind that despite the dramatic changes in Cuba’s and North Korea’s posi-
tion in the international system, and unlike other former socialist states, Cuba and
North Korea remained socialist. It is a combination of these aspects that split the
scholarly community into a “status quo” group and a “reformist” group regarding
the analysis of North Korean and Cuban foreign policy in the post-cold war era.
 The Operational Codes of Fidel Castro and Kim Il Sung 159

Operational Code Analysis and Learning Dynamics

The “reformist” and “status quo” scholars agreed on the importance of Fidel
Castro and Kim Il Sung when examining North Korean and Cuban foreign pol-
icy making. The literature indicates that Fidel Castro and Kim Il Sung are to be
categorized as predominant leaders (Hermann and Hermann 1989) establishing
them as indispensable actors (Greenstein 1967). Regarding Castro, for example,
it is argued that:

To understand Cuban foreign policy, one must first understand Fidel

Castro. Even though Cuba has a more institutionalized and complex
political order than before, Castro remains the architect of foreign
policy, the final arbiter of policy disputes, and the ultimate authority whose
backing must be secured by all political subordinates … he remains Cuba’s
undisputed Socialist caudillo and lider maximo.
(Gonzalez and Ronfeldt 1986: 1)

Similarly, regarding Kim Il Sung it is argued that he “exhibit[s] the classic char-
acteristic of the ‘personalized power’ leader … Governance in North Korea is a
case of personalized rule” in which no one aside from Kim Il Sung has power (Oh
and Hassig 2000: 103). These and similar evaluations feed into Ole Holsti’s (1976:
30) argument that individual level approaches are particularly promising and rel-
evant when “decisions [are] made at the pinnacle of the government hierarchy
by leaders who are relatively free from organizational and other constraints—or
who may at least define their roles in ways that enhance their latitude for choice.”
The theoretical and empirical call for an individual level approach suggests that
operational code analysis as discussed in the introductory chapters to this volume
provides a satisfying method for examining my research puzzle.
The hypothesis that I advance to explain Cuban and North Korean foreign
policy behavior in the post-cold war era is a learning hypothesis. I follow Levy’s
(1994) definition of experiential learning to refer to changes in beliefs (see also
Tetlock 1991; Walker et al. 2001). Conceptualized within an operational code
framework, learning occurs if and when Castro’s and Kim Il Sung’s beliefs about
the nature of the political universe and the most effective means to achieve politi-
cal goals are strengthened, weakened, or altered altogether. Operational code
analysis also allows me to distinguish among three levels of experiential learning:
simple, diagnostic, and complex (Deutsch 1963; Tetlock 1991; Leng 2000; Levy
1994; Nye 1987; Walker et al. 2001).
Simple learning is defined as changes in instrumental beliefs about the best
means to achieve goals, and diagnostic learning is defined as changes in phil-
osophical beliefs about the political universe. Complex learning occurs when
an actor’s key philosophical beliefs about political goals and key instrumental
beliefs about the most effective means to achieve them are modified so as to
160 Akan Malici

alter a leader’s strategic preferences. Walker et al. (2001) identify the leader’s
image of the political universe (P-1), the belief in the ability to control histori-
cal development (P-4), and the leader’s belief regarding the most effective strat-
egy for achieving political goals (I-1) as key beliefs (see also Walker et al. 2003:
231–235). Their corresponding indices aggregate the basic elements (attributions
and valences) used to construct the indices of all the remaining operational code
beliefs into summary indices of beliefs about the power relationship (P-4) and the
cooperative or conflictual orientations between self (I-1) and other (P-1). The
answer to my basic research question, “Did Fidel Castro and Kim Il Sung learn
from their cold war experiences, and if so, what did they learn?” lies in identify-
ing the learning that the leaders experienced, and evaluating what they learned at
each level of learning.
An important distinction is to be drawn between any substantive change and
a statistically significant change of beliefs (Achen 1982). While any substantive
changes inform us about a shift in a leader’s beliefs in the context of a learning
hypothesis, they could have occurred simply through random chance. Applying a
test of statistical significance to evaluate the change lessens this possibility, allow-
ing us to draw conclusions with more confidence. My yardstick for measuring
learning is thus a change in a leader’s beliefs at the p < .05 or p < .10 level of
significance, depending on whether or not it is a directional hypothesis being
tested.
One cautionary note is appropriate here, and that is learning means changes
in beliefs and does not automatically involve behavioral changes.6 “The latter,”
as Walker et al. (2001) point out, “could reflect a leader’s adaptation to chang-
ing circumstances (structural adaptation) or simply a reaction to a stimulus (social
learning) without a corresponding change in important beliefs” (see also Levy
1994). Such a distinction is important for the assessment of whether any Cuban or
North Korean departure from cold war practices was simply structural adaptation
due to a fundamental change of the international environment and Cuba’s and
North Korea’s positions within it after 1991, or just behavioral adjustment due to
social learning following the Soviet withdrawal.
If experiential learning is related to the conduct of Cuban and North Korean
post-cold war foreign policy, then Cuban and North Korean operational code
beliefs should experience a change following the end of the cold war and the
Soviet withdrawal. This main hypothesis is consistent with the shift of Cuban
and North Korean foreign policy towards some pragmatism for the time period
after 1991. It is stated so that it is possible to falsify the empirical interpretations of
either the “status quo” scholars or “reformist” scholars. More specifically, if the
hypothesis is confirmed, the interpretations of the “reformist school” bear more
validity. On the other hand, if the hypothesis is rejected, then the conclusion
would be that Fidel Castro and Kim Il Sung did not learn and that the interpreta-
tion of the status quo school bears more validity.
 The Operational Codes of Fidel Castro and Kim Il Sung 161

The Operational Codes of Fidel Castro and Kim Il Sung

In order to assess Fidel Castro and Kim Il Sung’s operational code beliefs I employ
the Verbs in Context System (VICS) as discussed in Chapter 4 of this volume.7
My strategy for assessing my research questions is to compare and contrast the
VICS scores of each leader’s operational code over time as well as to each other.
Thus, the independent variables for my analysis are time period and leader while
the operational code beliefs serve as the dependent variables.
I also compare the scores of Fidel Castro and Kim Il Sung to the mean scores of
a collection of 164 speech acts given by thirty different and diverse world leaders—
leaders of poor and rich states and weak and strong states. This collection is referred
to as a norming group. Making this set of comparisons allows me to put the Cuban
and North Korean leaders into a broader perspective, which is instructive because
it permits us to develop a better sense of how the leaders in my study—Third
World communist leaders who experienced significant political and economic
crises at the end of the cold war—compare to the “average leader.”8 I conclude
my analysis with a comparison of the two leaders in the post-cold war period.
The results from the difference of means test between Castro’s cold war (1980–
1985) operational code and that of the norming group, summarized in Table 9.1,
suggest that Castro’s beliefs differed significantly from those of the average leader
in a number of ways. All of Castro’s philosophical beliefs differed significantly
from the norming group such that Castro viewed the political universe in more
conflictual terms (P-1), was more pessimistic about the realization of political
values (P-2), thought the political future was less predictable (P-3), believed he
had less control over historical development (P-4), and thought chance played
a large role in political outcomes. Several of Castro’s instrumental beliefs were
also significantly different from the norming group. During the cold war, Castro
had a more conflictual strategic approach to goals (I-1), was less acceptant of
risk (I-3), had a significantly higher propensity to shift between cooperative
and conflictual tactics (I-4a) and between words and deeds (I-4b), and ascribed
significantly less utility to Appeal tactics (I-5c) and more utility to Threaten tactics
(I-5e) as a means of goal attainment.
The results from the difference of means test between Kim Il Sung’s cold
war (1980–1985) operational code and that of the norming group display a
somewhat similar pattern. During the cold war period, Kim Il Sung believed that
he had significantly less control over historical development (P-4), but he was
significantly more optimistic about the realization of political values (P-2) when
compared to the norming group. Like Castro, Kim was less acceptant of risk and
had a comparatively high propensity to shift between words and deeds (I-4b)
during the cold war period. Finally, with respect to the average leader, Kim Il
Sung ascribed significantly less utility to Appeal (I-5c) and Oppose (I-5d) tactics,
and significantly more utility to Reward (I-5a) and Punish (I-5f) tactics as a means
for attaining political goals during the cold war period.
162 Akan Malici

TABLE 9.1 Operational Codes of Castro and Kim Compared to Norming Group

Castro Norming Group Kim Il Sung

Philosophical and Instrumental Beliefs (n = 13) (n = 164) (n = 13)

P-1. Nature of Political Universe 0.081* 0.301 0.321

(Conflict/Cooperation)
P-2. Realization of Political Values –0.001* 0.147 0.239*
(Pessimism/Optimism)
P-3. Political Future 0.101* 0.134 0.123
(Unpredictable/Predictable)
P-4. Historical Development 0.161* 0.224 0.165*
(Low Control/High Control)
P-5. Role of Chance 0.983 0.968 0.980
(Small Role/Large Role)
I-1. Strategic Approach to Goals 0.154* 0.401 0.383
(Conflict/Cooperation)
I-2. Intensity of Tactics 0.094 0.178 0.202
(Conflict/Cooperation)
I-3. Risk Orientation 0.116* 0.332 0.194*
(Averse/Acceptant)
I-4. Timing of Action
a. Conflict/Cooperation 0.664* 0.503 0.566
b. Words/Deeds 0.694* 0.464 0.659*
I-5. Utility of Means
a. Reward 0.202 0.157 0.261*
b. Promise 0.106 0.075 0.096
c. Appeal/Support 0.269* 0.468 0.335*
d. Oppose/Resist 0.200 0.154 0.084*
e. Threaten 0.066* 0.034 0.054
f. Punish 0.158 0.112 0.171*

* Significant difference between each leader and the norming group at the p < .10 level (two-tailed
test).

In sum, Castro and Kim are distinct from the average world leader. However,
Castro’s beliefs deviate from the norming group more so than Kim’s. Castro is
distinct with respect to all of the key operational code beliefs (P-1, P-4, I-1) while
Kim is different with respect to one of them (P-4). Geopolitical reasons are likely
to figure in Castro’s more intense hostility. Cuba is in close proximity to the U.S.
superpower, while North Korea is located thousands of miles away from the Western
enemy. Nevertheless, both leaders show significant differences on a number of the
remaining operational code beliefs, demonstrating their unique identities.

Learning and Leader Comparisons

My main substantive concern in this chapter is the disagreement between the two
contending groups of scholars: Did the end of the cold war lead these leaders to
 The Operational Codes of Fidel Castro and Kim Il Sung 163

alter their beliefs as the reformist group of scholars argues? Or do Fidel Castro’s
and Kim Il Sung’s beliefs continue into the post-cold war period despite their
new, isolated position in the international system as the status quo group of
scholars suggest? Determining whether or not Castro and Kim learned requires
me to compare the leaders’ cold war operational codes to their post-cold war
operational codes. To test the reformist school’s learning hypothesis that the
two rogue leaders have shifted their foreign policy orientations in a cooperative
direction, I have conducted a one-way Analysis of Variance (ANOVA) for each
leader using time period as the independent factor and the operational code
indices as dependent variables.
When comparing Castro’s cold war (1980–1985) operational code to his post-
cold war (1991–1994) operational code, a few statistically significant differences
emerge. The ANOVA results in Table 9.2 suggest that Castro believed he had
significantly more control over historical development (P-4) in the post-cold war
period compared to the cold war period. This perception of increased control
is most likely the result of the collapse of the Soviet Union as the center of the
ideological battle against Western capitalism. Now he considered himself to be
in charge of continuing the battle. Only a few of Castro’s instrumental beliefs
changed significantly from the cold war to the post-cold war period—in the
post-cold war period Castro was less likely to ascribe utility to Promise tactics
(I-5b) and more likely to ascribe utility to Appeal (I-5c) tactics as a means of
attaining political goals than he was during the cold war period. Unlike Castro,
Kim Il Sung’s cold war and post-cold war operational codes in Table 9.2 show
no significant differences.
Although Castro perceived that he had significantly more control over historical
development and showed more inclination to appeal to others in the post-cold
war period, a consideration of his remaining operational code beliefs indicates
that overall he continued to maintain a conflictual stance. These results seem to
suggest that Castro engaged at best in modest learning, and Kim Il Sung did not
learn. That is to say, whereas one of Castro’s philosophical beliefs about political
goals and two of his instrumental beliefs about the most effective means to achieve
them were modified from the cold war to the post-cold war period, changes in
Kim Il Sung’s beliefs were not significant.
These findings lend little support to the reformist group of Cuban scholars and
strong support to the status quo group of North Korean scholars. Only one of
Castro’s key operational code indices (P-4) changed significantly between the cold
war and post-cold war period. For Kim Il Sung the conclusion is straightforward
since he did not engage in any learning. Therefore, I conclude that the changes
in Cuban and North Korean foreign policies following the cold war were more
likely due to structural adaptation to changing circumstances than to experiential
learning by Castro and Kim. The lack of learning by the two leaders is consistent
with the erratic pattern and the lack of lasting change in their foreign policies
despite radical shocks from their domestic and international environments.
164 Akan Malici

TABLE 9.2 Fidel Castro’s and Kim Il Sung’s Cold War and Post-Cold War Operational
Codes

Fidel Castro Kim Il Sung

1980–1985 1991–1994 1980–1985 1991–1994
Philosophical and Instrumental Beliefs (n = 13) (n = 11) (n = 13) (n = 8)

P-1. Nature of Political Universe 0.081 0.183 0.322 0.383

(Conflict/Cooperation)
P-2. Realization of Political Values –0.001 0.063 0.239 0.260
(Pessimism/Optimism)
P-3. Political Future 0.101 0.098 0.123 0.147
(Unpredictable/Predictable)
P-4. Historical Development 0.161* 0.205* 0.165 0.196
(Low Control/High Control)
P-5. Role of Chance 0.983 0.980 0.980 0.970
(Small Role/Large Role)
I-1. Strategic Approach to Goals 0.154 0.370 0.383 0.438
(Conflict/Cooperation)
I-2. Intensity of Tactics 0.094 0.182 0.202 0.249
(Conflict/Cooperation)
I-3. Risk Orientation 0.116 0.167 0.194 0.203
(Averse/Acceptant)
I-4. Timing of Action
a. Conflict/Cooperation 0.664 0.610 0.566 0.562
b. Words/Deeds 0.694 0.632 0.659 0.762
I-5. Utility of Means
a. Reward 0.202 0.213 0.261 0.271
b. Promise 0.106* 0.054* 0.096 0.065
c. Appeal/Support 0.269* 0.418* 0.335 0.384
d. Oppose/Resist 0.200 0.133 0.084 0.092
e. Threaten 0.066 0.059 0.054 0.079
f. Punish 0.158 0.123 0.171 0.110

* Significant difference between indices at the p < .05 level (one-tailed test).

In addition to identifying whether or not Fidel Castro and Kim Il Sung learned
from their cold war experiences, I am also interested in knowing if the two
leaders’ beliefs differ significantly from each other, and if they do, whether the
variance in beliefs between the two leaders is due (a) primarily to their individual
differences, (b) to the changes in the circumstances they faced at the end of the
cold war, or (c) a combination thereof. In order to examine this question, I
conducted a Multivariate Analysis of Variance (MANOVA) using time period
(cold war or post-cold war) and leader (Castro or Kim) as the independent factors
and the operational code indices as the dependent variables.
The results of the MANOVA showed that there were significant main effects
for both leader and time period; however, there were no significant interaction
 The Operational Codes of Fidel Castro and Kim Il Sung 165

effects. These findings suggest that the changes in the conditions that confronted
Cuba and North Korea at the end of the cold war (measured by the time period
factor) were substantial enough to lead to some belief differences independent of
the individual leaders’ identities. No matter whether the leader was Kim Il Sung
or Fidel Castro, the end of the cold war led to significant belief changes. The
leaders thought they had more control over historical development (P-4 = 0.201
vs. 0.163), and they were more likely to ascribe utility to Appeal tactics (I-5c =
0.404 vs. 0.302) as a means for attaining political goals in the post-cold war period
compared to the cold war period.
The MANOVA results also revealed that the significant main effects for leader
were more substantial than those for time period, suggesting that there are several
important differences between Fidel Castro and Kim Il Sung, independent of the
global changes accompanying the end of the cold war. In the realm of philosophical
beliefs, Castro viewed the nature of the political universe in significantly more
conflictual terms (P-1 = 0.128 vs. 0.345), was significantly more pessimistic about
the realization of political values (P-2 = 0.029 vs. 0.247), and thought the political
future significantly less predictable (P-3 = 0.099 vs. 0.132) than Kim Il Sung. In
the realm of instrumental beliefs, the results suggest that compared to Kim, Castro
was significantly more likely to ascribe utility to Oppose tactics (I-5d = 0.169
vs. 0.087) as a means of accomplishing political goals. The Strategic Approach
to Goals index (I-1) approaches significance (p < .109), and if interpreted as
significant, suggests that Castro had a much more conflictual approach to goals
than did Kim Il Sung (I-1 = 0.253 vs. 0.404). It seems that although the end of the
cold war had an impact on these two leaders, Fidel Castro and Kim Il Sung have
significantly different beliefs about political goals and the most efficient means to
achieve these goals.9

Conclusion
My analysis leads me to the following conclusions. First, the belief systems of
Kim Il Sung and Fidel Castro differed significantly from the average leader’s
with Castro displaying more dramatic deviations in the form of significant dif-
ferences for the key beliefs affecting general strategic orientation (I-1), image of
others (P-2), and control over historical development (P-4). Second, while Fidel
Castro shows some evidence of simple and diagnostic learning in the post-cold
war period, I argue for a cautious interpretation because only a few of his beliefs
changed significantly, and only one of the indices for key beliefs (P-4) shows
significant learning. Kim Il Sung’s beliefs did not change signficantly, indicating
that he did not engage in any learning. Third, the evidence from the MANOVA
presented above also suggests that the end of the cold war had only a modest
impact on the Cuban and North Korean leadership, independent of the specific
regional circumstances, personalities or identities of Fidel Castro and Kim Il Sung.
Instead, when I control for the tumultuous events that have accompanied the
166 Akan Malici

end of the cold war, there are significant differences between Castro’s and Kim’s
belief systems.
These results lead me back to a debate introduced at the beginning of this
chapter, namely the debate between “structuralists” and “cognitivists.” And it
is within the context of this debate that I want to set forth some more general
conclusions. “When do governments change course, and why?” is a question
that has received considerable attention in the international relations literature
(Hermann 1990; Wendt 1992). Structural approaches focus on changes in the
international environment and predict that they will impact the foreign policies
of individual states. During the late 1980s and early 1990s the configuration of
the international system, Cuba’s and North Korea’s position within it, and the
dyadic relationships between the Soviet Union vis-à-vis Cuba and North Korea
changed profoundly.
Despite forceful arguments made by structuralists (Waltz 1979; Wolfers 1962)
that these “external shocks” (Hermann 1990) should lead to changes in the for-
eign policy behavior especially of small states, Cuba and North Korea continue on
the status quo path with rather minor changes in their foreign policies. I contend,
therefore, that changes in foreign policy will be missing in the absence of change
at the individual level. In opposition to structural approaches, we conclude that
governments and their respective leaders do not necessarily “respond in a rational
manner to the reward and punishment contingencies of the international envi-
ronment” (Tetlock 1991: 24). Instead, the mechanisms for continuity and change
are located in the belief systems of leaders. I also agree with Wendt (1992) that a
change in leadership may be more potent than an environmental change, as in the
case of Mikhail Gorbachev’s accession to power in the USSR and his subsequent
initiatives to end the cold war.
An alternative strategy for foreign policy change may be to alter the Cuban and
North Korean leaderships’ perceived security context. Both consider the United
States as a main threat to their respective regimes. By employing a constructive
strategy of increased and repetitive engagement in the Caribbean and on the
Korean Peninsula, the United States could potentially contradict the perceptions
of the Cuban and North Korean leadership and thereby compel these states to
change their foreign policies. Specifically an altercasting strategy may be appro-
priate. This strategy entails continuous gestures (moves and tactics) that an oppo-
nent would not expect. It is similar to Charles Osgood’s (1960, 1962) Graduated
Reciprocation In Tension-reduction (GRIT) strategy. Osgood (1962: 96–103)
argued that unilateral cooperative initiatives should be unexpected surprise
moves, should explicitly invite reciprocation, and should be continued over a
considerable period of time even if reciprocation is not immediately forthcoming.
Most famously, Mikhail Gorbachev adopted such a strategy, changing President
Reagan’s and President Bush’s “enemy image” perception of the Soviet Union,
and was able to end the enduring rivalry between the United States and the Soviet
Union (see Cha 2002; Stein 1994; Walker 1992, 2000b; Wendt 1999).
 The Operational Codes of Fidel Castro and Kim Il Sung 167

Notes
* Reprinted with permission and with some alterations from Political Psychology, Vol. 26,
No. 3 (June 2005): 387–412.
1 In a comparison of great powers and small states, James Rosenau (1966: 47–48) for
example argues that features of the international environment are more potent regarding
a small state’s foreign policy decisions. This assertion is underlined by Kenneth Waltz’s
reasoning that small states are to be more attentive to external constraints because they
“operate on narrow margins” (Waltz 1979: 195). Similarly Robert Jervis argues that
since small states lack the resources of great powers, they also lack a “margin of time
and error” in their responses to occurrences in the international environment. The
international environment, therefore, dictates that leaders of small states be “closely
attuned” to it (Jervis 1978: 172–173). In a strikingly similar fashion Jack Snyder (1992:
317–318) argues that small states, as opposed to great powers, do not “enjoy a substantial
buffer from the pressures of international competition.” Small states are therefore
“more exposed to the vagaries of international security and economic competition.”
2 Regarding Gorbachev’s policy changes, Stein (1994) argues that Gorbachev changed
his policies by “trial and error” learning, and Wohlforth (1994/95: 115) argues that
“cognitive change was compelled by mounting negative feedback.” See also Kupchan
(1994: 495ff.).
3 Although I have made extensive efforts at data collection, I have encountered some
limits. I could not find a consistent measure of macro-economic growth across the two
cases, which is why I report real GNP growth rates for North Korea and real GDP
growth rates for Cuba. Additionally, I have been unable to locate real GDP growth
rates for Cuba for the years 1986, 1987, and 1988. For North Korea’s real GNP
growth, see Oh and Hassig (2000: 42). For Cuba’s real GDP growth from 1988–1994,
see Hernández-Catá (2000: 30), and for 1985, see United Nations (1997: 47).
4 Critics might counter and argue that Fidel Castro and Kim Il Sung were not entirely
free to learn and to change their attitudes as major concessions to the capitalist world
would have endangered their political standing at home. A slightly different argument
would be that if Castro or Kim were to make concessions toward the aggressively
portrayed capitalist world, they would no longer have the “Yankee imperialists” to
blame for domestic problems, and anti-Americanism would no longer serve to unify the
Cuban and North Korean peoples under the Castro and Kim regimes. However, both
counterarguments are mistaken. As cases in Eastern Europe efficiently demonstrated,
it is not concessions that endanger a regime, but rather the continuation of cold war
practices. Regarding the second argument, the idea that the Cuban and North Korean
people were unified behind their leaders is simply wrongheaded. Indeed, scholars have
often commented on the eroding political legitimacy of Castro’s and Kim’s regimes
(Mazarr 1991; Gonzalez 1996: 19–20, 28–32; Oh and Hassig 2000: 127–147). Hence,
the above assertion remains: the events between 1985 and 1991 provided ample
opportunity and incentive for Castro and Kim to learn.
5 For a detailed account of Cuban and North Korean foreign policy behavior in the
post-cold war period see Erisman (2000), Fernández (1994), Henriksen (1997), Lee
(2001).
6 For a contrasting view/definition see Jarosz and Nye (1989: 130) who define learning
as “the acquisition of new knowledge or information that leads to a change in
behavior.”
7 To the extent possible, we followed Walker et al. (1998: 182) who set the following
criteria for foreign policy speeches: “(1) the subject and object are international in scope;
(2) the focus of interaction is a political issue; (3) the words and deeds are cooperative or
conflictual.” The sample of speeches for Fidel Castro was randomly drawn from the Lexis
Nexis Academic Universe databank and the University of Texas Latin American Network
168 Akan Malici

Information Center (LANIC) Castro Speech Database. My final sample included thirteen
speeches for the cold war period (1981–1985) and eleven foreign policy speeches for
the post-cold war period (1991–1994). Due to the scarcity of speeches by Kim Il Sung,
random sampling was not possible. Databases, such as Lexis Nexis, and the Foreign Broadcast
Information Service (FBIS) served as my main sources for the sample. In addition, some
speeches in my sample were taken from Kim Il Sung: For an Independent World (1986), a
collection of his speeches and interviews. From these sources we selected those speeches
that meet the above criteria best. Despite this effort, the ratio between domestic and
foreign policy content contained in these speeches is somewhat more balanced than was
the case for Castro’s sample. My final sample of Kim Il Sung’s rhetoric included thirteen
speeches for the cold war period (1981–1985) and eight speeches for the post-cold war
period (1991–1994).
8 The data for the norming group were obtained from Professor Mark Schafer,
Department of Political Science, 240 Stubbs Hall, Louisiana State University, Baton
Rouge, LA 70803-5433.
9 A static comparison of just the post-cold war operational codes of Castro and Kim
reveals similar differences regarding philosophical beliefs but no significant differences
for instrumental beliefs. Castro’s view of the political universe is significantly more
conflictual (P-1), more pessimistic (P-2), and less predictable (P-3) than Kim’s.
10
STABILITY AND CHANGE
IN BELIEF SYSTEMS
The Operational Code of George W. Bush
from Governor to Second Term President

Jonathan Renshon*

Introduction
Those who study international politics and foreign policy decision making are
faced with a paradox. On one hand, the beliefs of leaders seem critical to under-
standing their foreign policy decisions. Empirically, they are important because
of the utility that they provide to political leaders by—among other things—sug-
gesting ready frameworks for analyzing both novel and common situations and
imposing order on what would otherwise be overwhelming amounts of infor-
mation. On a theoretical level they are important because of the influence they
exert on many facets of international politics—from how leaders think about and
respond to public opinion (Foyle 1999) to what type of international system is
constituted by the mutually reinforcing beliefs of world leaders (Wendt 1992).
There is also evidence that beliefs are relevant on multiple levels of analysis, from
the individual to the cultural level (Johnston 1995). In fact, unless one denies the
importance of looking at anything beyond objective capabilities, beliefs must take
a central place in one’s analysis as either independent or mediating variables.
On the other hand, despite its central importance, our understanding of when
and why beliefs change is far from perfect. A survey of the relevant literature
in fields of social and cognitive psychology, behavioral economics, and political
psychology reveals a surprising lack of attention paid to the question of when and
why beliefs change. Do beliefs shift imperceptibly over time? Do they change
instantaneously in response to traumatic events? Must a cluster of related beliefs
change at the same time? Or can specific beliefs change in isolation (implying a
more fragmented belief system)?
In this chapter I examine the issue of belief change by using the “operational
code” construct. As explained in much more detail in earlier chapters, operational
170 Jonathan Renshon

codes are a particular subset of an individual’s beliefs—those focused on the politi-

cal universe. These beliefs structure and order reality for decision makers, and help
them to sort the signals in their environment from the noise. They exert a signifi-
cant influence on how leaders interpret information, perceive the social environ-
ment, and make decisions. As such, they have a prima facie claim to relevance in
political leadership and decision making.
This chapter focuses on the changes in the foreign policy operational code
of President George W. Bush (hereafter GWB) over four distinct time periods:
(1) pre-presidential beliefs prior to inauguration (1998–January 19, 2001); (2)
pre-September 11th beliefs following inauguration (January 20, 2001–September
10, 2001); (3) post-September 11th beliefs covering the six-month period after
9/11 (September 11, 2001–March 11, 2002); and (4) end-of-term presidency
beliefs covering the last two years of his second term (January 20, 2007–July
20, 2007). These four periods constitute a natural experiment of sorts in which
we can examine the effects of three “classes” of events (three independent vari-
ables) on presidential beliefs. The transition from Phase 1 to Phase 2 represents a
change in Mr. Bush’s political role—from Governor and presidential candidate to
President—and thus allows us to examine whether or not beliefs change when
one’s role changes substantially. The terrorist attacks of September 11th, 2001, are
representative of a second class of belief-changing event: the “traumatic shock.”
The third class of potential belief-changing event is one that can be termed the
“learning in office” model of belief change. In this case, we can see the effect of
roughly six years of time spent as President of the United States, and examine
how belief change might occur in the absence of sudden shocks or role changes.
This sequence of phases allows us to examine how stable beliefs are over longer
periods of time.
Substantively, the empirical analysis of President Bush’s operational code and
strategic beliefs will contribute to a greater understanding of some of his more
consequential and controversial decisions. On a theoretical level, this chapter
aims at furthering our understanding of whether beliefs change and what types of
events are likely to effect that change. While several operational code analyses in
the past have indirectly examined some of these questions and provided some rel-
evant findings, we are still left without a strong theoretical basis for understanding
how and why beliefs change. The results of this analysis provide some surprising
perspectives on these crucial questions

What Are Beliefs and Why Are They Important?

Before proceeding to the analysis of GWB’s belief system, it is worthwhile to
briefly situate the operational code construct within the larger study of beliefs. In
doing so, there are several questions relevant to this study: What are beliefs? Why
are they important in the context of political decision making? And finally, what
do we know about how, when, and why they change?
 The Operational Code of George W. Bush 171

We must take care to clearly define what we mean when we say “belief.”
Beliefs are, in essence, that which we hold to be true. They may be propositions
about causal relationships or fundamental assumptions about the way the world
operates. There is also a strong element of probability in beliefs. That is, there are
varying degrees of certainty in beliefs; individuals believe some propositions to
be valid at a level approaching certainty while they may have relatively less con-
fidence in the validity of other propositions but still believe them to be generally
true (Fishbein 1967: 258). Beliefs are also distinct from both schemas (Larson
1994) and attitudes (Krech and Crutchfield 1948; Katz and Stotland 1959; Katz
1960). Attitudes have a “tripartite structure” that includes a cognitive component
(beliefs), but also includes conative (behavioral) and affective components, while
schemas differ from beliefs in being more “inclusive and individualistic” (Larson
1994: 18–19).
One main finding over the past half-century has been that individuals tend to
filter new information through their pre-existing beliefs in such a manner as to
maintain the consistency of their beliefs (Jervis 1976: 117; 2006: 651). Yet even
more important than the assumed outcome (belief stability) is the fact of the pro-
cess itself: beliefs act as a primary filter through which other perceptual processes
operate, whether the effect is to maintain the stability or not (Lane 1962; Knutson
1972). In the context of political decision making, leaders react not to an objec-
tive reality but to a subjective reality that is filtered through their belief system
(Holsti 1962: 244; Boulding 1969: 423).
Individuals do this because the world that they face is inherently complex. It
is full of contradictions, ambiguity, and surprises. Because of their relatively lim-
ited ability to process the vast amounts of information that they are faced with,
humans are, by necessity, “theory-driven” (Fiske and Taylor 1991; Simon 1985).
Beliefs provide essential utility for decision makers and facilitate decision mak-
ing by providing ready frameworks for analyzing novel situations and provide
answers to fundamental questions about the way the world works and the sources
of human behavior (Tetlock 1998; Rosati 2000). In short, they do nothing less
than help to impose meaning and order on what would otherwise be a “confusing
and overwhelming array of signals” picked up from our senses and the environ-
ment (George 1980: 57).
Despite the central importance of beliefs in social and cognitive psychology,
we are still far from understanding how, when, and why beliefs change.1 This is
because the study of beliefs has mostly focused on the mechanisms that individu-
als utilize to maintain stability in their beliefs, and the debate over the extent to
which beliefs systems are internally coherent. The study of beliefs has long been
dominated by the premise of “cognitive consistency.” The important features of
this model of cognition are that beliefs are stable over time, internally consistent/
interdependent, and hierarchical.
Cognitive consistency predicts the relative stability of core beliefs over time, as
a result of the strong tendency for people to assimilate new information in such
172 Jonathan Renshon

a manner as to conform to their pre-existing beliefs (Jervis 1976: 118; see also
Deutsch and Merritt 1965: 145). If beliefs act as a lens through which we view
reality, then that lens does not seem to be readily altered. Confirming evidence
is easily assimilated into our pre-existing beliefs, while dissonant information is
discredited, minimized, or ignored altogether.2 The tendency to assimilate infor-
mation in this particular way has the effect of keeping beliefs relatively consistent
over time.
The tendency for beliefs to remain stable has been demonstrated convincingly
in the laboratory. For instance, studies of “belief perseverance” have noted the
tendency for individuals to rigidly adhere to their initial beliefs about a given
proposition or causal relationship. Under experimental conditions, individuals
were given initial data about some casual relationship (for example, a firefighter’s
risk propensity and their success as a firefighter) upon which they based their
beliefs. Even when subjects are told or shown that the initial data were fictitious,
individuals largely clung to their beliefs despite the new information (Anderson et
al. 1980; Anderson 1983).
In addition to temporal stability, cognitive consistency also assumes a belief
system that is internally consistent. That is, different beliefs are consistent (or
logically coherent) with one another. Much of the research in this area has focused
on the strategies that individuals utilize to maintain internal coherence in their
belief systems (Festinger 1957; Heider 1958; Abelson 1967). More recently, a
debate has opened on whether individuals can be viewed as “consistency seekers”
(who are motivated to reduce discrepancies among beliefs) or “cognitive misers”
who are likely to have more internally fragmented belief systems.
Of course, to some extent the issues of internal and temporal consistency are
interrelated. The same strategies that individuals employ to maintain internal
consistency (to the extent that they strive to do so) at any given moment have
the effect of producing beliefs that are relatively stable temporally as well. The
last significant aspect of the cognitive consistency model is that beliefs are
hierarchical. This means that any given belief can be thought of as on a continuum
between “core” (more important/fundamental) and “peripheral” (less important/
fundamental). More fundamental, core beliefs are highly resistant to change, and
help to structure other, peripheral beliefs, which are more prone to change. When
a core belief does change, other peripheral beliefs should change as well in order to
maintain internal consistency.

Stability and Change in Operational Codes

Despite its clear import for the study of politics, the questions of whether beliefs
change, and if so, what causes them to do so has produced very little in the way
of empirical research or development of theories in political science. Because the
operational code is quantifiable and the results of its analysis are reproducible,
the examination of it can contribute a great deal to our basic understanding of
 The Operational Code of George W. Bush 173

belief change. Before proceeding to our analysis, it will be useful to summarize

the results of the few studies that have examined how (and whether) operational
codes change. However, the reader will notice that few of these studies sum-
marized below provide concrete or generalizable hypotheses about the nature of
belief change, i.e., which beliefs change, when are they likely to change, and why
do they change?
In the past, change has generally been conceptualized as any statistically sig-
nificant shift in the operational code. The statistical significance reveals whether a
change has occurred, and the magnitude of the change reflects how big a change
has occurred. However, there is a third dimension that is relevant, but often over-
looked: the type of change. Beliefs can change in one of two ways; they can be
reinforced, or they can be reversed. Both types of change are important, and wor-
thy of our attention, yet there is a clear difference between the two. In one case,
prior beliefs (“the world is friendly”) are reinforced (“the world is very friendly”),
while in the other case, they are changed in the opposite direction (“before I
thought the world was friendly, now I believe it to be more hostile”).3
The literature on operational code change can be divided into two major
groups—those that evaluate a given leader’s operational code longitudinally (and
are interested in how their beliefs evolve over time) and those that examine the
effect of specific events (e.g. the end of the cold war) on the leader’s beliefs. In
the first group, the most important trend to note is that operational codes (read:
beliefs) can change over time. Over the period between the 1970s and 1990s,
for instance, the philosophical beliefs of both Yitzhak Rabin and Shimon Peres
changed significantly: their generally pessimistic views of the world were amelio-
rated over the course of the two decades (Crichlow 1998: 695). In a similar vein,
Malici and Malici (2005a) found that the end of the cold war had no discernable
affect on the operational code of Kim Il Sung and a very small effect on that of
Fidel Castro’s belief in his control over historical development, and the utility of
promises and appeal/support tactics.
More recently, several scholars have examined the effects of specific events on
a leader’s operational code. Walker et al. (1998) analyzed the operational code
of President Jimmy Carter both before (1977–1979) and after (1980) the Soviet
invasion of Afghanistan. They found a large and statistically significant decrease
in his cooperative view of the political universe and in his optimism concerning
the realization of his goals (Walker et al. 1998: 185–186). This pattern illustrates
a “reversal” of beliefs in which Carter’s previously cooperative and friendly view
of the political universe in 1977–1979 changed in the opposite direction in 1980,
becoming notably more pessimistic and conflictual.
Walker and Schafer (2000b) found that the operational code of President
Lyndon Johnson shifted between two phases of planning for the war in Vietnam
in 1965–1965. Specifically, he exhibited a loss of control, attributed a greater
role to chance, and became more risk averse (Walker and Schafer 2000b: 537).
And most recently, Feng found that the Korean War affected all five of Mao’s
174 Jonathan Renshon

philosophical beliefs, which became more hostile and confrontational during the
conflict (Feng 2005a: 657). These changes are illustrative of the reinforcement of
prior beliefs. Mao’s views of the political universe were already on the hostile and
conflictual end of the spectrum, but these beliefs were reinforced and strength-
ened in wartime.
These earlier research projects provide a significant empirical foundation for
this project to build upon. The findings suggest that individuals’ operational codes
can change, and that they can do so in a rather limited time frame of one to two
years. Yet, while they provide an important foundation, it is easy to see that there
are still vast gaps in our understanding of how operational codes change. Only
a small handful of these projects were designed in such a way as to distinguish
between multiple potential causes of belief change. So while they were able to
provide support for the notion that operational codes were more flexible than
originally believed, a crucial first step in this research program, they were unable
to suggest conclusively how and why these changes occurred. This chapter builds
on those works, and the small number of recent projects that have examined
potential belief-changing events (such as the Korean and Vietnam wars), to sug-
gest several independent variables drawn from psychological theories of belief
change as well as operational code analysis.
An interesting trend that emerges from these studies concerns which beliefs
were stable and which were prone to change. The results above indicate that phil-
osophical beliefs are more prone to change than instrumental beliefs (Walker and
Schafer 2000a: 4). This pattern was found in analysis of the operational codes of
Bill Clinton, Mao Zedong, Deng Xiaoping, Jiang Zemin, and Hu Jintao (Schafer
and Crichlow 2000: 563; Feng 2005a: 657; Feng 2005b: 21).4 When instrumental
beliefs did change along with philosophical beliefs, as they did in the operational
code of Jimmy Carter, Lyndon Johnson, and Fidel Castro (Walker et al. 1998:
185; Walker and Schafer 2000a: 537; Malici and Malici 2005a: 403), the changes
tended to be of a very low magnitude.5
What is particularly interesting about these findings from operational code
research is that they directly contradict the social-psychological theories of belief
change. For instance, Tetlock argues convincingly that foreign policy belief sys-
tems are organized hierarchically. At the top are fundamental assumptions, below
which are “strategic policy beliefs” and then tactical beliefs. He argues that learning
takes place mostly at the lowest level, tactical beliefs, and that higher-level beliefs
are much less prone to change (Tetlock 1991: 27–31; see also Larson 1994: 26–27;
Rokeach 1968; Vertzberger 1990: 118). If this were correct, we would expect
change in beliefs to be relatively less likely in philosophical beliefs than in instru-
mental beliefs; however, exactly the opposite pattern has been found thus far. This
pattern and its implications for this chapter’s results will be taken up later.
The results noted above from both psychology and operational code analysis
are of minimal help in suggesting possible hypotheses concerning the change in
George W. Bush’s beliefs over the four periods of time under examination in
 The Operational Code of George W. Bush 175

this chapter. In some cases, such as the issue of central versus peripheral beliefs,
they suggest directly opposite predictions as to which beliefs are likely to change,
while in others, such as the effect of specific factors on belief change or whether
beliefs are likely to be reversed or reinforced, they suggest almost nothing at all.
Below I put forward some hypotheses based on the available research and relevant
literature.

Phase 1 to Phase 2 Hypotheses

There will be no major shifts in President Bush’s operational code from Phase
1 to Phase 2. To the extent that President Bush’s operational code does shift
from Phase 1 to Phase 2, the shifts will be in the philosophical beliefs (not the
instrumental beliefs) and the changes will reinforce, rather than reverse, his prior
views of the political universe as a friendly and cooperative place. As noted above,
the vast majority of evidence from psychology suggests the overall stability of
beliefs. Here, the question is whether a change in role, from Governor of Texas to
President of the United States, is likely to significantly alter GWB’s foreign policy
operational code. Feng’s results (2005b: 12) bear mentioning again here, as she
coded four Chinese leaders at different points in their political careers and found
that changes in their political role did correlate with significant alterations in their
operational codes. And though it is not directly comparable to the quantitative
studies, Holsti (1970: 155–156) found some evidence suggesting that J. F. Dulles
beliefs were influenced by his role.
However, all available sources on the initial months of the Bush presidency
agree that the President was not particularly focused on foreign policy issues
prior to September 11th, 2001 (see Daalder and Lindsay 2003a; Herrmann and
Reese 2004; Suskind 2004: 45). Additionally, a change in instrumental beliefs
(i.e., tactics) would imply that GWB’s prior instrumental beliefs had been tested
and found to be lacking in some way. Yet, both the relative lack of attention paid
to issues of foreign policy in his first nine months in office, and the short time span
of Phase 2 (precluding unambiguous feedback on the usefulness of various tactics)
suggest that GWB’s instrumental beliefs are likely to remain stable between Phase
1 and Phase 2.
Feng’s analysis (2005b) suggests that GWB’s philosophical beliefs might become
more conflictual upon taking office as president. However, observers of the Bush
Presidency have noted his surprisingly cooperative approach to international poli-
tics in his first nine months in office. Illustrative of this was Bush’s stance toward
Russia. While key foreign policy advisors Donald Rumsfeld, Paul Wolfowitz, and
Condoleezza Rice were busy “lambasting” Russia (Daalder and Linsday 2003a:
64; Perlez 2001), Bush took a dramatically more cooperative stance. During a
personal meeting with Premier Vladimir Putin, the two leaders discussed religion
and Bush famously “got a sense of his [Putin’s] soul” (Tyler 2001). Thereafter, it
became clear that the Bush administration’s policy toward its historical adversary
176 Jonathan Renshon

would be much more cooperative and trusting than most had expected (Mazarr
2003; Daalder and Lindsay 2003a).
Furthermore, optimism seems to be a key component of GWB’s personality.
Most of those who have spent time in his company have come away with the
impression of a confident, optimistic person (Frum 2003; Bruni 2002).6 Observers
and scholars have agreed with this character assessment of the president, despite
the antipathy some of them have expressed toward him (Mazarr 2003; Moens
2004; Mitchell 2000; Kristof 2000). It is thus possible that attaining the office of
President of the United States, arguably the most powerful position in the world,
might have exacerbated or intensified those qualities. For a man so confident in
his own abilities and optimistic about his future, gaining more power to accomplish
his goals would likely have the effect of reinforcing his prior beliefs about the
cooperative and friendly nature of the political universe.
Role theory supports this hypothesis as well. In novel situations actors are likely
to utilize a familiar role until events either reinforce that role conception or a new
role is learned (Walker 1979: 177). In the case of GWB, his new role as President
of the United States was sufficiently novel that he was unlikely to have at his dis-
posal a role conception appropriate to his new office, and therefore likely to retain
his old role conception (including his old beliefs) after taking office.

Phase 2 to Phase 3 Hypotheses

There will be statistically significant shifts in GWB’s operational code between
Phase 2 and Phase 3. The shifts will be in philosophical (and not instrumental)
beliefs, making his diagnosis of the political universe more conflictual and less
optimistic, reversing his prior beliefs. There has been considerable debate over the
extent to which the events of September 11th “changed” President Bush. Initially,
both GWB and his wife Laura Bush denied that the events had changed him in
any significant way (Hume 2002; see also Hughes 2004: 252).7 Some scholars have
agreed, arguing that the attacks merely confirmed GWB’s prior strategic beliefs
(Daalder and Lindsay 2003b: 101). Others have argued that while his public per-
sona was certainly transformed, the terrorist attacks gave him a sense of purpose
rather than truly changing any fundamental aspects of his beliefs about the world
(Draper 2007: 166). Still others argued that the terrorist attacks had given the presi-
dent not only a new sense of purpose but also dramatically shifted his views about
the fundamental nature of international politics (Heclo 2003; Renshon 2004: 152;
Jervis 2005: 39). Support for the notion that Bush changed significantly after 9/11
is also found by Robison (2006: 119) in his operational code analysis of GWB
and his advisors. Robison found several statistically significant changes in GWB’s
operational code, indicating that he became more conflictual, saw the world as
more hostile, and was less optimistic in achieving his political goals.
On the psychological side, the predictions are equally mixed. On one hand,
most theories of belief stability admit that beliefs do change sometimes, and
 The Operational Code of George W. Bush 177

suggest that large “spectacular events” might be a mechanism for that change
(Deutsch and Merritt 1965: 137; Jervis 1976: 218, 262). For instance, Peffley
and Hurwitz (1992) examined the effects of the 1987 and 1988 Arms Control
summits between U.S. and Soviet leaders. They found that public opinion polls
revealed that U.S. beliefs concerning the nature of the Soviet Union, which had
been relatively stable prior to these events, changed significantly, becoming more
trusting and “dovish” (Peffley and Hurwitz 1992).
Conversely, Tetlock (2005: 131–132) in a longitudinal study found that the
classification of an event as an “exogenous shock” (i.e. something dramatic and
unexpected) served as a defense mechanism for belief systems, and acted to consoli-
date prior beliefs rather than changing them in any way. Individuals in his study
claimed that the nature of the surprise event placed it outside of their “model”
(framework of analysis, beliefs, etc.) and thus they were not responsible for predict-
ing it and should not incorporate it into their model by changing their beliefs.
The debate mentioned earlier between those who see individuals as consis-
tency seekers and those who posit a more fragmented belief system also has rel-
evance here. If belief systems are generally internally coherent, then anything that
produced a change in a central or core belief would be likely to cause substantial
changes throughout the belief system. In other words, this theory would predict
generally stability and small shifts in beliefs, but once change comes, “it will come
in large batches; Several elements will change almost simultaneously” (Jervis 1976:
170). Here again there is conflicting evidence. While Ben-Zvi (1978) found dra-
matic shifts in the images of Japan held by American leaders after World War
II, only one operational code analysis has come close to finding such whole-
sale changes in a leader’s operational code. Even in that case, Malici found that
Ronald Reagan’s operational code in his last year in office evinced statistically
significant changes in only in two of the five philosophical indices and seven of
the eleven instrumental indices (Malici 2006: 140).
Clearly, mixed and contradictory predictions abound if one takes into account
all the relevant literature. However, the balance of the evidence (including Mr.
Bush’s own views on the matter) suggests that the events of September 11th
caused some fundamental change in GWB’s beliefs about the world. As for the
nature of the change in his operational code, it seems highly likely that to the
extent that change did occur, it would be a reversal of his prior beliefs, and GWB
would see the world as a more dangerous, conflictual place and have less opti-
mism for the prospects of accomplishing his goals. I do not anticipate wholesale
change in the entire code, however, and believe that the change will be confined
to the philosophical indices.

Phase 3 to Phase 4 Hypotheses

There will be no statistically significant shifts in President Bush’s operational code
from Phase 3 to Phase 4. At issue here is whether President Bush’s operational
178 Jonathan Renshon

code changed between March of 2001 (the end of Phase 3) and the end of his
second term (January–July 2007). During this six-year period, there were no
large exogenous shocks on the scale of the 9/11 attacks, and President Bush’s
role remained the same. This allows us to examine belief stability over a relatively
long duration.
On another level, we can further investigate the “learning in office” hypoth-
esis suggested by Schafer and Crichlow (2000: 562). That is, to what extent did
GWB’s foreign policy beliefs shift as he gained experience in office? No other job
is adequate preparation for the presidency, especially in an age of increased global
interconnectedness and American dominance. Thus, it is plausible to expect that
as presidents (and other high-level leaders) gain greater exposure to international
politics, and have to exercise greater command responsibility, their views on how
the international political system works, and the best tactical means to accomplish
their goals might change.
However, the “cognitive consistency” argument that has dominated the study
of beliefs for several decades suggests a different prediction. This model would
predict that new information would be assimilated in such a way as to maintain
the stability of President Bush’s belief system. If this were true, we would expect
Bush to minimize the importance of dissonant information, and interpret ambigu-
ous information in a manner consistent with his prior beliefs. In fact, this behavior
is precisely the pattern exhibited by President Bush with respect to Iraq. While
eventually admitting some misjudgments (particularly with regard to the lack of
adequate planning for Iraq’s reconstruction), he has been adamant in defending
the major decisions in this foreign policy area.
In addition, President Bush seems to have employed several cognitive and rhe-
torical strategies for maintaining his belief system in the face of dissonant informa-
tion. For instance, he has lengthened the relevant time frame (thus minimizing
the importance of immediate setbacks) as well as focused on historical analogies
that confirm the wisdom of his decisions (e.g. the reconstruction of Japan after
World War II) (Sanger 2005a, 2005b). And while he has seemed open to chang-
ing strategy within the context of Iraq, his overall message has remained constant: the
decision to go to war in Iraq was the correct decision (despite the difficulty of the
occupation and the failure to find weapons of mass destruction) because his essen-
tial view of the international system—in particular, the most dangerous threats to
the United States—has not changed (New York Times 2004; Bush 2004).

Methodology and Results

For this analysis the program Profiler Plus v. 5.7.0 was used to examine changes
in Bush’s operational code with the Verbs in Context System (VICS) of content
analysis. Both were provided by Social Science Automation, Inc. Content was
gathered from public speeches made by President Bush. In cases where GWB
answered questions from reporters or the audience after a speech, only the
 The Operational Code of George W. Bush 179

prepared remarks were used, following the precedent of Dille (2000) and Schafer
and Crichlow (2000), who found that spontaneous and prepared statements
generated slightly different results.
A common criticism of operational code analysis is that public speeches may
not reflect the “true” personal beliefs of leaders. One can respond to this on
three levels: practical, theoretical, and empirical. On the practical level, for
most leaders, the public record is far more extensive than the private. We often
have to wait years, or decades, for private material to become available (the
Foreign Relations of the United States series, for instance, waits roughly thirty years
before declassifying and publishing its documents). Thus, if one wishes for any
substantive reason to focus on a leader who still occupies or has recently left
political office, there are few alternatives but to rely on the available public
record. The second is the theoretical assumption of operational code analysis that
“a leader’s public behavior is constrained by his public image and that, over time,
his public actions will consistently match his public beliefs” (Walker et al. 2003:
223). That is, while beliefs and actions might diverge for a short time, the norm
is for beliefs and actions to converge.
The third rationale is recent evidence that suggests that the public statements of
leaders may reflect their true beliefs to a greater extent than skeptical observers would
predict. Renshon (2009) examined the operational code of President Kennedy
during the summer of 1962 based on (a) public statements and (b) transcripts
made from the secret taping system the president had installed in the Oval Office.
His results indicate that the two operational codes—public and private—were
virtually identical (see also Marfleet 2000). There is also evidence suggesting that
public speech can betray some types of information that leaders might consciously
try to suppress. For instance, there is evidence that the integrative complexity
of leaders (which tends to decline in the lead-up to wars) also declines prior to
surprise attacks, unknowingly giving clues to the impending attack to those who
know where (and how) to look (Suedfeld and Bluck 1988).
For Phases 2, 3, and 4, all speeches were obtained from the White House Press
Office (www.whitehouse.gov/news). All speeches made during the designated
time period that met the following two criteria were used. Each speech must
meet the minimum requirement of 1,500 words,8 and all speeches must focus
primarily on foreign policy issues, following Walker et al. (2003) who found that
operational codes may be domain-specific. There was no single, authoritative
source for the speeches President Bush made during the time period examined in
Phase 1, so speeches that met the same two criteria listed above for this phase were
compiled from a number of different sources (a list of sources for these speeches
is available from the author upon request).
Table 10.1 contains the results of the operational code analysis. It shows the sig-
nificant differences in philosophical and instrumental beliefs for George W. Bush
across four phases: (1) pre-election, (2) pre-9/11, (3) post-9/11, and (4) to the
end of his second term in office.9 The differences mark the effects of role change
180 Jonathan Renshon

TABLE 10.1 The Operational Code of George W. Bush

Phase 1 Phase 2 Phase 3 Phase 4

(n = 4) (n = 6) (n = 15) (n = 13)

P-1 Nature of the Political Universe +.37 +.51 + .21 +0.29

(–2.18*) (3.82***) (–1.22)
P-2 Realization of Political Values +.20 + .29 + .06 +.13
(–1.84*) (3.54***) (–1.15)
P-3 Predictability of Political Future .12 + .14 + .11 +.11
(–1.60) (1.84*) (–0.21)
P-4 Control over Historical Development .33 .33 .32 .25
(–.14) (.49) (2.51**)
P-5 Role of Chance .96 .95 .96 .97
(1.22) (–1.28) (–1.11)
I-1 Strategic Approach to Goals +.51 + .53 + .40 +.49
(–.09) (1.29) (–1.1)
I-2 Tactical Pursuit of Goals +.29 + .25 + .13 +.19
(.40) (1.70) (–1.14)
I-3 Risk Orientation .14 .19 .21 .26
(–.70) (–.32) (–1.18)
I-4 Timing of Action
a. Cooperation/Conflict .49 .49 .59 .51
(.09) (–1.25) (1.06)
b. Words/Deeds .38 .58 .65 .53
(–1.57) (–.59) (1.32)
I-5 Utility of Means
a. Reward .14 .18 .17 .15
(–1.49) (.49) (0.58)
b. Promise .26 .13 .05 .04
(1.57) (1.67) (0.64)
c. Appeal/Support .36 .46 .48 .56
(–.92) (–.37) (–1.49)
d. Oppose/Resist .11 .08 .08 .11
(.55) (.05) (–1.26)
e. Threaten .09 .05 .07 .02
(1.22) (–.53) (2.56**)
f. Punish .05 .11 .16 .12
(–.97) (–.98) (1.19)

Values are significant at the following levels: *** p ⱕ .01, ** p ⱕ .05, * p ⱕ .10.
Values in parentheses are t-statistics for change from previous phase.

(Phase 1/Phase 2), exogenous shock (Phase 2/Phase 3), and learning in office
(Phase 3/Phase 4). Each of these effects is summarized in Table 10.1.
As Table 10.1 indicates, there were statistically significant changes in two opera-
tional code indices: P-1 (Nature of the Political Universe) and P-2 (Realization of
Political Values) changed as Bush’s role changed from candidate to president. The
 The Operational Code of George W. Bush 181

results indicate that President Bush’s view of the political universe became more
optimistic in Phase 2, when he took office. P-1 changed from +.37 to +.51, which
in the quartile verbal descriptor categories would be a shift in his view of the politi-
cal universe from “somewhat friendly” to “definitely friendly.” P-2 changed from
+.20 to +.29, both of which indicate that Bush was “somewhat optimistic” con-
cerning the realization of his political values. Aside from these two indices, no other
aspect of GWB’s operational code shifted significantly from Phase 1 to Phase 2.
Because of the relatively small magnitude of the changes in P-1 and P-2, we
cannot classify this as a “major” shift in beliefs, but the fact that the changes in
both P-1 and P-2 were significant and in the same direction indicates that a shift
did occur and that the effect was to reinforce GWB’s prior beliefs about the
world. To the extent that GWB’s operational code did shift in Phase 2, the only
statistically significant shift was in the philosophical, not instrumental, indices, and
the changes were in the optimistic/cooperative direction.
The next substantive question concerns the impact of traumatic events on GWB’s
belief system. Table 10.1 indicates significant shifts in President Bush’s operational
code from Phase 2 to Phase 3. The first three philosophical indices—Nature
of the Political Universe, Realization of Political Values, and Predictability of
Political Future—all changed significantly. In addition, the changes in the first
two philosophical indices were of a substantial magnitude. P-1 decreased from
+.51 (“definitely friendly”) in Phase 2 to +.21 (“somewhat friendly”) in Phase 3,
decreasing even past his initial score when he took office (when it was +.37). He
was significantly more pessimistic in his beliefs concerning the prospects for the
realization of his political goals, the VICS index decreasing from +.29 to +.06
in Phase 3. Bush’s belief in the predictability of the political future also declined
significantly (from +.14 to +.11), though the shift was of a smaller magnitude.
The evidence suggests that traumatic shocks do have the capacity to effect
fundamental change in individuals’ belief systems. Furthermore, the change was a
reversal; whereas before GWB viewed the world as friendly and cooperative, after
9/11 those views became significantly more conflictual and hostile. This find-
ing provides further support for those who argued that President Bush’s beliefs
were significantly altered by the terrorist attacks of September 11th (e.g. Heclo
2003; Renshon 2004; Jervis 2005; Robison 2006). Conversely, it is difficult to
interpret this finding as supporting the hypothesis that 9/11 “confirmed” Bush’s
prior beliefs (Daalder and Lindsay 2003b) or that 9/11 simply gave him a “sense
of purpose” (Draper 2007).
However, while the first three philosophical indices are clearly quite impor-
tant—they are at the heart of how the individual views the political universe—it
is notable that they are the only indices that experienced a statistically significant
change. As a result, these changes, significant though they may be, do not wholly
disconfirm models that predict consistency in beliefs. Bush’s overall operational
code can still be described as relatively stable, with only three of the ten indices
experiencing a statistically significant change.
182 Jonathan Renshon

The change from Phase 3 to Phase 4 in Table 10.1 is notable for the lack of any
major shifts in President Bush’s operational code, which would have indicated
learning in office. While P-4 (Control over Historical Development) did decrease
slightly (from .32 in Phase 3 to .25 in Phase 4 ; significant at p ⱕ .05), the likely
explanation for this is that several years as president diminished GWB’s belief in
his personal ability to influence historical events and outcomes. Additionally, I-5
(Threaten) decreased from .07 to .02 in Phase 4 and was also significant at p ⱕ .05.
Both of these changes, though significant in the statistical sense, were not signifi-
cant in any practical or substantive sense. The change in P-4 was not of a large
enough magnitude to shift the verbal descriptor category (it remained “low” in
both Phase 3 and 4), while in I-5 the rank order of tactical preferences remained
identical to Phase 3 save for the switched positions of “promise” and “threaten.”
The very small magnitude of these changes indicates that though there may
have been some minor shifts in President Bush’s beliefs, we cannot conclusively
state that any major shift in his belief system took place. Thus, there is no learn-
ing in office by Mr. Bush. It is important here to note that “learning” is used
strictly in the descriptive (and not normative) sense of the term. It simply means
that President Bush’s foreign policy operational code did not change significantly
(after September 11th, 2001) while in office. Here we encounter a potential dis-
connect between “learn” as used in political science literature and the term as
it is generally understood. President Bush may have “learned,” in the general
sense, many things while in office. For instance, he is likely to have acquired new
information (he undoubtedly had more knowledge of international affairs after six
years in office, for instance). However, because there is no shift in his operational
code, we can conclude that he did not “learn” in the sense that the new informa-
tion effected a change in his operational code beliefs.

Discussion
We can now turn to the broader picture of how the beliefs of President Bush
changed over time, and how these results fit (or do not) with social-psychological
and operational code theories of belief change. Table 10.2 indicates that President
Bush’s beliefs changed significantly in two out of the possible three time periods
in which change could have occurred. These two shifts occurred when he took
office in early 2001, and in the aftermath of the terrorist attacks of September
11th, 2001. In the first case, changing role reinforced and strengthened GWB’s
existing beliefs, while in the latter case the events of September 11th reversed his
beliefs, pushing them in the opposite (negative, in this case) direction.
While these two major shifts seem contrary to the predictions made by social-
psychological theories of belief stability, it is difficult to assess with any degree of
certainty how contrary these findings truly are. Much of this has to do with the
frustratingly vague predictions of the theories, which seem to predict that beliefs
generally remain stable, except when they don’t. For instance, one could make
 The Operational Code of George W. Bush 183

the case that the results presented in this chapter are evidence for belief stability.
Evidence for this interpretation is that despite the many important changes in his
life (being elected president, taking office, the surprise terrorist attacks, launching
a preventive war in Iraq, etc.), only seven changes in GWB’s operational code
occurred. Discounting the relatively insignificant changes in Phase 4, this becomes
even more stark: only five changes occurred. By and large, GWB’s operational
code remained relatively stable over the tumultuous period under observation.
However, this interpretation—while technically correct—is also somewhat
misleading. This is because the five belief changes that did occur were of some
substantive significance. The shifts in GWB’s operational code in Phases 2 and
3 were changes in his most basic conception of the political universe: the nature
of the political world (whether hostile or friendly), the prospects for the realiza-
tion of his political goals (optimistic or pessimistic), and the extent to which the
future is predictable. Furthermore, the shifts were significant in the practical sense
as well, particularly in Phase 3. In Phase 2, upon assuming the presidency, the
change in P-1 was of sufficient magnitude to shift its verbal descriptor category
from “somewhat” to “definitely” friendly. In Phase 3, after the 9/11 attacks, the
changes in P-1 and in P-2 were both of sufficient magnitude to shift its verbal
descriptor.
Related to the issue of temporal stability is the question of how individu-
als respond to traumatic shocks over the long term. That is, do beliefs that are
reversed by traumatic events eventually “rebound” to pre-event levels, or are the
changes permanent? In the initial data analysis, Phase 3 was from September 11th,
2001, to March 11th, 2002, a six-month period. In order to examine the longer-
term effects of 9/11 on GWB’s belief system, I have extended Phase 3 to cover
the 18-month period from September 11th, 2001, to March 11th, 2003, using the
methods for data collection and analysis described earlier in Part III. The results of
this analysis are described in Table 10.2.
The results of this re-analysis suggest that traumatic events can permanently
alter belief systems, rather than simply causing a fleeting shift. Both P-1 and P-2
“rebound” slightly in the Extended Phase 3 from their more depressed levels in
the shorter Phase 3. However, the changes in these indices from Phase 2 are still
of a considerable magnitude, and statistically significant at p < .01. The value
of P-3 (predictability of the political future) also rebounds slightly from Phase 3
levels, though this change is no longer statistically significant (unsurprising given
the relatively small magnitude of the change that occurred in this index in Phase
3). All major indices seem to support the notion that traumatic events may cause
a severe reversal of certain key beliefs in the shorter term, and that those initially
severe changes may become slightly attenuated as the new belief system is con-
solidated over a longer period of time.
Beyond the empirical question of how the beliefs of President Bush shifted (or
remained constant) over time, the data from this operational code analysis allow us
some insight into the theoretical questions raised in earlier sections on the second
184 Jonathan Renshon

TABLE 10.2 Extended Analysis of Phase 3

Phase 2 Phase 3 Phase 3 Extended

P-1 Nature of the

Political Universe +.51 (–2.18*) + .21 (3.82***) +.27 (3.67***)
P-2 Realization of
Political Values + .29 (–1.84*) + .06 (3.54***) +.11 (3.47***)
P-3 Predictability of
Political Future + .14 (–1.60) + .11 (1.84*) +.12 (1.21)
P-4 Control over
Historical
Development .33 (–.14) .32 (.49) .31 (.56)
P-5 Role of Chance .95 (1.22) .96 (–1.28) .96 (–.85)
I-1 Strategic Approach
to Goals + .53 (–.09) + .40 (1.29) +.47 (.69)
I-2 Tactical Pursuit of
Goals + .25 (.40) + .13 (1.70) +.17 (1.32)
I-3 Risk Orientation .19 (–.70) .21 (–.32) .26 (–1.04)
I-4 Timing of Action
a. Cooperation/
Conflict .49 (.09) .59 (–1.25) .53 (–0.64)
b. Words/Deeds .58 (–1.57) .65 (–.59) .56 (.13)
I-5 Utility of Means
a. Reward .18 (–1.49) .17 (.49) .15 (1.48)
b. Promise .13 (1.57) .05 (1.67) .06 (1.34)
c. Appeal/Support .46 (–.92) .48 (–.37) .52 (–.94)
d. Oppose/Resist .08 (.55) .08 (.05) .09 (–.29)
e. Threaten .05 (1.22) .07 (–.53) .05 (.25)
f. Punish .11 (–.97) .16 (–.98) .13 (–.50)
Dates of Analysis 1/20/01–9/10/01 9/11/01–3/11/02 9/12/01–3/11/03

Values are significant at the following levels: *** p ⱕ .01, ** p ⱕ .05, * p ⱕ.10.
Values in parentheses are t-statistics for change from Phase 2.

aspect of belief consistency: internal coherence. That is, to what extent are differ-
ent component parts of a belief system or operational code logically coherent with
one another? Recall that earlier social-psychological accounts of beliefs assumed
individuals to be “consistency seekers” who by and large held beliefs that were
consistent with one another (McGuire 1967; Bem 1970: 13). Early operational
code analyses also operated under this assumption, assuming, for instance, the
direct link between philosophical and instrumental beliefs (George 1969; Holsti
1970: 154; 1977). An additional part of this model was the assumption that core
beliefs ordered and structured peripheral beliefs, and that these core beliefs were
more resistant to change (Tetlock 1991; Rokeach 1968; George 1969: 217). How
does George W. Bush fit into this paradigm?
 The Operational Code of George W. Bush 185

The data from this project clearly suggest that the different beliefs that indi-
viduals hold are not necessarily internally consistent. If a primary mechanism for
belief change is the drive to maintain internal coherence and reduce dissonance,
then we would have expected to see the instrumental and philosophical indices
change in tandem when change occurred. Even the argument that beliefs should
remain internally coherent only within either the philosophical or the instrumen-
tal indices cannot be supported by the results. We would expect, for instance, that
as GWB’s view of the nature of the political universe (P-1) became significantly
more negative in Phase 3, then his preferred tactics would change accordingly
(with, perhaps, a tactical orientation weighted further toward conflict than coop-
eration). Yet, it does not.
Table 10.1 indicates that the events of September 11th produced a fundamental
change in GWB’s views of the world, but he did not adjust his tactics in any way
in response to this diagnostic change. These data seem to contradict the notion
that individuals’ beliefs are internally consistent. If this had been the case, then
we should have seen significant changes in the instrumental indices in Phases 2
and 3. Additionally, given that the perceived utility of threats decreased in Phase 4,
we would have expected to see a somewhat different (perhaps more optimistic or
friendly) view of the political universe as well. In fact, no operational code analysis
to date—this one included—has found the wholesale changes in beliefs that are
predicted by cognitive consistency theory.
The next question concerns the relationship between the philosophical and the
instrumental indices in the operational code. Of the seven statistically significant
changes in the periods covered in this chapter, six were shifts in philosophical
beliefs—two shifted in Phase 2, two shifted in Phase 3, and one changed (mini-
mally) in Phase 4. The only statistically significant change in an instrumental belief
was in Phase 4. The obvious question that arises from this pattern—found in
other operational code analyses as well—is why are philosophical beliefs volatile
while instrumental beliefs are relatively stable over time? Before answering this
question, we must be explicit about what is being measured in the VICS system.
In fact, VICS was explicitly constructed so that the philosophical beliefs would
represent attributions about the “other” and instrumental beliefs would represent
attributions of the “self” (Walker et al. 1998; Schafer and Walker 2006: l31). That
is, scores for the philosophical indices represent beliefs about the nature of the
political system insofar as it is shaped by the actions and beliefs of other actors.
Instrumental indices in VICS represent a leader’s beliefs about how he should or
does act in the international system.
In other words, the philosophical beliefs in Table 10.1 represent GWB’s beliefs
about the behavior and actions of other actors in the international system, while
the instrumental indices reflect GWB’s beliefs about how he acts. With this
understanding, we can approach the puzzle of the volatile philosophical beliefs
and stable instrumental beliefs from a new angle. We can now conceive of the
instrumental aspects of the operational code as comprising part of an individual’s
186 Jonathan Renshon

identity, with the philosophical beliefs the results of “reality-testing” and “les-
sons-learned” about the nature of the political universe (this possibility is sug-
gested by Walker and Schafer 2000b: 4). In this model, it is GWB’s identity, and
beliefs about himself, that remained relatively stable, while his conception of the
“other” shifted dramatically.

Conclusion
This analysis has examined GWB’s operational code within the larger framework
of social-psychological theories of belief consistency. The findings presented here
cast further doubt on many aspects of the cognitive consistency model. In fact,
the results suggest that individuals’ belief systems are not as interdependent as has
often been assumed: the significant changes in what seem to be “key” or “core”
beliefs do not seem to impact other, derivative, beliefs at all. As for overall tem-
poral consistency, the results make clear that belief systems can, and do, experi-
ence major changes. Yet, the results also suggest the durability of many beliefs
over time, perhaps providing some confirmation that beliefs are, overall, relatively
stable.
However, the primary purpose in this chapter has not been to add to the cog-
nitive consistency debate, but rather to take a first step in considering explicitly
the causes of belief change. Three potential causes were explored: role change,
traumatic events, and learning in office. Of these, support was found for the first
two. There seems to be substantial support for the importance of traumatic events
in changing beliefs. GWB’s view of the political universe and other related philo-
sophical beliefs all reversed direction, becoming more negative and conflictual
after the terrorist attacks in 2001. The change in role also had an effect on GWB’s
beliefs, though it seemed to reinforce his prior beliefs, and strengthen them in
certain cases, rather than changing them.
Clearly, there is much scope for future research on this important subject. How
beliefs change has implications not just for the study of political leaders and deci-
sion making, but will also affect how we attempt to persuade others in the inter-
national system. For instance, the United States has evinced a desire for China
to become more democratically accountable, and to accept the responsibilities of
being a world power. Coercion might change their behavior, but it is obviously
preferable to effect a change in Chinese leaders’ beliefs about how they should act.
Yet, we still understand very little about whether such change is even possible, let
alone what policy offers the best hope of accomplishing these goals.
In order for such research to be useful, we must start to explicitly consider the
possible causes of belief change. Three candidates were explored in this chapter,
but more remain. How effective is persuasion, for instance, in changing beliefs?
How effective are threats? Furthermore, we must begin to consider the interac-
tion between personality and belief stability. It seems likely, for example, that
some individuals are more “closed” to new information and resistant to belief
 The Operational Code of George W. Bush 187

change than others. Some causes of belief change may be more effective on cer-
tain personality types, and less effective on others. Any of these, and many others,
would be fruitful avenues for further research in this area.

Notes
* Author’s Note: I would like to thank Stephen G. Walker for his invaluable assistance
and extensive comments on earlier drafts of this chapter. I would also like to thank
Iain Johnston for his assistance and guidance during the early stages of this project
and Roger Porter for his helpful comments on an earlier draft. Materials necessary
to replicate this analysis are available at the author’s website. This chapter is a slightly
modified and shortened version of “Stability and Change in Belief Systems,” Journal of
Conflict Resolution (2008) 52: 820–849. The author would like to thank the Journal and
Sage Publications for permission to reprint the article.
1 Related to the concept of “belief change” is learning. Some have used that term to
refer to changes in the overall posture of states over time (e.g. the Soviet Union toward
the end of the cold war, see Breslauer (1987) while others have used it to refer to the
particular lessons that decision makers learn from history (Khong 1992). In a review
article, Levy defined learning as “a change of beliefs (or the degree of confidence in
one’s beliefs) or the development of new beliefs, skills, or procedures as a result of the
observation and interpretation of experience” (Levy 1994: 283). This is a useful and
concise definition, though the emphasis of this article is on a concept that Levy does
not address: what causes learning, or changes in beliefs.
2 For more on strategies individuals use to maintain belief stability, see Nisbett and Ross
(1980); see Holsti (1967) and Carretta and Moreland (1982) for studies in the political
context illustrative of this concept.
3 While the explicit distinction between belief reversal and belief reinforcement is
introduced here, it is presaged by earlier studies that found patterns consistent with
this theoretical distinction. For instance, Walker et al. (2001) contains a discussion of
how the cooperative beliefs of Chamberlain and Halifax differentially adjusted over the
period 1937–1939, sometimes being weakened and other times being strengthened by
their understanding of current events.
4 Though not directly comparable to the modern quantitative studies, it is perhaps
suggestive that Johnson discovered the same pattern of belief modification in his
qualitative examination of Senator Frank Church’s operational code (Johnson 1977:
112).
5 That is, the changes either (a) did not change the “rank order” of the utility of means
or (b) did not move the VICS score into a different verbal category (e.g., “cooperative”
to “very cooperative”).
6 During the 2000 campaign, one Republican advisor to GWB even voiced concern
about the President’s “excessive optimism” (quoted in Bruni 2000).
7 Though GWB did later change his stance, commenting: “Prior to September 11th, we
were discussing smart sanctions.… After September 11th, the doctrine of containment
just doesn’t hold any water … my vision shifted dramatically after September 11th, because
now I realize the stakes, I realize the world has changed” [emphasis mine] (NYT 2003).
8 The first standard for operational code analysis was 1,500 words as a minimum length
for each speech act used. This was used so that the mean score for the combined
speeches would not weight one speech more heavily than another. More recent efforts
have pre-aggregated the speeches into one “big speech,” and as such are not so con-
cerned about individual speech length (as long as each speech act contains between
10 and 15 verbs that can be coded in VICS) (Schafer and Walker 2006c: 43–44). The
methodological issue of how one should compile content data for operational code
188 Jonathan Renshon

analysis reflects an important substantive issue. Operational codes can be conceptual-

ized as either a stable “personality trait” or a more transient “state of mind.” If it is
conceptualized as a stable personality trait, it is defendable to aggregate many smaller
speech acts for purposes of analysis, since there should not be significant shifts in the
operational code over time. Since the purpose of this chapter is to investigate whether
there is change in GWB’s operational code, the stability of these beliefs cannot be taken
for granted. For this reason, and because it allows the use of statistical analysis, mean
scores were utilized instead of aggregate scores.
9 It is worth noting here that the changes in Bush’s operational code described in
this section do not correspond to changes in his speechwriting staff, ruling out one
plausible alternative explanation. GWB’s chief speechwriter, Michael Gerson, worked
for President Bush from his first presidential campaign until June of 2006. Another
significant influence on GWB’s public communication, Karen Hughes, worked for
the President from his campaign through April of 2002 (Rutenberg 2006; Burke 2004;
Gerson 2007). The departures of these two administration officials do not correspond
to the phases examined in this analysis, and are thus unlikely to be confounding factors
in the results. In fact, these personnel shifts correspond inversely to changes in GWB’s
operational code. The only phase in which there were not significant changes (Phase
4) was also the only phase in which President Bush’s public communications were not
influenced by either Michael Gerson or Karen Hughes.
11
EXPERIENTIAL LEARNING BY U.S.
PRESIDENTS
Domestic and International Influences
in the Post-Cold War World

Samuel B. Robison

Introduction
Learning by political elites—defined here as a “change in beliefs” following from
a leader’s experiences or from a leader’s observation of events (Levy 1994)—is of
particular importance in analyses of foreign policy decision-making processes and
outcomes. If we are to understand how leaders interpret events, why certain types
of actors are viewed as more threatening when others are seen as less so, which
responses are generally preferred for given situations and why, and ultimately
why these leaders act as they do, then we must look beyond an exclusive focus on
domestic and international “constraints,” and examine the beliefs that leaders hold
and how these beliefs come to be. There already exists a fair amount of scholarship
examining individual-level “learning” in the foreign policy sphere. For instance,
Jervis (1976) evaluates the impact of cognitive dissonance, the “last” victory and
defeat, and the use (and misuse) of “generational effects,” amongst several other
factors, on belief change. Leng (1983, 2000) expands on Jervis’s examination of
the impact of conflict outcomes on “learning,” showing that losses provide pow-
erful incentives for cognitive and behavioral change. Khong (1991, 1992) and
Reiter (1994) discuss many of the ways in which “historical analogies” impact
leaders’ interpretations of important contemporary events.
However, these examples and most other works on learning in the foreign
policy domain use only a small number of cases to illustrate the various situations
and predispositions that contribute to learning or a failure to learn (Larson 1985,
1991; Breslauer and Tetlock 1991; Levy 1994; Stern 1997). These works have
all made significant contributions toward improving our understanding of the
nature of decision making and learning, but if we want to really understand these
systematic processes and phenomena, then we must expand our empirical scope
190 Samuel B. Robison

in examining their causes. In this chapter I examine the learning behavior of three
U.S. presidents over 168 monthly periods, examining belief changes from month
to month. The analysis itself is an evaluation of the relative impact of international
and domestic factors on the magnitude of presidential “absolute” belief change.
To begin, I examine factors that impact belief change on the three “master”
operational code beliefs—(P-1) the image of the external environment, (I-1) stra-
tegic orientation, and (P-4) belief in control over historical development (Walker
and Schafer 2006: 12). These specific beliefs were chosen because they are
hypothesized to be the most central components of one’s belief system, and thus
are the most important for us to understand. An example of “learning” behavior
here would be a change in “belief” values from one point in time to another on
the image index, which would illustrate the degree to which a leader “learned” to
see the world as more hostile or more friendly over time, presumably based upon
the influence of events experienced or observed by the leader.1
In addition to the simple form of learning noted above, or a change in the
“raw” level of a belief from one point in time to another—what I call “direc-
tional” belief change—it might also be useful to examine learning as a greater or
lesser absolute shift in one’s belief values, irrespective of the directionality of this
belief change—or what I will refer to as “absolute” belief change. This is the kind
of learning examined here. For instance, if a leader is concerned, narrowly, with
“adjusting” beliefs in order to adapt to the environment, and is not concerned so
much with the “directional” nature of this shift (so long as it is different from that
of previously held beliefs), this is the form of belief change that I am identifying.
An example of analyzing this “absolute” change is testing whether low levels
of public opinion convince leaders to shift their beliefs about the image of the
external political environment away from earlier beliefs. Leaders may see the
world as either more or less hostile than they did beforehand in an attempt to satisfy
the populace without necessarily anticipating a specific, directional shift one way
or the other, e.g., toward only seeing the world as more hostile, or toward only
seeing the world as more friendly. Related to this example, in this analysis I am
treating operational code beliefs as evaluating two distinct, but inter-related phe-
nomena. In their most basic, traditionally conceptualized form, these beliefs solely
reflect perceptions regarding the way that leaders interpret and choose to respond or
deal with the international environment. However, one’s “beliefs” as determined
via speech behavior may also reflect an “active” attempt by leaders to influence
others’ behavior, and thus may be more than simple, “passive” perceptions.
When engaging in foreign policy making, actors rely on others’ actions (includ-
ing verbal behavior) in order to determine the approach of others to politics, pref-
erences, and general ways of thinking. The “self” realizes this fact—that it exists
in an interactive political environment, where the way in which one presents
the “self” to “others” will, in turn, affect how these others behave toward the
“self.” This realization may then influence the expressed content of one’s beliefs,
potentially toward altering these beliefs as a means to influence others in order to
 Experiential Learning by U.S. Presidents 191

better achieve one’s desired policy goals. In other words, belief change may not
solely represent the “passive” process of updating one’s generalized understanding
of and preferences for dealing with the outside world—it may also represent an
attempt to “actively” influence the outside world via the means of one’s repre-
sentation of the “self” as it is observed and interpreted by others. Here I treat the
operational code as both a result and an influence in relationship to the actions of
others in the international and/or domestic political domains. Either or both of
these factors can dictate belief change.2
The hypotheses below will reflect both “passive” and “active” changes in
response to a president’s political environment. I examine factors potentially
influencing beliefs from both the international and domestic political spheres.
Additionally, I examine economic factors that lie at the intersection of these two
contexts. Below I present hypotheses about the influence of these three kinds of
factors on changes in the beliefs of U.S. presidents.

International Factors
Structural realists generally tend to downplay or ignore altogether the importance
of individuals and foreign policy in their evaluation of the factors that impact
international political outcomes. This generalization does not apply to all realists,
however, as some do recognize that such factors are important to take into account
if we are to have a reasonably nuanced understanding of why states behave as they
do.3 Perhaps most notable amongst those engaged in contemporary research of
this variety are the neoclassical realists, e.g., Randall Schweller, Fareed Zakaria,
and William Wohlforth. Harkening back to early realist thinkers such as Hans
Morgenthau, these individuals believe (as most realists do) that the international
political structure and one’s relative position in the hierarchy of states largely dic-
tates state action. However, neoclassical realists question the notion held by struc-
tural realists that individuals simply behave as a “smooth transmission belt” (Rose
1998: 158) who convert international contextual information into political action
as mindless automata might do. Instead, they argue that individual perceptions,
including subjective (and potentially irrational) perceptions, biases, and motivat-
ing factors will play a role in mediating the degree to which the input of external
events and factors contributes to the “output” of foreign policy action.
In addition, neoclassical realist thought suggests that domestic factors may play
a role in some situations, but that these influences almost always play a secondary
role to the power and actions of their international counterparts. The reason for
this is that international political interactions are critically important; a miscalcula-
tion or inappropriate action here could result in catastrophic outcomes. Although
domestic factors have always played a role in the foreign policy decision-making
calculus, they cannot be allowed to play a primary role given the stakes involved.
Thus, for those realists who find some utility in the analysis of foreign policy and
political psychology, they will likely expect that international factors would mat-
192 Samuel B. Robison

ter more than would domestic factors toward impacting political leaders’ beliefs
regarding foreign affairs.
One concrete realist hypothesis is that factors originating in the international
political environment will have a stronger expected influence on presidential
beliefs than will those originating in the domestic political environment. I test
this hypothesis by examining both the magnitude of the influence of each factor
in the international versus domestic domains on absolute belief change, and also
by the number of factors in each domain that yield a significant influence on this
change. International factors under examination include the level of international
conflict broadly defined (as measured by the number of militarized interstate dis-
putes occurring in a given year), the relative power of the United States as defined
by gross domestic product (ratio of U.S. GDP to worldwide GDP), and the
level of international conflict explicitly targeting the United States. Specifically, I
expect that higher levels of conflict both in the international system and toward
the United States lead to greater variation in each belief index of interest than is
the case during periods of less conflict, as the president adapts his beliefs in order
to better interpret and influence an increasingly dangerous and uncertain mili-
tary/political context.
Thus, during times of relative peace U.S. presidents will feel little reason to
alter their beliefs regarding international affairs (irrespective of what these beliefs
are) as these beliefs will (1) be perceived to appropriately correspond to the inter-
national environment as it exists, and/or (2) be perceived to facilitate the desirable
outcomes of reduced conflict directed towards both themselves and the interna-
tional environment broadly defined. These latter expectations about belief stabil-
ity derive from the assumption that operational code belief change may be in
part an active attempt to influence others. Thus, I expect that there is no strong
incentive for leaders to change their beliefs, nor is there any strong disincentive to
maintaining existing beliefs in a peaceful environment.
However, when conflict does occur, I expect that U.S. presidents will respond
by shifting their beliefs, as they “learn” that their previous beliefs were inaccurate,
or were inappropriate for interpreting and dealing with the international political
environment. Shifts toward an increased preference for cooperation or toward an
even more extreme preference for conflict may occur, reflecting either passive
change in beliefs better reflecting the nature of a changed situation or a tactical or
strategic means for attempting to reduce conflict. This response may or may not
be a preference for appeasement, as this change may not reflect actual intentions.
Conversely, initially held preferences for cooperation may be seen as inappropri-
ate for dealing with increased conflict, after which a change toward increased
conflict or for even greater cooperation may result.
Regarding the image (P-1) and locus of control (P-4) beliefs, my expectations
here may lead to counter-intuitive findings, which would prompt discussion. For
instance, why would a leader shift his beliefs regarding the image of the environ-
ment toward seeing the “other” as more friendly following increased hostility
 Experiential Learning by U.S. Presidents 193

abroad? Conversely, why would a leader potentially experience more control

over events following an increase in the level of international hostility? Intuition
suggests that an increased perception of hostility and reduced feelings of control
would only make most sense following increased conflict abroad. However, the
reason could be that these shifts are attempts to signal others in the international
political environment that if they will help to reduce the level of international
conflict, the United States is willing to take a less judgmental position by express-
ing increased beliefs of friendliness or adopt a less domineering strategy by articu-
lating reduced attributions of historical control to self.
Regarding the relative power variable, higher levels of relative GDP may lead
to more stable belief levels, as greater U.S. power reinforces the notion that pre-
existing beliefs are appropriate to maximizing the relative position of the United
States in the world. Conversely, lower levels of relative GDP may contribute to
more variation in beliefs, as less U.S. power suggests that one’s pre-existing beliefs
are inappropriate for dealing with the world “as it is.” Expectations here are in
the opposite direction to the expectations about the impact of greater hostility
from abroad—higher levels of relative GDP are more likely to reinforce faith in
the status quo of one’s beliefs, whereas lower levels are more likely to provide an
impetus to change, either due to inappropriate or incorrect perceptions or via an
attempt to influence the perceptions that others hold toward self’s beliefs, with the
hope of altering others’ behavior.

Domestic Factors
Beyond the realist perspective outlined above, there are a number of scholars
who seem to treat domestic influences as equivalent or as even more signifi-
cant than their international counterparts regarding the impact of these factors on
political leaders’ foreign policy beliefs. Rose (1998) refers to these individuals as
being of the Innenpolitik (domestic policy) school. Studies primarily focused on
the influence of public (Risse-Kappen 1991; Downs and Rocke 1994; Collier
and Sullivan 1995; Page and Barabas 2000; Sobel 2001) and congressional (Olson
1976; McCormick and Wittkopf 1990; Meernik 1993; Schraufnagel and Shellman
2001; Marshall 2003) support, bureaucratic and organizational political consider-
ations (Allison 1969; Holland 1999; De Castro 2000; Drezner 2000; Christensen
and Redd 2004), and domestic ethnic and business interests (Allison and Halperin
1972; Lindblom 1977; Mitchell 1997; Brener et al. 2004) are some prominent
examples.
This approach appears to stem from the fact that political leaders in democra-
cies are elected servants of the people, who frequently interact and have to com-
promise with other elected servants. Given this constraint, if the president does
not effectively fulfill the expectations of the electorate, then he will not retain the
political support to remain in office and/or to enact his policies of choice. As I am
concerned with U.S. presidents in the post-cold war world (George H.W. Bush
194 Samuel B. Robison

after 1990, Bill Clinton, and George W. Bush), we may expect stronger findings
here regarding the impact of domestic factors than if we were examining cold war
presidents. Given the reduced fear of nuclear annihilation or of a potential global
ideological shift toward communism following the demise of the Soviet Union,
these presidents are probably less constrained by the international environment
and more sensitive to domestic factors. Here I test whether these domestically
based concerns affect the stability of one’s foreign policy beliefs.
Therefore, the Innenpolitick hypothesis for post-cold war presidents is that the
factors originating in the domestic political environment have a stronger influ-
ence on presidential operational code beliefs than do those originating in the
international political sphere. Domestic political factors examined here include
public and congressional support levels for the U.S. president. Specifically, I
expect that when public and Congressional support levels for the president are
low, the president experiences more belief change than is the case when pub-
lic support for the president is high. The expectation here is that low levels of
domestic support will facilitate presidential belief change, as the president will
be unable to pursue his preferred set of policies without the necessary levels of
domestic support. Thus, he must appease the public and the Congress by shift-
ing his beliefs in periods where these beliefs do not match up with what these
groups expect.

Mixed Domestic and International Factors

There are some factors that are more explicitly a mix between international and
domestic than are the others examined above, and these are the economic con-
siderations of levels of inflation and domestic unemployment. The degree to
which these factors are perceived as existing in one of the two domains of inter-
est (international vs. domestic) here is likely to be affected by the perceptions of
the president in office, as well as the specific situation in which these factors are
perceived.4 As such, I classify these variables as being both internationally and
domestically based.
I do not have any expectations regarding the impact of these “mixed” factors
relative to international or domestic ones, as they might be expected by both real-
ists and scholars of the domestic policy school to impact beliefs. However, I do
have specific expectations for the directional impact of the economic situation on
“absolute” belief change. As inflation and unemployment levels rise, U.S. presi-
dents are more likely to shift their beliefs in order to better interpret or deal with
an increasingly undesirable economic context. Thus, if preferences for conflict
are seen to breed trade policies that are detrimental to U.S. economic well-being,
then a U.S. president might shift his beliefs toward preferring cooperation, and
may concurrently begin treating others in a more positive light. Conversely, if
in periods of economic downturn coddling others is seen to lead to attempts at
economic exploitation of the United States, the president may well shift these
 Experiential Learning by U.S. Presidents 195

beliefs toward a preference for conflict and may begin representing others as more
hostile.5
Finally, I expect that certain events targeting the United States will yield a
greater impact on a president than will others. These events are broken down
into three broad categories of policy domains: (1) All Behavior Domain defined
as all actions initiated against a state; (2) All Cooperation Behavior Domain defined
as all cooperative actions initiated against a state; (3) All Conflict Behavior Domain
defined as all conflict actions initiated against a state. At a given point in time
leaders might pay more attention to actions emanating from any one of these
domains over the others. I am interested in testing whether, on balance, we see
leaders respond as “realists” would expect or not, with the “realist” expectation
being that leaders will most strongly recognize and respond to conflict behavior.
This expectation stems from the idea that the achievement of state security (either
narrowly or broadly defined) is a critical goal in international affairs. As conflict
behavior can more readily threaten security, it will be viewed much more closely
and treated much more seriously than will less conflictual behavior.
Thus, if we assume that conflict targeting the United States will be selectively
perceived by a president as more important than other types of behaviors (as I
expect realists would predict), we should expect a closer relationship between
changes in the independent variables of interest here and the dependent variables
of absolute belief change when we examine in isolation the influence of conflict
actions on belief change. If the evidence suggests that it is not the case and that
cooperative actions influence beliefs to the same extent as conflict actions do, then
we might question the degree to which the realist perspective outlined here helps
us understand absolute changes in presidential beliefs. Therefore, our hypothesis is
that the influence of others’ actions directed toward the United States is strongest
when evaluated in the domain of conflict behavior.

Methods and Variables

The dependent variables of interest in this study are absolute belief change in the
image of others (P-1), the strategic orientation of self (I-1), and self’s belief regard-
ing the locus of control between self and others (P-4). Each variable is created by
first subtracting the VICS index value of the operational code belief for a given
month from the index value for the subsequent month ([op. code value at t + 1
minus op. code value at t] = change in operational code belief from “this” month
to “next” month). The absolute value of this number is then taken and associated
with the latter month evaluated in this measure. In other words, if a leader had a
P-1 value of .7 in January, and .5 in February, then he would initially get a –.2,
as there was a .2 decrease in this leader’s operational code value from January to
February. Given that I am interested here in the “absolute” change in beliefs,
however, this leader’s value would then be re-calculated into a non-negative .2.
For the purposes of analysis, this absolute value of .2 would then be listed as this
196 Samuel B. Robison

leaders’ absolute P-1 value change for the month of February (since the change
occurred in the time between January and February), and compared against inde-
pendent variable values occurring in January. Thus (excluding yearly measures), I
examine a one-month lag from the observation of independent variable values to
the observation of dependent variable values.
All independent variables are evaluated at the level of the month. Some vari-
ables occur more frequently than once a month in their raw state, and these values
are aggregated into monthly observations. For yearly measures the same yearly
value was applied to each month of a given year. For the measure evaluating the
impact of other states’ actions on U.S. presidential beliefs, I use event data taken
from Gary King’s “10 Million Dyadic Events” database.6 This database of dyadic
events is gathered from the first-sentence “lead” of all Reuters news stories from
1990 through 2004—thus, data here include actions initiated by every state in
the world. I re-scored each nominal-level “IDEA” event from the King data set
into the roughly interval-level, “Goldstein” format, which places each event on
a scale from –10 (most conflictual) to +10 (most friendly).7 Then I removed all
non-events from this analysis (those without a Goldstein value), and restricted the
data to focus exclusively on actions initiated by state actors. Finally, I averaged
out all events based upon their “Goldstein” values for each month, giving me the
“average” level of cooperation or conflict focused on the United States for a given
month across all states, and used this value for analysis.8 For the evaluation of data
from the cooperative versus conflictual event domains, I first restrict the data to
only those actions that were cooperative (Goldstein value greater than 0) or con-
flictual (Goldstein value less than 0), respectively, and then average the monthly
“Goldstein” value associated with these events for each month.
For an amount of conflict in the world measure, I use a count of the total number of
militarized interstate disputes (MIDs) that occur for a given year. MIDs are taken
from the Militarized Interstate Disputes V.3.10 database, part of the Correlates of
War project.9 For a measure evaluating U.S. relative GDP, I divide U.S. GDP by
World GDP (both measured in billions of dollars)—which is also a yearly mea-
sure. The U.S. GDP measure used here was taken from the U.S. Department of
Commerce, Bureau of Economic Analysis.10 The World GDP measure was taken
from the World Development Indicators Online.11
For domestic variables, public support (approval) for a president was taken from
the Roper Center for Public Opinion Research.12 I look at “approval” levels only
(thus, higher values equal more approval for the president). Congressional support
levels are taken from George C. Edwards’s work on Presidential-Congressional
relations.13 I use Edwards’s “overall support scores” from this database which gives
the percentage of votes for each Member of Congress (in both the House and
Senate) for a given year that match up with those policies for which the president
has openly given support. I average these values for all House and Senate mem-
bers for each year, respectively, and use two variables for analysis here—one for
the House of Representatives, and one for the Senate. Regarding the “mixed”
 Experiential Learning by U.S. Presidents 197

domestic and foreign variables, U.S. inflation measures were taken from the infla-
tiondata.com website.14 This rate was calculated based upon the Consumer Price
Index (CPI-U), which was taken from the U.S. Bureau of Labor Statistics. U.S.
unemployment measures were taken from the U.S. Department of Labor Statistics
website.15 These data reflect seasonally adjusted unemployment levels for persons
in the United States age 16 and over.
Presidential dummy variables are included to evaluate the impact of each presi-
dent on the overall findings. Given that the event data evaluated here only exist
for 1990 through 2004, George Herbert Walker Bush, Bill Clinton, and George
W. Bush are examined. George H.W. Bush provides the excluded variable in
these analyses. All statistical tests here are panel-corrected standard error (PCSE)
analyses (Beck and Katz 1995). This type of analysis is similar to an ordinary least
squares (OLS) regression, except that it corrects for potential issues of heteroske-
dasticity across the cases examined using panel data (as is done here).

Models and Results

Results for the statistical models are in Table 11.1. All relationships are examined
at the two-tailed level of significance in order to provide a more strenuous test of
these hypotheses. The first set of models looks at the domain of all actions target-
ing the United States, followed by other models for the cooperation actions and
conflict actions domains.
In terms of absolute belief change for (P-1) image of the other, U.S. presidents
are significantly influenced by actions targeting the United States from the all
actions domain. However, this influence is not in the expected direction (t = 2.44,
p<.05). The positive coefficient here suggests that when foreign actors treat the
United States in a friendlier manner, the president is more inclined to yield an
absolute shift in his beliefs than when the United States is treated in a more hostile
manner.
Thus, in the case of the (P-1) image belief, it appears that the president responds
to increased conflict with increased rigidity on this measure. Though contrary
to hypothesized expectations, this finding does make some sense, as U.S. presi-
dents could respond to periods of conflict and uncertainty by strengthening their
resolve—reinforced and signaled by a rigidity of beliefs. Conversely, in periods
when the president has little to fear from his peers (periods of low conflict), he
may possess the freedom of movement to hold more varied beliefs without fear
of serious ramifications for appearing to “waffle” under the duress of conflict.
Finally, the coefficients testing the level of all actions in the world and of U.S.
relative power as influences on the image of the other does not yield significant
values in the all actions domain. These findings give no real support for either real-
ist hypothesis concerning the (P-1) image belief.
Regarding the model of domestic variables in the all actions domain, the public
approval measure yields a significant, expected (negative) relationship (t = 4.70,
TABLE 11.1 Factors Influencing Presidential Learning

Image of the External Environment Strategic Orientation Historical Control

All Actions Cooperative Conflictual All Actions Cooperative Conflictual All Actions Cooperative Conflictual

All Actions 0.089 — — 0.120 — — 0.016 — —

(–) [2.44]** — — [2.46]** — — [1.13] — —
Cooperative — –0.119 — — –0.071 — — –0.034 —
Actions (–) — [2.13]** — — [0.91] — — [1.53] —
Conflictual — — 0.038 — — 0.050 — — 0.026
Actions (–) — — [1.19] — — [1.16] — — [2.12]**
MIDCount –0.003 –0.004 –0.003 –0.005 –0.006 –0.006 0.001 0.001 0.001
(+) [1.24] [1.59] [1.33] [1.81]* [2.25]** [2.01]** [1.00] [0.77] [1.11]
U.S. Relative 28.310 24.958 32.162 5.127 1.681 10.110 –7.655 –8.415 –4.108
GDP (–) [1.15] [1.01] [1.26] [0.14] [0.05] [0.27] [0.75] [0.83] [0.40]
Public Support –0.007 –0.007 –0.007 –0.004 –0.003 –0.003 0.000 0.000 0.000
(–) [4.70]*** [4.43]*** [4.64]*** [1.90]* [1.63] [1.71]* [0.25] [0.54] [0.42]
House Support 0.002 0.003 0.002 0.003 0.004 0.003 –0.000 –0.000 –0.001
(–) [0.64] [0.75] [0.52] [0.80] [0.87] [0.67] [0.33] [0.24] [0.53]
Senate Support 0.007 0.008 0.008 0.003 0.004 0.004 0.000 0.000 0.000
(–) [2.68]*** [2.75]*** [2.82]*** [1.00] [1.29] [1.27] [0.07] [0.14] [0.10]
Inflation –0.023 –0.028 –0.028 0.005 –0.001 –0.001 –0.005 –0.006 –0.006
(+) [1.29] [1.57] [1.53] [0.22] [0.04] [0.03] [0.71] [0.87] [0.87]
Unemployment –0.086 –0.072 –0.079 –0.049 –0.034 –0.040 0.009 0.012 0.009
(+) [3.04]*** [2.58]*** [2.78]*** [1.31] [0.89] [1.04] [0.78] [1.06] [0.79]
Clinton –0.120 –0.107 –0.100 –0.052 –0.030 –0.025 –0.058 –0.056 –0.054
Dummy [1.30] [1.14] [1.07] [0.49] [0.28] [0.23] [1.54] [1.53] [1.46]
W. Bush –0.291 –0.321 –0.306 –0.221 –0.261 –0.242 –0.044 –0.049 –0.039
Dummy [2.71]*** [2.99]*** [2.83]*** [1.84]* [2.19]** [2.01]** [1.08] [1.24] [0.99]
Constant 0.595 0.894 0.628 0.828 1.019 0.873 0.360 0.443 0.374
[1.90]* [2.59]*** [1.95]* [2.03]** [2.18]** [2.08]** [2.80]*** [3.24]*** [2.96]***
N 168 168 168 168 168 168 168 168 168
R-Squared 0.266 0.261 0.249 0.259 0.232 0.234 0.064 0.071 0.083

* p < .10, ** p < .05, *** p < .001.

 Experiential Learning by U.S. Presidents 199

p<.001) with image of the other; however, the Senate congressional approval measure
has a significant, unexpected (positive) relationship (t = 2.68, p<.001). The public
approval finding suggests that as the public views the president in a more nega-
tive light, he responds by shifting his image of others in the political universe to
a greater extent than is the case when he holds higher levels of public support.
Conversely, the Senate finding suggests that his beliefs about others are more
variable when he holds a high level of support here, and that his beliefs about the
political universe are more stable when Senate support is low. Thus, the public
opinion finding gives support to our domestic factors hypothesis, while the Senate
support finding runs counter to it.
Perhaps the U.S. presidents examined here feel that during periods of low
Senate support, the Senate will feel even less confidence in (or will present more
barriers to) presidential policies and agendas when these presidents waver in their
beliefs. This posture is in opposition to their stance toward the public, whom
presidents may perceive as being more responsive, in a positive sense, to presi-
dential maintenance of pre-existing beliefs during periods of high support, and to
attempts to adapt and change when the public supports them less.
For the “mixed” variables, unemployment yields a significant, negative shift (t =
3.04, p<.001), again in opposition to the expectations in our hypothesis. Higher
levels of unemployment result in more stable (P-1) image beliefs, whereas lower
levels of unemployment lead to more volatile beliefs in the all actions domain.
Again, perhaps this finding signals that in times of economic crisis, the president
prefers to demonstrate resolve, whereas during periods of economic well-being he
may possess the political “wiggle room” to shift his image beliefs more frequently
without fear of a backlash in support. Finally, the presidential dummy variable analy-
sis in the all actions domain for the (P-1) image model shows that the George W.
Bush measure is negative and statistically significant, suggesting that he held much
more stable perceptions of the outside political world than did either his father or
Bill Clinton (t = 2.71, p<.001).16
In summary, only one of the internationally based variables and two of the
domestically based variables are statistically significant in this model (along with
one mixed measure and one presidential dummy control variable). Further, the
domestic variables both yield a stronger correlation with the dependent variable
of absolute belief change than does the international measure of actions target-
ing the United States; the public opinion measure yields a relationship with the
dependent variable twice as strong as this international measure. However, I do
not believe that this demonstrates conclusive support for the idea that either inter-
national or domestic variables yield relatively stronger influences on the depen-
dent variable on (P-1) image of others. Thus, I do not find strong support for
hypotheses from either the realist or Innenpolitik theories regarding belief change
about the international environment.
The next model examined in the all actions domain is the (I-1) strategic ori-
entation model. Similar to the findings in the (P-1) image model, the actions
200 Samuel B. Robison

targeting the United States are positively and significantly correlated with the
dependent variable of absolute belief change (t = 2.46, p<.05), which means that
higher levels of conflict targeting the United States lead to increased belief stabil-
ity regarding presidential “strategic” orientations, and lower levels of conflict lead
to more variability. Although this result again runs counter to passive learning
expectations, it can be explained as an active attempt to maintain resolve during
times of trouble and to ensure that a predictable stance will reduce the room for
misinterpretation of behavior or beliefs.
In contrast to the (P-1) image model, the level of conflict in the international
system (MIDCount) is another international-level measure that yields a modestly
significant, unexpected finding (t = 1.81, p<.10) regarding (I-1) strategic orienta-
tion. The negative coefficient means that higher levels of international conflict
lead to increased belief stability on this measure of instrumental beliefs, and that
lower levels lead to more belief change, following the pattern observed thus
far with the actions targeting the United States measure. Thus, I do not find sup-
port for the realist hypothesis in the all actions domain for the strategic orientation
model.
In terms of domestically based variables, the public support measure in the all
actions domain for the model is negatively and significantly correlated with the
(I-1) strategic orientation dependent variable (t = 1.90, p<.10). Higher levels of
public opinion lead to more stability in strategic orientations while lower levels
of public opinion lead to more variation, as expected, supporting the Innenpolitik
hypothesis on this measure. However, neither of the congressional support measures
is significant here, failing to give support to the hypothesis for these measures.
Neither of the mixed economic measures is significant, though the George W.
Bush dummy variable is again negative and significant (t = 1.84, p<.10), suggesting
that Bush 43 holds much more stable strategic orientation beliefs than does George
H.W. Bush (“Bush 41”) or Bill Clinton.17
Overall, the international measures appear to have a somewhat greater influ-
ence on belief change in the strategic orientation model than do their domestic
counterparts, but as with the image model, these findings do not decisively show
one overshadowing the other. Thus, these findings do not give overwhelm-
ing support to hypotheses from either realist or Innenpolitik theories. Finally, an
examination of the presidential (P-4) locus of control model reveals that none of
the independent variables have a significant influence on the dependent variable.
Thus, I do not find support for any of the realist or Innenpolitik hypotheses in the
all actions domain.
The second set of models examines actions perceived in the cooperation actions
domain. I isolate only the cooperative actions targeting the United States, which
range on a scale from extremely cooperative to only slightly cooperative, then
aggregate these actions across actors at the level of the month, and include this
measure as an independent variable in the image, strategic orientation, and his-
torical control models. In other words, we are imagining that the president pays
 Experiential Learning by U.S. Presidents 201

selective attention to only cooperative actions, and we are evaluating only the
impact of these perceived actions on belief change.
In the cooperation actions domain the findings for the (P-1) image model are
precisely the same as those in the all actions domain—with the same variables
yielding a significant influence in the same direction. The one exception is for
the actions targeting the United States measure—the relationship here is still sig-
nificant, but this time in the direction expected by the realist hypothesis (t =
2.13, p<.05). When actions targeting the United States are cooperative in nature
but are less cooperative than what they could be, controlling for various other
factors, the president is more inclined to exhibit a change in beliefs than when
these actions are more cooperative. In the cooperation actions domain the results
for the (I-1) strategic orientation model reveal that the actions targeting the United
States measure loses significance, as does public approval. The count of MIDs
and the Bush 43 dummy variable analysis are both still significant and in the same
direction that as in the all actions domain. Finally, the historical control model
again yields no significant findings.
Lastly, in the conflict actions domain the actions targeting the United States mea-
sure are no longer significant for the (P-1) image model. There is no support for the
realist hypothesis regarding the impact of the level of conflict in the international
system (MID Count) or power position (U.S. relative GDP) on the image belief.
Taking the conflict actions domain into account, all other domestic and mixed
factors hold the same type and rough degree of influence that they did in the previ-
ous two models of belief change for the (P-1) image belief. In the conflict domain
the MID count (t = 2.01, p<.05), public approval (t = 1.71, p<.10), and Bush 43
dummy variables (t = 2.19, p<.05) are all significant in a negative direction for
the (I-1) strategic orientation model, just as they were in the all actions domain. The
conflict actions targeting the United States measure are again not significant.
However, the conflict actions targeting the United States measure are sig-
nificant in a hypothetically unexpected direction (t = 2.12, p<.05) for the (P-
4) historical control model, something that was not found in the all actions
domain. The positive coefficient here suggests that more conflict actions
coincide with more stable locus of control beliefs, and that these beliefs are
more prone to change when there is relatively less conflict. This pattern sup-
ports the notion that in the domain of conflict, resolve in terms of a stable
locus of control is seen to be more valuable when actions are relatively more
conflictual.
As with the other models, no other independent variable is significantly related
to the (P-4) locus of control as a dependent variable. However, the historical con-
trol model is notable for being the only one to give any support to the importance
of actions targeting the United States measure in the conflict actions domain—
despite the fact that the significant relationship exists in an unexpected direction.
To sum up, the findings in the conflict actions domain fail to give strong support
either to realist or Innenpolitik hypotheses.
202 Samuel B. Robison

Discussion
Although I do not find support for all of the hypotheses presented here, I do
discover important influences on “absolute” belief change in U.S. presidential
operational codes—findings that provide new insights into the nature of short-
term presidential “learning.” I expected that periods of conflict in the interna-
tional environment or instances of decreased domestic political support for the
president’s policies would lead to greater levels of belief change than would be
evident in less challenging times. What I mainly found regarding absolute changes
in the (P-1) image and (I-1) strategic orientation models was the opposite—when
times are tough, the president tends to maintain more stable beliefs than is the case
in other periods. This pattern seems to reflect a phenomenon across presidents
that periods of hardship (specifically in relation to the international political envi-
ronment and the Senate) are best addressed by maintaining an air of stability and
resolve. When times are less challenging, then the president may perceive more
room for autonomy with a reduced chance of negative consequences following
a change in the beliefs expressed in presidential rhetoric. In these times, he may
allow himself more freedom to shift beliefs, and he may experiment with the
use of different signals as a means to maximize U.S. bargaining leverage vis-à-vis
others or to better promote a given policy agenda. During periods of hardship
such changes may be viewed as too risky, and thus they may be supplanted by
rhetorical and belief stability as a means to better ensure security and political
well-being.
The exception to this generalization is in regard to domestic public opinion.
When public support is low, presidential beliefs regarding his representation of
the nature of both “others” and of the “self” in the foreign political sphere tend
to shift. When public support levels are higher, the president maintains his beliefs,
possibly out of fear that the public will yield less support for him and his policies if
he shifts his representation of the external political environment and the U.S. role
in it. Overall, domestic factors seem to have a slightly stronger influence regarding
the (P-1) image belief, and international factors seem to play a slightly stronger
role regarding (I-1) strategic orientation.
Ultimately, the findings here do not give overwhelming support to either the
realist or Innenpolitik hypotheses. Further, the president in the conflict actions
domain actually seems to be less sensitive to absolute belief change in image and
strategic orientation in response to relative changes in the level of conflict target-
ing the United States than is the case when conflict, cooperation, and neutral
actions are evaluated together. Thus, it does not appear that presidents pay par-
ticular, selective attention to conflict actions, at least as they affect absolute belief
change, with the sole exception of the (P-4) locus of control measure.
This chapter represents one step toward promoting a better understanding of
how and when the beliefs of political leaders change—an area of study that has
received a fair amount of theoretical investigation, yet has received little in the way
 Experiential Learning by U.S. Presidents 203

of multi-case, across-time quantitative analysis. Additionally, though researchers

often give lip service to the notion that a “multi-dimensional” approach to inter-
national relations research is preferable, we continually see exclusive foci on one
area of study or level of analysis at the expense of others. Here it appears that
factors from both the international and domestic levels play important roles in
influencing changes in beliefs. Thus, neglecting one level or the other would lead
to a less complete understanding of the nature of belief change than that captured
here. In this sense this study reinforces the call for a multi-dimensional approach,
if we are to take into account all of the most relevant factors contributing to phe-
nomena of interest in the foreign policy sphere.
Regarding the nature of belief change, many adhere to the argument laid out
by Jervis (1976) that the beliefs of political elites are established either prior to
coming into office or early during one’s time in office. From this perspective the
expectation is that changes to these beliefs are infrequent and are typically the
result of either major external “shocks” or of the steady accumulation of informa-
tion running counter to one’s existing view of the world. However, the findings
here suggest that belief change, at least in the short term, is fairly commonplace
when the international or domestic context demands it. What this analysis does
not address is the possibility that these changes are only ephemeral, and are not
truly substantial, enduring changes. It may be that Jervis and others are right and
that these beliefs are relatively stable over the longer term, fluctuating around
some “mean” point that ultimately does not change much. Nevertheless, these
beliefs clearly demonstrate a fair amount of variability at the level of the month,
which is explained in some measure by the international and domestic factors
examined here.

Notes
1 It is possible that “epiphanies” of some kind or brain damage to a leader will result,
impacting “beliefs” without a preceding change in one’s perceived environment.
Such changes would be examples of “non-learning” belief change. However, these
occurrences are expected to be extremely rare exceptions to the rule, and as such, any
instance of belief change seen here will be attributed to the impact of the experience
or observation of events.
2 Note that I use the term “active” here to delineate the idea that beliefs may change in
order to influence others (as opposed to suggesting that belief change only reflects a
changed way of perceiving the world). This is not to say that I assume these changes
are conscious choices. It is conceivable that, in some situations, these changes could
occur as an uncalculated, possibly subconscious response to external stimuli.
3 Note that I use the term “realist” here as a general “catch-all” to encompass those
scholars whose research and theories follow the tenets laid out by Hans Morgenthau,
E. H. Carr, and a number of other prominent international relations scholars who
believe that we live in a conflict-filled, “anarchic” world, where state security trumps
all other considerations in the foreign policy sphere.
4 Initially, given that these factors exist, narrowly, in the domestic sphere, one might be
inclined to classify them as domestic-level influences. However, given the increasingly
globalized nature of international trade and finance, it would be too simplistic to view
204 Samuel B. Robison

almost any “domestic” economic factor as being exclusively domestic in nature. The
foreign imposition of barriers to trade, “dumping” of cheaply made products into
the United States, the buying and selling of U.S. currencies abroad, and other factors
play critical roles toward impacting domestic considerations such as U.S. employment
rates.
5 Note that both inflation and unemployment are examined as these measures evalu-
ate very different things, and though the two are inter-related, they might each be
expected to exhibit independent effects on the U.S. economy, i.e., there is not always
a direct, linear relationship between the increase in one and the increase or decrease in
the other (see Block 1981).
6 Gary King and Will Lowe 2003. “10 Million International Dyadic Events”, hdl:1902.1/
FYXLAWZRIA UNF:3:dSE0bsQK2o6xXlxeaDEhcg== Murray Research Archive
[Distributor]; available: http://dvn.iq.harvard.edu/dvn/dv/king/faces/study/StudyPage.
jsp;jsessionid=a2bd6a40151a4f29aa5890d5bc42.dvnInstance1?studyId=505, accessed
04/20/08.
7 For more on Goldstein values, see Goldstein (1992), and “Modified WEIS Codes”
from the Kansas Events Data website—available: http://web.ku.edu/keds/data.dir/
KEDS.WEIS.Codes.html (accessed October 15, 2008). For the Goldstein association
with IDEA values, see King and Lowe (2003: 622–623).
8 Regarding the aggregation decisions on the events data variable here, I was interested in
gauging the “tenor” or “color” of the international political environment as it explicitly
targets the United States, particularly since this aggregation works well as a comparison
with the broad, “global” evaluation of the operational code as evaluated here. I use
all types of foreign policy-related speeches in order to construct the operational code
as used here, focused on a number of specific and non-specific actors. Therefore,
I believe that aggregating all actions targeting the United States at the level of the
month, across all actors (friendly and non-friendly; allied and non-allied) provides an
appropriate measure of behavior.
9 Available: http://www.correlatesofwar.org/COW2%20Data/MIDs/MID310.html
(accessed September 15, 2008).
10 Available at: http://www.bea.gov/national/nipaweb/TableView.asp?SelectedTable=
5&ViewSeries=NO&Java=no&Request3Place=N&3Place=N&FromView=YES&F
req=Year&FirstYear=1960&LastYear=2007&3Place=N&Update=Update&JavaBox
=no#Mid (accessed September 15, 2008).
11 Available at: http://ddp-ext.worldbank.org.libezp.lib.lsu.edu/ext/DDPQQ/member.
do?method=getMembers&userid=1&queryId=6 (accessed September 15, 2008).
12 Available: http://www.ropercenter.uconn.edu/ (accessed July 20, 2008).
13 Available: http://presdata.tamu.edu/ (accessed July 10, 2008).
14 Available: http://inflationdata.com/inflation/Inflation_Rate/HistoricalInflation.aspx?
dsInflation_currentPage=3 (accessed July 3, 2008).
15 Available: http://data.bls.gov/PDQ/servlet/SurveyOutputServlet (accessed July 3,
2008).
16 Other work by this author (Robison 2009) shows that these beliefs were stable and
negative, reflecting a strong perception of international hostility.
17 Some of my other work also demonstrates that these beliefs were stable and negative,
reflecting a strong orientation for conflict (Robison 2009).
12
COGNITIVE RESPONSES BY
U.S. PRESIDENTS TO FOREIGN
POLICY CRISES
Belief Changes in Response to Positive or
Negative Experiences

B. Gregory Marfleet and Hannah Simpson

Introduction
The adage that “those who do not remember history are condemned to repeat
it” resonates among international relations and foreign policy scholars (Bennett
1999; Pastor 2001). However, inquiries regarding the capacity of leaders to learn
something from experience have a checkered historical legacy of their own. Hegel
(1894) claimed that “[w]hat experience and history teach is this—nations and gov-
ernments have never learned anything from history, or acted upon any lessons they
might have drawn from it.” Evaluating claims like Hegel’s has been a central but
vexing challenge for observers of foreign policy. Learning is both difficult to define
and to study.
Many scholars who have approached the topic display ambivalence about evi-
dence that governments or leaders are able to learn (Stern 1997; Tetlock 1991;
Levy 1994). Nevertheless, assessing whether, or how, leaders, states, organizations
and institutions incorporate information acquired from direct experience, from
observing others, or from history, is a central concern of foreign policy scholar-
ship. The profusion of studies on learning, lesson-taking and the uses of history
by leaders suggests that scholars believe the reward is worth the risk (Zimmerman
and Axelrod 1981; Etheredge 1985; Breslauer and Tetlock 1991; Stein 1994;
Lebovic 1995; Leng 2000; Dyson and Preston 2006; Herrmann and Choi
2007).
A key conceptual problem for learning-centered research is deciding under
what conditions behavioral changes over time—and in response to stimuli—can
accurately be described as learning. Does a diagnosis of learning require evidence
of corresponding internal processes of a cognitive (in the case of leaders) or organ-
izational (in the case of bureaucracies) nature?1 Levy (1994: 286) insists that simple
206 B. Gregory Marfleet and Hannah Simpson

behavioral change is insufficient to support a claim that learning has occurred. He

defines behavioral changes without concomitant changes in internal processes
as either “simple” or “instrumental” social learning. In social learning, actors may
exhibit changes in tactics in response to the actions of others without altering their
underlying preferences and beliefs. In the instrumental form of social learning,
actors may change behavioral tactics without altering their overarching strate-
gies or goals. Neither involves a significant reconsideration of assumptions about
adversaries or environments. This presents an epistemological hurdle for those
attempting to straightforwardly track behavior and then draw inferences about
underlying processes.
For Levy (1994: 286), true learning is experiential learning, which involves some
dimension of attitude, goal or belief change. This type of learning is comprised of
two levels. The first is “diagnostic learning,” which occurs when an agent alters
his beliefs “about the definition of the situation or the preferences, intentions or
relative capabilities of others” (Levy 1994: 285). The second level of experiential
learning is “complex learning” or “causal learning.” This form of learning occurs
when an actor changes his or her mind (altering self’s preference rankings) or
significantly transforms his or her understanding about the nature of the political
system. Both levels of learning are experiential learning, i.e., a process of cognitive
change rather than behavioral change or social learning.
If we accept the principle that real learning requires cognitive change, then
our puzzle becomes one of identifying the types of stimuli likely to generate this
phenomenon. Associated with this question are issues of frequency and kind.
Are repeated stimuli necessary? Can a single, isolated, but major event generate
the type of change required to evince cognitive change? Must the stimulus be
negative feedback or can positive feedback provoke some sort of “reinforcement”
learning (and could this be measured)?
These questions delineate an ongoing debate between proponents of either a
“cumulative” or “traumatic” learning model. The latter approach insists that the
magnitude of the negative experience is the most important factor. The stronger
or more catastrophic the failure, the more salient it is and the greater the influence
on learning. For example, Vasquez (1987: 376) suggests that learning from war,
not only “whether the war was won or lost” but “the cost” and the “magnitude
of that cost,” will influence approaches to the next conflict. In contrast, defend-
ers of a cumulative model have noted that individuals often rationalize away
single instances—even large ones—if they fail to conform to prior beliefs (Jervis
1976). Isolated events may be insufficient to generate cognitive restructuring.
Multiple and recurring setbacks may be required for cumulative learning to occur.
Analogies have been made between this type of learning from repeated trials
and changing patterns of behavior in structured situations like iterated prisoners’
dilemma games or in other sequential games where agents “reconsider their basic
goals or objectives only after repeated strategic failures” (Levy 1994: 281, 286).
Similarly, cybernetic decision models and evolutionary models arising from the
 Cognitive Responses by U.S. Presidents to Foreign Policy Crises 207

literature on complex adaptive systems have offered an incremental conception

of learning through reactions to cumulative negative feedback (Steinbruner 1974;
Farkas 1996).
Of course, all of this attention to negative feedback (on either a small or large
scale) could also be a mistake. Learning can occur after success. However, this
type of learning might be especially problematic to measure if experience rein-
forces rather than alters cognitive patterns (Walker and Schafer n.d.). In fact,
the daunting methodological problems presented by reinforcement learning may
explain our attention to negative feedback. Imagine a situation where—after a
success—behavior or policy remained constant, and there was no measurable
change in beliefs or goals over time.2 A researcher attempting to study learning
from success would confront the task of trying to explain an invariant dependent
variable with an invariant independent variable—a very unconvincing research
design (King et al. 1994).
Conceptual confusion about when learning might occur, the substantial data
requirements needed to determine whether cognitive, belief, and/or goal changes
have occurred, and the issues presented by the cumulative and traumatic models
have led researchers in the field to resolve these difficulties by making simplifying
assumptions about (unobserved) internal processes. Rational learning models are
among the most prominent of these simplifications. From this perspective learn-
ing is the process of incorporating new information that allows the agent to revise
probabilistic expectations about outcomes and to re-evaluate the values of these
outcomes as part of their expected utility calculations (Herrmann and Choi 2007).
Typically, however, this process is not directly observed. Consistent with onto-
logical assumptions about the agents in the system, the learning process is “black
boxed” and the focus is shifted to behavior.
Russell Leng’s (2000) work on recurring crises is an example of this type
of analysis. In his examination of a series of repeated crisis interactions within
Soviet-American, Egyptian-Israeli and Indian-Pakistani rivalry dyads, he focuses
on sequential “patterns of behavior” by mapping the progression of crisis hostility
and observing the escalations or de-escalations signaled by the actions of the crisis
actors. To assess learning, Leng determines whether the states in the dyad altered
their influence strategies in response to success or failure in previous encounters.
Because these crises occur in the same rivalry, in the same region, and with similar
power arrangements, his research design can claim some measure of control over
many confounding factors.
However, the methodological focus on overt behavior has important limita-
tions. Most problematically, learning might not always result in behavioral change
that would be manifested in national foreign policy. For example, crises are often
studied because response behavior is especially likely to be influenced by lead-
ers’ beliefs. Crises “threaten high priority goals” and demand quick responses
that preclude institutional influences (Hermann 1979). Even in crises, change in
behavior might be suppressed if actors are impeded by domestic, economic or
208 B. Gregory Marfleet and Hannah Simpson

bureaucratic constraints. Change might also not occur if learning does not result
in immediately applicable alterations of beliefs or cognitive structures. Learning
may involve “the acquisition of some new skill, ability, or knowledge that can be
recalled and used on some future occasion” but which will not manifest itself in
immediate policy change (Hermann 1979: 10).
To complicate things further, behavioral or policy change may occur without
learning. This pattern is likely if a leader is simultaneously playing “an interna-
tional game and a domestic game with every move,” and is strategically using
“developments at one level to affect choices made at the other” (Hudson and
Vore 1995: 227; Putnam 1988). In a crisis situation, leaders may resort to policy
changes out of desperation, not learning: “if a crisis drags on or there is a series of
crises, the leadership will try to get out of a situation that portends to drain domes-
tic support by adopting any policy or strategy that may result in victory, despite
the risks” (Vasquez 1987: 378). Policy change may not follow external environ-
mental change straightforwardly simply because of the number of intervening
variables in the form of other collective or individual actors who influence deci-
sions, as well as exogenous factors (Welch 1992). Therefore, behavior-oriented
studies of learning run a high risk of both false positive and false negative findings
(Type I and Type II errors).
Given the complex relationship between environmental stimuli and behavioral
responses, the analytic neglect of the cognitive process of learning is a serious prob-
lem. While this oversight may be understandable given the historical absence of
reliable indicators of changes in leaders’ beliefs, attitudes and perceptions, the mat-
uration of “at-a-distance measurement” of elites presents opportunities to address
this gap (Schafer 2000). New content analysis schemes coupled with the technol-
ogy of computer-aided content analysis have been generating insights into how
foreign policy decision makers process information and exhibit belief change.

Learning and Operational Code Analysis

Operational code analysis is an innovative tool for examining learning that offers
a means to escape “black box” approaches. It has advanced as an individual-cen-
tered, cognitively oriented investigation of the relationship between actors and
their political environment. Although Holsti (1977) first viewed operational codes
as internally consistent across policy areas and over time, subsequent research indi-
cates that operational code beliefs are more complex than first imagined (Walker
and Falkowski 1984; Walker et al. 1998). Recent research has shown that these
beliefs may vary over time, by issue areas, and within different contexts (Walker
and Schafer 2006). The presence of this variance naturally leads to questions about
its underlying cause. It seems possible to describe learning in response to experi-
ence as an evolution of an individual’s general psychological perceptions of the
“nature of the political universe” and of the principal strategies employed by Self
(Ego) and Others (Alter) to achieve political goals in that environment.
 Cognitive Responses by U.S. Presidents to Foreign Policy Crises 209

As the previous chapters of this volume have described, operational code analy-
sis has evolved from a qualitative and subjective textual analysis technique into
a systematic, computer-aided content analysis methodology with the introduc-
tion of the VICS system by Walker and colleagues (1998). This evolution has
enhanced the potential for learning-oriented research for several reasons. First,
the VICS computer-aided system is more reliable than human coders and its abil-
ity to generate more output improves validity by reducing the impact of noise-
induced error. This potential for consistent repetition is essential to attempts at
measuring learning-related belief change. Second, as the VICS system has been
developed and been implemented through numerous studies (Walker et al. 1999;
Crichlow 1998; Malici 2005; Malici and Malici 2005a, 2005b), a central reposi-
tory for leader scores has also emerged. These data allow a researcher to identify
which VICS indicators are most stable and calculate “norming means” derived
from a relatively large sample of world leaders (Feng 2005a; Malici and Malici
2005a, 2000b; Schafer and Walker 2006).
Third, the incorporation of norming means and variance information into the
interpretations of leaders’ VICS scores has fostered the development of a typology
of beliefs based on measures of relative centrality or extremity. This typology has
been used to build the “Theory of Inferences about Preferences” (TIP), based on
beliefs. TIP defines preference rankings that can be combined into normal-form
games, which suggest how a leader’s ranked preferences can be linked to different
possible equilibria of settlement, domination, submission and deadlock (Walker
and Schafer n.d.; Marfleet and Walker 2006). From the perspective of an analysis
of learning, these games indicate what the leader perceives as the likely outcomes
of various combinations of cooperative or conflictual behavior by each side of a
strategic dyad. These subjective games also suggest—once assumptions have been
made about where the game begins and who moves first—the likely strategy of
the opponent and the type of behavior the leader should adopt in order to achieve
the best outcome given these parameters.
Of course, these subjective games are only “worlds in the minds” of the
agents—mental representations of the perceived strategic environment—and
they may or may not comport with the “real” environment. Should an actor
perceive the opponent’s preferences inaccurately, the subjective game would be
misspecified. Consequently, the prescribed behavior, anticipated response, and
expected outcomes of the game would be potentially sub-optimal from the per-
spective of a third-party observer with full information. This gap between the
prescribed moves that are indicated by an agent’s subjective game, the anticipated
and objective behavior of the opponent, and the expected and actual outcomes
generated by dyadic interaction provides a means through which we may be able
to map learning. By looking at repeated crisis interactions between dyadic pairs,
it is possible to determine whether there are indications of learning that might
be reflected in a better approximation of the “true” relationship between the
two actors. This task is done by measuring whether, and by how much, the gap
210 B. Gregory Marfleet and Hannah Simpson

between the subjective crisis outcomes predicted by a leader’s pre-crisis beliefs

and the actual objective crisis outcomes changes.
The careful applications of the operational code construct and VICS techniques
may allow us to evaluate several learning-related questions. Does unanticipated
behavior from an opponent and/or an unexpected crisis outcome prompt leaders
to rethink their beliefs? Does the magnitude of the event influence whether this
occurs? If an actor’s beliefs change following an event, does this make him or her
less likely to make subsequent errors? Does belief change portend future behavio-
ral changes? The essential steps to identify learning involve establishing empirical
links between actions, outcomes, belief changes and behavioral shifts.

Crisis Outcomes and Cognitive Change

Our analysis mirrors studies that have looked for behavioral changes through the
examination of sequences of foreign policy crises and we employ a modified ver-
sion of this classic, quasi-experimental, research design. Unlike these past efforts,
however, we attempt to measure explicitly whether belief changes occur in the
aftermath of an event and before behavior changes are manifested in subsequent
events by using VICS content analysis techniques. Crisis events are particularly
useful for studying learning because they typically occur over a fairly limited time
span with distinct temporal boundaries. Crises have an onset date and a termination
date that delineate the stimulus. Crises also have outcomes that are often designated
in ways that are amenable to game-theoretic categorization: they generate winners
and losers, compromises or stalemates, and usually end with varying degrees of
satisfaction among participants.
Crises also exhibit differences in levels of intensity. This feature provides useful
variation in the level of stimulus across episodes and allows us to assess the impact
of event magnitude on belief change. Unlike issues of “low politics,” crises have
a military and security dimension that attracts the attention of decision-making
officials at the highest level. Crisis responses are often crafted by leaders directly
or in collaboration with their closest advisors. All of these characteristics— tem-
poral bounding, clear outcomes, variance in stimuli and salience to leaders—make
interstate crisis events the most common avenue for studies of learning in for-
eign policy (Hermann 1979; Brecher and Wilkenfeld 2000; Stern 1997; Khong
1992).
For our study we have selected crises from the actor-level version of the
International Crisis Behavior (ICB) Project data set (version 2.6) assembled by
Brecher and Wilkenfeld (2000).3 This data set includes crisis data for 975 crisis
actors involved in over 400 crisis events between 1918 and 2003. From these data
we identified all crises since World War II for which the United States was an
actor.4 The reasons for this selection are multiple.
First, in order to measure foreign policy beliefs using the VICS measures for the
operational code, some record of public statements by the political leader must
 Cognitive Responses by U.S. Presidents to Foreign Policy Crises 211

be readily available. It helps, given that computer-based content analysis will be

conducted, if these documents are in electronic form. Also, VICS is currently
designed to grammatically parse and identify parts of speech for English text.
While translated text has been used in several studies, obtaining translations of
leader-generated statements for some countries can be difficult, time-consuming
and expensive (if not impossible). In addition, in order to generate valid indica-
tors, speech material must appear with some density. Isolated or infrequently
recorded public statements can starve the quantitative content analysis technique
of sufficient data.
Second, leaders must be easily identifiable. Most countries have a single indi-
vidual who acts as either head of state or the government, or one who is assigned
foreign affairs decision-making responsibility. It is not always easy to determine
when transitions in power occur or where political authority lies. Government
instability generates leader identification problems and introduces confounding
factors if political or decision-making institutions collapse or change radically over
extended periods. Continuity of government, well-designated authority and clear
transitions between leaders are desirable features when crafting a study of elites.
Finally, an important dimension to observing learning involves comparing the
actions of agents at multiple time points. This necessitates the selection of a set of
events large enough to allow for meaningful temporal comparisons. Moreover,
comparison is further enhanced if the events are not singular, one-time occur-
rences but also include repeated interactions with the same adversary. The set of
U.S. cases has all of these characteristics. The U.S. president is easily identified.
His public speeches are assiduously recorded as part of the Public Papers of the
President series, and there is virtually always sufficient density of material to gen-
erate high-quality VICS measures even over multiple, time-delimited periods.
Because the United States is the most commonly occurring crisis actor in the data
set, this subset of cases yields enough observations to track temporal changes even
within the tenure of most presidents for a significant number of recurring crises.
We focus on the following indicators from the variables included in the ICB
crisis data set: (1) the behavior of the crisis opponent (was the triggering act of
the crisis violent or not?); (2) the outcome of the crisis (what was the level of sat-
isfaction of the United States and its opponent with the crisis outcome—was the
United States, the other, both countries or neither satisfied?); and (3) factors that
indicate the overall salience and event magnitude of the crisis for the U.S. deci-
sion makers. These factors included the following variables: system-level (was the
crisis in the dominant system versus being in the regional subsystem or lower?);
proximity of crisis (distant or near?); threat level (high or low?); and the U.S. reac-
tion (did the United States under- or over-respond to the provocation?).
In addition to these factors we wanted to assess whether or not a leader’s accu-
racy (or inaccuracy) in anticipating how events might unfold would influence
the likelihood of belief change (Levy 1994: 292–294). To accomplish this goal,
we first measured each president’s operational code using the VICS system. All
212 B. Gregory Marfleet and Hannah Simpson

verbally delivered public statements made by a president with significant foreign

policy content (50 percent or better) were identified for one month before the
date of crisis onset and for one month following the date of crisis termination
(according to the ICB time variables) and cleaned of any extraneous materials
(press questions, comments by other leaders, etc.).5 We content analyzed this
material by computer and generated pre- and post-crisis VICS indicators for each
of the crisis events. For five of the 60 crises in the sub-sample this effort proved
problematic because of the absence of sufficient data or because a leadership
transition occurred between the onset and termination dates. These crises were
dropped from the analysis leaving an N of 60 cases (120 pre- and post-crisis obser-
vations, or about ten years of speech data). This time period yielded an average of
approximately 400 verbs before and after each crisis.6
The VICS scores for each month were then compared to the values indicated
by TIP, which identifies the subjective game the president perceives as represent-
ing the present international context. The two sets of perceived preferences for
Self and Other together define a game that can be depicted in strategic form and
which delineates the international context as it is currently understood by a U.S.
president.7 Given a particular game, the anticipated outcome of the game can
be derived in accordance with Steven Bram’s (1994) Theory of Moves (TOM).
Since the starting point of the TOM solution can be very important for games
without a single dominant outcome, a clear rule for determining the starting
quadrant was essential. The rule that we adopted for this analysis had two parts: (1)
if no prior crisis had ever occurred between the United States and the opponent
that represented the predominant threat, it was assumed that the starting position
for the game solution was mutual cooperation (CO, CO), and (2) if the United
States had experienced a crisis with this opponent in the past, then the outcome of
the last crisis defined the starting position for the solution of this game.8 For each
of the 60 crises in our data set, we identified the president’s subjective game, the
initial quadrant of that game (based on the above rule), the anticipated outcome
of the game, and the anticipated behavior of the opponent in the crisis.
We also conducted a similar analysis of each president’s perception of the post-
crisis context. From these data we identified two factors. First, we determined
whether the anticipated outcome and behavior matched the actual coding of
the crisis in the ICB data set. For the measure of behavior, we assessed whether
the president expected a cooperation or conflict strategy from the opponent and
compared this expectation to the crisis-triggering action as recorded in the ICB
data. We judged that violent triggering acts were indicative of a propensity for
conflict-laden tactics and that non-violent crisis triggers were an indication of
a more cooperative orientation.9 The president was coded as either accurate or
inaccurate in his pre-crisis beliefs according to whether he anticipated the oppo-
nent’s behavior.
To assess outcomes, we examined the anticipated final quadrant of the subjec-
tive game and compared that result to the ICB variable indicating satisfaction with
 Cognitive Responses by U.S. Presidents to Foreign Policy Crises 213

outcome. The ICB variable in question (outlev) indicates whether: (a) only the
United States, (b) only its opponent, (c) both or (d) neither country was satisfied
by the crisis outcome. We determined that the first condition would be antici-
pated correctly by the president if he expected the ordinal game-payoff for Self
to be either 4 or 3 while the opponent received 2 or less. The second condition
would be anticipated correctly if these payoffs were reversed (2 or less for Self and
3 or greater for Other). The third condition would be anticipated if the president
expected both sides to receive a 3 or better outcome, while the fourth condition
would be correctly predicted if the president believed that both Self and Other
were going to receive a payoff of 2 or less in the game. We experimented with
several systems for assessing accuracy and found the most robust assessment to be
a simple dichotomous measure indicating whether the president’s pre-crisis beliefs
regarding outcome were accurate (1) or not (0).
Second, we developed an indicator from the pre- and post-crisis subjective
games that measured whether there was any difference between them. If the
subjective game stayed the same (indicating no shift in anticipated strategy for Self
or Other, and therefore no major shift in VICS scores for the operational code)
then “no belief change” was recorded. If the perceived preferences and strategy
of Self or Other changed, then we determined that “a belief-changing event” had
occurred. This measurement of the presence or absence of belief change serves as
the dependent variable for the first portion of our analysis.

Crisis Induced Belief Change and Learning

Table 12.1 presents the logistic regression results from four nested models using
the dichotomous presence or absence of belief change variable as the dependent
variable. In Model 1, we introduced the factors related to the salience of a crisis.
These include whether it was a proximate crisis or a high threat crisis, and if it was
in the dominant system. We also included whether the United States under- or
over-reacted by comparing the triggering act and the crisis management response
in terms of level of militarization and violence. If the trigger was non-violent, but
the U.S. response was violent, then over-reaction was indicated. If the reverse was
the case, under-reaction was scored. We hypothesized that a high system-level,
close proximity, high threat and over-reaction would all indicate a highly salient
crisis where belief change (and consequently learning) might occur.
Model 1 explores the relationship between event magnitude and salience and
belief change. Model 2 explores the influence of leader expectations based on
beliefs. The third and fourth models represent “full” (unconstrained) and “best
fit” specifications. The results for Model 1 indicate that “U.S. reaction” and the
“level of threat” were not significant predictors of belief change. Level of threat
was probably not a useful indicator because of the lack of variance among the cases:
most were coded as high. The lack of impact of the U.S. reaction variable may
have resulted from our oversimplified expectations. It may be that under-reaction
214 B. Gregory Marfleet and Hannah Simpson

TABLE 12.1 Binary Logistic Regression Results for Post-Crisis Belief Change

Variable Model 1 Model 2 Model 3 Model 4

Constant –2.00 (1.4) 1.260 (.84) –.684 (1.78) 1.205 (.99)

Dominant System 1.21 (.63)* 1.21 (.65)* 1.04 (.61)*

Proximate –2.93 (1.1)** –2.96(1.14)** –2.80 (1.10)**
High Threat 1.51 (1.3) 1.32 (1.34)
Under Response .797 (.77) –.054 (.95)
Over Response .282 (.88) .835 (.90)

Accurate Expectations –1.61 (.85)* –1.57(1.06) † –1.72 (1.01)*

Regarding Outcome
Accurate Expectations –.015 (.54) .536 (.75)
Regarding Opponent
Behavior

–2LL 65.47 78.64 62.395 64.670

–2LL X 2 17.44 4.272 20.15 18.24
Cox & Snell R2 .252 .069 .29 .262
Naglekerke R2 .337 .092 .387 .350
% Correctly Predicted 75% 61% 75% 74%
N 60 60 60 60

Notes: N = 60 for all models. Dependent variable is presence/absence of change in Op-code derived
subjective games following crisis event. Percent correct prediction by constrained model 53.3%.
Figures represent raw Logistic Regression parameters; standard errors in parentheses. Wald statistic
significance results for coefficients † P < .15 * p < .10, ** P < .05, *** p < .01.

suggests caution in an especially risky or dangerous environment, while over- reac-

tion might suggest a low-risk context where the United States was dominant.
Of the two indicators that were significant in Model 1, the “dominant system”
indicator worked as predicted. Crises in the dominant system—more substan-
tial events involving higher-power opponents (usually the USSR)—were indeed
more likely to generate post-crisis belief shifts. Interestingly, the most powerful
predictor in the model—proximity—worked in the opposite direction of our
expectations. Proximate crises were those that occurred on the same continent.
In this data set these were entirely in Central America and the Caribbean. We
surmised that this result could be a consequence of stronger pre-crisis beliefs stem-
ming from familiarity with the region and/or culturally and historically rooted
perceptions that were unlikely to be changed by any single event. U.S. leaders
also seemed universally indisposed to learn anything from repeated crises in Latin
America perhaps because “the more powerful enjoy larger margins of safety in
dealing with the less powerful and have more to say about which games will be
played and how” (Waltz 1979: 194).10
 Cognitive Responses by U.S. Presidents to Foreign Policy Crises 215

As indicated in Table 12.1, overall model performance was strong with nearly
75 percent of cases correctly predicted. This was an improvement of about 22
percent over the baseline prediction generated by the modal (most common)
outcome. In the second model we introduced the two “accuracy of beliefs” fac-
tors. The first is an indicator of whether the president was accurate in anticipating
the outcome of the crisis; the second, whether the opponent’s behavior con-
formed to the president’s expectations. Both of these measures were derived from
the subjective games and the VICS scores. The results indicate that post-crisis
belief change was most strongly predicted by errors in anticipating outcomes.
Inaccuracy in anticipating the opponent’s behavior was not a significant factor.
Model 2 performs somewhat less well than the model that included only the sali-
ence and magnitude factors. This finding alone suggests that event prominence
may be a more important factor than surprise in generating belief change.
Models 3 and 4 present the results of a fully specified (unconstrained) model
and a best-performing model respectively. Notably, the “accuracy in predic-
tion of outcome” variable remained significant after the inclusion of the two
most influential crisis salience factors. Model 4, though it exhibits slightly lower
predictive accuracy than Model 3, is more parsimonious and displays several
other indications of model performance improvement compared to the other
models (–2LL value, Chi-Squared likelihood change, and pseudo R-Squared
statistics).
In substantive terms the models suggest that for a highly salient (dominant
system), non-proximate crisis where the outcome was anticipated accurately, the
probability of belief change following the crisis was approximately 63 percent; for
similar crises where the outcome was not correctly anticipated, this rises to nearly
90 percent. Figure 12.1 graphically depicts the relationship between the prob-
ability of belief change, crisis characteristics and accuracy in expected outcomes
across the range of possible situations.
The results of this analysis suggest that belief change should occur when crisis
characteristics increase salience and when there is low congruence between pre-
crisis expectations rooted in beliefs and actual outcomes. However, whether this
change can be understood as learning remains an open question. Since learn-
ing often implies improvement, we attempted to assess whether post-crisis belief
change would result in greater accuracy in understanding the environment and
the opponent, or in crafting a better-performing policy in subsequent crises that
involved the same crisis opponent. Because we conceptualize learning as an indi-
vidual-level cognitive phenomenon (not an organizational factor), we restricted
our examination of learning to subsequent crises involving the same U.S. presi-
dent. Based on the VICS indicators and the “source of threat to values” ICB indi-
cator, we assessed for each president in each crisis whether or not belief change
occurred in the previous crisis with the same opponent. Since in the first encoun-
ter with any opponent no prior learning measure was available, we excluded the
first crisis encounter with an opponent from this analysis. This step reduced the
216 B. Gregory Marfleet and Hannah Simpson

0.9

0.8

0.7

0.6

0.5 P

0.4

0.3

0.2

0.1
Inaccurate Expectations
Accurate Expectations 0
Distant, Lower Distant,
Proximate, Lower Proximate,
System Dominant
System Dominant
System
System
Crisis Characteristics

FIGURE 12.1 Probability of Belief Change

N of the study from 60 to only 26 repeated crises within presidencies. This small
sample size precluded the use of multivariate statistical techniques.
Table 12.2 presents the results of four cross-tabulation analyses of crises involv-
ing recurring opponents. Some were preceded by belief change in a prior crisis,
but in others belief change did not occur. We examined three indicators of learn-
ing: (1) whether the president in the current crisis exhibited predictive accuracy
in anticipating outcomes or opponent behavior; (2) whether the crisis outcome
was more likely to be a positive one for the United States (victory or compro-
mise versus stalemate or defeat); (3) whether the United States was more likely
to under- or over-react to a crisis (exhibit non-matching or non-reciprocal crisis
behavior). The table reveals several interesting patterns.
First, there appears to be no relationship between prior belief change and
accuracy in predicting opponent behavior, nor is there a relationship between
belief change in a prior crisis and accuracy of subsequent outcome predictions.
However, accuracy in prediction is uniformly high in both the belief change
and non-change conditions and both are higher than the predictive accuracy in
the original set of 60 crises (the president displayed 88 percent accuracy in these
repeated crises compared to the about 82 percent accuracy in the baseline of all
crises). Recall that this subset of 26 crises only includes those where the presi-
dent has experienced a crisis with this opponent before. This result suggests that
repeated crises lead to higher accuracy regardless of the presence of observed
belief change. Although the sample is small, it may well be the case that our use
of the VICS measures has missed some features of learning.
 Cognitive Responses by U.S. Presidents to Foreign Policy Crises 217

TABLE 12.2 Cross-tabulations of Belief Change with Learning Indicators

Did Belief Change Occur after the Did the President Accurately Anticipate the
Opponent’s
Prior Crisis with this Same Opponent? Behavior in Current Crisis?
No Yes
No 5 7
Yes 5 9
X2 = .097, n.s. tau–b = .061 n.s
Did the President Accurately Anticipate the Current
Crisis Outcome?
No Yes
No 2 10
Yes 1 13
X2 = .574, n.s. tau–b = .149 n.s
Was the Current Crisis Outcome Positive for the U.S.?
No Yes
No 3 9
Yes 2 12
X2 = .487, n.s. tau–b = .136 n.s
Did the U.S. Engage in Reciprocal (Matching) Crisis
Behavior in the Current Crisis?
No Yes
No 6 6
Yes 2 12
X2 = 3.896, p < .049, tau–b = .386, p < .038

While 88 percent predictive accuracy may appear extraordinarily high, we

should note that most U.S. presidents exhibited cooperative belief systems and
average or higher levels of perceived control. This combination implies a gen-
eral belief in the possibility of positive outcomes, which is borne out by the fact
that the United States tends to achieve satisfying outcomes in crisis situations.
According to the ICB data, it achieved victory or compromise in over 75 percent
of crises in the full set and over 80 percent in the repeated crisis set. Together this
combination of optimism (or confidence) and frequent success intersect as high
levels of congruence in prediction. However, the ubiquity of accuracy (and suc-
cess) may help to explain why inaccuracy seemed to have such a strong influence
on belief change. This pattern resonates with the view that learning arises from
deviations from expected events.
The second important pattern to note in Table 12.2 concerns behavioral
change. The presence of more matching behavior in repeated crises with the same
opponent is highly significant. This result parallels Leng’s finding that a move to
reciprocating influence strategies is commonly observed in the second and third
218 B. Gregory Marfleet and Hannah Simpson

episodes in a set of repeated crises (Leng 2000). However, the level at which this
occurs following belief change is especially pronounced. U.S. presidents appear
much more prone to adopt reciprocal crisis management strategies after crisis-
induced changes in beliefs. Overall, in the 26 repeated crises, a shift to matching
behavior occurred 66 percent of the time. When learning did not occur, the
likelihood drops to 50 percent, i.e., as likely as not. When belief change was indi-
cated, it reached almost 86 percent. This feature of the data is notable, because a
simple behavioral model might have generated a positive and statistically signifi-
cant finding regarding changes in behavior toward reciprocity in repeated crises.
However, a simple behavioral model would have been insensitive to the possibil-
ity that this pattern in the data was predominantly generated by those events that
followed a belief-changing episode.

Conclusion
We began this study with an account of the conceptual and measurement problems
associated with the study of learning and suggested that a concerted application of
the operational code construct might help to address these issues. The prospect of
measuring belief change raised some questions about the conditions under which
change might occur and the impact it might have on subsequent behavior and/or
cognition. One question revolved around unanticipated behavior from an oppo-
nent and unexpected outcomes. It appears that a failure to anticipate the result of
an interaction episode may indeed prompt leaders to rethink their beliefs. The
magnitude of the event also seemed to play a role in whether post-event belief
shifts occurred.
With regard to the question of whether belief change following an event makes
an actor less likely to make subsequent errors, our results were mixed. It appears
that experience itself may improve accuracy regardless of any change in beliefs
that we could measure. Whether this constitutes the kind of simple, incremental
learning that the cumulative learning model might expect, or whether it suggests a
rational/informational model is an open question worthy of further investigation.
A second area that remains open to inquiry is reinforcement learning. At this
juncture our operationalization of belief change is not calibrated to detect the
kind of learning that success might generate (for instance, increased intensity or
confidence in beliefs). A future challenge for the operational code community
is to consider how these measures might be used to assess whether beliefs have
become more deeply entrenched. Nevertheless, our findings indicate that belief
change does appear to be associated with behavioral changes in future crises (as
the shift toward more reciprocal, matching crisis-management techniques shows).
While it may be too early to claim that we have successfully identified learning,
we feel confident that the operational code approach provides a productive means
to explore it further.
 Cognitive Responses by U.S. Presidents to Foreign Policy Crises 219

Notes
1 This phrasing of these questions reveals a core ontological question about the locus of
learning. Some researchers contend that learning can occur at the level of the group,
organization or even system, while others view learning as an “axiomatically individual”
process (Stern 1997). Our position echoes Levy (1994) in that while learning may
be evinced at these higher levels, it undoubtedly does occur at the individual level.
Moreover, if organizations are going to learn, it will likely result from a process of
internal change resulting from the updated beliefs of its members.
2 A study of this type might be facilitated by a system whereby we could begin to assess
change in belief intensity and not simply categorical differences.
3 These data and narrative case summaries are available at the following website: http://
www.cidcm.umd.edu/icb/
4 Initially there were 65, but this number was reduced to 60 as a result of speech data
availability discussed below.
5 A one-month temporal window for public statements seemed to be the least arbitrary
way to ensure sufficient data for a quality measurement and we adhered to this rule
even though it resulted in the loss of a few cases (from 65 to 60). We also surmised
that the signal-to-noise problem raised earlier with regard to operational code scores
derived from speech data might be analogous to the one that is present in international
events data (such as the Conflict and Peace Data Bank (COPDAB) and World Events
Interaction Survey (WEIS)) and that our measures would benefit from a similar level of
aggregation. The one-month period we selected follows the example set by Goldstein
and Freeman (1990) in their comprehensive analysis of international events involving
the United States. In the VICS context, a similar window has been employed by
Schafer (1999).
6 Mean N of verbs for the sample of months was 426 with a standard deviation of 392.65.
This standard deviation is substantially inflated by four one-month periods during
which the loquacious President Clinton generated over 1,500 codable verbs! Without
these four months, the mean verb count drops to slightly below 400 with a standard
deviation of about 200.
7 Consult Schafer and Walker (2006), Marfleet and Walker (2006) for a more thorough
explanation of the logic of TIP.
8 In judging whether prior crisis events had occurred we referred to the ICB indicator
of the “primary source of threat to values.” This actor was not always the specific crisis
initiator. This decision reflects the notion that throughout much of the cold war era,
minor crisis actors stood in as proxy opponents for the superpowers and repeated crises
were then seen as a series of contests within the U.S.-Soviet or U.S.-China rivalry and
not with the minor crisis irritant.
9 It might be noted that since we are examining interstate crisis events, which the ICB
data collectors note are potential pre-war episodes, the notion that the opponent might
be cooperative is not appropriate. However, crisis activity can alternatively be viewed
as a type of bargaining behavior where the agent is using provocative acts to attempt
to achieve a beneficial outcome. Avoidance of violence at the outset of a crisis suggests
that the likely objective of the interaction is not conflict itself and that the crisis act is a
means to achieving a more favorable end through a negotiated outcome.
10 Instructive analysis in support of this conclusion comes from Cottam’s (1994) study
of the impact of images on U.S. foreign policy makers in Central America, which
suggests policy intransigence even among less hawkish advisors.
PART IV

Foreign Policy Dynamics

13
DUELING WITH DICTATORS
Explaining the Strategic Interaction Patterns
of U.S. Presidents and Rogue Leaders

Stephen G. Walker and Mark Schafer

Introduction
Are the beliefs of leaders pivotal in explaining and anticipating the processes of
strategic interaction between states? We argue in this chapter that beliefs have al-
ways mattered and are not just agents of change in the post-cold war world (Jervis
1994). Even in the bipolar world of the cold war, the attitudes and beliefs of lead-
ers were important in explaining strategic interactions between the superpowers.
They failed to resolve conflicts that to outside observers were not insurmountable
save for psychological or ideational impediments (Larson 1985, 1997; Lebow and
Stein 1994).
More generally, we subscribe (with important rationalist qualifications) to the
claim shared by the constructivist and decision-making approaches that “anarchy
is what states make of it” (Wendt 1992). We assume that the immediate source
of a state’s actions is the “definition of the situation” in the beliefs of the state’s
leaders, which may or may not accurately represent the actual situation (Snyder et
al. 1962; Brecher et al. 1969). We assume as well that one state’s actions may de-
fine another state’s situation in important respects (Wendt 1999; Lake and Powell
1999b). We shall employ this pair of assumptions in tandem with assumptions
from rational choice theory to inform a model of strategic interactions between
the United States and the rogue states of Iraq and Serbia. Signorino (1999: 279)
defines joint strategic interaction processes as characterized by: (1) the intersub-
jective recognition of each other’s options and goals, and (2) the subsequent in-
terdependent conditioning of behavior.
Our conceptualization of foreign policy behavior is that it is constituted by
the words and deeds carried out on behalf of one state toward other states and
informed by the beliefs and preferences of individuals acting alone or in larger
224 Stephen G. Walker and Mark Schafer

groups within the state (McClelland and Hoggard 1969; Hermann 1971). These
words and deeds display the exercise of power in world politics, as they are de-
signed to establish domination, submission, mutual cooperation, or conflict rela-
tionships regarding issues between states in the international system (McClelland
1966; Morgenthau 1967; Mansbach and Vasquez 1981). It is desirable to analyze
these kinds of behavior and their consequences at different levels of decision, i.e.,
as moves, tactics, and strategies. We are interested in explaining: the intensity of
the words and deeds that constitute a move by one state; the sequence of moves
that constitute a tactic pursued by one state over time; the direction of a tactical
sequence that constitutes a strategy (Snyder and Diesing 1977).

Data Collection and Measurement Procedures

Our general model links beliefs, stimuli, and preferences with foreign policy deci-
sions and political outcomes and is based on Steven Brams’s sequential game the-
ory (Brams 1994). As a rational choice theorist, he prefers initially to assume the
condition of two-sided information in modeling his Theory of Moves (TOM),
which reduces the role of beliefs to mirroring the shared reality of the players. We
recognize this two-sided information assumption as modeling the process of sub-
stantive rationality, in which cognitive processes have an endogenous causal role
in a transparent environment (Simon 1957, 1985). In our application of TOM,
however, we treat this assumption as an empirical question and allow for the
possibility that the actors are playing different “subjective games” (Maoz 1990).
Brams (1994: 53–57) also recognizes this possibility with his model of “anticipa-
tion games.”
The core of our research design is to investigate the attributions of preferences
to self and other by each actor in a strategic interaction episode, compare the na-
ture and degree of overlap between these subjective games, and thereby explain
the outcomes of strategic interactions between the players. Along the way we
should also be able to explain the moves and tactics that implement the strate-
gic choices of each player and determine the power relationships that constitute
outcomes of strategic interaction episodes between the players. The VICS auto-
mated content analysis software parses the text of public statements by a leader
and employs the dictionaries of verbs to code them into indices of the leader’s
operational code beliefs (Schafer 2000; Young 2001; Schafer and Walker 2006).
A sample of public statements for 35 world leaders from different eras and regions
provides a norming group with which to standardize the interpretation of scores
for our subjects.
We theorize that different combinations of the key VICS Indices for Self
(I-1, P-4a) and Other (P-1, P-4b) specify the different preference orders for each
member of a dyad regarding the four outcomes of domination, submission, settle-
ment, and deadlock. The (I-1, P-4a) combination reveals Self’s ranked preferenc-
es while the (P-1, P-4b) combination reveals each actor’s beliefs about the other’s
 Strategic Interaction Patterns of U.S. Presidents and Rogue Leaders 225

ranked preferences. Our theoretical propositions about rank-ordered preferences

for the outcomes of Settle, Submit, Dominate, and Deadlock reflect three as-
sumptions. (1) Actors with cooperative I-1 indices (+) prefer to settle while actors
with conflictual I-1 indices (–) prefer to dominate as their first preference. (2) No
rational actor prefers to submit when the locus of control over historical develop-
ment is relatively symmetrical (P-4a = P-4b). (3) Asymmetries (> or <) in the lo-
cus of control attributed to Self and Other result in different preference orderings
that reflect the imbalances. These assumptions are the basis of propositions in the
Theory of Preferences about Inferences (TIP) reported in Marfleet and Walker
(2006) and discussed at length elsewhere in this volume. The expansion of these
propositions into hypotheses about a set of 2 × 2 games with ordinal preferences
guides our analysis of dyadic interactions in the Persian Gulf and Kosovo conflicts
after the cold war.
Words and deeds directed by each state toward the other in chronological
order form an event data series. We have developed software that partitions an
event series into a series of moves by each state toward the other with each actor’s
moves bounded by the intervening words and deeds of the other. This procedure
yields three basic types of moves directed by each actor toward the other: (1) a
single category of word or deed, (2) a variety of either conflict or cooperation words
or deeds, (3) a mixture of both conflict and cooperation words and deeds. Since
we are interested in the sequence of the moves as part of a strategic interaction
process between a pair of states, we recode each move as either escalatory (E) or
de-escalatory (D) according to the following two rules based on assumptions from
information theory regarding signal detection (McClelland 1968, 1972; Garner
1962). All conflict moves are escalatory, and all cooperative moves are de-escalatory. This
rule applies to the first two basic types of moves identified above. All mixed moves
are coded the opposite of the preceding move by the same actor. If the preceding move is
E, then the mixed move is D; if the preceding move is D, then the mixed move
is E. This rule applies to the third basic type of move above. The application of
these procedures and rules to an event series generates a sequence of escalatory
and de-escalatory moves between a pair of states, which can be analyzed for their
intensity, tactical sequence, and strategic direction.
At the lowest level of decision (moves), we employ statistical regression and
ANOVA models to account for variations in the level of intensity. The level
of intensity for each state’s moves is analyzed by reference to variations in the
direction (+ or –) and intensity of the stimulus provided by the other state’s im-
mediately preceding move and to variations in the philosophical and instrumental
beliefs in each leader’s operational code. We will also build into the model other
factors to control for historical context, e.g., power and interest asymmetries be-
tween states (Waltz 1979; Walt 1987; Bueno de Mesquita 1981; Schweller 1998;
Walker et al. 1999). For extended interactions between states, it is possible to
search for learning effects as a leader’s beliefs and corresponding prescriptions for
cooperation or conflict change or remain the same over time.
226 Stephen G. Walker and Mark Schafer

At higher levels of decision we will use formal models to test the effects of
beliefs, stimulus, and context on tactical sequences of escalatory and de-escalatory
moves between two states leading to a strategic equilibrium that either maintains
or alters the relationship between them. These models are constructed as “sub-
jective games” from the strategic preferences inferred from key indices in each
leader’s operational code. Predictions from cognitivist and rationalist models are
derived for 2 × 2 sequential games with ordinal preferences. The general model of
the 2 × 2 game in normal form appears in Figure 13.1. The purpose of this model
is to delimit the possible sequences of choices by each state at different levels of
decision, which define a strategic interaction episode over time. The range of
choices is defined by the rule of alternating moves from an “initial state” within

d+ e–

CO* ← OTHER → CF**

D D D E E E
Reward Promise Support Oppose Threat Punish
(+3) (+2) (+1) ( –1) (–2) (–3)

D Reward (+3) SELF & OTHER SELF SUBMITS

CO* D Promise (+2) (CO, CO) (CO, CF)

D Support (+1) SETTLE OTHER DOMINATES

d+ ↑

Self

e–
↓
E Oppose (-1) SELF DOMINATES SELF & OTHER

CF** E Threat (-2) (CF, CO) (CF, CF)

E Punish (-3) OTHER SUBMITS DEADLOCK

*CO = COOPERATE in various ways (support, promise, reward) to exercise power in the form of
positive (D = de-escalatory) sanctions: d+ represents de-escalatory moves within the CO category.
**CF = CONFLICT in various ways (oppose, threat, punish) to exercise power in the form of
negative (E = escalatory) sanctions; e– represents escalatory moves within the CF category.

FIGURE 13.1 Levels of Decision and Strategic Interaction in the Normal Form of a
2 × 2 Game
 Strategic Interaction Patterns of U.S. Presidents and Rogue Leaders 227

the 2 × 2 Cooperation (CO) and Conflict (CF) Matrix in Figure 13.1 for Ego
(Self) and Alter (Other), according to the Theory of Moves (TOM) developed
by Brams (1994).
We define formally the initial phase of a strategic interaction episode as four
alternating choices by the two actors, which can be four decisions to “move,”
four decisions to “stay,” or any other combination of “move” and “stay” choices
totaling four. This family of tactical sequences gives each party an opportunity to
exchange information about their preferences to “stay” or “move” from any of
the four cells in the matrix. The initial phase of a strategic interaction episode may
also be the final phase if the party who makes the fifth move in a sequence chooses
“stay.” But if each party continues to choose “move” rather than “stay,” the final
phase of the episode may extend as far as three more alternating moves before
cycling back to the cell in the matrix defined by the intersection of the third and
fourth choices in the initial phase (Marfleet and Walker 2006).
The alternating choices in the initial phase represent an exchange of new in-
formation between the parties about their respective responses to each other’s
choices, and they express tactics of reciprocity, e.g., a sequence of DDDD or
departures from reciprocity, e.g., a DEDE sequence. The sequences of De-escala-
tory and Escalatory moves can also occur within the basic categories of Coopera-
tion and Conflict as well as across them, which are represented in Figure 13.1 as
d+ and e– extensions of the former categories the entire length of the Row and
Column categories in the matrix.
This information allows each party both to make a prediction about the other’s
next move and make a strategic choice either to affirm (choose “stay”) or alter
(choose “move”) their relationship. If they choose the former, then the final
phase and the initial phase end with the fourth choice. If they choose the latter,
then the final episode will extend until one party chooses “stay” or they recycle
back to the “initial state” that began the final phase of the strategic interaction
episode (Brams 1994).

Levels of Decision and Data Analysis Steps

Our goal is to explain strategies, tactics, and moves and thereby explain patterns
of strategic interaction and the resulting international relations between pairs of
states over time. We expect to focus on beliefs, context, and stimuli as explana-
tory variables within each strategic interaction episode. Beliefs may have mirroring
effects in choosing strategies or steering effects in overriding rival strategic choices
predicted from the historical context and the stimulus of other’s behavior in the
strategic interaction episode. We shall also test for interactive effects between
these variables from the rival explanations associated with cognitivist and rational-
ist models of strategic choice and interaction at different levels of decision. Finally,
we will look for learning effects in the form of a change in beliefs leading to a sub-
sequent change in strategies, tactics, and moves by each state.
228 Stephen G. Walker and Mark Schafer

Testing for each of these effects will require us to take a series of steps in our
data analysis at each level of decision: moves, tactics, and strategies. Our main
assumption is that beliefs and moves are consistent with one another. Therefore,
we hypothesize that the direction and intensity of each actor’s moves are gener-
ally consistent with the valence and intensity of the VICS Index for I-1 Approach
to Strategy. Since we will not have independent measurements of I-1 for every
move, our data analysis will regress mean scores for the intensity of moves by
different actors with different I-1 scores within the context of a panel design for
organizing the data analysis (Walker et al. 1999; Walker et al. 2001).
Our analysis of these variables will also test a model of contingent conditions
for interactive effects between beliefs and context specified by combining cogni-
tivist and rationalist models. Condition 1 is the more transparent the information
from the environment, the more influential is the new information. Condition 2
is the greater the receptivity to new information from the environment, the more
influential is the new information. These two conditions hypothesize interaction
effects between environmental features and beliefs in which the influence of new
information is enhanced or suppressed, depending on environmental transparency
and the receptivity of the actor.
Our index of transparency is agreement between the valence of the leader’s
philosophical beliefs and the valence of the stimulus from the environment.
Our index of receptivity is agreement between the valence of the leader’s in-
strumental beliefs and the valence of the stimulus from the environment. Low
transparency and low receptivity exist when the valences do not have the same
signs while valences with the same signs indicate high transparency and high
receptivity conditions. In our analysis of moves, we will also test with a multiple
regression model for the additive effects of different combinations of power and
interests, stimuli, and beliefs to see which variables are more influential when
controlling for the others.
At the tactical and strategic levels of decision the dependent variable shifts from
the mean intensity of a particular set of moves to the sequence and direction of moves
selected by an actor. Our analysis of strategies by each actor takes into account
the perceived preferences for different outcomes inferred from attributions by
each actor to Self and Other. The inferences about preferences are used to con-
struct “subjective games” (Maoz 1990), which are analyzed with the Theory of
Moves (TOM) for sequential game theory (Brams 1994). The subjective games
are games of bounded rationality (Simon 1957, 1985), in which the boundaries of
rational choice are set by the actor’s strategic beliefs about Self and Other.
These beliefs are the basis for attributing ranked preferences for the outcomes
of domination, submission, settlement, and deadlock and then calculating wheth-
er to stay at one of these outcomes or move to another, as shown in Figure 13.2.
Depending upon an actor’s beliefs about the locus of control and each other’s
strategic orientation toward cooperation or conflict, different rankings regarding
these outcomes are attributed to Self and Other. Then predictions of rational
 Strategic Interaction Patterns of U.S. Presidents and Rogue Leaders 229

SUBJECTIVE GAMES BASED ON BELIEFS

Other Other Other Other

D E D E D E D E
D 4,4 2,3 D 4,2 2,4 D 4,4 1,2 D 4,3 1,4
Self Self Self Self
E 1,1 3,2 E 1,1 3,3 E 2,1 3,3 E 2,1 3,2
(I– 1, P– 4a) Self: (+, <) (I–1, P–4a) Self: (+, <) (I–1, P–4a) Self: (+, =) (I–1, P–4a) Self: (+, =)
(P–1, P– 4b) Other: (+, >) (P–1, P–4b) Other: (–, >) (P–1, P–4b) Other: (+, =) (P–1, P–4b) Other: (–, =)

Other Other Other

D E D E D E
D 3,2 2,4 D 3,4 2,3 D 3,3 1,4
Self Self Self
E 4,1 1,3 E 4,1 1,2 E 4,1 2,2
(I–1, P–4a) Self: (–, <) (I–1, P–4a) Self: (–, <) (I–1, P–4a) Self: (–, =)
(P–1, P–4b) Other: (–, >) (P–1, P–4b) Other: (+, >) (P–1, P–4b) Other: (–, =)

STRATEGIC INFERENCES BASED ON COMMUNICATION TACTICS

Tactics: Appease / Bluff / Reward / Deter / Compel / Punish / Exploit / Bully
Sequences: DED EED DDD DEE EDD EEE DDE EDE
Inferences: (+, <) (–, <) (+, =) (+, =) (–, =) (–, =) (+, >) (–, >)

A THEORY OF INFERENCES ABOUT PREFERENCES: SIX PROPOSITIONS

(1) If (+, <), then Settle>Deadlock>Submit>Dominate. (2) If (+, =), then Settle>Deadlock>Dominate>Submit.
(3) If (+, >), then Settle>Dominate>Deadlock>Submit. (4) If (–, <), then Dominate>Settle>Submit>Deadlock.
(5) If (–, =), then Dominate>Settle>Deadlock>Submit. (6) If (–, >),thenDominate>Deadlock>Settle>Submit.

OBJECTIVE GAMES BASED ON NATIONAL INTERESTS & BALANCE OF POWER

Other Other Other Other
D E D E D E D E
D 4,4 2,3 D 4,2 2,4 D 4,4 1,2 D 4,3 1,4
Self Self Self Self
E 1,1 3,2 E 1,1 3,3 E 2,1 3,3 E 2,1 3,2
(NI, BOP) Self: (+, <) (NI, BOP) Self: (+, <) (NI, BOP) Self: (+, =) (NI, BOP) Self: (+, =)
(NI, BOP) Other: (+, >) (NI, BOP) Other: (–, >) (NI, BOP) Other: (+, =) (NI, BOP) Other: (–, =)

Other Other Other

D E D E D E
D 3,2 2,4 D 3,4 2,3 D 3,3 1,4
Self Self Self
E 4,1 1,3 E 4,1 1,2 E 4,1 2,2
(NI, BOP) Self: (–, <) (NI, BOP) Self: (–, <) (NI, BOP) Self: (–, =)
(NI, BOP) Other: (–, >) (NI, BOP) Other: (+, >) (NI, BOP) Other: (–, =)

FIGURE 13.2 Internal and External Focal Points for Inferring Strategic Preferences

strategic choices, defined by the beliefs of the actor, are derived and tested in the
final stage of a strategic interaction episode against the actual outcomes of each
episode.
The subjective games shown in the top part of Figure 13.2 represent subsets
of the 78 different 2 × 2 games with structurally different solution properties
(Rapoport and Guyer 1966). Two of them are in the set of no-conflict games
with a mutually best (4,4) outcome. When both parties rank the same outcome as
the highest, then there is no conflict between them, meaning that the final state
of each game will be the (4,4) outcome. Five of them are in the set of remaining
230 Stephen G. Walker and Mark Schafer

conflict games in which there is a disagreement on the highest ranking outcome.

The nonmyopic equilibrium solutions for the seven subjective games in Figure
13.2 may vary, depending on the initial state and which player has the next move
for a particular strategic interaction episode. With this information we can com-
pare the actual outcome of a given strategic interaction episode and the predicted
outcome of the subjective game defined by the beliefs of the player with the next
move.
We also juxtapose an “objective” game against the “subjective” game for each
strategic interaction episode to see which game predicts the actual outcome and
whether the latter is nested within the former (Maoz 1990; Leng and Walker
1982; Snyder and Diesing 1977). These objective games are derived from balance
of power and balance of threat models associated with realist theories of conflict
dynamics (Bueno de Mesquita 1981; Walt 1987). The roster of objective games
appears in the lower part of Figure 13.2, defined according to the contextual
variables of the distribution of negotiable (+) and non-negotiable (–) interests
and the balance of power (> < or =) between the players. The measures of vital
(non-negotiable) vs. secondary (negotiable) interests are the presence or absence
of involvement of a player’s possessions or alliance commitments regarding oth-
ers’ possessions in a strategic interaction episode while the balance of power is
indexed by the military capabilities of each player (Bueno de Mesquita 1981;
Schweller 1998; Brecher et al. 1988). As with the subjective games based on
beliefs, the objective games based on the balances of interests and power can take
on symmetrical or asymmetrical relationships between the players. Preferences
for domination, submission, settlement, or deadlock in both kinds of games are
ranked according to the six propositions in the Theory of Inferences about Prefer-
ences in Figure 13.2.
To what extent do the subjective beliefs and corresponding preferences in our
subjective cognitivist model correspond with the predictions postulated by objec-
tive rationalist models based on contextual variables? Do these models explain the
actual choices of the actors (and does one perform better than the other)? That
is, do the actors make rational strategic choices as defined either by the external
focal points provided by the balance of power and the balance of interests or by
the internal focal points provided by operational code beliefs? We anticipate that
a within-dyad analysis over time will explain the strategic interaction processes
between members of a dyad that lead to the outcomes of domination, submission,
settlement, and deadlock for strategic interaction episodes. This analysis should
also answer questions about learning effects as the choices and the beliefs of the
actors respond to shifts in power/interests relationships between members of each
dyad, to changes in their respective leaders, and to variations in the outcomes of
strategic interaction episodes over time.
The dynamic version of the strategic interaction model allows us to analyze
learning effects at different levels of decision. In order to clarify the possible re-
lationships between learning and choice, we identify levels of learning as well
 Strategic Interaction Patterns of U.S. Presidents and Rogue Leaders 231

as levels of decision. We also attempt to differentiate between different kinds of

learning as mechanisms of change in the relations between states. Following Levy
(1994), we distinguish between social learning and experiential learning (Walker
et al. 2001). Social learning refers to variations in Self’s behavior in response to
environmental stimuli, e.g., a shift in Other’s behavior from cooperation to con-
flict. Experiential learning refers to significant changes in beliefs that may or may
not lead to behavioral changes (Levy 1994). Social learning effects from the tacti-
cal exchanges in the initial phase of a strategic interaction episode can manifest
the operation of pre-play communication processes. These external processes may
affect different levels of decision directly or influence them indirectly through
changes in beliefs.
Conversely, for several well-known reasons in the literature on cognitive pro-
cesses, beliefs can resist reorganization and continue to influence behavior in spite
of being outdated by new information (Jervis 1976; Rosati 2000). The division
of a strategic interaction episode into two phases allows us to consider the initial
phase as a communication game in which the members of a dyad exchange infor-
mation about each other’s tactics. In the middle part of Figure 13.2 are the com-
munication tactics for possible sequences of escalatory and de-escalatory signals
in the initial phase of a strategic interaction episode and the corresponding infer-
ences about strategic preferences that they signal. We can analyze the experiential
learning effects of changes in beliefs as a causal mechanism to explain shifts in sub-
jective games and corresponding tactical and strategic choices leading to turning
points in dyadic relations. We shall distinguish three levels of experiential learning
in our analysis. They correspond to (a) changes in beliefs about self (simple learn-
ing), (b) changes in beliefs about other (diagnostic learning), and (c) simultaneous
changes in beliefs about both self and other (complex learning).

The Persian Gulf and Kosovo Conflicts

During the first half of the 1990s, two U.S. presidents faced challenges to the emerg-
ing post-cold war order from the Iraqi leader Saddam Hussein and the Serbian leader
Slobodan Milosevic. Both leaders used military force to violate international norms
against ethnic cleansing of minority groups and to expand their territorial rule.
George Bush and Bill Clinton led coalitions that attempted to resolve the ensuing
conflicts. In the end, both U.S. presidents used military force to reverse the gains of
these rogue leaders. We focus on these two cases to test our models and measures of
strategic interaction. The data sources include public statements by the leaders from
the public papers of U.S. presidents and Lexis/Nexis, event data collected by the
authors from the Washington Post for the Persian Gulf conflict, and event data for
the Kosovo conflict from the Balkans event data set collected by Joshua Goldstein
and Jon Pevehouse (1997; see also Goldstein 1995).
The small number of cases carries with it some methodological problems re-
garding selection bias and what we can claim about the results of our analysis
232 Stephen G. Walker and Mark Schafer

(Geddes 1991; King et al. 1994). Our primary methodological purpose here is
to specify candidate theories with a plausibility probe rather than to engage in a
crucial test of their relative merits. In fact, we suspect that they indeed may not
be rivals but are instead likely partners that act as a “theory complex” to offer a
better collective explanation of strategic interaction than any one of them can do
by itself (Eckstein 1975; Laudan 1977; Walker 2003).
Given this methodological goal and our suspicion about complementary rather
than rival relationships between rational choice and cognitive consistency theo-
ries, our analytical task is not simply to insert variables from each theoretical
orientation into an equation and see which ones are more robust predictors. We
are interested in whether they interact with one another or are related in a hier-
archical form under contingent antecedent conditions. Moreover, we want to see
whether they relate to one another differently as the dependent variable changes
across levels of decision and the unit of analysis expands over time from an isolated
move by one state to a sequence of moves between states demarcating a strate-
gic interaction episode. The following analysis begins with an inspection of the
standing causes that act as potential external focal points for rational choice for the
attacker and defender in the local dyads (Iraq-Kuwait and Serbia-Kosovo). Then
we examine whether the external and internal focal points for rational action are
influential once attention shifts to the U.S.–Iraq and U.S.–Serbia dyads following
the invasions of Kuwait and Kosovo by their respective attackers.
The objective game defines the situation in which the attacker decided to es-
calate the dispute with the defender in each of our cases and bring the attention
of the United States to focus on a militarized dispute. In both cases the attacker
and defender have vital interests to protect or achieve, as the focus of each dispute

Defender Defender

CO CF CO CF
CO “2,3” ← 1,4 CO “2,3” → 1,4

Attacker ↓ ↑ Attacker ↑ ↓

CF 4,2 → 3,1 CF 4,2 ← 3,1

Attacker’s Strategy Defender’s Strategy

Initial State for this game is in quotation marks. Preferences of attacker and defender are
ranked from 4 = Highest to 1 = Lowest. Predicted final outcome from “initial state” when
each actor has next move is underlined. Each actor has a choice of either “stay” (→|) or
“move” (→) from the “initial state.” When one actor prefers “move” while the other prefers
“stay,” the actor who prefers “move” takes precedence. The final outcome (4,2) is both a
Nash equilibrium and a non-myopic equilibrium for this game (see Brams 1994).

FIGURE 13.3 Local Attacker-Defender Objective Game in Persian Gulf and Kosovo
Conflicts
 Strategic Interaction Patterns of U.S. Presidents and Rogue Leaders 233

was territorial rule by the attacker vs. independence or local rule by the defender.
The balance of power between attacker and defender favored the former in each
case. We construct the game in Figure 13.3 from Propositions 4 and 6 from the
Theory of Inferences about Preferences (TIP) in Figure 13.2. This game has a
dominant strategy for the attacker who will choose Conflict (CF) no matter what
the defender chooses as a strategy. Given an initial state of negotiations (CO, CO)
between the two parties, the defender’s best strategy is to choose “stay” while the
attacker’s best strategy is to choose “move.” In both cases these predicted choices
are reflected in actual escalation into a militarized dispute and a domination out-
come by the attacker.
At this point the United States faced a decision in each case regarding whether
to intervene or live with the outcome. The United States did not have either ma-
terial possessions as stakes in either conflict or an alliance commitment to defend
the possessions of others; however, the balance of power was in its favor. Propo-
sition 3 specifies its rank order for the different outcomes of settle, dominate,
deadlock, and submit. Proposition 4 specifies the rank order of these outcomes for
a weaker power with possessions as its stakes. The objective game derived from
these propositions defines the situation between the defender and attacker in each
dyad (U.S.–Iraq and U.S.–Serbia) and appears in Figure 13.4.
In this game, the United States is the defender and the other member of the
dyad is the attacker. There is no Nash equilibrium for this game; however, there
is a Pareto optimal equilibrium (4,3) in which the joint gains for both parties
exceed any of the other three outcomes. The attacker clearly has an incentive to
negotiate a settlement with the stronger potential defender. The outcomes in the
sidebar show what will happen if the attacker chooses “stay” at (1,4); the defender
will “move” to (3,2) in order to avoid its worst outcome. According to Brams’s
(1994) Theory of Moves (TOM) for sequential 2 × 2 games, rational players
can achieve a nonmyopic equilibrium of (4,3) through the exchange of credible

Attacker Attacker

CO CF CO CF
↓

CO 4,3 “1,4” CO 4,3 “1,4”

Defender
↓
↓ Defender ↓ ↓
↓

CF 3,2 2,1 CF 3,2 2,1

↓

Defender’s Strategy Attacker’s Strategy

Initial State for this game is in quotation marks. Preferences of defender and attacker are ranked
from 4 = Highest to 1 = Lowest. Predicted final outcome from “initial state” when each actor has
next move is underlined. Each actor has a choice of either “stay” ( ) or “move” ( ) from the
↓

“initial state.” When one actor prefers “move” while the other prefers “stay,” the actor who prefers
“move” takes precedence (Brams 1994).

FIGURE 13.4 Global Defender–Attacker Objective Game in Persian Gulf and Kosovo
Conflicts
234 Stephen G. Walker and Mark Schafer

threats to defect to (3,2) or (1,4), respectively, if a negotiated settlement is not

reached and maintained. However, it did not occur without a resort to arms in
either the Persian Gulf or Kosovo conflicts.
In this realist account there is a symmetrical two-sided information assump-
tion. That is, both players define the situation the same way. Each recognizes
that the defender is stronger than the attacker and there is a shared recognition
of who has possessions (vital interests) as the stakes. Further, both players make
the corresponding inferences about preferences for the outcomes of settlement,
domination, deadlock, and submission attributed to defender and attacker. Fi-
nally, each player is able to calculate the appropriate strategy and corresponding
choices of “move” or “stay” and make the threats necessary to make their choices
effective and establish a “nonmyopic” equilibrium (Morrow 1997; Brams 1994).
While rationalist theories of strategic interaction initially make at least some form
of the two-sided information assumption, cognitivist theories initially view all of
these assumptions as problematic. Information asymmetries and biases in the form
of different beliefs and flawed information-processing skills make it likely that
decision makers will not define the situation the same way, creating information
asymmetries, faulty inferences about preferences, and myopic calculations about
the consequences of alternative choices.
The consequences may be different outcomes than the ones predicted by the
rationalist model’s objective game or at least a less elegant process of reaching the
final outcome. Decision makers may eventually converge on a shared definition
of the situation and reach the predicted outcome, but not before they acquire the
necessary information and alter pre-existing beliefs, i.e., engage in experiential
learning (Snyder and Diesing 1977; Leng and Walker 1982; Levy 1994; Leng
1993). Even if they acquire new information, there is no guarantee that they will
draw the appropriate lessons about ranked preferences and corresponding strate-
gies and tactics leading to an optimum outcome prescribed by game theory (Leng
and Walker 1982; Walker et al. 1984; Leng 2000). These possibilities account for
wars of miscalculation that “nobody wanted,” such as World War I and perhaps
the current war in Iraq based on misperceptions and faulty intelligence (Holsti
1972).
Therefore, it remains an empirical question whether the situations in our two
cases were transparent to the protagonists. If so, then the weaker party in each
dyad should have acquiesced to the power asymmetries represented in the dictum
that in a realist world of power politics, “the weak do what they must, and the
strong do what they want.” The choices facing the Kuwaiti government in the
Persian Gulf conflict were ultimately to submit to Iraqi demands or withdraw
into exile. The same choices confronted the weaker party in the Kosovo conflict.
Similar Hobbesian choices faced the Serbs and Iraqis once the United States de-
cided to try and reverse the occupation of Kuwait and the oppression of Kosovo.
While the final outcomes predicted by the objective games for the two conflicts
are roughly what eventually transpired, the events leading to these outcomes do
 Strategic Interaction Patterns of U.S. Presidents and Rogue Leaders 235

not appear to reflect the paths predicted by these models. In particular, the de-
fenders and attackers in the U.S.–Iraq and U.S.–Serbian dyads did not apparently
make the credible threats or select the appropriate tactics and strategies necessary
to proceed directly from the initial state to a Pareto optimum outcome.
Instead, the two parties in each case moved to a bloody war with numerous
casualties suffered primarily by the attacker before reaching the Pareto outcome
of settlement. Why? A simple answer is to hypothesize that not all leaders are
realists—at least not all of the time. That is, they may make decisions based on
other considerations than the distribution of interests and the balance of power.
They may misperceive the interest and power configurations and/or draw in-
appropriate inferences about preferences and strategies. In order to address this
hypothesis, we turn now to an investigation of each leader in the U.S.–Iraq and
U.S.–Serb dyads to see if their beliefs mirror the power and interest configura-
tions specified in the objective game by realist models of power and interests.

Analysis and Results

We have collected data for each leader from several public statements at different
points in time before and during the two conflicts. We first compare and contrast
below the responsiveness of the leaders in the two conflicts to the external focal
points of power and interests vs. the internal focal points of philosophical and in-
strumental beliefs in selecting the intensities of their moves toward one another.
Next we examine points of overlap between the subjective games in their public
statements with the objective game analyzed above. Then we compare the relative
degree of fit between these two kinds of games and the actual patterns of strate-
gies and tactics during the strategic interaction episodes in the two conflicts. We
conclude with an analysis of learning by these leaders during and across strategic
interaction episodes.
Are moves responses to external or internal focal points? To investigate the im-
pact of moves of the configuration of power and interests in the external environ-
ment vs. the constellation of philosophical and instrumental beliefs of the leader,
we have constructed a regression model with the following indices of external
and internal focal points.

• Move (Y) = the mean of events (actions) by Self that are bounded by events
(actions) by Other. We use the scale for WEIS event data developed by
Goldstein (1992).
• Context (X1) = 0 for Bush and Clinton; 1 for Hussein and Milosovic to
represent the two configurations of power and interests represented by the
defender and attacker in the objective game.
• Stimulus (X2) = the Move (or mean of events) by Other that immediately
precedes Self’s Move. We use the scale for WEIS event data developed by
Goldstein (1992).
236 Stephen G. Walker and Mark Schafer

• Other (X3) = the Philosophical beliefs that Self has about the nature of Oth-
er’s cooperative/conflictual orientation and about Other’s ability to control
historical development. Comparison data are derived from a norming group
of world leaders collected by Walker et al. (2003). The four-position scale
constructed from these indices ranks Other from least cooperative/most con-
flictual (1) to most cooperative/least conflictual (4).
• Self (X4) = the Instrumental beliefs that Self has regarding cooperative/con-
flictual orientation toward Other and about Self’s ability to control historical
development. Comparison data are derived from a norming group of world
leaders collected by Walker et al. (2003). The four-position scale constructed
from these indices ranks Self from least cooperative/most conflictual (1) to
most cooperative/least conflictual (4).

The regression model constructed with these variables hypothesizes with one
exception that each of the independent variables is positively related to the de-
pendent variable, i.e., as each independent variable increases or decreases in value,
there is a corresponding increase or decrease in the dependent variable as follows:
Y = a – bX1 + bX2 + bX3 + bX4. The exception is the Context (X1) variable’s
negative relationship to Move(Y) in which we expect Hussein and Milosovic to
be less cooperative/more conflictual and Bush and Clinton to be more coop-
erative/less conflictual, because the attacker-defender objective games specifies a
more conflictual strategic orientation to the rogues than to the superpowers.
Therefore, the four variables are hypothesized to covary as follows with the
dependent variable Move (Y):

H-1. Context: strong leaders with secondary interests at stake (Bush and Clinton)
are more likely than weak leaders with vital interests at stake (Hussein and
Milosovic) to make a cooperative move.
H-2. Stimulus: the more cooperative the stimulus, the more cooperative the lead-
er’s move.
H-3. Other: the more cooperative the leader’s belief about other, the more coop-
erative the leader’s move.
H-4. Self: the more cooperative the leader’s belief about self, the more coopera-
tive the leader’s move.

The first two independent variables in this model are external focal points in the
political universe independent of the leader while the second two independent
variables are internal focal points in the leader’s belief system.
Our tests of the bivariate directional hypotheses support one of the hypotheses
about external focal points and both of the hypotheses about internal focal points.
The relationship between Stimulus and Move is both positive and significant
(b = .300, SE = .07, R² = .096, p < .001). However, the relationship between
Context and Move has the wrong sign and is not significant (b = .516, SE = .07;
 Strategic Interaction Patterns of U.S. Presidents and Rogue Leaders 237

R² = .006, p = .145). The relationship between beliefs about Other and Move is
positive and significant (b = .002, SE = .24; R² = .073, p < .001), as is the rela-
tionship between beliefs about Self and Move (b = .840, SE = .25; R² = .057, p
= .001). Overall, there is more and stronger support for the influence of internal
focal points associated with cognitive models than for the external focal points
associated with rationalist models.
A multiple regression analysis incorporating all of these variables reveals that the
intensity of each move is related significantly (p = < .05) to the stimulus variable
and to the leader’s beliefs about Other’s cooperative or conflictual orientation in
the political universe. The estimations for the constant a and the b-values for each
independent variable in the equation are as follows, with the one-tailed signifi-
cance test in parentheses: Y = –4.84 + .498X1(p = .167) + .255X2( p < .001) +
.541X3 (p = .028) + .232X4(p = .229). One external focal point (Stimulus) and
one internal focal point (beliefs about Other) are significant while the other two
variables (Context and beliefs about Self) are not. The internal focal point X3 is a
little over twice as influential as the external focal point X2 (.541 vs. .255).
To test our transparency and receptivity hypotheses about the conditional ef-
fects of the stimulus and beliefs on actions, we create two new variables, X2,X3
Transparency and X2,X4 Receptivity, where the values for each X variable are
the valences (+ or –) of their respective indices. This results in four values for the
new variables of Transparency and Receptivity: +,+; +,–; –,+; –,–. The mixed
values (+, – and –, +) for each new variable—representing the conditions of low
transparency and low receptivity—are grouped as one level while the +,+ and
–,– values are treated as separate levels in the following three-level ANOVAs.
The mean score of Y (Move) is the dependent variable. According to our hypoth-
eses, we should expect to find:

• Transparency Hypothesis: the more consistent the stimulus is with the leader’s
philosophical beliefs about the environment, the more transparent the infor-
mation from the environment to the leader and the more likely the leader’s
move will correspond to the stimulus. Consistent valences (–,– or +,+) of
stimulus and philosophical beliefs define the environment as more transpar-
ent and result in the most hostile or least hostile move scores, respectively,
while inconsistent or mixed valences define a non-transparent environment
and result in middling move scores.
• Receptivity Hypothesis: The more consistent the stimulus is with the leader’s
instrumental beliefs predisposing him toward one kind of action or another,
the more receptive the leader to information in the environment and more
likely the leader’s move will correspond to the stimulus. When consistent va-
lences (–,– or +,+) exist for stimulus and instrumental beliefs, leaders should
be more receptive to information and result in most hostile and least hostile
move scores, respectively, while inconsistent or mixed valences define low
receptivity and result in middling move scores.
238 Stephen G. Walker and Mark Schafer

TABLE 13.1 Three-Level ANOVA Results for the Transparency and Receptivity Hypotheses

TRANSPARENCY CONDITIONS*
High Negative Transparency Low Transparency High Positive Transparency

Means** –3.611 –2.615 –.613

N 60 81 41

* High Negative = (– –) level; Low = (+– and –+) level; High Positive = (++) level.

RECEPTIVITY CONDITIONS*
High Negative Receptivity Low Receptivity High Positive
Receptivity

Means** –4.127 –2.392 –1.083

N 43 96 43

* High Negative = (– –) level; Low = (+– and –+) level; High Positive = (++) level.

Table 13.1 shows that the means for the (–,– ) and (+,+) levels vary in opposite
directions away from the grand mean while the means for the mixed conditions
fall between the other two for both the Transparency and Receptivity variables.
The results support both of the hypotheses about the conditional effects of in-
teraction between the external and internal focal points of stimulus and different
kinds of beliefs. For Transparency, when the environment is transparent and neg-
ative, move scores are the most hostile; when it is transparent and positive, move
scores are least hostile. When the environment is not transparent, scores are mixed
and between the other two. The differences among these three mean scores,
(F (179,2) = 11.19, p < .001), are statistically significant. For Receptivity, when
the leader is receptive to the other’s move, he tends to respond in kind with more
hostile moves in response to negative stimuli and less hostile moves in response
to positive stimuli. When the leader is not receptive, he tends to respond with
a more mixed level of moves. The differences among these three mean scores,
(F (172,2) = 10.07, p < .001), are also statistically significant.
Do beliefs have mirroring effects? Is the objective game between defender and
attacker mirrored by the subjective games in the beliefs systems of the leaders
during the Persian Gulf and Kosovo conflicts? The results of our analysis show
that there is virtually no overlap between the objective game postulated by realist
balance of power and national interest premises and the subjective games in the
public statements of the leaders of each state. Out of 76 definitions of the situ-
ation retrieved from their public statements, only ten subjective games (13.1%)
mirrored the objective attacker-defender game for the Persian Gulf and Kosovo
conflicts. All except one were found in President Bush’s statements early in the
Persian Gulf conflict; the remaining occurrence was in President Clinton’s rheto-
ric during the penultimate episode of the Kosovo conflict. There was no episode
 Strategic Interaction Patterns of U.S. Presidents and Rogue Leaders 239

in either conflict in which the two leaders shared a common definition of the
situation as the attacker–defender game. The overall pattern across the two con-
flicts reveals little transparency between the objective game of power and inter-
ests and the subjective games of the decision makers in these conflicts. A shared
subjective definition of the situation between the leaders in each conflict is also
relatively rare.
Therefore, the mirroring effects of beliefs during the two conflicts were not
common occurrences. Instead, the leaders in these conflicts articulated 11 dif-
ferent definitions of the situation. The most frequent one for the Persian Gulf
conflict was the Bully–Appeasement game in which a Great Power confronted a
Small Power. It occurred in 16 (61.5%) of the 26 episodes for either Bush or Hus-
sein. However, it was a shared definition of the situation for Bush and Hussein
during only 4 (15.4%) of the 26 strategic interaction episodes during the conflict.
The most frequent definition of the situation between a Great Power and a Small
Power for the Kosovo conflict was the Stag Hunt-Appeasement game found in
Clinton’s public statements for 10 (83.3%) of the 12 strategic interaction episodes.
Milosovic tended to define the situation as a Prisoner’s Dilemma-Assurance game
in 9 (75%) out of the 12 strategic interaction episodes.
Do beliefs have steering effects? Given these results we would expect a relative-
ly poor fit between the strategies of these leaders and the predictions of the objec-
tive game of power and interests. A tally of the 38 strategic interaction episodes
across the two conflicts shows that the objective game’s predictions of the next
move and final outcome during the final stage of each episode were correct no
more than 50% of the time. Nineteen out of 38 next moves (50%) fit the model’s
predictions and only 16 out of the 38 outcomes (42.1%) were consistent with the
model. Do the beliefs of the leaders predict the strategies of the leaders any better?
To answer this question we examine the subjective games for the leader with the
next move toward a final outcome within each strategic interaction episode. Self
can either choose to “stay” at the initial outcome generated by the exchange of
tactics between each party in the conflict or choose “move” in the expectation
that Other will respond according to the logic of the preference ordering at-
tributed to Other in Self’s subjective game. This expectation may or may not be
“realistic,” depending on whether Self and Other share the same subjective game
as their definition of the situation.
The results of our analysis show that the leader with the next move chooses
“stay” or “move” in accord with the logic of his subjective game 29 (76.3%) out
of 38 strategic interaction episodes across the Persian Gulf and Kosovo conflicts.
Twenty-six (68.4%) of the 36 final outcomes were consistent with the predicted
outcome of the subjective game for the leader with the next move. This latter
figure does not demonstrate that the other player for a particular strategic inter-
action episode shared the leader’s definition of the situation. It merely indicates
that the leader’s subjective game is a better predictor of Other’s strategy than the
objective game of power and interests. The reason for the more modest inference
240 Stephen G. Walker and Mark Schafer

TABLE 13.2 Predictive Powers of Objective and Subjective Games

% Correct Next Move* % Correct Outcomes*

Predictions Objective Subjective %d** Objective Subjective %d**

Conflicts
Both 50.0 (38) 76.3 (38) +26.3 42.1 (38) 68.4 (38) +26.3
Persian Gulf 53.8 (26) 80.8 (26) +27.0 42.3 (26) 69.2 (26) +26.9
Kosovo 41.6 (12) 75.0 (12) +33.4 33.3 (12) 75.0 (12) +41.7
Leaders
Bush 57.1 (21) 76.2 (21) +19.1 52.3 (21) 61.9 (21) +9.6
Hussein 40.0 (5) 100.0 (5) +60.0 20.0 (5) 100.0 (5) +80.0
Clinton 33.3 (3) 33.3 (3) 0.0 00.0 (3) 33.3 (3) +33.3
Milosovic 44.4 (9) 77.7 (9) +33.3 44.4 (9) 77.7 (9) +33.3
Means 45.7% 74.2% 28.4% 33.5% 69.3% +35.9%

* Numbers of cases are in parentheses.

** %d = Percent difference between subjective and objective games.

is that different subjective games can prescribe the same strategy for Other from
a given initial outcome.
Overall, a comparison in Table 13.2 of subjective games defined by a leader’s
beliefs with the Attacker–Defender game of power and interests shows the former
to be a much better predictor than the latter of both a leader’s next move and
the final outcome. This generalization holds true for the four leaders as a group
across the two conflicts and for each leader within each conflict with only one
exception. President Clinton’s subjective game does not predict his strategy any
better than the objective game of power and interests for the Kosovo conflict. On
average, the predictions of the leaders’ subjective games are 28.5% more accurate
(74.2% vs. 45.7%) for the next move and are 35.9% more consistent (69.3% vs.
33.5%) with the final outcome no matter whether one is interested in the fit for
both conflicts, just one conflict, or a particular leader.
What do leaders learn? To answer this question, we examine the fit between
the tactical signals exchanged between the two players and their solution strate-
gies to end each strategic interaction episode. We also look at how players alter
their beliefs over successive interaction episodes. These two research questions tap
different kinds of learning identified by Levy (1994) who distinguishes between
social and experiential learning. Social learning is altering behavior in response
to changes in the stimulus. If one player’s tactics signal an appeasement strategy
while the other player’s signals a deterrence strategy, does the player with the op-
portunity to end a strategic interaction episode act in accord with the definition
of the situation contained in the exchange of signals? Experiential learning is the
alteration of beliefs over successive interaction episodes.
There are three levels of experiential learning: simple, diagnostic, and complex.
Simple learning is the alteration of Self’s instrumental beliefs so as to change the
 Strategic Interaction Patterns of U.S. Presidents and Rogue Leaders 241

TABLE 13.3 Predictive Powers of Subjective and Communications Games

Percent Correct Next Move

Predictions Communications Subjective %d**

Conflicts
Both 76.3 (38) 76.3 (38) 0.0
Persian Gulf 84.6 (26) 80.8 (26) –3.8
Kosovo 58.3 (12) 75.0 (12) +16.7
Bush 87.5 (21) 76.2 (21) –11.3
Hussein 80.0 (5) 100.0 (5) +20.0
Clinton 0.0 (3) 33.3 (3) +33.3
Milosovic 77.7 (9) 77.7 (9) 0.0
Means 66.3% 74.2% +7.8%

* Numbers of cases are in parentheses.

** %d = Percent difference between subjective and communications games.

leader’s preference rankings for the outcomes of settlement, domination, sub-

mission, or deadlock. Diagnostic learning is the alteration of Self’s philosophi-
cal beliefs about Other’s preference rankings for these same outcomes. Complex
learning is the simultaneous altering of both sets of preference rankings. Any of
these three changes may lead to a different strategy attributed to Self and Other in
the leader’s subjective game with a corresponding change in the leader’s decisions
during the conflict.
Evidence to answer the first question appears in Table 13.3. It shows that the
predicted strategy from the communications game defined by the exchange of
tactical signals between the players is highly consistent with the solution strategy
by the player with the next move. The percentages of correct predictions are very
close to the ones in Table 13.2 based on the subjective games in each player’s
beliefs. They are equal in two instances (for both conflicts and for Milosovic),
higher in two others (for the Persian Gulf cases and for Bush), and lower in the
remaining three instances (for the Kosovo conflict, Hussein, and Clinton). On av-
erage, the subjective games model is 7.8% more accurate (74.2% vs. 66.3%). The
small overall difference between the two models is primarily due to overlapping
predictions by each model for the same interaction episodes.
In 24 (63.2%) of the episodes the two models made the same prediction for the
player with the next move. In 10 games the two models disagreed and the player
with the next move followed one of the predictions. The subjective games model
failed to predict Bush’s next move in 5 of those games where he followed the
prediction of the communication games model. In the other 5 games where the
two models disagreed, Bush followed the subjective games model in 3 cases while
Hussein followed the subjective games model in the remaining 2 cases. Finally,
in 4 episodes the two models agreed on their predictions, but the player with the
242 Stephen G. Walker and Mark Schafer

next move chose a different strategy (Milosovic in 2 episodes and Clinton in 2

episodes).
The overdetermination evident in the two models probably indicates a couple
of forces at work. One is a tendency of leaders to send tactical signals consistent
with their subjective beliefs about strategies. Another is the principle of equifinal-
ity in which different models may predict the same outcome. For example, any
2 × 2 no-conflict game ranks mutual cooperation (settlement) as the most pre-
ferred outcome by both players and predicts it as the final outcome no matter how
much the two players differ in their preferences for the remaining outcomes of
deadlock or domination by one player at the expense of the other. The same pos-
sibility of equifinality is present to a lesser degree in mixed-motive games when
only one player ranks cooperation as the highest outcome and in conflict games
where both players rank domination or deadlock as their highest outcome.
Turning to whether leaders alter their beliefs over successive interaction epi-
sodes, thereby exhibiting experiential learning, Table 13.4 indicates that two
of the four leaders tended to exhibit complex learning. Clinton and Milosovic
were exceptions to this generalization, indicating that complex learning was
55% more (%d = .55) characteristic of the Persian Gulf conflict than the Kosovo
conflict. When leaders did alter their beliefs, they were almost equally likely to
shift their own preferences toward conflictual or cooperative outcomes. The
same is true for changes in their assignment of preferences for conflictual or
cooperative outcomes to others. The distributions in Table 13.5 are based on
changes in each leader’s preference rankings for Self and Other along a six-
position scale formed by the Theory of Preferences in Figure 13.2. The six
propositions in the theory anchor the following scale from most coopera-
tive/least conflictual to least cooperative/most conflictual strategic preferences
arrayed as follows: the Appeasement (+,<) strategy in Proposition (1), the
Assurance strategy (+,=) in Proposition (2), the Stag Hunt strategy (+,>) in
Proposition (3), the Chicken strategy (−,<) in Proposition (4), the Prisoner’s

TABLE 13.4 Levels of Experiential Learning by Leaders

Regional Conflict
Learning* Persian Gulf Kosovo

Simple or Diagnostic 20% (4) 75% (3)

Complex 80% (16) 25% (1)
Totals 100% (20) 100% (4)
%d = .55**

* Simple learning is a change in self attributions regarding self’s belief about the most effective strategic
orientation. Diagnostic learning is a change in other attributions regarding self’s belief about other’s
strategic orientation. Complex learning is a change in both self and other attributions regarding self’s
beliefs about the most effective strategic orientation and other’s strategic orientation.
** % difference between the types of learning in the two conflicts.
 Strategic Interaction Patterns of U.S. Presidents and Rogue Leaders 243

TABLE 13.5 Direction of Learning by Leaders

Learning
Change of Direction Simple Diagnostic
Toward Conflict 48% (10) 48% (10)
Toward Cooperation 52% (11) 52% (11)
Totals 100% (21) 100% (21)
%d = .00

%d = Percent difference between simple and diagnostic learning.

Dilemma strategy (−,+) in Proposition (5), and the Deadlock strategy (−,>) in
Proposition (6). Depending on the leader’s initial position on the scale, a shift
in one direction or the other becomes a move toward a cooperative strategy or
a move toward a conflictual strategy attributed to Self or Other.

Conclusion
In formal models of strategic interaction by rational choice theorists, there are
two approaches designed to provide shared definitions of the situation under
the condition of incomplete information. Both operate so that both parties can
choose rationally, i.e., make their respective choices based on their own prefer-
ences for different outcomes as well as predictions about the other party’s pref-
erences for the same set of outcomes. One approach emphasizes external focal
points, such as power distributions or the rules of institutions, which provide a
context for inferring each other’s preferences from common knowledge about
their respective power positions or institutional roles (Lake and Powell 1999b).
The other approach is to allow parties to communicate and learn before they
select their final strategies. This communication can take the form of preliminary
moves that frame the main choices within a strategic interaction episode (O’Neill
1999; Snyder and Diesing 1977). Either mechanism implies at least a tacit com-
munication and coordination game between the parties in a strategic interaction
situation prior to the game of cooperation or conflict that determines the political
outcome and equilibrium between them (Chwe 2001; O’Neill 1999: appendix
A; Schelling 1960). In the absence of such external mechanisms, each player can
make choices based on internal beliefs about self and other that are not necessarily
shared with one another. All three mechanisms—external focal points, commu-
nication, and beliefs—are possible frames of reference that make decisions rational
in the absence of certainty, i.e., without perfect and direct information about each
other’s preferences.
We contend that the debate between rationalists and cognitivists in world
politics is precisely about which of these mechanisms work either together or in
opposition to explain foreign policy choices and the evolution of dyadic relations
244 Stephen G. Walker and Mark Schafer

over time. Our analysis in this chapter is designed to investigate the interaction
of these mechanisms as they shaped decisions for war and peace in world politics.
We find that external focal points are the least effective mechanisms for explain-
ing strategic interactions between protagonists in the Persian Gulf and Kosovo
conflicts. The effects of power and interests are virtually non-existent. Direct
effects on decisions by leaders were absent, as were indirect mirroring effects on
decisions through the beliefs of leaders. The effect of the other player’s move
as an external focal point in our model is statistically significant at the level of
moves, though conditioned strongly by interaction with the beliefs of the decision
makers. The internal focal points of beliefs about self and other are strongly re-
lated to a leader’s actions in theoretically consistent ways at all levels of decision:
moves, tactics, and strategies. On average, these beliefs offer slightly better predic-
tions of strategies selected by leaders than the tactical signals exchanged between
them. It also appears that at higher levels of decision the internal focal points of
beliefs and the external focal points of tactical signals work together to explain
strategic interactions in military conflict situations.
14
BINARY ROLE THEORY
Reducing Uncertainty and Managing
Complexity in Foreign Policy Analysis

Stephen G. Walker

Introduction
This chapter contains a review of the four major concepts employed in the opera-
tional code analysis research program, which were elaborated and applied in previ-
ous chapters to analyze foreign policy decision-making and learning processes in
world politics. These concepts provide the foundations for constructing a binary
role theory of uncertainty reduction and management, which is the central dynamic
in world politics. This theory describes how and explains why the two-worlds-of-
interaction proposition operates to integrate the world of mental events (beliefs) at
the micro level of foreign policy analysis with the world of social events (behaviors)
at the macro level of international systems. Binary role theory’s solutions to the
agent-structure and level-of-analysis problems are presented in this chapter as oper-
ational code analyses of the dynamics of symbolic and strategic interaction between
systems of thought and action within dyads, which are then extended and expanded
beyond the strategic dyad to more complex ensembles of world politics.
We began our study of foreign policy analysis and international relations with
a distinction between beliefs (thoughts) and behavior (actions) and suggested that
these two concepts represent phenomena for investigation at two levels of analy-
sis: the mental world of beliefs constituted by the thoughts of agents and the
social world of behavior constituted by the actions of agents interacting with one
another as dyads, triads, or even larger ensembles.
We also introduced operational code analysis as a neobehavioral systems
approach to the study of interactions within each world and between them.
We argued that the core pair of concepts in operational code analysis, rationality
and power, retrieve and specify the same phenomena for analysis in each world
through the medium of texts—micro-belief inventories in the public and private
246 Stephen G. Walker

statements by political leaders and macro-event chronologies in the observa-

tions by journalists and historians of the interactions between states. The concept
of rationality specifies the cognitive processes of symbolic interaction (decision
making and learning) in the world of mental events while the concept of power
specifies the social processes of strategic interaction (cooperation and conflict) in
the world of behavioral events.
Both analyses of mental and social events are captured in the same language
of texts, which means that the two events are potentially isomorphic, i.e., they
describe the same phenomena in the same form within two different contexts
(mental and social). Even though their external shapes may differ, their internal
logical structures may be the same (Hanrieder 1967). Texts are a medium and
can also be considered a third world of language in addition to the other two
worlds of thoughts and actions (Popper 1994: 5–10; Habermas 1984; Der Derian
and Shapiro1989; Weber 1995). Unlike some authors, we treat this third world
as transparent and endogenous in relation to the other two worlds. It is simply a
medium or channel within which processes of communication and control occur
between the mental world of belief systems and the social world of behavioral
systems. The language symbols that occur in a belief inventory or behavioral
chronology are commensurate with one another, which make their contents iso-
morphic within this medium of language. That is, both thoughts and actions
are reduced to information (patterns of events) in the form of linguistic symbols
(Deutsch 1966, 1968; Shannon and Weaver 1964; Garner 1962).
The isomorphic character of beliefs and behavior within the medium of texts
makes it possible to assess whether and to what extent they correspond with one
another. This possibility is addressed by the third pair of concepts in our analysis:
congruence and balance. Congruence simply means that the relationships between
the elements of two systems match up. Balance is a special case of congruence and
refers to whether the valence for the exercise of power within the world of beliefs
or within the world of behavior matches up with the same sign (+ or −) between
the two agents Ego and Alter, conditioned by whether the distribution of power
matches up symmetrically with the same sign (= or ≠) between them. In more
complex relationships between dyads or larger ensembles, all of the relationships
are congruent when they match up with one another in more complex patterns,
i.e., when positive and negative valences converge or diverge as predicted within
and between the world of beliefs and the world of actions.
The mathematical symbols associated with the concepts of balance and con-
gruence represent these relationships and processes isomorphically, i.e., in the
same symbolic form, and thereby amalgamate the world of beliefs and the world
of behavior. These binary symbols collectively define and represent a fourth pair
of concepts that operate in both worlds: entropy and uncertainty. Entropy refers
to the possible states for each of these systems. A state of maximum entropy is
also assumed to be totally unorganized without structure in a state of maximum
uncertainty, i.e., a state in which any one of the possible states of organization
 Binary Role Theory and Foreign Policy Analysis 247

is equally probable in each system (Carroll 2010; Shannon and Weaver 1964;
Milburn and Billings 1972). The concepts of rationality and power in the context
of a systems analysis of beliefs and behavior operate to specify the level of relative
entropy and uncertainty of mental and social events in these two worlds. The
following effects reduce the entropy in these systems by making some states
(mental and social events) more probable (less uncertain) than others:

• Steering Effects: a reduction in entropy in the world of beliefs via the cogni-
tive process of thinking rationally leads to a reduction of entropy in the world
of behavior.
• Learning Effects: a reduction in entropy in the world of behavior via the
behavioral process of exercising power leads to a reduction of entropy in the
world of beliefs.
• Congruence Effects: the levels of entropy in the two worlds tend to converge
(balance) over time.

Binary Role Theory

Role theory as a theory of uncertainty reduction is capable of describing and
explaining the dynamics that produce these three effects. Its basic concepts include
role conception as a mental element, role enactment as a behavioral element, and a
structural element that defines role and counter-role as subsystems of two agents (Ego
and Alter) interacting to form a social system (dyad). These dyads are conceptual-
ized as role sets in which role and counter-role are congruent; they match up so that
they instantiate emergent properties, i.e., properties of a system that do not exist in
the absence of its components and organization (Cummins 1983: 14–27), and are
coupled as social systems with shared role expectations. These cognitive and social
systems interact as subsystems within a larger system or an environment of con-
straints and opportunities, which limit or facilitate the selection and enactment of
possible roles and counter-roles. These possibilities are represented in role theory
by the concept of role demands (Walker 1992, 2010).
All of these concepts are inter-related in a process represented by the concept of
role location, in which agents seek to match role and counter-role in their respec-
tive repertoires of role sets (Sarbin and Allen 1968; Walker 1979). This process
may be complicated by conditions of role strain (difficulty in role enactment), role
competition (difficulty in choosing a role), and role conflict (determining which of
two contradictory roles is called for in a situation). In a binary role theory these
features (role conceptions, role expectations, role enactments, and role demands)
are represented as congruent with one another by the binary symbols of positive
(+) and negative (−) valences or symmetrical (=) and asymmetrical (≠) valences
(Walker 1992, 2010).
We argued in Part I of this book that the emergent properties of peace and war
between large-scale social units such as nation-states should be understood as the
248 Stephen G. Walker

product of interactions between them under the constraints imposed by the distri-
butions of power and interests between them and their respective choices to exer-
cise power as positive (+) or negative (−) sanctions. In this realist model of world
politics the power and interest distributions were also conceptualized as binary,
i.e., as symmetrical or asymmetrical distributions of power and vital or secondary
national interests. The binary quality of these variables make them compatible as
well with a role theory of uncertainty reduction that contains binary role concep-
tions, role expectations, and role enactments as either positive (+) or negative (−)
strategic orientations in the operational codes of political leaders regarding the
exercise of power and the symmetric (=) or asymmetric (≠) role demands regard-
ing the distribution of control between Self and Other.
The following steps in constructing a binary role theory of uncertainty reduc-
tion that incorporates both the macroscopic variables of a state’s power and inter-
ests and the microscopic variables of a leader’s strategic beliefs about Self and
Other also allow for the analysis of the three kinds of effects (steering, learning,
and convergence), which we previously identified as occurring in the interactions
between the cognitive system of beliefs within each agent and the social system of
behaviors between each agent.

The Elements of Binary Role Theory

We begin with the equation from Chapter 4 that integrates the state-level, macro-
interactions between Ego and Alter with the individual-level, micro-interactions
between Self and Other: [E | A] =[SE |A | OE | A]. The emergent interactions in the
macro-level systems of large-scale state dyads [E | A] can ultimately be explained
by the interactions in the micro-level belief systems of Self and Other as leaders
within each dyad regarding SE |A and OE | A. The dyadic strategic interaction game
between the two states Ego and Alter is solved over time by the intersection of the
subjective games of Self [SE|A] and Other [OE|A] as the leaders of the two states.
The subjective games of Self and Other are defined by their respective opera-
tional code beliefs that assign a role to Ego and a counter-role to Alter, respec-
tively, which may be altered by new information during a strategic interaction
episode between Ego and Alter. In the absence of past operational code beliefs,
incoming information from the social world may be the sole basis for the con-
struction of new roles for Ego and Alter in the mental world of each agent. This
information may be received as cues from the actions of the other member of the
dyad or internalized as information from the environment about the distributions
of power and interests between Ego and Alter.
We shall refer to these two sources of role conceptions and role expectations
as boundary or scope conditions for employing role theory based on a particu-
lar solution to the agent-structure problem in international relations. In the first
instance, the old information stored as operational code beliefs suggests the Past
Hypothesis as a structure-oriented solution to the agent-structure problem (Carroll
 Binary Role Theory and Foreign Policy Analysis 249

2010: 176–178; Wendt 1999: 328–331; Kydd 2005: 34–36). According to this
hypothesis, the entropy in the belief systems of each agent has been reduced by
past interactions with each other or by interactions with other agents so that roles
for Ego and Alter have already been constructed for Ego and Alter to enact in
their present encounter.
However, if new information is the basis for roles assigned to Ego and Alter by
Self and Other, then the First Encounter Hypothesis is relevant as a solution to
the agent-structure problem (Wendt 1999: 187–190). According to this hypoth-
esis, the entropy in the belief systems of each agent is reduced by their present
interactions so that roles for Ego and Alter are constructed de novo for their pres-
ent encounter. A general relativity solution to the agent-structure problem is a
structurationist solution that allows both hypotheses to work: depending on the
location in space and time of the agents and the outside observer, one solution is
bracketed while the other is employed (see Wendt 1987; Archer 1988; Carlnaes
1992). Role theory is flexible enough to employ a structurationist solution by
varying the dimensions of space and time with the static and dynamic metrics of
behavior introduced in Part I of this book.

The Past Hypothesis

Under the simplest version of the Past Hypothesis condition, agency is bracketed
and the prevailing assumption is that roles for Ego and Alter are already structured
and assigned by old information shared by both agents as shown in the static game
matrix in Figure 14.1. The binary roles for Ego and Alter in the static matrix are
either Cooperation (CO+) or Conflict (CF–), and the power relations between
Ego and Alter are specified by the Theory of Inferences (TIP) as asymmetrical (≠)
with Ego as the stronger (>) agent and Alter as the weaker (<) agent. The initial
state in quotation marks is the status quo at a particular point in time. The final state
underlined is also the nonmyopic equilibrium (NME) according to the Theory of
Moves (TOM) for this game (Brams 1994). The prediction is for no intermediate
states with the initial and final states being identical as both Ego and Alter choose
“stay” rather than “move.”
Given these assumptions of the Past Hypothesis as the initial state and the
accompanying environmental conditions below, each player shares a common
definition of the situation as located in the CO,CO cell marked by the exercise of
power by both Ego and Alter as positive sanctions in the context of an asymmetri-
cal distribution of power between them. Ego has no incentive to choose “move”
from (4,3), which is Ego’s highest-ranked outcome. The constraint specified by
the power distribution specifies that Alter also has no incentive to choose “move”;
the final state would be a return to (4,3) as shown in the following analysis.
If Alter chooses “move” from (4,3) to (1,4), then Ego will choose “move”
to (2,1), Alter will then choose “move” to (3,2), and Ego will move to (4,3)
unless either Ego changes its role from CO to CF with a corresponding change
250 Stephen G. Walker

Alter (A−,<) D,D ↑ Alter ↑ D,E

← D E →
CO CF
1,4 (e–)
“4,3”
CO
D S5 S1 → S2

Ego (E+,>) Origin ↑ (d+)

(d+) ↑ ↓ (e )
Ego −
“S0”
CF 3,2 2,1

← E S4 ← S3→
(d+)
E,D ↓ ↓ E,E

Role Enactment Process Role Location Process

Static Game Matrix Dynamic Game Matrix
(Past Hypothesis) (First Encounter Hypothesis)
TIP & TOM Inferences CUE Inferences
If Ego (+,>), then TIP is CO, CO > CF, CO > CF, CF > CO, CF. Ego: If (DEE), then CUE is Deter (+,=).
If Alter (–,<), then TIP is CO, CF > CO, CO > CF, CO > CF, CF. Alter: If (EED), then CUE is Bluff (–,<).
If initial state is (4,3), then Ego’s TOM choice is “Stay.” Step0. No Ego, Alterinteraction.
If initial state is (4,3), then Alter’s TOM choice is “Stay.” Step1. Ego: De-escalation (d+).
Step2. Alter: Escalation (e−).
Step3. Ego: Escalation (e−).
Step4. Alter: De-escalation (d+).
Step5. Ego: De-escalation (d+).

CO = Cooperation; CF = Conflict; D = De-escalatory; E = Escalatory. Initial state is in

quotation marks and final outcome is underlined. Brams Non-Myyopic Equilibrium (NME)
is in bold. S1 (Step1)…S5 (Step5) constitute Stage I, and subsequent decisions to stay at
D,D or move to D,E would constitute Stage II or a repeated play of the game constructed in Stage I.

FIGURE 14.1 Static and Dynamic Matrix Analyses of Role Location Processes

in preference rankings for the different outcomes from settlement > domination
> deadlock > submission to domination > deadlock > settlement > submission.
The outcome of the strategic interaction episode would be (CF, CO) in this case
with this outcome (domination) ranked highest for Ego and being the NME for
the new asymmetrical power game.
In order to insure that (4,3) remains a stable NME in this static game, TOM
specifies that Ego should employ threat power in communicating with Alter to
deter the latter’s defection from (4,3). If exogenous events should change the
credibility of Ego’s deterrent threat, then it is possible for Alter to choose “move.”
Domestic conditions inside Ego or an asymmetry in interests between Ego and
Alter might create role strain (increased difficulty) for Ego to shift from coop-
eration to conflict in this example. Alter’s role is CF−, which is consistent with
Alter’s vital (worth fighting for) interests while Ego’s role is CO+, which is con-
sistent with Ego’s secondary (not worth fighting for) interests.
These differences in role demands between the two agents may make
Alter willing to risk a “move” from (4,3) and undermine the credibility of Ego’s
 Binary Role Theory and Foreign Policy Analysis 251

deterrent threat. Simultaneous involvement by Ego in other dyads, i.e., role compe-
tition, may also dilute its ability to make a credible deterrent threat in its relations
with Alter and lead to a “move” from (4,3) by the weaker member of the dyad.
Finally, Ego may experience role conflict in its relations with Alter, as an escalation
in conflict between them may jeopardize ongoing cooperation between them or
with other agents, which would be costly to Ego if Alter were the source of a
valuable commodity or was a valuable ally in the conduct of Ego’s relations with
other states in the region.
To illustrate these possible dynamics, consider that the static game in Figure
14.1 is the same one as the Global Defender-Attacker game between Iraq
and the United States in Figure 13.4. In this case role demands leading to role
strain, role competition, and role conflict may have operated to make it pos-
sible and plausible for a relatively weak state (Iraq) to choose “move” and invade
Kuwait in August 1990 even though it would risk changing Iraq’s cooperative
relations with the United States. Relative uncertainty about the definition of
the situation in this case in the form of role demands on the United States may
have led Iraq to underestimate the resolve of the United States and the cred-
ibility of its deterrent position in this situation. These role demands included
distractions and disruptions in U.S foreign policy accompanying the end of the
cold war, American aspirations to continue good relations with Iraq as a
source of oil for the United States and as a buffer against Iranian expansion in
the Persian Gulf, and insufficient specificity in the U.S. Ambassador to Iraq’s
response to Bagdad’s inquiry about the posture of the United States in the event
of an escalation in tensions between Iraq and Kuwait (Freedman and Karsh
1993).

The First Encounter Hypothesis

In contrast, the dynamic game matrix in Figure 14.1 shows the logic of the First
Encounter Hypothesis as the prevailing “initial state” condition for interactions
between Ego and Alter. This condition may apply when Ego first interacts: (a) with
a new state (Alter) in the international system (Thies 2001); (b) when Ego and Alter
first interact following a change of leaders in either Ego or Alter (Hermann 1980);
(c) when Ego and Alter interact following an exogenous shock that “resets” their
previous relations in a new social and environmental context, e.g., the end of the
cold war in the case of the United States (Ego) and (Alter) Russia or U.S. (Ego)
relations with a variety of other states (Alters) following the 9/11 terrorist attacks
by Al Qaeda (Hermann 1990); (d) when Ego and Alter expand their relations into
a new area of interaction, e.g., nuclear proliferation or global warming, to cite two
relatively new issue areas in the history of international relations.
It is also possible simply to invoke the First Encounter Hypothesis based on the
a-historical assumption that the shadow of the past does not count in the present.
If so, then Ego and Alter are located at the origin of the dynamic matrix in Figure
252 Stephen G. Walker

14.1 for every encounter between them—at least in the sense that each one can
make the next move in any direction spatially through time. This possibility is
depicted in Figure 14.1 as S0, a location representing the “initial state” as no
previous interaction between Ego and Alter located at the origin of the dynamic
matrix. S1 represents Ego’s next move vertically in a de-escalatory (d+) direction
and is followed by Alter’s horizontal escalatory (e−) move. These two moves
locate Ego and Alter at S2, which is a (d+, e−) state. Ego’s next move is escalatory
(e−) so that Ego and Alter are located at S3, which is an (e−, e−) state. Alter’s subse-
quent move to S4 is de-escalatory (d+) to an (e−, d+) state. When Ego de-escalates
(d+) to S5, the relationship between Ego and Alter is (d+, d+).
Steps S1 through S4 construct intermediate states leading to S5 as the final
state of a role location process with a dyadic “initial state” beginning at S2. These
steps in the role location process create a brief history of time in which there is
interaction between Ego and Alter. Their initial moves signal their respective
binary role conceptions regarding the exercise of power as positive (d+) or nega-
tive (e−), and the subsequent exchange of moves signals their exercise of power
as conditional or unconditional. The sequences of cue patterns in Figure 14.1 are
for Ego (+,−,−) Deter and for Alter (−,−,+) Bluff, in which the first and third
move in each sequence is by the agent in question and the intervening move is
by the other agent for a total of four moves (two by each agent). Ego then moves
to S5, which is the final state of (D,D) predicted by TIP and TOM for a dyad
constructed by this exchange of cues.
The analysis of the Persian Gulf conflict in Chapter 13 between the United
States and Iraq employed the First Encounter Hypothesis as well as the Past
Hypothesis in analyzing the sequences of moves between the two protagonists.
The fit between the Past Hypothesis was slightly better than for the First Encounter
Hypothesis as an assumption when the “past” is defined as the world of beliefs
for each agent’s political leaders (George H.W. Bush and Saddam Hussein). The
analysis showed that each leader’s operational code beliefs prescribed different
roles for the United States and Iraq, which had steering effects on their sequences
of moves. The two leaders also experienced learning effects as they updated their
respective belief systems in an exchange of cues, and there were converging con-
gruence effects generated as final outcomes in the series of strategic interaction
episodes between the United States and Iraq.
The rules of binary role theory provide an understanding of how role concep-
tions and role expectations interact, conditioned by role demands, to describe
and explain continuity and change in the states that exist within the mental world
of beliefs and the social world of action. We can depict these worlds as systems
of thought and action with a field diagram, as in Figure 14.2, which shows that
the status quo location of interactions between Ego and Alter in the social world
of political behavior is in the region of mutual cooperation (CO,CO) and can
move dynamically in a de-escalatory (d+) or escalatory direction (e−) as a series of
alternating moves by Ego or Alter. Depending on the magnitude of these moves
 Binary Role Theory and Foreign Policy Analysis 253

CO CF
Barrier Barrier
D+ W+ ← Alter → W– D–

D+ [d+,d+] [d+,e–]

CO Barrier [Ego | Alter]

W+ [e–,d+] [e–,e–]

↑
Ego
↓

W–
CF Barrier

D–

W = Words and D = Deeds; CO = Cooperation and CF = Conflict;

e– = Escalation and d+ = De-escalation.

FIGURE 14.2 A Dynamic Political Field Diagram of Role Location

the new locations for this dyad may remain in the CO,CO quadrant or move
into one of the other four quadrants in the field diagram. It is possible to traverse
these quadrants either in a non-linear path of moves around the perimeters of
the rectangle in this figure or diagonally through the medium of prior (pre-play)
communication inside the boundaries of the rectangle formed by the Word games
within the field.
Ego and Alter can debate and reach an agreement within the boundaries of this
rectangle, which could function as a short cut or worm hole to move their loca-
tion in any direction from the origin by redefining their respective roles and cor-
responding strategic interaction game. Communication can also take the form of
a mix of Words and Deeds as signals and moves, which could steer the two agents
into or out of this rectangle over time. It is possible as well for the two agents to
orbit indefinitely within the rectangle of Word games or in one of the quadrants
of Deed games. These Word games and Deed games are located in regions in the
political field in which the distinction between Words and Deeds act as barriers
or thresholds within the even more basic distinctions between Cooperation and
Conflict domains. Shifts within these regions are moves within a 2 × 2 game
matrix defined either by these distinctions or by moves across regions of Words
or Deeds in which the game matrix is redefined within the broader categories of
Cooperation and Conflict.
254 Stephen G. Walker

Looking back at the clockwise set of moves S0 through S5 within the dynamic
matrix in Figure 14.1 and locating its pattern within Figure 14.2, these steps could
in principle begin at the origin and traverse (a) only the region within the rect-
angle of Word space in Figure 14.2; (b) move in and out of this space around the
boundaries of the game matrix in Figure 14.2; (c) traverse only the region outside
the rectangle in one of the quadrants of the game matrix in Figure 14.2, e.g., the
location of Ego and Alter marked in Figure 14.2; or (d) begin with a clockwise
pattern within the rectangle, e.g., [e−,e−] in Figure 14.2, and break out diagonally
at some point by mutual agreement into the Deeds space within one of the quad-
rants in the game matrix for Figure 14.2. The rules of binary role theory govern-
ing these patterned sequences of communications, interactions, and background
conditions between Ego and Alter are TOM, TIP, and CUE, which are listed in
Figure 14.3 (Walker 2010).
Clearly, the possible patterns of interactions between Ego and Alter through
time and across the space defined by the political field diagram are non-linear,
complex, and vary in scope, rate, and direction. Therefore, role theory needs to
be flexible and complex enough in order to describe and explain these possibili-
ties. These sets of rules govern the various relationships between Ego and Alter,
which are illustrated in Figure 14.4 as a series of steps, states, and stages. The
Congruent Role Dyad (−,−) in Figure 14.4 is a microscopic version of the [Ego,

TOM TIP CUE

Rules of Play Rules of Inference Rules of Learning
(Theory of Moves) (Theory of Preferences) (Theory of Cues)

1. Either player can choose to 1. If Appease, then (+, <): Settle > 1. If (+, –, +), then (DED) Appease.
‘move’ or ‘stay’ from the Deadlock > Submit > Dominate.
‘initial state.’ The player who
moves is P1.
2. Player 2 (P2) can respond by 2. If Reward or Deter, then (+, =): 2. If (+, +, +), then (DDD) Reward;
choosing ‘move’ or ‘stay.’ Settle > Deadlock > Dominate if (+, –, –), then (DEE) Deter.
> Submit.
3. Alternating responses continue 3. If Exploit, then (+, >): Settle > 3. If (+, +, –), then (DDE) Exploit.
until a Player with the next choice Dominate > Deadlock > Submit.
chooses “stay,” terminating the
game at a final outcome state.
4. A player will not move from 4. If Bluff, then (–, <): Dominate 4. If (–, –, +), then (EED) Bluff.
the initial state if itleads to a less- > Settle > Submit > Deadlock.
preferred final state as the final
outcome.
5. A player will not move from an 5. If. Compel or Punish, then (–, =): 5. If (–, +, +), then (EDD) Compel;
initial state if it leads to a less- Dominate > Settle > Deadlock > if (–, –, –), then (EEE) Punish.
preferred final state or cycles play Submit.
to the initial state as the final
outcome.
6. If one player chooses ‘stay’ and 6. If Bully, then (–, >): Dominate > 6. If (–, +, –), then (EDE) Bully.
the other player chooses ‘move’ Deadlock > Settle > Submit.
from the initial state, then the
choice of ‘move’ overrides ‘stay.’

TOM Source: Brams 1994; TIP Source: Marfleet and Walker 2006; CUE Source: Walker 2010.

FIGURE 14.3 Rules of Symbolic and Strategic Interaction in Binary Role Theory
 Binary Role Theory and Foreign Policy Analysis 255

A Congruent Static Role Dyad*

Self’s Role −,− Other’s Role
Conception/Expectation Expectation/Conception

RME RTE RTA RMA

Role (Ego) − − Counter-Role (Alter)

− −
Role Enactment Role Enactment
−,−

Steps and Stages in the Construction of a Dynamic Role Dyad

− −
+
RC/RX RX/RC RC/RX RX/RC RC/RX RX/RC
+ −
+ + + − − −
+ − − −
+ −
+
RE RE RE
RE RE RE
− −
Stage I…Step 1 (Ego Moves) Step 2 (Alter Moves) Step 3 (Ego Moves)

−
RC/RX RX/RC
Alter
– D+ E− CUE Stage I Sequences
– − Ego: {E(+) A(−) E(−)} Deter
– − D+ 4,3 1,4
Alter: {A(–) E(−) A(−)} Punish
–
Ego
↑
E− 2,1← “3,2”* TOM Stage II Predictions
RE RE Ego chooses “stay” at (3,2)
−
TIP Inferred Game Alter chooses “move” to (2,1)
Step 4 (Alter Stays). . . . . . . Stage II Initial State. Ego chooses “move” to (4,3)
Alter chooses “stay” at (4,3).

*Congruent is either (+,+) or (–,–) relations between Role (Ego) and Counter-Role (Alter) in a Dyad
in which the power distribution is symmetrical (=,=). RC is Role Conception, RX is Role Expectation,
RE is Role Enactment. RM is Role Making and RT is Role Taking. E is Ego and A is Alter. A positive cue (+)
is de-escalatory and a negative cue (−) is escalatory. Stage I is Steps 1–4, and Stage II occurs if the outcome
at Step 4 is not a Nonmyopic Equilibrium (NME). Initial state is in quotation marks and final state is underlined.
Nash equilibrium is asterisked and Non-myopic equilibrium is in bold.

FIGURE 14.4 Role Location and Strategic Interaction Processes for a Role Dyad

Alter] location in Figure 14.2, magnified statically as a role set of relationships in

a behavioral field and expanded dynamically as a series of steps and stages to show
the processes of de-escalation (d−) and escalation (e−) in play to alter or sustain
the role location of the [Ego, Alter] dyad. Let us consider in more detail first the
static, magnified view of this dyad in space and then the dynamic, expanded view
of the same dyad through time.
According to the equation cited earlier in this chapter, [E | A] = [SE |A | OE | A ],
the Congruent Role Dyad [Ego | Alter] in Figure 14.4 is ultimately equal to the
respective role conceptions and role enactments plus the shared role expectations
of Self and Other, namely, the political leaders of the agents occupying the role
256 Stephen G. Walker

of Ego and the counter-role of Other. Applied to the macroscopic interactions of

large social systems, such as states in international relations, [E |A] is generated,
sustained, and altered by the microscopic foreign policy role conceptions, role
enactments, and role expectations of the leaders of those states. A static view of
these relationships is represented in the top half of Figure 14.4, which shows bal-
anced (−, −) relationships between the role conceptions and role expectations of
Self and Other and between their respective role enactments.
The role location process that generates this congruent role dyad is represented
by the arrows from role conceptions to role enactments and from role enactments
to role expectations. The role-making (RM) processes by Self and Other generate
negative (–) valences from role conception to role enactment for Self and Other.
When each agent enacts a negative role, the role-taking process (RT) internalizes
the transmission of this information as role expectations for each agent. The role-
making processes represent the steering effects of beliefs on behavior while the
role-taking processes represent the learning effects of behavior on beliefs.
The immediate sources of role conceptions and role expectations for Ego
and Alter are the operational code beliefs of Self and Other and the cues (role
enactment behaviors) exchanged between these leaders. The interaction patterns
among these agents are also conditioned by various role demands, particularly the
distributions of power and interests between the occupants of the Ego (role) and
Alter (counter-role) positions. If there is convergence between these elements of
role theory from past interactions (the Past Hypothesis), then the rules of TOM
are sufficient to regulate the present interactions depicted in the top half of Figure
14.4 (Malici 2006, 2008).
However, if there is divergence between the subjective games of Ego and Alter
occupants specified by their respective operational code beliefs, then the intersec-
tion of these subjective games is modeled dynamically by TIP and TOM. If there
is also no past set of beliefs to represent [Ego, Alter] interactions (or at least no rel-
evant role set), then the rules of CUE (Cognitive Uncertainty Exchange) govern
the construction of a brief symbolic interaction history of their interactions and
specify the outcome of the strategic interaction episode between Ego and Alter.
It is also possible for Ego or Alter to engage in an altercasting strategy in this brief
history, which may or may not reconstitute the relevant role set (Malici 2006,
2008). In that case the interaction episode extends beyond four moves from Stage
I into Stage II with a further exchange of cues until the party with the next move
chooses “stay” or until the next move would return them to the “initial state”
constructed at the end of Stage I.
These possibilities are illustrated in the dynamic series of steps in the bottom
half of Figure 14.4, which show how the static Congruent Role Dyad above
them could have been constituted. Step 1 shows how Ego initiates a role-mak-
ing process and a role enactment process with positive (+) cooperation behavior,
and Alter experiences a role-taking process in which Ego’s positive (+) behavior
enacts a de-escalatory (d+) cue for Alter. Step 2 shows Alter’s response, which is
 Binary Role Theory and Foreign Policy Analysis 257

to make and enact a negative (−) conflict role, and Ego experiences a role-taking
process in which Alter’s enactment is a negative (e−) cue for Ego. Step 3 shows
that Alter’s response to Ego in Step 2 was an instance of successful altercasting in
which a negative cue from Alter prompted a change in Ego’s role conception and
enactment from a positive to a negative role as Ego learned via the role-taking
process that Alter did not want to cooperate. Step 4 shows that Alter responded to
this change by re-enforcing its previous negative conflict role.
These four steps were generated in Stage I of the Congruent Role Dyad in the
upper half of Figure 14.4, but there is more to this brief history through time. The
Past Hypothesis in conjunction with the rules for CUE and TIP says that Ego and
Alter have constructed a subjective game that has mutual escalation (e−, e−) as an
initial state; however, it is predicted by TOM’s rules to have mutual de-escala-
tion (d+,d+) as a final state of nonmyopic equilibrium. This prediction extends
the present from the immediate past into the immediate future. The sequences of
cues on which this game was constructed in Stage I show a Deter (+, −, −) tactic
by Ego that signals a strategic preference for settlement > deadlock > domination
> submission and a Punish (−, −, −) tactic by Alter that signals a strategic prefer-
ence for domination > settlement > deadlock > submission. These signals and the
subjective game constructed from TIP’s rules are shown in the lower right corner
of Figure 14.4 along with TOM’s predictions that Alter will choose “move” from
an initial state of (3, 2) at the end of Stage I/beginning of Stage II to (2, 1) and
Ego will respond by choosing “move” to (4, 3) as a final state and the nonmyopic
equilibrium for this strategic interaction episode.

Solving Two-Sided and Three-Sided Interaction Problems

The entropy of possible states for TOM, TIP, and CUE working together as a
general theory to understand the interaction of rationality and power between
the world of beliefs and the world of behavior is defined by considering (a) the
binary combinations of positive and negative cues specified as tactics by CUE, (b)
the binary inferences about the exercise and distribution of power between Ego
and Alter specified by TIP, and (c) the possible interaction sequences between
Ego and Alter specified by TOM derived for the possible games constructed from
CUE and TIP. These combinations of attributions and games and their links were
identified earlier in Figure 2.4 and Figures 3.4 and 3.5. The TOM solutions (final
states) for the intersection of the possible pairs of subjective games between Ego
and Alter with four different initial states are in Table A.2 at the end of this book.
This entropy matrix shows that there are twelve possible subjective games for Ego
paired against the same twelve possible subjective games for Alter for a total of 576
possible outcomes.
When these two sets of subjective games are played against one another from
their four possible initial states to a final state, the 576 possible outcomes represent
the state of maximum entropy for the theoretical universe defined by binary role
258 Stephen G. Walker

theory according to the rules of CUE, TIP, and TOM (Marfleet and Walker
2006). These outcomes identify possible locations for the [Ego | Alter ] dyad as
social events in the political field diagram and how those states can be generated
from different initial states through intermediate states to final states according
to the rules of TIP and TOM and the Past Hypothesis. In the absence of a past
record of interactions, CUE’s rules can be employed to specify the subjective
games for Ego and Alter from their First Encounter interactions.
In making the transition from a focus on dyadic systems considered separately
to their aggregation as triads and larger ensembles of agents that constitute more
complex systems with third-party and n-party consequences, we confront directly
the limits associated with moving immediately from determinate solutions of
two-body problems to indeterminate solutions of three-body problems and indi-
rectly the extension of this limiting condition to n-body problems (Goldstein et
al. 2002; Barrow-Green 1997). As Kenneth Waltz (1979: 192–193) notes, “The
three-body problem has yet to be solved by physicists. Can political scientists or
policymakers hope to do better charting the courses of three or more interacting
states? Cases that lie between the simple interaction of two entities and the sta-
tistically predictable interactions of very many are the most difficult to unravel.”
Robert Jervis (1997: 3–4) comments further, “We all know that social life and
politics constitute systems and that many outcomes are the unintended conse-
quence of complex interactions ... [but] … Few if any realms of human conduct
are completely determined at the systems level. Actors choices are crucial and …
are influenced by beliefs about how the system operates.”
Are there either basic rules for solving these more complex problems or strate-
gies for applying the present rules of TOM, TIP, and CUE for solving two-body
problems to three-body and n-body problems? It turns out that there are rules for
triadic, three-body problems as well as dyadic, two-body problems (Jervis 1997;
Harary 1961, 1969; Hartmann 1982; Burt 1992). With these additional rules and
a strategy of disaggregation from the larger, complex system of n-bodies to a series
of dyads and triads, it is possible to reduce and analyze these relations for analysis
with role theory’s triadic distinctions among Ego (1st Party), Alter (2nd Party),
and Audience (3rd Party) and the dyadic game theory models in this book (Thies
2010; Newell and Simon 1972; Voss 1998).
The overall strategy for employing these rules is consistent with the analytical
strategies used in theories of complexity associated with complex adaptive systems
analysis (Axelrod and Cohen 1999; Mitchell 2009). The link between the binary
role theory employed to reduce uncertainty (entropy) and the complex adaptive
systems approach is that the analysis of complexity is the other side of the coin
in relation to the analysis of entropy. Agents interact and organize into patterns
constituting a system of thought or action. Their patterns become complexly
organized instead of random or nonexistent, and information (patterned relations)
replaces entropy as a description of their relations. These patterns can be very
complex, varying in their static configurations and dynamics of change along such
 Binary Role Theory and Foreign Policy Analysis 259

dimensions of complexity as hierarchy and decomposability (Simon 1973; Jervis

1997; Mitchell 2009).
According to the logic of systems analysis in the physical universe, the ultimate
final state of any system is maximum entropy—a state in which the elements of
the system no longer interact, thereby making any pattern possible and equally
likely in the future. Physicists believe that this will be the final state of our uni-
verse, which began with a “big bang” from a state of low entropy and low com-
plexity (simplicity) and has since evolved toward higher complexity in the near
term and high entropy in the long term (Hawking with Mlodinow 2005; Carroll
2010). The laws of thermodynamics governing the physical universe predict that
this trend will eventually end with maximum entropy and low complexity. High
complexity in this account is an intermediate state in the life cycle of the universe.
Although time as we experience it only runs forward in one direction from the
big bang, it is possible that a “multiverse” preceded our universe, i.e., a state of
higher entropy among the elements of a physical system interacted to form our
known universe with a “big bang” whose effects form an “arrow of time” from
the past through the present into the future (Carroll 2010).
Physicists posit the effects of “free energy” to account for the existence of local
complexity in a physical universe that holistically is evolving toward maximum
entropy. Free energy comes from environmental sources or interactions within
a larger system that can provide usable energy for elements of a local subsystem.
If the subsystem is actually a closed system so that there is no impact from the
environment or from membership in a larger system, it will follow the Second
Law of Thermodynamics and eventually reach maximum entropy in which the
patterned interaction among its elements disappears and the system disintegrates.
However, if a system is an open system, then the trend toward entropy may be
offset so long as the elements of the local subsystem can extract energy from the
larger system or the environment to replace what is used to constitute and sustain
the patterned interactions that make those elements a complex system (Carroll
2010: 194–201; Mitchell 2009).
Living biological systems are open systems that have the capacity to extract free
energy from their environments and convert it into usable forms that maintain
life. By extension it is possible to think of social systems of interaction among
living biological systems, which operate the same way. If the agents in biologi-
cal and social systems have this capacity plus the capacity to repair and maintain
the mechanisms (processes) that perform these tasks in the context of a changing
environment, then in principle they can postpone their disintegration. Explaining
the development of these capacities in the world of biological systems is the focus
of evolutionary theory, which accounts for them as complex adaptive systems
that are organized sufficiently into patterns of interaction that vary from simple
to complex, depending on the environments in which they operate, historical
contingencies, biological constraints, and capacities for self-organization (Carroll
2010; Wright 2000; Mitchell 2009; Popper 1982).
260 Stephen G. Walker

According to Wright (2000: 244–245), the random mutation and evolution

of the biological elements in the social systems of animals and humans has led to
inherited traits that define their capacity for strategic interaction and the exercise
of power based on rational thought processes. The path to this point in human
evolution begins theoretically with random interactions among lower (less com-
plex) systems of matter and energy in which entropy is reduced by higher pat-
terns of complexity in the form of emergent properties from varying degrees of
non-random interaction, which becomes the organization of a system of interest
(Axelrod and Cohen 1997: 15–20).
This coupling into patterns is what Wright (2000: 4–5) calls a “non-zero-
sum” relationship by which he means (252) that the elements of the system are
now linked “in the same boat” in a regular rather than a random pattern. He
notes further (337–343, 4–5) that both social scientists and evolutionary biolo-
gists divide such patterns in social systems and biological systems into zero-sum
and non-zero-sum systems or games. In the artificial world of game theory, a
zero-sum game is one in which one player’s loss is the other player’s gain (thus
summing to zero) while the outcomes of a non-zero-sum game include outcomes
in which one player’s gain is not necessarily the other’s loss (thus not summing to
zero). This emergent property makes non-zero-sum games more complex than
zero-sum games and a source of potential energy. In Wright’s words:
Non zero-sumness is a kind of potential. Like what physicists call “potential
energy,” it can be tapped or not tapped, depending on how people behave.
But there’s a difference. When you tap potential energy—when you, say,
nudge a bowling ball off a cliff—you’ve reduced the amount of potential
energy in the world. Non-zero-sumness, in contrast, is self-regenerating. To
realize non-zero-sumness—to turn the potential into positive sums—often
creates even more potential, more non-zero-sumness. That is the reason that
the world once boasted only a handful of bacteria and today features IBM,
Coca-Cola, and the United States.
(Wright 2000: 339)
Depending on the distribution of information (and therefore the distribution of
power) among the players and their information-processing skills in social and
biological systems, the outcomes of a non-zero-sum game can be win-win or lose-
lose as well as win-lose or lose-win. A living system like human societies persists
by the process of natural selection so long as its pattern of complexity enables it to
adapt to its social and biological environment. This dynamic is the one underlying
the theory of biological evolution of living systems and its extension to the cultural
evolution of social systems. Such patterns reduce the entropy (possible states) of
the living system with a particular configuration (an actual state), and uncertainty is
thereby replaced by complexity (Wright 2000: 345–347).
Therefore, the order (organization) in a system of interest may be the result of
interactions between elements (agents) in a closed system or the product of forces
 Binary Role Theory and Foreign Policy Analysis 261

introducing energy into an open system, i.e., one that is open to the introduction
of energy (power) in the form of information from the system’s environment.
Where the system of interest ends and the environment begins is a matter of
choice by the complex adaptive systems analyst. For example, it is possible to
focus on a dyad as the system of interest and relegate other agents in the larger
system to the environment. It is also possible to identify three dyads with overlap-
ping membership and focus on this triad as the system of interest and by extension
expand this focus to overlapping triads.
Role theory is simultaneously a theory of uncertainty and a theory of complex-
ity in which the analytical goal is to reduce uncertainty and thereby introduce
complexity, i.e., replace random variety with ordered selection in a complex
adaptive social system through the process of role location. Once the boundar-
ies of the system of interest are specified, there are three sets of rules that act as
theories of complexity to understand, explain, and predict the interaction pat-
terns. One set of rules are the rules of congruence associated with Cognitive
Uncertainty Exchange (CUE) in the role location process while the other sets
of rules are the rules of inference and interaction associated with the Theory of
Inferences about Preferences (TIP) and the Theory of Moves (TOM). The role
location rules governing the exchange of cues are already suited to the analysis
of triads and have been applied previously by social psychologists to cognitive
balancing processes in the world of beliefs (Sarbin and Allen 1968). These logical
rules are also the basis for triadic analyses of coalitions in the world of events sum-
marized by the Arab proverb: “the friend of my friend is my friend; the enemy of
my friend is my enemy; the enemy of my enemy is my friend” (Jervis 1997: 211;
see also Schweller 1998; Zinnes 1970; Caplow 1968). Harary (1961, 1969) has
formalized them into general mathematical rules with signed graph theory.
In order to understand how the process of role location works in dyadic and
triadic systems with these rules, let us consider the concept of role location more
deeply. It is a concept with social, temporal, and spatial dimensions. It refers to
the relationship of mutual awareness between Ego and Alter expressed by the role
location process modeled earlier in Figure 14.4 in which their respective cogni-
tive and behavioral worlds converge and diverge socially through time within the
spatial context of the distribution of power between them. The social dimension
is distinctive for living systems such as human societies while the other dimensions
of location are shared with non-living systems, which lack the emergent property
of mutual awareness that make living systems social (Searle 1995, 1998, 2001).
The role location process of four moves by Ego and Alter in binary role the-
ory communicates awareness along all three of these dimensions: temporally by
the alternating exchange of cues; socially by the initial pair of positive or nega-
tive cues in the sequence that signal cooperation or conflict; and spatially by the
subsequent pair of cues in the sequence that signal symmetrical or asymmetrical
power between Ego and Alter. The same dimensions are embedded in the cogni-
tive architecture of the operational codes of Self and Other who engage in the
262 Stephen G. Walker

role location process and constitute Ego and Alter as a role set. The positive (+)
and negative (–) valences for the key beliefs of (I-1) Strategic Orientation of Self
and (P-1) Orientations of Other in the Political Universe, plus (P-4) Historical
Control of Self and Other as symmetrical (=) or not (≠) symbolize the same kind
of information in the cognitive belief systems of each agent in a social system.
When we speak of cognitive or behavioral systems as operating within a field,
therefore, we are conceptualizing the respective relationships that constitute each
system as locating the system within a three-dimensional field of cognitive or
social space-time. The binary isomorphic qualities shared by each type of system
make them comparable and their properties commensurate enough to trace the
interactions between the two systems of thought and action via the role location
process specified by binary role theory’s rules as modeled by CUE, TIP, and
TOM.
The role location rules for triads are logically consistent with the rules for
determining whether dyads are balanced (+,+ or −, −) versus imbalanced (+,−
or −,+), because a triad ABC can be reduced to three dyads AB, BC, and AC. As
we have seen in Chapter 2, a macroscopic triad of three agents A, B, and C can
be reduced to a set of three dyads {A|B, B|C, A|C} and each segment of the
triad solved as a separate two-body problem. This possibility is shown in Figure
14.5 where the agents in the upper part of this figure for three segments of the
triad ABC have identical subjective games in a role set with a complementary role
and counter-role represented by their respective supra-scripts for the positive (+)
or negative (−) exercise of power and the symmetrical (=) or asymmetrical (⫽)
distribution of power between them.
The triadic solution rules are illustrated in the remainder of Figure 14.5 along
with the underlying dyadic game solutions for the segments {A|B, B|C, A|C} of
a symmetrical ABC triad in which the power distribution is equal between each
dyadic segment and an asymmetrical triad where A > B, B > C, and therefore
A > C. The game solutions (equilibria) for each segment in each triad are in the
lower part of Figure 14.5 for three types of games: no-conflict, mixed-motive,
and conflict games. The valences (+ or –) for each segment of the triad are indi-
cated by just one of these symbols when they are balanced (+,+ or –,–) and by
both symbols when they are imbalanced, leaving out the directional lines for each
segment, e.g., A → B and B ← A, and replacing them with a single line without
arrows.
Inspection of the valences for the segments of a particular triad indicates
whether the triad as a whole is also stable (balanced), and if unstable, what the
prospects for stability are as a final state given the initial conditions indicated by
the valences for each segment of the triad. A stable triad has an even number of
negative (–) valences while an unstable triad has an odd number of negative (–)
valences, as defined by the mathematical theory of signed graphs (Harary 1961).
If a triad is unstable, the solutions for stability depend on the available equilibria
for the dyadic power games in each segment. For the symmetrical power triad in
 Binary Role Theory and Foreign Policy Analysis 263

ABC Triad Power Distributions

Symmetric Segment Asymmetric

Dyadic Games Outcome Cells*
A|B No-Conflict Game A+,= | B+,= A B A+,> | B+,< Settle(0) Submit/
Dominate(1)
B|C Mixed-Motive Game B–,= | C+,= B C B–,> | C+,<
Dominate/ Deadlock (2)
Submit (3)
A|C Conflict Game A–,= | C–,= A C A–,> | C–,<

Types of Dyadic Games for Triad Segments with Different Power Distributions
*TIP Preference propositions for constructing the outcome cells of each game are:
(1) If (+, <), then 0 > 2 > 1 > 3; (2) If (+,=), then 0 > 2 > 3 > 1; (3) If (+, >), then 0 > 3 > 2 > 1; (4) If (–, <),
then 3 > 0 >1> 2; (5) If (–,=), then 3>0>2 >1; (6) If (–, >), then 3 > 2 > 0 > 1 (Marfleet and Walker 2006).

Symmetrical Power Triads** Asymmetrical

A A
A+,= | B+,= + +/– A–,= | C–,= A+,> | B+,< + – A–,> | C–,<
4,4 1,2 3,3 1,4 4,4 1,1 2,3 1,4
2,1 3,3* B C 4,1 2,2* 3,2 2,3* B C 4,2* 3,1
+/– ±
A (0,1,2,3/0) A (0/0 & 1,2,3/2) A (0,1,2,3/0) A (0,1,2,3/3)
B (0,1,2,3/0) C (0/0 & 1,2,3/2) B (0,1,2,3/0) C (0,1,2,3/3)
B–,= | C+,= B–,> | C+,<
Cells
3,4 1,2 2,4 1,1
Settle(0) Submit/
4,1 2,3* 4,2 3,3*
Dominate(1)
B (0,2/0 & 1,3/2) B (0,1,2/2 & 3/3)
C (0,1/0 & 2,3/2) C (0/0 & 1,2,3/2)
Dominate/ Deadlock (2)
Submit (3)

Triadic Outcomes Menu

Unstable Triads Stable Triads

A A A A
_ _ + + _ _ + +

B _ C _ B B C B
+ +

**The valences indicate the possible final states (equilibria) for the game in each segment of a triad as balanced (+ or –)
settlement / deadlock or imbalanced (±) domination / submission. *Nash equilibria are asterisked and Brams equilibria
are in bold for each game. **Cells below each game are numbered 0 to 3 (upper-left clockwise to lower-left) as initial
states with corresponding final states separated by a slash mark when each player has the next move
(see Marfleet and Walker, 2006: 58–62). The row player is listed first and the column player is listed second below
each game. The final states of each triad are more or less balanced (stable) according to the menu of triadic outcomes
(Harary 1961).

FIGURE 14.5 General Rules for Power Triads with Agent Strategies for Each
Segment

this figure, the multiple equilibria for the A|C and B|C games suggest that: (a) if
either game shifts to a positive (+) equilibrium, the other will follow; (b) if either
game shifts to a negative (–) equilibrium, the other will follow. For the asym-
metrical power triad, the multiple equilibria for the B|C game suggest that a shift
from (+ ) to (−) by C will produce stability.
The validity of these propositions depends on the skills of the respective players
for each game and also on which player has the next move from an initial state
264 Stephen G. Walker

of play. While there are structural pressures from the triadic system toward stabil-
ity, players as agents in the system ultimately have the choice to accept or resist
this pressure. And some players are in pivotal roles within a triad, meaning that
they are positioned to exercise greater power over whether the triad is stable or
unstable. For example, Player C occupies one type of pivotal role at the bottom
of Figure 14.5, because C is a member of both triads in each half of the triadic
outcomes menu (Jervis 1997; Burt 1992).
The analytical solutions to the three-sided interaction problems in Figure 14.5
assume that each half of the dyads in this figure have convergent worlds of beliefs,
i.e, they agree cognitively that they are playing the same strategic games. A com-
plication to this kind of macroscopic solution occurs if the two agents that form
each leg of the ABC triad have leaders with different subjective games and have
not interacted through the process of role location within and between agents so
as to reach a stable equilibrium at the macro level of analysis. An additional com-
plication is what time-space scales are employed in the observation of these dyads.
As the focus shrinks temporally and spatially from larger and slower change (CO+
to CF−) to smaller and faster change (d+ to e−) in either beliefs or behavior, there
is more likely to be divergence rather than convergence between the beliefs and
behavior of Self and Other and less congruence between their respective roles of
Ego and Alter.
The inevitable result is variation in the precision and the dynamism of observa-
tions, depending on whether they are taken at the macroscopic level of interna-
tional relations in which the dyads and triads are states and other large organizations
or the microscopic level of foreign policy analysis in which the dyads and triads are
leaders and small groups. A shift in the level of aggregation from states to individu-
als reduces the precision and increases the dynamic character of the observations
while a shift in the other direction increases precision and reduces dynamism.
One level resembles Newton’s world of classical physics in which large objects
of matter and energy move in relatively predictable ways while the other level
shares characteristics of Einstein’s world of modern physics in which small quan-
tum packets of energy and information act in relatively unpredictable ways.
Integrating these two levels of analysis in physics depends on the vantage point
of the observer and the timescales employed for observations. Einstein’s theory of
relativity addressed both vantage point and timescale, but he was not ultimately
successful in understanding the behavior of very small quanta of energy moving
at high rates of speed. The attempts by subatomic physicists to achieve Feynman’s
goal of amalgamation have failed to encompass the behavior of energy as both
waves and particles with string theory under the same set of principles (Feynman
1995; Hawking with Mlodinow 2005: 117–137). The same fate seems likely in
attempts by social scientists to amalgamate the understanding of agents and larger
systems under a single set of principles as the horizons of time and space are
reduced from years to days and from states to individuals in the observations of
thoughts and actions.
 Binary Role Theory and Foreign Policy Analysis 265

This argument is qualified by the level of aggregation at which analysis occurs.

As the focus moves from higher to lower levels of analysis both spatially and tem-
porally, the degree of uncertainty increases. Smaller bits of behavior and informa-
tion occurring more rapidly or parsed more minutely lead somewhat paradoxically
to less precise estimates of reality—even the reality reflected in the language world
of texts. Just as physicists shift from point estimates to area estimates of location in
the atomic and subatomic physical universe, the estimates of thoughts and actions
in the political universe are perhaps better based on areas rather than points in the
political universe.
The cognitive and behavioral field diagrams for locating beliefs and behavior
have an origin and quadrants in which the signs and scores for Verbs in Context
(VICS) indices of balance, average, and dispersion are the coordinates estimating
the words and deeds representing an agent’s beliefs and behavior. These indices
can be disaggregated temporally by date and spatially by issue, but as their location
within a text is narrowed to the order of what is uttered and distinguished by its
specific content, these indices become more erratic and even more so as the cat-
egories for their classification become more refined from cooperation v. conflict
to words v. deeds and then scaled as different kinds of words and deeds.
This parsing of time and space into smaller metrics increases entropy and the
corresponding uncertainty of the cognitive and behavioral fields in two ways.
First, the static entropy (possible spatial states) for each field becomes greater.
Second, the dynamic entropy (possible temporal paths) increases by which prop-
erties at higher levels of aggregation emerge (Carroll 2010: 26–44, 143–178). As
leaders think and act at lower levels of aggregation, the tracing of their thoughts
and actions as processes becomes a more complex task, because of the principle of
equifinality that governs the explanation of the emerging structure (organization)
of their thoughts and actions at higher levels of aggregation (George and Bennett
2005).
For example, if the categories of beliefs and behavior are simply positive (+) or
negative (−), then there are only two possible kinds of beliefs and behavior (CO+
and CF−) and directional change in thoughts and actions are either (d+) or (e−). A
state of maximum uncertainty is when either kind of mental or behavioral event is
equally likely, p = .50 (1 out of 2), but subdividing the categories from two (CO,
CF) to four (CO Deeds, CO Words, CF Words, CF Deeds) increases maximum
uncertainty from p = .50 to p = .25 (1 out of 4) and also increases the possible
paths to CO and CF at a higher level of aggregation (McClelland 1968, 1972).
The paradox of attempting to manage uncertainty is that the more detailed and
precise the efforts at analysis, the more uncertain and indeterminate are the results
as entropy increases both statically at lower levels of aggregation and dynamically
at higher levels of aggregation. This paradox seems to hold for increasing the
number of elements in a system of large objects that change slowly as well as for
observing a few small particles with a larger number of possible states that change
rapidly. Both are n-body problems in which the possible relationships among
266 Stephen G. Walker

the elements in the system of interest increases exponentially. For example, if

the number of elements changes from two (dyad) to three (triad), the number of
dyadic relationships increases from two to three; however, if the number of ele-
ments increases to four, the number of dyads jumps to six, and for five elements
the number increases to ten (Deutsch and Singer 1964). The number of indirect
paths from one element to another also increases dramatically (Harary 1969).

Conclusion
The existence of the uncertainty paradox makes it likely that the analysis of mon-
ads (single agent), dyads (two agents), and triads (three agents) may demarcate the
interface for integrating Foreign Policy Analysis as an agent-centered subfield and
International Relations as a structure-oriented field of world politics. A focus on
triads may also be the optimum level of aggregation to demonstrate that a binary
role theory can be a unifying theory for the amalgamation of the subfield of FPA
and the field of IR. There are at least two reasons that make triads attractive for
this purpose.
First, the triad is an important system for generating the processes of balanc-
ing and mediation to manage conflict and cooperation and resolve role con-
flicts among complex systems of agents across several social and political arenas
(Caplow 1968; Zinnes 1970; Burt 1992; Jervis 1997; Schweller 1998). Second,
causal connections among three agents in a system, as the simplest of the inde-
terminate n-body problems, remain relatively local and unmediated by others.
Triads maintain the feature of the strategic dyad as a system in which each agent
has direct and adjacent contact with the others while adding only one intermedi-
ate (indirect) relationship per agent. The close (local) connections and the rela-
tively few, indirect (distant) connections make more plausible the assumption that
a triad is a “nearly decomposable system.” As Simon (1973: 23) says, “Everything
is connected, but some things are more connected than others. The world is a
large matrix of interactions in which most of the entries are very close to zero”
(quoted in Jervis 1997: 260, citing Simon 1981: 209–219).
Simon’s “nearly decomposable” assumption allows more confidence in a solu-
tion of the indeterminate, three-body problem of triadic interactions by decom-
posing and representing it as a sequence of determinate, two-body problems with
sequential game theory solutions (Newell and Simon 1972; Voss 1998). This
approach to the segments of the triad thereby honors the premise that causality is
ultimately local. In turn, this feature means optimal understanding and maximum
human control in managing uncertainty are likely to be leveraged by focusing
on strong local connections among a relatively small number of elements, which
constitute subsystems of foreign policy analysis agents within a larger system of
more distant and weakly connected international relations (Jervis 1997: 260).
15
THE INTEGRATION OF FOREIGN
POLICY ANALYSIS AND
INTERNATIONAL RELATIONS
Stephen G. Walker

Introduction
A chronic feature of the human condition is the relative inability to solve “n-
body” problems or even solve “two-body” problems at very low levels of spatial
and temporal aggregation where a single large body such as the nation-state
turns out also to be an ensemble of elements that constitute a complex cog-
nitive and social system (Feynman 1995; Deutsch 1966: appendix, 258–261).
Solutions will ultimately remain approximate rather than precise, but this analy-
sis does not imply paralysis. This final chapter discusses the macromechanisms of
role-making and role-taking in social systems of behavior and the micromecha-
nisms of steering and learning in cognitive systems of beliefs (Walker 2002). It
ends with a consideration of the ontological and methodological implications of
employing binary role theory and operational code analysis as a neobehavioral
approach to the search for causal mechanisms and the integration of Foreign
Policy Analysis and International Relations (Walker and Post 2003: 63–65;
Walker 2010).
The individual is still capable of acting as a complex adaptive subsystem and
managing uncertainty by means of learning and steering processes of communica-
tion and control within larger social and ecological systems (Deutsch 1966). Our
thoughts and actions are not random, and we are capable of being rational, i.e.,
purposeful (goal-oriented), mindful (means-oriented), and empathetic (other-ori-
ented). Rationality bestows communication and control within limits as emer-
gent properties of human interactions with one another and within the context
of the social and physical environment. Those limits mean that humans are not
subject to “cast-iron” laws like the laws of Newton’s physics but to “plastic” laws
of control more akin to the laws of Einstein’s atomic physics or to the approxi-
268 Stephen G. Walker

mate generalizations in quantum physics (Popper 1972; Almond and Genco 1977;
Feynman 1995; see also Cummins 1983).

Understanding Clocks and Clouds

Karl Popper’s metaphor for understanding human social behavior is to situate its
rules between the cast-iron laws of classical physics that govern the behavior of
“clocks” and the indeterminacy models of modern physics that govern the behav-
ior of “clouds.” At one end of this continuum are determinate systems, and at the
other end are indeterminate systems. Examples of the former are “the solar system
… pendulums, precision clocks, and motor cars.” An example of the latter is “a
cluster of gnats or small flies in which each insect moves randomly except that
it turns back toward the center when it strays too far from the swarm. Near this
extreme we would find gas clouds, the weather, schools of fish, human societies
and, perhaps a bit closer toward the center, individual human beings and animals”
(Almond and Genco 1977: 489–490).
Popper’s Problem is how to amalgamate the heterogeneous nature of reality
with a common set of principles representing a philosophy of science. His solution
is to recognize both the determinism of classical physics represented by Newton’s
theory and the indeterminism of modern physics represented by quantum theory
while also holding out for the autonomous power of human ideas (Almond and
Genco 1977: 490–491; Popper 1972: 212–226; see also Popper 1982, 1994):

What we need for understanding rational human behavior—and indeed

animal behavior—is something intermediate in character, between perfect
chance and perfect determinism—something between perfect clouds and
perfect clocks. ... For obviously what we want is to understand how such
non-physical things as purposes, deliberations, plans, decisions, theories, inten-
tions, and values, can play a part in bringing about physical changes in the
physical world.
(Popper 1972: fn.1, 228, 229. Emphasis in original. Quoted in Almond
and Genco 1977: 491)

Almond and Genco (1977: 491) go on to point out that Popper’s Solution is evo-
lutionary and consistent with the principles of variation and selection in general
systems analysis (see also Campbell 1969):

His [Popper’s] conjecture is that the problem is one of control, i.e., the control
of behavior and other aspects of the physical world by human ideas or mental
abstractions. Thus, he states that “the solution must explain freedom; and it
must also explain how freedom is not just chance but, rather, the result of a
subtle interplay between something almost random or haphazard, and something
like a restrictive or selective control—such as an aim or standard—though certainly
not a cast-iron control.” Accordingly, he restricts the scope of acceptable
 The Integration of Foreign Policy Analysis and IR 269

solutions to those that “conform to the idea of combining freedom and control,
and also to the idea of ‘plastic control,’ as I shall call it in contradistinction to a
‘cast-iron’ control.”
(Almond and Genco 1977: 491, quoting Popper 1972: 231–232. Emphasis in
original)

Popper’s Solution directly addresses the central propositions that we have used
in this book to solve the fundamental problem of understanding the reduction
of entropy and uncertainty in the worlds of beliefs and behavior that constitute
the political universe. We have linked them recursively with the two-worlds-
of-interaction proposition, namely, the steering and learning propositions that
dynamic interactions between the two worlds occur and a reduction in uncer-
tainty in one world causes a reduction in uncertainty in the other world. The
Popperian solution governing the problem of the relationship between the world
of ideas and the world of behavior is plastic control: “We are not forced to submit
ourselves to the control of our theories, for we can discuss them critically, and we
can reject them freely if we think that they fall short of our regulative standards.
Not only do our theories control us, but we can control our theories (and even
our standards): there is a kind of feedback here” (Popper 1972: fn.1, 240–241.
Emphasis in original. Quoted in Almond and Genco 1977: 491).
The notion of plastic control is consistent with a political universe constituted
by a world of beliefs and a world of behavior linked as complex adaptive systems.
As Almond and Genco (1977: 492. Emphasis in original) point out,

At the center of this complex system are choices and decisions—decisions to

command, obey, vote, make demands. The political universe has organiza-
tion; elites make decisions to command or not to command, what to com-
mand, how to implement commands. Citizens and subjects make decisions
to comply or not to comply; to make demands, how to make demands, or
not to make demands.

In short, individuals and groups exercise social power as we have conceptualized

it in this book and, as Almond and Genco note (492), “the behavioral reper-
toires of elites and citizens are not fixed repertoires. … Memory, learning, goal
seeking, and problem solving intervene between ‘cause’ and ‘effect,’ between
independent and dependent variables ... [that form] a complex array of constraints
and opportunities.”
Almond and Genco (1977: 497ff.) infer from Popper’s analysis that the notion
of causality associated with social power is flawed, because sometimes the exercise
of positive and negative sanctions by Ego does not have the effects on Alter that
Ego intends. They recommend (518–522) political scientists abandon the search
of classical behavioral political science for universal principles of amalgamation
modeled after the principles of classical physics that identify regularities or unifor-
mities in the political universe in favor of a more modest approach to the study of
270 Stephen G. Walker

politics that employs complex adaptive systems analysis. They identify the central
task of political science as

the analysis of choice in the context of constraints … [that] …would place

the search for regularities, the search for solutions to problems, and the
evaluation of these solutions on the same level. They would all be parts of
a common effort to confront man’s political fate with rigor, with the nec-
essary objectivity, and with an inescapable sense of identification with the
subject matter which the political scientist studies.
(Almond and Genco 1977: 522)

We have argued in this book that binary role theory is able to capture the process
of plastic control that characterizes the exercise of social power under the environ-
mental constraints and opportunities that political scientists study, which interact
within the open and complex systems of beliefs and behavior that constitute the
political universe. In our view, Almond and Genco (1977: 502–505) appear to
make a category error in mistaking a recursive (two-way interaction) relationship
for a non-recursive (one-way action) relationship and thereby underspecify the
causal relationship in their discussion of the exercise of social power between an
agent (Ego) and a patient (Alter). Although it may be based partly on this misap-
prehension, we do endorse their conclusion that the basic goals of political science
are discovering plastic laws (rules) of explanation and prescription, which allow
for exceptions, rather than engaging in a misplaced emphasis on the search for
cast-iron covering laws of generalization and prediction.
We contend that role theory is a theory of social interaction that focuses on
the exercise of social power in a way that is compatible with the goals of com-
plex adaptive systems analysis, which Almond and Genco (1977: 511–520) also
endorsed a generation ago as an approach already gaining momentum in several of
the social sciences as part of a post-behavioral movement in political science (Ricci
1984). Complex adaptive systems analysts ask the normative question posed by
Axelrod and Cohen (1999: xi): “In a world where many players are all adapting
to each other and where the emerging future is extremely hard to predict, what
actions should you take?” This question is essentially the same question posed by
role theory and answered by the process of role location in which Ego and Alter
reach an equilibrium in the mutual exercise of social power with an exchange of
cues resulting in the construction of a shared definition of the situation between
them that is dynamic and reversible, i.e., subject to Popper’s principle of plastic
control.
In this account, role location is both a process of navigation and construction.
The interactions of Ego and Alter communicate, define, construct, and alter a defi-
nition of the social self-in-situation. The rules of binary role theory can both explain
and prescribe the symbolic and strategic interaction processes of role location. While
they are not able to predict exactly what will occur in the future between Ego and
 The Integration of Foreign Policy Analysis and IR 271

Alter, these rules can explain what did occur in the past and prescribe what can or
ought to occur in the future. The complexity of the details of interaction processes
and their local circumstances makes it difficult, even impossible, to anticipate with
high precision the details of future interactions for complex systems. While it is
possible in principle to specify the probability of a future mental or social event
with sufficient knowledge of local conditions in the present, it is often not feasible
in practice to expect full knowledge to be available to the observer until after the
event in question has occurred.
Therefore, a general theoretical understanding of complex systems consists pri-
marily of explaining past events and modeling good and bad futures. If all causal-
ity is ultimately local, even theoretical explanations are likely to be approximate
in the absence of local details and the operation of linear or summative processes
leading to events. Depending on the complexity of the system that generates
events and the observer’s access to knowledge of local conditions, it is feasible
to make only relatively modest predictions about the probable occurrence of
immediate events within relatively simple subsystems, such as those governed
by the immediate choices of single predominant leaders or such additive mecha-
nisms as voting within elite decision units or in mass elections. Even under these
conditions, overconfidence or other cognitive and motivational biases of human
estimators may lead to false probability judgments (Jervis 1976; Kahneman and
Tversky 1979; Vertzberger 1990; Fiske and Taylor 1991; Tetlock 2005).
The prediction of single events based on more complex, non-linear interac-
tions within bureaucracies and institutions or in the natural world is relatively
uncertain no matter whether the event is social or physical, e.g., a war or an earth-
quake, unless a detailed temporal narrative of adjacent (local) events connecting
an initial state (event) with a final state (event) is available to construct and trace
a causal story (Little 1991; Edelman 1992; Feynman 1995; Allison and Zelikow
1999; Dessler 2004: 386–394; George and Bennett 2005). It is possible to reduce
uncertainty with probability estimates but not to eliminate it in human affairs.
We offer a quantum of solace regarding this conclusion in recalling that precise
prediction is over-rated as an aspiration in any of the sciences and necessarily sub-
ordinate to understanding as an intellectual goal (Feynman 1995).

The Microfoundations of Foreign Policy Decisions

Foreign policy and international relations events are the products of very com-
plex, agent-level subsystems of thoughts and actions interacting as larger social
systems of varying complexity. It is possible with sufficient access to detailed
belief inventories and historical behavior chronologies to construct a causal story
to understand and explain a single mental or behavioral event. Diplomatic histo-
rians undertake such explanations-in-detail to explain the thoughts and actions of
leaders under the conditions of different constraints and opportunities and may
even draw inferences about general regularities across these narratives (Elman and
272 Stephen G. Walker

Elman 2001). It is also possible to begin with a general narrative template linking
past, present, and future as an explanation-in-principle and fill in the details for
the narrative of a particular event (Dessler 2004), as we have done with an alliance
of cognitive and behavioral systems theories in this book.
The future research agenda for employing this alliance of cognitive consis-
tency and game theories, forged with the concepts and rules of binary role theory
as a general theory of symbolic and strategic interaction, would appear to be a
combination of normative and explanatory tasks. On the one hand, a structure-
oriented focus on prescription and systems design can prescribe assistance in the
design and conduct of international relations among larger ensembles of agents
and dyads, which may help achieve good outcomes and avoid bad outcomes. On
the other hand, agent-centered foreign policy analyses focusing on explanation
and causal mechanisms can explain the dynamics of strategic dyads over time and
across issues. We shall close with a discussion of how these two tasks can be car-
ried out within the theoretical context provided by the concepts and propositions
of binary role theory.
The social power map in the upper-left of Figure 15.1 is a template of the infor-
mation necessary for the binary role theory in this book to manage the uncer-
tainty problems presented by past hypothetical relations among agents ABCD in
a regional international system, which are also depicted in the upper-right of this
figure as a set of interconnected strategic triads. The Past Hypothesis accounts for
the existence of these relations as the initial state for the analytical narrative below
and represented in the remainder of Figure 15.1 as a series of Pivot intervention
models. In this narrative a new agent intervenes as a Pivot in this regional system
either in a First Encounter with the other agents in the regional system or as an
agent whose leaders decide to “reset” relations with the others. This agent is
called the Pivot after its strategic role as the third body in the possible role set of
strategic triads with the rest of the agents in the system (Jervis 1997).
Who can and should enact a Pivot role in this regional system? The determi-
nants of a pivotal position in a system of world politics include location, technol-
ogy, and a pragmatic ideology. These factors affect the type of pivot role available
to a state as well as the prospects for enacting the role with success. Jervis argues
that it is possible to act either as an internal (more dense) or external (less dense)
pivot role within a system, depending on whether the state’s location is deeply
enmeshed within the system of interest or is relatively autonomous in a structural
hole (Jervis 1997: 177–209; Burt 1992; Joffe 1995).
A fourth determinant of success is the strategy that the potential pivot enacts.
A binary role of cooperation (+) or conflict (−) in the strategic exercise of power
is crucial in establishing the criteria for success by specifying preference rankings
for the possible outcomes that follow from the strategic enactment of the role,
which we have identified as settlement, deadlock, domination, or submission. In
turn, the determinants of power distribution and the exercise of power as posi-
tive or negative sanctions include role demands in the form of such constraints
 The Integration of Foreign Policy Analysis and IR 273

Ego|Alter A = B > C = D A
A n+/– n± s± n+/– n±
B n+/– n+ s± +
C n± n+ s+ n
B C
D s± s± s+
s±
Social Power Map s± s+
D
Strategic Triads

Pivot Effects Strategic Games

A B A–, = | B–, = A–, >| C–, < A+, > | D–, <
– – 3,3 1,4 2,3 1,4 2,3 1,4
(+) –
4,1 2,2* 4,2* 3,1 4,2* 3,1
+ A (0,3/ 0 & 1,2/ 2) A (0,1,2,3/ 3) A (0,1,2,3/ 3)
(–)
B C D C B (0,1/ 0 & 3,2/ 2) C (0,1,2,3/ 3) D (0,1,2,3/ 3)
+ +
PIV+, = | A+, = PIV+, = | B+, = PIV–, = | C–,=
+ (–) (–) – 4,4 1,2 4,4 1,2 3,3 1,4
Pivot Pivot 2,1 3,3* 2,1 3,3* 4,1 2,2*
Effects Effects PIV (0,1,2,3/ 0) PIV (0,1,2,3/ 0) PIV (0,3/ 0 & 1,2/ 2)
Model (1) Model (2) A (0,1,2.3/ 0) B (0,1,2.3/ 0) C (0,1/ 0 & 3,2/ 2)
B C B+, > | C+, < B–, >| D+, < D+, = | C+, =
(+) – – + 4,4* 1,2 2,4 1,1 4,4* 1,2
– 3,1 2,3* 4,2 3,3* 2,1 3,3*
B (0,1,2,3/ 0) B (3/ 3 & 0,1,2/ 2) D (0,1,2,3/ 0)
A D A D
– C (0,1,2,3/ 0) D (0,1,2,3/ 2) C (0,1,2.3/ 0)
+ –
+ (–) + (–)
Pivot Pivot
Effects Effects
Model (3) Model (4)

*Nash equilibria are asterisked and Brams equilibria are in bold. **Valence outcomes are in parentheses
and antecedent valence conditions are not. Strategic valences are: mutual cooperation (+); mutual conflict
(–); cooperation or conflict (+/–); submission/domination (±). Agents are states ABCD in a regional system.
Locations are neighbors (n) and strangers (s). PIV is the role of pivot by an agent inside or outside of the
regional system.

FIGURE 15.1 Pivot Effects Models for Strategic Triads among Agents ABCD

and opportunities as location and technology, plus the ideologies and strategies of
other actors than the pivot (Jervis 1997).
A final determinant of success is the normative criterion for identifying suc-
cess, as posed earlier by Axelrod and Cohen’s (1999: xi) question regarding the
purpose of complex adaptive systems analysis: “In a world where many players
are all adapting to each other and where the emerging future is extremely hard
to predict, what actions should you take?” The answer depends on what your
normative goal is—it could be to stabilize a region characterized initially by coop-
eration (+) or conflict (−) among its members and a distribution of power among
them that is symmetrical (=) or asymmetrical (≠). In this example, therefore, we
shall make the normative policy assumption that a region is preferred with rela-
tions ranked as no-conflict > non-zero-sum conflict > zero-sum conflict and a
274 Stephen G. Walker

distribution of power in which the outcomes of settlement or domination are

preferred possibilities over the outcomes of deadlock or submission in dyadic
relations.
Skillful implementation of a strategy is also a requirement for enacting a suc-
cessful role as a Pivot, which requires a well-designed map of the dynamics in the
system of interest in order to implement it. It is important to remember that part
of the task in managing social relations is their construction as well as their naviga-
tion. In both instances, the goal is the reduction of uncertainty and the manage-
ment of risk associated with avoiding bad outcomes (deadlock or submission) and
achieving good outcomes (settlement or domination).
In this book we have argued primarily for an alliance of game theory and
role theory as a theory complex to act as an analytical compass for “mapping”
(navigating and constructing) the terrain of world politics and for steering agents
in a complex adaptive system toward optimum or at least satisfactory outcomes.
Together they offer navigational aids and architectural plans for constructing net-
works of non-zero-sum relations, which are necessary but not sufficient for the
goal of sustainable cooperation in world politics (Wright 2000). We have also
provided some rules of diagnosis and prescription, CUE, TIP, and TOM, which
are keys to eliciting and constructing cooperation from non-zero-sum strategic
situations and for converting zero-sum to non-zero-sum relations and no-conflict
relations.
The strategic situations constructed by the models in Figure 15.1 are analyzed as
a series of strategic interventions by the Pivot from antecedent conditions through
intermediate states in chronological order with final outcomes as the Pivot effects.
The results of the analysis indicate that if the Pivot pursues an intervention strat-
egy of conditional cooperation first toward A and if this intervention produces
the consequences predicted by signed graph theory, then the ensuing results are
consistent: (a) as intermediate states with strategic triads balanced with coopera-
tion between A and the Pivot; (b) as final outcomes in the form of eventual joint
conflict toward C and D. This top-down approach by the Pivot to constructing
balanced triads and non-zero-sum games between B and A specifies that if the
Pivot pursues a successful strategy of conditional cooperation toward each of these
actors in Model (1), the predictions are that: (a) the relations between B and A
will shift from conflict (+/–) to cooperation (+); (b) relations between B and C
will shift from cooperation (+) to conflict (–); (c) relations between the Pivot and
C will be conflict (–).
These predictions are consistent with the initial conditions for each dyad speci-
fied in their strategic games as relations in each triad, a Pivot strategy of condi-
tional cooperation, and the balance theorems of signed graph theory applied to
political dyads and triads under these initial conditions (Harary 1961). The {PIV,
A, B} triad becomes stable (balanced) after the shift in [A | B] relations occurs
in accord with the proposition that a balanced triad has all plus signs. The {A,
B, C} triad becomes stable (balanced) after the shift in [B | C] relations occurs
 The Integration of Foreign Policy Analysis and IR 275

in accord with the proposition that a balanced triad can have an even number of
minus signs. The same proposition holds for [PIV | C] relations in the {PIV, A,
C} triad.
In Model (2), the antecedent conditions are specified by the negative rela-
tions between B and D, the shift in relations between B and C in Model (1), and
the Pivot’s intervention strategy in Model (1). The prediction is that relations
between D and PIV will be conflict (–). The same prediction for [B | A] in
Model (1) and [D | PIV] in Model (2) are evident in Model (3) and Model (4),
respectively, when relations with C are not considered within the triads of inter-
est in Model (3) and when relations with B are omitted from the triads of interest
in Model (4).
This narrative analysis specifies the theoretical consequences of a binary role
of regional cooperation by the Pivot that begins with a conditional cooperation
strategy toward A; the Pivot’s relations with B in Figure 15.1 may either precede
or follow this step. It is the Pivot’s role to facilitate a shift in relations from (+/–)
to (+) between A and B. The Pivot and A would then together encourage a shift
in relations between B and C in Model (1) corresponding to their own relations
with C. The Pivot’s role in Model (2) would be to establish cooperation with B
and then balance the {PIV, B, D} triad by establishing conflict relations with D if
they did not already exist. The sequence of steps for the incumbent of the Pivot
role is generally the same in Model (3) and Model (4).
In order to assess the feasibility of this macrostructural analysis, it is desirable to
investigate the separate segments of these triads. This task involves surveying the
microfoundations of their interactions that locate relations between these agents
and answering the following questions. How stable are those relations? Are they
products of the same subjective games played by the leaders of each agent? This
analysis involves an inventory of the respective belief systems for the agents of
each dyad that should extend over time and across issues to find out if the initial
state specified in Figure 15.1 for each dyad is likely to emerge from the micro-
foundations of decisions by the agents in each dyad.
This judgment and an assessment of stability are desirable in order to determine
the feasibility of the prescriptions for intervention by a Pivot in this regional
system. An assessment of stability should also take into account the local
conditions that act as role demands and make stability more or less likely. Agents
with different cultures and technologies may be more or less likely to select and
maintain a positive (+) role of cooperation toward one another, as the demo-
cratic peace literature suggests, while other cultures may be less likely to adopt
cooperation as a stable role (Schafer and Walker 2006b; Feng 2007; Malici 2006;
Lebow 2008). Some regional systems may be more institutionalized with shared
mental models that provide rules for mutual interaction (Malici 2008; He 2009).
Such internal and external features can qualify the logic of role location and sug-
gest new or additional rules to the ones embodied by our CUE, TIP, and TOM
models.
276 Stephen G. Walker

The analysis of these possibilities is likely to focus on phenomena identified

by role theory as conditions and macroprocesses associated with the concepts of
role-making, role-taking, role enactment, role strain, role competition, and role
conflict. If the agents in foreign policy analysis are individuals and small groups
interacting within larger units such as institutions and organizations that aggregate
to the state as the agent in international relations, then the concepts representing
these relationships in role theory should be unpacked and the microprocesses
among these levels of analysis also need to be theorized as causal mechanisms
(Bunge 1967; Hedström and Swedberg 1998; Walker and Post 2003). Elster
(1998: 45. Italics his) conceptualizes such mechanisms in a way consistent with
Popper’s notion of plastic control as “intermediate between laws and descriptions.
... [T]hey are frequently occurring and easily recognizable causal patterns that are triggered
under generally unknown conditions or with indeterminate consequences.”
Such processes in foreign policy analysis and international relations include
the various macroprocesses associated with role-making produced by the steering
effects of decision making and with role-taking produced by the learning effects
of social awareness. Smith (1968) identifies such individual microprocesses as
projection, ego defense, object appraisal, and mediation of self-other relations,
shaped by the personality traits of leaders and often occurring unconsciously
within leaders, which influence the more social role-making process. These psy-
chological analyses of leadership style, e.g., Hermann (1980, 2003) and Winter
(1973, 2003), offer important insights into the dispositions and motivations that
explain why a leader selects the enactment of a positive or negative role concep-
tion (Greenstein 1969; Walker and Falkowski 1984; Schafer 2000; Post 2003;
Walker 2009).
Examples of causal mechanisms that are more overtly conscious, cognitive, and
social processes within the larger decision-making process of role-making include
the framing effects of realistic empathy regarding rogue states in Chapter 5, the
risk-acceptant and risk-averse calculations identified by the middle-range, situ-
ational logic of prospect theory in the analysis of North Korea’s nuclear strategy in
Chapter 6, the groupthink syndrome and its antecedent conditions in the analysis
of small groups and leaders in Chapter 7, and the external threat-internal cohesion
mechanism guiding the intra-state and inter-state comparisons of NATO allies in
Chapter 8. All of these possible decision-making mechanisms may have diagnostic
and steering effects within the role-making process that leads to the selection and
enactment of a binary role conception of either cooperation or conflict and its
subsequent enactment as de-escalation or escalation in interactions with others.
The tactical variants of the two strategies of cooperation or conflict in the
initial exchange of cues that constitute the larger process of role location are
causal mechanisms associated with the concept of role enactment. The sequences
of Appease, Deter, Reward, Exploit, Bluff, Compel, Punish, and Bully tactics
are social processes within the larger strategic interaction processes embedded in
game theory (Elster 1998; Cowen 1998), as are the processes specified by the rules
 The Integration of Foreign Policy Analysis and IR 277

of TOM to account for multiple equilibria in sequential games and the processes
specified by the rules of TIP and CUE to link social and cognitive events.
Other causal mechanisms specify learning effects associated with the process
of role-taking. Examples in this book include the dissonance reduction process of
cognitive consistency and the adaptive processes of simple, diagnostic, and com-
plex learning reported for Castro and Kim Il Sung in Chapter 9 and for George
W. Bush in Chapter 10 to account for continuity and change in role conceptions
attributed to self and role expectations attributed to others, plus the inventory of
domestic and international mechanisms specified and tested for learning effects in
the form of non-directional (absolute) belief change by U.S. presidents in Chapter
11. The differential effects of success or failure on the complex learning and steer-
ing processes of U.S. presidents and presidential advisory systems within crises in
Chapter 7, across successive crises in Chapter 12, and during crises with rogue
states in Chapter 13, all illustrate various causal mechanisms in both the cognitive
world of beliefs and the social world of behavior.
As the analysis of foreign policy dynamics expands from monads to dyads to
triads, other concepts from role theory become relevant, including role strain,
role competition, and role conflict. These concepts can inform a search for causal
mechanisms surrounding the strategies of bandwagoning, appeasement, balanc-
ing, and hegemony associated with general structural theories of international
relations (Walker 2007b). These role concepts also appear to engage varying levels
of rationality. Role strain occurs at the level of rationality (purposefulness) when
role demands frustrate Ego’s goal of enactment of a single role. Role competition
occurs at the level of instrumental rationality (mindfulness) when role demands
make it difficult to enact multiple roles. Role conflict occurs at the highest level
of strategic rationality (empathy) and makes the enactment of multiple roles an
exercise in reconciling contradictory behavior.
Ego may experience role strain in adjusting the enactment of a role in the face of
changing conditions in either the strategic environment, e.g., new leaders or new
states, or changes in Ego’s power position within a regional or global system, e.g.,
Germany after World War I, Britain after World War II, Russia and the United
States after the cold war (Gilpin 1981; Kennedy 1987; Doran 1991; Kupchan
1994; Malici 2008; He 2009). Variants of prospect theory, image theory, and
social identity theory may offer insights into the microprocesses of role-making
and role-taking at these turning points in foreign policy dynamics (Welch 2006;
Lebow 2008; Boulding 1959; Herrmann et al. 1997; Cottam 1977, 1986; Wendt
1999; Larson and Shevchenko 2010). Maoz and Mor (2002) have applied TOM
rules of sequential game theory to model the role strain of managing strategic
adjustments during an enduring rivalry at different periods in its history, and oth-
ers have suggested that image theory may be expanded to understand strategic
interaction (Herrmann and Fischerkeller 1995; Herrmann 2003).
The concept of role competition (difficulty in selecting and enacting multiple
roles) focuses on the resource management problem of how to allocate cognitive
278 Stephen G. Walker

attention and social power to the enactment of multiple roles when Ego does not
have sufficient cognitive and social resources to enact both of them and achieve
role satisfaction (Walker 1992). This shortfall creates a “two-good” problem for
Ego, which is solved with the classic instrumental rationality mechanism of cal-
culating trade-offs and reaching an equilibrium of allocating cognitive and social
resources between the two goods (Palmer and Morgan 2006).
The main contribution of this two-good theory of foreign policy decision
making is to close the spatial gap between different foreign policy decisions
regarding different targets and issues with trade-off and portfolio concepts that
link disparate policy choices within the same cost/benefit calculations and invest-
ment decisions. This solution may range from completing suppressing one role
to achieve a satisfactory performance of the other one to reaching a minimally
satisfactory performance level of both roles (Palmer and Morgan 2006; Hartmann
1982; Walker 1992).
Cognitive models of search, evaluation, and calculation identify several other
causal mechanisms for managing attention and calculation as cognitive resources
in addition to the classic instrumental rationality model, which are employed as
cognitive shortcuts in information processing and cognitive management by deci-
sion makers as a function of available cognitive resources and the complexity of the
task (Fiske and Taylor 1991; Lau 2003; Suedfeld et al. 2003). Two cognitive deci-
sion-making models combine different models of instrumental rationality across
spatial domains or temporal stages and test them with laboratory experiments
and historical case studies of foreign policy decisions. Prospect theory models
cognitive processes spatially by framing risk calculations and choices as occur-
ring within a domain of gains or losses (Kahnemen and Tversky 1979; Farnham
1975; McDermott 1998; Taliaferro 2004). Poliheuristic theory represents cogni-
tive processes in two temporal stages. The first stage is a non-compensatory stage
in which a satisficing search strategy identifies those choices that meet a minimal
criterion of acceptability. The second stage is a compensatory stage in which
the surviving options are subjected to a trade-off assessment consistent with the
two-good model of choosing among alternatives (Mintz et al. 1997; Mintz and
DeRouen 2010; see also Farnham 1997).
A closely related concept to role competition is role conflict (difficulty in selecting
and enacting multiple contradictory roles). The two empirical phenomena often
occur together, and both may take either a bilateral or a multilateral form (Walker
1987b: 86). Ego may experience: (a) bilateral difficulty in choosing between two
roles toward the same Alter, e.g., cooperation with Alter regarding economic
issues and conflict with Alter regarding security issues; (b) multilateral difficulty
in cooperating with Alter 1 while engaging in cooperation with Alter 2. If the role
strain is due to insufficient cognitive or social resources, then these two instances
are forms of multiple role competition. If the role strain is due to the contradic-
tory opposition between the two roles, then these two examples are forms of
multiple role conflict.
 The Integration of Foreign Policy Analysis and IR 279

Role theory identifies four strategies as mechanisms for resolving role conflict
(Walker 1987b: 86–88; see also Sarbin and Allen 1968: 538): “(1) alternation of
two or more roles within the period of observation; (2) the merging of two or
more roles so they are indistinguishable in terms of role enactment behavior within
the period of observation; (3) the interpenetration of roles without their merger in
behavioral terms within the period of observation.” Altercasting is a fourth strategy,
which is a causal mechanism that resolves role conflict by changing the expecta-
tions of Alter regarding Ego’s behavior and thereby changing the identity of one
of the conflicting roles (Walker 1987b: 88–89). A model of an instrumental ratio-
nality mechanism to resolve role conflict is the application of prospect theory to
choose between two contradictory roles, in which the emphasis is on the distinc-
tion between the two alternatives to maintain or alter the status quo and the choice
is a function of Ego’s location in the domain of gains or losses (Welch 2006).
Examples of bilateral role conflict include when Ego initiates an altercasting
strategy, e.g., transforming relations with Alter from enmity to amity (Kupchan
2010), or when Ego responds to an altercasting strategy by Alter, e.g., Neville
Chamberlain’s attempts to maintain friendly relations in the face of negative
feedback as Great Britain attempted to deal with hostile behavior from the gov-
ernments of Germany and Italy in the 1930s (Eden 1960; Walker et al. 2008).
Examples of multilateral role conflict are in the analyses of the strategic triads in
Figure 15.1, which resolve multilateral role conflicts with the rules of binary role
theory for balancing triads.

Conclusion
The theoretical integration of Foreign Policy Analysis and International Relations
in the end requires that conceptual and empirical anomalies be resolved across
the different levels of aggregation from monads to dyads to triads and even larger
ensembles of agents in a social system of interest. We have identified in this chap-
ter a dual approach to solving this problem of integration with beliefs from the
operational code construct: one is the search for a unifying theory and the other
is the search for causal mechanisms (Walker 2002; Walker and Schafer 2010).
Binary role theory is compatible with both of these approaches as a solution to
this problem.
The preceding discussion of explanation and causal mechanisms illustrates how
middle-range causal mechanisms at the agent level of foreign policy analysis can
be subsumed as specific instances of more general cognitive and social processes
represented by the concepts of role theory. It is less clear from the preceding
discussion of prescription and systems design how role theory unifies the insights
of other system-level theories of international relations. One way to demonstrate
how role theory works to close this gap is to ask two basic questions: one is an
ontological question about the nature of reality while the other is a methodologi-
cal question about how to study reality. In this context, these questions become:
280 Stephen G. Walker

What is the nature of (world) politics? How does particular theory (X) approach
the study of world politics?
We have addressed these questions implicitly throughout this book and explic-
itly in these last two chapters with (X) = binary role theory. Binary role theory
answers these questions as follows: (a) the nature of world politics is the inter-
play of power between agents in a social system; (b) an agent-centered approach
emphasizing the exercise of behavior as social power and rationality as cognitive
power to reduce uncertainty and manage complexity is one way to understand
this reality. Considering these questions and answers for binary role theory more
generally and in comparative theoretical perspective prompt the following addi-
tional questions.
Is there agreement on the nature of world politics as the interplay of power
among agents in a social system among general theories of International Relations?
How do general theories of International Relations deal with the problem of
managing uncertainty? More specifically, how does each theory handle the two
worlds of interaction between mental and social events? The answers to these
comparative questions are embodied in each theory’s solutions to the agent-
structure and level-of-analysis problems, which role theory solves by instantiating
agents and then expanding to higher levels of analysis.
Answers by political scientists and political activists to the ontological ques-
tion reveal a common definition of the nature of politics. Easton (1953) defines
politics as the authoritative allocation of values for a society; Lasswell (1958)
defines politics as who gets what, when, and how; Morgenthau (1978: 29) states,
“International politics, like all politics, is a struggle for power.” Bismarck cir-
cumscribes politics as “the art of the possible” (Knowles 1999). Lenin answers
the ontological question with another question in the form of the cryptic phrase
“Kto-kovo?” (Who-whom)? It refers to “Who (will destroy, will control, will
utilize) whom?” and focuses on the exercise of social power by agents who inter-
act strategically to determine who gets what, when, and how in a struggle for the
authoritative allocation of values in a social system (Leites 1953: 27–29; Walker
2003: 246).
Operational code analysis as a neobehavioral research program employs the
concepts of cognitive rationality and social power to focus on the world of beliefs
and the world of behavior regarding the interaction of agents in a social system.
Binary role theory formalizes the relationships between the cognitive system of
beliefs and the social system of behavior with rules for articulating symbolic and
strategic interaction processes between them. The ontological location or “start-
ing point” for binary role theory is ultimately an assumption of radical indetermi-
nancy, represented as a “non-location” by the origin in our cognitive and social
field diagrams for the two political worlds. Politics as the exercise of social power
is movement in any direction from the origin to instantiate relations as an emer-
gent property of a system between agents in the form of a social Self and a social
Other, i.e., as Ego and Alter, which reduces entropy and uncertainty and creates
 The Integration of Foreign Policy Analysis and IR 281

a complex structure of cognitive and social power relations known ontologically

as politics.
In comparing role theory’s ontological and methodological positions with other
general IR theories, one contrast is that realist, liberal-institutionalist, constructiv-
ist or neo-Marxist approaches have different theories of agency. In realist theory,
agents may be individuals or states who seek power or survival (Morgenthau
1978; Waltz 1979; Schweller 2003; Lobell et al. 2009). In neoliberal-insitutional-
ist theory, agents as groups or organizations seek to protect or achieve institutional
interests (Keohane and Martin 2003; Moravcsik 2003). In neo-Marxist theory,
individuals and classes as agents seek to enhance their wealth (McGowan and
Walker 1981; Thompson 1983). In constructivist theory, agents as members of
gender, ethnic, or religious communities and societies seek to express and protect
their cultural and social identities (Tickner 1997; Wendt 1999; Lebow 2008).
A similarity is that with some exceptions these other theories solve the agent-
structure problem by privileging structure over agency, i.e., they begin with the
Past Hypothesis rather than the First Encounter Hypothesis in their approach to
world politics. Exceptions are classical and neoclassical realism, some economic
variants of liberalism and some post-structural variants of constructivism, which
emphasize psychological attributes of personality or rationality as primitive char-
acteristics of agency that precede anarchy, markets, or society (Schweller 2003;
Morovcsik 2003; Lebow 2008; Doty 1996). Role theory entails these various
solutions to the agent-structure and level-of-analysis problems by being able to
privilege either the Past Hypothesis or the First Encounter Hypothesis as a bound-
ary condition and thereby offer a structurationist solution to the agent-structure
ontological problem (Wendt 1986).
Role theory is also compatible with all of these approaches by entailing their
particular theories of agency, i.e., deriving them as elements in a larger task of
uncertainty reduction. Entailment is the solution of a conceptual (theoretical)
problem “in the form of movement along a continuum from inconsistency or
implausibility between theories toward compatibility, reinforcement, or ulti-
mately the ‘entailment’ (derivation) of one theory from another one” (Walker
2003: 270; see also Laudan 1977: 48–54). Role theory entails and unifies these
different approaches in varying degrees by breaking down the ontology of politics
into three elements within the framework of Lenin’s kto-kovo question of who
(controls) whom, which encompass the central questions posed by each of these
general IR theories. These elements are identities, interests, and interactions. The
identities question is: Who are Ego and Alter? The intentions question is: What do
Ego and Alter want to do? The interactions question is: Can Ego and Alter do what
they want? Cultural constructivists focus mainly on the identities question while
neoliberal and neo-Marxist institutionalists are focused on the intentions question
and realists address the interactions question.
Role theory entails all three elements with a formal model of symbolic and stra-
tegic interaction (Walker 1992; McCall and Simmons 1978). In Stage I, Ego and
282 Stephen G. Walker

Alter communicate symbolically with a sequential exchange of cues their respec-

tive identities, intentions, and interaction capabilities as roles of Cooperation (+)
or Conflict (–) and then interact strategically in Stage II to reach a stable equi-
librium as a final outcome of their encounter (Walker 2004a, 2004b). The use
of game theory as a binary model to formalize these processes of social commu-
nication and control not only captures the theory kernels in these other general
theories. It is also unifying in the sense that game theory is a model of agents and
systems that extends across the natural and social sciences (Wright 2000).
While binary role theory offers a way to integrate the understanding of Foreign
Policy Analysis and International Relations with a common set of rules, it is
clearly not the end point for inquiry into these domains of politics. As Feynman
(1995: 24) reminds us about the science of physics: “Actually, we do not have
all the rules now. ... Aside from not knowing all of the rules, what we really can
explain in terms of those rules is very limited, because almost all situations are so
enormously complicated that we cannot follow the plays of the game using the
rules, much less what is going to happen next.” The same comment is true about
rules for the science of world politics.
Therefore, it is appropriate to end a book about the theoretical understand-
ing of uncertainty in the subfield of Foreign Policy Analysis and the field of
International Relations via plastic laws of communication and control with the
following reminder that even the cast-iron laws of macroscopic physics are sub-
ject to revision. New results from the study of planets in distant star systems “are
calling into question current theories about how solar systems are formed and suggesting that
more complex theories must be developed.”

... [R]ecently observed extrasolar planets are revolving around their stars
in the opposite direction from the stars’ spin. ... [T]hey must reach their
final orbits not through the familiar process of accumulation of dust, “but
through the much more dramatic and exciting process of gravitational bil-
liards” … [in which] … planets orbiting a star would result in some planets
getting sling-shotted into highly eccentric orbits, in some cases even back-
ward (or retrograde) ones.
(Maugh 2010. Italics added)

We can expect even more complex interactions than billiard ball models and
corresponding theoretical challenges when states collide and leaders decide in the
social and cognitive systems of world politics.
APPENDIX: FORMAL MODELS
OF SYMBOLIC AND STRATEGIC
INTERACTION
Stephen G. Walker

Introduction
The entropy matrix of theoretically possible, dyadic interaction states for the
binary role theory in this book is in Table A.2 at the end of this appendix, which
contains a slightly revised and corrected analysis by Gregory Marfleet of the simu-
lation results appearing in Marfleet and Walker (2006). This political universe does
not include all of the possible mathematical patterns for the interaction sequences
between row and column players in a 2 × 2 game; the results are bounded for
binary role theory by its rationality assumptions about how to rank the possible
outcomes of settlement, deadlock, domination, and submission. The assumptions
are that it is rational for Ego and Alter to rank settlement over deadlock and domi-
nation over submission, conditioned by the interaction between their respective
beliefs about the distributions of power and interests between them.
An illustration is in Table A.1, which contains this information for the inter-
section of the subjective games at the four corners of Table A.2. The latter table
contains the expected outcomes and simulation-generated outcomes (final states)
generated by all possible pairs of agents and starting quadrants (initial states) for
the intersection of their subjective games specified by binary role theory. The
analysis of the intersection of the smaller number of role sets in Table A.1 shows
what happens when a player with an Appeasement (+) strategic orientation (Ego:
+,< ; Alter: +,>) and a player with a Bully (−) strategic orientation (Ego: −,>;
Alter:−,<) encounter one another from the four possible initial states of a 2 × 2
game.
These states are denoted in clockwise order as starting quadrants (0 = upper-left
cell, 1 = upper-right cell, 2 = lower-right cell, 3 = lower-left cell), which cor-
respond to states of settlement, submission, deadlock, and domination for the row
284 Stephen G. Walker

TABLE A.1 Intersections of Subjective Games for Appeasement (+) and Bully (–) Role Sets

Column
App+ Bully–
0|0 0|0 0|2 0|2
App+ ———
0|0 0|0
| ——— |
0|2 0|1
Row
3|3 3|2 3|2 3|2
Bully– ———
3|3 3|2
| ——— |
3|2 3|2

Cells indicate expected and actual (E|A) outcome by starting quadrant (0 = upper-left cell, 1 = upper-
right cell, 2 = lower-right cell, 3 = lower-left cell) of the 2 × 2 subjective games for Row and Column.
These outcomes assume that Row has the next move, and bold text indicates outcomes that deviate
from that expected by the row player as first mover. These outcome matrices are taken from the four
corners of Table A.2.

player. If the two players both have an Appeasement orientation, the expected
and actual final state (outcome) is congruent; however, in all the other cases there
are deviations from expected and actual outcomes with only two exceptions. If
the player with a Bully (−) role set has the starting move from an initial state of
either mutual cooperation (0) or domination (3), then there is a convergence
between the expected outcome for that player’s subjective game and the actual
outcome of the intersection of the subjective games for the two players.
These results illustrate the problems of uncertainty reduction and the manage-
ment of complexity associated with the role location process modeled by the
rules of binary role theory. While our binary role theory does not privilege either
side of this dyadic interaction process over the other side, there are a couple of
anomalies from the perspective of realist theory. They pose some conceptual and
empirical problems for binary role theory regarding two-sided interaction prob-
lems, which may make it desirable to expand its rules in order to entail them as
solved problems (Laudan 1977).
Binary role theory specifies only two basic strategic roles regarding the exer-
cise of power: positive (+) and negative(−) sanctions conceptualized statically as
Cooperation (CO+) or Conflict (CF−) with their respective dynamic gradients of
de-escalation (d+) or escalation (e−), plus two tactical rules of conditional (recip-
rocate d and e) or unconditional (always choose d or always choose e) sequences
for each strategy. However, there are four possible ranked outcomes {settlement;
domination; submission; deadlock} for the intersection of these strategies and
tactics: the static states of {CO,CO; CO,CF; CF,CO; CF,CF} and their corre-
sponding dynamic states of {d,d; d,e; e,d; e,e}.
Excluded from the maximum level of entropy in binary role theory are distinct
outcomes resulting from ranking either submission or deadlock as the top prefer-
ence. These two families of strategic options appear at first glance to correspond
 Formal Models of Symbolic and Strategic Interaction 285

to the well-known strategies identified by realist IR theory as bandwagoning,

i.e., always choose cooperation with submission as the highest-ranked outcome
and balancing, i.e., always choose conflict with deadlock as the highest-ranked
outcome (Waltz 1979; Schweller 1994, 2006). This omission may turn out to
be more apparent than real, depending on their empirical frequency, because
these outcomes occur as Nash myopic or Brams nonmyopic equilibria within
the intersections of other families of strategies that rank highest one of the other
outcomes.
For example, the prisoner’s dilemma game [E (–,=) |A –,=)] in the entropy
matrix for binary role theory shows that if CF,CF is the initial state, then CF,CF
is also the final state, as both Ego and Alter will choose “stay” in repeated plays
of the game. This “always choose conflict” pattern is consistent with a balancing
strategy by both players of ranking deadlock (CF,CF) highest. Bandwagoning is a
strategy with a top-ranked preference for submission, an outcome that can occur
frequently as a second-best or even lower-ranked outcome for one agent in the
interactions between two agents with asymmetrical power distributions (Waltz
1979; Schweller 1994, 1998, 2006).
While the outcomes may be empirically frequent and the families of strat-
egies are mathematically possible, balancing and bandwagoning strategies may
not be valid solutions as explanations for their respective outcomes of deadlock
and submission that occur frequently in the empirical world. Wohlforth et al.
(2007) report very few instances of balancing as an explicit strategy in a survey
of international relations across regions and eras, and historians are divided about
its frequency of occurrence and its importance (Schroeder 1994; Gulick 1955).
Ranking these two outcomes highest also violates the assumptions of rationality
that rank winning (domination or settlement) over losing (deadlock or submis-
sion) as outcomes, which are the theoretical basis for binary role theory’s entropy
matrix.
Game theorists have a mixed record in recognizing widely either strategy per
se as an important rational strategy in its own right. Brams (1994: 215) notes that
there are 78 possible 2 × 2 games that are structurally distinct, but he eliminates
consideration of the 21 no-conflict games as relatively uninteresting because of
their obvious solution, which would mathematically include balancing games in
which both agents rank deadlock as their highest outcome. Stein (1990) includes
some of these no-conflict games in his analysis of why nations cooperate; however,
he does not explicitly consider the case in which both agents choose bandwagon-
ing games with submission as their highest-ranked outcome because this outcome
is entailed by ranking mutual cooperation in the form of mutual withdrawal as the
highest-ranked outcome.
Snyder and Diesing (1977) implicitly identify 2 × 2 games of bandwagoning
and balancing theoretically and empirically in the context of intra-alliance rela-
tions of cooperation. Schweller (1994, 1998) and Walker (2007b) address band-
wagoning as important explicitly in the context of extreme power disparity. Still
286 Stephen G. Walker

others have refined variants of different strategies of conflict and cooperation

beyond the logical possibilities circumscribed by ranking the four outcomes pos-
sible with a binary theory (Christensen and Snyder 1990; Schweller 1998, 2006;
Nye 1990; He and Feng 2008; He 2010).
It is possible to revise and expand binary role theory’s present strategic rules to
deal with the phenomena of bandwagoning and balancing processes, and the math-
ematics of these families of sequential games is already worked out (Rapoport and
Guyer 1966; Walker 2007b). It is relatively straightforward to expand the rules of
CUE and TIP in order to distinguish bandwagoning and balancing tactical cues
from cues for other types of strategies, as shown in Figure A.1. However, the fol-
lowing solution omits explicit inclusion of the prisoner’s dilemma and chicken
games, which both game theorists and IR theorists have traditionally considered
substantively important (Snyder 1971; Snyder and Diesing 1977; Brams 1994).
In Figure A.1 Ego pursues either an unconditional or conditional strategy
of submission (bandwagoning), depending on the interests at stake for Ego and
constrained in any event by being less powerful than Alter. If the power distri-
bution between Ego and Alter is equal, Ego can pursue either a strategy of settle-
ment (appeasement) or a strategy of deadlock (balancing) and, depending on the
national interests at stake, prefer submission or domination as the least optimum
outcome. If the power distribution favors Ego over Alter, then Ego can pursue
a strategy of domination (hegemony), an outcome that Ego may either forgo or
exploit if secondary interests are the stakes.
It would appear to be an empirical question as to exactly where in the behav-
ioral field diagram of politics and the cognitive field diagram of beliefs these
expanded rules of inference for CUE and TIP would apply. Candidate loca-
tions in Figure A.2 for bandwagoning and hegemony strategies would be at the
extreme corners of the fields where cooperation or conflict is very high and con-
trol is very asymmetrical between agents, which may be associated with unusual

Grand Cooperation Strategies Grand Conflict Strategies

Bandwagon Appease Balance Hegemony
Submit Bluff Reward Deter Compel Punish Exploit Bully
DED EED DDD DEE EDD EEE DDE EDE
Alter Alter Alter Alter Alter Alter Alter Alter
D E D E D E D E D E D E D E D E
D 3 4 3 4 4 2 4 1 3 2 3 1 4 1 2 1
Ego
E 1 2 2 1 1 3 2 3 1 4 2 4 3 2 4 3
NI Sec Vit Sec Vit Sec Vit Sec Vit
Pow Ego < Alter Ego = Alter Ego = Alter Ego > Alter
E = Escalation; D = De-escalation. Power (Pow) Distributions: Equal (=); Less Than (<);
Greater Than (>). National Interests (NI) for Ego: Sec = Secondary; Vit = Vital. Preference
rankings 4 (highest)…1 (lowest) are for Ego only.

FIGURE A.1 Grand Strategies and Inferred Preferences from Power and Interest
Distributions
 Formal Models of Symbolic and Strategic Interaction 287

constraints or opportunities in the environment for these agents to think and act.
The locations for appeasement and balancing strategies would likely be closer to
the origin formed by the distributions of power and interests at stake because of
the symmetrical power distribution between Ego and Alter.
Computer simulations of games constructed from these new rules may pro-
vide some answers to these questions, as previous simulations have analyzed the

CO (+)
NI (Sec)

A
Sub p Rew Com B Exp
DED B p DDD EDD a H DDE
a e l e
n s a g
d e n e
Pow (<) Pow (>)
w m c m
a e i o
g n Det Pun n n
o t DEE EEE g y
Blu n Bul
EED EDE

NI (Vit)
CF (−)

Role Enactment (TOM) Role Making (TIP) Role Taking (CUE)

1. Either player can choose 1. Bandwagoning Strategy. 1. Bandwagoning Strategy (E < A)
to ‘move’ or ‘stay’ from the a. If Submit (+, <) and NI (Sec): a. If (+, –, +), then (DED) Submit.
‘initial state.’ The player who Sub > Set > Ded > Dom.
moves is P1. Player 2 (P2) can b. If Bluff (+, <) and NI (Vit): b. If (–, –, +), then (EED) Bluff.
respond by choosing ‘move’ Sub > Set > Dom > Ded.
or ‘stay.’

2. Alternating responses continue 2. Appeasement Strategy. 2. Appeasement Strategy (E = A)

until a Player with the next choice a. If Reward (+, =) and NI (Sec): a. If (+, +, +), then (DDD) Reward.
chooses ‘stay,’ terminating the Set > Ded > Sub > Dom.
game at a final outcome state. b. If Deter (+, =) and NI (Vit): b. if (+, –, –), then (DEE) Deter.
Set > Ded > Dom > Sub.

3. A player will not move from the 3. Balancing Strategy. 3. Balancing Strategy (E = A)
initial state if it leads to a less- a. If Compel (–, =) and NI (Sec): a. If (–, +, +), then (EDD) Compel.
preferred state as the final outcome. Ded > Set >Sub > Dom.
b. If Punish (–, =) and NI (Vit): b. If (–, –, –), then (EEE) Punish.
Ded > Set > Dom > Sub.

4. A player will not move from the 4. Hegemony Strategy (E > A) 4. Hegemony Strategy (E > A)
initial state if it leads to a less- a. If Exploit (+, >) and NI (Sec): a. If (+, +, –), then (DDE) Exploit.
preferred final state or cycles play Set > Dom > Ded > Sub.
to the initial state as the final b. If Bully (–, >) and NI (Vit): b. If (–, +, –), then (EDE) Bully.
outcome. Dom > Ded > Set > Sub.

5. If one player chooses ‘stay’

and the other player chooses ‘move’
from the initial state, then ‘move’
overrides ‘stay.’
Set = Settle; Dom = Dominate; Ded = Deadlock; Sub = Submit

FIGURE A.2 Binary Role Theory Locations and Rules for Grand Strategies
288 Stephen G. Walker

conditional conflict and cooperation games of prisoner’s dilemma and stag hunt
(Axelrod 1984; Skyrms 2004; Marfleet and Walker 2006; Mitchell 2009: 218–
220). The three sets of rules in Figure A.2 may relate recursively or non-recur-
sively, depending on the assumptions of causal direction in the analyses. They
specify the interaction of the structure-oriented Past Hypothesis (role demands
rules), an agent-centered Present Hypothesis (role conceptions rules) and the
future-directed First Encounter Hypothesis (role enactments rules). Statistical
models for large-N studies that make either non-recursive or recursive multi-
stage assumptions among the three sets of rules are available (Blalock 1964, 1969;
Thies and Sobek 2010).
It is also possible with small-N, structured-focused case studies to check the fit
between relations specified in these rules by taking successive “cuts” at the case(s)
from the vantage point of the rules for each temporal hypothesis (Allison 1969;
Allison and Zelikow 1999; Walker 2004b; George and Bennett 2005). Such ana-
lytical historical narratives of particular combinations of (a) power and interest
distributions as role demands; (b) beliefs about the utility of cooperation and
conflict strategies and the distribution of historical control as role conceptions; (c)
cues exchanged between Ego and Alter may validate these revised and expanded
rules for modeling TIP and CUE.
Binary role theory’s present CUE rules can be expected to evolve over time
and be flexible in their application to local conditions as have the older rules
for TIP and TOM (Walker and Schafer 2006, 2007; Brams 1975, 1985, 1994).
Following Feynman’s (1995: 23–27) advice of formulating simple rules first, test-
ing them, and evolving more complicated rules in light of empirical research,
the rules of TOM, TIP, and CUE have evolved from simple to more complex
propositions and models and are likely to continue to do so in the future (Laudan
1977; Brams 1994; Walker 2004a, 2004b).
TABLE A.2 Entropy Matrix for Intersections of Subjective Games Specified by Binary Role Theory
APP + APP – REW/ REW/ EXP + EXP – BLUFF + BLUFF – COMP/ COMP/ BULLY + BULLY –
DET + DET – PUN + PUN –

APP + 0|0 0|0 0|0 0|2 0|0 0|0 0|0 0|2 0|0 0|0 0|0 0|2 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|2 0|2 0|2 0|2 0|2
0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|1 0|0 0|1 0|0 0|1 0|0 0|1 0|2 0|1 0|2 0|1
APP – 2|0 1|0 2|0 1|0 2|0 1|0 2|0 1|0 2|0 1|0 2|0 1|0 2|0 1|1 2|0 1|1 2|0 1|1 2|0 1|1 2|1 1|1 2|1 1|1
2|0 2|0 2|0 2|2 2|0 2|0 2|0 2|2 2|0 2|0 2|0 2|2 2|0 2|0 2|0 2|0 2|0 2|0 2|0 2|2 2|1 2|2 2|1 2|2
REW/DET + 0|0 0|0 0|0 0|2 0|0 0|0 0|0 0|2 0|0 0|0 0|0 0|2 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|2 0|2 0|2 0|2 0|2
0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|1 0|0 0|1 0|0 0|1 0|0 0|1 0|2 0|1 0|2 0|1
REW/DET – 0|0 2|0 0|0 2|2 0|0 2|0 0|0 2|2 0|0 2|0 0|0 2|2 0|0 2|0 0|0 2|0 0|0 2|0 0|0 2|2 0|2 2|2 0|2 2|2
0|0 2|0 0|0 2|2 0|0 2|0 0|0 2|2 0|0 2|0 0|0 2|2 0|0 2|0 0|0 2|0 0|0 2|0 0|0 2|2 0|2 2|2 0|2 2|2
EXP + 0|0 0|0 0|0 0|2 0|0 0|0 0|0 0|2 0|0 0|0 0|0 0|2 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|2 0|2 0|2 0|2 0|2
0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|1 0|0 0|1 0|0 0|1 0|0 0|1 0|2 0|1 0|2 0|1
EXP – 0|0 3|0 0|0 3|2 0|0 3|0 0|0 3|2 0|0 3|0 0|0 3|2 0|0 3|0 0|0 3|0 0|0 3|0 0|0 3|2 0|2 3|2 0|2 3|2
0|0 0|0 0|0 0|2 0|0 0|0 0|0 0|2 0|0 0|0 0|0 0|2 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|2 0|2 0|2 0|2 0|2
BLUFF + 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|0 0|1 0|0 0|1 0|0 0|1 0|0 0|1 0|1 0|1 0|1 0|1
1|0 0|0 1|3 0|0 1|0 0|0 1|0 0|0 1|0 0|0 1|0 0|0 1|3 0|1 1|3 0|1 1|1 0|1 1|1 0|1 1|1 0|1 1|1 0|1
BLUFF – 1|0 1|0 1|0 1|0 1|0 1|0 1|0 1|0 1|0 1|0 1|0 1|0 1|0 1|1 1|0 1|1 1|0 1|1 1|0 1|1 1|1 1|1 1|1 1|1
1|0 1|0 1|3 1|0 1|0 1|0 1|0 1|0 1|0 1|0 1|0 1|0 1|3 1|1 1|3 1|1 1|1 1|1 1|1 1|1 1|1 1|1 1|1 1|1
COMP/ 0|0 2|2 0|0 2|2 0|0 2|2 0|0 2|2 0|0 2|2 0|0 2|2 0|0 2|3 0|0 2|3 0|0 2|2 0|0 2|2 0|2 2|2 0|2 2|2
PUN + 2|0 0|0 2|3 0|0 2|0 0|0 2|0 0|0 2|0 0|0 2|0 0|0 2|3 0|1 2|3 0|1 2|1 0|1 2|1 0|1 2|1 0|1 2|1 0|1
COMP/ 0|0 2|2 0|0 2|2 0|0 2|2 0|0 2|2 0|0 2|2 0|0 2|2 0|0 2|3 0|0 2|3 0|0 2|2 0|0 2|2 0|2 2|2 0|2 2|2
PUN – 2|3 2|2 2|3 2|2 2|3 2|2 2|2 2|2 2|3 2|2 2|2 2|2 2|3 2|3 2|3 2|3 2|3 2|2 2|2 2|2 2|2 2|2 2|2 2|2
BULLY + 2|3 2|2 2|3 2|2 2|3 2|2 2|2 2|2 2|3 2|2 2|2 2|2 2|3 2|3 2|3 2|3 2|3 2|2 2|2 2|2 2|2 2|2 2|2 2|2
3|3 2|2 3|3 2|2 3|3 2|2 3|2 2|2 3|3 2|2 3|2 2|2 3|3 2|3 3|3 2|3 3|3 2|2 3|2 2|2 3|2 2|2 3|2 2|2
BULLY – 3|3 3|2 3|3 3|2 3|3 3|2 3|2 3|2 3|3 3|2 3|2 3|2 3|3 3|3 3|3 3|3 3|3 3|2 3|2 3|2 3|2 3|2 3|2 3|2
3|3 3|2 3|3 3|2 3|3 3|2 3|2 3|2 3|3 3|2 3|2 3|2 3|3 3|3 3|3 3|3 3|3 3|2 3|2 3|2 3|2 3|2 3|2 3|2

Note: Cells indicate Expected and Actual (E | A) outcome for the row player by quadrant (see Figure 14.5 for quadrant numbering convention) of the interaction between paired agents
moving according to the logic dictated by their subjective games. Cells are grouped into blocks of four representing the four possible starting quadrants in a normal game. The proportion of
outcomes that are correctly anticipated by the row player is 56% over 576 possible configurations of agents and starting quadrants.
Source: Adapted from Marfleet and Walker (2006: 64–65).
BIBLIOGRAPHY

Abelson, R. (1967) “Modes of Resolution of Belief Dilemmas,” in M. Fishbein (ed.)

Readings in Attitude Theory and Measurement, New York: John Wiley and Sons.
Achen, C. (1982) “Interpreting and Using Regression,” in J. Sullivan and R. Niemi (eds.)
Series: Quantitative Applications in the Social Sciences, Newbury Park, CA: Sage.
Adler, R. (2002) Science Firsts, New York: John Wiley and Sons.
Ahn, B. (1992) “The Possibilities of Change in North Korea,” Korea and World Affairs,
16: 421–435.
Ahn, B. (1997) “Constraints and Objectives of North Korean Foreign Policy,” in T.
Henriksen and J. Mo (eds.), North Korea after Kim Il Sung, Stanford, CA: Hoover
Institution Press.
Allison, G. (1969) “Conceptual Models and the Cuban Missile Crisis,” American Political
Science Review, 63: 689–718.
Allison, G. and Halperin, M. (1972) “Bureaucratic Politics,” World Politics, 23(Suppl.):
40–79.
Allison, G. and Zelikow, P. (1999) Essence of Decision, 2nd edn, New York:
Longman.
Alker, H. (1966) “The Long Road to International Relations Theory,” World Politics, 18:
623–655.
Almond, G. and Genco, S. (1977) “Clouds, Clocks, and the Study of Politics,” World
Politics, 29: 489–522.
Anderson, C. (1983) “Abstract and Concrete Data in the Perseverance of Social Theories,”
Journal of Experimental Social Psychology, 19: 93–108.
Anderson, C., Lepper, M., and Ross, L. (1980) “Perseverance of Social Theories,” Journal
of Personality and Social Psychology, 39: 1037–1049.
Anderson, J. (2003) “American Hegemony after 11 September,” Geopolitics, 8: 35–60.
Appleton, S. (1968) United States Foreign Policy, Boston: Little, Brown.
Archer, M. 1988) Culture and Agency, New York: Cambridge University Press.
Ashby, R. (1960) Design for a Brain, New York: John Wiley.
 Bibliography 291

Ashley, R. (1981) “Political Realism and Human Interests,” International Studies Quarterly,
25: 204–236.
Ashley, R. (1984) “The Poverty of Neorealism,” International Organization, 38: 225–286.
Axelrod, R. (1984) The Evolution of Cooperation, New York: Basic Books.
Axelrod, R. and Cohen, M. (1999) Harnessing Complexity, New York: Free Press.
Azar, E. (ed.) (1975) Theory and Practice of Events Research, New York: Gordon and
Breach.
Bachrach, P. and Baratz, M. (1962) “The Two Faces of Power,” American Political Science
Review, 56: 947–952.
Bailey, T. (1958) A Diplomatic History of the American People, New York: Appleton-Century
Crofts.
Baldwin, D. (1989) Paradoxes of Power, New York: Blackwell.
Baldwin, D. (1993) Neoliberalism and Neorealism, New York: Columbia University.
Baldwin, D. (2002) “Power and International Relations,” in W. Carlsnaes, T. Risse and
B. Simmons (eds.) Handbook of International Relations, London: Sage.
Bales, R. (1950) Interaction Process Analysis, Cambridge, MA: Addison-Wesley.
Ball, T. (1987) “Deadly Hermeneutics,” in T. Ball (ed.) Idioms of Inquiry, New York:
SUNY Press.
Bandura, A. (1969) Principles of Behavior Modification, New York: Holt, Rinehart, and
Winston.
Barnet, R. (1972) Roots of War, New York: Athenum.
Barrow-Green, J. (1997) Poincare and the Three Body Problem, Providence, RI: American
Mathematical Society.
Beal, T. (2005) North Korea, London and Ann Arbor: Pluto Press.
Beck, N. and Katz, J. (1995) “What To Do (and Not To Do) with Time-Series Cross-
Section Data,” American Political Science Review, 89: 634–647.
Becker, J. (2005) Rogue Regime, Oxford: Oxford University Press.
Bechtol, B. (2007) Red Rogue, Washington, DC: Potomac Books.
Beer, F. and De Landtsheer, C. (eds.) (2004) Metaphorical World Politics, East Lansing:
Michigan State University Press.
Bem, D. (1970) Beliefs, Attitudes, and Human Affairs, Belmont, CA: Brooks/Cole Press.
Ben-Zvi, A. (1978) “The Outbreak and Termination of the Pacific War,” Journal of Peace
Research, 15: 33–49.
Bennett, A. (1999) Condemned to Repetition? Cambridge, MA: MIT Press.
Berkowitz, L. (ed.). (1969) Roots of Aggression, New York: Atherton Press.
Berkowitz, L. (1978) “Whatever Happened to the Frustration-Aggression Hypothesis?”
American Behavioral Scientist, 21: 691–707.
Berkowitz, L. (1989) “Frustration-Aggression Hypothesis,” Psychological Bulletin, 106: 59.
Bertalanffy, L. and Rapoport, A. (eds.) (1956) General Systems, Vol. 1, Ann Arbor:
University of Michigan Press.
Biddle, B. (1979) Role Theory, New York: Academic Press.
Blacker, C. (1993) Hostage to Revolution, New York: Council on Foreign Relations.
Blair, T. and Chirac, J. (2003) Press Conference in Le Touquest, France (4 February).
Available at: http://www.lexisnexis.com/
Blalock, H. (1964) Causal Inferences in Non-Experimental Research, Chapel Hill: University
of North Carolina Press.
292 Bibliography

Blalock, H. (1969) Theory Construction, Englewood Cliffs, NJ: Prentice-Hall.

Blasier, C. (1993) “The End of the Soviet-Cuban Partnership,” in C. Mesa-Lago (ed.)
Cuba After the Cold War, Pittsburgh, PA: University of Pittsburgh Press.
Blau, P. (1964) Exchange and Power in Social Life, New York: John Wiley.
Block, F. (1981) “Unemployment: Causes and Cures,” CATO Policy Analysis (4). Available:
http://www.cato.org/pubs/pas/pa004.html
Bobrow, D., Chan, S., and Kringen, J. (1979) Understanding Foreign Policy Decisions, New
York: Free Press.
Boulding, K. (1959) “National Images and International Systems,” Journal of Conflict
Resolution, 3: 120–131.
Boulding, K. (1962) Conflict and Defense, New York: Harper and Row.
Boulding, K. (1969) “National Images and International Systems,” in James N. Rosenau
(ed.) International Politics and Foreign Policy, New York: Free Press.
Bowers, P. et al. (2003) Iraq: Developments since UN Security Council Resolution 1441,
International Affairs and Defence Section, London: House of Commons Library.
Brams, S. (1975) Game Theory and Politics, New York: Free Press.
Brams, S. (1985) Rational Politics, Boston: Academic Press.
Brams, S. (1994) Theory of Moves, Cambridge: Cambridge University Press.
Brams, S. (2002) “Game Theory in Practice,” in M. Brecher and F. Harvey (eds.) Millennial
Reflections on International Studies, Ann Arbor: University of Michigan Press.
Brams, S. and Kilgour, M. (1986) “Is Nuclear Deterrence Rational?” Political Science, 19:
645–651.
Brecher, M. and Wilkenfeld, J. (2000) A Study of Crisis, Ann Arbor: University of
Michigan.
Brecher, M., Steinberg, B., and Stein, J. (1969) “A Framework for Research on Foreign
Policy Behavior,” Journal of Conflict Resolution, 12: 75–101.
Brecher, M., Wilkenfeld, J., and Moser, S. (1988) Crises in the Twentieth Century, Oxford:
Pergamon.
Brener, P., Haney, P., and Vanderbush, W. (2004) “Intermestic Interests and U.S. Policy
Toward Cuba,” in E. Wittkopf and J. McCormick (eds.) The Domestic Sources of American
Foreign Policy, Lanham, MD: Rowman & Littlefield.
Brenner, M. (2003) “The CFSP Factor. A Comparison of United States and French
Strategies,” Cooperation and Conflict, 38: 187–209.
Breslauer, G. (1987) “Ideology and Learning in Soviet Third World Policy,” World Politics,
39: 429–448.
Breslauer, G. (1992) “Explaining Soviet Policy Changes,” in G. Breslauer (ed.) Soviet Policy
in Africa, Berkeley: University of California Press.
Breslauer, G. and Tetlock, P. (1991) Learning in U.S. and Soviet Foreign Policy, National
Research Council (U.S.), Committee on International Conflict and Cooperation,
Boulder, CO: Westview.
Brown, C. (2002) “Narratives of Religion, Civilization and Modernity,” in K. Booth and
T. Dunne (eds.) Worlds in Collision, New York: Palgrave Macmillan.
Bruni, F. (2002) Ambling into History, New York: HarperCollins.
Brush, S. (1988) The History of Modern Science, Ames: Iowa State University Press.
Bueno de Mesquita, B. (1981) The War Trap, New Haven, CT: Yale University
Press.
 Bibliography 293

Bueno de Mesquita, B. and Lalman, D. (1992) War and Reason, New Haven: Yale
University Press.
Bueno de Mesquita, B. and McDermott, R. (2004) “Crossing No Man’s Land,” Political
Psychology, 25: 275–287.
Bueno de Mesquita, B. et al. (2003) The Logic of Political Survival, Cambridge, MA: MIT
Press.
Bunge, M. (1967) Scientific Research of Studies of the Foundations, Methodology, and Philosophy
of Science, Berlin: Springer-Verlag.
Burgess, P. and Moore, D. (1973) “Inter-nation Alliances,” in J. A. Robinson (ed.) Political
Science Annual, Vol. 3, Indianapolis: Bobbs-Merrill.
Burke, J. (2004) Becoming President, Boulder, CO: Lynne Rienner.
Burt, R. (1992) Structural Holes, Cambridge, MA: Harvard University Press.
Burton, J. (1969) Conflict and Communication, New York: Macmillan.
Bush, G. (2001) Address to a Joint Session of Congress and the American People (September).
Available at: http://www.whitehouse.gov/news/releases/2001/09/20010920-8.html
Bush, G. (2002) Address to the United Nations General Assembly on 12 September.
Available at: http://www.whitehouse.gov/news/releases/2002/09/20020912-1.html
Bush, G. (2004) “We Will Finish the Work of the Fallen,” New York Times, April 14:
A12.
Buzo, A. (1999) The Guerrilla Dynasty, Boulder, CO: Westview.
Camerer, C. (2003) Behavioral Game Theory, Princeton, NJ: Princeton University Press.
Campbell, D. (1969) “Variation and Selective Retention in Socio-cultural Evolution,”
General Systems Yearbook, 14: 69–85.
Campbell, D. (1998) Writing Security, Minneapolis: University of Minnesota Press.
Caplow, T. (1968) Two against One, Englewood Cliffs, NJ: Prentice-Hall.
Carlsnaes, W. (1992) “The Agency-Structure Problem in Foreign Policy Analysis,”
International Studies Quarterly, 26: 245–270.
Carney, T. (1972) Content Analysis, Winnipeg: University of Manitoba Press.
Carretta, T. and Moreland, R. (1982) “Nixon and Watergate,” Personality and Social
Psychology Bulletin, 8: 446–453.
Carroll, S. (2010) From Eternity to Here, New York: Dutton.
Carter, W. (2002) “Back to the Brink,” Washington Post, Oct. 20.
Centeno, M. (1997) “Cuba’s Search for Alternatives,” in M. Centeno and M. Font (eds.)
Toward a New Cuba? Boulder, CO: Lynne Rienner.
Central Intelligence Agency (1992) Cuba: Handbook of Trade Statistics, Washington, DC:
CIA.
Central Intelligence Agency (1998) Cuba: Handbook of Trade Statistics, Washington, DC:
CIA.
Cha, V. (2002) “Hawk Engagement and Preventive Defense of the Korean Peninsula,”
International Security, 27: 40–78.
Cha, V. and Kang, D. (2003) North Korea: A Debate on Engagement Strategies, New York:
Columbia University Press.
Chirac, J. (2003) “La France est contre une nouvelle resolution, quelleque soient les
Circonstances,” Le Monde, 11 March.
Christensen, E. and Redd, S. (2004) “Bureaucrats Versus the Ballot Box in Foreign Policy
Decision Making,” Journal of Conflict Resolution, 48: 69–91.
294 Bibliography

Christensen, T. and Snyder, J. (1990) “Chain Gangs and Passing Bucks,” International
Organization, 44: 137–167.
Chwe, M. (2001) Rational Ritual, Princeton, NJ: Princeton University Press.
Collier, K. and Sullivan, T. (1995) “New Evidence Undercutting the Linkage of Approval
with Presidential Support and Influence,” Journal of Politics, 52: 939–963.
Cook, T. and Campbell, D. (1979) Quasi-Experimentation, Chicago: Rand McNally.
Cortright, D., Millar, A., Lopez, G., and Gerber, L. (2003) Unproven: the Controversy
over Justifying War In Iraq, The Joan B. Kroc Institute for International Peace Studies,
University of Notre Dame.
Coser, L. (1956) The Function of Social Conflict, New York: Free Press.
Cottam, M. (1986) Foreign Policy Decision Making, Boulder, CO: Westview.
Cottam, M. (1994) Images and Intervention, Pittsburgh, PA: University of Pittsburgh Press.
Cottam, R. (1977) Foreign Policy Motivation, Pittsburgh, PA: University of Pittsburgh
Press.
Cotton, J. (1998) “The Raijin-Sonbong Free Trade Zone Experiment,” in S. Kim (ed.)
North Korean Foreign Relations in the Post-Cold War Era, New York: Oxford University
Press.
Cowen, T. (1998) “Do Economists Use Social Mechanisms to Explain?” in P. Hedstrom
and R. Swedberg (eds.) Social Mechanisms, Cambridge: Cambridge University Press,
125–146.
Cox, M. (2003) “Martians and Venutians in the New World Order,” International Affairs,
79: 523–532.
Crichlow, S. (1998) “Idealism or Pragmatism? An Operational Code Analysis of Yitzhak
Rabin and Shimon Peres,” Political Psychology, 19: 683–706.
Cumings, B. (2004) North Korea: Another Country, New York: New Press.
Cumings, B. (2005) Korea’s Place in the Sun, New York: W. W. Norton.
Cummins, R. (1983) The Nature of Psychological Explanation, Cambridge, MA: MIT
Press.
Daalder, I. (2003) “The End of Atlanticism,” Survival, 45: 147–166.
Daalder, I. and Gordon, P. (2002) Euro Trashing. Washington Post Online, 29 May.
Available at: http://www.brook.edu/views/op-ed/daalder/20020529.htm
Daalder, I. and Lindsay, J. (2003a) America Unbound, Washington, DC: Brookings Press.
Daalder, I. and Lindsay, J. (2003b) “Bush’s Foreign Policy Revolution,” in F. Greenstein
(ed.) The George W. Bush Presidency, Baltimore: Johns Hopkins University Press,
100–137.
Dahl, R. (1957) “The Concept of Power,” Behavioral Science, 2: 201–215.
Dahrendorf, R. (1959) Class and Class Conflict in Industrial Society, Stanford, CA: Stanford
University Press.
Dahrendorf, R. (1968) Essays in the Theory of Society, Stanford, CA: Stanford University
Press.
Daley, S. (2001) “For First Time, NATO Invokes Joint Defence Pact with OS,” New York
Times, 12 September.
De Castro, R. (2000) “Whither Geoeconomics?” Asian Affairs: An American Review, 26:
201–221.
De Jonge Qudraat, C. (2003) “Combating Terrorism,” The Washington Quarterly, 26:
163–176.
 Bibliography 295

Delpech, T. (2002) “International Terrorism and Europe,” Chailot Papers 56, Institute for
Security Studies, Paris. Available at: http://aei.pitt.edu/518/
Denzau, A. and North, D. (1994) “Shared Mental Models,” Kyklos, 47: 3–31.
Der Derian, J. and Shapiro, M. (1989) International/Intertextual Relations, Lexington, MA:
Lexington Books.
Dessler, D. (2004) “Explanation and Scientific Progress,” in C. Elman and M. Elman (eds.)
Progress in International Relations Theory, Cambridge, MA: MIT Press.
Deutsch, K. (1966) The Nerves of Government, London: Free Press.
Deutsch, K. (1968) The Analysis of International Relations, Englewood Cliffs, NJ:
Prentice-Hall.
Deutsch, K. and Merritt, R. (1965) “The Effects of Events on National and International
Images,” in H. Kelman (ed.) International Behavior, New York: Holt, Rinehart and
Winston.
Deutsch, K. and Singer, D. (1964) “Mulipolar Power Systems and International Stability,”
World Politics, 16: 390–406.
Diamond, J. (1992) The Third Chimpanzee, New York: Harper-Collins.
Dille, B. (2000) The Prepared and Spontaneous Remarks of Presidents Reagan and Bush,”
Political Psychology, 21: 573–585.
Dollard, J., Dobb, L. Miller, N, Mowrer, O., and Sears, R. (1939) Frustration and Aggression,
New Haven, CT: Yale University Press.
Domínguez, J. (1989) To Make a World Safe for Revolution, Cambridge, MA: Harvard
University Press.
Domínguez, J. 1993) “The Political Impact on Cuba of the Reform and Collapse of
Communist Regimes,” In C. Mesa-Lago (Ed.), Cuba After the Cold War. Pittsburgh, PA:
University of Pittsburgh Press.
Domínguez, J. (1997) “US-Cuban Relations,” Journal of Interamerican Studies and World
Affairs, 39, 49–76.
Doran, C. (1991) Systems in Crisis, New York: Cambridge University Press.
Doty, R. (1996) Imperial Encounters, Minneapolis, MN: University of Minnesota Press.
Downs, G. (1989) “The Rational Deterrence Debate,” World Politics, 41: 225–237.
Downs, G. and Rocke, D. (1994) “Conflict, Agency, and Gambling for Resurrection,”
American Journal of Political Science, 38: 362–380.
Doyle, M. (1986) “Liberalism and World Politics,” American Political Science Review, 80:
1151–1169.
Draper, R. (2007) Dead Certain, New York: Free Press.
Drezner, D. (2000) “Ideas, Bureaucratic Politics, and the Crafting of Foreign Policy,”
American Journal of Political Science, 44: 733–750.
Duke, S. (2003) “The Hyperpower and the Hype,” European Institute of Public
Administration, Working Papers No. 2003/W/1.
Dunne, M. (2003) “The United States, the United Nations and Iraq,” International Affairs,
79: 257–277.
Dyson, S. and Preston, T. (2006) “Individual Characteristics of Political Leaders and the
Use of Analogy in Foreign Policy Decision Making,” Political Psychology, 27: 265–288.
Easton, D. (1953) The Political System, New York: Alfred Knopf.
Eberstadt, N. and Ellings, R. (eds.) (2001) Korea’s Future and the Great Powers, Seattle, WA:
National Bureau of Asian Research.
296 Bibliography

Eckstein, H. (1975) “Case Study and Theory in Political Science,” in F. Greenstein and
N. Polsby (eds.) Handbook of Political Science, Vol. 7, Reading, MA: Addison-Wesley.
Edelman, G. (1992) Bright Air, Brilliant Fire, New York: Basic Books.
Eden, A. (1960) Full Circle, Boston: Houghton-Mifflin.
Ellison, H. (2001) “Russia, Korea, and Northeast Asia,” in N. Eberstadt and R. Ellings
(eds.) Korea”s Future and the Great Powers, Seattle, WA: National Bureau of Asian
Research.
Elman, C. (1996) “Horses for Courses,” Security Studies, 6: 7–53.
Elman, C. and Elman, M. (eds.) (2001) Bridges and Boundaries, Cambridge, MA: MIT
Press.
Elman, C. and Elman, M. (eds.) (2003) Progress in International Relations Theory, Cambridge,
MA: MIT Press.
Elman, M. (1995) “The Foreign Policies of Small States,” British Journal of Political Science,
25: 171–217.
Elster, J. (1998) “A Plea for Mechanisms,” in R. Hedstrom and E. Swedberg (eds.) Social
Mechanisms, Cambridge: Cambridge University Press.
Emerson, R. (1972a) “Exchange Theory Part I,” in J. Berger, M. Zelditch, and
B. Anderson (eds.) Sociological Theories in Progress, Boston: Houghton-Mifflin.
Emerson, R. (1972b) “Exchange Theory Part II,” in J. Berger, M. Zelditch, and
B. Anderson (eds.) Sociological Theories in Progress, Boston: Houghton-Mifflin.
Erisman, H. (1985) Cuba’s International Relations, Boulder, CO: Westview.
Erisman, H. (2000) Cuba’s Foreign Relations in a Post-Soviet World, Gainesville: University
Press of Florida.
Etheredge, L. (1985) “Government Learning,” in S. Long (ed.) The Handbook of Political
Behavior, Vol. 2, New York: Plenum Press.
Eulau, H. (1963) The Behavioral Persuasion in Politics, New York: Random House.
Evangelista, M. (1991) “Sources of Moderation in Soviet Security Policy,” in P. Tetlock et
al. (eds.) Behavior, Society, and Nuclear War, New York: Oxford University Press.
Farhang, M. (1993) “U.S. Policy Toward the Islamic Republic of Iran,” in Hooshang
Amirahmadi (ed.) The United States and the New Middle East, New York: SUNY Press.
Farkas, A. (1996) “Evolutionary Models in Foreign Policy Analysis,” International Studies
Quarterly, 40: 343–361.
Farnham, B. (ed.) (1975) Avoiding Losses/Taking Risks, Ann Arbor: University of Michigan
Press.
Farnham, B. (ed.) (1994) Avoiding Losses/Taking Risks: Prospect Theory and International
Conflict, Ann Arbor: University of Michigan Press.
Farnham, B. (ed.) (1997) Roosevelt and the Munich Crisis, Princeton, NJ: Princeton
University Press.
Feng, H. (2005a) “The Operational Code of Mao Zedong,” Security Studies, 14:
637–662.
Feng, H. (2005b) “The Operational Codes of Four Generations of Chinese Leaders,”
paper presented at the annual meeting of the American Political Science Association,
September 1–4, Washington, DC.
Feng, H. (2007) Chinese Strategic Culture and Foreign Policy Decision-Making, New York:
Routledge.
Fernández, D. (1994) “Continuity and Change in Cuba’s International Relations in the
 Bibliography 297

1990s,” in J. Pérez-López (ed.) Cuba at a Crossroads, Gainesville: University Press of

Florida.
Festinger, L. (1957) Theory of Cognitive Dissonance, Evanston, IL: Row, Peterson.
Feynman, R. (1995) Six Easy Pieces, Reading, MA: Addison-Wesley.
Fishbein, M. (1967) “A Consideration of Beliefs, and Their Role in Attitude Measurement,”
in M. Fishbein (ed.) Readings in Attitude Theory and Measurement, New York: Wiley.
Fiske, S. and Taylor, S. (1991) Social Cognition, New York: McGraw-Hill.
Forward, N. (1971) The Field of Nations, Boston, MA: Little, Brown.
Foyle, D. (1999) Counting the Public In, New York: Columbia University Press.
Franck, T. and Weisband, E. (1971) Word Politics, New York: Oxford University Press.
Frauenfelder, H. et al. (2009) “A Unified Model of Protein Dynamics,” PNAS, 13:
129–134.
Freedman, L. and Karsh, E. (1993) The Gulf Conflict, 1990–1991, Princeton, NJ: Princeton
University Press.
French, J. and Raven, B. (1959) “Bases of Social Power,” in D. Cartwright (ed.) Studies in
Social Power, Ann Arbor: University of Michigan Press.
French, P. (2007) North Korea: The Paranoid Peninsula, London and New York: Zed
Books.
Freud, S. and Bullitt, W. (1967) Thomas Woodrow Wilson, London: Weidenfeld and
Nicholson.
Frieden, J. (1999) “Actors and Preferences in International Relations,” in D. Lake and R.
Powell Strategic Choice and International Relations, Princeton, NJ: Princeton University
Press.
Fritz, C. and Williams, H. (1957) “The Human Being in Disasters,” Annals, 309: 42–51.
Frum, D. (2003) The Right Man, New York: Random House.
Gaddis, J. (1992) “The Cold War’s End Dramatizes the Failure of Political Theory,”
Chronicle of Higher Education, 23: A44.
Galhofer, I. and Saris, W. (1979) “Strategy Choices of Foreign Policy Decision Makers,”
Journal of Conflict Resolution, 23: 425–445.
Galtung, J. (1967) Theory and Methods of Social Research, New York: Columbia University
Press.
Galtung, J. (1967) “On the Effects of International Economic Sanctions,” World Politics,
19: 378–416.
Garner, W. (1962) Uncertainty and Structure as Psychological Concepts, New York: John
Wiley.
Geddes, B. (1991) “How the Cases You Choose Affect the Answers You Get,” in J.
Stimson (ed.) Political Analysis, Ann Arbor: University of Michigan Press.
Gelb, L. (2010) Power Rules, New York: Harper.
George, A. (1969) “The ‘Operational Code,’” International Studies Quarterly, 13: 190–222.
George, A. (1979) “The Causal Nexus between Cognitive Beliefs and Decision-making
Behavior,” in L. Falkowski (ed.) Psychological Models in International Politics, Boulder,
CO: Westview.
George, A. (1980) Presidential Decisionmaking in Foreign Policy, Boulder, CO: Westview.
George, A. (1993) Bridging the Gap, Washington DC: United States Institute for Peace.
George, A. and Bennett, A. (2005) Case Studies and Theory Development, Cambridge, MA:
MIT Press.
298 Bibliography

George, A. and George, J. (1956) Woodrow Wilson and Colonel House, New York: John
Day.
George, A. and George, J. (1981/82) “Woodrow Wilson and Colonel House: A Reply to
Weinstein, Anderson, and Link,” Political Science Quarterly, 96: 641–655.
Gerson, M. (2007) Heroic Conservatism, New York: Harper One.
Geva, N. and Mintz, A. (eds.) (1997) Decisionmaking on War and Peace, Boulder, CO:
Lynne Rienner.
Gilpin, R. (1981) War and Change in World Politics, New York: Cambridge University
Press.
Glennon, M. (2003) “Why the Security Council Failed,” Foreign Affairs, 82: 16–35.
Goffman, E. (1959) The Presentation of Self In Everyday Life, New York: Doubleday.
Goffman, E. (1969) Strategic Interaction, Philadelphia, PA: University of Philadelphia Press.
Goh, E. (2003) “Hegemonic Constraints,” Australian Journal of International Affairs, 57:
77–97.
Goldstein, H., Poole, C., and Safko, J. (2002) Classical Mechanics, San Francisco: Addison-
Wesley.
Goldstein, J. (1992) “A Conflict-Cooperation Scale for WEIS Events Data,” Journal of
Conflict Resolution, 36: 369–385.
Goldstein, J. (1995) “Great-Power Cooperation Under Conditions of Limited Reciprocity,”
International Studies Quarterly, 39: 453–478.
Goldstein, J. and Freeman, J. (1990) Three-Way Street, Chicago: University of Chicago
Press.
Goldstein, J. and Keohane, R. (eds.) (1993) Ideas and Foreign Policy, Ithaca, NY: Cornell
University Press.
Goldstein, J. and Pevehouse, J. (1997) “Reciprocity, Bullying, and International
Cooperation,” American Political Science Review, 91: 515–529.
Gonzalez, E. (1996) Cuba: Clearing Perilous Waters? Santa Monica, CA: Rand.
Gonzalez, E. and Ronfeldt, D. (1986) Castro, Cuba, and the World, Santa Monica, CA:
Rand.
Gordon, P. (2001/2002) “NATO After 11 September,” Survival, 43: 69–106.
Gordon, P. (2002) France, the United States and the “War on Terrorism,” The Brookings
Institution. Article available at: http://www.brook.edu/fp/cusf/analysis/terrorism.htm
Gowa, J. (1994) Allies, Adversaries and International Trade, Princeton, NJ: Princeton
University Press.
Green, D. and Shapiro, I. (1994) The Pathologies of Rational Choice, New Haven, CT: Yale
University Press.
Greenstein, F. (1967) “The Impact of Personality on Politics,” American Political Science
Review, 61: 629–641.
Greenstein, F. (1969) Personality and Politics; Problems of Evidence, Inference, and
Conceptualization, Chicago: Markham.
Gulick, E. (1955) Europe’s Classical Balance of Power, Ithaca, NY: Cornell University Press.
Guttieri, K., Wallace, M., and Suedfeld, P. (1995) “The Integrative Complexity of
American Decision Makers in the Cuban Missile Crisis,” Journal of Conflict Resolution,
39: 595–621.
Habermas, J. (1979) Communication and the Evolution of Society, Boston: Beacon Press.
Habermas, J. (1984) The Theory of Communicative Action, Boston: Beacon Press.
 Bibliography 299

Hamblin, R. (1958) “Group Integration during a Crisis,” Human Relations, 11: 67–76.
Haney, P. (1997) Organizing for Foreign Policy Crises, Ann Arbor: University of Michigan
Press.
Hanrieder, W. (1967) “Compatibility and Consensus,” American Political Science Review,
61: 971–982.
Harary, F. (1961) “A Structural Analysis of the Situation in the Middle East in 1956,”
Journal of Conflict Resolution, 5: 167–178.
Harary, F. (1969) Graph Theory, Reading, MA: Addison-Wesley.
Harnisch, S. (2010) “Dialogue and Emergence: George Herbert Mead’s Contribution
to Role Theory and his Reconstruction of International Politics,” paper presented at
the Annual Meeting of the International Studies Association, New Orleans, February
16–20.
Harrison, S. (2002) Korean Endgame, Princeton, NJ: Princeton University Press.
Hart, P. ‘t (1990) Groupthink in Government, Amsterdam: Swets & Zeitlinger.
Hartmann, F. (1982) The Conservation of Enemies, Westport, CT: Greenwood.
Hartmann, F. (1983) The Relations of Nations, New York: Macmillan.
Hawking, S. with Mlodinow, L. (2005) A Briefer History of Time, New York: Random
House.
He, K. (2009) Institutional Balancing in the Asia Pacific, New York: Routledge.
He, K. (2010) “Undermining Adversaries,” paper presented at the Annual Meeting of the
International Studies Association, New Orleans, LA.
He, K. and Feng, H. (2008) “If Not Soft-Balancing, Then What?” Security Studies, 17:
363–395.
Heclo, H. (2003) “The Political Ethos of George W. Bush,” in F. Greenstein (ed.) The
George W. Bush Presidency, Baltimore, MD: Johns Hopkins University Press.
Hedström, P. and Swedberg, R. (eds.) (1998) Social Mechanisms, Cambridge: Cambridge
University Press.
Hegel, G. (1894) Lectures on the Philosophy of History, London: Bell Press.
Heider, F. (1958) The Psychology of Interpersonal Relations, New York: John Wiley.
Henriksen, T. (1997) “North Korea’s Two Kims and American Foreign Policy,” in
T. Henriksen and J. Mo (eds.) North Korea after Kim Il Sung, Stanford, CA: Hoover
Institution Press.
Hermann, C. (1969) Crises in Foreign Policy, Indianapolis, IN: Bobbs-Merrill.
Hermann, C. (1971) “What is a Foreign Policy Event?” in W. Hanrieder (ed.) Comparative
Foreign Policy, New York: David McKay.
Hermann, C. (1972) International Crises, New York: Free Press.
Hermann, C. (1979) “Some Issues in the Study of International Crises,” in C. Hermann
(ed.) International Crises, New York: Free Press.
Hermann, C. (1990) “Changing Course,” International Studies Quarterly, 34: 3–21.
Hermann, M. (1980) “Explaining Foreign Policy Behavior Using the Personal
Characteristics of Political Leaders,” International Studies Quarterly, 24: 7–46.
Hermann, M. (1984) “Personality and Foreign Policy Decision Making,” in D. Sylvan and
S. Chan (eds.) Foreign Policy Decision Making, New York: Praeger.
Hermann, M. (1999) “Assessing Leadership Style,” Social Sciences Automation.
Hermann, M. (2001) “How Decision Units Shape Foreign Policy,” International Studies
Review, 3: 47–81.
300 Bibliography

Hermann, M. (2003) “Assessing Leadership Style,” in J. Post (ed.) The Psychological

Assessment of Political Leaders, Ann Arbor: University of Michigan Press.
Hermann, M. and Hermann, C. (1989) “Who Makes Foreign Policy Decisions and How,”
International Studies Quarterly, 33, 361–388.
Hernández-Catá, E. (2000) “The Fall and Recovery of the Cuban Economy in the 1990s,”
in Cuba in Transition, Vol. 10, Papers and Proceedings of the 10th Annual Meeting of
the Association for the Study of the Cuban Economy, Miami, FL, available at: http://
lanic.utexas.edu
Herrmann, R. (1985) Perceptions and Behavior in Soviet Foreign Policy, Pittsburgh, PA:
University of Pittsburgh Press.
Herrmann, R. (2003) “Image Theory and Strategic Interaction in International Relations,”
in D. Sears, L. Huddy, and R. Jervis (eds.) Oxford Handbook of Political Psychology, Oxford:
Oxford University Press.
Herrmann, R. and Fischerkeller, M. (1995) “Beyond the Enemy Image and Spiral Model,”
International Organization, 49: 415–450.
Herrmann, R. and Kun Choi, J. (2007) “From Prediction to Learning,” International
Security, 31: 132–161.
Herrmann, R. and Reese, M. (2004) “George W. Bush’s Foreign Policy,” in C. Campbell
and B. Rockman (eds.) The George W. Bush Presidency, Washington, DC: CQ Press.
Herrmann, R., Tetlock, P., and Visser, P. (1999) “Mass Public Decisions to Go to War,”
American Journal of Political Science, 93: 553–573.
Herrmann, R., Voss, J., Schooler, T., and Ciarrochi, J. (1997) “Images in International
Relations,” International Studies Quarterly, 41: 403–433.
Hewitt, J. (1988) Self and Society, Boston: Allyn and Bacon.
Hoffmann, S. (1960) Contemporary Theory in International Relations, Englewood Cliffs, NJ:
Prentice-Hall.
Hoffmann, S. (2003a) Contribution to “Freedom Fries: The French-American Rift over
Iraq” conference. Columbia University, May 22.
Hoffmann, S. (2003b) “US-European Relations: Past and Future,” International Affairs, 79:
1029–1036.
Holland, J. Holyoak, K., Nisbett, R., and Thagard, P. (1986) Processes of Inference, Learning
and Discovery, Cambridge, MA: MIT Press.
Holland, L. (1999). “The U.S. Decision to Launch Operation Desert Storm,” Armed Forces
& Society, 25: 219–23.
Hollis, R. (2003) “Getting Out of the Iraq Trap,” International Affairs, 79: 23–35.
Holsti, K. (1970) “National Role Conceptions in the Study of Foreign Policy,” International
Studies Quarterly, 14: 233–309.
Holsti, O. (1962) “The Belief System and National Images,” Journal of Conflict Resolution,
6: 244–252.
Holsti, O. (1965) “East-West Conflict and Sino-Soviet Relations,” Journal of Applied
Behavioral Science, 1: 115–130.
Holsti, O. (1966) “External Conflict and Internal Consensus,” in P. Stone (ed.) The General
Inquirer, Cambridge, MA: MIT Press.
Holsti, O. (1967) “Cognitive Dynamics and the Image of the Enemy,” in J. Farrell and
A. Smith (eds.) Images and Reality in World Politics, New York: Columbia University
Press.
 Bibliography 301

Holsti, O. (1969) Content Analysis for the Social Sciences and Humanities, Reading, MA:
Addison-Wesley.
Holsti, O. (1969) “External Conflict and Internal Cohesion,” in J. Triska (ed.) Communist
Party States: Comparative and International Studies, New York: Bobbs-Merrill.
Holsti, O. (1970) “The Operational Code Approach to the Study of Political Leaders,”
Canadian Journal of Political Science, 3: 123–157.
Holsti, O. (1972) Crisis, Escalation, War, Montreal: McGill-Queens University Press.
Holsti, O. (1976) “Foreign Policy Viewed Cognitively,” in R. Axelrod (ed.) Structure of
Decision, Princeton, NJ: Princeton University Press.
Holsti, O. (1977) The “Operational Code” as an Approach to the Analysis of Belief Systems,
Durham, NC: Duke University Press.
Holsti, O. Brody, R., and North, R (1964) “Measuring Affect and Action in International
Reaction Models,” Journal of Peace Research, 1: 170–189.
Holsti, O., Hopmann, P., and Sullivan, J. (1973) Unity and Disintegration in International
Alliances, New York: Wiley.
Homans, G. (1961/1974) Social Behavior, New York: Harcourt, Brace.
Hopmann, T. (1967) “International Conflict and Cohesion in the Communist System,”
International Studies Quarterly, 11: 212–236.
Houghton, D. (2007) “Reinvigorating the Study of Foreign Policy Decision Making,”
Foreign Policy Analysis, 3: 24–45.
Howorth, J. (2003) “ESDP and NATO,” Cooperation and Conflict, 38: 235–254.
Ho-yol, Y. (1992) “North Korea’s Economic Policy and the Role of Political Leadership,”
Vantage Point, 15: 1–10.
Hoyt, P. (1997) “The Political Manipulation of Group Composition,” Political Psychology,
18: 771–790.
Hudson, V. (2005) “Foreign Policy Analysis,” Foreign Policy Analysis, 1: 1–30.
Hudson, V. and Vore, C. (1995) “Foreign Policy Analysis Yesterday, Today, and
Tomorrow,” Mershon International Studies Review, 39: 209–238.
Hudson, V., Schrodt, P., and Whitmer. R. (2008) “Discrete Sequence Rule Models as a
Social Science Methodology,” Foreign Policy Analysis, 4: 105–126.
Hughes, K. (2004) Ten Minutes From Normal, New York: Penguin Books.
Hume, B. (2002) Brit Hume Goes One-On-One with First Lady Laura Bush—Transcript,
September 10. Available at: www.foxnews.com
Hybel, A. (1993) Power over Rationality, Albany: State University of New York
Press.
Hymans, J. (2008) “Assessing North Korean Nuclear Intentions and Capacities,” Journal of
East Asian Studies, 8: 263.
Ikenberry, J. (2002) “America’s Imperial Ambition,” Foreign Affairs, (September/October):
44–60.
James, P. (2007) “Behavioral IR,” International Studies Review, 9: 162–165.
Janis, I. (1972) Victims of Groupthink, Boston: Houghton Mifflin.
Janis, I. (1982) Groupthink, Boston: Houghton Mifflin.
Jarosz, W. and Nye. J. (1989) “The Shadow of the Past,” in P. Tetlock et al. (eds.) Behavior,
Society, and International Conflict, New York: Oxford University Press.
Jervis, R. (1970) The Logic of Images in International Relations, Princeton, NJ: Princeton
University Press.
302 Bibliography

Jervis, R. (1976) Perception and Misperception in International Politics, Princeton, NJ: Princeton
University Press.
Jervis, R. (1978) “Cooperation under the Security Dilemma,” World Politics, 30: 167–214.
Jervis, R. (1988) “Realism, Game Theory, and Cooperation,” World Politics, 40: 317–349.
Jervis, R. (1994) “Leadership, Post-Cold War Politics, and Psychology,” Political Psychology,
15: 769–778.
Jervis, R. (1997) System Effects, Princeton, NJ: Princeton University Press.
Jervis, R. (2005) American Foreign Policy in a New Era, New York: Routledge.
Jervis, R. (2006) “Understanding Beliefs,” Political Psychology, 27(5): 641–663.
Jervis, R., and Snyder, J. (eds.) (1991) Dominos and Bandwagons, New York: Oxford
University Press.
Joffe, J. (1995) “Bismarck or Britain?” International Security, 19: 94–117.
Johnson, L. (1977) “Operational Codes and the Prediction of Leadership Behavior,” in
M. Hermann and T. Milburn (eds.) A Psychological Examination of Political Leaders, New
York: Free Press.
Johnston, A. (1995) “Thinking about Strategic Culture,” International Security, 19: 32–64.
Kagan, R. (2003) Paradise and Power, London: Atlantic Books.
Kahneman, D. and Tversky, A. (1979) “Prospect Theory: An Analysis of Decision Under
Risk,” Econometrica, 47: 263–291.
Kang, D. (1995) “Rethinking North Korea,” Asian Survey, 35: 253–267.
Kang, D. (2003) “International Relations Theory and the Second Korean War,” International
Studies Quarterl, 47: 301–324.
Kaplan, M. (1957) System and Process in International Politics, New York: John Wiley.
Katselli, E. and Shah, S. (2003) “September 11 and the UK Response,” International and
Comparative Law Quarterly, 52: 245–255.
Katz, D. (1960) “The Functional Approach to the Study of Attitudes,” Public Opinion
Quarterly, 24: 163–204.
Katz, D. and Stotland, E. (1959) “A Preliminary Statement to a Theory of Attitude
Structure and Change,” in S. Koch (ed.) Psychology: A Study of Science, New York:
McGraw Hill.
Kennedy, P. (1987) The Rise and Fall of the Great Powers, New York: Random House.
Keohane, R. (1983) “Structural Realism and Beyond,” in A. Finifter (ed.) Political
Science: The State of the Discipline, Washington, DC: American Political Science
Association.
Keohane, R. (1984) After Hegemony, Princeton, NJ: Princeton University Press.
Keohane, R. and Martin, L. (2003) “Institutional Theory as a Research Program,” in C.
Elman and M. Elman (eds.) Progress in International Relations Theory, Cambridge, MA:
MIT Press.
Kerr, P. (2003) “Top U.S. Officials Voice Concern About Syria’s WMD Capability,”
Arms Control Today, (May): 32–33.
Khong, Y. (1991) “The Lessons of Korea and the Vietnam Decisions of 1965,” in G.
Breslauer and P. Tetlock (eds.) Learning in U.S. and Soviet Foreign Policy, Boulder, CO:
Westview.
Khong, Y. (1992) Analogies at War, Princeton, NJ: Princeton University Press.
Kim, P. (1994) “Reform Measures in North Korea and Mongolia,” Asian Perspectives, 18:
55–83.
 Bibliography 303

Kim, S. (1998) North Korean Foreign Relations in the Post-Cold War Era, New York: Oxford
University Press.
King, G. (1989) “Event Count Models for International Relations,” International Studies
Quarterly, 33: 123–147.
King, G. and Lowe, W. (2003) “An Automated Information Extraction Tool for
International Conflict Data with Performance as Good as Human Coders,” International
Organization, 57: 617–642.
King, G., Keohane, R., and Verba, S. (1994) Designing Social Inquiry, Princeton, NJ:
Princeton University Press.
Kissinger, H. (2003) “Role Reversal and Alliance Realities,” Washington Post, Feb. 10.
Knowles, E. (1999) Oxford Dictionary of Quotations, 5th edn, New York: Oxford University
Press.
Knutson, J. (1972) The Human Basis of the Polity, Chicago: Aldine Press.
Kohut, H. (1971) The Analysis of the Self, New York: International Universities Press.
Kohut, H. (1977) The Restoration of the Self, New York: International Universities Press.
Kohut, H. (1984) How Does Psychoanalysis Cure? Chicago: University of Chicago Press.
Kowert, P. (2002). Groupthink or Deadlock? Albany: SUNY Press.
Kowert, P. and Hermann, M. (1997) “Who Takes Risks?,” Journal of Conflict Resolution,
41: 611–637.
Krauthamer, C. (2005) “In Iran, Arming for Armageddon,” Washington Post, Dec. 16.
Krech D. and Crutchfield, R. (1948) Theory and Problems of Social Psychology, New York:
McGraw-Hill.
Kristof, N. (2000) “For Bush, His Toughest Call Was The Choice To Run,” New York
Times, Oct. 29.
Kuhn, T. (1962) The Structure of Social Science Revolutions, Chicago: University of Chicago
Press.
Kunkel, J. (1977) “The Behavioral Perspective of Social Dynamics,” in R. Hamblin
and J. Kunkel (eds.) Behavioral Theory in Sociology, New Brunswick, NJ: Transaction
Books.
Kunkel, J. and Nagasawa, R. (1973) “A Behavioral Model of Man,” American Social Review,
38: 530–543.
Kupchan, C. (1994) The Vulnerability of Empire, Ithaca, NY: Cornell University Press.
Kupchan, C. (2010) How Enemies Become Friends, Princeton, NJ: Princeton University
Press.
Kydd, A. (2005) Trust and Mistrust in International Relations, Princeton, NJ: Princeton
University Press.
Lakatos, I. (1970) “Falsification and the Methodology of Scientific Research Programmes,”
in I. Lakatos and A. Musgrave (eds.) Criticism and the Growth of Scientific Knowledge,
Cambridge: Cambridge University Press.
Lakatos, I. and Musgrave, A. (eds.) (1970) Criticism and the Growth of Knowledge, Cambridge:
Cambridge University Press.
Lake, D. and Powell, R. (1999a) Strategic Choice and International Relations, Princeton, NJ:
Princeton University Press.
Lake, D. and Powell, R. (1999b) “International Relations,” in D. Lake and R. Powell
(eds.) Strategic Choice and International Relations, Princeton, NJ: Princeton University
Press.
304 Bibliography

Landsberger, H. (1958) Hawthorne Revisited, Ithaca: New York State School of Industrial
and Labor Relations.
Lane, R. (1962) Political Ideology, New York: Free Press.
Lanzetta, J., Haefner, D., Langham, P., and Axelrod, H. (1954) “Some Effects of Situational
Threat on Group Behavior,” Journal of Abnormal and Social Psychology, 49: 445–453.
Lapid, J. and Kratochwil, F. (1996) The Return of Culture and Identity to IR Theory, Boulder,
CO: Lynne Rienner.
Larson, D. (1985) Origins of Containment, Princeton, NJ: Princeton University Press.
Larson, D. (1988) “Problems of Content Analysis in Foreign-Policy Research,” International
Studies Quarterly, 32: 241–255.
Larson, D. (1991) “Learning in U.S.–Soviet Relations,” in G. Breslauer and P. Tetlock
(eds.) Learning in U.S. and Soviet Foreign Policy, Boulder, CO: Westview.
Larson, D. (1994) “The Role of Belief Systems and Schemas in Foreign Policy Decision
Making,” Political Psychology, 15: 17–33.
Larson, D. (1997) Anatomy of Mistrust, Ithaca, NY: Cornell University Press.
Larson, D. and Shevchenko, A. (2010) “Status Seekers,” International Security, 34: 63–96.
Lasswell, H. (1958) Politics: Who Gets What, When, How, New York: Meridian Books.
Lau, R. (2003) “Models of Decision-Making,” in D. Sears, L. Huddy, and R. Jervis (eds.)
Oxford Handbook of Political Psychology, New York: Oxford University Press, 19–59.
Laudan, L. (1977) Progress and Its Problems, Berkeley: University of California Press.
Laughlin, R., Pines, D., Schmalian, J., Stojkovic, B., and Wolynes, P. (2000) “The Middle
Way,” PNAS, 97: 32–37.
Lebow, R. (1994) “The Long Peace, the End of the Cold War, and the Failure of Realism,”
International Organization, 48: 249–277.
Lebow, R. (2008) A Cultural Theory of International Relations, Cambridge: Cambridge
University Press.
Lebow, R. and Risse-Kappen, T. (eds.) (1995) International Relations Theory and the End of
the Cold War, New York: Columbia University Press.
Lebow, R. and Stein, J. (1994) We All Lost the Cold War, Princeton, NJ: Princeton
University Press.
Lebovic, J. (1995) “How Organizations Learn,” American Journal of Political Science, 39:
835–863.
Lee, C. (2001) “Conflict and Cooperation,” in N. Eberstadt and R. Ellings (eds.) Korea”s
Future and the Great Powers, Seattle, WA: National Bureau of Asian Research, 51–87.
Leites, N. (1951) The Operational Code of the Politburo, New York: McGraw-Hill.
Leites, N. (1953) A Study of Bolshevism, New York: Free Press.
Leites, N. (1964) “Kremlin Moods,” RM 3535-ISA (January). Santa Monica, CA: Rand.
Leng, R. (1983) “When Will They Ever Learn?” Journal of Conflict Resolution, 27: 379–419.
Leng, R. (1993) Interstate Crisis Behavior, 1816–1980, Cambridge: Cambridge University
Press.
Leng, R. (2000) Bargaining and Learning in Recurring Crises, Ann Arbor: University of
Michigan Press.
Leng, R. and Walker, S. (1982) “Comparing Two Studies of Crisis Bargaining,” Journal of
Conflict Resolution, 26: 571–591.
Lepgold, J. and Lamborn, A. (2001) “Locating Bridges,” International Studies Review, 3:
3–31.
 Bibliography 305

Lepgold, J. and Miroslav, N. (2002) Beyond the Ivory Tower, New York: Columbia
University Press.
Lesch, D. (1996) “The 1957 American-Syrian Crisis,” in David Lesch (ed.) The Middle East
and the United States, Boulder, CO: Westview.
Leverett, F. (2005) Inheriting Syria, Washington, DC: Brookings Institution Press.
LeVine, R. and Campbell, D. (1972) Ethnocentrism, New York: John Wiley.
Levy, J. (1994) “Learning and Foreign Policy,” International Organization, 48:
279–312.
Levy, J. (1997) “Prospect Theory, Rational Choice, and International Relations,”
International Studies Quarterly, 41: 87–112.
Levy, J. (2003) “Political Psychology and Foreign Policy,” in D. Sears, L. Huddy, and
R. Jervis (eds.) Oxford Handbook of Political Psychology, New York: Oxford University
Press.
Lewin, K. (1951) Field Theory in Social Science, New York: Harper and Row.
Lewis, W. (2002) “The War on Terror,” Mediterranean Quarterly, (Fall): 21–37.
Lindblom, C. (1977). Politics and Markets, New York: Basic Books.
Liska, G. (1962) Nations in Alliance, Baltimore: Johns Hopkins University Press.
Little, D. (1991) Varieties of Social Explanation, Boulder, CO: Westview.
Little, D. (1998) Microfoundations, Method, and Causation, New Brunswick, NJ:
Transaction.
Lobell, S., Ripsman, N., and Taliaferro, J. (2009) Neoclassical Realism, the State, and Foreign
Policy, Cambridge: Cambridge University Press.
Lott, A. J. and Lott, B. E. (1965) “Group Cohesiveness and Interpersonal Attraction,”
Psychological Bulletin, 64: 253–309.
Lukes, S. (1974) Power, London: Macmillan.
Lupia, J., McCubbins, M., and Samuel, P. (2000) Elements of Reason, Cambridge:
Cambridge University Press.
McAllister, J. (1994) “Pyongyong’s Dangerous Game,” Time (April 4).
McCall, G. and Simmons, J. (1978) Identities and Interactions, New York: Free Press.
McClelland, C. (1966) Theory and the International System, New York: Macmillan.
McClelland, C. (1968) “Access to Berlin,” in D. Singer (ed.) Quantitative International
Politics, New York: Free Press.
McClelland, C. (1972) “The Beginning, Duration, and Abatement of International Crises,”
in C. Hermann (ed.) International Crises, New York: Free Press.
McClelland, C. and Hoggard, G. (1969) “Conflict Patterns in the Interactions among
Nations,” in J. Rosenau (ed.) International Politics and Foreign Policy, 2nd edn, New York:
Free Press.
McCormick, J. and Wittkopf, E. (1990) “Bipartisanship, Partisanship, and Ideology in
Congressional-Executive Foreign Policy Relations, 1947–1988,” Journal of Politics, 52:
1077–1100.
McDermott, R. (1998) Risk-Taking in International Politics, Ann Arbor: University of
Michigan Press.
McDermott, R. (2004) “Prospect Theory in Political Science,” Political Psychology, 25:
289–312.
McGowan, P. and Walker, S. (1981) “Radical and Conventional Models of U.S. Foreign
Economic Policy Making,” World Politics, 33: 347–382.
306 Bibliography

McGuire, W. (1967) “The Current Status of Cognitive Consistency Theories,” in

M. Fishbein (ed.) Readings in Attitude Theory and Measurement, New York: Wiley.
Majeski, S. (2004) “Asymmetric Power among Agents and the Generation and
Maintenance of Cooperation in International Relations,” International Studies Quarterly,
48: 450–470.
Malici, A. (2005) “Discord and Collaboration between Allies,” Journal of Conflict Resolution,
49: 90–119.
Malici, A. (2006) “Germans as Venutians,” Foreign Policy Analysis, 2: 37–62.
Malici, A. (2007) “Thinking About Rogue Leaders,” Whitehead Journal of Diplomacy and
International Relations, 8: 103–111.
Malici, A. (2008) When Leaders Learn and When They Don’t, Albany: SUNY.
Malici, A. and Buckner, A. (2008) “Empathizing with Rogue Leaders,” Journal of Peace
Research, 45: 783–800.
Malici, A. and Malici, J. (2005a) “The Operational Codes of Fidel Castro and Kim Il
Sung,” Political Psychology, 26: 387–412.
Malici, A. and Malici, J. (2005b) “When Will They Ever Learn?” Psicología Política, 31:
7–22.
Manning, R. (1998) “The United States in North Korean Foreign Policy,” in S. Kim (ed.),
North Korean Foreign Relations in the Post-cold War Era, New York: Oxford University
Press.
Mansbach, R. and Vasquez, J. (1981) In Search of Theory, New York: Columbia University
Press.
Maoz, Z. (1990) National Choices and International Processes, New York: Cambridge
University Press.
Maoz, Z. and Mor, B. (2002) Bound by Struggle, Ann Arbor: University of Michigan.
Marcus, G. (2003) “The Psychology of Emotion in Politics,” in D. Sears, L. Huddy, and
R. Jervis (eds.) Oxford Handbook of Political Psychology, New York: Oxford University
Press.
Marfleet, B. (2000) “The Operational Code of John F. Kennedy during the Cuban Missile
Crisis,” Political Psychology, 23: 545–558.
Marfleet, B. and Miller, C. (2005) “Failure after 1441,” Foreign Policy Analysis, 1:
333–359.
Marfleet, B. and Walker, S. (2006) “A World of Beliefs,” in M. Schafer and S. Walker
(eds.) Beliefs and Leadership in World Politics, New York: Palgrave.
Marshall, B. (2003) “Presidential Success in the Realm of Foreign Affairs,” Social Science
Quarterly, 84: 685–703.
Mastanduno, M. (1997) “Preserving the Unipolar Moment,” International Security, 21,
49–88.
Maugh, T. (2010) “Distant Planets Rattle Theories with Their Irregular Orbits,” Arizona
Daily Sun, April 14.
Mazarr, M. (1991) “Orphans of Glasnost,” Korea and World Affairs, 15: 58–84.
Mazarr, M. (2003) “George W. Bush, Idealist,” International Affairs, 79: 503–522.
May, E. (1973) Lessons of the Past, New York: Oxford University Press.
Mead, G. (1947) Mind, Self and Society, Chicago: University of Chicago Press.
Mearsheimer, J. (1990) “Back to the Future,” International Security, 15: 5–56.
Meernik, J. (1993) “Presidential Support in Congress,” Journal of Politics, 55: 569–588.
 Bibliography 307

Mercer, J. (2005) “Prospect Theory and Political Science,” Annual Review of Political
Science, 8: 1–21.
Mesa-Lago, C. (1993a) “Introduction: Cuba, the Last Communist Warrior,” in C. Mesa-
Lago (ed.) Cuba After the Cold War, Pittsburgh, PA: University of Pittsburgh Press.
Mesa-Lago, C. (1993b) “The Economic Effects on Cuba of the Downfall of Socialism
in the USSR and Eastern Europe,” in C. Mesa-Lago (ed.) Cuba After the Cold War,
Pittsburgh, PA: University of Pittsburgh Press.
Mesa-Lago, C. (1998) “The Cuban Economy In 1997–98: Performance and Policies,” in
Cuba in Transition, Vol. 8, Papers and Proceedings of the 8th Annual Meeting of the
Association for the Study of the Cuban Economy, Miami, FL. Available: http://lanic.
utexas.edu
Milburn, T. and Billings, R. (1972) “Decision-Making Perspectives from Psychology,”
American Behavioral Scientist, 20: 111–126.
Miller, N. et al. (1941) “The Frustration-Aggression Hypothesis,” Psychological Review, 48:
337–342.
Mintz, A. (2007) “Why Behavioral IR?” International Studies Review, 9: 157–162.
Mintz, A. and DeRouen, K. (2010). Understanding Foreign Policy Decision Making,
Cambridge: Cambridge University Press.
Mintz, A., Nehemia, G., Redd, S., and Carnes, A. (1997) “The Effect of Dynamic and
Static Choice Sets on Political Decision Making,” American Political Science Review, 91:
553–566.
Mitchell, A. (2000) “The 43rd President: George Walker Bush,” New York Times,
December 13: A1.
Mitchell, M. (2009) Complexity, New York: Oxford University Press.
Mitchell, N. (1997) The Conspicuous Corporation, Ann Arbor: University of Michigan
Press.
Moens, A. (2004) The Foreign Policy of George W. Bush, Burlington, VT: Ashgate.
Monroe, K. (ed.) (2005) Perestroika! New Haven, CT: Yale University Press.
Moon-Young, H. (1994) “North Korea’s Foreign Policy in Transition,” Vantage Point,
17(7–8): 21–29, 19–27.
Moravcsik, A. (2003) “Liberal International Relations Theory,” in C. Elman and M.
Elman (eds.) Progress in International Relations Theory, Cambridge, MA: MIT Press.
Morgan, P. (1998) “New Security Arrangements Between the United States and North
Korea,” in D. Suh and C. Lee (eds.) North Korea After Kim Il Sung, Boulder, CO: Lynne
Rienner.
Morgenthau, H. (1967) Politics Among Nations, 4th edn, New York: Alfred Knopf.
Morgenthau, H. (1978) Politics Among Nations, 5th edn, New York: Alfred Knopf.
Morrow, J. (1994) Game Theory for Political Scientists, Princeton, NJ: Princeton University
Press.
Morrow, J. (1997) “A Rational Choice Approach to International Conflict,” in A. Mintz
and N. Geva (eds.) Decisionmaking in War and Peace, Boulder, CO: Lynne Rienner.
Mulder, M. and Stemerding, A. (1963) “Threat, Attraction to Group, and Need for Strong
Leadership,” Human Relations, 16: 317–334.
Nash, J. (1951) “Non-cooperative Games,” Annals of Mathematics, 54: 286–295.
Neustadt, R. and May, E. (1986) Thinking in Time, New York: Free Press.
New York Times (2003) “Facing Common Enemy,” February 1.
308 Bibliography

New York Times (2004) “The Candidates 2nd Debate,” October 9.

Newell, A. and Simon, H. (1972) Human Problem Solving, Englewood Cliffs, NJ: Prentice
Hall.
Nincic, M. and Lepgold, J. (2000) Being Useful, Ann Arbor: University of Michigan
Press.
Nisbett, R. and Ross, L. (1980) Human Inference, Englewood Cliffs, NJ: Prentice-Hall.
Nye, J. (1987) “Nuclear Learning and U.S. Security Regimes,” International Organization,
41: 371–402.
Nye, J. (1990) Bound to Lead, New York: Basic Books.
Nye, J. (2008) Understanding International Conflicts, New York: Longman.
Obama, B. (2005) “America’s Nuclear Non-Proliferation. Policy,” May. Available at:
http://Obamaspeeches.com
Oh, K. (1990) North Korea’s Response to the World, Santa Monica, CA: Rand.
Oh, K. and Hassig, R. (2000) North Korea through the Looking Glass, Washington, DC:
Brookings Institution.
Olson, W. (1976) “President, Congress and American Foreign Policy: Confrontation or
Collaboration,” International Affairs, 52: 565–582.
O’Neill, B. (1999) Honor, Symbols, and War, Ann Arbor: University of Michigan Press.
O’Reilly, K. (2007) “Perceiving Rogue States,” Foreign Policy Analysis, 3: 295–315.
Osgood, C. (1960) Graduated Reciprocation in Tension Reduction, Urbana: University of
Illinois Press.
Osgood, C. (1962) An Alternative to War and Surrender, Chicago: University of Illinois
Press.
Osgood, C., Suci, G., and Tannenbaum, P. (1957) The Measurement of Meaning, Champaign:
University of Illinois Press.
Oye, K. (1986) Cooperation under Anarchy, Princeton, NJ: Princeton University Press.
Page, B. and Barabas, J. (2000) “Foreign Policy Gaps Between Citizens and Leaders,”
International Studies Quarterly, 44: 339–364.
Paige, G. (1968) The Korean Decision, New York: Free Press.
Palmer, G. and Morgan, C. (2006) A Theory of Foreign Policy, Princeton, NJ: Princeton
University Press.
Pastor, R. (2001) Not Condemned to Repetition, 2nd edn, Boulder, CO: Westview.
Peers, S. (2003) “EU Responses to Terrorism,” International and Comparative Law Quarterly,
53: 227–244.
Peffley, M. and Hurwitz, J. (1992) “International Events and Foreign Policy Beliefs,”
American Journal of Political Science, 3: 431–461.
Penska, S. and Mason, W. (2003) “EU Security Cooperation and the Transatlantic
Relationship,” Conflict and Cooperation, 38: 225–280.
Perlez, J. (2001) “Cordial Rivals,” New York Times, June 18: A10.
Peterson, B., Winter, D., and Doty, R. (1994) “Laboratory Tests of a Motivational-
Perceptual Model of Conflict Escalation,” Journal of Conflict Resolution, 38: 719–748.
Phillips, W. (1974) “Where Have All the Theories Gone?” World Politics, 26: 155–188.
Pickel, A. (1998) “Is Cuba Different?” Communist and Post-Communist Studies, 31: 75–90.
Planas, J. (1993) The Impact of Soviet Reforms on Cuban Socialism,” in E. Baloyra and
J. Morris (eds.) Conflict and Change in Cuba, Albuquerque: University of New Mexico
Press.
 Bibliography 309

Polk, W. (1997) Neighbors and Strangers, Chicago: University of Chicago Press.

Popper, K. (1972) Objective Knowledge, London: Oxford University Press.
Popper, K. (1982) The Open Universe, Totowa, NJ: Rowman and Littlefield.
Popper, K. (1994) Knowledge and the Mind-Body Problem, London: Routledge.
Posen, B. (1984) The Sources of Military Doctrine, Ithaca, NY: Cornell University Press.
Post, J. (1980) “The Seasons of a Leader”s Life,” Political Psychology, 2: 35–49.
Post, J. (1983) “Woodrow Wilson Re-examined,” Political Psychology, 4: 289–306.
Post, J. (2003) The Psychological Assessment of Political Leaders, Ann Arbor: University of
Michigan Press.
Preston, T. (2001) The President and His Inner Circle, Ann Arbor: University of Michigan
Press.
Putnam, R. (1988) “Diplomacy and Domestic Politics,” International Organization, 42:
427–460.
Rapoport, A. (1960) Fights, Games, and Debates, Ann Arbor: University of Michigan
Press.
Rapoport, A. and Guyer, M. (1976) The 2 × 2 Game, Ann Arbor: University of Michigan
Press.
Rapoport, A. and Guyer, M. (1966) “A Taxonomy of 2 × 2 Games,” General Systems:
Yearbook for the Society of General Systems Research, 11: 203–214.
Reiter, D. (1994) “Learning, Realism, and Alliances,” World Politics, 46: 490–526.
Renshon, J. (2009) “When Public Statements Reveal Private Beliefs,” Political Psychology,
30(4): 649–661.
Renshon, S. (2004) In His Father’s Shadow, New York: Palgrave Macmillan.
Ricci, D. (1984) The Tragedy of Political Science, New Haven, CT: Yale University Press.
Rich-Kaplowitz, D. (1993) “Cuba’s Foreign Policy: Old Goals, New Strategies,” in D.
Rich-Kaplowitz (ed.) Cuba’s Ties to a Changing World, Boulder, CO: Lynne Rienner.
Richardson, N. (1987) “Dyadic Case Studies in the Comparative Study of Foreign Policy,”
in C. Hermann, C. Kegley, and J. Rosenau (eds.) New Directions in the Study of Foreign
Policy, Boston: Allen and Unwin.
Risse-Kappan, T. (1991) “Public Opinion, Domestic Structure, and Foreign Policy in
Liberal Democracies,” World Politics, 43: 479–512.
Ritter, A. (1994) “Cuba’s Economic Strategy and Alternative Futures,” in J. López (ed.)
Cuba at a Crossroads, Gainesville: University Press of Florida.
Robison, S. (2006) “George W. Bush and the Vulcans,” in M. Schafer and S. Walker (eds.)
Beliefs and Leadership in World Politics, New York: Palgrave Macmillan.
Robison, S. (2009) “Experiential Learning in the Foreign Policy Decision-Making
Context,” Dissertation Manuscript. Louisiana State University.
Rokeach, M. (1968) Beliefs, Attitudes and Values, San Francisco: Jossey-Bass.
Rosati, J. (2000) “The Power of Human Cognition in the Study of World Politics,”
International Studies Review, 2: 45–75.
Rosecrance, R. (1963) Action and Reaction in World Politics, Boston: Little, Brown.
Rose, G. (1998) “Neoclassical Realism and Theories of Foreign Policy,” World Politics,
51: 141–172.
Rosenau, J. (1966) “Pre-theories and Theories of International Politics,” in B. Farrell
(ed.) Approaches to Comparative and International Politics, Evanston, IL: Northwestern
University Press.
310 Bibliography

Rosenau, J. (1968) “Comparative Foreign Policy: Fad, Fantasy, or Field?” International

Studies Quarterly, 12: 296–329.
Rothstein, R. (1968) Alliances and Small Powers, New York: Columbia University Press.
Ruggie, J. (1983) “Continuity and Transformation in the World Polity,” World Politics,
35: 261–285.
Rummel, R. (1975–81) Understanding Conflict and War, Volumes 1–5, Beverly Hills, CA:
Sage.
Rumsfeld, D. (2003) Press briefing at the foreign press center, January 22. Available at:
http://www.defenselink.mil/news/Jan2003/t01232003_t0122sdfpc.html
Rutenberg, J. (2006) “Adviser Who Shaped Bush’s Speeches is Leaving,” New York Times,
June 15.
Sang-Woo, R. (1991) “North Korea in 1990,” Asian Survey, 31: 71–78.
Sanger, D. (2003) “Iraq Makes U.N. Seem ‘Foolish,’” New York Times, October 29.
Sanger, D. (2005a) “Bush Likens Iraqi Action to Transition in 40s Japan,” New York Times,
Dec. 10.
Sanger, D. (2005b) “The President’s First Draft of the Iraq War,” New York Times, Dec.
18.
Sarbin, T. and Allen, V. (1968) “Role Theory,” in G. Lindzey and E. Aronson (eds.) The
Handbook of Social Psychology, Reading, MA: Addison-Wesley.
Schachter, S. (1959) The Psychology of Affiliation, Stanford, CA: Stanford University Press.
Schafer, M. (1999) “Explaining Groupthink,” International Interactions, 25: 1–31.
Schafer, M. (2000) “Issues in Assessing Psychological Characteristics at a Distance,” Political
Psychology, 21: 511–528.
Schafer, M. and Crichlow, S. (1996) “Antecedents of Groupthink,” Journal of Conflict
Resolution, 40: 415–435.
Schafer, M. and Crichlow, S. (2000) “Bill Clinton”s Operational Code,” Political Psychology,
21: 559–571.
Schafer, M. and Crichlow, S. (2002) “The Process-Outcome Connection in Foreign
Policy Decision Making,” International Studies Quarterly, 46: 45–68.
Schafer, M. and Crichlow, S. (2010) Groupthink vs. High-Quality Decision Making in
International Relations, New York: Columbia University Press.
Schafer, M. and Walker, S. (2001) “The Operational Code of Vladimir Putin: Analyzing
a New Global Leader with a New Automated Coding System,” paper presented at the
Annual Meeting of the International Studies Association, Chicago, February 21–24.
Schafer, M. and Walker, S. (eds.) (2006a) Beliefs and Leadership in World Politics, New York:
Palgrave Macmillan.
Schafer, M. and Walker, S. (2006b) “Democratic Leaders and the Democratic Peace,”
International Studies Quarterly, 50: 3.
Schafer, M. and Walker, S. (2006c) “Operational Code Analysis at a Distance,” in M.
Schafer and S. Walker (eds.) Beliefs and Leadership in World Politics, New York: Palgrave
Macmillan.
Schafer, M., Robison, S., and Aldrich, B. (2006) “Operational Codes and the 1916 Easter
Rising in Ireland,” Foreign Policy Analysis, 2: 63–82.
Schelling, T. (1960) The Strategy of Conflict, Cambridge, MA: Harvard University Press.
Schraufnagel, S. and Shellman, S. (2001) “The Two Presidencies 1984–1998,” Presidential
Studies Quarterly, 31: 699–707.
 Bibliography 311

Schrodt, P. et al. (1994) “Political Science: KEDS: A Program for the Machine Coding of
Event Data,” Social Science Computer Review, 12: 561–588.
Schroeder, P. (1994) “Historical Reality versus Neo-Realist Theory,” International Security,
19: 108–148.
Schweller, R. (1994) “Bandwagoning for Profit,” International Security, 19: 72–107.
Schweller, R. (1998) Deadly Imbalances, New York: Columbia University Press.
Schweller, R. (2003) “The Progressiveness of Neoclassical Realism,” in C. Elman and M.
Elman (eds.) Progress in International Relations Theory, Cambridge, MA: MIT Press.
Schweller, R. (2006) Unanswered Threats, Princeton, NJ: Princeton University Press.
Scott, A. (1967) The Functioning of the International Political System, New York:
Macmillan.
Searle. J. (1995) The Social Construction of Reality, New York: Free Press.
Searle. J. (1998) Mind, Language, and Society, New York: Basic Books.
Searle. J. (2001) Rationality in Action, Cambridge, MA: MIT Press.
Sears, D., Huddy, L., and Jervis, R. (eds.) (2003) Oxford Handbook of Political Psychology,
Oxford: Oxford University Press.
Shannon, C. and Weaver, W. (1964) The Mathematical Model of Communication, Urbana:
University of Illinois Press.
Sherif, M. and Sherif, C. (1953) Groups in Harmony and Tension, New York: Harper and
Row.
Sherif, M. et al. (1961) Intergroup Conflict and Cooperation, Norman: University of Oklahoma
Press.
Shinagel, M. (ed.) (1994) Robinson Crusoe, New York: W.W. Norton.
Sigel, L. (1998) Disarming Strangers, Princeton, NJ: Princeton University Press.
Signorino, C. (1999) “Strategic Interaction and the Statistical Analysis of International
Conflict,” American Political Science Review, 93: 279–298.
Simmel. A. (1979) “Small-Group Dynamics and Foreign Policy Decision-Making,” in L.
Falkowski (ed.) Psychological Models in International Politics, Boulder, CO: Westview.
Simmel, G. (1955) Conflict, New York: Free Press.
Simon, H. (1957) Models of Man, New York: John Wiley.
Simon, H. (1973) “The Organization of Complex Systems,” in H. Pattee (ed.) Hierarchy
Theory, New York: Braziller.
Simon, H. (1981) Sciences of the Artificial, Cambridge, MA: MIT Press.
Simon, H. (1985) “Human Nature in Politics,” American Political Science Review, 79:
293–304.
Simon, H. (1986) “Rationality in Psychology and Economics,” Journal of Business, 59:
209–224.
Skyrms, B. (2004) The Stag Hunt and the Evolution of Social Structure, Cambridge: Cambridge
University Press.
Smith, B. (1968) “A Map for the Analysis of Personality and Politics,” Journal of Social
Issues, 24: 15–28.
Smith, C. (1992) Motivation and Personality, New York: Cambridge University Press.
Smith, D. (2006) Deterring America, Cambridge: Cambridge University Press.
Smith, S. (1988) “Belief Systems and the Study of International Relations,” in R. Little
and S. Smith (eds.) Belief Systems and International Relations, Oxford: Blackwell.
Smith, T. (1988) “The Spirit of the Sierra Maestra,” World Politics, 41: 98–119.
312 Bibliography

Snyder, G. (1971) “Prisoner’s Dilemma and Chicken Models in International Politics,”

International Studies Quarterly, 15: 66–103.
Snyder, G. and Diesing, P. (1977) Conflict Among Nations, Princeton, NJ: Princeton
University Press.
Snyder, J. (1992) Myths of Empire, Ithaca, NY: Cornell University Press.
Snyder, J. and Jervis, R. (1993) Coping with Complexity in the International System, Boulder,
CO: Westview.
Snyder, R., Bruck, H., and Sapin, B. (eds.) (1954) Foreign Policy Decision-making as an
Approach to the Study of International Politics, Princeton, NJ: Princeton University Press.
Snyder, R., Bruck, H., and Sapin, B. (eds.) (1962) Foreign Policy Decision Making, New
York: Free Press.
Snyder, S. (1999) Negotiating on the Edge, Washington, DC: USIP Press.
Sobel, R. (2001) The Impact of Public Opinion on U.S. Foreign Policy Since Vietnam, New
York: Oxford University Press.
Sprout, H. and Sprout, M. (1956) Man-Milieu Relationship Hypotheses in the Context of
International Politics, Princeton, NJ: Princeton University Press.
Sprout, H. and Sprout, M. (1969) “Environmental Factors in the Study of International
Politics,” in J. Rosenau (ed.) International Politics and Foreign Policy, New York: Free
Press.
Stein, A. (1976) “Conflict and Cohesion,” Journal of Conflict Resolution, 20: 143–172.
Stein, A. (1990) Why Nations Cooperate, Ithaca, NY: Cornell University Press.
Stein, J. (1988) “Building Politics in Psychology,” Political Psychology, 9: 245–272.
Stein, J. (1994) “Political Learning by Doing,” International Organization, 48: 155–183.
Stein, J. (2007) “Foreign Policy Decision-making,” in S. Smith, A. Hadfield, and T.
Dunne (eds.) Foreign Policy: Theories, Actors, Cases, Oxford: Oxford University Press.
Steinbrunner, J. (1974) The Cybernetic Theory of Decision, Princeton, NJ: Princeton
University Press.
Stern, E. (1997) “Crisis and Learning,” Journal of Contingencies and Crisis Management, 5:
69–86.
Stryker, S. and Statham, A. (1985) “Symbolic Interaction and Role Theory,” in G.
Lindzey and E. Aronson (eds.) The Handbook of Social Psychology, New York: Random
House.
Suchlicki, J. (2000) “Castro’s Cuba,” Studies in Comparative International Development, 34:
123–135.
Suedfeld, P. and Bluck, S. (1988) “Changes in Integrative Complexity Prior to Surprise
Attacks,” Journal of Conflict Resolution, 32: 626–635.
Suedfeld, P., Guttieri, K., and Tetlock, P. (2003) “Assessing Integrative Complexity at
a Distance,” in J. Post (ed.) The Psychological Assessment of Political Leaders, Ann Arbor:
University of Michigan Press.
Sullivan, M. (1976) International Relations, Englewood Cliffs, NJ: Prentice-Hall.
Suskind, R. (2004) The Price of Loyalty, New York: Simon and Schuster.
Suskind, R. (2009) The Way of the World, New York: Harper.
Swede, S. and Tetlock, P. (1986) “Henry Kissinger’s Implicit Theory of Personality,”
Journal of Personality, 54: 617–648.
Taliaferro, J. (2000/01) “Security Seeking under Anarchy,” International Security, 25:
128–161.
 Bibliography 313

Taliaferro, J. (2004a) “Power Politics and the Balance of Risk: Hypotheses on Great Power
Intervention in the Periphery,” Political Psychology, 25: 177–211.
Taliaferro, J. (2004b) Balancing Risks, Ithaca, NY: Cornell University Press.
Tanter, R. (1998) Rogue Regimes, New York: St. Martin’s Press.
Tetlock, P. (1986) “Psychological Advice on Foreign Policy,” American Psychologist, 41:
557–567.
Tetlock, P. (1991) “In Search of an Elusive Concept,” in G. Breslauer and P. Tetlock
(eds.) Learning in US and Soviet Foreign Policy, Boulder CO: Westview.
Tetlock, P. (1998) “Social Psychology and World Politics,” in D. Gilbert, S. Fiske, and
G. Lindzey (eds.) The Handbook of Social Psychology, Volume II, 4th edn, New York:
McGraw-Hill.
Tetlock, P. (2000) “Coping with Trade-Offs,” in A. Lupia, M. McCubbins, and S. Popkin
(eds.) Elements of Reason, Cambridge: Cambridge University Press.
Tetlock, P. (2005) Expert Political Judgment, Princeton, NJ: Princeton University Press.
Tetlock, P. and Levi, A. (1982) “Attribution Bias,” Journal of Experimental Social Psychology,
18: 68–88.
Thaler, R. (1994) Quasi-Rational Economics, New York: Russell Sage.
Thies, C. (2001) “A Social Psychological Approach to Enduring Rivalries,” Political
Psychology, 22: 693–725.
Thies, C. (2010) “Role Theory and Foreign Policy,” in R. Denemark (ed.) The International
Studies Encyclopedia, Oxford: Blackwell.
Thies, C. and Sobek, D. (2010) “War, Economic Development, and Political Development
in the Contemporary International System,” International Studies Quarterly, 54: 267–
288.
Thompson, W. (1983) Contending Approaches to World Systems Analysis, Beverly Hills, CA:
Sage.
Thompson, W. and Rapkin, D. (1981) “Collaboration, Consensus, and Détente,” Journal
of Conflict Resolution, 25: 615–637.
Thornhill, J. and Ward, A. (2002) “The Puzzle of How to Deal with North Korea’s
Strange Autocrat,” Financial Times (15 December): 7.
Tickner, J. (19970 “You Just Don’t Understand,” International Studies Quarterly, 41:
611–632.
Toloraya, G. (2002) “Russia and North Korea,” in T. Akaha (ed.) The Future of North
Korea, New York: Routledge.
Tow, W. (2003) “Apocalypse Forever?” Australian Journal of Politics and History, 49:
314–325.
Tsebelis, G. (1990) Nested Games, Berkeley: University of California Press.
Tsebelis, G. (2002) Veto Players, Princeton, NJ: Princeton University Press.
Tuchman, B. (1984) The March of Folly, New York: Random House.
Tucker, R. (1977) “The Georges’ Wilson Re-examined,” American Political Science Review,
71: 606–618.
Twain, M. (Samuel L. Clemens) (1939) Tom Sawyer, Cambridge, MA: The University
Press.
Tyler, P. (2001) “Bush and Putin Look Each Other in the Eye,” New York Times,
June 16.
UK Ministry of Defence (2003) Operations in Iraq, London, UK.
314 Bibliography

United Nations (1997) World Statistics Pocketbook, New York: United Nations.
Vasquez, J. (1983) The Power of Power Politics, New Brunswick, NJ: Rutgers University
Press.
Vasquez, J. (1987) “Foreign Policy, Learning, and War,” in C. Hermann, C. Kegley, and
J. Rosenau (eds.) New Directions in the Study of Foreign Policy, Boston: Allen and Unwin.
Vasquez, J. (1997) “The Realist Paradigm and Degenerative versus Progressive Research
Programs,” American Political Science Review, 91: 899–912.
Vasquez, J. and Elman, C. (2003) Realism and the Balancing of Power, Upper Saddle River,
NJ: Prentice-Hall.
Vertzberger, Y. (1990) The World in Their Minds, Stanford, CA: Stanford University
Press.
Voss, J. (1998) “On the Representation of Problems,” in D. Sylvan and D. Voss (eds.)
Problem Representation in Foreign Policy Decision Making, Cambridge: Cambridge
University Press.
Wagner, H. (1993) “What Was Bipolarity?” International Organization, 47: 77–106.
Walker, R. (1993) Inside/Outside, Cambridge: Cambridge University Press.
Walker, S. (1977) “The Interface between Beliefs and Behavior,” Journal of Conflict
Resolution, 21: 129–168.
Walker, S. (1979) “National Role Conceptions and Systemic Outcomes,” in L. Falkowski
(ed.) Psychological Models in International Politics, Boulder, CO: Westview.
Walker, S. (1983) “The Motivational Foundations of Political Belief Systems,” International
Studies Quarterly, 27: 179–202.
Walker, S. (1987a) “Role Theory and the International System,” in S. Walker (ed.) Role
Theory and Foreign Policy Analysis, Durham, NC: Duke University Press.
Walker, S. (1987b) “The Correspondence between Foreign Policy Rhetoric and
Behavior,” in S. Walker (ed.) Role Theory and Foreign Policy Analysis, Durham, NC:
Duke University Press.
Walker, S. (ed.) (1987c) Role Theory and Foreign Policy Analysis, Durham, NC: Duke
University Press.
Walker, S. (1990) “The Evolution of Operational Code Analysis,” Political Psychology, 11:
403–418.
Walker, S. (1992) “Symbolic Interactionism and International Politics,” in M. Cottam and
C. Shih (eds.) Contending Dramas, New York: Praeger.
Walker, S. (1995) “Psychodynamic Processes and Framing Effects in Foreign Policy
Decisionmaking,” Political Psychology, 16: 697–718.
Walker, S. (2000a) Forecasting the Political Behavior of Leaders with the Verbs in Context System
of Operational Code Analysis, Hilliard, OH: Social Science Automation.
Walker, S. (2000b) “Role Identities and the Operational Codes of Leaders,” paper pre-
sented at the annual meeting of the International Society of Political Psychology, Seattle,
WA, July 1–4.
Walker, S. (2000c) “Assessing Psychological Characteristics at a Distance,” Political
Psychology, 21: 597–603.
Walker, S. (2002) “Beliefs and Foreign Policy Analysis in the New Millennium,” in M.
Brecher and F. Harvey (eds.) Conflict, Security, Foreign Policy, and International Political
Economy, Ann Arbor: University of Michigan Press.
Walker, S. (2003) “Operational Code Analysis as a Scientific Research Program,” in C.
 Bibliography 315

Elman and M. Elman (eds.) Progress in International Relations Theory, Cambridge, MA:
MIT Press.
Walker, S. (2004a) “Role Identities and the Operational Codes of Political Leaders,” in M.
Hermann (ed.) Political Psychology as a Perspective on Politics, London: Elsevier.
Walker, S. (2004b) “The Management and Resolution of International Conflict in a
Single Case,” in Z. Maoz et al. (eds.) Multiple Paths to Knowledge in International Relations,
Lanham, MD: Lexington Books.
Walker, S. (2007a) “Back to the Future: Behavioral IR as a Case of Arrested Development,”
International Studies Review, 9: 165–172.
Walker, S. (2007b) “Generalizing about Security Strategies in the Baltic Sea Region,”
in O. Knudsen (ed.) Security Strategies, Power Disparity and Identity, Burlington, VT:
Ashgate, 149–176.
Walker, S. (2009) “The Psychology of Presidential Decision Making,” in G. Edwards
and W. Howell (eds.) The Oxford Handbook of the American Presidency, Oxford: Oxford
University Press.
Walker, S. (2010) “Binary Role Theory and Foreign Policy Analysis,” paper pre-
sented in the Foreign Policy Analysis Workshop, Integrating Foreign Policy Analysis
and International Relations Through Role Theory, at the Annual Meeting of the
International Studies Association, New Orleans, February 16–20.
Walker, S. and Falkowski, L. (1984) “The Operational Codes of U.S. Presidents and
Secretaries of State,” Political Psychology, 5: 237–266.
Walker, S. and Malici, A. (forthcoming) U.S. Presidents and Foreign Policy Mistakes, Stanford,
CA: Stanford University Press.
Walker, S. and Post, J. (2003) “The Search for Causal Mechanisms,” in J. Post (ed.) The
Psychological Assessment of Political Leaders, Ann Arbor: University of Michigan Press.
Walker, S. and Schafer, M. (2000a) “The Operational Codes of Bill Clinton and Tony
Blair,” presented at the Annual Meeting of the American Political Science Association,
September, Washington, DC.
Walker, S. and Schafer, M. (2000b) “The Political Universe of Lyndon B. Johnson and His
Advisors,” Political Psychology, 21: 529–544.
Walker, S. and Schafer, M. (2001) “The Operational Code of Vladimir Putin,” paper
presented at the Annual Meeting of the International Studies Association, Chicago,
February 21–24.
Walker, S. and Schafer, M. (2003) “The Contributions of Operational Code Analysis
to Foreign Policy Theory,” paper presented at the International Studies Association
Midwest regional meeting. St. Louis, Missouri.
Walker, S. and Schafer, M. (2004) “Dueling with Dictators,” paper presented at the
Annual Meeting of the American Political Science Association, Chicago, September
2–September 5.
Walker, S. and Schafer, M. (2006) “Belief Systems and Causal Mechanisms in World
Politics,” in M. Schafer and S. Walker (eds.) Beliefs and Leadership in World Politics, New
York: Palgrave.
Walker, S. and Schafer, M. (2007) “Theodore Roosevelt and Woodrow Wilson,” Political
Psychology, 28: 747–776.
Walker, S. and Schafer, M. (2010) “Operational Code Theory,” in R. Denemark (ed.) The
International Studies Encyclopedia, London: Blackwell.
316 Bibliography

Walker, S. and Schafer, M. (n.d.) “The Contribution of Operational Code Analysis to

Foreign Policy Theory,” in A. Mintz (ed.) Advances in Foreign Policy Analysis (under
review).
Walker, S., Schafer, M., and Marfleet, B. (2001) “The British Strategy of Appeasement,”
paper presented at the Annual Meeting of the American Political Science Association,
August 29–September 2, San Francisco, CA.
Walker, S., Schafer, M., and Marfleet, B. (2008) “The British Strategy of Appeasement,”
in C. Hermann (ed.) When Things Go Wrong: Responding to Adverse Feedback in Foreign
Policy. (under review).
Walker, S., Schafer, M., and Young, M. (1998) “Systematic Procedures for Operational
Code Analysis,” International Studies Quarterly, 42: 173–188.
Walker, S., Schafer, M., and Young, M. (1999) “Presidential Operational Codes and
Foreign Policy Conflicts in the Post-Cold War World,” Journal of Conflict Resolution,
43: 610–625.
Walker, S., Schafer, M., and Young, M. (2003) “Profiling the Operational Codes of
Political Leaders,” in J. Post (ed.) The Psychological Assessment of Political Leaders, Ann
Arbor: University of Michigan Press.
Walker, S., Bohlin, D., Boos, R., Cownie, D., Nakajima, H., and Willson, T. (1984)
“Evidence of Learning and Risk Orientation during International Crises,” British Journal
of Political Science, 14: 33–51.
Wallace, W. (2001) “Europe, the Necessary Partner,” Foreign Affairs, 80: 16–34.
Walt, S. (1987) The Origins of Alliances, Ithaca, NY: Cornell University Press.
Waltz, K. (1979) Theory of International Politics, Reading, MA: Addison-Wesley.
Watson, G. and McGaw, D. (1980) Statistical Inquiry, New York: John Wiley.
Weber, C. (1995) Simulating Sovereignty, Cambridge: Cambridge University Press.
Weiner, T. (2007) Legacy of Ashes, New York: Doubleday.
Weinstein, E. (1981) Woodrow Wilson, Princeton, NJ: Princeton University Press.
Weinstein, E., Anderson, J., and Link, A. (1978/1979) “Woodrow Wilson’s Political
Personality,” Political Science Quarterly, 93: 585–598.
Welch, D. (1992) “The Organizational Process and Bureaucratic Politics Paradigms,”
International Security, 17: 112–146.
Welch, D. (2006) Painful Choices, Princeton, NJ: Princeton University Press.
Weller, L. (1963) “The Effects of Anxiety on Cohesiveness and Rejection,” Human
Relations, 16: 189–197.
Wendt, A. (1987) “The Agent Structure Problem in International Relations Theory,”
International Organization, 41: 335–370.
Wendt, A. (1992) “Anarchy is What States Make of It,” International Organization, 46:
391–425.
Wendt, A. (1994) “Collective Identity Formation and the International State,” American
Political Science Review, 88: 384–396.
Wendt, A. (1999) Social Theory of International Politics, Cambridge: Cambridge University
Press.
White, R. (1984) Fearful Warriors, New York: Free Press.
White, R. (1991) “Empathizing with Saddam Hussein,” Political Psychology, 12: 291–308.
White, S. (1987) “Toward a Critical Political Science,” in T. Ball (ed.) Idioms of Inquiry,
New York: SUNY Press.
 Bibliography 317

Winter, D. (1973) The Power Motive, New York: Free Press.

Winter, D. (2003) “Personality and Political Behavior,” in D. Sears, L. Huddy, and R.
Jervis (eds.) Oxford Handbook of Political Psychology, New York: Oxford University
Press.
Winter, D. (2005) “Measuring the Motives of Political Actors at a Distance,” in J. Post
(ed.) The Psychological Assessment of Political Leaders, Ann Arbor: University of Michigan
Press.
Winter, D., Hermann, M., Weintraub, W., and Walker, S. (1991) “The Personalities of
Bush and Gorbachev Measured at a Distance,” Political Psychology, 12: 215–245.
Wohlforth, W. (1994/95) “Realism and the End of the Cold War,” International Security,
19: 91–129.
Wohlforth, W. (1998) “Reality Check,” World Politics, 50: 655–680.
Wohlforth, W. et al. (2007) “Testing Balance-of-Power Theory in World History,”
European Journal of International Relations, 13: 155–185.
Wolfers, A. (1962) Discord and Collaboration, Baltimore: Johns Hopkins University Press.
Wriggins, H. (1969) The Ruler’s Imperative, New York: Columbia University Press.
Wright, Q. (1942) A Study of War, Chicago: University of Chicago Press.
Wright, R. (2000) Nonzero, New York: Random House Vintage Books.
Youn, H. S. (1999) “Changes in DPRK-Russia Relations 1989–1999,” Journal of East
Asian Affairs, 13: 434–463.
Young, M. (2001) “Building Worldviews with Profiler+,” in G. Barnett (ed.) Progress in
Communications Sciences, Westport, CT: Ablex Publishing.
Zimmerman, W. and Axelrod, R. (1981) “The ‘Lessons’ of Vietnam and Soviet Foreign
Policy,” World Politics, 34: 1–24.
Zinnes, D. (1970) “Coalition Theories and the Balance of Power,” in S. Groennings, E.
Kelly, and M. Leiserson (eds.) The Study of Coalition Behavior, New York: Holt Rinehart
and Winston.
Zinnes, D. (1976) Contemporary Research in International Relations, New York: Macmillan.
INDEX

actors 12, 37, 131, 230; individuals as 7, 174–5, 183, 236–7; philosophical 6,
17, 22–5, 29, 159, 208; states as 7, 17, 55–6, 76, 159, 174–5, 185–6, 228,
22–5, 29, 135 236–7; prescriptive 134–6, 143–9
Agreed Framework 97, 109–10, 158 Bush, George H. W. 197, 200, 252
Al Qaeda 143, 145 Bush, George W. 126–8, 170, 175–86,
alliances 39, 101, 131–3 199–200; optimism 176–7
Almond, Gabriel 268–70
assurance, strategy of 91, 94, 242 Carter, Jimmy 97–8, 108, 110, 173
attitudinal prism 86 Castro, Fidel 87–9, 155, 158–66
Axelrod, Robert 220, 274 Cha, Victor 99, 102
Clinton, Bill 110, 198–9, 235–6, 240–2
balance proposition 49–50 coercive bargaining 99–100, 102–4,
balance of power 134, 229–30, 233 109–10
balance of threat 134, 230 cognitions 9, 12, 33–4, 171, 218
behavioral IR 6–10, 21–4, 29 cognitive consistency see belief stability
belief change 170, 173–5, 182–3, 185–7, cognitive dissonance 55, 185, 189
189–92, 194–5; absolute 190, 192, Cognitive Uncertainty Exchange
194–5, 199–200, 202, 277; active 190, (CUE) 250, 256–8, 261, 286–8
192; directional 190; passive 192 Cohen, Michael 270
belief in ability to control events 118, 126 conceptual complexity 115, 118, 120,
belief stability 171, 176, 186, 200, 216; as 126–8
internal consistency 172; as temporal congruence 50, 136, 246, 261
consistency 172, 186 Consumer Price Index (CPI-U) 197
belief systems 19, 55–6, 59, 87, 105, containment 92–3
134, 165, 174, 177, 186, 249; learning content analysis 26, 44, 61, 70–1, 115,
effects 15, 230–1, 247, 256, 277; 209–11, 224
mirroring effects 227, 236, 239, 249; Correlates of War Project 196
steering effects 15, 227, 239, 252, 276
beliefs: as hierarchical systems 46, 171–2, deceptive bargaining 98, 102–4, 106–7,
174; as opposed to schemas 171; 110–11
diagnostic 134–40, 143–9; decision-processing faults 117–23
instrumental 55–6, 75–6, 162, 164, distrust 115–17, 119, 122–8
 Index 319

Easton, David 280 games: attacker-defender 232, 236,

Edwards, George C. 196 238–40; communication 231,
Einstein, Albert 24, 79, 264, 267 241; conflict 60, 262–3, 285;
Elster, Jon 276 coordination 243; Deed 29–31, 34,
empathy 20, 86; behavioral 91; 45–8, 253–4; mixed-motive 242–62;
psychological 91; realistic 84–5, 87, no-conflict 262, 285; non-zero-
90–1; strategic 58 sum 260, 274; objective 230,
endogeneity problem 116, 129 232–6, 238–40; power 46, 75, 250,
energy 29–31, 70–1, 79–80, 259–61, 264; 262; statesman-revolutionary 67–8;
free 259; potential 260 subjective 17, 59, 61, 65–7, 92–3, 143,
engagement: strategy of 73, 93, 166 209, 212–13, 238–41, 248, 257, 281;
entailment 281 two-player 35; Word 29–31, 34, 45–8,
entropy 13, 15, 65, 79, 246, 249, 253–4; zero-sum 260, 273
258–9, 265; level of 15–16, Genco, Stephen 268–70
58–9, 247; matrix 257, 283, 285, generational effects 189
289; maximum 15, 246–7, 259; Gerson, Michael 188
propositions 13 Goldstein, Joshua 231, 235
equilibrium: Brams non-myopic Goldstein scale 196, 235
(NME) 40, 42, 60–1, 65, 93, 232, 234, grand strategies: appeasement 91,
249–50, 285; Nash myopic 40, 42, 192, 240, 277, 287; balancing 277,
60–1, 65, 92–3, 143, 146, 233, 285; 285–7; bandwagoning 277, 285–7;
Pareto optimal 40, 42, 233, 235 contingent 41–2; dominant 41, 233;
events data 204, 219 hegemony 277, 286–7
exogenous shock see traumatic events Gross Domestic Product (GDP) 192
explanandum 15, 33, 156 group-structural faults 117, 119–22, 126
explanans 15, 32–3, 45, 154 groupthink 112, 276

Feng, Huiyin 97, 173, 175 Habermas, Jürgen 33

Feynman, Richard 10–12, 63–4, 71, 78, Heisenberg, Werner 71
80, 264, 282, 288 Hermann, Margaret 115, 118, 154
field: and subfield 5–10, 21–2, 31, 266, Herrmann, Richard 154
282; cognitive 286; definition of highly enriched uranium (HEU) 97
8–9; dynamic 12, 65; political 30–1, historical analogies 178, 189
39, 43, 64, 253–4, 258; static 8, 30, Holsti, Ole 55–6, 133, 159, 175, 208
249–50 Hudson, Valerie 27, 29
First Encounter Hypothesis 249–52, 281, Hughes, Karen 188
288 Hussein, Saddam 90, 236, 239–41
focal points 55; external 32, 229–30, 232,
235–7, 243–4; internal 32–3, 229–30, inflation 194, 197–8
232, 235–8, 244 Innenpolitik 193–4, 200–2
Foreign Policy Analysis (FPA): Integrated Data for Events Analysis
macrofoundations of 37, 80; (IDEA) 196
microfoundations of 24, 29, 43, 80, International Atomic Energy Agency
129, 271; subfield of 5–10, 21–2, 31, (IAEA) 85, 97
266, 282 International Relations (IR): behavioral
frustration-aggression hypothesis/ model of 17, 28–9, 43–61, 79, 218;
syndrome 51, 87 billiard ball model of 17, 21–43, 78–9;
fundamental attribution error 87, 148 field 9, 22, 29, 31, 266
Iraq 130, 135–41, 145–8, 178, 232–5,
game theory 17–19, 35, 39, 57, 59–60, 251–2
68, 80, 260; contributions of 5, 19, isomorphic 246
68–9, 80, 234, 274, 277, 282;
ordinal 17, 43; rationality rule 68 Janis, Irving 112
320 Index

Jervis, Robert 189, 203, 258, 272–3 Morgenthau, Hans 14, 23, 37, 191, 280
Johnson, Lyndon 173–4 motives 28, 54–5, 60, 113, 116, 119,
128–9, 148, 242, 262–3
Kang, David 100–4
Kennedy, John F. 3, 179 national identity conception (NIC) 101
Khong, Yuen 189 national interest: secondary 38–9, 114,
Khrushchev, Nikita 3, 54 230, 236, 248, 250, 286; vital 38–9,
Kim Il Sung 20, 98–9, 101, 104–5, 107– 230, 232, 234, 236, 248, 250
11, 153–65, 173, 277; Great Leader’s need for power 53–4, 116–17, 119,
Absolutism 107 122–5, 127–8
Kim Jong-il 84, 87–9, 92, 99, 101, network 5; action 33; language 33;
110–11 neural 33–4
King, Gary 196, 207 Newton, Sir Isaac 24, 79, 264, 267–8
Kosovo 231–4, 238–44 nonmyopic equilibrium 35, 40, 42, 61,
65, 68, 74–5, 92–3, 233–4, 249, 255,
Lasswell, Harold 280 257, 285
laws: cast-iron 267–70, 282; plastic 267, norming group 87–9, 91, 136, 161–2,
269–70, 276, 282 224, 236
leadership style 77–8, 276; dogmatic 77; North Korea: nuclear tests 97, 101, 103;
erratic 77; pragmatic 77 first nuclear crisis 97–100, 104, 108,
Leadership Trait Analysis (LTA) 113–17, 110; second nuclear crisis 103, 111
120, 125–7, 129 North Vietnam (NVN) 67–70
learning: complex 159, 206, 231, 241–2, NPT (Non-Proliferation Treaty) 85, 103,
277; diagnostic 159, 165, 206, 231, 107–8, 110, 158
241; experiential 22, 55, 61, 159–60,
163, 189–203, 206, 231, 234, 240, 242; Operational Code: as a neobehavioral
instrumental 206; simple 159, 231, approach 6–7, 80, 267; Barack
240; social 55, 61, 160, 206, 231, 240; Obama’s 75–7; beliefs 57–8, 136–8;
structural adaptation 53, 57, 61, 160, 163 Bolshevik 55; Britain’s 142, 145;
Lenin, Vladimir I. 45, 47, 54, 70, 79, construct 91, 116, 125, 127, 129, 169–
280–1 70, 210, 218, 279; Cuban 160; Fidel
levels of conflict 117–21, 125, 127–8, Castro’s 156, 159, 161–4, 166–8, 173;
192, 200–2 France’s 142; George W. Bush’s 170,
Levy, Jack 159, 189, 205–6, 211, 231, 240 175, 177–8, 181–2, 188; Henry
Kissenger’s 66–7; Jimmy Carter’s 173–
Marfleet, Gregory 205–19, 283 4; John F. Kennedy’s 179; Kim
method: backward induction 60, 65–6; Il Sung’s 104–5, 161, 163, 165;
scientific 10–11 Lyndon B. Johnson’s 173–4;
microfoundations thesis 27 North Korean 160; typology 56;
militarized interstate disputes (MIDs) 192, variables 125–6; Woodrow
196, 201 Wilson’s 54
Milosovic, Slobodan 235–6, 239–42
Mintz, Alex 6, 10, 22–3 Past Hypothesis 248–50, 252, 256–8, 272,
models: complex adaptive systems 5, 281, 288
207, 258–9, 261, 267, 269–70, 273–4; perceptions 80, 84, 87, 89–90, 92–3,
formal 20, 57–61, 226, 243, 283–9; 95, 104, 131, 133–5, 139, 147–9, 166,
rational choice 7, 16, 18, 22, 28, 190–1, 193–4, 199, 208, 214; threat 89,
70; social-psychological 5–6, 10, 28, 132–4, 139, 148
132, 174, 182, 184, 186; strategic Peres, Shimon 173
interaction 5, 13–14, 17–18, 28, 40, Persian Gulf 225, 231–4, 238–42, 244,
57, 67–8, 70, 78, 93, 223–44, 246, 248, 251–2
250, 252–7, 260, 270, 272, 276–7, 280– personality 26, 54–5, 57, 61, 86, 89, 113,
7; symbolic 5, 20, 245–6, 283–9 128, 176, 186–7, 276, 281
 Index 321

Pevehouse, Jon 231 structured-focused comparison 19, 63,

physics: classical 24, 34, 264, 268–9; 80, 88
modern 19, 24, 264, 268; quantum 35, research programs: behavioral IR 3–24,
268 28–9, 80, 112; neobehavioral 6–7,
Pivot Effects Model 273 10, 18–19, 78, 80, 245, 267, 280;
Planck, Max 79 rational choice 7, 9, 16, 18–19, 22–3,
Popper, Karl 268–9 28, 32, 37, 51, 70, 80, 223–4, 228, 232,
power: balance of 100, 134, 229–30, 233, 243
235, 238; cognitive 280; distribution Rice, Condoleezza 175
of 17, 36–8, 79, 100, 156, 246, 249, Robison, Samuel 176, 189–203
257, 260–2, 273–4; elements of 14–15; role: binary 20, 245–66, 270, 272,
matrix 36–7; politics 32, 37–9, 59, 275–6, 279–80, 282–9; change 179,
66, 70, 78–9, 234; social 46, 269–70, 186; competition 50, 247, 251,
272–3, 278, 280–1; struggle for 14, 23, 276–8; conception 49–50, 52, 176,
31, 37, 280; theorem 13–14 247, 255–7, 276; conflict 50, 247,
prisoner’s dilemma 91–3, 96, 239, 285–6, 251, 276–9; counter-role 247–8,
288 255–6, 262; cues 47–50, 250, 252,
problems: n-body 35, 65, 78, 80, 258, 254, 256–62, 274–5, 277, 286–8;
265–6; three-body 34–6, 40, 43, 78, demands 247–8, 250–2, 256, 272,
258, 266; three-sided 20, 257, 264; 275, 277, 288; enactment 49–50, 53,
two-body 34–7, 39–40, 43, 48, 65, 78, 57, 247, 250, 255–6, 276, 279, 287–8;
80, 258, 262, 266–7; two-sided 17, 20, expectations 247–8, 252, 255–6, 272;
42–3, 68, 224, 243, 257, 284 location 247, 250, 252, 255–6, 261–2,
Profiler Plus 116, 129 264, 270, 275–6, 284; making 256,
prospect theory-sequential game 267, 276–7; set 255–6, 262, 272, 284;
model 98, 104–5, 110 strain 247, 250–1, 276–8; taking
psychological milieu 86 256–7, 267, 276–7
psychological characteristics 10, 13, 18, role theory 5, 18, 20, 48–54, 57, 66, 176,
22, 87, 112–29, 210–11, 215–16, 264, 245–67, 270–89
281 Roper Center for Public Opinion
public statements 16; use in content Research 196
analysis 52–4, 57, 72, 179, 210–12, Rose, Gideon 191, 193
224, 231, 235, 238–9 Rosenau, James 8
Putin, Vladimir 73, 175 Rumsfeld, Donald 146, 175
Russia 3, 72–4, 100, 105, 156, 175, 251,
Rabin, Yitzhak 173 277
rationality: bounded 13, 17–18, 28–9, 43–
5, 58, 70, 78–9, 228; instrumental 32, Schafer, Mark 110–29, 134–6, 173–4,
277–9; strategic 20, 37, 69, 143, 145, 178–9, 186, 190, 207–9, 223–44
277; substantive 13, 17, 28, 43–4, 70, Schweller, Randall 191, 285
78–9, 224; theorem 10, 13–14, 17, September 11, 2001 169, 175–7, 181–3,
19–20, 23–4, 28, 38–9, 43, 53–4, 69, 185, 187
79–80; thick 22–3; thin 22 Signorino, Curtis 223
realism 23–4, 131; neoclassical Simon, Herbert 13, 17–18, 266
realism 134, 139, 191, 281; structural Six Party Talks 98
realism 134 Smith, M. Brewster 276
regression models 119, 197, 213–14, 225, Snyder, Glenn 285
228, 235–7 Soviet Union see Russia
Reiter, Dan 189 states of play: initial 60, 65, 68, 78;
Renshon, Jonathan 169–88 intermediate 59, 249, 252, 258–9, 274;
research designs: quasi-experimental final 16, 58–60, 64–6, 249, 252, 254–9,
panel 63, 80, 210; repeated- 262–3, 271, 283–7
measures 63, 80; single-case 19, 63, 72; Stein, Arthur 285
322 Index

strategic interaction episode: Stage I 250, theory complex 232, 274; Theory
255–7, 281; Stage II 250, 255–7, 282 of Inferences about Preferences
strategic orientations 58, 61–2, 74, 132–6, (TIP) 65, 91, 105, 132, 136, 142, 209,
139–44, 147–8, 200, 248; appeasement 229–30, 233, 261; Theory of Moves
tactics 56, 91, 192, 239–42, 283–4, (TOM) 65, 104, 212, 224, 227–8,
286–7; bully tactics 276; compel 233, 249, 254, 261; two-good 278; of
tactics 64–5, 77, 86, 166, 229, 254, uncertainty 245, 247–8, 261
276, 286–7; deter tactics 64–5, 67, 77, threat: balance of 134; external 131–
100, 229, 250, 252, 254–5, 257, 276, 3, 135; threat-cooperation
286–7; exploit tactics 64, 77, 229, proposition 132
254, 276, 286–7; punish tactics 47, time constraints 121
64–5, 72, 77, 88, 137, 140–1, 161–4, Transparency Hypothesis 237
180, 184, 226, 229, 254–7, 276, 286–7; traumatic shock/event 170, 181, 183
reward tactics 47, 64–5, 77, 88, 161–2, typology: belief system 59; leadership
164, 180, 184, 226, 229, 254, 276, style 77–8
286–7
strategies: Assurance 91, 94, 239, 242; U.S. Congress 194, 196; House of
Chicken 91, 242, 286; Prisoner’s Representatives 196; Senate 51–4,
Dilemma 91–3, 96, 239, 242, 285–6, 196, 198–9, 202
288; Stag Hunt 91, 235, 242, 288 uncertainty: principle of 71
structural approaches/explanations 153, unemployment 194, 197–9
166 United States (USA) 67–8, 70
Sung, Kim Il see Kim Il Sung
Systems: closed 259–60; complex Verbs in Context System (VICS) 57, 59,
adaptive 5, 207, 258–61, 267, 70, 87, 115, 161, 178
269–70, 273–4; general 5, 242;
of interest 15, 260–1, 272, 274, Walker, Stephen 1
279; living biological 259; nearly Waltz, Kenneth 27–8, 258
decomposable 266; open 259 White, Ralph 84, 91
Wohlforth, William 191, 285
task focus 116–17, 119–24, 127–8 Wolfowitz, Paul 175
temporal stability 24, 122, 183 world: of beliefs 6, 12–15, 34, 64–6,
Tetlock, Philip 83, 174, 177 245–7, 252, 257, 261, 269, 277, 280;
texts: statements; events 26, 30, 33–4, 54, of events 5–6, 12–15, 26, 28, 30,
245–6 33–4, 44, 64–9, 79; of language 33–4;
Theory: complexity 261; mental 34, 245–6, 248; social 245–6,
constructivist 281; exchange 48–54, 248, 252, 277; two worlds of interaction
57, 66; information 80; liberal- proposition 13, 15, 245, 269
institutionalist 281; neo-marxist 281;
poliheuristic 278; prospect 66–9, Yongbyon 85, 98–9, 107–8, 111
97–9, 102–5, 109–11, 276–9;
realist 37, 43, 45, 59, 79, 281, 284; Zakaria, Fareed 191