OXFORD STUDIES IN AGENCY AND

RESPONSIBILITY VOLUME 1
This page intentionally left blank
Oxford Studies
in Agency and
Responsibility
Volume 1
Edited by
DAVID SHOEMAKER

1
 3
Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
# the several contributors 2013
The moral rights of the author have been asserted
First Edition published in 2013
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
British Library Cataloguing in Publication Data
Data available
ISBN 978–0–19–969485–3 (Hbk)
ISBN 978–0–19–969486–0 (Pbk)
Printed in Great Britain by
CPI Group (UK) Ltd, Croydon, CR0 4YY
 Contents

Notes on Contributors vii

Introduction 1
David Shoemaker
1. The Possibility of Action as the Impossibility of Certain
Forms of Self-Alienation 12
Sarah Buss
2. The Fecundity of Planning Agency 47
Michael E. Bratman
3. Can I Only Intend My Own Actions? Intentions and the
Own Action Condition 70
Luca Ferrero
4. Regret, Agency, and Error 95
Daniel Jacobson
5. Phenomenal Abilities: Incompatibilism and the Experience
of Agency 126
Oisı́n Deery, Matt Bedke, and Shaun Nichols
6. Reasons-Responsiveness, Agents, and Mechanisms 151
Michael McKenna
7. Responsibility, Naturalism, and “The Morality System” 184
Paul Russell
8. The Three-Fold Significance of the Blaming Emotions 205
Zac Cogley
9. Unwitting Wrongdoers and the Role of Moral
Disagreement in Blame 225
Matthew Talbert
10. Partial Desert 246
Tamler Sommers
11. Values, Sanity, and Responsibility 263
Heidi L. Maibom
12. Fairness and the Architecture of Responsibility 284
David O. Brink and Dana K. Nelkin

Index 315
This page intentionally left blank
 Notes on Contributors

Matt Bedke is Assistant Professor of Philosophy, University of British

Columbia
Michael E. Bratman is U. G. and Abbie Birch Durfee Professor in the School
of Humanities and Sciences, and Professor of Philosophy, Stanford University
David O. Brink is Professor of Philosophy, University of California, San Diego
Sarah Buss is Associate Professor of Philosophy, University of Michigan
Zac Cogley is Assistant Professor of Philosophy, Northern Michigan University
Oisı́n Deery is PhD candidate in Philosophy, University of British Columbia
Luca Ferrero is Associate Professor of Philosophy, University of Wisconsin,
Milwaukee
Daniel Jacobson is Professor of Philosophy, University of Michigan
Heidi L. Maibom is Associate Professor of Philosophy, Carleton University
Michael McKenna is Professor and Keith Lehrer Chair, Department of
Philosophy and Center for the Philosophy of Freedom, University of Arizona
Dana K. Nelkin is Professor of Philosophy, University of California, San Diego
Shaun Nichols is Professor of Philosophy, University of Arizona
Paul Russell is Professor of Philosophy, University of British Columbia
David Shoemaker is Associate Professor, Department of Philosophy, Murphy
Institute, Tulane University
Tamler Sommers is Associate Professor, Philosophy Department and Honors
College, University of Houston
Matthew Talbert is Assistant Professor, Department of Philosophy, West
Virginia University
This page intentionally left blank
 Introduction
David Shoemaker

It is my pleasure to introduce the inaugural volume of Oxford Studies in

Agency and Responsibility, a new entry in the Oxford Studies series. The
papers in this volume are excellent illustrations of the important cutting-
edge work being done in this field.
“Agency and Responsibility” is a new label for an old set of unified, yet highly
diverse, concerns. It involves investigation into such familiar questions as these:
 What does it mean to be an agent?
 How (if at all) does the nature of personhood and personal identity across
time bear on questions of agency?
 What is the nature of moral responsibility? Of criminal responsibility?
What is the relation between moral and criminal responsibility (if any)?
Is there a relevant conception of responsibility generally?
 What is the relation between responsibility (moral, criminal, or general)
and the metaphysical issues of determinism and free will?
 What do various psychological disorders (e.g. autism, psychopathy, mental
retardation, insanity, and dementia) tell us about agency and responsibility?
 How do moral agents develop? How does this developmental story bear
on questions about the nature of moral judgment and responsibility?
 What is the will, willpower, and weakness (or strength) of will? What role
can psychology play in the investigation of these issues generally?
 What do the results from neuroscience imply (if anything) for our
questions about agency and responsibility?
 What is the nature of political agency (e.g. citizenship), and how does it
differ, if at all, from moral agency?
 What is the nature of autonomy and how is it related to agency and
responsibility?

Work on these questions, while more or less having an uneasy home base in the
world of moral philosophy, draws from a diverse range of cross-disciplinary
2 David Shoemaker

sources, including moral psychology, psychology proper (experimental,

developmental, etc.), philosophy of psychology, communicative disorders,
philosophy of law, legal theory, metaphysics, neuroscience, neuroethics,
political philosophy, and more. But it is nevertheless unified by its focus on
who we are as deliberators and (inter)actors, embodied practical agents
negotiating (sometimes unsuccessfully) a world of moral and legal norms. It
is an attempt to understand both what it takes to be moral and prudential
beings—agents who reason about and act (or fail to act) on various practical
norms—as well as what it takes to be responsible beings—agents who are
responded to in distinctive ways in virtue of their reasoning about and
acting on such norms. We are engaged, then, to borrow a phrase from
Harry Frankfurt, in the project of “philosophical anthropology”
(Frankfurt 1999, p. ix). This project thus doesn’t fit squarely into any of
the standard subdivisions of moral philosophy—applied ethics, normative
ethics, and metaethics—because in important respects it is presupposed
by, or provides the background conditions for, all of them.
The new papers collected herein were drawn from presentations at the
first New Orleans Workshop on Agency and Responsibility (NOWAR) in
November 2011, and they all grapple with one or more of our questions.
The first half of the papers are roughly about “Agency” and the second half
are roughly about “Responsibility,” although given the interrelated nature
of the topics, several papers take up questions on both sides of the agency/
responsibility map, so the division is somewhat artificial.

Sarah Buss’s paper, “The Possibility of Action as the Impossibility of

Certain Forms of Self-Alienation,” kicked off the workshop and provides
our jump start as well. Buss is interested in the general conditions of
rational agency, and she reveals some serious problems with both the
standard instrumentalist account and the two alternative, constitutivist
accounts. She begins with a simple truth: to act for reasons includes both
taking various features of our circumstances as reasons to do certain things
and being motivated to do certain things by various considerations. But the
relation between these justifying and motivating senses of acting for reasons
is extremely complicated, and ultimately quite puzzling. As part of her
approach, she presents a stratagem: if I am to govern myself, I must render
my motivations authoritative with respect to self-determination, so that my
actions are genuinely attributable to me, the ruling self. But if my ruling self
consists in reason, and my to-be-controlled motivational part consists in
my indifferent-to-reason desires, then how could it be sensible to make
demands of desire based on reasons? One kind of purported solution restricts
reason to a merely instrumental role in aiding our motivational parts’
execution, but this will not work insofar as we can question the desirability
of satisfying any of our desires. The other kind of purported solution
 Introduction 3

attempts to find a way to give reason a greater role in determining the goals
we pursue, so that it is already, in a way, built into the motivational part.
But this kind of solution is problematic, argues Buss, insofar as its advocates
do not build enough into the motivational part. We are left with a real
challenge to agency theorists: how might we understand how the goal-
directed behavior of creatures like us is transformed into behavior we cause
in pursuing those goals? Buss urges that this challenge cannot be met absent
metaethical vexation about the nature of reasons.
Michael Bratman’s essay, “The Fecundity of Planning Agency,” may
be construed as an attempt to answer this challenge (although that is not
how it is explicitly framed). As agents, we are capable of (1) acting—and
organizing our actions—across time; (2) joint action (or shared intention-
ality); and (3) self-governance. A theory of rational agency will gain in
plausibility to the extent that it can provide a unified account of these three
remarkable capacities. Bratman’s conjecture, fleshed out in insightful ways
here, is that these three capacities are grounded in our core capacity for
planning agency. The issue engaging most directly with Buss is that of self-
governance. Bratman attempts to avoid worrisome homuncular models of
the self by locating it (or at least locating its representative voice) in the
agent’s plan-like commitments to what matters (such that they carry weight
in our deliberative thought). As Bratman puts it, “[W]hen these plan states
guide, the agent governs.” And such commitments obviously build on the
agent’s planning agency. The question, then, is to what extent locating the
agent (or the agent’s voice) in such commitments can avoid the worries
articulated by Buss.
Luca Ferrero is also interested in an important aspect of attributability,
of what is involved in an attitude’s being mine. In particular, Ferrero is
interested in one sort of attitude—intention—which is commonly thought
to take as its proper object only the intender’s own action. Call this the
“own action” condition. Ferrero’s aim in “Can I Only Intend My Own
Actions? Intentions and the Own Action Condition,” is to show that the
“own action” condition is false. The relevant aspects of agency deemed
necessary to intention may actually be met without restricting the objects of
the agent’s intention to her own actions. He arrives at this conclusion by
first considering “aimings,” a simpler kind of practical attitude, and he
attempts to show that the own action condition isn’t necessary to aimings,
and to the extent that intentions are not relevantly different in key aspects,
the own action condition isn’t necessary to them either. This conclusion
bears significance for several skirmishes in the field of agency, including
how to properly characterize the relation between intentions and actions,
the fate of the causal action theory, and the nature of joint, or shared,
intention and agency. This last, of course, is one of the three agential
4 David Shoemaker

capacities ostensibly unified by Bratman’s more basic theory of planning

agency.
In “Regret, Agency, and Error,” Daniel Jacobson focuses on a different
psychological aspect of the attributability of actions to agents, namely, what
Bernard Williams called “agent-regret.” For Williams, regret is the general
emotional response one has to the thought, “How much better if it had
been otherwise,” and agent-regret is this sort of response with respect to
one’s own actions and their consequences, i.e. a response to the thought,
“How much better if I had done otherwise.” This construal is representative
of the view Jacobson calls a cognitivist theory of the emotions, and one of
the aims of his rich paper is to cast doubt on this theory. He does so, in part,
by pointing out ways in which a cognitivist theory like Williams’s (and
Stocker’s) conflates distinct sentiments (e.g. regret and guilt) under the
same constitutive thought and also yields some allegedly distinct specific
emotions that have neither coherent expression in action nor any clear
connection to motivation. (And for the important distinction between
emotions and sentiments, see the paper.) Another commitment of the
cognitivist theory is that certain emotions are irrational in virtue of the
falsity of their constitutive thought. This is thought to be illustrated by
Williams’s famous lorry driver case, in which a faultless driver kills a child.
He should feel something bad, we think, and Williams posits agent-regret as
the emotion it is rational for him to feel, given that he can think, “How
much better if I had done otherwise” about “his” action, despite its being a
complete accident. In response to this stance, Jacobson argues persuasively
that the regret of the lorry driver may well be irrational, given that he did in
fact do nothing wrong. Nevertheless, his regret may still be admirable: we
would expect such a response in the virtuous. We might, then, view the
rationality of emotions in terms of their fittingness to the properties of their
objects (and not responses to evaluative judgments), and then preserve our
general sentimental categories of regret and guilt without proposing special
emotions in order to account for Williams’s special cases.
The thought, “Ahh, to have been able to do otherwise!” will be resonant
to those familiar with the long-standing dispute over the relation between
free will and determinism. Taking up a lot of room in this dispute has been
the wrangling over the Principle of Alternate Possibilities, or PAP, which
states that one is morally responsible for some action A only if one
could have done other than A (Frankfurt, 1988, 1). While many theorists
continue to think PAP is true, they strenuously disagree over the
interpretation of the nature of such an ability and, relatedly, what reason
there may be to think we have it. In taking up both issues, many libertarians
(those who believe that we are free agents in a way that is incompatible with
determinism) think phenomenological considerations favor their view. In
 Introduction 5

particular, they say, we experience ourselves from the inside as having the
ability to do otherwise when engaged in moral deliberation and/or action,
an ability that is incompatible with determinism and so provides some
evidence for our incompatibilist freedom. Compatibilists, in return, may
grant that we have some such experience, but deny that it implicates any
sort of incompatibilist ability; rather, they say, the experience we have of
being able to do otherwise is simply of being able to do otherwise if the
circumstances had been different. But if this is all my experience is of, then it
merely implicates an ability that is perfectly compatible with determinism.
Oisı́n Deery, Matt Bedke, and Shaun Nichols take this to be an empir-
ical dispute and so ripe for the methods of experimental psychology. In
their paper “Phenomenal Abilities: Incompatibilism and the Experience of
Agency,” they discuss a series of experiments they performed investigating
how people actually regard what is going on in their experiences of moral
deliberation and decision-making. As it turns out, they found remarkably
consistent results: people across the board overwhelmingly (a) view them-
selves as having the ability to do otherwise in a variety of practical (includ-
ing moral) circumstances, and (b) regard this ability explicitly in
incompatibilist terms. When testing various compatibilist reinterpret-
ations, the authors found that they just weren’t what subjects meant by
their experienced ability to do otherwise, an experience that remained
firmly incompatibilist. This paper is a model of experimental philosophy,
for it identifies an empirical matter on which a certain aspect of a long-
standing philosophical dispute hangs and subjects it to scientific scrutiny. It
then discusses the findings in distinctly (and insightful) philosophical
terms. If we accept the authors’ findings, we must move on from wrangling
over the interpretation of our phenomenal ability to do otherwise to the
role the phenomenology ostensibly plays in support of libertarianism: is it
really evidence for the view?
Alternatively, perhaps the ability to do otherwise isn’t necessary for
freedom at all. This is the course taken by many compatibilists who have
become convinced by the so-called “Frankfurt cases,” scenarios in which a
counterfactual intervener stands at the ready to ensure that some agent
decides as he (the intervener) wants her to (and so removes her ability to do/
decide otherwise) but never does a thing, as she decides all on her own to do
the hoped-for action. But if PAP is false, what does the sort of control
implicit in freedom and responsibility require? One popular move has been
to view free agents as those who are reasons-responsive. The problem with
doing so is that it looks as if agents in Frankfurt cases actually aren’t
reasons-responsive, as they wouldn’t have been sensitive to the reasons
not to perform the hoped-for action. In responding to this worry, Fischer
and Ravizza (1998) have argued that their favored notion of guidance
6 David Shoemaker

control over some action consists in its issuing from the agent’s own
reasons-responsive mechanism, something which can be sensitive to
reasons to do otherwise in Frankfurt cases, even if the agent herself isn’t.
Michael McKenna, in his intricate paper, “Reasons-Responsiveness,
Agents and Mechanisms,” traces this dialectic with great care before discuss-
ing some serious problems with a mechanism-based approach to the nature of
compatibilist control, in particular the problem of individuating the specific
mechanisms operative in cases of free action. This problem, McKenna argues,
is really symptomatic of a deeper, structural problem with mechanistic
theories generally: if (as seems plausible) we allow that various sub-mechan-
isms clearly interact with one another to issue in different types of action, we
seem driven to attribute actions to “the mechanism” that can somehow
include all such simultaneously-operating sub-mechanisms, but then this
mechanism looks an awful lot like a full-fledged agent once more. So if we
are forced back to an agent-based reasons-responsive view, are we also back to
accepting PAP (given that it seems these types of views give the wrong answer
in Frankfurt cases)? No, argues McKenna: it is possible for an agent to be
reasons-responsive in the Frankfurt case without being able to react to reasons
to do otherwise when given sufficient reason to do so. Developing and
defending this intriguing proposal takes up the remainder of his essay.

Turn, then, from a focus on agency to a focus on responsibility. In recent

years, most discussions of moral responsibility jump off from P. F. Strawson’s
deeply influential “Freedom and Resentment” (Strawson 1962), and ours is
no different. In “Responsibility, Naturalism and ‘the Morality System’,” Paul
Russell offers a thoughtful discussion of Strawsonian naturalism about moral
responsibility, as part of his defense against construing both Strawson and the
general notion of moral responsibility too narrowly, constrained by and
within “the moral system” and so defined by concepts like obligation and
wrongness. One worry about this construal is that it has us dealing with
positive and negative actions in an asymmetrical fashion, for it views our
responsibility-assessments-via-reactive-emotions as roused only by violations
of norms, but this is a one-eyed view of our rich emotional and interpersonal
lives. Another worry is that this approach is too locally biased, a responsibility
conception steeped exclusively in modern, Western, Christian culture,
implausibly marking us off conceptually from the ancient Greeks, say, and
other shame-based cultures, whose activities we must now regard as non-
responsible, as alien to our own sort of ethical lives. But this has the cost of
removing any apparatus enabling us to criticize and engage with those cultures,
which we seem quite able to do. The third and most serious worry, however, is
that construing responsibility so narrowly, as so culturally located, generates
the possibility that our responsibility practices need an external justification, a
 Introduction 7

possibility that opens the door again to deep skepticism about responsibility
generally. Russell, in the remainder of his essay, tries to close this door by
offering a middle-path naturalism about our ethical reactive attitudes—
motivated by Bernard Williams’s critique of “the morality system”—one
that charts a path between Strawson’s own overly wide approach (given his
inclusion of too many emotional reactions as ostensibly relevant to moral
responsibility, e.g. love and friendly feeling) and R. Jay Wallace’s overly
narrow one (which distorts our understanding of human ethical life and
looks conceptually imperialistic).
One of the many innovations of Strawson’s approach was to view being
morally responsible as somehow a function of being regarded as morally
responsible, where to be regarded as such is just to be the target of various
sorts of reactive emotions (although there are significantly different ways to
interpret this view; see Brink and Nelkin’s paper for a realist—in contrast
to a response-dependent—understanding of Strawsonian responsibility).
Strawson divided the sorts of pleas that get one off the hook as a target of
such reactions into two groups, what Watson (1987) has helpfully called
“excusing” and “exempting” pleas. Excusing pleas—“It was an accident”;
“I didn’t know”—are those that get the agent off the hook for a particular
action but continue to leave her vulnerable to reactive emotions generally.
Exempting pleas—“She’s only a child”; “He’s a hopeless schizophrenic”—
remove the offender from the community of morally responsible agents
altogether. What Strawsonian theorists have searched for, then, are ways of
theoretically unifying these categories. In other words, is there any unified
grounding for the inappropriateness of resentment for accidents, ignorance,
insanity, and the like? Some have thought there is theoretical unity given
the sanctioning aspect of (negative) reactive attitudes, where the unity
conditions are grounded in the fairness of such sanctions. Others have
sought the appropriateness conditions in the expressive nature of the
reactive emotions (such that some emotions are unfitting in virtue of their
failure to meet the felicity conditions of certain forms of communication).
Others have found unifying ground by focusing on how the reactive
emotions target certain qualities of will.
If Zac Cogley, in “The Three-Fold Significance of the Blaming Emo-
tions,” is right, this search for a unified treatment may be quixotic. Cogley
first draws from recent psychological research to articulate three distinct
functions of the (negative) reactive emotions: appraisal, communication,
and sanction. Second, Cogley shows that each of these three functions
imports its own distinct set of appropriateness conditions: while a blaming
emotion may meet the conditions rendering it an accurate or fitting
appraisal of responsibility, say, it may not meet the conditions rendering
it a fair or deserved sanction. This point leads to a third: different (unifying)
8 David Shoemaker

theories of responsibility are typically motivated by a focus on different

functions of the blaming emotions. This fact of course could explain their
disagreements about the nature of moral responsibility, but it may also
help explain the long-standing dispute between compatibilists and in-
compatibilists: if compatibilists focus on appraisal and/or communicative
functions of blame, whereas incompatibilists focus on the sanctioning
functions (and each has different appropriateness conditions), it is no
wonder they wind up with radically different metaphysical views.
Cogley’s sort of analysis provides a tantalizing glimpse of progress in
this well-entrenched debate.
Turn, then, from the general significance of blame to its specific appro-
priateness conditions. As already mentioned, one of the assumed excusing
conditions is ignorance. This assumption is widely held across all three sorts
of theories of moral responsibility. In particular, the thought is that if one is
morally ignorant—non-culpably ignorant of moral principles rendering
what one is doing is morally wrong—then it would be inappropriate to
blame one, regardless of the differential functions of the blaming emotions.
Matt Talbert, in “Unwitting Wrongdoers and the Role of Moral Disagree-
ment in Blame,” argues strenuously against this assumption. For Talbert, to
blame someone is to have certain attitudes responding to features of the
blamed agent’s actions/attitudes that are objectionable from the perspective
of the blamer, and often the features we find blameworthy are judgments
that our interests don’t count (or don’t count sufficiently) to the blamed
agent. But these features may well be present in those who act from non-
culpable ignorance, and so unwitting wrong-doers may well be blame-
worthy. What Talbert’s move does is undercut a recent kind of skepticism
about moral responsibility premised on the assumption about the excusing
status of moral ignorance. It also reveals the importance and role of moral
disagreement: what actually matters in our moral disputes is our judgment
about the significance of the various interests and needs being affected by
the offending agent, and not whether this agent knows (or it was reasonable
for her to know) the moral principles rendering her actions wrong.
Tamler Sommers, in “Partial Desert,” also focuses on features external to
blamed agents that may nevertheless ground blame. While he admits that
facts about the agent’s knowledge and ability to control her actions specify
a range of appropriate responses, what Sommers adds to the mix is
the thought that the victim’s responses are relevant to narrowing down
reactions within that desert range. To make the view more intuitively
appealing, Sommers draws on the moral luck literature in describing two
cases. In both, a drunk driver kills a girl who happens to run out in the
road, then keeps on driving. In the first variation, the state catches him and
puts him to death by firing squad (it’s Utah). In the second, he’s tracked
 Introduction 9

down by the father of the girl, who shoots him. In both cases, the driver
dies, but only in the second might we think he deserved his fate. This
reaction, Sommers argues, reveals our commitment to there being facts
external to the punished agent and what he knew or did that are relevant
to assessments of desert. In particular, it matters whether the offended
parties carry out (or at least contribute to a determination of) the punish-
ment. His defense of this proposal carries him into interesting territory,
including an important exploration of whether the proportionality
required for just legal sentencing and moral sanctioning must be ordinal
as well as cardinal, i.e. an exploration of whether like cases really must be
treated alike.
While Sommers exclusively focuses on moral desert and responsibility,
some have thought that we can achieve real insight into features of moral
responsibility by thinking about criminal responsibility. There are, for
instance, several explicitly articulated aspects of criminal responsibility—
excuse, justification, mens rea, fault, and liability—that seem to play an
analogous and important role in moral responsibility too (for doubts
about this ostensible relation, however, see Shoemaker Forthcoming).
Susan Wolf, for example, has famously appealed to the legal notion of
insanity to ground a normative competence condition for moral responsi-
bility. She discusses the case of JoJo, the son of an evil dictator, who
soaks up his daddy’s values and, when grown, treats the peasants in the
same horrible ways as dad. Wolf ’s thought is that, insofar as JoJo isn’t
viewed as morally responsible (a dubious finding, however; see Faraci
and Shoemaker 2010), it is because his deep self (his reflecting, valuing
self) isn’t normatively sane, i.e. he is unable to recognize the correct
moral values. Insofar as previous compatibilist theories hadn’t taken this
fundamental orientation to the good as necessary to moral responsibility,
they were incomplete.
Heidi Maibom, in “Values, Sanity, and Responsibility,” sees the Wolfian
position as dilemmatic: either this understanding of insanity is supposed to
capture the legal notion or it is not. If it is supposed to do so, then it
overreaches, insofar as legal insanity typically incorporates only those
suffering from delusions and hallucinations, and not those who merely
have different values than ours; indeed, the legally insane usually com-
pletely share our values. If, on the other hand, the Wolfian view advances a
different understanding of moral insanity, then it is wrong-headed, as it
would require us to cease regarding as responsible those we clearly do,
namely, those who are merely intransigent in holding different values than
ours. As long as there is an available inferential route for them from their
own values to beliefs about wrongness we can share, they are eligible for
responsibility assessments for failing to do so. If we didn’t allow for this
10 David Shoemaker

possibility, then we would have to limit our responsibility assessments only

to those who acted wrongly-while-fully-aware-what-they-were-doing-was-
wrong, but this is to implausibly restrict our responsibility community.
In the final selection of the volume, “Fairness and the Architecture of
Responsibility,” David Brink and Dana Nelkin further explore and draw
from the relation between criminal and moral responsibility to provide a
theory of the framework of general responsibility, a conception that is
unified, on their view, by its commitment to the target’s having a fair
opportunity to avoid wrong-doing. On Brink’s and Nelkin’s picture,
drawing from both criminal and moral responsibility enables us to avoid
some unfortunate elisions in each realm. For example, moral theorists often
overlook the importance of situational factors in determinations of respon-
sibility, something which legal theorists tout. Even when agents are norma-
tively competent, there may be features of the situation—e.g. involving
coercion or duress—that undermine their ability to avoid wrong-doing.
But there are also features emphasized by moral theorists that legal theorists
sometimes overlook or wrongly discount, for example, the importance of
volitional impairments (e.g. paralyzing fear or depression) as providing
excuses from responsibility. On their overall model, cognitive and vol-
itional elements are of equal importance in enabling people to have the
fair opportunity to avoid the sort of wrong-doing to which we typically
respond with moral blame and, in the criminal realm, legal sanctions. And
because the various elements in their architectonic are scalar, there is a
coherent notion of partial responsibility ultimately implied by their view,
and they offer interesting suggestions for how this notion may be applied
and dealt with in the criminal realm.

In each case above, I have only skimmed the surface of these rich essays. But
what I have said should be sufficient to reveal both the variety of topics
falling under the rubric of “agency and responsibility” as well as their
interesting and variegated relations to one another. While most writing
on responsibility focuses solely on the negative, blaming, reactions to the
deeds of others, my reaction to the contributions of the authors herein is
nothing but philosophical and personal gratitude.1

1 My thanks to all of the authors, both for their contributions to this volume and for
their comments on an earlier draft of these introductory remarks. I am also deeply
grateful to Steve Sheffrin and the Murphy Institute at Tulane University for financing
the first workshop on agency and responsibility in November 2011, and to Nathan
Stout, Jason Tyndal, and Frankie Worrell for compiling the index. Finally, thanks to
Peter Momtchiloff at Oxford University Press for his early and enthusiastic support of
this new series.
 Introduction 11

REFERENCES

Faraci, David, and Shoemaker, David. (2010). “Insanity, Deep Selves, and Moral
Responsibility: The Case of JoJo.” Review of Philosophy & Psychology 1: 319–32.
Fischer, John Martin, and Ravizza, Mark. (1998). Responsibility and Control.
(Cambridge: Cambridge University Press).
Frankfurt, Harry G. (1988). The Importance of What We Care About. (Cambridge:
Cambridge University Press).
—— (1999). Necessity, Volition, and Love. (Cambridge: Cambridge University Press).
Shoemaker, David. Forthcoming. “On Criminal and Moral Responsibility.” In
Mark Timmons, ed., Oxford Studies in Normative Ethics, Volume 3 (Oxford:
Oxford University Press).
Strawson, P. F. (1962). “Freedom and Resentment.” Proceedings of the British
Academy 48: 1–25.
Watson, Gary. (1987). “Responsibility and the Limits of Evil: Variations on a
Strawsonian Theme.” In Ferdinand Schoeman, ed., Responsibility, Character, and
the Emotions: New Essays in Moral Psychology (Cambridge: Cambridge University
Press, 1987): 256–86.
 1
The Possibility of Action as the
Impossibility of Certain Forms
of Self-Alienation
Sarah Buss

I begin with an extremely simple observation. Whenever we do things

for reasons, we do what we do because we regard certain features of our
circumstances as reasons to behave in certain ways and not others, and
because we are motivated to behave in certain ways and not others. But
what is the relation between these two elements of acting for a reason? What
is the relation between our capacity to respond to reasons and our practical
impulses? In trying to answer this question, we discover that what seems so
very simple is really quite complicated.
The complications that interest me here arise from the fact that agents
who do things for reasons (“rational agents”) can reflect on their own
psychological states, and in so doing, relate to these states as features of
their circumstances. This means that they can relate to their own impulses
to do things (their inclinations, desires, and whatever other conditions
move them to initiate change) as features of their circumstances. And this
means that they can be alienated from themselves in a way that threatens
their capacity to act. An instance of behavior constitutes an agent’s action
only if the agent herself is its direct cause. But if it is possible for an agent to
dissociate herself from, and even repudiate, her own practical impulses,
then it is possible for these impulses to cause her to behave in certain ways
without this behavior being attributable to her.
Given this possibility, an account of what is required in order for rational
beings to act must differ significantly from an account of what is required in
order for nonrational beings to act. Precisely because the latter sort of
beings cannot be alienated from their own psychic states in the way
I have just indicated, we do not need to tell a story about why behavior
caused by various elements of their psyche qualifies as their actions. But if
 The Possibility of Action 13

a being can be self-alienated in the relevant way, then the fact that she
engages in instrumentally rational, goal-directed behavior does not suffice
to ensure that the goals she pursues are her own. The fact that she behaves
in a way that would qualify as an action if she were not a rational being is
compatible with the fact that she is alienated from her behavior; and so, it is
compatible with the fact that her behavior does not qualify as her action;
and so, an adequate account of what is required in order for her to act will
have to be an account of the conditions under which she is not alienated
from herself in the relevant way.1
In the first half of this paper, I will press the case against the assumption that
rational beings succeed in acting as long as they engage in instrumentally

1
The point is not, as Jennifer Hornsby puts it, that we are under “pressure to add an
extra ingredient in order to reveal [a rational agent] as a more or less reasonable,
conscious being” (Hornsby 2004: 185). The point, rather, is that we are under
pressure to identify the extra ingredient that is present when a rational being does not
dissociate herself from her motives. It is perhaps worth noting at the outset some of the
ways in which my critique of the standard picture of action is at once similar to and
different from the sort of challenges that target what Hornsby calls “the naturalistic
assumptions” at the heart of this picture. Like Hornsby, I will be arguing that there is a
form of self-alienation which is a threat to even less-than-“robust” agency. More
particularly, I share the concern that insofar as one relates to oneself as a feature of the
world, one is a bystander to one’s own motives, and so one is not the direct cause of
anything that is directly caused by these motives; i.e. their causal efficacy is not an
instance of one’s own agency. Whereas, however, Hornsby and others who challenge so-
called “causal” accounts of action think it is a mistake to characterize bodily actions as
events caused by agents, I do not challenge this assumption. (This assumption, it is
important to stress, is not the assumption that “an action . . . is something on the physical
side of a supposed mental/physical divide and called a bodily movement.”(180) Rather,
the claim is that bodily actions are bodily movements insofar as they have a certain
nonphysical status—which includes, but (as the requirement of instrumental
rationality makes clear) is not exhausted by, their status as the effects of mental causes.
Note that, Hornsby’s suggestion to the contrary notwithstanding, the same story applies
to actions that consist of an agent’s intentionally refraining from moving her body in
various ways.)
In addition to these respects in which my criticism resembles and differs from the
typical challenges to causal theories of action, there are others. Whereas Hornsby thinks
that the problems with the standard conception reveal the misguidedness of any attempt
to account for action in terms of the “mental states” of the agent, I think that we can offer
a diagnosis of what has gone wrong without ruling out the possibility of such a reduction.
In other words, it seems to me that taking an “external perspective” on oneself can “pose
a genuine threat to our agency,” even if our actions can be “unproblematically accom-
modated in the naturalistic explanatory order” (181). (This is true, even if, as I suggest, to
do something for a reason, one must take oneself to be responsive to reasons, and even if
it is a mistake to offer a “naturalistic” account of reasons.) Finally, as I hope to make
clear, one can claim that “X’s desiring something and believing something [can be]
translated into talk of items with causal potential” without claiming that “X’s having a
reason is . . . a matter of the existence of a pair of states” (180). Indeed, as far as I can tell,
very few advocates of the standard conception endorse the latter claim.
14 Sarah Buss

rational, goal-directed behavior. I will challenge this assumption by calling

attention to the problem posed by the capacity for self-alienation. In so doing,
I will side with those who insist that an adequate account of action (not just
“robust,” or “full-blooded,” or “self-governing” action, but whatever it is that
suffices to qualify a person’s behavior as voluntary) will be an account of what
is distinct about someone who “identifies with,” or “endorses,” or “shares” the
goals that her desires (and other psychic conditions) move her to pursue.
Whatever form such an account will take, it must explain how (i) a
person’s assumptions (conscious or unconscious) about what she has reason
to do and (ii) her nonrational impulses (whatever inclinations, desires, etc.
are neither constitutive of her rationality nor the product of her reasoning)
can be sufficiently unified, or integrated, to constitute two aspects of a
single point of view on her behavior. Unless an account of action accom-
plishes this task, it will not explain why a rational being’s behavior can be
attributed to her, the one who reflects on her impulses and wonders
whether she has reason to endorse them. In the paper’s second half, I will
consider two proposals regarding how this integration might be affected.
Both accounts fail, I will argue, and for the same reason: they assume that
an agent’s reason and her nonrational practical impulses play entirely
different roles in the production of action; and they thus lack the resources
necessary for explaining how these elements interact to constitute an agent.
When the advocates of these accounts refer to “the agent’s” exercise of
authority over herself, or “the agent’s” understanding of what she herself is
inclined to do, they help themselves to a self-reflexive relation they cannot
make sense of in their own terms.
Rational action can be an agent’s way of “constituting herself,” I will
argue, only if the relationship between a rational being’s reason and her
nonrational desires to bring about various changes is a partly internal
relationship—only if, that is, these two parts of the soul are already partially
integrated. In order for desires to figure among the constituents of a “self”
that is committed to responding to reasons, desiring must render the desirer
vulnerable to the commitments and demands she makes in her rational
capacity. Qua subject of desires, she cannot be indifferent to her normative
assessments, even if her desires cannot be attributed to these assessments.2

2
This is a point Aristotle makes in the Nicomachean Ethics: “In the soul too there is
something contrary to the rational principle, resisting and opposing it. . . . Now even this
seems to have a share in a rational principle, as we said; at any rate in the continent man it
obeys the rational principle—and presumably in the temperate and brave man it is still
more obedient; for in him it speaks on all matters, with the same voice as the rational
principle. Therefore the irrational element also appears to be twofold. For the vegetative
element in no way shares in a rational principle, but the appetitive, and in general the
desiring element in a sense shares in it, in so far as it listens to and obeys it; this is the
 The Possibility of Action 15

It follows, I will argue, that rational beings cannot engage in goal-directed

behavior without assuming that they are responding to mind-independent
reasons to behave in this way. They must make this assumption because the
self-integration necessary for them to initiate their behavior requires them to
regard themselves as responding to the same sort of thing—reasons—in both
their capacity as rational beings and their capacity as beings with substantive
goals. If a person’s nonrational impulses do not render her open to the
demands of her reason, then insofar as she is the subject of these demands,
these impulses are just so many features of her circumstances. As such, they
pose no more threat to her integrity than anything else she must “take into
account” in deciding what to do. Accordingly, she has no need to integrate
them into herself in order to act.

THE STANDARD ACCOUNT: ACTION

AS INSTRUMENTALLY RATIONAL
GOAL-DIRECTED BEHAVIOR

Most people who think about what is involved in doing something volun-
tarily assume that an adequate account of action will apply equally to
(i) those, like squirrels, to whom it is a mistake to attribute any opinions
about what they have reason to do and (ii) healthy, mature human beings
who form such opinions quite frequently, even though they rarely do so
self-consciously. According to this widespread assumption, when it comes
to plain old acting, a rational being need not satisfy any conditions other
than those that suffice for the action of a nonrational being.
But this assumption is false. For it overlooks the fact that the capacity to
reason is the capacity to dissociate oneself from the very sort of psychic
forces whose causal role would yield an action if they were attributable to a
nonrational being. Whereas having a behavior-determining impulse to
bury a nut is all that it takes for a squirrel to have the goal of burying a
nut, a rational being need not share the goal of her desire at a given moment
in time, no matter how powerful this desire may be, no matter how nicely it

sense in which we speak of ‘taking account‘ of one’s father or one’s friends, not that in
which we speak of ‘accounting’ for a mathematical property. That the irrational element
is in some sense persuaded by a rational principle is indicated also by the giving of advice
and by all reproof and exhortation. And if this element also must be said to have a
rational principle, that which has a rational principle (as well as that which has not) will
be twofold, one subdivision having it in the strict sense and in itself, and the other having
a tendency to obey as one does one’s father.” (Aristotle, Nicomachean Ethics, Bk. 1, ch.
12, 1102b25–1103a3)
16 Sarah Buss

may cohere with her other desires and commitments, and no matter how
confident she may be that some considerations do count in favor of the
goal-directed behavior. Any conception of agency that suggests otherwise is
thus inadequate. It is inadequate because it implies that a form of self-
alienation is compatible with agency, even though in relating to one’s
behavior in this way, one would be no less passive than one is when one
suffers a blow.
So, at any rate, I hope to show. I will argue, first, that it is not possible to
act if one dissociates oneself from the motivating force of whatever psychic
condition directly causes one’s behavior. I will then point out what this
implies: in order for a rational being to act, it is not enough that she engages
in instrumentally rational, goal-directed behavior.
Let me begin by calling attention to a metaphor. Most philosophers
seem to agree that the paradigm case of an agent who is not responsible for
what she does is someone who is, as Harry Frankfurt puts it in characteriz-
ing the “unwilling” addict, a “passive bystander” to her own motives
(Frankfurt 1988). But the metaphor of the “passive bystander” reveals a
problem: if there is anything that being an agent is not, being an agent is not
being a “passive bystander” to the direct mental causes of one’s behavior.
Insofar as one is a passive bystander to an event, this event is not one’s own
doing. It is not one’s own action.
Of course, it is possible for me to reflect on my actions even as I am
engaged in them: here I am, typing away at my keyboard, even as I am
observing the movements of my own fingers. But, as this example makes
clear, the point of the metaphor is not simply that we can take a detached
stance toward our own actions, but that we can be opposed to what we
ourselves do, even as we are doing it for a reason. Indeed, there is more to it
than this. Even if we very much wished that we had better alternatives, and
even if we keenly appreciated what compelling reasons we have to act
otherwise, we would not necessarily be opposed to our behavior in the
relevant sense. The point of the metaphor is that—with eyes wide open, as
they say—we can prefer to do one thing under the circumstances, even as we
are intentionally doing another. We can do this, moreover, in cases where
the action is not merely “reflexive”—cases, that is, in which what we do is
not a spontaneous response to a sudden change in our circumstances.3
Since, as Frankfurt reminds us, it would be possible to satisfy the
stipulated conditions only if it were possible for the forces that move us
to render us passive bystanders to their motivating power, and since insofar

3
For an example of a reflexive movement, think of a case in which one quickly moves
one’s hand in front of one’s face when a baseball suddenly appears in close proximity to
one’s head.
 The Possibility of Action 17

as someone is a passive bystander to an event, she is not the agent of this

event, the metaphor reveals that no such self-alienated intentional action is
conceptually possible. The behavior of someone who was alienated from
her motives in this way would not be caused by her—the one who is
committed to responding to reasons. It would not be attributable to her
even if, as is surely the case, she took herself to have some reason to behave
this way. As we are to imagine the case, her assumption about what it makes
sense for her to do cannot account for the force of the desire that moves her
to do it. Accordingly—and this is the point of the metaphor—the force that
“moves her” is a force that prevents her from moving herself. And so—and
this is what the metaphor reveals—this force prevents her from acting.
The same point can be made in a way that anticipates the divided soul models
I will discuss in the paper’s second half. Qua rational being, Sarah’s goal is to do
what she takes herself to have sufficient reason to do; qua nonrational being, her
goal is to do otherwise (though not, of course, under that description!). But no
one can wittingly commit herself at one and the same time to realizing two
incompatible goals. We can have conflicting desires, wishes, hopes, fantasies,
etc. But we cannot regard something as our goal while wittingly committing
ourselves to behaving in a way that will ensure that we do not achieve it. This is as
impossible as walking off in two different directions at once.4
This image calls attention to the fact that the conceptual limitation on
the possibility of self-alienation can be expressed in the functionalist terms
popular among many theorists of action. As Jay Wallace notes, future-
directed intentions can play their role in long-term planning only if they are
in harmony with the agent’s normative judgments. This is, roughly,
because if I believe that it would be bad for me to do X in two weeks,
then I hope I will not do X in two weeks. But if I hope I will not do X in two
weeks, then I am open to being moved by any considerations against doing
X in two weeks. But then I am not in the mental state definitive of the
future-directed intention to do X. I am not in the state whose function it is
to lead me not to be moved by the considerations (or at least most of the
considerations) that count against my doing X (Wallace 2001: 8).
Like most other philosophers, Wallace thinks that a similar argument
cannot be constructed for tying present-directed intentions to normative
judgments (Wallace 2001: 8). But I don’t see why. Why should the
addition of a longer time interval make a difference? If I believe that I
have overriding reason not to do X right now, then I am in a state whose

4
For a discussion of the principle of instrumental rationality in which I argue that it is
not possible to wittingly will to do what is incompatible with achieving one’s end under
this description, see Buss, “Norms of Rationality and the Superficial Unity of the Mind,”
Unpublished Manuscript.
18 Sarah Buss

function it is to lead me to hope that I will not do X right now; and so I am

in a state whose function it is to lead me to be moved by the considerations
against doing X right now—whichever such considerations I am right now
aware of. Of course, I can form two incompatible intentions when I am
unaware of doing so. But, again, just as it is impossible for me to walk in
two different directions at once, so too, it is impossible for me to knowingly
commit myself to walking in two directions at once. This means that
I cannot knowingly commit myself to—at one and the same time—being
moved by the considerations against doing X right now and not being so
moved. And this means that I cannot commit myself to contravening my
own contemporaneous normative judgment.
More carefully, I cannot wittingly commit myself to contravening my own
contemporaneous normative judgment insofar as I am committed to respond-
ing to reasons. Under special circumstances I might appraise the desirability of
a given course of action out of idle curiosity, with no intention of treating my
own verdict as a guide to my action. In the ordinary case, however, I am
committed to being so guided. So were I to intend to do X without altering
my conviction that I have overriding reason not to do X, I would wittingly
commit myself to doing what I am committed to not doing. And so, again,
I would be making a commitment I cannot myself credit as such.
Note that in making this point, I am not endorsing the controversial
thesis that “the will is (just) reason in its practical capacity.” Though in the
usual case, a person forms a judgment concerning what she has reason to do
in response to her commitment to being guided by such judgments, it
hardly follows that all her commitments are determined by her normative
judgments. To the contrary, a person’s action is always one of several
possible actions that would have been compatible with her all-things-
considered verdict concerning what she has sufficient reason to do.
(Think, for example, of the countless possible variations on the theme of
“going up stairs.”) Every action thus involves an “exercise of will” that
cannot be reduced to the practical employment of reason. In committing
oneself to complying with a given normative verdict in-this-way-rather-
than-that, one is doing something that cannot be fully explained by the
normative verdict itself.5 My point is simply that it is not possible to will to
act contrary to one’s contemporaneous normative judgment. Though the
conceptual limits of willing are not determined by the conceptual limits of
our normative judgments, when we impose a normative constraint on our
will, we impose a conceptual constraint as well.6

5
Buridan’s ass choices are a special instance of this phenomenon.
6
For an account of weakness of will that does justice to this relationship between
practical reason and the will, see Buss 1997.
 The Possibility of Action 19

Interestingly, support for this thesis can be found in the testimony of people
who are widely agreed to be the paradigm examples of agents overpowered by
inclinations they wish they could resist: those suffering from obsessive-com-
pulsive disorder (OCD). Consider, for example, the anguished testimony of a
man who could not overcome his powerful compulsion to return, over and
over, to the place where he thought he might have hit someone with his car:
The pain is a terrible guilt, that I have committed an unthinkable, negligent act.
At one level, I know this is ridiculous, but there’s a terrible pain in my stomach
telling me something quite different. . . . The anxious pain says to me, “You Really
Did Hit Someone.” The attack is now in full control. Reality no longer has meaning.
My sensory system is distorted. I have to get rid of the pain. Checking out this
fantasy is the only way I know how.
I start ruminating, “Maybe I did hit someone and didn’t realize it. . . . Oh my
God! I might have killed somebody! I have to go back and check.” Checking is the
only way to calm the anxiety. It brings me closer to truth somehow. I can’t live with
the thought that I actually may have killed someon—I have to check it out.
My fantasies run wild. I desperately hope the jury will be merciful. I’m particu-
larly concerned about whether my parents will be understanding. After all, I’m now
a criminal. I must control the anxiety by checking it out. Did it really happen?
I think to myself, “Rush to check it out. Get rid of the hurt by checking it out.”
I’m now close to hysteria because I honestly believe someone is lying in the brush
bleeding to death. Yes,.. the pain makes me believe this. (Rapoport 1991: 24)
In short, even though there is an obvious sense in which someone responding
to the characteristic obsessions and compulsions of OCD wishes that she did
not behave as she did (and even though there is usually “a part of [her] psyche
[that] keeps telling [her] that this checking behavior is ridiculous” (Rapoport
1991: 26)), there is an equally obvious sense in which, at the moment
of action, she is convinced that she has no better alternative, under the
circumstances. She believes that nothing is more important than behaving
as she does; and her understanding of why it is important incorporates
both the irrational “fantasies” and the very real pain that accompanies the
fantasies. Most importantly for our purposes here, her experience as of being
overpowered by her compulsion is not the experience of being passive in
relation to her behavior. To be sure, she is passive in relation to the pain and
anxiety. But her response to her agony is the response of someone who is
absolutely (and not unreasonably, given the intensity of the pain, and given
that nothing else is likely to stop it) convinced that she really has no viable
alternative. Her compulsion dominates her by determining her point of view.
Her problem is precisely that she cannot dismiss her fantasies as alien to her
own beliefs and concerns.7

7
For a development of this account of how “compulsions” prevent agents from being
accountable for their actions, see Buss 2012.
20 Sarah Buss

To see that reports of what it is like to be in the grip of compulsions call

our attention to a conceptual limitation on self-alienation, it may help for
us to focus our attention on how things change when we cross the border
from the voluntary to the involuntary. Imagine, then, that someone is
hanging from a high ledge. She really, really, really does not want to fall
to the ground. Nonetheless, at some point, she lets go. What has happened?
It could be that the growing pain and discomfort of resisting the pull of
gravity, together with the dawning conviction that no one is going to show
up to help, led her to give up, or give in—to decide (however unself-
consciously) to stop trying to resist. Alternatively, it could be that she never
wavered in her commitment to hold on, but that eventually her finger
muscles simply lost their strength. According to the first hypothesis, letting
go was an intentional action. According to the second hypothesis, letting go
was involuntary behavior.
Is there a third possible scenario? One that appeals to an irresistible desire
to let go? So as to distinguish the “pull” of such a desire from the “pull” of
gravity, let us consider a second case. In this case, someone really, really,
really does not want to satisfy her powerful desire to take drugs. Nonethe-
less, at some point, she goes off in search of a dealer, purchases the desired
substance, and very deliberately injects it into her veins. What has
happened? The obvious hypothesis is that she gave up, or gave in: the
pain and discomfort of resisting the pull of her desire was simply too great
relative to the benefit. So it seemed to her, at any rate; and so she decided
(however unselfconsciously) to stop resisting.
But isn’t it possible that she never wavered in her commitment to resist
the pull of the desire? No, it is not. If never wavering in her commitment
to resisting were compatible with the fact that she was moved to satisfy
the desire, then it would have to be possible for the desire, like gravity,
to so tax her body that her limbs would no longer be capable of obeying
her commands.8 But if this were to occur, then the resulting bodily
movements would be—like the relaxation of her fingers in the second
version of the previous story—involuntary. Of course, the desire to take
the drugs would itself play a key role in causing the bodily movements.
But though, unlike gravity, a desire is a goal-directed state, according to
the conditions we are here imagining, it is the force of the desire that
explains why she does not do what she takes herself to have sufficient

8
The image of “the will as a muscle” is misleading, at best. “Will power” is not
analogous to the power we manifest when we hold on for dear life; it is not the power of
“the will.” Rather, it is the very sort of power we manifest when we hang on for dear life.
The difference simply has to do with what is being resisted, and—as I will soon have
occasion to stress—with what form it is possible for defeat to take.
 The Possibility of Action 21

reason to do. So, according to the stipulated conditions, her desire

explains her behavior because it plays the same role that gravity plays in
the earlier story—the role of a force the agent is trying to resist, a force
that is external to her will, a force that is opposed to her agency. If, then,
we are to imagine that she loses whatever physical power is necessary to
resist this force, without thereby ceasing to be committed to resisting it,
we must conclude that the resulting behavior is—like the behavior
directly caused by muscle fatigue—involuntary.
It seems, then, that an addict cannot intentionally take drugs without
giving up her effort to resist her desire to take them. More generally, our
desires cannot directly overcome our efforts of resistance in the manner of
an irresistible physical force; we cannot relate to our desires as if they
were irresistible physical forces.9 In order to play this role, desires would
have to overwhelm our agency, even while—in their capacity as desires—
they moved us to pursue a goal. To put the same point the other way
around: in order for desires to play the role of gravity, they would have
to be, at one and the same time, the sort of causes that prevent our
behavior from being involuntary and the sort of causes that overpower
our agency.10
This conclusion represents a challenge to any conception of agency
according to which doing something for a reason is compatible with
being alienated from one’s behavior in the way I have here been discussing.
This means that it is a challenge to the standard conception of agency.
Again, the point is not that satisfying the conditions spelled out in the
standard conception never suffices for acting. (For all I am claiming, a
squirrel acts as long as it satisfies these conditions.) But nor is the point
simply that there is a more “robust” form of action which requires the
satisfaction of additional conditions. The point is that agents who are
capable of being alienated from their own desires must satisfy additional
conditions if they are to qualify as acting; and this is because they must
satisfy whatever conditions ensure that they are not alienated from the
psychic conditions that move them.

9
This is a point that Gary Watson stresses in his more recent work. (See Watson
2004) There he rejects the model of unfree action I am criticizing here. For an alternative
account of the distinction between self- and other-determined action that is key to
assessments of moral responsibility, see Buss 2012.
10
Note what this implies about cases in which our demands on ourselves are not met
with obedience: the defiance necessarily takes the form of forcing a reappraisal of our
options—even if (as in the case of akrasia) we are unpleasantly aware that this reappraisal
is a mere rationalization.
22 Sarah Buss

In order to drive home the critique of the standard conception already

implicit in the preceding discussion, let me quote at some length from
Michael Smith’s recent attempt to clarify this conception by explaining
what distinguishes intentional actions from behavior that is caused by
an internal wayward causal chain. According to Smith (who is here
appealing to work by Christopher Peacocke), if a person is to act
intentionally, then “the movement of [his] body mustn’t just be caused
by a suitable desire and belief pair, but . . . it must also be the case that,
abstracting away from the causal role played by factors external to the
agent’s psychology, if the agent had had a range of desires and beliefs
that differed ever-so-slightly in their contents from those he actually has,
he would still have acted so as to realize the contents of these desires,
given these beliefs” (Smith 2012: 399). “What [this] differential sensitivity
requirement guarantees,” Smith explains, “is that when an agent acts
intentionally, he doesn’t just try to realize the desires he actually has, given
the means-end beliefs he actually has, but that he would have tried to realize
his desires, given his means-end beliefs, in a range of nearby possible worlds
in which he had desires and means-end beliefs that differ ever-so-slightly in
their contents. . . . [The] . . . requirement thus suggests . . . that, for someone
to act intentionally at all, she must not just possess but also exercise a certain
capacity to be instrumentally rational.” (Smith 2012: 398). With a nod to
Hempel, Smith continues:
There are three distinct psychological states that play a causal role whenever an
agent acts . . . [C]ausal roles are played . . . by the agent’s desire that the world be a
certain way, and his belief that the thing done is a way of making the world that
way, [and] by . . . the agent’s possession and exercise of the capacity to be instru-
mentally rational [so that] he [can] put his desire and belief together in the way in
which they need to be put together if they are to cause an action. . . . [I]f the agent
doesn’t do what she does because of her possession and exercise of her capacity to
be instrumentally rational, then her desires and beliefs with ever-so-slightly
different contents will not stand in the modally rich pattern of connections
that they must stand in if they are to satisfy the differential sensitivity require-
ment. . . . (Smith 2012: 399)
Smith concludes that the standard story can do justice to the fact that
“those who act ‘intervene’ between their desires and their beliefs and their
bodily movements” (Smith 2012: 399). Accordingly, the story can make
sense of the fact that agents are the cause of their actions. If, however, the
preceding observations are correct, then Smith is mistaken: a goal pursued
in an instrumentally rational way is not necessarily the agent’s own goal.
More carefully, if these two goals cannot come apart, this is because
 The Possibility of Action 23

whenever someone does something intentionally, she satisfies whatever

additional condition ensures that she is not alienated from her own goal-
directed states.
To reinforce the point that Smith’s account leaves out something
important I want now to offer two little thought experiments:
[Thought Experiment 1] Imagine an artifact, built so that it can achieve
two different goals. Consider, for example, a rather special clock. It can
either tell the time or indicate what the time will be fifteen minutes later. It
is, moreover, so exquisitely (and diabolically!) constructed that it can
substitute the one goal for the other; and when it does so, the mechanism
relevant to satisfying the alternative goal will be engaged. (In other words,
nothing about the causal chain initiated by the one goal-seeking mechan-
ism prevents the alternative causal mechanism from being engaged when
the clock’s goal changes. In other words, our clock is “differentially sensi-
tive” to the change in its goals.) Having (let us imagine) stipulated that the
clock’s goal-seeking behavior is not attributable to wayward causal chains,
we can now ask: Have we imagined an agent?
Perhaps we are inclined to answer “no” because our clock lacks a “point
of view.” So, let us give this already magnificent clock a point of view.
Now, we could be rather stingy about this, and merely give it an
awareness of the goal “built into” it by the manufacturer, and of what it
is doing—and even what it must do—in order to achieve this goal. Would
this be enough to make it an agent? More importantly for our purposes,
would it be enough to make it the case that when its hands do what they
do “in order to tell the time,” it—the clock—is acting?
Rather than stopping to answer this question, I want to add something
to our clock’s point of view. I want to give it an interest in what it is
doing, a concern about which goals it has reason to pursue. With this
important addition, let us now suppose that it has concluded that it
doesn’t much like the goal imposed on it by the manufacturer. Perhaps
it can see that there is quite a lot to be said for telling and foretelling the
time. But, at least now and then, it believes it would be nice to use its
hands for other purposes. It would be nice, for example, if its hands would
enact a little choreography it has thought up during the many boring
hours it has spent diligently marking time. Indeed, it believes that the
reasons in favor of behaving this way now and then outweigh whatever
reasons it has to devote itself wholeheartedly to the cause of accurately
representing (or predicting) the time. Yet, let us also imagine, it cannot
translate this belief into reality; it lacks the power to put its all-things-
considered judgment into action. Should we insist that it nonetheless has
24 Sarah Buss

the power to act? Or should we concede, instead, that when its hands
move in order to indicate that the time is now five past midnight, and
when this happens despite its conviction that it has overriding reason to
do otherwise, these movements no more qualify as its actions than do the
hand movements of an ordinary clock?
Of course, in the case as described, the clock’s opinions about “its”
behavior are so causally isolated from this behavior as to disqualify the
behavior from being its own. But this is just the point. Nothing of
significance would change, moreover, if many of its opinions did have an
effect on its behavior. As long as its judgment regarding the all-things-
considered desirability of the two-handed choreography was inefficacious,
the regular movements of (what we can now safely call) its hands would not
be its doings. They would not be its actions.
This situation resembles a case in which a person’s body drags her
around, or does other things, “against her will.” Apparently, something
like this does sometimes happen. When a person suffers from anarchic
or alien hand syndrome, for example, one of her hands may undo the
buttons of her shirt, without her having any intention of doing this.11
So as to contrast this behavior with a mere spasm, we could refer to it as
an “action.” Since, however, the goal being pursued is not a goal that
anyone “has in mind,” it is not anyone’s intentional action. And in any
case—and this is what matters to me here—it is not an action
attributable to the person herself. A person is passive in relation to
what her anarchic fingers are doing to the buttons of her shirt—no
less passive than she would be if these fingers were on somebody else’s
hand.12
[Thought Experiment 2] Consider a second goal-seeker. Call him
A. A has no capacity whatsoever to reason. He has no capacity to ask
himself whether anything is worth doing. Nonetheless, he has dispositions

11
People suffering from anarchic or alien hand syndrome “lose the ‘sense of
agency’ associated with the purposeful movement of the limb while retaining a
sense of ‘ownership’ of the limb. They feel that they have no control over the
movements of the ‘alien’ hand, but that, instead, the hand has the capability of
acting . . . independent of their voluntary control. The hand effectively has a ‘will of
its own.’ ” “The alien hand is directed toward goals of which the patient is not
consciously aware.” (“Alien Hand Syndrome,” Wikipedia <http://en.wikipedia.org/
wiki/Alien_hand_syndrome>)
12
“Sufferers of alien hand will often personify the rogue limb, for example believing
it to be ‘possessed’ by some intelligent or alien spirit or an entity that they may name or
identify.” (“Alien Hand Syndrome,” Wikipedia <http://en.wikipedia.org/wiki/Alien_
hand_syndrome>)
 The Possibility of Action 25

to pursue various goals. And he also has the disposition to be guided by B’s
theoretical reasoning about what it would take in order to satisfy these
goals. When B is aware of a goal A has, she engages her reason in the task of
understanding what it would take to achieve this goal. And when she
reaches a conclusion on this point, A—who also has a special capacity to
read her mind!—adjusts his goal-pursuing accordingly.
By stipulation, this story contains two distinct individuals: one (A)
disposed to pursue goals in an instrumentally rational way, despite having
no opinion whatsoever about the desirability of doing so; the other (B)
disposed to employ reason to determine which events are likely to give rise
to which other events. Is A an agent? In contemplating this question, we can
consider some others: Would it make a difference if B deliberately “fed”
A the results of her reasoning—because (motivated, perhaps, by the interest
in understanding things) she was curious to see what he would do? Would
it matter if we implanted B inside A—so that they now shared a single
body? Perhaps this would suffice to render B an agent—if we also gave her
the higher-order goal of achieving A’s goals. Note, however, that in this
case, B’s goal-seeking behavior would constitute an action only because
B would endorse this behavior; the fact that it was instrumentally rational
goal-seeking behavior would not suffice.
This point is even more obvious if we stipulate that B strongly repudiates
the goals pursued by the body she now shares with A. Under these
circumstances, who is the agent of the resulting behavior? Should we not
concede that no one is, and that, accordingly, the behavior does not qualify
as an action, despite satisfying Smith’s conditions?
We cannot solve the problem revealed by these examples by simply
introducing more complex causal feedback mechanisms. As long as, despite
these greater complexities, the explanation thereby provided does not rule
out the possibility that the behavior is unresponsive to the contempor-
aneous normative judgments of the rational being whose action it would
otherwise unproblematically be, it will not qualify as an action. I thus
conclude that we must reject the standard conception of agency.
Before turning to consider an alternative conception, I want to warn
against reading too much into this conclusion. I am not claiming that
acting always requires reviewing the considerations pro and con. I am not
claiming that when we do things for reasons, we generally have a vivid,
determinate conception of why what we are doing is really worth doing.
I am sure that neither of these claims is true. Though, as Talbot Brewer
eloquently reminds us, we can, and often do, make a point of pursuing
certain activities with the aim of gaining an ever-greater appreciation of
their value (Brewer 2009), for the most part, our actions involve relatively
unreflective attempts to continue the activities we have already initiated,
26 Sarah Buss

often without ever having given the matter much thought. As Hilary Bok
points out in discussing the Milgram experiments, our capacity to act
without regarding every change in our circumstances as a new occasion for
deliberation and choice makes it possible for us to become accomplices to evil
without endorsing any evil ends: if we don’t pay attention—and, again, we
often don’t—continuing on the path we have set ourselves can lead us to a
place we did not know we were headed (Bok 1996). Less dramatically,
philosophers of action are surely right to stress the extent to which we are
inclined to give ourselves permission to act “from habit”; and David
Velleman and Peter Railton (among others) are right, too, to note that
even instrumental reasoning is often superfluous: our habits of response
include dispositions to adjust our means to our immediate ends without
engaging in any more “reasoning” than occurs when the average backyard
squirrel scampers from branch to branch (Velleman 2009 and Railton 2011).
We can concede each of these points, however, while insisting that even
the most ho-hum instances of agency are incompatible with the sort of self-
alienation I have here been exploring. Insofar as (however unconsciously)
we—we reasoning beings—take a stance on what we are doing (insofar as
we take ourselves to have sufficient reason for what we are doing), we endorse
what we are doing—not in the sense that we think especially well of
ourselves, or are especially pleased with the options we face—but in the
sense that we—the ones who bother forming an opinion about whether
what we are doing makes sense—accept the goals of our behavior as our
own. An adequate account of rational agency will thus be an account of this
self-relation, and of the normative judgment that is inseparable from it.13

AN ALTERNATIVE ACCOUNT: ACTING AS A

WAY OF INTEGRATING CONTINGENT
GOAL-DIRECTED MENTAL STATES
WITH THE CAPACITY TO REASON

I am not alone in drawing this conclusion. Other philosophers have called

attention to the fact that rational beings can intelligibly ask themselves

13
There is a vast philosophical literature devoted to making sense of the fact that
rational agents “identify with” their motives. Some philosophers appeal to higher-order
desires. (For the classic defense of this approach, see Frankfurt 1971 and Frankfurt
1977.) Others appeal to evaluative judgments (for the classic defense of this approach, see
Watson 1975. Others appeal to hierarchies and intentions, or other “plan-like” states.
(See Bratman 2002.) By the end of this paper, I hope it will be clear why my sympathies
lie with Watson’s approach, even though I do not endorse his account.
 The Possibility of Action 27

whether they have sufficient reason to do what they are contingently

inclined, committed, or otherwise disposed to do. Indeed, they have
stressed that we can intelligibly ask whether we have sufficient reason to
do what we would be contingently inclined to do, if we were aware of all the
nonnormative, nonevaluative facts that are “relevant” to the satisfaction of
our noninstrumental desires. Unless, these philosophers insist, our motives
reflect our answer to this question, they lack our endorsement. And if our
motives lack our endorsement, then we are not the cause of their effects;
their effects could not be our actions—even if they qualified as instances of
goal-directed behavior.14
To acknowledge this possibility is to acknowledge that reason is not
simply “a slave of the passions.” This is compatible with acknowledging
that no one can reason without falling under the causal influence of
nonrational motivating states. And it is compatible with acknowledging
that, as a matter of contingent fact, some (most) rational agents believe they
have reason to satisfy their (strongest, highest-order, most coherent) desires.
The point is simply that rational agents can intelligibly ask themselves
whether they have sufficient reason to satisfy any or all of their contempor-
aneous desires. And this means that they can be divided from themselves in
a way that would not be possible if their reason were simply a slave of
whatever psychic forces move them to pursue certain substantive goals.
How do self-reflective rational beings manage to achieve the necessary
integration of their reason and their practical impulses if the former is not a
slave of the latter? In the second half of this paper I want to consider two
answers that have recently been offered by philosophers who take the
problem of self-alienation seriously. According to the first account, we
cannot make sense of rational action without replacing the alleged power
relation between master and slave with an authority relation: a rational
being constitutes herself when and only when her practical impulses obey
the orders of her reason. According to the second account, the self-consti-
tuting relationship between these two parts of the soul is more like a
partnership between two agents with two different, but ultimately compat-
ible, agendas: a rational being constitutes herself when and only when her
nonrational motivating states accommodate themselves to her defining
interest as a reasoner. On both accounts, the integration of the two parts
is accomplished when an intention is formed. On both accounts, there is

14
It is important to distinguish this point from the claim that agents must reflect on
(“foreground”) their desires. The point is that they can reflect on—and call into ques-
tion—what these desires represent as to-be-done. (For a defense of the thesis that desires
are not typically among the things agents take into account, See Pettit and Smith 1990.
28 Sarah Buss

thus an important sense in which until there is an intention, there is no

agent.15
In suggesting that determining one’s action is the necessary means of
constituting oneself, the alternative accounts share an assumption with the
standard accounts they reject: they contrast the goals we have insofar as
we are rational beings (e.g. avoiding incoherence, discovering causal rela-
tions,. . . .) with the substantive goals we have insofar as we are moved to
initiate change. According to this division of labor, the roles played by each
part of the soul do not give them a common interest; the person herself does
not have a single interest insofar as she is thus divided. More carefully, the
two parts do not have a common interest if, as the alternative accounts also
assume, a person’s reason is not limited, in its practical role, to figuring out
how to satisfy her desires. I will try to show that this is precisely why we
cannot accept the alternative accounts: they fail to explain how a soul thus
divided can assemble itself into a whole.
In making my case, I will also be exploring the relationship between (i)
the claim that acting involves unifying one’s practical and rational parts and
(ii) the claim that the conditions necessary for agency provide the criteria
for what we have reason to do. Many philosophers have offered compelling
objections to the second claim. Before I press my own objection, however,
I want to stress how little one needs to assume in order for it to seem as
though there is no plausible alternative. One need merely embrace cogni-
tivism about practical reasons, while conceding both that (i) we can intelli-
gibly wonder whether we have reason to satisfy any of our desires, even
when we understand the likely effects of so doing, and that (ii) there is no
mind-independent substantive criterion for what counts as a reason for
what. Given (i), it seems that we must locate the criterion for practical
reasons in a commitment or goal that does not depend on which (contin-
gent) substantive desires a prospective agent happens to have. But given
(ii), it seems that the criterion must be grounded in one or more of the
prospective agent’s commitments or goals. Is there, then, any alternative to
acknowledging that what someone has reason to do is determined by
whatever goal is inseparable from her identity as a prospective agent? And
isn’t this just the goal of not being a passive bystander to one’s behavior?
Isn’t it the goal of “constituting oneself” as an agent?
In short, from rather plausible premises, it is reasonable to conclude that
the commitment to being responsive to reasons is a commitment to being
governed by formal principles (whatever these may be), compliance with

15
Korsgaard is quite explicit about this: “I am going to argue that in the relevant sense
there is no you prior to your choices and actions, because your identity is in a quite literal
way constituted by your choices and actions (Korsgaard 2009: 19).
 The Possibility of Action 29

which ensures that the ends associated with our practical impulses are truly
our own. According to the first of the two accounts that endorse this
conception of practical reasons, this means that the criteria for what counts
as a reason for action are the necessary conditions of self-determined
behavior. According to the second account, the criteria are the necessary
conditions of nonobservational knowledge of one’s behavior. If, as I will argue,
we must reject these accounts, then we must reconsider the premises that
appear to support them. We must reconsider the assumption that action is
possible if agents endorse the conception of practical reasons these premises
express.

Version 1: Constituting Oneself

as a Self-determining Cause
Consider the story often repeated by philosophers who note that acting
makes special demands on beings with the capacity to wonder what they
have reason to do. According to this story, the capacity to reason is (among
other things) the capacity to question the desirability of being moved by
whatever impulses we happen to have. We rational beings thus need to
discover a reason to do what we find ourselves inclined to do. Unless we
discover such a reason, our point of view as rational beings will be insuffi-
ciently unified with the point of view constituted by our dispositions to
initiate change. Under these circumstances, the effects of our desires will
not be attributable to us—the ones who are searching for reasons. Their
motivating force will not be our motivating force.
According to this story, whenever we do things for a reason, we have
managed to transform a mere collection of mental states, in relation to
whose motivating force we rational beings would otherwise be mere passive
bystanders, into a unified rational agent—the sort of rational being we are
when we cause our behavior. How do we manage to do this? According to one
compelling and influential version of the story, rational beings incorporate
their various impulses into a unified whole by ensuring that these impulses
obey the laws that are constitutive of their rationality, where these include laws
that determine which ends they have reason to pursue. They employ their
reason, not only to determine how to satisfy their noninstrumental desires, but
also to determine whether to do so. They put their reason to practical use in
order to “govern” their nonrational dispositions to pursue certain ends. As
Christine Korsgaard puts it, “the reflective structure of self-consciousness
inevitably places us in a relation of authority over ourselves, and . . . we are
as a consequence accountable to ourselves. . . . I act under my own authority
30 Sarah Buss

as lawgiver, and I am accountable to myself if I do not. So my reasons—and

indeed, practical reasons in general—are grounded in the authority the
human mind necessarily has over itself ” (Korsgaard 2007: 10–11).
But why are the laws of our reason (whatever these are) authoritatively
binding on any motives that are independent of these laws, and independent
of the commitment to discovering reasons for action that is essential to “the
reflective structure of self-consciousness”? How could we possibly regard
these laws as authoritatively binding insofar as we are constituted by inclin-
ations that are independent of our need to discover reasons for acting? I want
to argue that this is not possible unless the authority of these directives is
grounded in considerations that are relevant from the point of view of the
motives themselves. Otherwise, there is no intrapersonal authority relation
after all, and so we cannot act. Unless our normative verdicts have an
authority for us that is not simply grounded in our identification with our
reason, we are in no position to expect any desires that are independent of
our reason to obey the demands that are grounded in the verdicts we reach
when we reason about the desirability of satisfying these desires; and so, we
(the ones who make these demands on ourselves) are in no position to ensure
that the motivating force of our desires can be attributed to us.
In pressing this charge, let me begin by noting that there do seem to be
occasions on which we exercise authority over ourselves: we find ourselves
tempted to do something we think we have overriding reason not to do, and
we order ourselves not to do it. (“Sarah, do not reach your hand toward
that cookie!” “Sarah, keep working on that paper for the conference
on agency and responsibility!”) Surely, however, this is not the way
things usually work—at least not for most of us. Exercising authority over
oneself is something one does when one’s rational agency is under threat. It
is not what occurs in the paradigm instances of rational agency. When
things are going well, our reasons for action do not include the fact that we
have ordered ourselves to act this way, nor the fact that we issue this order
in our capacity as rational beings.
This is what Bernard Williams gets right when he says that giving oneself
permission to save one’s wife would generally involve “one thought too
many” (Williams 1981: 18). Most of us rarely need to defer to an exercise
of authority in order to act. And because we have no need to defer to
authority, we have no need to assert authority either.
In this respect, we are no different from the purely rational beings to
whom Kant attributes a “divine,” or “holy,” will. (Kant 1956: 81) Because
such beings are wholly identified with the verdicts of their own reason
(because their point of view just is the point of view constituted by these
verdicts), they are not sufficiently divided from themselves to impose
demands on themselves; they are not subject to any imperatives; no
 The Possibility of Action 31

obligations can figure among their reasons for action. So, too, even if we
less-than-divine rational beings can be divided from ourselves in a way that
enables us to give and take the same orders, most of us are usually
sufficiently identified with the verdicts of our own reason—i.e., these
verdicts are usually sufficiently motivating—to make such assertions of
authority superfluous. For us, too, the thought that our reason permits or
requires a certain course of action is usually one thought too many.
This is, I think, an important objection to the conception of rational
agency as the product of an intrapersonal authority relation. I want, how-
ever, to focus on a deeper problem with what we can call the “authority-
based” conception. To put it bluntly, on this conception of rational agency,
insofar as we are desiring beings, we could not care less about whether the
motivating force of these desires can be attributed to an agent. And so we
could not care less about the fact that in defying the demands of our reason,
we would be sabotaging our agency. And so, we have no motive to obey
these demands. And so, we cannot obey them.
Of course, in describing the possibility of “sabotaging our agency,”
I have been forced to speak about what we would do, and so I have been
forced to imagine a situation in which we are sufficiently unified to perform
various agent-sabotaging acts. But this merely brings home the point that it
is not possible to make sense of action if we posit the division of labor that is
essential to the authority-based account. If, as the authority-based concep-
tion assumes, the laws of our reason and the ends of our desires stand in no
internal relation to each other, then there is no possible way for us to
exercise authority over our desiring selves.
The argument for this claim went by very quickly. I want to rehearse it
again more slowly, beginning with an observation by the person who has
done more than any other philosopher working today to develop and
defend the authority-based account. “If I am to govern myself,” Korsgaard
tells us, “there must be two parts of me, one that is my governing self, my
will, and one that must be governed and is capable of resisting my will”
(Korsgaard 2008: 60). According to this division of labor, the governing
self is the agent insofar as she is identified with her reason, and more
specifically, with whatever formal principles of rationality impose
constraints on which desires she has reason to satisfy. Accordingly, the
governed self—the self on the receiving end of reason’s “laws”—must be the
agent insofar as she is identified with motives that are not grounded in this
commitment.16 And here is where things get tricky. As we have learned

16
Note that these motives could nonetheless be responsive to reasons. Indeed, this
could well be true of even our deepest instinctual desires.
32 Sarah Buss

from the literature on authority, in order for one person to exercise

authority over another, the latter must be capable of recognizing the
demands of the former as distinct (preemptive) reasons for action. This
means that in order for my rational self to exercise authority over my
nonrational self, the latter must be capable of regarding the demands of
the former as (preemptive) reasons. And this means that it must be possible
for me (= that aspect of my identity which is constituted by a concatenation
of motives not grounded in my reason) to regard the directives given me
by myself (= that aspect of my identity which is constituted by the verdicts
and demands of my reason) as distinct reasons to do certain things and
not others—distinct reasons to behave as directed. And this means that
the nonrational motives themselves—the very motives whose independence
from my reason prompts me to give orders to myself—must be such
as to enable me (= the one constituted by these motives) to appreciate
the normative force of my demands (= the demands I make as the
representative of reason).
How is this possible? According to Korsgaard, “the acting self concedes
to the thinking self its right to government. And the thinking self, in turn,
tries to govern as well as it can. [This is] a relation which we have to
ourselves. And it is a relation not of mere power but rather of authority.”
(Korsgaard 1996: 104) But what is it about the thinking self that gives it the
“right” to expect obedience from the acting self? What justifies its claim to
authority? The point cannot simply be (the relatively trivial point) that
insofar as I am identified with my thinking self (and its commitments),
I cannot call its authority into question. We need to know why that
aspect of me which is not identified with my thinking self is nonetheless
bound to treat the demands of this self as reasons to respond as demanded.
If, qua rational being, I demand compliance/obedience from myself
qua nonrational being, if I base this demand on nothing more than my
assumption (which I cannot fail to make insofar as I employ my reason)
that the laws of my reason are to-be-obeyed, and if I hear this demand from
a point of view from which I am entirely indifferent to the verdicts of
my reason and its guiding principles, then my attempt to assert authority
over myself fails. But this means that insofar as “exercising authority over
myself” is not merely a manner of speaking—insofar as it is not merely
what I do whenever I reason about which of my desires it really makes most
sense for me to satisfy—it is not possible for me to exercise authority over
myself under the conditions stipulated by the account here under consider-
ation. Asserting authority over myself under these conditions would be a
fraud of the very sort Williams thinks he sees in every attempt to get
someone to respond to “external reasons” to do something (Williams
1981: 101–13). It would be a fraud precisely because on the view we are
 The Possibility of Action 33

here considering, the alleged authority of reason cannot be justified without

appealing to considerations (e.g. what is (allegedly) necessary in order to
act) whose normative force is invisible from the point of view of the mental
states over which the authority is asserted.
Of course, as my reference to our “manner of speaking” acknowledges,
there is a perfectly unproblematic sense in which I cannot fail to regard the
laws of my own reason as authoritative: I cannot intelligibly challenge the
authority of the laws of my own reason even as I am reasoning about what
to do. But the reference here to “the laws of my own reason” is not a
reference to some distinct aspect of myself that asserts authority over me as
I reason. When I reason, the “laws” of my thought are not among the things
I take into account—nor is the fact that they govern my thinking, nor the
fact that I impose them on myself. Though there is an obvious sense in
which I treat these laws as authoritative, and an obvious sense in which, in
so doing, I am imposing certain demands on myself, the fact that I am
making these demands is not an additional—trumping—reason for me to
act in the ways spelled out in the demands. And so, my relation to myself is
not a relation of practical authority; it is not a relation in which I defer to,
or obey, myself.
Insofar as my desires reflect my normative judgments, I have no need to
unify myself in order for my goal-directed behavior to qualify as my action.
My acting self and my thinking self are one and the same.17 If, on the other
hand, my desiring self is completely distinct from my thinking self, then my
thinking self has no choice but to regard the relevant desire as a feature of
its circumstances—one more fact to be “taken into account.” Accordingly,
under these circumstances, too, my thinking self must be able to generate
its own motives; and so, it must already be my acting self. In order, then,
for my actions to result from the integration of two parts of my soul—in
order for acting to involve “self-constitution”—the two distinct parts
cannot be entirely independent. As a subject of desires which do not
themselves reflect my commitment to being governed by my reason,
I must nonetheless be capable of recognizing the verdicts of my reasoning
as authoritatively binding.

17
Can my thinking self be alienated from itself? Not in the sense that matters here.
No sooner do I call my contemporaneous normative judgment into question, than it
ceases to be my judgment. No sooner do I disavow a given assessment of the consider-
ations for and against a given action than I endorse a different assessment—if only one
that can be expressed in terms of the disavowal. This having been said, it does seem
possible to be alienated from one’s normative judgments in a different way: one can call
their legitimacy into question, not on normative, but on metaphysical, grounds. This is
the sort of self-alienation familiar from discussions in metaethics.
34 Sarah Buss

This conclusion has an important implication: “practical reasons [cannot

be] grounded in the authority the mind has over itself.” For consider. How
could I (qua subject of reason-independent desires) come to regard the verdicts
of my reasoning as authoritatively binding unless I (qua subject of reason-
independent desires) could take there to be some reason to regard myself (qua
reasoner) as having authority over myself (qua subject of desires)? And how
could I (qua subject of reason-independent desires) have a reason to regard
myself (qua reasoner) as having this authority unless I had practical reasons for
doing so that are not grounded in the authority my mind has over itself? These
questions lead me to conclude that Korsgaard’s account of agency does not
“answer” “the realist objection—that we need to explain why we must obey [a]
legislator.” It does not identify “a legislator whose authority is beyond question
and does not need to be established.” It does not show us how to make sense of
“the authority of [our] own mind and will” (Korsgaard 1996: 104).
There are two respects in which the authority of “our own mind and
will” is, for us, “beyond question.” First, we are in no position to question
this authority insofar as we are identified with our own mind and will. But
under these circumstances, we lack the distance from ourselves to stand in
an authority relation to ourselves. Second, even when we do exercise
authority over ourselves, this is possible only because, however strongly
we may be inclined to defy the orders of our own mind and will, it is not
possible for us—the ones thus ordered—to reject the order as an abuse of
authority, a brute assertion of power. This means that, even in our nonra-
tional capacity, we must acknowledge the force of whatever reasons there
are to regard the order as authoritative.
How are we to make sense of this double-minded stance? How is it
possible for a self to be just divided enough and just unified enough to
exercise authority over itself?18 Joseph Raz’s work on interpersonal
authority (Raz 1985) suggests one possible answer: even insofar as we
apprehend our options from a point of view that is not grounded in our
reason, we acknowledge the epistemic authority of the judgments that are so

18
It is interesting to compare my answer to this question—and, more generally, my
critique of Korsgaard’s account of action as self-constitution—with Tamar Schapiro’s
attempt to make sense of the Kantian thesis that when someone does something
intentionally, she “incorporates” her desire (“inclination”) into her action-guiding
principle. Schapiro notes that insofar as I have an inclination that is not grounded in
my reason, “part of me sees the world through [the] eyes of [this inclination] and
responds to the world [accordingly], and part of me is aware of itself as not being the
source of this way of seeing and responding” (Schapiro 2011: 164). So far, so good. My
point, however, is that if an inclination really is internal to my point of view, then even
though “it has no capacity to demand reason or justification for its way of seeing and
responding to the world” (156), I cannot be indifferent to such demands insofar as I am
the subject of this desire.
 The Possibility of Action 35

grounded.19 I am not sure whether this is the only way things could work.
I want, however, to explain what I have in mind in suggesting that
satisfying this condition would suffice to make an intrapersonal authority
relation possible. In so doing, I hope to reinforce the two related lessons
I have just drawn from my discussion of Korsgaard: (1) a genuine
intrapersonal authority relation presupposes that there are facts about
what we have reason to do that are not grounded in our commitment to
acting; and (2) a genuine intrapersonal authority relation requires that in
our capacity as the one who is governed, we are not so alienated from our
governing self that we cannot appreciate any reason for regarding the
demands of the governing self as authoritatively binding.
Suppose, then, that I have reached the conclusion that I have overriding
reason not to eat that second piece of pie. And suppose that I would not
have bothered to employ my reason on this occasion if I had not assumed
that its verdicts would have a greater epistemic authority with respect to
what is really to-be-done than any desire prompted by the close proximity
of the pie. Alas, I might nonetheless persist in desiring to eat that second
piece of pie, and this desire might be stronger than any others. If to be in
this state is to represent the pie as to-be-eaten, then when I am in this state,
I am opposed to being guided by the verdict of my reason(ing). Insofar,
however, as I represent the pie in this way, I cannot dismiss as irrelevant the
belief that it is not to-be-eaten-by-me. And so, I cannot be indifferent to
which of these representations has greater epistemic authority. If, moreover,
I believe that the verdicts of my reason have greater epistemic authority,
then even as I am disposed to act contrary to these verdicts, I am also
disposed to acknowledge the authority of the demand to act otherwise. In
short, precisely because, qua pie-desirer, I am not wholly identified with my
reason, I am in a position to regard the epistemic authority of my own
normative judgments as a distinct reason to grant practical authority to
whatever demands reflect these judgments.
Under the stipulated circumstances, my desire to eat the pie puts me at
odds with myself; it moves me to defy the verdicts of my own reason. To
this extent, I regard the directives that are grounded in these verdicts as
external constraints—even as I am the one who is imposing them. At the
same time, however, since I also regard the directives as coming from a
source that has epistemic authority with respect to the very point at issue
between me and myself (since—rightly or wrongly—even as I desire the
pie, I regard the contemporaneous verdicts of my reason as more likely to

19
I agree with those who reject Raz’s appeal to epistemic authority to account for
relations of practical authority between agents. My suggestion here is that intrapersonal
authority relations are importantly different.
36 Sarah Buss

track the reasons I really have than are any pie-related desires that are not
grounded in my reason20), I assume that I (the person I am insofar as
I employ my reason to figure out what to do) am justified in asserting
practical authority over myself (the person I am insofar as I am not identified
with my reason). And so I regard the directives as authoritatively binding.
I regard these orders as authoritatively binding because I regard them as
reflecting my own superior epistemic position with respect to the very issue
my desire represents as being at stake when it represents the piece of pie as
really, really, really to-be-eaten-by-me.

Version 2: Constituting Oneself as

Someone Whose Behavior One does
not Find Perplexing
The preceding sketch suggests one way in which a rational being’s relation
to her own motives could be just external enough to allow for a relation of
authority, but not so external as to reduce all assertions of authority over
herself to futile attempts to bully herself. I leave for another day the task of
considering how this sketch might be filled in, and what other conceptions
of the relation of self to self might also strike the right balance. For my
purposes, the most important points to stress are these. If someone is as
divided from herself as the self-reflective pre-agent appears to be on
Korsgaard’s account, then there is nothing that this deformed being can
possibly do to “pull herself together in order to act.” Under these circum-
stances, if her “thinking self” is not already capable of generating motives
for action—perhaps in response to certain (nonrational) motivating states
that were not so generated, and so cannot be attributed to her (the
thinker)—then it is not possible for her to act. But if for less-than-holy
wills, the possibility of rational agency is inseparable from the possibility of
intrapersonal authority, then agents with less-than-holy wills (agents
whose desires are not necessarily in harmony with the verdicts of their
reason) cannot act unless they assume that there are reasons for action
whose normative status is independent of both the commitments consti-
tutive of their rationality and whatever other commitments they happen
to have.

20
It is important to stress that I could be mistaken about this: sometimes one’s reason
leads one astray. The important point here is that insofar as I am the subject of a certain
normative judgment, I take myself to have reason to rely on it. If I did not believe that the
judgment was reliable, then I would see things differently.
 The Possibility of Action 37

I suspect that any account of rational agency is bound to come up short if

it fails to do justice to this last requirement. I want now to lend support to
this hunch by considering a second account of the conditions under which
rational agents avoid self-alienation. This is the account developed with
great ingenuity by David Velleman. According to Velleman, “reflection on
the phenomena of action reveals that being the subject of causally related
attitudes and movements does not amount to participation of the sort
appropriate to an agent” (Velleman 1992: 463). More particularly, if
desires and means-end beliefs “cause an intention, and intention causes
bodily movements,” and if there is nothing more to the causal history of
our behavior than this, then “nobody—that is, no person—does
anything; . . . the person serves merely as the arena for [the psychological
and physiological events that take place inside]: he takes no active part”
(Velleman 1992: 461).
Like Korsgaard, Velleman assumes that in order for rational beings to
take an “active part” in the production of their actions, they must deter-
mine which of their desires they have reason to satisfy, and the “drive” to be
guided by this normative verdict must play a decisive causal role in their
behavior. In order to act, rational beings must do what they do because they
are moved to do it in their capacity as the ones who demand a reason for
satisfying their desires; they must be motivated by whatever motive “drives”
their “critical reflection on, and endorsement or rejection of, the potential
determinants of [their] behavior” (Velleman 1992: 477).
On Velleman’s account, too, there is a tight connection between (i) the
need to locate a criterion of practical reasons in the necessary conditions of
agency and (ii) a picture of agency that divides the pre-agent in two. But,
according to Velleman, for agents who can question the desirability of
satisfying their desires, acting is not the result of giving oneself a reason to
satisfy a given desire by asserting practical authority over one’s practical
impulses. Rather, it involves using one’s reason to make predictions that
elicit the cooperation of that aspect of oneself which is constituted by one’s
nonrational, practical impulses.
Since on this account the reasoning that yields the predictions is
grounded in the desire to understand things in a relatively straightforward
way, the division within the self that one overcomes whenever one cooper-
ates with oneself in this way is not the division between one’s reason and
one’s desires. It is the division between a desire one cannot fail to have
insofar as one is rational and one’s other (contingent) desires. According to
Velleman, insofar as a person is identified with her reason, she has one
guiding aim: to understand things. And insofar as she is disposed to behave
in certain ways in order to bring about certain states of affairs, she is not
responsive to reasons.
38 Sarah Buss

Before we consider how, according to Velleman, these two “parts”

interact to produce action, it is important to see just how far apart their
job descriptions ensure that they are. To this end, imagine for a moment
that you occupy a point of view from which nothing you see can strike
you as the way it ought or ought not to be, or as in any way good or bad.
Imagine that you are curious to discover the way things are, but in no way
“invested” in what you discover. Imagine, in other words, that it makes
no difference to you what you come to understand. You are indifferent,
for example, to whether you understand that human beings lack the
intelligence to slow global warming or whether you understand that
slowing global warming is well within their intellectual and emotional
capacities. Now, imagine that your response to the world is very differ-
ent. You find yourself drawn to make various propositions true, without,
however, having any capacity to form any opinion about these behavioral
dispositions. Finally, imagine that when you occupy the first point of
view, one of the aspects of the world that comes into view is the dispos-
itions that constitute the second point of view, as well as their actual
and likely effects. Is there anything about these two ways of relating to
reality that could enable them to relate to each other so as to generate
behavior that qualifies as your acting for a reason? (Having explored this
question, I will consider the possibility that no cooperative relation is
necessary—since, in effect, the desire to understand our own behavior
can generate new (rational, “motivated”) desires to perform certain
actions and not others. On this alternative account, action does not
involve self-integration.)
By stipulation, none of the contingent, nonrational behavioral dispos-
itions our self-reflective rational being discovers in herself could care less
about whether it plays a role in producing an action. None of these
dispositions could care less about whether the behavior to which it gives
rise is intelligible to anyone. Accordingly, insofar as someone is identified
with these dispositions, she does not care about whether her behavior is
intelligible. This seems to imply that, on the account we are here consider-
ing, if we are ever to overcome the divide between us and ourselves, both
the aim of overcoming this divide and the capacity to overcome it must
somehow be built into our interest in understanding things. But how could
this be? Why would (how could) a rational being, construed as one for
whom all considerations have normative significance only insofar as they
are relevant to understanding things care about whether the world to be
understood is a world in which at least some practical impulses are respon-
sive to this theoretical aim?
Of course, since we do, in fact, act, we do, in fact, understand what we
are doing in a way that is different from the way we understand the actions
 The Possibility of Action 39

of others.21 So there must be something that accounts for this

understanding, and its distinctive features. But this puts the cart before
the horse. We need an account of how, on the view we are here considering,
we come to act.
Again, we might wonder how the desire for understanding could suffice to
pull the trick off. After all, we want to understand other people’s behavior too;
and to achieve this aim, we certainly do not need to understand their
behavior as we would understand it if it were our action. So, again, why
should we think that we have this aim insofar as we aim to understand our
own behavior? Why does the desire to understand things give us any more
reason to object to being surprised by our own behavior than we have reason
to object to being surprised by the behavior of others? Wouldn’t such
behavior simply provide us with different material to understand?
Velleman’s answer to this question is, first, that insofar as an event is at
odds with one’s expectations, one does not really understand why it
occurred, and second, that when the motivations are one’s own, one has
a means of avoiding such surprises, which one does not have when one’s
predictions concern the behavior of others. As he puts it in the Precis of The
Possibility of Practical Reason,
a normal person is aware . . . of being identical with an especially salient member of
the objective order—identical, that is, with the creature walking in his shoes
sleeping in his bed, eating his meals, and so on. That creature is certainly of great
interest to a person, and its doings consequently become the object of the person’s
intellectual motives. But the person’s awareness of being identical with that creature
opens up an obvious shortcut to the cognitive goal. The subject can know what that
creature is doing simply by doing what he conceives of the creature as doing. . . .
Practical knowledge thus supplants theoretical knowledge, as a more secure route to
the same cognitive goal. (Velleman 2005: 229)
But what grounds this identity, and the awareness of it? It would seem that,
on Velleman’s account, the only ground I could have for making a special
proprietary claim to the behavior of a particular “member of the objective
order” is that I have a special mode of access to the psychic forces that cause
this behavior, namely, I am the subject of the qualitative experiences to
which these forces directly give rise. Since my position in relation to these
forces is, on this account, an exclusively receptive one, it is hard to see why
I would think that their behavioral effects are my movements. And so, it is
hard to see why I would want them to be intelligible to me in the way they
would be if they were my own.

21
For the classic discussion of our special epistemic relation to our own actions, See
Anscombe 1957. See also Moran 2001.
40 Sarah Buss

But perhaps this line of questioning misses the point. After all, Velleman’s
comment suggests that what matters is not that we will better understand our
behavior if we understand it in the way that we do when we understand it as
our action. What matters is that we are in a privileged position to ensure that
our behavior will not surprise us. According to Velleman, our beliefs about
what we will do are based on what we know about our behavioral dispos-
itions. Given our desire to understand our behavior, these beliefs will
generally be responsive to what we are motivated to do; and for the same
reason, they will generally be sufficiently motivating for the resulting behav-
ior to make them true. When this happens, it is we who are the behavior’s
determining cause. We are the cause of our behavior because the desire that
causes it would not do so if we thought that this made no sense in light of
everything else we believe and desire. Because the motivating force of the
desire is constrained by our desire for self-understanding, the aim it moves us
to seek is our aim too.
But something still seems missing from this account. In particular: why
would the aspect of myself that is indifferent to reasons be moved to
“accommodate” the “naturally inquisitive” aspect of myself “by enacting
ideas of what it would be intelligible for [me] to do”? (Velleman 2009: 18)
The problem identified here is evident when we take a careful look at
Velleman’s most recent attempt to explain his conception of action as the
product of a cooperative relationship between a rational being (qua
rational) and certain reason-independent behavioral dispositions. In
particular, his use of pronouns reveals that, on his account, it is not
possible for this relationship to qualify as the reflexive self-relation that
only genuine unity/identity will support. Here is a representative passage:
The cognizable object [i.e. the person of whom someone ( = the inquirer) is aware in
being aware of himself and his dispositions] is disposed to instantiate what the
inquirer already understands: it is so disposed because it consists in the inquirer
himself, with his drive toward self-understanding. The inquirer learns that he can make
sense of himself by [his] making sense to himself — that is, by [his] doing what makes
sense to him. (Velleman 2009: 17; the underlined words reflect Velleman’s emphasis)
In this passage, I have used bold type for those pronouns that refer to the
aspect of the person constituted by the contingent behavioral dispositions
that are the objects of the person’s reflection. And I have used italics for
those pronouns that refer to the person insofar as he is the subject of this
reflection, motivated by the desire to understand his behavior. Velleman
asks us to think of the relationship between these two aspects of the pre-
agent as analogous to the relationship between an improvising actor and the
character whose responses to the circumstances the actor is improvising.
Because, he argues, the character and the improviser are one and the same
 The Possibility of Action 41

person, the character is capable of accommodating the improviser; and for

the same reason, he is motivated to accommodate him. Against this
suggestion, I have been arguing that the division of labor between the
two does not support the identity; and that, accordingly, it provides no
basis for the claim that the improviser’s aims are shared by the character.
Velleman is keen to stress that in our capacity as the ones who improvise the
responses it would make sense for us to make if we really did have a certain
character, we do not merely relate to this aspect of ourselves as the object of our
reflection. We also try to see how things look from this point of view:
He gathers materials for understanding possible courses of action by reacting to
external circumstances in consciously valenced thoughts—thoughts that are con-
sciously desirous or worried or fearful or joyous, as occasioned by the circumstances.
Thoughts that are consciously desirous or fearful may illuminate intelligible lines of
conduct better. (Velleman 2009: 22)
But since the character he is trying to understand does not attribute
normative significance to anything, “reacting to external circumstances in
[the character’s] consciously valenced thoughts” does not involve seeing
facts as having any normative significance. So, these reactions are necessar-
ily external to the commitment that constitutes his point of view—namely,
the commitment to responding to his circumstances in a way that makes
sense. So, these reactions are not themselves his take on what makes sense.
So, they are, for him, mere facts to be considered in determining how he has
reason to react. To put the same point the other way around: a character
and an improviser of this character’s reactions are one and the same person,
insofar, and only insofar, as they share reason between them. If they are
two, then the best that the improviser can do is to act as if he were the
character. In short, under the stipulated circumstances, it is not true that
“he [is] fully merged with his character” (Velleman 2009: 16).
If action requires the cooperation of two parts of a pre-agent, one of
which has no motive for cooperating, then action is not possible. Perhaps,
however, we can abandon the model of intrapersonal cooperation without
rejecting the basic elements of Velleman’s account. (Many of Velleman’s
own comments suggest as much.) According to this suggestion, our desire
for self-understanding generates intentions (self-fulfilling beliefs about
what we will do) by generating new desires in response to our beliefs
about what—independent of our desire for self-understanding—we are
disposed to do. Given, for example, that I believe that I am a kind person
who desires to help others in need, my desire for self-understanding will
give rise to a desire to help this particular person who is struggling with her
groceries; and so (assuming that I am aware of no stronger conflicting
desires), I will come to believe that I will help this particular person who is
42 Sarah Buss

struggling with her groceries; and so, given my desire to find my behavior
intelligible, I will be moved to help her. (Alternatively, the desire for self-
understanding will directly generate the belief that I will help this person,
and this will then generate the desire to help her.)
On this account, action does not require a pre-agent’s nonrational self to
cooperate with her rational self. Accordingly, it does not require the pre-
agent to overcome her alienation from the practical impulses she finds
herself with. No such self-integration is necessary because, on this account,
the aspect of the self from which the subject of self-reflection is alienated is
relegated to the status of a very special feature of her circumstances: it is the
feature of her circumstances that shares with her the power to directly affect
the movements of her body, the feature that shares a body with her. On this
account, integrating oneself with one’s reason-independent practical
impulses is not essential to one’s identity as an agent. So, the self-alienation
that persists is not an impediment to one’s agency (provided, of course, that
one’s beliefs about one’s reason-independent impulses are sufficiently on
target to generate reason-dependent motivations that can exert a decisive
influence on the movements of one’s body). Nonetheless, the fact that, on
this account, every agent is necessarily alienated from herself in this way
seems to be a defect. (Is it really essential to rational agency that a rational
being’s understanding of why her action makes sense is completely inde-
pendent of the way she represents this action to herself insofar as she is the
subject of desires that are not grounded in her reason?)
More importantly, it seems to me that this account is plausible only
insofar as it attributes to rational agents a substantive assumption about
what is really worth doing, independent of their commitment to acting, or
any other commitments they may have. Insofar as my desire to help the
person struggling with her groceries is a response to my desire to under-
stand myself, it is not simply one more disposition I find myself with when
I reflect on the reason-independent constituents of my mind; it is not just
one more mental state for me to take into account in deciding how it would
be intelligible for me to behave. Nor is it simply the disposition to regard
intelligibility as the key to what it is for a fact to be a reason for action. Nor
is it simply the effect of such a disposition. As I have described the case, the
desire to help the person struggling with her groceries is grounded in my
belief that I desire to help people, together with my disposition to regard
the intelligibility of my helping her as a reason to help her—a distinct,
substantive consideration in favor of helping. As far as I can tell, unless my
motivation can be construed in this way, my identity as a rational being
cannot be unified with my identity as the subject of practical impulses in
the way these two identities must be unified if the motivating force of my
impulses is to be attributable to me—the one who is committed to acting
 The Possibility of Action 43

for reasons. My behavior is my action when I am moved by my desire to

help the person struggling with her groceries because and only because my
desire to help is “under the guise of the intelligible,” and because it thus
reflects the very commitment that is constitutive of the point of view from
which I reflect on the normative significance of my desires. To put the same
point the other way around, in the case as I have described it, I am the agent
of my behavior precisely because I—the one who demands a reason for
doing what I do—regard the intelligibility of what I am doing “under the
guise of the good.” (Note that this point is analogous to a point I made in
discussing Korsgaard, namely, insofar as a rational being does not stand in
the sort of authority relation to herself that requires her to distance herself
from her own normative verdicts, she is “governed” by these verdicts
because she shares the commitment (to being governed by reason) that
underlies them. Her “identification” with her reason just is her endorse-
ment of the goals she sets herself insofar as she has the more basic goal of
responding to reasons.)

The idea that the necessary conditions of action are the ultimate criteria of
practical reasons is a very intriguing one. I have chosen to discuss two
fascinating developments of this idea not only because they are fascinating
(though this certainly played a part in my choice), but, more importantly,
because they represent proposals regarding what rational agency could be if
we assume both that (i) no rational agent qualifies as acting intentionally
simply in virtue of engaging in instrumentally rational, goal-directed
behavior, and that (ii) our reason is not a capacity for discovering mind-
independent facts regarding which substantive goals we have reason to
pursue. I do not know whether there are any alternative accounts of the
source of normativity that are compatible with these conceptual commit-
ments—any alternatives, that is, to (i) an account according to which practical
reasons are grounded in laws whose practical authority is itself grounded in
nothing but the fact that no one can coherently challenge this authority if she is
committed to being the author of her behavior, or (ii) an account according to
which practical reasons are grounded in the constitutive aim of theoretical
reason, and the more specific aim of not being perplexed about why one is
being moved to satisfy a given desire. I strongly suspect, however, that whatever
other options there may be, they will have the same structural defects. They will
presuppose that our identity as rational beings is distinct from our identity as
beings with noninstrumental practical goals. And so they will lack the resources
necessary to make sense of the possibility of rational agency.
Something has gone seriously wrong if agency itself (“constituting one-
self as an agent”) is conceived as the goal of every being with practical
impulses—not, to be sure, a goal that such pre-agents conceive under the
44 Sarah Buss

description “constituting myself as an agent,” nor, indeed, any goal they

have “in view” (Velleman 1996), but a goal nonetheless. There is an
obvious—and trivial—sense in which we bring ourselves into existence as
agents every time that we act. But, as far as I can tell, our actions are not
events that occur because, and only because, we pursue the particular ends
given us by our desires as a way of (or a means to) satisfying a condition
without which we fail to act. To act is not to do things as a way of (or a
means to) governing or understanding ourselves. Acting is something that
just happens when we do things for reasons.
To be sure, actions are not just any old happenings. To act is to govern
one’s behavior in such a way that one knows what one is up to without
observing oneself. This means, I have argued, that (1) there is more to the
action of rational beings than the instrumentally rational pursuit of goals.
As I have also tried to show, however, (2) no rational being can identify
with her motivating desires if she is so alienated from them that she regards
the conditions necessary for identification as a constraint on what she has
reason to do. If I am right about this, then (3) an adequate account of
agency will have to endorse an alternative account of practical reasoning—
an account according to which our interest in responding to reasons is an
interest in responding to considerations whose normative force depends on
something more than our own commitments, be they our commitment to
acting or our commitments to achieving certain substantive goals.
These three conclusions emerge from a train of thought that can be
summarized as follows. (1) We need an account of rational agency that does
justice to the possibility of inner conflict. This means that (2) an account of
rational agency must do justice to the fact that if a rational being can
question the desirability of satisfying her desires, then she does not qualify
as an agent simply in virtue of engaging in instrumentally rational, goal-
directed behavior; satisfying this condition does not suffice to ensure that
the causal power of her motives can be attributed to her. From (1) it also
follows, however, that (3) an account of rational agency must not assign to
the “parts” of the soul essential to rational agency two roles so independent
from each other that there is no way to explain how they could be two
aspects of a single agent; if someone were alienated from herself in a way she
could not overcome without acting, then she would lack the resources
necessary for constituting herself as an agent. More particularly, (4) on
any adequate account of rational agency, the nonrational part of the soul
must be capable of appreciating the force of the claims/interests/concerns of
the rational part, even though what is distinctive about it is precisely that it
is not defined by these claims/interests/concerns. This means that (5) there
must be something about the content of our nonrational desires/inclinations
that makes them open to cooperate with, or obey, our normative verdicts.
 The Possibility of Action 45

That is, there must be something about us in our nonrational capacity that
inclines us to cooperate with, or obey, the interests, or demands, that define
us as “rational.” It also means that (6) we cannot be so alienated from
ourselves that the conditions necessary for unifying our normative verdicts
and our nonrational dispositions—the conditions necessary for being an
agent—determine what we have reason to do. In particular, (7) insofar as
our point of view is not constituted by our normative verdicts, we must take
ourselves to have reason to accommodate ourselves to these verdicts, and so
we must take there to be reasons for action that are not grounded in the
conditions necessary for agency.

REFERENCES

Anscombe, G.E. (1957). Intention, 2nd edn. (Ithaca, NY: Cornell University Press).
Bok, Hilary (1996). “Acting Without Choosing.” Noûs 30 (2): 174–96.
Bratman, Michael (2002). “Hierarchy, Circularity, and Double Reduction.” In
Sarah Buss and Lee Overton (eds.), The Contours of Agency (Cambridge, MA:
MIT Press), 65–85.
Brewer, Talbot (2009). The Retrieval of Ethics. (Oxford: Oxford University Press).
Buss, Sarah (1997). “Weakness of Will.” Pacific Philosophical Quarterly 78: 13–44.
—— (2012). “Autonomous Action: Self-Determination in the Passive Mode.”
Ethics 122: 647–91.
—— “Norms of Rationality and the Superficial Unity of the Mind” (Unpublished
manuscript).
Frankfurt, Harry. (1971). “Freedom of the Will and the Concept of a Person.” The
Journal of Philosophy LXVIII: 5–20. Reprinted in Frankfurt 1988, 11–25.
—— (1977). “Identification and Externality.” In Amelie Oksenberg Rorty (ed.),
The Identities of Persons (Berkeley, CA: University of California Press). Reprinted
in Frankfurt 1988, 58–68.
—— (1988). The Importance of What We Care About. (New York: Cambridge
University Press).
Hornsby, Jennifer (2004). “Agency and Alienation.” In Mario De Caro and David
Macarthur (eds.), Naturalism in Question (Cambridge, MA: Harvard University
Press), 173–87.
Kant, Immanuel (1956). The Groundwork of the Metaphysics of Morals. Trans..
H. J. Paton. (London: Hutchinson & Co).
Korsgaard, Christine (1996). The Sources of Normativity. (Cambridge: Cambridge
University Press).
—— (2007). “Autonomy and the Second Person Within: A Commentary on
Stephen Darwall’s the Second-Person Standpoint.” Ethics 118 (1): 8–23.
—— (2008). “The Normativity of Instrumental Reason.” In The Constitution of
Agency. (Oxford: Oxford University Press), 27–68.
46 Sarah Buss

—— (2009). Self-Constitution: Agency, Identity, and Integrity. (Oxford: Oxford

University Press).
Moran, Richard (2001). Authority and Estrangement. (Princeton, NJ: Princeton
University Press).
Pettit, Philip and Smith, Michael (1990). “Backgrounding Desire.” The Philosoph-
ical Review XCIX: 565–92.
Railton, Peter (2011). “The Affective Dog and Its Rational Tail.” Unpublished
manuscript.
Rapoport, Judith (1991). The Boy Who Couldn’t Stop Washing. (New York: Penguin
Putnam Inc.).
Raz, Joseph (1985). “Authority and Justification.” Philosophy and Public Affairs
14 (1): 3–29.
Schapiro, Tamar (2011). “Foregrounding Desire: A Defense of Kant’s Incorpor-
ation Thesis.” Journal of Ethics 15: 147–67.
Smith, Michael (2012). “Four Objections to the Standard Story of Action (And
Four Replies).” Philosophical Issues 22, Action Theory: 387–401.
Velleman, David (1992). “What Happens When Someone Acts?” Mind 101 (403):
461–81.
—— (1996). “The Possibility of Practical Reason.” Ethics 106 (4): 604–726.
—— (2005). “Precis of the Possibility of Practical Reason.” Philosophical Studies
121 (3): 225–38.
—— (2009). How We Get Along. (New York: Cambridge University Press).
Wallace, R. Jay (2001). “Normativity, Commitment, and Instrumental Reason.”
Philosophers’ Imprint 1 (3): 1–26.
Watson, Gary (1975). “Free Agency.” The Journal of Philosophy 72: 205–20.
—— (2004). “Disordered Appetites: Addiction, Compulsion, and Dependence.”
Reprinted in Gary Watson, Agency and Answerability (Oxford: Oxford University
Press): 59–87.
Williams, Bernard (1981). Moral Luck, 1–19. (Cambridge: Cambridge University
Press).
 2
The Fecundity of Planning Agency
Michael E. Bratman

1. THREE PRACTICAL CAPACITIES

As normal adult human agents we have a remarkable trio of capacities.

First, we are capable of acting over time in ways that involve important
forms of intentional cross-temporal organization and coordination. Think
about growing vegetables in your garden. You need to prepare the soil,
plant, water, cultivate, and harvest. These activities take place over time,
and each of them is infused with the agent’s understanding of and commit-
ment to the larger temporally extended arc of the activity. This is a case of
temporally extended intentional agency.1
Second, we are capable of acting together with others in ways that go
significantly beyond standard forms of strategic interaction. Think about a
piano quartet. It is not just that each is acting in a way that best promotes
what she wants given her knowledge of what the other agents are doing, and
given that the other agents’ thoughts and actions have a parallel structure.
Instead, each sees herself and her partners as acting together in ways that
involve distinctive forms of commitment and responsiveness to the joint
activity and so to each person’s participation in that joint activity. This is a
case of shared intentional activity.2
Third, we are capable of self-governance. On reflection, however, this
can seem mysterious. The picture of a self stepping back from the psychic
flow and putting a thumb on the scales is deeply problematic even though it
does echo aspects of our experience of our agency. But in the absence of
such a homunculus, what is it for the agent herself to be governing?
A theory of human agency should include an understanding of this trio
of capacities: capacities for temporally extended, for shared, and for self-

1 A related idea is in Ferrero 2009.

2 Indeed, it is what I call a shared cooperative activity. See Bratman 1999a. But here
I will work with the somewhat broader concept of a shared intentional activity.
48 Michael E. Bratman

governed intentional agency. And in each case we have three interrelated

concerns. First, we want to know what concepts best support our theoret-
ical understanding of these phenomena. Second, we want to understand
how it is that we realize these capacities. Third, we want to understand the
distinctive normative aspects of these capacities. Our concerns are, then,
conceptual, metaphysical, and normative.
And my response to these concerns is to seek to understand these three
human practical capacities by appeal to a common, core capacity for
individual planning agency. This core, planning capacity itself suffices
for the capacity for temporally extended intentional agency. Further,
when appropriately supplemented, this core capacity suffices for capacities
for shared intentional and self-governed agency; and these supplemental
resources are broadly continuous with the resources already available within
our theory of individual planning agency. In this way we provide a bridge
from the conceptual, metaphysical and normative resources of our model of
our individual planning agency to conceptual, metaphysical, and normative
resources adequate for a model of each of the trio of practical capacities that
we have highlighted. We thereby articulate structures of planning agency
that suffice for central cases of temporally extended, shared, and self-
governed intentional agency. The conjecture of the fecundity of planning
agency is that this trio of human practical capacities, and perhaps others as
well,3 is in this way grounded in our capacity for planning agency.4

2. PRACTICAL THINKING AND TIME

Not all agents are planning agents. There can be goal-directed agents who
do not structure their thought and action over time by way of planning.
Dogs and very young human children are likely examples. Such agents
engage in goal-directed activity in which their behavior tracks a goal—say,
getting a drink of water. And this may involve complex plasticity of
response—a thirsty dog may respond sensibly to your moving the bowl
of water. But it seems unlikely that the dog, or the fifteen-month human,
settles on plans for a temporally extended period, and thinks about and

3 Allan Gibbard sees our capacity for planning agency as at the bottom of normative
thinking (Gibbard 2003). And Scott Shapiro sees law as grounded in our planning
capacities (Shapiro 2011). In these ways both Gibbard and Shapiro aim to extend the
fecundity of planning agency.
4 Bratman 2010. What is central is the claim to provide sufficient conditons for
central cases of these forms of human agency. The conjecture of the fecundity of agency
leaves open the possiblity of multiple such sufficient conditions.
 The Fecundity of Planning Agency 49

guides his behavior over time in the light of those plans. Of course, whether
the dog, or the young child, does or does not guide his thought and action
in this plan-infused way is an empirical issue; perhaps psychological
research on these matters will surprise us. But the central point here is
the distinction between, on the one hand, temporally local goal-directed
agency and, on the other hand, plan-guided diachronic agency.5
In planning we take seriously what to do later, as well as what to do now.
But why worry about what to do later? Why not just worry about what to
do now and cross your bridges when you come to them later? The
commonsense answer is that we need to figure out what to do later in
order to figure out what to do now and to coordinate our earlier with our
later activities. Once I settle on going to New Orleans next month it is clear
that I need soon to get an airline ticket. In contrast, if I just ask right now
“should I right now get a ticket to New Orleans for next month?”
I normally cannot sensibly answer without settling on whether to go
there then. In the interest of cross-temporal coordination, I need many
times to settle now what I will do later, and then plan accordingly about
what to do now. In settling now to go to New Orleans next month I come
to be in a plan state that is future-directed. And we understand what such a
plan state is by explaining its role in the rational dynamics of planning
agency.
This pressure in the direction of future-directed planning is especially
acute for resource-limited agents like us.6 We frequently need to settle now
what to do later in part because we cannot reasonably trust that when the
later time arrives we will be able, given the pressures of action, to sort out all
the complexities efficiently and reach a sensible decision from scratch.
In developing this idea of the cross-temporal coordinating role of plan-
ning we need also to do justice to two further, commonsense ideas. The first
is the idea of choice among alternative options each of which would be
sensible—one aspect of, as we might say, the idea of “will.”7 We make such
choices all the time, from different routes to a destination, to different
careers, to the choice made by Sartre’s young man between aiding his
mother and fighting with the Free French.8 And these choices, while
underdetermined by the agent’s view of her reasons, nevertheless settle
relevant practical matters and play central, forward-looking roles in cross-

5 For the idea of planning agency as a species of agency, see my use of Paul Grice’s
strategy of “creature construction.” Grice 1974–5 in Bratman 2007 f.
6 Simon 1983.
7 A second aspect is the idea of willpower. See Holton 2009. And here too, as Holton
argues, the planning theory can be of use.
8 Sartre 1975, 354–6.
50 Michael E. Bratman

temporal coordination. The ability to make choices and settle such matters
seems a basic feature of human agency, one that helps explain why we are not
constantly suffering the plight of Buridan’s ass. This is an ability to make a
transition from a state of indecision between several alternatives, each of
which seems sensible, to a state of being settled on one of those alternatives in
particular. We shed light on this ability by saying more about the output state
of being settled on one of those alternatives in particular. And we understand
this output state in large part in terms of its role in planning agency: to be
settled on an option is, or anyway is closely related to,9 being in a plan state,
where, again, we understand what a plan state is by explaining its role in the
rational dynamics of planning agency. In this way we resist an overly simple
time-slice conception of the will by embedding our account of the will within
a larger model of diachronic planning agency.
There is, second, the very idea of intention. Here we need to be careful to
distinguish adverbial forms—as in “he did it intentionally”—from the verb
“to intend.” The idea that is central here concerns the latter, and the
corresponding phenomenon of intending; and we can allow for a more or
less complex relation between X-ing intentionally and intending to X.10
The central thought here is that to intend to do something is to be in a plan
state, where—again—we understand what a plan state is by explaining its
role in the rational dynamics of planning agency. Intending leads to action
in ways that normally involve diachronic planning structures.
So a model of the rational dynamics, and the associated cross-temporal
coordinating roles, of planning agency would also be a theory of the will and a
theory of intending.11 This promises a deeper understanding of the forward-
looking nature of the will, and of the complex diachronic interconnections
between our thought and our action. And the conjecture of the fecundity of
planning agency is that such a planning model sheds light on the trio of
agential capacities with which we began.

9 This qualification is needed to allow for the complex relation between choice and
intention that I discuss in Bratman 1987, ch. 10.
10 This is where I have argued against the “Simple View” according to which it is
always true that in intentionally A-ing one intends to A. (Bratman 1987, ch. 8). In that
discussion I also considered the “single phenomenon view” according to which when one
intentionally A’s one intends something, though perhaps not, in particular, A. I said that
something like this will be true in “the vast majority of cases,” but I left it open that there
might be marginal cases in which one acts intentionally and yet there is no intention—no
plan-state—involved. (pp. 126–7.) What is important here is the conjecture, built into
the planning theory, that cases of intentional action in which relevant plan states are
completely absent, if such there be, are theoretically marginal for our understanding of
our adult human agency. For a challenge to this conjecture See Velleman 2007.
11 Holton 2009, 5, argues that “there is solid psychological evidence that we have
intentions along something like these lines.”
 The Fecundity of Planning Agency 51

3. PLANNING AGENCY

Let me now sketch a model of our individual planning agency that I have
discussed elsewhere.12 We begin with plans of action. Such a plan specifies
ways of acting now and into the future. It ties these ways of acting, together
with what one believes about the world, into a more or less consistent and
coherent web, though a web that will normally be both partial in important
ways and to some extent conditional in structure. To plan to do something
is to be appropriately committed—committed in a distinctive, practical
way—to a plan of action that says to do that. Planning agents are
systematically involved in such commitments to relevant plans, and these
commitments shape their practical thought and action over time.
Intending to do something is, I have said, being committed in the
relevant way to a plan that says to do that (perhaps conditionally). We try
to explain this idea of being appropriately committed to such a plan of
action by appeal to two general ideas. First: we try to characterize the
normal roles—the normal ways of functioning—that are characteristic of
such plan states; and second we try to articulate basic planning norms,
the at-least-implicit acceptance of which by the agent is involved in those
roles.
Begin with these planning norms. Intentions—that is, plan states—are
subject to a characteristic quartet of norms. Intentions are to be internally
consistent and consistent with the agent’s beliefs. It needs to be possible to
agglomerate one’s various intentions together into an overall intention that
is itself consistent and consistent with one’s beliefs. Intentions in favor of
ends engage demands of means-end coherence in the direction of filling
in one’s—normally, partial—plans appropriately, and as time goes by, with
means and preliminary steps.13 And structures of plan states, though
potentially a target of reconsideration, are subject to a norm of stability
over time: there is some sort of defeasible presumption in favor of
one’s prior plan states. In short, there are planning norms of consistency,
agglomeration, means-end coherence, and stability. A planning agent will
at least implicitly accept these norms and this will tend to guide that agent’s
thought and action in ways that support conformity to these norms.
To accept these norms is not simply to be disposed to conform to them.
One is also set to see divergence from these norms as breakdowns, and so to

12 See esp. Bratman 1987.

13 Having noted the category of preliminary steps, I proceed to ignore it and focus on
means.
52 Michael E. Bratman

respond to such cases with a kind of “Darn it!” reaction. One is thereby set
to be guided in the direction of conformity with the norm.14
This norm-guidance is part of the explanation of characteristic roles that
such plan states play. Given the pressure from an accepted planning norm
to avoid means-end incoherence, prior, partial plan states pose problems
about means; and the agent will be set to fill in her plans accordingly. This
will normally induce end-means reasoning that is structured by prior,
partial plans. Given the pressure from accepted planning norms in favor
of agglomeration and consistency, plan states will tend to filter out from
deliberation options intending which would introduce new plan inconsist-
encies. And given the pressure from an accepted planning norm of stability,
a prior intention will tend to persist, other things equal. This persistence
is sometimes a result of the snowball effect: once one begins acting on a
prior plan one changes the world in ways that tend to support continuing
with that plan.15 And sometimes this is because the costs of reconsidering
prior plans—and/or the risks that reconsideration poses of undermining
important coordination previously forged—are too high in the absence of
recognized reasons for change; and a prior plan will tend to persist unless it
is reconsidered.16 But our planning agency also involves, I think, a
distinctive norm of stability that directly says to treat one’s prior plans as
a defeasible default.17
So we have a package of characteristic roles together with associated
norms that are at least implicitly accepted. Such a planning psychology
induces and supports complex forms of temporal organization of thought
and action. And a plan state in this planning psychology can favor a specific
option despite known underdetermination by prior reasons. Once Sartre’s
young man settles, say, on a plan to work with the Free French, demands of
consistency and means-end coherence come to bear and support corres-
ponding forms of temporally downstream functioning. This is true even
though his decision in favor of the Free French was, by his own lights,
underdetermined by his prior reasons. And we can make a similar point
about less dramatic examples of career decisions, or even of different routes
to an intended destination.
If all goes well, planning structures induce cross-temporal referential
connections that are both forward and backward looking. My present

14 For discussion of norm acceptance See Gibbard 1990. See also Railton 2006.
15 See Bratman 1987, 82, where I note that I owe to John Etchemendy the label
“snowball effect.”
16 Yet another potential explanation of this persistence is that we are frequently
guided by habits of nondeliberative nonreconsideration, habits that are, in general,
useful to us.
17 Bratman 2010, 2012.
 The Fecundity of Planning Agency 53

plan to paint the house this week at least implicitly refers to my later, then-
present-directed intention to put on the final coat of paint; and that later
intention at least implicitly refers back to my earlier intention. And the
stability of my intention in favor of painting this week helps support a
coordinated flow of activity over time. These cross-temporal referential
interconnections and constancies help support an effective temporally
extended structure of partial plans. These partial plans provide a back-
ground framework for further deliberation about means, deliberation
shaped by accepted norms of coherence, consistency, and agglomeration.
This idea of cross-temporally stable and referentially interlocking atti-
tudes is familiar from the Lockean tradition of understanding personal
identity over time—or, at least, “what matters” in such identity18—by
appeal to overlapping strands of continuities of attitude and referential
connections across attitudes.19 The standard functioning of plan states in
our planning agency involves such broadly Lockean cross-temporal ties.20
Now, in speaking of accepted planning norms I have so far not raised the
question whether these norms do indeed have normative force or signifi-
cance. I have, up till now, only appealed to the explanatory role of their
acceptance in a planning psychology.21 But I do not think we should rest
content simply with such a “positivist” theory of planning agency. There
are two reasons for this. The first is that we ourselves are planning agents:
our theory of planning agency is a theory of our agency. So the question
whether these norms have normative force is one we face in such theorizing.
After all, there are likely many patterns of thought that are pervasive in
human agency as we know it, but which we would not, on reflection,
endorse.22 So we will want to know whether, and if so why, these planning
norms do pass reflective muster. Second, it would be an important
theoretical fact about planning agency if these structures of agency were
indeed robust in the face of reflection on its basic principles. One important
way to ascertain whether our planning agency has such robustness is
directly to ask whether these planning norms really do have normative
force.23 And a positive answer would defuse a concern that the planning

18 Parfit 1984, 217.

19 See John Locke, An Essay Concerning Human Understanding Bk. 2 ch. 27.
20 The idea is only that these planning ties are among the broadly Lockean ties, which
also include, most prominently, memory. So one can survive a breakdown in these
planning ties.
21 For a closely related distinction See Schroeder 2003. And also See Broome 2007.
22 See e.g. the essays in Kahneman et al. 1982. For a discussion of related matters, See
Morton 2011.
23 In saying this I remain neutral about the extent to which it is up to us whether to
continue to be planning agents.
54 Michael E. Bratman

theory mis-describes our practical thinking insofar as it ascribes to us the

acceptance of norms that are not defensible.24
A closely related issue is whether, and in what sense, these planning
norms are norms of rationality. We can begin by observing that synchronic
planning norms of consistency, agglomeration, and means-end coherence
have obvious parallels with synchronic norms of consistency, agglomer-
ation, and theoretical coherence of belief; and there is perhaps also a parallel
between the diachronic planning norm of stability and a diachronic
norm of belief conservation. I have argued elsewhere that the practical
planning norms are not simply a matter of theoretical norms on belief.25
Nevertheless, there remains an important parallel. Given their tight tie
to belief, there is an initial case for seeing these norms of belief as norms
of theoretical rationality for agents who have beliefs. Similarly, given
their tight tie to intending/planning there is an initial case for seeing
these planning norms as norms of practical rationality for planning
agents; though, since not all agents are planning agents, they need not be
rationality norms for all agents.26
More needs to be said; but we need first to return to the trio of practical
capacities with which I began this essay.

4 . T E M P O R A LL Y E X T E N D E D A G E N C Y

The capacity for temporally extended intentional agency is the capacity to

guide one’s actions in light of one’s grasp of their location in a larger
temporally extended structure of activity to which one is practically com-
mitted. One plants the seeds as a part of a larger activity of growing beans in
one’s garden, a larger activity to which one is practically committed. This
cross-temporal organization of this activity is mind-infused: this is not
merely the cross-temporal biological organization of seeds developing
into beans.
In what sense mind-infused? Here my proposal is that for us the normal
explanation of this kind of temporally extended intentional activity
involves the agent’s plans and planning: one’s plan is to grow the food by

24 For a version of this concern See Kolodny 2009.

25 See esp. Bratman 2009a.
26 This idea of rationality norms for a specific kind of agent fits well within the
strategy of Gricean creature construction that lies behind the planning theory. Kieran
Setiya discusses a related idea, with skepticism, under the rubric of “pluralistic rational-
ism” in Setiya forthcoming. I am very much indebted to Setiya’s discussion, though
I reach different conclusions.
 The Fecundity of Planning Agency 55

cultivating the garden; one recognizes that this needs to involve relevant
means such as planting the seeds and watering; and so one includes those
means within the plan to which one is practically committed. One’s
practical commitment to this plan involves one’s being set intentionally
to act in these specified ways in the course of following through with one’s
overall plan. In this way our temporally extended intentional activity is an
exercise of our planning capacities.
Michael Thompson has criticized what he calls “the tendency of students
of practical philosophy to view individual human actions as discrete or
atomic or point-like or eye-blink-like units.”27 Now, I think that the
planning theory in my 1987 book is not guilty of this purported tendency:
a central concern of that book was, after all, to understand how planning
rationally supports our temporally extended agency. In the present context
the point is that the first dimension of the fecundity of planning agency—its
support of our capacity for temporally extended intentional agency—
continues to avoid an atomistic tendency that, as I agree with Thompson,
we should avoid.28

5. SHARED AGENCY

Turn now to our capacity for shared intentional activity. The proposal here
is that our capacity for shared intentional activity is grounded in our
individual planning capacities in the sense that the proper functioning of
those planning capacities, given relevant special contents of the plans,
contexts of the plans, and interrelations among the participating planning
agents, would constitute a basic case of shared intentional activity.
As anticipated, this does not mean that simply by having the capacity for
planning agency one thereby has the capacity for shared intentionality. To
get from planning capacities to the capacity for shared intentionality we
need further resources in the form of distinctive contents, contexts, and
interrelations. So there remains the possibility of planning agents who are
not capable of shared intentionality. Indeed, it is some such gap between
planning agency and shared intentionality that Michael Tomasello and his
colleagues have highlighted as a potential key to a fundamental difference
between humans and the great apes.29 Nevertheless, the further resources—
conceptual, metaphysical, and normative—involved in the step from

27 Thompson 2008, 91.

28 Though Thompson would not agree with my way of avoiding such atomism.
29 See e.g. Tomasello 2009.
56 Michael E. Bratman

individual planning agency to shared intentional agency will be available

within our theory of individual planning agency.30 Or so I conjecture.
The key is the idea of shared intention. When we paint the house
together, or dance together, or play a quartet together, or have a conversation
together, or perform an experiment together, or—in Margaret Gilbert’s
example—walk together,31 our activity is explained by our shared
intention in favor of our so acting. We are walking together because that
is what we intend to do—where talk of what we intend to do is talk of our
shared intention so to act. Or at least this connection between shared
activity and shared intention holds in the kinds of cases that are our
central concern. In this respect shared intentional activity parallels
individual intentional activity: in each case, the explanatory role of
intention (individual or shared) is fundamental. And the idea is to
articulate structures of interconnected individual planning agency whose
rational functioning would constitute the rational functioning of shared
intention.
If we are going to see certain cases of interconnected rational functioning
of individual planning agents as constituting the rational functioning of
shared intention, we need to say more about what such rational functioning
of shared intention is. So let us ask the same questions about shared
intention we asked about individual intention: why do we bother with
shared intentions? What fundamental roles do they play in our lives, and
what norms are associated with those roles?
My response to these queries is to highlight analogies with the coordin-
ating and structuring roles of intentions and plans in the individual case,
and with the associated norms of individual intention rationality. The
characteristic roles of a shared intention to X will include interpersonal
coordination of action and planning in pursuit of X, and the structuring of
related bargaining and shared deliberation concerning how we are to X.
And these roles will be associated with norms of social consistency and
agglomeration, social coherence and social stability. Roughly, it should be
possible to agglomerate relevant intentions into a larger social plan that is
consistent, that in a timely way adequately specifies relevant means, and
that is associated with appropriately stable social psychological structures.
Failure to satisfy these social norms will normally undermine the distinctive
social roles of shared intention.
We want then to specify a structure of interconnected plan states of
individuals (and other related attitudes of those individuals) in appropriate

30 Though see the qualification below about “out in the open.”

31 Gilbert 1990.
 The Fecundity of Planning Agency 57

contexts that would, when functioning in the norm-guided ways high-

lighted by the planning theory of individuals, play these roles characteristic
of shared intention in part by way of conformity with these associated
norms of social rationality.32 This would allow us to say that such
individualistic planning structures constitute shared intention, or at least
one important kind of shared intention.
We can call such a broadly individualistic structure of interconnected
planning agents a construction of shared intention, and the view on order as
a kind of constructivism about shared intention. The idea is not that the
participants themselves construct the shared intention: it is we, the theor-
ists, who do the constructing.33 And the proposal is that some such
construction will constitute shared intention, or at least a central kind of
shared intention.
I have tried to develop such a construction elsewhere.34 Here I proceed in
broad strokes. What we need are a number of ideas that are within the
domain of our theory of individual planning agency but when put together
allow us to characterize a central form of shared intention and shared
intentional activity. In particular, and focusing on our shared intention to
paint the house, the initial ideas we need are:
a. Each of us intends that we paint the house.
b. Each of us intends that we paint the house by way of the intention of the
other that we paint the house.35 In this sense our intentions interlock.
c. Each of us intends that we paint the house by way of mutual responsive-
ness between us in sub-plan and in action. So, in particular, each intends
that our sub-plans for our painting mesh with each other, in the sense of
being co-possible.

32 In putting the idea in this way I am signaling an asymmetry. In the individual case
the basic explanation of conformity to the individualistic rationality norms involves the
at-least-implicit acceptance of those very norms by the relevant individuals. In the shared
case, in contrast, the initial explanation of conformity to the cited social norms appeals to
the at-least-implicit acceptance of the norms of individual plan rationality by the relevant
individuals. However, once these interconnected planning structures are on board it may
well be that the individual participants also accept, and are guided by, the social norms
themselves. I discuss these matters in Bratman forthcoming a.
33 Though in special cases there will also be a sense in which the participants
themselves construct the sharing. The theoretical construction I am envisaging would
be a part of a larger strategy of Gricean creature construction.
34 See e.g. the quartet of essays on this subject in Bratman 1999a, Bratman 2009c,
and Bratman forthcoming a.
35 I also think we need the idea that each intends that we paint by way of his own
intention that we paint. But to keep things more manageable I put aside here this
condition of reflexivity.
58 Michael E. Bratman

d. Each of us believes that our intentions in a. are interdependent in their

persistence, and that given these intentions we will indeed paint the
house.
e. There is in fact interdependence in persistence of the intentions in a.
f. These conditions are out in the open among us.
These conditions describe a structure of interconnected intentions of each,
and associated beliefs. The constructivist proposal is that when this struc-
ture of attitudes functions in accordance with the norms of the planning
theory of individual agency it thereby realizes the roles characteristic of
shared intention in part by way of conformity with the associated social
rationality norms. Further, when this structure in fact leads to our painting
the house by way of the cited kinds of mutual responsiveness, there will be
shared intentional activity. Such shared intentional activity is the activity of
interconnected planning agents.
Without trying fully to defend this proposal here, let me make some
brief comments. First, to avoid problems of circularity, the concept of our
painting the house that appears in the contents of these intentions of each
should (at least in basic cases) be understood as neutral with respect to
shared intentionality. We might paint together in this neutral sense in a
case in which we are each merely individually painting the same house
while keeping an eye out not to bump into each other. (What will be crucial
for shared intentionality is that this neutrally characterized pattern of our
individually intentional activities is explained by the kinds of intentions
cited in the construction.)
Second, these conditions appeal to the phenomenon of my intending
that we act in a certain way: not all intending is intending to.36 I think that
once we understand intending within the framework of the planning theory
this becomes a fairly natural idea,37 so long as each participant also believes
that her own intention appropriately settles whether they are going to act
accordingly. And my conjecture is that each can sensibly have such a belief
when each has the beliefs cited in d. about interdependence and efficacy.38
Third, the condition of intention-interlocking precludes cases in which
each intends that the joint action proceed in a way that does not involve the
intention of the other, as when a certain Mafioso, intending to throw the
other into the trunk of the car, announces that “we are going to New
Orleans together.” In contrast with such a mafia case, the idea here is that

36 Davis 1984. Though, as Randolph Clarke emphasized in conversation, a rational

agent who intends that we J will normally also intend to act in certain ways.
37 As Philip Cohen once suggested in conversation.
38 For this last point See Bratman 1999b.
 The Fecundity of Planning Agency 59

each intends that the joint activity proceed by way of the relevant intention
of the other participant, and by way of associated mutual responsiveness
between the intentions and actions of the participants. And this intended
mutual responsiveness will need to involve mesh between the relevant sub-
plans of each.
A fourth observation concerns the condition that all this is out in the
open.39 A central reason for this condition is that we want to model a kind
of shared reasoning that is characteristic of shared intention, namely: shared
reasoning about how to carry out our shared intention to paint the house
together.
I leave it as an open question whether the relevant idea of “out in the
open” goes somewhat beyond the conceptual resources of our theory of
individual planning agency. Except for that possible qualification, however,
I think that the conceptual, metaphysical, and normative resources at work
in this proposed construction are within the domain of the resources of our
theory of individual planning agents: the construction is conceptually,
metaphysically, and normatively conservative.
The basic conjecture, then, is that these interdependent and interlocking
intentions of each, taken together with the cited beliefs and in a public
context, will, in responding to the rational pressures associated with indi-
vidual planning agency, function in ways that constitute the social-norm-
conforming social functioning of shared intention. So this structure of
interdependent and interlocking intentions constitutes at least one central
form of shared intention.
In partial support of this basic conjecture, note that we are supposing
that each intends the joint painting by way of the other’s intention, mutual
responsiveness, and meshing sub-plans. It is not just that each intends his
part and merely expects the other to play her part.40 This means that the
rational pressure on each to make her own plans coherent and consistent
ensures rational pressure on each to support the success of the joint activity
and the meshing role of the other in that activity. Given my intention that
we paint by way of your intention, mutual responsiveness, and meshing
sub-plans, I am under rational pressure to coordinate with you, to support
your role—perhaps by way of helping actions—and to avoid ways of acting

39 It is here that talk of common knowledge and/or mutual belief is potentially apt.
40 This contrasts with a case in which each only intends his own activity while
expecting the others to perform their part, and the social organization is out-sourced
to some external managerial group. In such a case the participants will not each be
committed to the joint activity in a way that is normally involved in shared deliberation
about how to proceed. (Scott Shapiro discusses large-scale versions of this latter sort of
case in Shapiro forthcoming.)
60 Michael E. Bratman

that are incompatible with all that. And given your analogous intention,
you too are under analogous rational pressures. These rational pressures on
each, given these distinctive contents and interrelations, induce pressures
in the direction of social coherence and consistency, and associated
coordination and effectiveness, pressures that are characteristic of shared
intentionality.
This approach to shared intentionality highlights conceptual, metaphys-
ical and normative continuities with individual planning agency. This
contrasts with views that see the step to shared intentionality as necessarily
involving a basic new conceptual and/or metaphysical and/or normative
resource. For example, John Searle thinks that shared intentionality
involves a special kind of “we-intention” in the heads of the individual
participants. We-intentions, on Searle’s view, are distinctive attitudes, not
to be identified with ordinary intentions that concern our own activity.41
Again, Margaret Gilbert thinks that shared intentionality involves a
primitive interrelation of “joint commitment” between the participants,
an interrelation that essentially involves distinctive nonmoral obligations.42
Without pausing to examine the details of either of these views, we can say
that if the plan-theoretic model I have proposed does succeed in providing
sufficient conditions for shared intentionality,43 then the burden of proof
is on both Searle and Gilbert to explain why we also need these new
primitives.44

6 . SE LF - GO V E R N AN C E

Turn now to our capacity for self-governance. Here my proposal is that a

key to a nonhomuncular model of our self-governance is the idea that
certain plan states both concern what matters in the sense of having weight
in our deliberative thought,45 and when functioning properly tie together
our thought and action in relevant ways, both synchronically and
diachronically. These plan-like commitments to weights speak for the
agent, and when these plan states guide, the agent governs. This capacity
for self-governance involves the capacity for guidance of thought and action

41 Searle 1990.
42 Gilbert 2009.
43 What about the obligations that are commonly present in cases of shared intention
and shared intentional action? On the plan-theoretic model these will normally be
familiar moral obligations, such as an obligation to follow through with an assurance,
or in accordance with reliance that is intentionally induced.
44 As Michael Smith noted in conversation.
45 I put to one side other possible ways in which certain considerations can matter.
 The Fecundity of Planning Agency 61

by plan-like commitments to weights. Not all planning agents have this

latter capacity. But its addition is a conservative extension of our model of
planning agency, and is an important step towards sufficient conditions for
self-governance.
I am led to this view by reflection on the idea that in self-governance the
agent’s relevant practical standpoint appropriately guides her thought and
action. Such a practical standpoint serves as an anchor for deliberation,
thereby supporting the idea that its guidance is a form of governance. And
given the overall role of this standpoint in the agent’s psychic economy, its
appropriate guidance constitutes the agent’s governance of his thought and
action.46
Should we say that the agent’s practical standpoint is constituted by her
evaluative or normative judgments? Here I aim to be neutral as between
different metaethical accounts of the nature of such judgments. I will,
however, assume that such judgments are not merely personal preferences
but are in some sense in the domain of the intersubjective.47 Different
metaethical theories will interpret this idea differently. Some will appeal to
the idea of rational convergence; some will appeal to the idea of enjoining
others to converge. But I aim to be neutral about these differences.
That said, I think that there are difficulties with the proposed appeal to
evaluative or normative judgment. First, some such judgments may not be
a part of one’s relevant standpoint. As David Velleman observes, one can be
alienated from one’s evaluative or normative judgments.48 Perhaps this is
what we should say about Huck Finn’s judgment that he ought to turn in
Jim.49 A related point is that one can sometimes be, as Allan Gibbard puts
it, “in the grips” of a normative or evaluative judgment.50 The subjects in
Milgram’s famous experiment may have believed that they should do what
the person in the white coat told them to do; yet their wrenching conflict
may indicate that this belief is not located securely in their standpoint. Yet
further, one can judge that something is a good thing, but have no interest
in it at all.51 There are, after all, many good things and not that much time.
In these cases there is reason to exclude certain normative or evaluative
judgments from the agent’s practical standpoint. There are also cases in

46 Bratman 2007b. This general approach—though not the details—is in the spirit of
work of Harry Frankfurt. See Frankfurt 1988a and 1999a. (An important difference is
my emphasis on the role of these commitments to weights in deliberation.) Concerning
the agent’s guidance, see Velleman 1992.
47 Bratman 2007e, 151–4.
48 Velleman 1992, 472.
49 Bennett 1974, Arpaly and Schroeder 1999, and Driver, 2001, ch. 3.
50 Gibbard, 1990, ch. 4. The example to follow comes from Gibbard’s discussion.
51 See Harman 2000b, 129.
62 Michael E. Bratman

which important elements in that standpoint resist identification with

evaluative or normative judgment. Think about the kinds of love that
have been highlighted by Harry Frankfurt.52 What I love will normally
play a central role in my practical standpoint. But in many cases it seems
strained to see such loving as identical with a normative or evaluative
judgment. Further, even when one seeks to respond to one’s judgments
about the right and the good, these judgments may underdetermine one’s
practical commitments. Sartre’s young man may see considerations of loyalty
to his mother and of loyalty to the Free French as noncomparable, and yet
arrive at a commitment to give significantly more weight to the interests of
his mother. Or he might be struck by broad disagreement in reflective views
about the relative importance of these different forms of loyalty, and so be
led, by way of a kind of humility of judgment, away from an intersubjectively
accountable evaluative judgment in favor of one over the other. Yet he
still might go on to a commitment to giving significantly more weight
to one rather than the other. In each case such practical commitments
go beyond evaluative judgment, but may still loom large in the agent’s
practical standpoint.53
These considerations suggest that a central element of a planning agent’s
practical standpoint will be plan-like commitments to weights in deliber-
ation. Such commitments are missing in cases of alienated or noninterested
value judgment. In contrast, love involves plan-like commitments to give
significant weight to the interests of the beloved. (Which is not to say that it
only involves that.) Commitments to certain personal ideals or projects
involve plan-like commitments to give weight to relevant considerations.
These commitments to weights will frequently be associated with relevant
judgments about the right and the good. Nevertheless, these commitments
are not in general identical with or ensured by such judgments, and they
may settle matters in response to underdetermination by such judgments.
The role of these commitments is to settle where one stands on questions
about how to live one’s own life; and they need not be intersubjectively
accountable in the way in which evaluative judgment is.54

52 Frankfurt 2006.
53 Bratman 2007b, 233–8.
54 Bratman 2007e, 153. Since these plan-like commitments to weights will normally
have an important generality, I have called them self-governing policies; and I have also
emphasized that they will include policies about the significance of one’s own desires
and/or what those desires are for. See Bratman 2007b and, 2007c. There are important
similarities between my appeal to policies about weights and Alan Gibbard’s appeal to
plans “of ‘treating R as weighing in favor of doing X’,” in Gibbard 2003, 189. Gibbard’s
main concern is his metaethical proposal that judgments about normative reasons are
expressions of such plans. My main concern is with the role of such plan states in self-
governance.
 The Fecundity of Planning Agency 63

What explains why such plan-like commitments speak for the agent55
and are such that when they guide the agent governs? I have said that the
agent’s practical standpoint serves as an anchor for deliberation, and so
guidance by this standpoint is a form of governance. But I have also alluded
to other roles of that standpoint, roles that help support the idea that when
it guides the agent governs. What roles are these?
Here my proposal is to embed the role of anchor in deliberation within
the overall role of knitting together in distinctive ways the agent’s practical
thought and action, both synchronically and diachronically. What ways?
Well, what we learn from broadly Lockean approaches to personal identity
over time is that there is a close connection between such personal iden-
tity—or, anyway, “what matters” in such identity—and overlapping
strands of continuities of attitudes at different times, and cross-reference
among attitudes at different times. If certain attitudes that anchor delibera-
tive thought had the role of knitting together thought and action in these
broadly Lockean ways—in ways that essentially involve such diachronic
continuities and cross-references—these attitudes would organize that
agent’s life over time and thereby help constitute the cross-temporal iden-
tity56 of the agent whose life thereby has this organization. And my
proposal is that it is in playing this role of anchoring deliberation in ways
that constitute and support such broadly Lockean cross-temporal
organization of one’s thought and action that, in the absence of relevant
conflicts,57 an attitude speaks for that agent.58
And now the point is that plan-like commitments to weights do indeed
realize these “design specifications”: they anchor deliberation in ways that
involve these broadly Lockean roles in the diachronic organization of the
agent’s practical life. This follows from the content of these commit-
ments—namely, what is to have weight in deliberation—together with
the general theory of the nature of plan states. Given their distinctive
contents these plan-like commitments can provide premises about weights,
premises that shape deliberation. And plan states organize our thought and
action over time in Lockean ways. So there is a case for the idea that these
plan-like commitments to weights, in the absence of relevant conflicts, do
indeed speak for the agent. When plan-like commitments to weights
function properly, they weave together the agent’s practical thought and

55 As I once put it, why do these commitments have “agential authority”? See
Bratman 2007d.
56 Or anyway, “what matters” in such identity.
57 This appeal to the absence of relevant conflict draws on Harry Frankfurt’s idea of
satisfaction in Frankfurt 1999b; and See Bratman 2007c, esp. 34–5, 44.
58 A related idea is in Yaffe 2000, ch. 3.
64 Michael E. Bratman

action in ways that, in the absence of relevant conflict, help constitute

where the agent stands: their guidance is the agent’s self-governance.
Granted, it might be that certain kinds of evaluative judgment also realize
these design specifications.59 But not all evaluative judgments do this; not
only evaluative judgments do this; there are systematic ways in which a
person’s evaluative judgments will tend to underdetermine the practical
commitments that do this; and these practical commitments are in
important cases not intersubjectively accountable in the way in which
evaluative judgment is. So it is reasonable to see plan-like commitments to
weights as central to our nonhomuncular model of human self-governance.

7. WHY PLAN RATIONALITY MATTERS

Return now to the norms of individual plan rationality. Let’s focus on the
synchronic norms of consistency, agglomeration, and means-end coher-
ence. And here reflection on the fecundity of planning agency gives us
something to say about the force of these norms. After all, accepting and
being guided by these norms is, on the theory, part and parcel of being a
planning agent. Planning agency is fecund in the sense that it helps support,
inter alia, important forms of temporally extended, shared, and self-
governed agency. And, I take it that we sensibly value these forms of agency:
they significantly enrich our lives. This means that we have good reason to
value being a planning agent, and so to value our continued acceptance of
these norms. Giving up these planning structures, if we could, would
cascade through our lives and come at a high price.
This is a broadly pragmatic rationale for continuing to be a planning
agent, and so for continuing to accept these synchronic norms—in a sense
of “pragmatic” that includes both instrumental and noninstrumental rela-
tions between such planning agency and what we sensibly value. This
pragmatic support for our continued acceptance of these planning norms
does draw on a form of instrumental reasoning; but it need not draw
specifically on the norm of means-end coherence of intentions that is
among the norms it seeks to support. So there need be no circle.
There remains, however, a problem, one familiar from J. J. C. Smart’s
concerns about “rule worship.”60 Even if there are significant advantages
that accrue to the acceptance of these general synchronic norms, it seems
that there can still be particular occasions in which one does better—better
as judged by the very values to which our pragmatic argument appeals—by

59 Bratman 2007b, 249. 60 Smart 1967.

 The Fecundity of Planning Agency 65

violating one or more of these norms. Perhaps there are, on specific

occasions, unusual benefits of inconsistency and/or incoherence. Given
this possibility we cannot in general infer directly from the pragmatic
support for the acceptance of these norms to normative support for
conformity in the particular case.61 So we are still without an explanation
of the normative force of these norms in each particular case, an
explanation to which a reflective planning agent can appeal to support
her conformity to these norms in her particular case and to rebut the charge
of unjustified consistency or coherence “worship.”
One response to this is to see the idea that these norms in general have
force in the particular case as a “myth.”62 In contrast, I think that our
reflections on the relation between planning agency and self-governance
point to a less skeptical view. Recall that a self-governing planning agent
will have a plan-structured practical standpoint that speaks for him and
whose relevant guidance is his self-governance. And note that for your plan-
structured standpoint to speak for you about a particular practical matter it
had better not be inconsistent or incoherent in relevant ways. After all, if
you intend E, know that this requires that you intend means M, and yet do
not intend M, you have no clear stance on E; there is, as Harry Frankfurt
might say, no place where you stand with respect to E.63 And similarly if
you intend E but also intend F while knowing that E and F are not co-
possible. In such cases you might manage to act intentionally and for a
reason; but you have no clear standpoint concerning E whose guidance of
your conduct can constitute your self-governance with respect to E. In this
way the violation of these consistency and/or coherence norms precludes
relevant self-governance for planning agents like us.
This is a point about the metaphysics of self-governance in the case of
planning agents; it is not a point that depends on our having already
established that conformity to these norms has normative force. So we
can use this point about the metaphysics of self-governance to explain an
aspect of this normative force. And we can do that if we suppose that a
planning agent has a normative reason in favor of her self-governance in
each particular context of intentional agency.64
We begin with the point, just noted, about the role of plan consistency
and coherence in the metaphysics of self-governing planning agents. We
then appeal to a principle that if there is normative reason for S to achieve

61 So we cannot take for granted what Michael Thompson calls a “general-to-

particular transfer principle.” (2008), 171.
62 Raz,2005, Kolodny, 2008.
63 Frankfurt,1988b, 166.
64 Though we may need to qualify this supposition for contexts of trivial decision.
66 Michael E. Bratman

E, and if S has the capacity to achieve E, and if X is constitutively necessary

for E, then there is normative reason in favor of X.65 We then infer, given
our assumption of a reason for self-governance, that if you are a planning
agent who has the capacity for relevant self-governance then you have a
reason of self-governance to conform to the planning norms of consistency
and coherence in this particular case.66
In partial summary: The fecundity of planning agency leads to a prag-
matic rationale for continuing to be a planning agent, one who accepts the
cited norms. Given that we are—as we have pragmatic reason to be—
planning agents, our self-governance involves, as a necessary constitutive
element, relevant plan consistency and coherence. So, on the assumption
that we have reason to govern our lives in the particular case, and that this
self-governance is in our power, we have a reason of self-governance to
conform to these synchronic norms in the particular case, a reason over and
above the specific reasons we have for the specific options in play. In this
sense these capacities of planning agency are self-reinforcing: if we do have
these planning capacities (capacities supported by a pragmatic rationale),
and if we do have the capacity for relevant self-governance, then we thereby
have a distinctive reason of self-governance to conform in the particular
case to the synchronic norms whose acceptance is an element in these very
planning capacities.67
In this way a reason for self-governance, taken together with the meta-
physics of self-governing planning agency, supports a tendency toward
equilibrium between one’s pragmatically grounded acceptance of the
cited planning norms and one’s reasons for conformity to those norms in
the particular case. And my tentative conjecture is that this combination of
pragmatic support, self-reinforcement, and tendency toward normative
equilibrium lends further support to the idea that these norms are, indeed,
norms of practical rationality for planning agents. In these multiple ways,
the fecundity of our planning agency helps us understand the complex
normative significance of norms the acceptance of which is at the heart of
that planning agency.68

65 This is not to say that ought works this way. Nevertheless, there are hard issues that
I cannot discuss here about this proposed principle.
66 Bratman 2009b. Compare David Copp’s thought that “rationality is in the service
of self-government” (Copp 2007, 351), and Kenneth Stalzer’s thought that means-end
incoherence is a failure of “self-fidelity.” (Stalzer 2004, ch. 5).
67 I discuss a related idea of self-reinforcement, in the context of a discussion of views
of David Gauthier, in Bratman forthcoming b.
68 Thanks to the participants at the 2011 New Orleans Workshop on Agency and
Responsibility. Special thanks to David Shoemaker both for his deft organization of the
workshop and for his helpful comments on an earlier draft.
 The Fecundity of Planning Agency 67

REFERENCES

Arpaly, Nomy and Schroeder, Timothy (1999). “Praise, Blame and the Whole
Self,” Philosophical Studies 93, 161–88.
Bennett, Jonathan (1974). “The Conscience of Huckleberry Finn.” Philosophy 49,
123–34.
Bratman, Michael E. (1987). Intention, Plans, and Practical Reason (Cambridge,
MA: Harvard University Press; reissued 1999 by CSLI Publications).
—— (1999a). Faces of Intention (New York: Cambridge University Press).
—— (1999b). “I Intend that We J.” In Bratman, 1999a, 142–61.
—— (2007a). Structures of Agency: Essays (New York: Oxford University Press).
—— (2007b). “Three Theories of Self-Governance.” In Bratman, 2007a,. 222–53.
—— (2007c). “Reflection, Planning, and Temporally Extended Agency.” In
Bratman, 2007a, 21–46.
—— (2007d). “Two Problems About Human Agency.” In Bratman, 2007a,.
89–105.
—— (2007e). “A Desire of One’s Own.” In Bratman, 2007a, 137–61.
—— (2007f). “Valuing and the Will.” In Bratman, 2007a, 47–67.
—— (2009a). “Intention, Belief, Practical, Theoretical.” In Robertson (ed.), 2009,
29–61.
—— (2009b). “Intention, Practical Rationality, and Self-Governance.” Ethics 119,
411–43.
—— (2009c). “Modest Sociality and the Distinctiveness of Intention.” Philosophical
Studies 144, 149–65.
—— (2010). “Agency, Time, and Sociality,” Proceedings and Addresses of the
American Philosophical Association 84, 7–26.
—— (2012). “Time, Rationality, and Self-Governance,” Philosophical Issues
(Action Theory) 22.
—— (Forthcoming a). Shared Agency (New York: Oxford University Press).
—— (Forthcoming b). “The Interplay of Intention and Reason,” Ethics.
Broome, John (2007). “Is Rationality Normative?” Disputatio 2, 161–78.
Cohen, Philip R., Morgan, Jerry, and Pollack, Martha E. (eds.) (1990). Intentions in
Communication (Cambridge, MA: MIT Press).
Copp, David (2007). Morality in a Natural World (New York: Cambridge Univer-
sity Press).
Davis, Wayne (1984). “A Causal Theory of Intending.” American Philosophical
Quarterly 21, 43–54.
Driver, Julia (2001). Uneasy Virtue (New York: Cambridge University Press).
Ferrero, Luca (2009). “What Good is a Diachronic Will?” Philosophical Studies 144,
403–30.
Foot, Philippa (ed.) (1967). Theories of Ethics (New York: Oxford University Press).
Frankfurt, Harry G. (1988a). The Importance of What We Care About (New York:
Cambridge University Press).
68 Michael E. Bratman

—— (1988b). “Identification and Wholeheartedness.” In Frankfurt, 1988a,

159–76.
—— (1999a). Necessity, Volition, and Love (New York: Cambridge University
Press).
—— (1999b). “The Faintest Passion.” In Frankfurt, 1999a, 95–107.
—— (2006). Taking Ourselves Seriously and Getting It Right (Stanford, CA:
Stanford University Press).
Gibbard, Allan (1990). Wise Choices, Apt Feelings (Cambridge, MA: Harvard
University Press).
—— (2003). Thinking How to Live (Cambridge, MA: Harvard University Press).
Gilbert, Margaret (1990). “Walking Together: A Paradigmatic Social Phenom-
enon.” Midwest Studies in Philosophy 15, 101–14.
—— (2009). “Shared Intention and Personal Intentions.” Philosophical Studies
144, 167–87.
Grice, Paul (1974–5). “Method in Philosophical Psychology (From the Banal to the
Bizarre).” Proceedings and Addresses of the American Philosophical Association 48,
23–53.
Harman, Gilbert (2000a). Explaining Value and Other Essays in Moral Philosophy
(New York: Oxford University Press).
—— (2000b). “Desired Desires.” In Harman, 2000a, 117–36.
Holton, Richard (2009). Willing, Wanting, Waiting (New York: Oxford University
Press).
Kahneman, Daniel, Slovic, Paul, and Tversky, Amos (eds.) (1982). Judgments
Under Uncertainty: Heuristics and Biases (Cambridge: Cambridge University
Press).
Kaufmann, Walter (ed.) (1975). Existentialism: from Dostoevsky to Sartre (New
York: Meridian/Penguin).
Kolodny, Niko (2008). “The Myth of Practical Consistency.” European Journal of
Philosophy 16, 366–402.
—— (2009). “Reply to Bridges.” Mind 118, 369–76.
Leist, Anton (ed.) (2007). Action in Context (Berlin: de Gruyter/Mouton).
Morton, Jennifer (2011). “Towards an Ecological Theory of the Norms of Practical
Deliberation.” European Journal of Philosophy 19, 561–84.
Parfit, Derek. 1984. Reasons and Persons (New York: Oxford University Press).
Railton, Peter (2006). “Normative Guidance.” In Shafer-Landau, 2006, 3–33.
Raz, Joseph (2005). “The Myth of Instrumental Rationality,” Journal of Ethics and
Social Philosophy 1, 1–28.
Robertson, Simon (ed.) (2009). Spheres of Reason: New Essays on the Philosophy of
Normativity (New York: Oxford University Press).
Sartre, Jean Paul (1975). “Existentialism is a Humanism.” In Kaufmann (ed.)
(1975), 345–69).
Schroeder, Timothy (2003). “Donald Davidson’s Theory of Mind is Non-Normative.”
Philosophers’ Imprint 3, 1–14.
Searle, John R. (1990). “Collective Intentions and Actions.” In Cohen, Morgan,
and Pollack (eds.) (1990), 401–15.
 The Fecundity of Planning Agency 69

Setiya, Kieran (Forthcoming). “Intentions, Plans, and Ethical Rationalism.” In

Vargas and Yaffe (eds.) (forthcoming).
Shafer-Landau, Russ (ed.) (2006). Oxford Studies in Meta-Ethics 1.
Shapiro, Scott (2011). Legality (Cambridge, MA: Harvard University Press).
—— (Forthcoming). “Massively Shared Agency.” In Vargas and Yaffe (eds.)
(forthcoming).
Simon, Herbert (1983). Reason in Human Affairs (Stanford, CA: Stanford Univer-
sity Press).
Smart, J. J. C. (1967). “Extreme and Restricted Utilitarianism.” In Foot (ed.)
(1967), 171–83).
Stalzer, Kenneth (2004). On the Normativity of the Instrumental Principle
(Ph.D. Thesis, Stanford University).
Thompson, Michael (2008). Life and Action (Cambridge, MA: Harvard University
Press).
Tomasello, Michael (2009). Why We Cooperate (Cambridge, MA: MIT Press).
Vargas, Manuel and Yaffe, Gideon (eds.) (Forthcoming). Rational and Social
Agency: Essays on the Philosophy of Michael Bratman (New York: Oxford Univer-
sity Press).
Velleman, J. David (1992). “What Happens When Someone Acts?” Mind 101,
461–81.
—— (2007). “What Good is a Will?” In Leist (ed.) (2007), 193–215.
Yaffe, Gideon (2000). Liberty Worth the Name: Locke on Free Agency (Princeton, NJ:
Princeton University Press).
 3
Can I Only Intend My Own Actions?
Intentions and the Own Action Condition

Luca Ferrero

1. INTRODUCTION

1.1
The possible objects of one’s desires and wishes seem to be virtually
unlimited: one might desire unattainable states of affairs including,
perhaps, even known logical impossibilities. By contrast, the proper
objects of one’s intentions appear to be much more limited. As Baier
(1976: 214) writes: “My intentions must not only be ‘made’ by me, when
I make up my mind, they must be directed upon, they must concern,
my own future actions.” It seems that the agent can only intend to do
something herself : intentions appear to be necessarily de actu suo (Wilson
1989).
This restriction on the proper object of intentions is reflected in the
grammar of attributions of intention: an agent is usually said “to intend to
do such-and-such.” The syntactical complement of attributions of
intention is usually an infinitival verb phrase whose subject is implicitly
understood to be the very agent of that intention. Although it is not
ungrammatical to say that an agent S intends that p be the case, it seems
that the propositional complement “that p” should be interpreted as
making implicit reference to the agent’s own action. That is, “S intends
that p” is elliptical for “S intends to do what it takes for her to bring about
that p.”
This restriction on the admissible objects of intentions—which some-
times goes under the name of the “own action condition”—suggests
 Can I Only Intend My Own Actions? 71

that there is a special relationship between an agent, her intentions, and her
own actions.1 The own action condition (OAC, hereafter) might strike the
reader as trivial, in the sense of both obviously true and uninteresting.
Nonetheless, this thesis has not gone utterly uncontested.2 In addition,
many philosophers claim that the status of OAC bears on important issues
in the philosophy of agency. For instance, according to both Thompson
(2008) and Setiya (2011) a proper characterization of the relation between
intentions and actions depends on the fate of OAC. Boyle and Lavin
(2010) defend OAC as central to their rejection of causal theories of
action and their defense of the “guise of the good” thesis. Finally, OAC
matters for the characterization of shared intentions in joint agency (See
Bratman 1997 and Stoutland 2002).
Hence, in spite of its apparent triviality, OAC is worth a closer look. In
fact, in this paper I will argue against it: the genuine object of intentions is
neither necessarily nor primarily restricted to the agent’s own actions.
Although the agent’s own agency plays a necessary role in carrying out her
intentions, this role is not reflected in the restriction of the intention’s objects
to her own actions. The object of an intention is not necessarily cast in the
infinitival/agential form—“I intend to do so-and-so.” The more inclusive
propositional clause—“I intend that such-and-such”—is actually the proper
characterization of the logical form of intentions. Or so I will argue.3

1.2
I will begin with some preliminary considerations on the notion of the
object of an attitude in general (Section 2). I will then discuss how to
characterize the content of a simpler kind of practical attitude, which
I call “aiming.” I will argue that OAC does not hold of aimings (Section 3).
Then I will move to intentions and show that the features that make them
different from aimings are not sufficient to support the application of OAC
to intentions (Section 4). I will then consider whether one’s own actions
might still be taken to be the standard although not necessary object of

1
OAC is defended by Baier (1970: 650), Meiland (1970), Castañeda (1972: 140),
Castañeda (1975: 25, 169–75), Baier (1976: 214), Searle (1983: 100, 105), Gustafson
(1986: 104, 121, 204), Wilson (1989), Perloff (1991: 403), Velleman (1997), Stoutland
(2002), Thompson (2008: 120–3, 127–8, 130–1), Moran and Stone (2009: 143, 147).
2
See Bratman (1997), Tuomela (2005), Setiya (2011).
3
In this paper, I use “propositional” as an umbrella term that covers various possible
characterizations of content—whether in terms of propositions, sentences, states of
affairs, and the like. For present purposes, what matters is only the contrast between
the characterization in broadly conceived propositional terms and the one in the agential/
infinitival form.
72 Luca Ferrero

intention. I will argue that although one’s own actions might play a
prominent role they do not do so as distinct pieces of conduct and separate
targets of one’s intentions (Section 5). Finally, I will consider the different
degrees to which one’s own agency might be involved in intentions. I will
claim that one’s own agency is necessarily involved but in a way that is
formal and generic, and as such lends no support to OAC (Section 6).
Space constraints prevent me from discussing how my conclusions bear on
the various philosophical disputes that invoke the fate of OAC but for the
discussion of the distinction between intending and acting. I will argue that
my rejection of OAC puts pressure on the tenability of a stark distinction
between intending and acting. This is a somewhat surprising development
of the standard dialectic, since this substantive claim is usually supported
on the basis of the acceptance of OAC (5.8).

2. THE OBJECT OF AN ATTITUDE

2.1
Talk of the “object” or “content” of an intention is not as straightforward as
it might appear. Grammatically speaking, the object is the syntactic com-
plement of expressions of intention, but what we are interested in here is
rather the object as the complement of the logical form of intentions.
Before investigating this logical form, however, we need to make some
preliminary considerations about the notion of the content of attitudes in
general and various notions of the conditions of satisfaction or success of an
attitude.
Let’s begin with the simpler case of belief. Take the belief that p. The
content of this belief—the proposition p—individuates the attitude. It
differentiates this particular belief from other beliefs, such as the belief
that q. (This individuating content is also something that can be “shared”
with attitudes of a different kind, for instance, the desire that p, the
assumption that p, etc.) The individuating content of a belief also seems
to indicate its conditions of success: the belief that p succeeds as a belief when
p obtains.
At first, it seems that the same holds true of practical attitudes. Consider
the desire that p. The proposition p individuates it by differentiating from
other particular desires. In addition, a desire that p is satisfied—succeeds as
a desire—only when p obtains. Likewise for intentions. The object of the
intention (whether characterized in infinitival or propositional terms)
seems to both individuate a particular intention and indicate its conditions
of success.
 Can I Only Intend My Own Actions? 73

These considerations, however, are too hasty since they do not pay
attention to the different senses in which an attitude can be successful.
First, an attitude is constitutively successful when it meets the standards
that govern the acquisition, retention, and abandonment of attitudes of its
kind. For instance, beliefs are regulated by the standards of veridicality: the
belief that p is constitutively successful when p is true. This explains why
the obtaining of the conditions of individuation of a belief coincides with
the obtaining of its conditions of “constitutive” success.
This coincidence, however, might not hold of other attitudes: conditions
of constitutive success need not be the same as the attitude’s individuating
content. For instance, assumptions are not regulated by veridicality: the
assumption that p is individuated by p but, unlike the belief that p, it might
be constitutively successful even if p is false.
Something similar holds of desires. The constitutive standards of desires
are those of “desirability”: one is supposed to acquire, retain, and abandon a
desire in light of the “desirability” of its content—whatever that turns out
to be according to the substantive accounts of the nature of desires. The
desire that p can be constitutively successful even if p does not obtain and
never will. The obtaining of p, however, matters for a different kind of
success, the desire’s fulfillment or satisfaction.

2.2
The satisfaction of a desire is a form of success distinctive of practical attitudes,
but it is not to be confused with another kind of practical success, the
“achievement” as the success of executive attitudes. An executive attitude, such
as an intention, is an attitude that moves the agent toward a goal. When the
agent has an executive attitude aimed at the state of affairs g, the mere obtaining
of g is not sufficient for the attitude’s executive success. The obtaining of g does
not count as an achievement if g comes about independently of the attitude or
via its deviant operation (although the agent’s non-executive desire that g,
which might accompany the executive attitude toward g, can still be success-
ful—in the sense of “satisfied”—no matter how g comes about).
Thus, for executive attitudes like intentions, there are three kinds of
conditions: (i) of individuation (formulated in terms of the goal g ), (ii) of
constitutive success (to be met in order for the agent to be correct in having
that attitude), (iii) of executive success (the non-deviant obtaining of g by
way of the proper operation of the executive attitude). The distinction is
not simply between these notions but also between the substantive condi-
tions: the goal g might individuate the attitude, but its obtaining is neither
necessary for constitutive success nor sufficient for the executive one.
74 Luca Ferrero

The difference between these conditions matters for the attitude’s logical
form. Talk of its “object” should be cast in terms of individuating content.
The conditions of individuation are those that matter for the discussion of
restrictions on the proper objects of an attitude. Considerations about
constitutive and executive success pertain instead to discussions about
kinds of attitudes. They matter for the understanding of the distinctive
operation of attitude-types—their distinctive mode or force—rather than
the content of their tokens.
It can be quite tempting to conflate the different conditions, especially
when talking about the objects of attitudes. Particularly dangerous is the
nowadays common philosophical talk of the attitudes’ aims. This expres-
sion when properly used indicates what regulates attitudes as a matter of
their constitutive conditions (for instance, to say that beliefs aim at truth is
to single out veridicality as their constitutive standard). This “constitutive
aim,” however, should not be confused with the “individuating aim” of
particular executive attitudes, that is, with the particular goal g that the
agent has when she is set on a particular pursuit. Only executive attitudes
have individuating aims, since having aims in this sense is what makes them
executive. (Additionally, executive attitudes are also under conditions of
executive success and, as a result, there might be a sense in which they “aim”
at achievement but the target of this aiming is not to be confused with the
substantive goal that provides their individuating content—see 5.7.)

3. AIMING

3.1
Before discussing whether OAC holds of intentions, let’s consider whether
it is true of a simpler kind of executive attitude, which I call “aiming.”
Aiming is the distinctive attitude of the basic form of representational
agency. When an agent aims at a state of affairs g, she has g as her goal,
that is, she is set on making g true in light of her representation of it.
The pursuit of g by aiming at it consists of the combined exercise of
several executive powers.
1. Power of self-motion. The subject as a whole is the source of its conduct
(the conduct is not the mere product of external forces or uncoordinated
operations of the subject’s subsystems).4

4
See Frankfurt (1978), Burge (2009).
 Can I Only Intend My Own Actions? 75

2. Power to represent the goal and orient one’s conduct in view of it.5
3. Power to respond to interferences by (i) adjusting conduct in the face of
perturbations and (ii) persisting in the pursuit in the face of some
setbacks.
4. Practical ingenuity and opportunism: the ability to take advantage of
favorable conditions, including: (a) refraining from interfering with
advantageous courses of events, (b) reliance on other agents, (c) explor-
ation of novel ways to progress toward the goal.
5. Two-way powers. The exercise of executive powers might not be merely
reactive : the subject might not simply react automatically and in fixed
ways. It might respond to circumstances in light of its perception of or
belief about the fit between its response and the circumstances, rather
than being simply triggered by them (See Kenny 1992: 70).
Some of these powers might be exercised in isolation from each other and,
as such, underpin even simpler forms of agency. In addition, agents might
have these powers in domain-specific forms and exert them with different
degrees (if any) of conceptual sophistication and reflection. Nonetheless,
I contend that some combination of these powers, even if in the absence of
conceptual sophistication and reflection, is constitutive of basic goal-
directed agency, of the agency exhibited in “aiming.”

3.2
A particular aiming is individuated in terms of its goal g, the state of
affairs toward which the aiming is executively directed. When g obtains
nondeviantly by way of the combined operation of executive powers
constitutive of aiming, the aiming at g is executively successful—it achieves g.
The conditions of executive success are not part of the individuating
content, of the goal g: when one aims at g, one is orienting the executive
capacities toward making the states of affairs g obtain. One is not orienting
them toward “the bringing about of g in a nondeviant manner and by way
of this very aiming.” The nondeviant bringing about of g is not the
individuating goal of the attitude. Rather it spells out what aiming consists
in as an attitude kind. Hence, it needs to be spelled out only when
articulating the characteristic force of aiming as a kind, rather than the
distinction between token aimings.

5
The orientation might be minimal: it requires neither a plan toward g, nor the
representation of g as a goal. At bottom, it amounts to the capacity of taking instrumental
steps and registering one’s progress toward g (at least, registering when g obtains so as to
stop its pursuit).
76 Luca Ferrero

These considerations suggest that the following is the canonical logical

form of aiming:
(A) S aims at g
where g stands for the state of affairs whose obtaining constitutes the aim’s
achievement—g is the goal of this executive attitude. The “object” of the
aiming is specified in (broadly conceived) propositional terms, not in
agential/infinitival ones.

3.3
Limiting the content of the simpler form of goal-directed agency to one’s
own actions is overly restrictive. Although it is possible to aim just at one’s
own actions, this restriction does not seem supported by reflection on the
operation of goal-directed agency. Executive capacities can be oriented
toward goals quite remote from the agent’s own actions, that is, from the
agent’s exercises of these capacities. As long as the cognitive and concep-
tual abilities allow for it, the agent’s goals can extend as far as the remote
consequences of her actions, including the results of her omissions, the
actions of other agents, and the effects of other agents’ actions. That a
state of affairs g might obtain either by omission or the mediation of
others’ actions does not make it illegitimate as a goal.
A state of affairs qualifies as a goal as long as it can provide suitable
orientation for the exercise of executive capacities. There might be limita-
tions on how remote the object might be to qualify as a goal rather than the
object of a wish. But this is a concern with the upper boundary of aiming’s
possible objects, which puts no pressure on restricting acceptable goals to
the lower boundary of the agent’s own actions.
These considerations rule out OAC for aimings. If, as I maintain, talk of
the object of an attitude should be interpreted in terms of individuating
content, and an aiming’s individuating content is its goal, then the object of
aimings cannot be restricted to the agent’s own actions.
One might try to reject this conclusion by interpreting talk of the aiming’s
“object” in terms of what is especially relevant in the attitude’s psychological
operation rather than in its individuating content. For instance, one might
argue that the main focus of the exercise of one’s executive capacities in the
aiming at g is not the (explicit) representation of g but of the instrumental
actions to be taken toward it. And one might continue with the claim that the
object of an attitude is to be equated with the content of this focus. But this
suggestion is problematic. It is not sufficiently systematic and comprehensive.
First, the focus of the psychological operation of aimings might vary widely;
second, oftentimes the remote goal rather than the means seems to be at the
 Can I Only Intend My Own Actions? 77

center of one’s attention (this is especially so, I surmise, for the simpler
executive attitudes like aimings).6 In addition, even if the goal is not the
focus of orientation, it is hard to deny that it plays some role in the actual
psychological operation of aimings.

3.4
To reject OAC for aimings is not to deny that they display important de se
features: these features can be found both (i) in the ownership and exercise
of executive capacities, and (ii) in the path toward the achievement of g.
First, the immediate exercise of the executive powers that underpins the
agent’s aiming cannot but be the agent’s own. For it is constitutive of the
subject’s identity as an agent that she is the locus of this immediate exercise.
This is an instance of a more general phenomenon: the fundamental de se
involvement that characterizes the existence and identity of subjects as
separate loci of individual psychologies. A subject is none other than the
locus of the unmediated exercise of psychological powers, including the
executive ones (See Burge 2000).
Second, the path toward achievement originates in the agent. Execution
originates in the agent and proceeds from her: from her specific location
in space and time and from her point of view on her surroundings. In
addition, execution is subjected to the limitations imposed by the agent’s
present executive powers and circumstances. In this sense, execution is
always ego-centered and perspectival.
These features do not entail that the object of one’s aiming should be
restricted to the exercise of one’s executive capacities. Execution is neces-
sarily ab se, but its object/target need not be de se.
The psychological work is, ultimately, always of one’s own. For the
subject of this working is nothing other than the locus of the unmediated
psychological operations. This is not to say that one’s executive attitudes
are either exclusively or primarily oriented toward these operations or
their immediate effects. The distal and outward-looking orientation of
aiming is rather a manifestation of the proper operation of goal-directed
agency.
The ab se character of execution does not bear directly on the object of
executive attitudes, but it matters for determining the conditions of both
executive and constitutive success. First, executive success can only be
secured via the initial and unmediated exercise of one’s executive capacities.

6
The focus of the psychological operation on instrumental steps might affect the
content of intentions as more complex executive attitudes; see plans-as-recipes in 4.5.
78 Luca Ferrero

It must originate in the agent. Second, constitutive success is related to what

counts as an acceptable goal, which is partly a matter of what is attainable
by the agent ab se from her present circumstances.
The ab se character of execution is undeniable, but it does not support
OAC. For it is the de se aspect of agency in its source rather than in its
target. Although any achievement has to go through the agent’s contribu-
tion via the exercise of her executive powers, her targets can be at much
remove from that origin.
A state of affairs might not count as a genuine goal if the connection
between its eventual occurrence and the exercise of executive capacities is
excessively remote—too thin and fragile. But a concern about excessive
distance gives no reason to push the objects of executive attitudes all the
way back to the immediate exercises of executive powers.7
The relation between the exercise of executive powers and the obtaining
of the goal g can be quite indirect. For g might just be a remote and indirect
effect (including via the intermediation of other agents) of processes initi-
ated by these exercises. The ability to aim at such distal states of affairs is
one of the remarkable products of practical ingenuity and opportunism,
when coupled with the agent’s predictive ability.
Opportunistic capacities and the two-way volitional powers also make
it possible to pursue goals by relying on already favorable circumstances.
Agents often make progress toward their goal simply by refraining from
“antagonistic” interventions in the natural course of events. These nonanta-
gonistic omissions can contribute to genuine achievement as much as
acts of commission. In the limiting case, an agent might achieve g simply
by monitoring the unfolding of a favorable course of events. She is to
monitor it with an eye toward making possible corrective interventions,
which she is able and ready to perform but might not to be required.8 This
nonantagonistic attainment is still a bona fide achievement, i.e. an executive
success.

3.5
To sum up: the discussion of aiming shows that there are fundamental de se
elements in the location and working of executive attitudes. But these elem-
ents do not necessarily make the content of these attitudes to be exclusively
or primarily about one’s own actions: the admissible goals of aimings are

7
My claims concern the logical form not the metaphysics of executive attitudes. It might
well be that possessing an executive attitude amounts to nothing more than a disposition to
or the actual exercise of certain immediate executive capacities.
8
Compare the case of the driver going downhill in Frankfurt (1988).
 Can I Only Intend My Own Actions? 79

not limited to the immediate exercises of one’s executive powers or their

proximal outcomes. OAC is not true of aimings.

4. INTENTION

4.1
Let’s return to intentions. Intentions are executive attitudes. They are
considerably more complex than aimings, but does this difference affect
their logical form and support OAC?
Intentions differ from aimings in four ways:
1. Intentions are under distinctive rational pressures for stability and
agglomeration, whereas aimings are not rationally criticizable when
unstable or unagglomerable (See Bratman 1987).
2. Intentions extend the temporal reach of agency, by allowing the pursuit
of very distal goals and engagement in temporally extended and
“unified” activities (See Ferrero 2009b).
3. Full-fledged intentional agency usually goes together with more sophis-
ticated conceptual capacities. It is the agency characteristic of agents
who are (a) able to articulate their goals and their plans to reach them,
(b) in the business of both offering and asking for folk-psychological
explanations, and (c) capable of at least some reflection about their
agency and psychology.
4. Intentions are planning attitudes (Bratman 1987): (a) they frame further
deliberation and help coordinate conduct over time (in part by fixing
expectations about future conduct); (b) they often come with plans as
(partial and hierarchical) recipes that list some of the steps to be taken
toward one’s goal.
None of these distinctive features of intentions support OAC. This is easy to
show for the first two features. First, the rational pressures on the intention
do not concern the form of their individuating content. Likewise for the
temporal reach of agency. It does not affect the content’s form, although it
allows for the pursuit of more complex and distal substantive goals.

4.2
Perhaps, one might devise a way to support OAC starting from the de se
character of the alleged self-referential nature of intentions, which relates to
the third feature of intentions, their alleged conceptual sophistication.
80 Luca Ferrero

Some philosophers maintain that the content of an intention includes the

role that the intention plays in securing its executive success.9 If so, its
logical form is something like:
(I-sr) S intends that: this very intention nondeviantly
results in the obtaining of g.
The content of (I-sr) has a necessary although indirect de se element. For the
intention included in the content is none other than the agent’s own
intention (that is, “S intends that: this very intention of one’s own non-
deviantly results in the obtaining of g”). But this de se element does not
support OAC. For it does not bear on the characterization of what the
intention is supposed to result in. One needs a separate argument to prove
that g is to be restricted to the agent’s own actions. The self-referentiality of
the intention only pertains to the origin of executive success. It spells out
the ab se character of achievement (see 3.4).
In any event, I find the self-referentiality of intentions problematic for
two reasons. First, it is based on a confusion between the individuating
content of a particular intention (which is not self-referential) with the
reflective articulation of the conditions of executive success of intentions
as a kind of attitude.10 Second, it is psychologically unrealistic insofar as
it demands that only agents with the capacity for entertaining self-referential
content can have genuine intentions.11
The last concern ultimately undercuts any strategy of attempting
to derive OAC from the conceptual sophistication of subjects capable
of intentional agency. Although intentions might be the characteristic
executive attitudes of reflective and conceptually sophisticated agents, it is

9
See Searle (1983: 85), Harman (1986: 86), Harman (1993: 141).
10
This confusion underlies the “content satisfaction view” used by Searle (1983) to
support self-referentiality. According to him, the content of an attitude should be
equated with its conditions of satisfaction. But as shown above (2.2), the idea of
satisfaction/success is ambiguous. The content satisfaction view seems to work for
some attitudes, such as beliefs and (nonexecutive) desires—the content of beliefs
happens to correspond to the conditions of their constitutive success; the content of
desires happens to correspond to the conditions of their success as satisfaction. But this
equivalence is accidental; there is no principled reason to think that it holds of attitudes
in general, let alone of executive attitudes given that the conditions of constitutive and
executive success of executive attitudes are much more complex than their individuating
content. (My point here is stronger than the one made by Mele (1987: 316–17), who
only claims that the content satisfaction view does not hold of intentions but might be
fine for beliefs and desires.)
11
See the criticisms of Mele (1992: 204–6), Kapitan (1995: 154, fn. 8), Roth (2000),
and Harman’s (1993: 145) acknowledgment of the problem. In addition, Kapitan
(1995: 163) correctly remarks that the efficacy of intending does not seem to depend
on the representation in its content of the conditions of its success.
 Can I Only Intend My Own Actions? 81

problematic to limit their possession to these agents. Conceptual and

reflective sophistication might help with shaping one’s agency by helping
in devising more complex goals and improving one’s rate of success thanks
to one’s understanding of how the intention contributes to the attainment
of one’s goals. But explicit understanding of executive success does not
change the logical form of intentions. In particular, it does not induce the
agent to take her own achievement of g as her goal: having the goal g already
equates with being set on achieving g, on obtaining it in the mode of
executive success (see 5.7).12

4.3
Consider now the last distinctive feature: the planning character of inten-
tions. Their role in framing and coordinating deliberation and conduct
does not concern their content. But the plan-as-recipe component does. An
agent might not simply intend to pursue a goal g but to pursue it by way of
a plan-as-recipe r ; that is, she might pursue g guided by a specification of
some of the steps that she expects to take toward it. The intention with a
plan-as-recipe plays a somewhat different role in organizing the agent’s
deliberation and conduct than an intention directed at the same goal
without a plan or via a different one. To this extent, the recipe is part of
the individuating content of the intention.
However, this qualification of the content does not support OAC. First,
the change does not restrict the goal, which still needs not to be formulated
in terms of one’s own actions. Second, a recipe need not be formulated in
terms of the agent’s own actions. Sometimes it only specifies intermediate
nonagential goals on the way to g.

4.4
Putting these considerations together, we might conclude that none of the
ways in which intentions differ from aimings make a difference to the

12
A conceptually sophisticated agent might have second-order intentions about the
efficacy of her first-order intentions. For instance, she might become aware of failures of
her executive capacities and address them by intending to improve her rate of executive
success. By engaging in the self-policing of her executive capacities, she acquires new
goals that make reference to the role of her intentions in securing executive success. But
she does not thereby change the content of her first-order intentions. In addition, the
second-order intentions about the success of one’s first-order intention do not have as a
goal the securing of their own conditions of executive success (if not in the indirect sense
in which they themselves might benefit from the self-policing of executive capacities that
the second-order intentions might put in place).
82 Luca Ferrero

structure of their content. Intentions are just a more psychologically and

normatively sophisticated form of aimings, but they retain the basic logical
form of the individuating content of aimings. Therefore, by analogy with
(A) in 3.2, I maintain that the canonical form of intentions is propositional
rather than agential:
(I) S intends that g
where g stands for a state of affairs in the role of the goal of intending as an
executive attitude.
Are there restrictions on the acceptable goals of intentions? The restrictions
parallel those imposed on the content of aimings. At most, there might be an
upper limit: the goal must be in principle attainable or, more restrictively,
its attainment should not be too remote from the exercise of the agent’s
executive powers (and the remoteness, like for aimings, allows for executive
success even when the agent relies on nonantagonistic waiting, monitoring,
and the intermediation of other agents, see 3.4). But this limit does not
impose the lowest possible restriction to the agent’s own actions.
Likewise for de se features. The executive powers whose exercise under-
pins intending are necessarily de se. The subjects of intentions are none
other than the loci of possession and exercise of these executive powers. But
this constitutive relation between agents and their executive powers only
makes the path to the intention’s success ab se rather than the intention’s
object de actu suo.

4.5
The only modification to the logical form warranted by the distinctive
features of intentions is the introduction of a recipe component. Some
intentions have the following form:
(I+r) S intends that: (by way of r) g
where r stands for a plan-as-recipe, the specification of some of the instru-
mental steps that the agent expects to take in the pursuit of g.13
The recipe’s instrumental character is indicated by the parenthesis.
Unless r is entirely composed of necessary means to g, it is possible
genuinely to achieve g even if the agent does not follow the recipe (not to
be confused with the accidental or deviant obtaining of g, which does not
count as an achievement). In this sense, the agent might be executively

13
For a similar suggestion about the role of plan components in the content of
intentions, see Mele (1987: 326).
 Can I Only Intend My Own Actions? 83

successful in carrying out the more generic intention (I) directed at the
same goal but without the specification of a plan of action. (For recipes that
are partly constitutive of the goal and incorporated into it, see 5.4.)

4.6
To sum up, in this section, I argued that there are no differences in logical
form and the de se involvement between simpler aimings and intentions.
Aimings and intentions share these features in virtue of their common
executive character. Their only differences concern (a) the rational norms
that govern them and the specific ways in which the executive powers are
called upon to meet these norms; (b) the possible presence of the recipe
component in intentions. None of these differences, however, support
OAC for intentions.

5. ACTION AS THE OBJECT OF INTENTION

5.1
Rejecting OAC does not imply that one’s own actions might not be the
object of one’s intentions. One might actually argue that they are the
paradigmatic objects at least of future-directed intentions, which ordinarily
seem directed at one’s own future actions. A future-directed intention to ç
at f seems to have as its genuine object the action of ç-ing, which is to be
initiated at the future time f and whose inception marks the transition
from the mere intending to ç to the actual ç-ing.
This reading might seem obvious at first, but it needs to be handled
carefully since it might induce a misleading picture of the relation between
intending and acting, a picture that exaggerates their differences by taking
the intention to be directed at the action as a truly distinct piece of conduct.
This distinction is perfectly in order when an agent intends that another
agent does something. For one intends that the other agent acquires a
separate intention and carries it out successfully as a matter of a separate
course of action. But as I am going to argue, a similar distinction is
problematic for ordinary intentions directed at one’s own actions.
In this section, I will discuss how best to understand the relation between
an intention to ç and one’s ç-ing. This discussion will reinforce my case
against the OAC but also offer a diagnosis of its intuitive appeal. In
addition, it will allow a more fine-grained characterization of the de se
elements of intention.
84 Luca Ferrero

5.2
As soon as one acquires the intention that g, one puts oneself under various
rational pressures, including the continuous demand to secure the possibil-
ity of eventual success and, when appropriate, to make actual progress
toward it. What is specifically required to meet these demands depends
on one’s particular circumstances (including one’s present and expected
skills, opportunities, and information). At times, one has to take specific
steps, including making particular bodily movements and using specific
tools; at other times, one might simply take advantage of favorable condi-
tions and let the natural course of events unfold unperturbed. At times, one
might have to engage in particular deliberations about implementation and
coordination; at other times, one might automatically implement prior
plans and policies, or let habits determine one’s conduct. Discharging the
rational pressures of intention is a matter of the agent’s continuous “intelli-
gent guidance” toward g, which requires a mix of antagonistic interven-
tions, nonantagonistic monitoring, and the management of attention and
deliberation. This mix of bodily and mental events is what I will call a
Course of Active Intelligent Guidance (CrAIG, henceforth) directed at g.
Responding to the rational pressures of the intention that g is thus a matter
of engaging in the appropriate CrAIG.
Although a CrAIG is produced by exercising one’s executive powers, not
all portions of the CrAIG need to be “actions” in the sense of antagonistic
bodily interventions. In the limiting case, an agent might secure that g
simply by monitoring the favorable unfolding of a natural course of events
that eventuates in g without requiring any antagonistic intervention. In
this case, there is no action in the narrow sense of some antagonistic
intervention, but the achievement is still the agent’s doing.

5.3
Imagine that I intend that I be at a party tonight but I don’t care about how
I get there (this includes not caring about getting there by antagonistic
interventions of mine; it would be fine, say, if someone were to take me
there while I am asleep). However, given my circumstances, it is reasonable
to expect that some antagonistic intervention of mine is required. Yet, I do
not need to commit to any specific implementation. My intention only
commits me to exercising my executive powers toward g, i.e. to engage in a
CrAIG directed at g—whatever shape this CrAIG might take in response to
the specific demands imposed by my present and future circumstances until
the goal is either achieved or abandoned.
 Can I Only Intend My Own Actions? 85

Imagine now that my goal is still simply to be at the party, but I also plan
on driving there as the most reasonable or likely means to my goal. Not
only does my intention include a recipe (my driving) but it might also be
expressed in its terms as “I intend to drive to the party,” even if the driving
is only instrumental to my goal (that is, “I intend that: (by way of my
driving) I be at the party”). The driving-as-recipe provides the default focus
for the organization of my executive capacities. As I embark in the CrAIG
directed at g, I exert these capacities toward g in large part by being oriented
toward my driving. As a result, I am going to engage in an episode of
“driving” as a course of action with a characteristic mix of antagonistic
interventions (including the use of various tools and the performance of
distinctive bodily movements), nonantagonistic monitoring, and correlated
management of deliberation and attention.
A plan-as-recipe offers a default yet revisable orientation for especially
salient albeit instrumental stages of an otherwise continuous CrAIG
directed at g. The CrAIG does not necessarily consist only of my driving,
nor does it necessarily start or end with my driving. My actual driving
should not be equated with the executive success of the intention. But my
starting to drive marks the point where the progress toward g begins to
unfold as “intended” in the sense of “according to the plan-as-recipe;” the
point where the course of active intelligent guidance directed at my being at
the party begins to take its expected shape.

5.4
There are cases where the recipe becomes part of the goal. For instance,
I intend to be at the party only by my driving (maybe, I want to impress the
partygoers by showing off my new car and I do not care to be at the party
otherwise). In these cases, the intention takes this form:
(I+gr) I intend that: g-by-way-of-r
Although g might be a goal by itself, here it is only a portion of the goal,
which now also includes some of the means to g—the recipe r. When so,
the inception of r marks a more momentous transition in the CrAIG:
when r begins, the achievement of the intention begins as well. From that
moment on, the intention is “in achievement,” so to say. But this
transition does not mark the acquisition of a new intention or the
inception of a novel CrAIG. It is not a transition between the intention
and a genuinely separate action. Rather, the transition is only the
beginning of the internal culmination of the original CrAIG, the incep-
tion of its finale.
86 Luca Ferrero

Oftentimes, the recipes that are part of the goal are what might be called
“performances”: specific and characteristic combinations of mostly antag-
onistic interventions in the form of certain bodily movements and the
distinctive uses of specific tools—for instance, playing a piano sonata or
dancing the tango. The beginning of the performance marks the inception
of the achievement but it is still a transition internal to the CrAIG that
began when the intention was first acquired (which might occur well in
advance of the performance’s inception).
Whenever the execution of a recipe begins (whether it is instrumental to
or constitutive of the goal), there is no break in the continuity of the agent’s
active guidance toward the goal. The execution of any of the actions
included in the recipe does not mark the termination of the intending
and the inception of a distinct piece of conduct, the acting. It is only an
internal transformation of the CrAIG’s shape, as required by the dynamics
of intentional progress toward g.
To claim that there is a single course of active intelligent guidance that
begins with the acquisition of the intention and continues until the inten-
tion is given up, voided, or carried out is not to deny that this course of
guidance has distinct stages. These stages might be quite different from
each other given that the specific demands on the agent’s executive capaci-
ties might vary widely as she progresses toward g. For instance, if she uses
her planning abilities well, earlier stages of a CrAIG tend to demand few, if
any, antagonistic interventions. At the outset, many well-planned projects
require minimal, if any, interference with the natural course of events. The
agent might only need to monitor for the persistence of currently favorable
circumstances. At the earlier stages, the specific effects of the intention tend
to be limited to the framing and organization of further practical deliber-
ation rather than antagonistic interventions in the outer world. Conversely,
later stages tend to keep the agent busier with antagonistic interventions
since more interference is usually needed at the later stages to secure that the
course of events eventuates in the intended state of affairs. But this is only a
simplified illustration of a typical but not necessary progression of the
demands imposed by an intention. The distribution of these demands
can differ widely (for instance, the agent might be kept busier earlier rather
than later, or cycle through the different stages).
The transitions between the stages can vary from the smooth and subtle
to the abrupt and sudden, depending on the changes in the demands
imposed by the circumstances. In any event, these transitions do not
mark a hard and fast metaphysical boundary between two utterly distinct
kinds of practical engagement with the world—intending vs. acting. When
describing the unfolding of an intentional pursuit, the distinction between
merely intending and actually acting is between two stages of an underlying
 Can I Only Intend My Own Actions? 87

unitary process of active intelligent guidance, stages whose boundaries need

not be hard and fast.14

5.5
The temptation to think of the acting stage as a distinct item to which the
intention is directed might arise from the possibility of engaging in the
same kind of conduct independently of the future-directed intention. It is
often possible to imagine circumstances where one might initiate the action
of ç-ing without any prior intention directed at it. But from this it does not
follow that in acquiring a prospective intention to ç at f one aims at
initiating a distinct piece of conduct in the future. The prospective inten-
tion and its correlated CrAIG are not successfully carried out by “passing
the baton” to another intention and its correlated CrAIG at f. What occurs
at f is rather a transition to a different stage within the same intention
and CrAIG. It is a matter of internal transformation dictated by the
dynamic unfolding of the pursuit of the goal g to which the ç-ing is directed.
When one acquires the prospective intention to ç at f one is—so to say—
“stretching over time” the pursuit of g by bringing its inception forward in
time. One is not preparing in advance for a distinct future undertaking.
Rather, one is advancing the time when one first acquires the goal g
and, thereby, puts oneself under the rational pressures distinctive of that
particular intentional pursuit. In turn, if one is rational, this is also the time
when one begins discharging these pressures by engaging in a CrAIG
directed at g.

5.6
We are now in a position to better appreciate what it means for an intention
to be directed at an action as a genuinely distinct item, as it happens when
one intends the action of another agent. The latter intention is directed at
the initiation of a separate course of active guidance. The target of my
intention that you ç is the success of your distinct CrAIG directed at your
goal. Your ç-ing is not a stage of my CrAIG. When an action is a genuinely
distinct goal, the inception of the action marks a break in the continuity of
intentional guidance rather than a transformation induced by its internal
dynamics.

14
One might think of the two stages of intending and acting as akin to phase-sortals,
and to their transitions as akin to metamorphosis (See McDowell (2010)).
88 Luca Ferrero

In the first-person case, the continuity of active guidance is what normally

precludes taking one’s own actions as genuinely distinct items targeted by
one’s intentions. For one’s ç-ing to be a genuinely distinct object of one’s
intention, either (i) one intends only to prepare for ç-ing without being
already committed to its actual pursuit, or (ii) one is alienated from one’s
future self and treats one’s future conduct third personally, i.e. as if it were of
another agent. Unless one is in either of these two scenarios, it is paradoxical
to intend to initiate an action of one’s own but not to be successful at carrying
it out. This is not problematic, instead, for the genuinely distinct action of
another agent (or of an alienated future self ).15 This is another way to show
the lack of separation between having an intention and the normal first-
personal engagement in the correlated CrAIG.
Under normal circumstances, when one’s own actions are presented,
both in thought and speech, as the objects of one’s intentions, they are not
the distinct targets of one’s intentional agency. Rather, they are just the
descriptions of the more specific shapes—that is, of the characteristic
patterns of bodily movement, tool-use, monitoring, attention manage-
ment, and appreciation of the situation—that the CrAIGs are supposed
to take at some crucial stages in their unfolding (whether as a matter of
instrumental recipes or of goal-constituting performances).

5.7
Let’s consider one last attempt at defending OAC. One might concede all
the points I made but claim that there is still a sense in which the object of
intention is necessarily one’s own action. The argument would revolve
around the necessary ab se character of executive success. As argued above
(3.4), executive success can be secured only via the agent’s exercise of her
own executive capacities: achievement is necessarily ab executione sua, so to
say. Hence, any achievement amounts to the culmination of an actual doing
on the agent’s part, that is, the culmination of a CrAIG directed at g (i.e. of
a sequence of exercises of the agent’s own executive capacities that eventu-
ates in the nondeviant obtaining of g ). Why can’t we claim that the object
of the agent’s intending is necessarily this CrAIG, which is by its very

15
There are some special cases in which one might genuinely intend to pursue a goal
g and yet hope, without any paradox, that one will never be in a position to actually
succeed at carrying it out. For instance, I am fully committed to go the hospital if I get a
life-threatening injury but also hope that I will never have to carry out this precautionary
conditional intention. If I can have some control on the antecedent, I might even intend
to avoid that it ever comes true (that I ever get a life-threatening injury), see Ferrero
(2009a).
 Can I Only Intend My Own Actions? 89

nature always of her own? In other words, according to this suggestion, the
object of intentions is necessarily the agent’s own doing, in the broad sense
of “doing” that encompasses the various modes of intelligent guidance.
The problem with this suggestion is that it indicates as the object a formal
and generic target. The course of action that leads to the achievement is not
something to which one can aim in the substantive and specific way in
which one aims at a particular goal g. As the necessary object of the
intention, the CrAIG in question is nothing other than the very operation
of an executively successful intention; it is the same as the successfully
completed process that constitutes the “perfection” of one’s intending that
g. A formal target cannot be the individuating goal of an intention, since
there is nothing specific to it. The executive capacities are oriented toward
substantive targets, not toward executive success as such. In intending that g
one necessarily “aims” at the formal target as well, at the achievement as the
successful culmination of one’s own doing. But in this purely formal sense
of aiming, it is uncontroversial but also uninformative to claim that in
intending one aims necessarily at the doing of one’s own that would
amount to the executive success of that intention.

5.8
Time to take stock. In the first part of the paper, I discussed whether OAC
could be supported on the basis of considerations on the individuating
content of intentions and their conditions of success. I started by consider-
ing whether OAC might hold of the simpler kind of executive intentions—
aiming. I argued that it doesn’t. Intentions are a special kind of aiming. Their
distinctive features, however, do not make a difference to the structure of
their individuating content. Hence, I argued that OAC does not hold
of intentions either.
I then moved to a distinct but related question. Could action at least be
the paradigmatic target of future-directed intentions? I argued that this is
not so, if the action is understood to be the separate target of the intending.
The actions of other agents are genuinely distinct pieces of conduct that can
be made into the proper object of one’s intentions. But one’s own actions
do not normally play this role. In the first person mode of full temporal
identification, there is a deeper continuity between intending and acting.
This continuity is deeper than it might appear at first, especially if one were
to read the logical structure of intentions out of the grammatical structure
of ordinary expressions of intentions with their infinitival complement,
since the latter seems to suggest that one is targeting one’s actions as distinct
pieces of conduct.
90 Luca Ferrero

My conclusion supports a picture of diachronic agency that takes as

fundamental the unity of what I call courses of active intentional guidance
(CrAIG), which correspond to sequences of exercises of the agent’s plan-
ning executive capacities directed toward a particular goal. According to
this picture, intending is not directed at acting. Rather, intending is a
matter of engaging in courses of active intentional guidance, some stages
of which we describe as actions or activities. These stages provide a focal
point for the organization of the courses of active guidance (whether as
central recipes or as goal-constituting performances). Further development
of this picture must be left to another occasion, but let me notice that this
conclusion is a somewhat surprising twist in the debate about OAC. The
picture that I have just sketched has some affinity with the one championed
by some on the basis of the very views that I have rejected: the infinitival
reading of the object of intentions and the defense of OAC (See Thompson
2008, Boyle and Lavin 2010, and Moran and Stone 2009). Hence, more
work needs to be done to explore the implications of my rejection of OAC
on those additional issues where the fate of OAC is supposed to bear, such
as the status of the causal theory of action, of the “guise of the good” theory,
and the nature of shared intentions (see 1.1).

6 . D E G R E E S O F D E S E I N V O LV E M E N T

6.1
Although I have argued that the content of intentions is neither necessarily
nor paradigmatically cast in terms of one’s own actions, I do not deny that
intending necessarily involves one’s own agency. This necessary involvement
is a matter of the metaphysics of executive attitudes, which are necessarily of
their own agent, both in ownership and exercise. Any executive attitude is
necessarily (but also trivially) de executione sua: possession of an executive
attitude is a matter of the agent’s exercise of her relevant executive powers.
This de se involvement of agency is fundamental but also unspecific. This is
just the ab se dimension of execution (see 3.4), a dimension that does not
impose any formal restriction on the objects of executive attitudes.
This fundamental but generic degree of de se involvement of agency is
not reflected in the individuating content. It is rather implicit in the
reference to the subject of the executive attitude: in a first-person attribu-
tion of intention, we might say that it is implicit in the “I” of the “I intend.”
There are two other degrees of involvement of one’s agency in intending.
They concern the role that the agent might play in the individuating
content of her intentions. The agent might plan on playing an instrumental
 Can I Only Intend My Own Actions? 91

role in the pursuit of g. That is, she plans on relying on her taking certain
specific steps (say, her ç-ing) toward g but she is open to the possibility that
someone else might take her place in promoting g. Alternatively, an agent
might take specific exercises of her own agency to be constitutive of her
goal—putting herself under a rational pressure to prevent anyone else from
taking those steps instead.
Here is how the three degrees of de se involvement of agency appear (in
boldface) in the logical form of intentions as formulated in the first person:
(1st) I intend that: g
(2nd) I intend that: (by way of my ç-ing) g
(3rd) I intend that: g-by-way-of-my-ç-ing
It is only in the first degree that one’s own agency is necessarily involved,
although in the form of the generic exercise of one’s executive capacities. By
contrast, one’s specific agential involvement in the recipes and goals of
one’s intentions is not required. These goals or recipes might actually
involve other agents, both instrumentally and constitutively (for instance,
I might intend that: (by way of your ç-ing) g; or I might intend that: g-by-
way-of-your-ç-ing).

6.2
One might be involved to the second and third degree in a temporally
“alienated” form. Let’s imagine that, out of a concern for my own health,
I intend to work out tomorrow. My intention is that my body undergoes
the strenuous exercise. However, the satisfaction of my desire to be healthy
does not require that I work out directly out of my prior intention rather
than just as a result of it. Given that I expect that tomorrow I will be so lazy
that I might irrationally abandon my intention, I can still carry out my
intention by setting up a pre-commitment device that manipulates me into
working out tomorrow in spite of my reluctance to do so at that time.
In this case, my intention involves me to the third degree (the goal is my
performance rather than someone else’s) even if tomorrow I need not
endorse the considerations that supported my original intention. I might
act in response to that intention in the same way as another agent might be
cajoled by it. If so, my lack of full identification with myself in the past
makes my working out a distinct action, which is the genuine object of my
prior intention. This action is the culmination of a separate CrAIG guided
by a new intention, the intention to work out that I acquire tomorrow as a
result of the manipulating effect of my prior intention.
Standard intentions are not of this alienated kind. We ordinarily take
ourselves to continue to identify over time, to continue to embrace and
92 Luca Ferrero

sustain the same intention throughout its unfolding and, thereby, to engage
in a single continuous CrAIG without self-directed goading, cajoling, or
manipulating. Hence the strongest degree of de se involvement is that of full
temporal identification rather than that of mere (and potentially alienated)
temporal identity.16

6.3
Although the second and third degrees of de se involvement are not necessary,
it is very common for ordinary intentions to take these forms.17 Hence, many
of our recipes and goals are formulated in terms of specific manifestations of
our own agency (usually in the nonalienated form). I surmise that this
explains why ordinary expressions of intention usually take the infinitival
form. For two reasons:
First, verbal expressions help characterize the specific ways in which the
agent’s executive capacities are involved in pursuing her goals. Nongeneric
verbs of action describe distinct and characteristic patterns of bodily move-
ment, tool-use, monitoring, attention management, and appreciation of
the situation. Ordinary expressions usually make explicit the communi-
catively most salient elements of the intention, in terms of (some aspects) of
its recipe and/or goal. But we should not expect these expressions necessar-
ily and fully to articulate the content of the attitude.
Second, usually the intention’s most salient elements concern the agent’s
exercise of her executive capacities in the mode of full temporal identification.
Hence it is often pleonastic to make explicit the subject of the infinitival

16
To mark this stronger form of first-personal transtemporal relation, one might
introduce an augmented quasi-indicator, S**. Hence, “S intends that: S* works out
tomorrow,” leaves it open that one might manipulate one’s future self, whereas “S
intends that: S** works out tomorrow” doesn’t.
17
The existence of the various degrees of de se involvement raises one important
question. Even if many of our ordinary projects involve by default the agent to the third
degree, should this involvement matter to us? Are there intrinsically personal projects—
projects that cannot be pursued but de se to the strongest degree? And if there are, should
we care about them? (See Perry 1976, Whiting 1986.) Reflection on the nature of the
fundamental first-degree of de se involvement does not seem to help with these questions.
The fundamental form of de se involvement makes it metaphysically impossible for the
source of agency to be but particular agents involved in the immediate exercise of their own
executive capacities. What does it take for this involvement to extend over time? That is, to
extend over its immediate and momentary exercises? Is some temporal extension
metaphysically required by the nature of agency itself ? (See Burge (2004).) And if not,
what does it take to secure this extension? And is it worth it? These questions cannot be
addressed in this paper, but the discussion of the object of intentions and their relations to
the first person makes them particularly vivid. (For some initial considerations about what
makes extended intentional agency valuable to us, see Ferrero 2009b.)
 Can I Only Intend My Own Actions? 93

complement. In saying that “I intend to ç,” it is implicit that I am talking

about my own ç-ing. But this does not imply that the agent is necessarily
involved in the recipes and/or goals of her intentions. The only necessary de se
involvement is that of the first degree, which is not reflected in the content of
the attitude, but in its subject. The standard grammatical form of the expres-
sions of intention in the infinitival form lends some initial support to the
“own action condition,” but—as I hope to have shown—these expressions,
although perfectly in order in their everyday use, are a misleading guide to the
logical form of intentions.18

REFERENCES

Baier, Annette (1970). “Act and Intent.” Journal of Philosophy 67: 648–58.
—— (1976). “Mixing Memory and Desire.” American Philosophical Quarterly 13:
213–20.
Boyle, Matt and Lavin, Doug (2010). “Goodness and Desire.” In Sergio Tenenbaum
(ed.), Desire, Good, and Practical Reason. (Oxford: Oxford University Press),
171–91.
Bratman, Michael (1987). Intentions, Plans, and Practical Reason. (Cambridge, MA:
Harvard University Press).
—— (1997). “I intend that We J.” In R. Tuomela and G. Holmstrom-Hintikka
(eds.), Contemporary Action Theory, Volume 2 (Dordrecht: Kluwer), 49–63.
Burge, Tyler (2000). “Reason and the First Person,” In C. Wright, B. Smith, and
C. Macdonald (eds.), Knowing Our Own Minds. (Oxford: Oxford University
Press).
—— (2004). “Memory and Persons.” Philosophical Review 112: 289–337.
—— (2009). “Primitive Agency and Natural Norms.” Philosophy and Phenomeno-
logical Research 79: 251–78.
Castañeda, Hector-Neri (1972). “Intentions and Intending.” American Philosophical
Quarterly 9: 139–49.
—— (1975). Thinking and Doing. (Dordrecht: Reidel).
Ferrero, Luca. (2009a). “Conditional Intentions.” Noûs 43: 700–41.
—— (2009b). “What Good is a Diachronic Will?” Philosophical Studies 144: 403–30.
Frankfurt, Harry (1978). “The Problem of Action,” In Harry Frankfurt, The
Importance of What We Care About (Cambridge: Cambridge University Press),
69–79.
Gustafson, D. F. (1986). Intention and Agency. (Dordrecht: Reidel).
Harman, Gilbert (1986). Change in View: Principles of Reasoning. (Cambridge, MA:
MIT Press).

18
Thanks to Michael Bratman, Rahul Kumar, Matthias Haase, David Shoemaker,
the audience at the NOWAR meeting in 2011, and two anonymous referees.
94 Luca Ferrero

—— (1993). “Desired Desires.” In R. G. Frey (ed.), Value, Welfare, and Morality.

(Cambridge: Cambridge University Press).
Kapitan, Thomas (1995). “Intentions and Self-Referential Content.” Philosophical
Papers 24: 151–66.
Kenny, Anthony (1992). The Metaphysics of Mind. (Oxford: Oxford University
Press).
McDowell, John (2010). “What is the Content of an Intention in Action?” Ratio
23: 415–32.
Meiland, Jack W. (1970). The Nature of Intention. (London: Methuen).
Mele, Alfred R. (1987). “Are Intentions Self-referential?” Philosophical Studies 52:
309–29.
—— (1992). The Springs of Action. (Oxford: Oxford University Press).
Moran, Richard and Martin Stone (2009). “Anscombe on Expression of Intention.”
in C. Sandis (ed.), New Essays on the Explanation of Action (New York: Palgrave
Macmillan).
Perloff, Michael (1991). “STIT and the Language of Agency.” Synthese 86: 379–408.
Perry, John (1976). “The Importance of Being Identical.” In A. Rorty (ed.), The
Identity of Persons (Berkeley, CA: University of California Press),. 67–90.
Roth, Abraham (2000). “The Self-Referentiality of Intentions.” Philosophical Studies
97: 11–51.
Searle, John R. (1983). Intentionality. (Cambridge: Cambridge University Press).
Setiya, Kieran (2011). “Intention.” In Edward N. Zalta (ed.), The Stanford Encyclo-
pedia of Philosophy, Spring 2011. <http://plato.stanford.edu/archives/spr2011/
entries/intention/>.
Stoutland, Frederick (2002). “Critical notice of Bratman’s Faces of Intention.”
Philosophy and Phenomenological Research 65: 224–38.
Thompson, Michael (2008). Life and Action. (Cambridge, MA: Harvard University
Press).
Tuomela, Raimo (2005). “We-intentions Revisited.” Philosophical Studies 125:
327–69.
Velleman, J. David (1997). “How to Share an Intention.” Philosophy and Phenom-
enological Research 57: 29–50.
Whiting, Jennifer (1986). “Friends and Future Selves.” The Philosophical Review 95:
547–80.
Wilson, George M. (1989). The Intentionality of Human Action. (Stanford, CA:
Stanford University Press).
 4
Regret, Agency, and Error*
Daniel Jacobson

“How much better if it had been otherwise.” Bernard Williams characterizes

this proposition as the constitutive thought of regret in general. Although he
never explicates the notion, presumably this is a thought that must somehow
be credited to an agent, though perhaps not consciously entertained, in order
for her to be in that emotional state. According to Williams (1976: 27): “In
this general sense of regret, what are regretted are states of affairs, and they
can be regretted, in principle, by anyone who knows of them.” Thus one can
regret unfortunate states of affairs to which one bears no special relationship.
Although Williams only briefly characterizes generic regret, before narrowing
his focus to those cases directed at one’s own action—which he calls agent-
regret—it’s easy to think that he has gotten off to a false start by describing
even the more general state too broadly.1
Since the thought supposedly constitutive of regret is available to
anyone, it follows that, in Williams’s view, everyone has equal standing
to regret any unfortunate event. He recognizes no difference between agent
and spectator when it comes to what they can (aptly) regret. Yet one might

* This paper issues from an ongoing collaborative project undertaken with Justin
D’Arms. Many of its arguments were developed in discussion with him, though respon-
sibility for the paper’s faults falls to me alone. I am grateful to Bruce Brower, Sarah Buss,
Rae Langton, Michael McKenna, Peter Railton, Tim Scanlon, David Shoemaker, two
anonymous referees, and audiences at the New Orleans Workshop on Agency and
Responsibility, Syracuse University, and the University of Michigan for helpful com-
ments on previous drafts. My greatest debt is to Bernard Williams and Michael Stocker
whose work not only inspired this paper but sparked much of my interest in moral
philosophy. This paper was supported by a grant from the John Templeton Foundation.
1
It should be noted that Williams’ primary concern is with the narrower notion of
agent-regret. The problems I will raise concern agent-regret too, but this initial com-
plaint reflects mostly a semantic (and hence superficial) difference with Williams: we are
working with different conceptions of regret. But it should become clear that we are not
merely fighting over words.
96 Daniel Jacobson

think for instance that Neville Chamberlain, the architect of the 1938
appeasement of Hitler in Munich, who returned to England triumph-
antly declaring “peace with honour,” was peculiarly well situated to
regret that policy in light of the German blitzkrieg of Poland in 1939.
By contrast, Winston Churchill famously announced, on Chamberlain’s
return from Munich: “You were given the choice between war and
dishonour . . . you chose dishonour and you will have war.” Obviously
Churchill thought it would have been much better had Britain con-
fronted Hitler earlier and under more favorable conditions. As Williams
understands regret, Churchill and Chamberlain both have reason to
regret the Munich agreement—but this seems odd. Indeed, Williams
can hardly claim that Chamberlain has more reason to regret than does
Churchill, and hence that his feelings should be more intense, since even
in retrospect it was surely Churchill who believed more strongly that
another course of action would have been preferable. Yet agent-regret
might seem to solve this problem on the cheap, since, by definition, it is
an emotion that Chamberlain but not Churchill could feel about the
Munich appeasement.
However, I will argue that the deep worries in this neighborhood cannot
be met in this way, by designing new emotions to meet a philosophical job
description. In particular, I reject the method Williams employs in charac-
terizing generic regret and then constructing agent-regret out of it. This
method of type-identifying emotions is characteristic of the cognitivist
theory of the emotions, which identifies a constitutive thought, belief (or
perhaps some weaker propositional attitude) in which is a necessary condi-
tion for having any given emotion.2 Williams does just this by explicitly
identifying a constitutive thought of regret and implicitly suggesting
another for agent-regret. Perhaps having this thought isn’t sufficient for
being in a state of regret, since it might need to be accompanied by the right
sort of feeling. Nevertheless, Williams (1976: 27) expressly claims that “in
principle” anyone who can make the judgment can have the emotion
and—since the judgment he associates with regret expressly pertains to
states of affairs—“anyone who knows of them” can make the relevant
judgment. While I will not argue directly against cognitivism here, I do
want to use this discussion to illustrate two serious problems with the

2
Such cognitivist theories come in several varieties, most notably neo-stoicism
(Solomon 1988, Nussbaum 2001), judgmentalism (Taylor 1985), and quasi-
judgmentalism (Greenspan 1988, Roberts 1988). These theories share the defining
propositions methodology of type-identifying the emotions, though they differ on
whether the thought must be accepted or merely entertained, and whether it is wholly
or only partly constitutive of the emotion.
 Regret, Agency, Error 97

theory, and to suggest an alternative conception of how emotions are to be

type-identified and when they are rational.3
The first problem is evident in Williams’s characterization of generic
regret by way of its constitutive thought, without regard to whether the
emotional states that can be expressed with this thought resemble each
other in their affective and motivational aspects. As a result, his notion of
generic regret turns out to be a genus rather than a species of emotion. It
includes not just agent-regret and what Williams refers to as the regret of
the spectator, but also such disparate sentiments as guilt, shame, sorrow,
and anger—all of which can (sometimes) be manifested in the thought,
“How much better if it had been otherwise.” Philosophers can type-identify
emotions however they like, of course, but not all such constructions will
be fruitful and some prove misleading. The defining-proposition metho-
dology of cognitivism threatens to obscure matters by conflating some
importantly different emotions, as well as other practical attitudes that
lack the distinctive affective and motivational features of emotion. So it is
with agent-regret, which inherits this problem with generic regret and has
another significant problem of its own.
This second problem reflects a more general difficulty facing the cogni-
tivist theory of emotion: it threatens to create chimerical emotions when-
ever a sentiment’s underlying motivational component—its action tendency
(Frijda 1986)—conflicts with its posited constitutive thought. This
problem emerges dramatically for agent-regret, because the rationality
conditions implicitly given by its constitutive thought contradict
Williams’s claims about his two primary examples. In light of what
Williams says about agent-regret, its constitutive thought must be: “How
much better, in some respect, if I had done otherwise.” Notice that this
thought concerns the outcome of one’s agency rather than any decision about
acting; and that it does not imply any conviction that things would be
better otherwise, all things considered. I will suggest that the chimerical
emotion here is regret accompanied by the thought: “While I regret doing
what I did, I endorse doing it again in similar circumstances.” This would
make good sense if directed at a bad but unlucky outcome of one’s action,
but it would be confused if directed at one’s decision. Hence part of my
dispute with Williams concerns whether regret is fundamentally about bad
outcomes or bad decisions, loss or error. A closely related disagreement

3
The issue of what is meant by calling an emotion rational will be taken up presently.
In order to follow the literature under discussion, I will use the term “rational” to refer to
what D’Arms and I elsewhere called its fittingness, in order to disambiguate this
judgment from other kinds of assessment of the appropriateness of emotions.
98 Daniel Jacobson

with both Williams and Michael Stocker concerns whether one can
rationally regret taking (what one considers) the best option available.
Nevertheless, their discussion of regret and its cognates exhibits
an important virtue, which I greatly respect and wish to maintain: its
psychological realism. Since I find this one of the characteristic virtues in
the work of both Williams and Stocker, my two principal foils here,
I should begin by noting my admiration for the psychological insight
that informs their moral philosophy. Two facets of this realism are
particularly significant for present purposes, what I’ll call the agency
principle and the residue principle, both of which concern the relationship
between acting and feeling. These two principles are illustrated respect-
ively by Williams’s two paradigms of agent-regret: the lorry driver who
blamelessly kills a child, and Agamemnon who chooses to sacrifice his
daughter in order to save his becalmed fleet.4
Consider first the agency principle: the claim that we are connected,
through our feelings, to even the involuntary aspects of what we do—as
opposed to what simply happens around us. Williams has in mind things
for which we are merely causally responsible, but in virtue of what we
intentionally did. He illustrates this insight with his famous lorry driver
case. The driver who blamelessly kills a child can surely be expected to feel
something different than does a mere spectator to the accident, though
neither is culpable for the tragedy. This core insight is one of his primary
motivations for differentiating the “regret of the agent” from that of a mere
spectator. Whereas Williams insists that the lorry driver feels rational agent-
regret, I will argue that he can be expected to feel irrational but praise-
worthy guilt. This is not to deny that the driver might have another bad
feeling, directed at the state of affairs or even his role in bringing it about;
but this emotion should not be considered a form of regret either. Although
I thus reject both Williams’s claim that the driver feels regret and that this
emotion is rational, I think he grasps an important kernel of truth. The
agency principle, understood as an empirical rather than a normative claim,
is a psychological insight with profound philosophical implications. But
the further claim that such predictable feelings are rational needs to be
considered separately, as does the question of exactly what emotion we
should expect from the driver in this scenario.

4
Before considering these scenarios, however, it’s worth noting that they are chosen
for their philosophical interest. This strategy proves double-edged. In trying to gloss an
emotion—that is, to understand its distinctive concern—I contend that one should look
first to commonplace rather than novel cases. Though philosophers tend to focus on
philosophically interesting examples, the more mundane cases are also more perspicuous,
and unusual cases can be misleading—often for the very reason that they’re interesting.
 Regret, Agency, Error 99

Second, both Williams and Stocker observe that sometimes when we face
conflicting values or obligations, which cannot be jointly satisfied, some
residue of bad feeling (or emotional “remainder”) can be expected even
when we have done the best we could.5 This residue principle contains a
crucial psychological insight as well, though I will describe it differently
and draw different conclusions from it than do Williams and Stocker.
Williams (1965: 181) adduces as an example Agamemnon, who “ought to
discharge his responsibilities as a commander, further the expedition, and so
forth . . . [but] ought not to kill his daughter.” We are of course to imagine
that he can satisfy either obligation but not both. Williams insists further that
both incompatible obligations bind Agamemnon, and that therefore it is
rational for him to regret whatever he does, despite thinking he has most
reason to do it. In fact, Williams and Stocker go so far as to claim that
anything Agamemnon may be able to do in this tragic situation would be
wrong.
While I do not accept this conclusion, the cases on which I will focus
concern conflicts of value rather than obligation, so as to avoid issues of
guilt and wrongness. The problematic claim here, developed most per-
spicuously by Stocker, is that in some such conflicts of value it will be
rational to regret choosing what one knows is the better option: the greater
good or the lesser evil. Let us call this residual regret. As with the agency
principle, one can accept the residue principle as an empirical insight
without granting the rationality of residual regret. I will argue, analogously,
that some of these cases don’t involve regret but another emotion, which
may or may not be rational; and others are cases of irrational regret that is
nevertheless, in a different sense, an appropriate emotional response. This
normative corollary to the residue principle, concerning the putative
rationality of residual regret, plays a crucial role in a recent debate over
conflicts of value, to be discussed later. But this argument obscures the
insight at the heart of the residue principle: the phenomenon of loss. I grant
that something significant can be lost in making the better choice, and that
this loss can indeed justify residual bad feeling in various respects.
A similar problem plagues discussion of both the agency and residue
principles. The problem lies not in the principles themselves but with the

5
Sometimes this remainder is supposed to be a new obligation arising out of the old
(forsaken) one. But that kind of remainder poses no problem to those who think that the
lesser of two conflicting (prima facie) obligations does not bind, and hence that violating
it would not be wrong. It is easy to grant that when I miss our lunch date for some
obviously good reason, I incur an obligation to explain myself to you, to reschedule at
your convenience, or even to pick up the tab. But these new obligations can be seen as
issuing from the inconvenience I’ve caused you, however justifiably, rather than from any
wrongdoing on my part.
100 Daniel Jacobson

conception of regret (or agent-regret) these philosophers adopt, and the

normative conclusions this conception leads them to draw about its ration-
ality. I contend that agent-regret is a chimerical emotion, an amalgam of
two fundamentally different sentiments with disparate conditions of ration-
ality: guilt and regret. But if we adopt a more narrowly focused conception
of regret and pay closer attention to the different respects in which emo-
tions can be evaluated, that will allow us to accommodate these authors’
psychological insights while avoiding certain philosophical excesses. Spe-
cifically, the agency principle does not imply that it is rational to regret
tragedies for which one was merely causally responsible; and the residue
principle does not imply that it is rational to regret making the correct
choice. Both these conclusions should be rejected.

1 . R E G R E T A S A SE N T I M E N T

As a purely semantic point, it must be granted that the word “regret” can be
used extremely broadly. However strange it sounds for a present-day histor-
ian to say that she regrets the Munich agreement, she could make use of
what Williams elsewhere calls the thoroughly adverbial sense of the word to
say: “The Allies regrettably chose appeasement at Munich.”6 Even if any
unfortunate situation can be called regrettable, though, should we really
conclude that everyone has equal standing to regret it? It will be more
perspicuous to focus on when it is apt to ascribe regret to an agent, rather
than on what could be called regrettable but might just as well be called
unfortunate. Indeed, if “regrettable” and “unfortunate” are not synonymous,
when used in this broadest sense, then it is because regret involves someone’s
agency or choice, whereas misfortune need not. This linguistic observation
suggests that Williams’s notion of generic regret, which is directed at states
of affairs rather than actions, is a misnomer. The fact that the storm did
so much damage is more felicitously termed unfortunate than regrettable.
Although in my view all regret is agential, one can certainly judge another
person’s action regrettable, meaning thereby that the agent should
(rationally) regret it.
Perhaps the best lesson to draw from these reflections is that ordinary
language is especially misleading with respect to emotion terms. Even if the
adverbial usage counts as a literal ascription of regret, certain familiar

6
Williams (1973b: 112, fn. 1) writes of the “thoroughly adverbial” account of
pleasure, on which anything that can be done pleasurably counts as a form of pleasure.
 Regret, Agency, Error 101

“expressions of regret” could not qualify by any plausible standard.7 Many

of these broad uses of emotion terms should not be understood as
ascriptions of any actual emotion, understood (as is standard) as an
occurrent, affect-laden, object-directed state. Understanding regret as an
emotion helps explain the oddness of placing Chamberlain, Churchill,
and the imagined historian in the same position with respect to regretting
the Munich agreement. I will use these three characters to explicate the
distinction between evaluative judgments, emotions, and sentiments.
The historian could write a speculative alternative history, detailing how
much better it would have been had Hitler not been appeased but con-
fronted in 1938, while making this conjecture dispassionately.8 Though she
thereby has the thought supposedly constitutive of regret, she is unlikely to
be seriously bothered by it. If any unwanted feeling keeps her up at night, it
is more likely anxiety about how her work will be received than obsessive
thoughts about what might have been, if not for Munich. Indeed it would
be neurotic for her to be greatly pained by disastrous decisions in the distant
past to which she bears no special relationship. This is a fundamentally
aretaic assessment of when emotional responses are admirable or not. So the
historian makes an evaluative judgment without having the emotional
response characteristically (but not inevitably) associated with it. Neither
the intimacy nor the disparity between evaluation and emotion should
be surprising. When one has the emotion, one typically makes the
characteristic judgment; and when one makes the evaluative judgment,
one sometimes has the emotion. The latter connection is weaker because
various contextual features such as one’s place in the scenario, one’s
temperament, and even temporal factors play a crucial role.
By contrast with the historian, Churchill was pained by the Munich
agreement precisely because he thought appeasement so much worse than
the alternative. Undoubtedly Churchill had an emotional response to the
event, in virtue of believing the supposedly constitutive thought. But was it
regret? There is another emotion term ready to hand, which also involves
being moved—both in the sense of having some feeling and being motiv-
ated—by the thought of how much better things might have been. We

7
My favorite example comes from a postcard used by a famous scholar to reply to
requests which begins, “Edmund Wilson regrets that it is impossible for him to: Read
manuscripts . . . ,” and then goes on to list some twenty other scholarly tasks that he
was, regrettably, unable to perform. We should all be so fortunate as to have such regrets.
8
But cf. Deigh (2000) who suggests that this assertion must issue from a Humean
theory of mind on which only passions and not judgments can motivate. On the
contrary, this is a commonsense observation about specific cases. Indeed, I claim to
make the relevant judgment about Munich myself, without affect. It seems ad hoc to
deny that I make this judgment sincerely.
102 Daniel Jacobson

might instead say that Churchill felt dismay over the Munich agreement (in
addition to whatever else he may have felt, such as anger at Chamberlain’s
self-righteousness, contempt for his naivety, fear for England’s future, and
so forth). Whereas the historian’s judgment that things would have been
better otherwise is simply an evaluation, dismay and regret are emotions
because they are ways of feeling. The question is whether they are the same
emotion. Is the regret that Chamberlain can be supposed to have felt, in
retrospect, about his decision to appease Hitler at Munich simply a “par-
ticularly important species” of the same emotion Churchill felt about what
happened despite his objections, as Williams implies? More abstractly, is
this question a matter of psychological fact, as it purports to be, or simply
linguistic stipulation about “regret” and “dismay”?
At this point I need to broach two additional claims, one general and the
other more specific, for which I cannot yet argue. These are substantial
claims, open to falsification, and ultimately in need of elaboration and
defense. The first concerns a distinction between emotions and sentiments.
Although the disparate class of states called emotions have little in common
(compare approval and surprise, for example), my focus here will be on a
core group I’ll term sentiments, each of which constitutes a natural psycho-
logical kind of state whose nature can be discovered empirically rather than
stipulated semantically.9 The second claim specifically concerns regret:
that it is a sentiment concerned with the agent’s own errors. If this gloss is
correct, then the reason it would be odd to attribute regret over Munich to
Churchill is simply that he didn’t make the mistake of trying to appease
Hitler—Chamberlain did. But since both Williams and Stocker’s central
claims about regret are inconsistent with my gloss, it would beg the
question against them to assume it. For now I simply aspire to make
these two claims plausible: that regret is a sentiment, and that it is best
glossed as being about one’s own error.
What am I claiming by calling regret a sentiment? Like other emotions,
sentiments are syndromes of thought, feeling, and motivation; but unlike
some members of the disparate class of emotional states, the sentiments
play a distinctive role in the human mental economy.10 Sentiments are
discrete sources of motivation, which can be in tension with our overall

9
Philosophers of the emotions as disparate as Amélie Rorty (1978: 104) and Paul
Griffiths (2004) doubt that emotion counts as a natural kind. Although I agree,
I contend both that the sentiments as a group, and each individual sentiment such as
regret, forms a natural psychological kind. In part, this claim amounts to adopting a
broad construal of the emotions and a narrow, more technical notion of the sentiments.
Cf. Frank (1988), Frijda (1986).
10
See D’Arms and Jacobson (2003, 2006). Note the unfortunate change of terminology:
what we call natural emotions in (2003) become the sentiments in (2006).
 Regret, Agency, Error 103

beliefs and desires. For this reason, two related features distinguish the
sentiments from mere emotions: they are prone to stable recalcitrance, and
they issue in acting without thinking in a sense to be explicated. First,
whereas emotions that are not sentiments evaporate when one disbelieves
their associated judgment, the sentiments can be recalcitrant, in that an
agent can be in the grip of a sentiment contrary to his better judgment.11
Those who are afraid of flying despite knowing that their flight is safer than
the drive to the airport, which does not frighten them, are prone to stably
recalcitrant fear. By contrast, the emotion of resentment, understood as a
form of anger that involves a critical moral judgment, cannot coexist with
the belief that no wrong was done; what can persist recalcitrantly is just
anger.
The phenomenon of recalcitrance reflects the epistemic aspect of the
sentiments’ partial encapsulation from our beliefs and desires; whereas
acting without thinking reflects the motivational aspect. For instance, the
sentiment of anger is notoriously insensitive to whether the behavior it
motivates coheres with our overall ends. This gives anger and the other
sentiments an essential role in psychological explanation of behavior that
would otherwise be mysterious. These explanations are obvious, not espe-
cially insightful. They might be called shallow (as opposed to depth)
psychological explanations, so long as this isn’t taken pejoratively. Obvious
explanations need not be false.
Consider for example the behavior of the French footballer Zidane,
perhaps the greatest player of his era, in his final match: the 2006 World
Cup Final. This was not only the most important match of Zidane’s career,
it was his last, and the hopes of the French side depended almost entirely on
him. Yet when an Italian player taunted him, Zidane responded with a
flagrant and illegal head-butt, certain to get him thrown out of the match,
thus ending his career in selfishness and disgrace. He thereby ruined any
chance of attaining what had to be his greatest desire: leading France to a
World Cup victory in the final match of his spectacular career. Instead he
got the immediate gratification of revenge at the cost of defeat and igno-
miny. Why would anyone behave so irrationally—that is, so contrary to his
overall beliefs and values? The answer should be obvious: he was in a rage
and acted on the characteristic motivation of anger, namely retaliation.
Sentiments thus explain the phenomenon of acting without thinking in this

11
See D’Arms and Jacobson (2003) for discussion of recalcitrance and why there isn’t
recalcitrant “fear of flying,” understood as an emotion, only recalcitrant fear. The crucial
claim here is that whenever an emotion figures in such an explanation, some underlying
sentiment does the real explanatory work.
104 Daniel Jacobson

specific sense: the agent does not consider how his action coheres with his
aims.12
In short, the sentiments are a core class of emotions that can be found
across cultures and times, albeit with some variation in their eliciting
conditions, which have some characteristic motivational tendency.13
We can thus say, with Aristotle, that anger is about undeserved slights—
even though what counted as a slight for Aristotle differs from what a
modern youth would consider a diss (i.e. show of disrespect). Whereas any
combination of thought, feeling, and motivation counts as an emotion, the
sentiments figure essentially in psychological explanations and predictions
of behavior. Only some of the varied states that get called emotions will
count as sentiments, while others cut across different sentiments. For
example, we have stipulated that one can be dismayed in various ways, all
of which can be expressed by a thought such as, “How much better if it had
been otherwise”; but sometimes such dismay will involve sorrow, anger, or
shame rather than regret. Although we can say that dismay is an emotion
that includes various sentiments expressed by that constitutive thought, this
way of speaking can be treacherous. Each of these sentiments has distinctive
content and issues in different characteristic behavior. Hence what explains
why Jack lashed out in his dismay, while Jill tried to make amends and Jan
withdrew completely, is that Jack’s dismay was a form of anger, Jill’s of
guilt, and Jan’s of shame. What I’m calling dismay resembles the emotional
manifestations of Williams’s notion of generic regret, but I contend that
regret is best understood as a sentiment.
Whatever Chamberlain actually felt after the blitzkrieg shattered the
illusion of “peace in our time,” there is a familiar psychological syndrome
that he might have suffered but Churchill could not. Only Chamberlain
could have had a bout of the sentiment I will hereafter simply call regret:
the syndrome of painful feelings of self-reproach focused on his blunder
and its disastrous consequences, accompanied by the wish to undo the error
and the intention to act differently next time. Part of the regret syndrome
thus involves a motivation toward what might loosely be called policy
change, although in this case there could be no question of Chamberlain
getting another chance. (Indeed, the motivation should be expected even
when it is futile and manifests itself only in symbolic gestures and fantasy;

12
This is not always to condemn such action, even when it conflicts with considered
judgment, much less to bemoan the fact that we humans have such discrete, fast-and-frugal
motivational systems in addition to slower and more deliberative ways of thinking (Gigerenzer
and Todd 1999).
13
See Ekman (2003) on the universality of facial expression of the sentiments (or
basic emotions).
 Regret, Agency, Error 105

this partly explains the social importance of such gestures and the psycho-
logical importance of such fantasy.) This characterization is loose because
in some cases the regretted act does not accord with any previous policy:
either it was done without thinking; or not as an expression of a general
policy; or even, in cases of weakness of will, contrary to one’s policy. In
all these cases, though, regret has a motivational component concerning
one’s intentions for future action. The lessons learned from regret involve
dwelling painfully, perhaps obsessively, on some (putative) mistake and the
decision leading up to it.14 In sum, our historian makes a dispassionate
evaluative judgment, and Churchill’s painful dismay is an emotion, but
only Chamberlain’s hypothesized regret counts as a sentiment. Moreover,
agent-regret will prove to be chimerical because it manifests itself in
disparate sentiments, which can be in tension both with Williams’s
characterization of the state and with his paradigm cases: the lorry driver
and Agamemnon.

2. WHAT IS AGENT-REGRET?

I have suggested that regret is a familiar, painful sentiment focused on

error and motivating the intention to act differently. This gloss is admit-
tedly contentious, since Williams and Stocker expressly assert that to regret
an act, even rationally, need not involve repudiating it as mistaken. As
Williams (1976: 31) writes:
Regret necessarily involves a wish that things had been otherwise, for instance that
one had not had to act as one did. But it does not necessarily involve the wish, all
things taken together, that one had acted otherwise. An example of this . . . is offered
by the cases of conflict between two courses of action each of which is morally
required, where either course of action, even if it is judged to be for the best, leaves
regrets—which are, in present terms, agent-regrets about something voluntarily
done. We should not assimilate agent-regret and the wish, all things taken together,
to have acted otherwise.
My proposal for how to understand regret closely resembles the conception
(of agent-regret) Williams rejects, but it is important to note that I am not
assimilating regret and the wish to have acted otherwise. One can find
oneself regretting something without wishing, all things considered, that

14
The word “putative” does important work here, since there is no guarantee that the
action genuinely regretted was really mistaken. Indeed, when things go badly, one can
expect that the reasons against doing what was done will become especially salient, and
the reasons in favor will appear correspondingly diminished.
106 Daniel Jacobson

one acted otherwise. When this is really regret about the choice rather than
dismay about the outcome, however, I hold that it is recalcitrant regret,
which is irrational by the agent’s own lights. Hence the criterion for the
rationality of regret is whether or not one erred. That is one of my central
disagreements with Williams and Stocker.
Recall that Williams coins agent-regret out of his capacious conception
of regret in general, with the additional restriction that it must focus on the
agent’s own actions and their consequences, and the insistence that this
need not be an all-things-considered wish. He thus implies that agent-
regret has a constitutive thought along the lines, “How much better, in some
respect, if I had done otherwise.” Moreover, he (1976: 27–8) writes:
The sentiment of agent-regret is by no means restricted to voluntary agency. It can
extend far beyond what one intentionally did to almost anything for which one was
causally responsible in virtue of something one intentionally did. Yet even at deeply
accidental or non-voluntary levels of agency, sentiments of agent-regret are different
from regret in general, such as might be felt by a spectator, and are acknowledged in
our practice as being different. The lorry driver who, through no fault of his, runs
over a child, will feel differently from any spectator, even a spectator next to him in
the cab, except perhaps to the extent that the spectator takes on the thought that he
himself might have prevented it, an agent’s thought.
Here Williams seems to be making a psychological claim about what we
can expect an ordinary, decent person in the unfortunate driver’s position
to feel: something different from what a mere spectator would feel. He also
claims, plausibly, that we would expect the driver’s bad feelings to move
him to make some gesture of reparation to the victim. As Williams (1976:
28) notes, “The lorry-driver may act in some way which he hopes will
constitute or at least symbolize some kind of recompense or restitution, and
this will be an expression of his agent-regret.” While a spectator would
likely be moved to console the victim’s parents, it would be strange for her
to offer restitution for the accident.
Moreover, although Williams grants that people are prone to “irrational
and self-punitive excess” in this area, he insists on the propriety of some
such expression from the driver. Despite the driver’s (stipulated) blameless-
ness, he claims that such a response would not only be normal but proper,
indeed rational. “[I]t would be a kind of insanity never to experience
sentiments of this kind toward anyone,” Williams (1976: 29) writes,
“and it would be an insane conception of rationality which insisted that a
rational person never would.” Thus the leading thought of his discussion of
the lorry driver case—with which he introduces the notion of agent-
regret—is what I’ve called the agency principle. As he (1993: 92) notes
elsewhere, it is an “utterly familiar fact that what happened to others
 Regret, Agency, Error 107

through our own agency can have its own authority over our feelings, even
though we brought it about involuntarily.” This seems especially likely
when the outcome is negative.
Williams moves quickly from the predictability of such a response to the
insanity of being entirely immune to it, and then to the rejection of any
conception of rationality on which a rational person would be so immune.
Surely “insanity” is hyperbole, and what he really means is something more
like inhumanity. Any decent person will display such tendencies. Thus
Williams commits himself to the rationality of the lorry driver’s regret
(which I will deny), on the grounds that it would be unacceptable to adopt
standards of rationality for the emotions that fail to cohere with human
psychology (which I accept). One issue between us concerns the specific
respect in which one should endorse this regret. This will be the topic of
Section 3. Another question, to which I will now turn, is whether we should
accommodate the agency principle by positing a novel emotion, namely
agent-regret.
Williams’ notion of agent-regret seems to differ from guilt, which is
typically taken to be rationally restricted to the voluntary—or, at any rate,
the blameworthy.15 As Williams (1993: 89) himself observes, “What
arouses guilt in an agent is an act or omission of a sort that typically
elicits from other people anger, resentment, or indignation.” This is just
what distinguishes the blameless driver from the negligent or malicious
one: no one has any justified complaint against him. Although Williams
(1976: 27) considers agent-regret to be a species of the more capacious bad
feeling he calls regret in general, he also claims that it “is not distinguished
from regret in general solely or simply in virtue of its subject-matter,”
namely one’s own action, but is also characterized by a different form
of expression in motivation: the desire to make reparations. But note
that this is exactly the motivation characteristic of guilt.16 In the throes of
guilt, one feels like one owes a debt, and that discomforting feeling issues in
the desire to make reparations. Thus Williams’s notion of agent-regret
seems precariously placed between (generic) regret and guilt. What then
is the point of giving the emotion we can expect from the lorry driver a
distinct name? And why call it agent-regret?

15
It is worth noting that, years later, Williams (1993: 93) tosses off the remark that
the lorry driver’s agent-regret is “psychologically and structurally a manifestation of
guilt.” On one hand, this suggests that my treatment of the case does not differ as
much from Williams as first appears; on the other, this seems an admission that even his
restricted notion of agent-regret conflates cases of regret with those of guilt.
16
See, e.g. Gibbard (1990), Tangney and Dearing (2002), Baumeister et al. (1994,
1995).
108 Daniel Jacobson

This question becomes even more puzzling when we consider what some
other philosophers have written about the lorry driver case. Gabrielle
Taylor describes the driver’s emotion simply as guilt and concludes that
guilt concerns causal rather than moral responsibility. “Causal responsi-
bility is the type that is sufficient for guilt, and that much is also necessary,”
Taylor (1985: 91) writes; hence, guilt “cannot be vicarious, and feelings of
guilt similarly cannot arise from the deeds or omissions of others,” not even
one’s children or compatriots. But almost no one else agrees with her, and
for good reason: counterexamples seem rampant.17 Taylor’s gloss of guilt
forces her to hold that vicarious guilt is not merely irrational but impossible,
which would make survivor guilt and liberal guilt into counterexamples
to her theory—if they are indeed forms of guilt, not just states casually
called by that name.18 Of course, she could always posit another emotion
to accommodate these phenomena rather than denying their possibility
outright. But to posit something like “liberal compunction” to cover cases
of vicarious guilt is an ad hoc maneuver designed to salvage the theory by
stipulation. Worse, she would no longer be talking about guilt, the natural
kind of psychological state, but a subset of its instances chosen precisely to fit
her claims. This renders tautological a claim—that guilt cannot be
vicarious—which seems to be put forward as an empirical observation. It
would be better to advance a claim that is neither empirical nor semantic but
normative: that causal responsibility is necessary for guilt to be rational. This
would make survivor guilt possible, even explicable, but not rational; and
that combination of claims seems quite plausible. Moreover, Taylor’s
implicit suggestion that we should expect the lorry driver to feel guilt
rather than some variety of regret is exactly right.
Though we can evaluate claims about the sentiments, those about
constructed emotions are much more difficult to assess, because they
seem precariously placed between empirical hypothesis and semantic stipu-
lation. Consider Marcia Baron’s (1988) discussion of agent-regret and the
lorry driver, in which she claims to detect a difference between regret and
agent-regret that Williams fails to note. Baron (1988: 262) writes: “Agent-
regret has an ethical dimension which plain regret lacks. Agent-regret is felt
toward the sorts of things which, if done deliberately, would properly
occasion guilt.” Baron clearly means to be talking about the same state as

17
See Greenspan (1992) for counterargument. When Taylor is being more careful
she allows that the requirement is, rather, that an agent must believe himself causally
responsible for something in order to feel guilty over it. Even so, many counterexamples
arise.
18
The fact that these states motivate apology and attempts at reparation, even where
there is no causal responsibility, is a reason to think that they are indeed forms of guilt.
 Regret, Agency, Error 109

Williams, not just expropriating an emotion term he coined. Whereas for

him agent-regret is something like regret about what I did, she thinks it
more like guilt about the involuntary. What Baron gets right is that we
should expect ordinary decent people to feel guilt over terrible results of
their actions even when they are involuntary. This is what I’ve called
Williams’s agency principle, but with the connection to guilt made
explicit. Even so, not all of Williams’s paradigms of agent-regret look like
guilt over the involuntary—the Agamemnon case does not.
The ethical dimension Baron claims to find in agent-regret also involves
a threshold of significance, below which she thinks it impossible to have the
emotion. In contrast with what she calls plain regret, Baron (1988: 262;
emphasis added) holds that “agent-regret simply toward my failure to buy
myself a pair of shoes that I liked does not seem possible. What I failed to do
was too trivial and too remote from ethical concerns.” Yet surely a
sufficiently shallow person could regret forgoing the shoes, and it is
unclear how Baron can deny that this is agent-regret except by definition.
Although she can restrict her use of “agent-regret” stipulatively to cases that
are of ethical import and reach a threshold level of significance, it is hard to
see what this gains. Not only would it trivialize her (seemingly empirical)
impossibility claim, she would then simply be talking about a different
emotion than is Williams, while taking herself to be correcting him about
its nature. Again the normative hypothesis seems more tenable. Why not
say instead that only a little regret is rational over minor errors?19
Moreover, Williams does not merely “fail to note” that agent-regret must
have an ethical dimension, as Baron contends; he denies it outright. Her
claim contradicts his initial characterization of the emotion as a refinement
of regret in general. If one can (rationally) feel agent-regret toward anything
about which one can truthfully think, “How much better if I had done
otherwise,” then surely this can be thought and felt about missing out on a
nice pair of shoes at a great price. Indeed, Williams expressly acknowledges
that one can feel agent-regret over mistakes that harmed only oneself. Since
he thinks that the lorry driver’s emotion will be expressed in the desire to
make restitution, he has to adduce a different sort of expression for these
cases. “Granted that there is no issue of compensation to others in the
purely egoistic case,” Williams (1976: 32–3) writes, agent-regret can only
be expressed in “the agent’s resolutions for his future deliberations.” This,

19
This point illustrates another infelicity of ordinary language: some words do have
connotations of great significance, such as “remorse” and “misery,” both of which suggest
great intensity. But surely this is just a semantic point and, what is more, it does not hold
about “regret.”
110 Daniel Jacobson

of course, is the motivation characteristic not of guilt but regret—that is,

regret understood in my sense, as a sentiment that focuses on error.
However, Williams does fail to notice that this expression too will be out
of place in certain cases of putative agent-regret: those where the agent does
not repudiate her action, either because she chose the least-bad alternative
in a bad choice situation, or because she made the right decision but things
happened to turn out badly nonetheless. In those cases there is no oppor-
tunity for either reparation or repudiation, so agent-regret cannot seem to
find any appropriate expression in action. And yet the thought supposedly
constitutive of agent-regret is very much in order. It would be much better,
at least in some respect, if I had done (or had been able to do) otherwise.
These cases strongly suggest that the most striking psychological phenom-
enon in the neighborhood is not one that can attach to any way things
might have been better. Rather, it is Stocker’s notion of loss, which we will
consider in more detail in Section 4.20
The problem for Williams is that his posited notion of agent-regret has
no coherent expression in behavior. Worse, his paradigms have two quite
different expressions, which cut across the distinction between egoistic and
altruistic cases, and which happen to be precisely the motivations charac-
teristic of regret and guilt. This strongly suggests that Williams has con-
flated (at least) two distinct sentiments under his broad class of generic
regret, characterized in terms of a constitutive thought. The problem infects
agent-regret as well, since that subspecies is sharpened out of the more
capacious notion by the stipulation that it concerns one’s own action. Since
both guilt and regret typically focus on one’s own action, agent-regret too
will cut across the distinction between the two sentiments. This explains the
disparate action tendencies found in the different cases Williams groups
together as agent-regret. Such chimerical emotions, which are not distinct
sentiments but constructed groupings, cannot sustain empirical generaliza-
tions. Although philosophers can type-identify emotions however they
like—and hence there is no point in denying that agent-regret counts as
an emotion—only some emotions turn out to be sentiments which explain
human behavior that would otherwise be mysterious, such as Zidane’s
enraged head-butt.
Although there are no definitive arguments in this domain, I can offer
some reasons to recognize regret as a sentiment that plays an important and

20
When one makes the right decision but it turns out badly, one loses out on
something genuinely valuable, with only the cold comfort of having chosen correctly.
And when one makes a hard choice, where there are different reasons on both sides of the
question—even if the reasons for one choice clearly outweigh the other—then we often
do painfully feel the loss of the goods rationally forgone.
 Regret, Agency, Error 111

familiar role in the explanation of human behavior. The function of regret

is to focus us on our ends and to help us avoid temptation and distraction.
The regretful agent recalls his action painfully and without bidding; his
regret motivates him to undo the action if possible, and to change policy for
the future. This gloss of regret locates its concern in putative mistakes. If
this syndrome of thought, feeling, and motivation seems familiar and
useful in shallow psychological explanation, then that would give us reason
to posit a sentiment in the neighborhood of the pre-theoretic notion of
regret. Further evidence that regret is a sentiment should be found in the
presence of stably recalcitrant regret, and in regretful agents being prone to
acting without thinking. First, to be in a state of recalcitrant regret would be
to be motivated to undo some previous action despite believing it to have
been the best available option. Do people excoriate themselves over things
they’ve done—in particular, things that have turned out badly—despite
knowing that there wasn’t anything mistaken about their decision-making?
Surely they do. Second, for regret to issue in acting without thinking would
be for it to move people, in the throes of regret, to punish themselves
counterproductively over their putative mistake, even to the extent of
harming themselves more than did the regretted error. This too seems
quite familiar.
Notice that because regret has obvious similarities with another senti-
ment, guilt, the two distinct states are often confused. An agent in the
throes of guilt also focuses on his action painfully and without bidding, but
guilt primarily functions to placate the anger of others. It has a social
function of reconciliation that regret does not. Although guilt is typically
glossed as being about wrongness, I actually think this too simple; but for
present purposes I will adopt the simpler gloss in order to avoid unnecessary
complications.21 Recalcitrant guilt would arise when someone feels guilty
despite thinking he has not acted wrongly. The lorry driver case is just such
an example. Guilt also motivates acting without thinking, the other
hallmark of a sentiment, when people are moved to make reparations
contrary to their overall set of beliefs and desires. Are guilt-ridden people
prone to apologize, for instance, even when so doing actually hurts the
wronged person even more? Again this seems all too familiar.22

21
I think guilt should instead be given a dual aspect interpretation, as about either
impersonal wrongness or personal betrayal. This dual-aspect view can explain the
rationality of guilt in situations like Agamemnon’s, where the demands of duty conflict
intractably with our deepest personal commitments and relationships. See D’Arms and
Jacobson (1994), Baumeister et al. (1994, 1995), Tangney and Dearing (2002).
22
While some of these cases call for depth psychological explanation, such as when
the confession is best seen as a further act of hostility rather than atonement, surely this is
not always the case.
112 Daniel Jacobson

One reason why guilt and regret are so often confused—or conflated by
way of an artificially constructed emotion like agent-regret—is that typically
when one feels guilty, one will also feel regret (though not vice versa). That
is to say, painful thoughts of having done wrong, which motivate the desire
to make reparations and thereby placate the anger of others, tend to be
accompanied by painful thoughts that one has made a mistake, which issue
in repudiation of the action and resolve to act differently in the future.
Nevertheless, guilt need not be accompanied by regret, since it seems
psychologically possible to think that one acted immorally without repudi-
ating what one did.
Because agent-regret conflates two distinct sentiments, it is a misleading
emotion category that obscures more than it illuminates. However, its
construction is not a mere mistake; it is motivated by the insights of the
agency and residue principles. Yet Williams’s attribution of agent-regret to
the lorry driver carries some tacit normative implications which, when
made fully explicit, point in a different direction than he suggests. He
aims to find an emotion that it would be rational for the lorry driver to feel,
but which cannot rationally be felt by a mere spectator to the event. I’ve
already granted the plausibility of Williams’s agency principle, which holds
that the driver can be expected to have bad feelings disproportionate to
those of a mere spectator. If we assume, further, that the difference between
agent and spectator is not merely a matter of degree but of what kind of
emotion they can be expected to feel—as Williams clearly implies by
contrasting agent-regret with the regret of the spectator—then his motiv-
ation for coining agent-regret becomes clear.
Calling the lorry driver’s emotion agent-regret suggests that he need not
be blameworthy in order to feel it rationally. According to Williams, one
can feel rational regret over actions with bad outcomes for which one is
merely causally responsible, in virtue of what one intentionally did—such
as driving the lorry blamelessly. And calling it agent-regret suggests that a
mere spectator cannot feel it rationally, since he could feel that only if he
thought himself to be a participant in the action (which really would be
insane). Agent-regret thus seems to fit the job description perfectly. Yet
I will argue in the following section that this is a specious victory. Nothing
important is gained by attributing agent-regret rather than guilt or dismay
to the lorry driver, and Williams’s (hedged) commitment to defending the
rationality of the lorry driver’s emotion is understandable and plausible but
ultimately proves misguided.
 Regret, Agency, Error 113

3. THE RATIONALITY OF REGRET

Williams does not argue for the rationality of the lorry driver’s regret so
much as peremptorily reject any conception of rationality that denies it.
Since there are several different respects in which philosophers have claimed
emotions to be appropriate—as rational, admirable, fitting, and so forth—
it isn’t immediately obvious what Williams thereby rejects. We might
differentiate between these forms of criticism by saying that an emotion
is unreasonable whenever there is most reason not to feel it, that it is not
admirable when it conflicts with what a virtuous person would feel under
the circumstances, and that it is unfitting when it misrepresents its object. It
is possible that an emotion could be fitting, in this sense, despite neither
being what a virtuous person would feel nor what one has most reason to
feel. Because cognitivist theories of the emotions posit a constitutive
thought to each, they have an obvious affinity for accounts of fittingness
in terms of the truth of that proposition. Although I here follow Williams
and Stocker in speaking of the rationality of emotions such as regret, I take
them to be talking about fittingness. And I understand them to hold that an
emotion is irrational when its (supposedly) constitutive thought is false.23
Williams suggests that there would be something wrong with the lorry
driver if he felt merely what a spectator could feel over the tragedy for which
he was causally responsible. While acknowledging the likelihood of exces-
sive self-reproach in such situations, he insists that a driver should be prone
to an initial bout of bad feeling that expresses itself in the impulse to make
(even symbolic) reparations. But he also expresses some proper caution
about this conclusion. Williams (1976: 28) writes:
Doubtless, and rightly, people will try, in comforting him, to move the driver from
this state of feeling . . . to something more like the place of a spectator, but it is
important that this is seen as something that should need to be done, and indeed
some doubt would be felt about a driver who too blandly or readily moved to that
position.
I can accommodate Williams’s leading thought about this case, which
corresponds to the modest, empirical version of the agency principle
alongside an ethical norm endorsing those feelings. We can expect even
the blameless driver to feel worse than a spectator, and to want to make
some kind of restitution; moreover, in some sense he should have such

23
See D’Arms and Jacobson (2000). An additional reason not to use “rational” in this
sense arises from cases of misleading evidence, where the epistemic flavor of that term is
inapt. But we will not be considering such cases here.
114 Daniel Jacobson

feelings. Yet Williams wants to defend the driver’s emotion as rational

regret, and this motivates him to coin the notion of agent-regret so as to
construct an emotion whose constitutive thought is true of the driver but
not the spectator. The blameless driver can think “How much better if
I had done otherwise” without imagining that he is at fault, but the
spectator cannot think this—supposing that she could not have averted
the accident. But if the driver simply feels regret or guilt (or most likely
both), then his sentiment is irrational: in fact he made no mistake and did
nothing wrong. Nevertheless, I can accept that it should be endorsed as
admirable.
The first thing to note about Williams’s discussion of the case is that he
displays some ambivalence about what the driver should feel. While he
suggests that we would feel “some doubt” about a driver who was not
inclined to some participatory feeling, he also thinks that we should try to
talk him out of such feelings and into something more like what a spectator
would feel. Someone might hold this view for merely strategic reasons—
such as the consideration that there is nothing to be gained by blaming
himself—but I see no reason to attribute to Williams the mistake of
conflating the reasonable and the rational emotion. Instead, I suggest that
his core normative conviction is aretaic. One would have doubts about the
character of a driver who too easily adopted the attitude of a spectator, when
he was the causal agent of some disaster. People tend to beat a hasty retreat
from responsibility for bad outcomes, often under the cover of the passive
voice. In light of this dubious tendency, we should acknowledge that there
are grounds for self-scrutiny whenever one is the causal agent of a bad
outcome.24 An admirable person in a real-life situation would be prone
to search for grounds of self-reproach: he cannot help himself to the
stipulation that he is blameless.
In fact, Williams utilizes something very much like this thought when he
considers what someone should feel about cases of conflicting obligations.
There he (1965: 173) insists that the standards for “an admirable moral
agent cannot be all that remote from that of a decent human being.” He
goes on to claim, plausibly, that decent human beings will be disposed to
feel badly about doing what is best in some such situations, for instance
when that entails breaking a serious promise. I suggest an analogous
conclusion about the lorry driver case: a decent person who is causally
responsible for some disaster will be prone to guilt, at least initially, though
he should be movable to something more like the position of a spectator.
(And he will be so movable, if he is not subject to “self-punitive excess.”)

24
I take this to be the leading thought of Rosebury (1995) on moral luck.
 Regret, Agency, Error 115

Moreover, if this is what decent people feel, then any notion of what an
admirable agent will feel cannot deviate too far, on pain of becoming overly
idealized and unrealistic. As Williams (1965: 175) holds about cases of
conflicts of obligation,
A tendency to feel regrets . . . at having broken a promise even in the course of acting
for the best might well be considered a reassuring sign that an agent took his
promises seriously. At this point, the objector might say that he still thinks the
regrets irrational, but that he does not intend ‘irrational’ pejoratively: we must
rather admit that an admirable moral agent is one who on occasion is irrational.
This, of course, is a new position: it may well be correct.
But this is precisely my view of the lorry driver scenario. It is admirable for
the driver to feel guilt about the accident that is, strictly speaking, irrational.
The driver’s guilt attributes some grounds for blame to himself that are not
present, by stipulation—a stipulation to which the driver cannot help
himself. The driver’s initial tendency should be considered, to paraphrase
Williams, as a reassuring sign that he takes his agency seriously.25
This solution will remain elusive, however, so long as the different forms
of normative assessment are conflated with the term “rational” (or any
other word). Only by differentiating between the rational and the admir-
able can we explicate what Williams calls the non-pejorative sense of
irrationality. Moreover, once this distinction is made, it no longer seems
necessary to coin an emotion that the driver can rationally feel; hence there
is no need to coin agent-regret to do that job. And this, in turn, would
allow Williams to acknowledge that it is irrational guilt that we can expect
the driver to feel, not rational agent-regret. It also solves the problem noted
earlier, that his catch-all notion of agent-regret makes it mysterious why
that emotion has such disparate forms of expression in action: it sometimes
issues in the motive to make reparations, sometimes to change policy, and
other times to do neither. That can be easily explained once we notice that
the emotion is sometimes guilt and sometimes regret, and that it is not
always rational even when it is predictable, and indeed admirable, to feel.
Even if guilt can be rationally felt only over actions for which we
are responsible, it can be reliably predicted in other circumstances as well.
The most admirable sort of person—within the constraints imposed by a
realistic moral psychology—will initially be prone to unjustified guilt when
he is an innocent causal agent of disaster; but he will also be “moveable”

25
I want to thank Michael McKenna for pressing the point that the agency principle
does not operate solely under conditions of uncertainty. In my view, uncertainty is
sufficient for justifying some such feelings, but I do not hold that it is necessary. Much
will depend upon the context, and people can disagree about what virtue demands.
116 Daniel Jacobson

from that guilt in the face of sufficient evidence of his blamelessness.

Though I claim this guilt to be irrational, I have not adopted an insane
conception of rationality. It would be closer to the truth to say that I have
explicated the notion that Williams refers to, approvingly but opaquely, as
irrationality in the non-pejorative sense.
Suppose one accepts the distinction between sentiments and (mere)
emotions, and my account of regret and guilt as sentiments. One might
still wonder if there is an emotion that would be rational for the lorry
driver to feel but which would not be rational from a mere spectator.
After all, there are actions that are appropriate for the driver but not the
spectator, so why not feelings as well?26 Note first that, in my view, there
is a sentiment (guilt) that would be in some sense appropriate (that is,
admirable) for the driver to feel, at least initially—though I agree with
Williams that he should not remain in that state. Our thoughts about the
appropriateness of actions and especially feelings need to be focused; the
question is whether any such emotion is rational in the sense we have fixed
upon. My answer is yes and no. There is no such sentiment, I claim—not
regret, guilt, shame, or sadness (the most obvious candidates). Each of
these is either too narrow to be rational for the faultless lorry driver or too
broad not to be rational for the spectator. But there is an emotion that can
fit the bill—or at any rate one can be constructed. It will be useful to
consider this point further.
I have used the term “dismay” stipulatively to refer to an emotion
generically directed at bad outcomes, which can naturally be expressed
with the thought, “How much better if it had been otherwise.” By the
agency principle, we are more inclined toward dismay when we have some
salient causal role in the outcome, even if we made no error and thus are not
at fault. Consider a case where there can be no question of culpability,
unlike the lorry driver, because the outcome is so distant from one’s agency.
Suppose you have parked your car on the street outside your house, and it is
thrown by a tornado into your neighbor’s house, killing their child. You
will be dismayed by this event, as no doubt will other uninvolved neigh-
bors. The thought of how much better it would have been otherwise is
available to all of you. But the agential thought, “How much better if I had
done otherwise”—say, parked on the other side of the street—is available
only to you. Given the permissive stance I have taken toward the category
of emotions, we can use the constitutive thought methodology to coin an
emotion that only you can feel: call it agent-dismay. Hence there is an

26
I am indebted to Tim Scanlon for this way of putting the worry, which I find
particularly challenging. It has been extremely helpful to me to think through my
response to this challenge.
 Regret, Agency, Error 117

emotion that you can feel rationally but the mere spectator cannot. Yet this
constructed emotion explicates nothing. We might just as well have said
that although both you and the spectator can be dismayed at this terrible
event, only you can (rationally) be dismayed while having the thought that
it would have been better had you parked elsewhere.
Contrast dismay with regret. Dismay has no coherent motivational
component; there is no characteristic action-tendency common to its
disparate instances. One can be dismayed about all sorts of misfortune,
and one will act and express such dismay very differently depending on the
context. These actions and expressions will be explicable in terms of the
underlying sentiment producing them: if I feel guilty, I will act on my
dismay in one way; if I am ashamed, in another; and if I am angry, in an
altogether different manner. Moreover, when dismay is recalcitrant, it will
be recalcitrant as a sentiment: as regret, guilt, shame, or anger. And when it
issues in acting without thinking, the action motivated will again be the
characteristic motivation of some particular sentiment. Of course what
I have been calling agent-dismay is precisely Williams’s notion of agent-
regret. But the dispute over what to call this emotion is not mere semantics,
because calling it regret obscures what is distinctive about regret as a
sentiment: that it focuses on our errors and motivates policy change.
This discussion has also aimed to illuminate why, although it is cheap to
construct new emotions by their constitutive thoughts, that approach does
not advance our understanding.

4. CONFLICTS OF VALUE AND RESIDUAL REGRET

Thus far I have argued for a particular conception of regret as a sentiment

that concerns error, and I have focused on the agency principle and
Williams’s first paradigm of agent-regret, the lorry driver. Now I propose
to turn to the residue principle and a variation on Williams’s second
paradigm, where I will draw an analogous conclusion. The second case
Williams uses to characterize agent-regret is Agamemnon’s tragic choice
between his daughter and his fleet, love and duty, personal and impersonal
obligation. My argument for diagnosing the conflicted agent’s residual bad
feeling as guilt is strongest when there are incompatible obligations in play,
as here, but I grant that Agamemnon might feel both guilt and regret
simultaneously. In order to avoid this complication, I will focus on conflicts
of value rather than of obligation, where guilt is not at issue. Suppose that
someone who made a hard choice regrets his decision despite thinking it the
best of his options. If my gloss of regret is correct then this emotion, like the
118 Daniel Jacobson

lorry driver’s, must be irrational though it might be appropriate in some

other sense. (We would not want a parent to be too ready to sacrifice his
child even for the best of reasons.) Such conflict of value cases present the
strongest objection to my gloss of regret in terms of error.
The objection is that my conception of regret cannot make sense of the
intuition, insisted upon by both Williams and Stocker (1990: 191), that in
such cases it is “rational to regret what clearly is to be done.” Recall that the
empirical version of the residue principle predicts that we will often feel
residual regret upon making hard choices between (somehow) competing
values. Situations where one has strong reasons for making each of two
incompatible choices can be taken to support an alternative conception of
regret, suggested by Stocker, on which it is an emotion that fundamentally
concerns not error but loss. When one makes the right decision but it turns
out badly, one loses out on something genuinely valuable, with the cold
comfort of having chosen correctly. And when one makes a hard choice,
where there are strong conflicting reasons, then we often painfully feel the
loss of those forgone goods. This much seems impossible to deny. But are
they cases of regret, and is it rational?
Consider the ramifications of any conception of regret broad enough to
vindicate such residual regret as rational. Stocker clearly finds the norma-
tive corollary of the residue principle intuitively obvious. “It is beyond
argument,” he (1990: 125) writes, “that the world as we know it gives
us grounds for regret and conflict even if we do what is to be done.” In
order to grasp his view of regret, one must appreciate his concern with
the psychological state of being conflicted: emotionally ambivalent. This
state of psychological conflict is not justified whenever there are reasons
on both sides of a choice, as is typical even of easy choices. Rather,
for Stocker (1990: 87) the distinction between “rational conflict” and
“conflict over unimportant matters” marks the crucial difference between
choices that should be emotionally conflicting and those where one
should simply forgo some good without regrets. He considers this the
rationality criterion for residual regret. Hence Stocker’s conception of
regret is fundamentally agential, but it focuses not on error but significant
loss of value.
Were regret understood as being about the kind of loss that often renders
us conflicted even about doing what clearly should be done, then obviously
it would sometimes be rational no matter what we do. Specifically, in
certain circumstances where you have chosen the greater good or the lesser
evil, it is nevertheless rational to feel regret. But in order for regret to be
rational over doing what clearly is to be done, Stocker (1990: 267) writes,
“[t]he agent must see something good in the lesser option not seen in the
 Regret, Agency, Error 119

greater, which gives an evaluative reason to do the lesser rather than the
greater.”27 This suggests a gloss of regret in terms of the thought, “How
much better, in some respect, if I had done otherwise”—which is exactly
the constitutive thought of agent-regret according to Williams.
We should not accept an account of regret that marginalizes the central
scenario in order to capture the peripheral, even if the latter are more
philosophically interesting. The gloss of regret in terms of this thought
threatens to do just that. Since the thought is true both when the agent errs
and when he makes the right choice in a hard case, the account assimilates
error and loss. This threatens the idea that regret serves as a sentiment
whose function involves learning from one’s mistakes, because cases of loss
without error should not motivate any change of policy. However we use
the term regret, we must not lose sight of this sentiment or fail to notice its
importance. Even in difficult choice situations, where we will be emotion-
ally conflicted no matter what we do, our aim is surely to make the right
decision. In deliberation we seek to avoid error and the sentiment con-
cerned with sanctioning errors, even if a painful emotion concerned with
loss is unavoidable.
Like Williams’s agency principle and the lorry driver, Stocker’s argument
moves very quickly and with great rhetorical insistence from the predictabil-
ity of an emotion to its rationality. Indeed, we ordinarily grant a presumption
of warrant to our actual feelings and, especially, to what we are stably prone
to feel. Hence the burden lies with those arguing against the rationality of
feelings to which almost everyone would be prone. Yet recall the lesson of the
lorry driver example. Even though we can predict that ordinary, decent
people will feel guilty over being the causal agent of disaster despite their
lack of culpability, there is an explanation ready-to-hand that undermines the
rationality of such predictable guilt. This explanation invokes the deeply
unrealistic idealization of such examples. Someone actually in the driver’s
unfortunate situation cannot be certain of his lack of culpability, let alone
stipulate it. The trouble with the lorry driver case is that it is designed to be
unusual and, in this crucial respect, deeply unrealistic.
A similar problem infects the dilemma cases. In ordinary life, we make
decisions under conditions of uncertainty. Even if the reasons I’m aware of

27
This is the core intuition behind Stocker’s (1996) argument from rational regret
for value pluralism. However, I do not think that the distinction between choices that
justify emotional conflict and conflict over unimportant matters tracks the distinction
between kinds of value. So although I accept value pluralism, I do not find this argument
for it compelling, because the sort of loss that can rationally be emotionally conflicted
can involve one value instantiated in qualitatively different ways or distributed to
different people. In just this respect, I agree with Hurka (1996a, 1996b), but see fn. 29.
120 Daniel Jacobson

favor buying the country house rather than renting in the city, one never
knows—the well could be about to run dry. The general problem with
much of this literature about conflicts and dilemmas is that it focuses on the
most extreme sort of cases and ignores the commonplace uncertainty under
which we make decisions. I suggest we look to more mundane scenarios
where we can expect to feel regret no matter what we do, such as buying a
house. Psychological realism here dictates that we can expect to vacillate
between buyer’s remorse and renter’s regret (as it were) without any new
evidence. This vacillation should not be mysterious, as it has an explicable
source. Our anxiety over such large but hardly heroic or tragic decisions
makes the problems with whatever is currently the likely outcome more
salient, as well as the attractions of the alternative likely forgone. When the
likelihoods change so does the salience, even when nothing else about the
choice alters.
This rather obvious point goes underappreciated in this literature. There
are many circumstances in which you can reliably expect to feel regret no
matter what you do, not because of some deep fact about value or morality,
but because of familiar human tendencies such as those present when
deliberating over buying a house. No important lesson for value theory
can be drawn from the all-too-human phenomenon of shifting salience of
goods and the anxiety produced by prospective regret.28 The fact that
you will feel regret no matter what you do must serve as a defeater of
the presumption of warrant one normally accords to one’s emotions and
dispositions to feel. If I will regret whatever I do, then it would be a mistake
to accord that regret any evidential weight with respect to my decision.
How can I know if it is ordinary regret over making the wrong choice or
residual regret over the right choice? Moreover, we can reliably predict
regret over the better choice when, for reasons that could not be foreseen,
the choice turns out badly. The recognizable human tendency to commit a
“bad outcome, therefore bad decision” fallacy also serves as a defeater in
many mundane cases concerning regret. These are strong reasons not to
accept the normative corollary vindicating the rationality of residual
regrets, even where we accept the residue principle predicting them.
Nevertheless, a profound insight underlies the examples Stocker and
Williams consider, which might be taken to motivate a broader conception
of regret than my error-focused gloss. Their insight is that there is an
ethically important difference between true sacrifices and mere forgone
goods. Consider this example drawn from Stocker. Suppose a man must

28
To be clear, some very interesting results in decision theory and experimental
economics might issue from such phenomena. But these are psychological not axiological
conclusions.
 Regret, Agency, Error 121

give up his favored occupation, to which he has committed himself—

perhaps it is philosophy—in order to support his family by some more
mundane but lucrative enterprise. It seems he may well look back on his
life and think he made the right choice, all things considered, yet feel a
deep sense of loss at the personal good sacrificed for the sake of his family.
By contrast, someone unable to satisfy her desire for a convertible because
her family needs a minivan, although she gives up something greatly
desired, would be shallow to dwell on the foregone good. This is not a
point about materialism, and it can be framed in other terms. But I’m
prepared to grant that the convertible could be a long-standing desire, a
potential source of considerable pleasure, and not in principle an
indefensible indulgence. Even so, to bemoan this hardly tragic sacrifice
appears less than admirable. There seems an important distinction
between significant loss and mere doing without, even when the choice
was for the best in both cases. But I contend that this point marks a
fundamentally aretaic distinction. There are sacrifices an admirable
person can regret honorably, and those he must accept gracefully as yet
another way in which the world does not bend to our wishes. If I am right
then no coherent conception of regret can track Stocker’s fundamental
distinction, because the difference between conflicting and insignificant
loss is not a matter of value but virtue.
Does my argument imply that no painful emotion over making the best
choice in a situation of value conflict can ever be rational? As with the lorry
driver, the answer is yes and no. No such sentiment is rational. The claim
that regret is rational in such cases assimilates making a better choice that is
not preferable in every respect with making a worse choice that has some-
thing going for it. In almost every choice situation there will be some reason
to choose the lesser good, so this view leads inevitably to a proliferation
of rational regrets.29 But what we care about in deliberation is making
the better choice: avoiding error. Even in cases where we can expect to be
emotionally conflicted no matter what we choose, this is the crucial concern
of the sentiment that deserves the name regret.
Stocker’s key insight—that we do feel conflicted when forced to make
some such choices and sacrifices, and it seems that in some sense we should

29
The counterargument given by Hurka (1996a, 1996b) looks like a reductio ad
absurdum of this claim. In his view, whenever you choose the best vacation out of ten
good choices, for example, you should rationally regret not choosing the other nine
(albeit not “to excess”). But any actual bad feeling about the unlimited number of lesser
but still good alternatives would be unseemly. Hurka’s (2001) view makes regret not an
emotion but an evaluative judgment, which is just what he does with love as well,
following Brentano (1969).
122 Daniel Jacobson

feel this way—does not apply to many cases where we sacrifice one kind of
value for another, simply because the good forgone is insufficiently import-
ant. Yet it does hold about some cases where the same kind of value is
at stake in both options, when the choices are qualitatively different
in important respects.30 To be seriously dismayed over genuine losses
of value is compatible with being an admirable person facing an
uncooperative world that demands sacrifice; whereas to be dismayed over
not getting everything you want, though a recognizably human tendency,
falls into the sphere of vice. As Stocker has aptly put the point: it makes you
a whiner. While we could stipulate that dismay is rational only in cases of
significant loss, this will be true by definition; moreover, the rationality
conditions of this residual emotion will be parasitic on an aretaic judgment
about the agent.
Williams’s lorry driver scenario and Stocker’s argument about conflicts
of value share certain virtues and vices. Keen psychological insight underlies
their core claims, particularly about human emotional propensities. These
insights are sometimes obscured, though, by an insufficiently discrimin-
ating theory of emotion. The failure to differentiate adequately between
regret, guilt, and dismay prevents them from drawing the right philosoph-
ical lesson from the psychological data. It is true that we can expect the
decent lorry driver to feel something different from a mere spectator to the
tragedy. He will feel guilt, even though his guilt is (by a stipulation
unavailable to him) unfitting. It is irrational guilt that an admirable person
can nevertheless be expected to feel, at least until he becomes convinced of
his blamelessness. Analogously, Stocker’s cases of conflicts of value reflect
the insight that bad feelings are often driven by uncertainty and anxiety
over our decisions. Sometimes they reflect a good forgone that deserves
tribute from our feelings, but is not best understood as regret. Here we can
say that bad feeling over making the best choice is either rational dismay
about some good forgone or irrational regret over a correct decision—or
perhaps some vacillation between the two.

30
We would need a coherent account of the distinction between value monism and
pluralism to make this argument rigorous, and I’m not sure that one can be found. See
Chang (2001) for doubts along these lines. It suffices for present purposes that different
varieties of sensual pleasure—which must be granted to be instances of the same type of
value if anything is to count as monism—can be sufficiently qualitatively different as to
justify some dismay over forgone choices. This much seems clear to me, though I am
reluctant to go into detail about cases.
 Regret, Agency, Error 123

REFERENCES

Baron, Marcia (1988). “Remorse and Agent-Regret.” In Midwest Studies in Philosophy

vol. XIII: “Ethical Theory: Character and Virtue,” eds. French, Uehling, and
Wettstein. (Notre Dame, IN: University of Notre Dame Press).
Baumeister, R. F., A. M. Stillwell, and T. F. Heatherton (1994). “Guilt: An
Interpersonal Approach.” Psychological Bulletin 115: 243–67.
—— —— —— (1995). “Interpersonal Aspects of Guilt: Evidence from Narrative
Studies.” In Self-Conscious Emotions: The Psychology of Shame, Guilt, Embarrass-
ment, and Pride, eds. June Tangney and Kurt Fischer. (New York: Guilford Press).
Bittner, Rüdiger (1992). “Is it Reasonable to Regret Things One Did?” The Journal
of Philosophy 89: 262–73.
Brentano, Franz (1969) [1889]. The Origin of our Knowledge of Right and Wrong,
ed. Roderick Chisholm. (London: Routledge & Kegan Paul).
Chang, Ruth (2001). “Value Pluralism.” In International Encyclopedia of the Social and
Behavioral Sciences,” eds. Neil Smelser and Paul Baltes. (New York: Elsevier Press).
D’Arms, Justin and Daniel Jacobson (1994). “Expressivism, Morality, and the
Emotions.” Ethics 104: 739–63.
—— —— (2000). “The Moralistic Fallacy: On the ‘Appropriateness of
Emotions.” Philosophy and Phenomenological Research 61: 65–90.
—— —— (2003). “The Significance of Recalcitrant Emotions (or, Anti-
Quasijudgmentalism).” In Philosophy and the Emotions, ed. Anthony Hatzimoysis.
(Cambridge: Cambridge University Press).
—— —— (2006). “Anthropocentric Constraints on Human Value.” Oxford
Studies in Metaethics 1: 99–126.
—— —— (2009). “Regret and Irrational Action.” In Reasons for Action, eds. David
Sobel and Steven Wall. (Cambridge: Cambridge University Press).
—— —— (2010). “Demystifying Sensibilities: Sentimental Values and the
Instability of Affect.” In The Oxford Handbook of Philosophy of Emotion, ed.
Peter Goldie. (Oxford: Oxford University Press).
—— —— “Guilt and Wrongness Reconsidered.” (Unpublished manuscript).
Deigh, John (1994). “Cognitivism in the Theory of Emotions.” Ethics 104: 824–54.
—— (2000). “Symposium: The Works of Martha C. Nussbaum: Nussbaum’s
Defense of the Stoic Theory of Emotions.” Quinnipiac Law Review 19: 293–307.
de Sousa, Ronald (1974). “The Good and the True.” Mind 83: 534–51.
—— (1987). The Rationality of Emotion. (Cambridge, MA: MIT Press).
Ekman, Paul (2003). Emotions Revealed. (New York: Henry Hold and Co).
Frank, Robert (1988). Passions Within Reason: The Strategic Role of the Emotions.
(New York: W. W. Norton and Co).
Frijda, Nico (1986). The Emotions. (Cambridge: Cambridge University Press).
Gibbard, Allan (1990). Wise Choices, Apt Feelings. (Cambridge, MA: Harvard
University Press).
124 Daniel Jacobson

Gigerenzer, Gerd and Peter Todd (1999). “Fast and Frugal Heuristics: The Adap-
tive Toolbox.” In Simple Heuristics That Make Us Smart, eds. Gigerenzer, Todd,
et al. (New York: Oxford University Press).
Greenspan, Patricia (1988). Emotions and Reasons: An Inquiry into Emotional
Justification. (London: Routledge).
—— (1992). “Subjective Guilt and Responsibility.” Mind 101: 287–303.
Griffiths, Paul (2004). “Is Emotion a Natural Kind?” In Emotion, Evolution and
Rationality, eds. P. Cruse.and D. Evans. (Oxford: Oxford University Press).
Hume, David (1978) [1739]. A Treatise of Human Nature, 2nd edn., ed.
L. A. Selby-Bigge. (Oxford: Clarendon).
Hurka, Thomas (1996a). “Monism, Pluralism, and Rational Regret.” Ethics 106:
555–75.
—— (1996b). “Reply to Critics.” BEARS Symposium, eds. Jamie Dreier and David
Estlund. <http://www.brown.edu/Departments/Philosophy/bears/homepage.html>.
—— (2001). Virtue, Vice, and Value. (New York: Oxford University Press).
Jacobson, Daniel (1997). “In Praise of Immoral Art.” Philosophical Topics 25:
155–99.
—— (2005). “Seeing by Feeling: Virtues, Skills, and Moral Perception.” Ethical
Theory and Moral Practice 8: 387–409.
Knobe, Joshua (2006). “The Concept of Intentional Action: A Case Study in the
Uses of Folk Psychology.” Philosophical Studies 130: 203–31.
Nussbaum, Martha (2001). Upheavals of Thought: The Intelligence of Emotion.
(Cambridge: Cambridge University Press).
Roberts, Robert (1988). “What an Emotion Is: A Sketch.” The Philosophical Review
97: 183–209.
Rorty, Amélie (1978). “Explaining Emotions.” In Explaining Emotions, ed. Amélie
Rorty. (Berkeley, CA: University of California Press, 1980).
Rosebury (1995). “Moral Responsibility and Moral Luck.” The Philosophical Review
104: 499–524.
—— (1980). “Agent Regret.” In Explaining Emotions, ed. Amélie Rorty. (Berkeley,
CA: University of California Press).
Solomon, Robert (1998). “On Emotions as Judgments,” reprinted in Not Passion’s
Slave: Emotions and Choice. (Oxford: Oxford University Press, 2003).
Stocker, Michael (1971). “ ‘Ought’ and ‘Can’.” Australasian Journal of Philosophy
49: 303–16.
—— (1990). Plural and Conflicting Values. (Oxford: Clarendon Press).
—— (1996). “Symposium: Monism, Pluralism, and Rational Regret.” BEARS
Symposium, eds. Jamie Dreier and David Estlund. <http://www.brown.edu/
Departments/Philosophy/bears/homepage.html>.
Tangney, June (1995). “Shame and Guilt in Interpersonal Relationships.” In Self-
Conscious Emotions: The Psychology of Shame, Guilt, Embarrassment, and Pride,
eds. June Tangney and Kurt Fischer. (New York: Guilford Press).
—— and Ronda Dearing (2002). Shame and Guilt. (New York: Guilford Press).
 Regret, Agency, Error 125

Taylor, Gabrielle (1985). Pride, Shame, and Guilt: Emotions of Self-assessment. (New
York: Oxford University Press).
Williams, Bernard (1965). “Ethical Consistency,” reprinted in Williams (1973a).
—— (1973a). Problems of the Self. (Cambridge: Cambridge University Press).
—— (1973b). “Against Utilitarianism.” In J. J. C. Smart and Bernard Williams,
Utilitarianism: For and Against. (Cambridge: Cambridge University Press).
—— (1976). “Moral Luck,” reprinted in Williams (1981).
—— (1979). “Conflicts of Values,” reprinted in Williams (1981).
—— (1981). Moral Luck. (Cambridge: Cambridge University Press).
—— (1993). Shame and Necessity. (Berkeley, CA: University of California Press).
 5
Phenomenal Abilities: Incompatibilism
and the Experience of Agency*
Oisı́n Deery, Matt Bedke, and Shaun Nichols

1 . B A C K G R O UN D

1.1 Introduction
Agents act. They buy detergent at the store, they go to work, they celebrate
holidays, they cheat on their taxes. Sometimes we hold agents morally
responsible for what they do, or what they fail to do, meting out credit
and blame as the occasion merits. In typical cases, when agents act they are
thought to have an ability to do otherwise. This is a point on which most
parties to the free-will debates agree. When it comes to characterizing the
ability to do otherwise and asking whether this ability is compatible with
determinism, however, there is no consensus.
In the ensuing debates, the experience as of having an ability to do
otherwise occupies a central role.1 Many libertarians, for instance,
maintain that the ability experienced is incompatible with determinism
(Campbell 1951; O’Connor 1995). Of course, some compatibilists have
challenged this idea (Mill 1865; Grünbaum 1952; Nahmias et al. 2004).
Despite the centrality of the phenomenology of agency in all this, there has
been strikingly little work on its characteristics. Of particular significance,
there is almost no empirical work on whether the experience of agency

* We thank Eddy Nahmias, Dana Nelkin, Paulo Sousa, David Shoemaker, Uriah
Kriegel, and Bertram Malle for comments on earlier drafts.We are also grateful to partici-
pants at the New Orleans Workshop on Agency and Responsibility (November 3–5, 2011)
and two anonymous reviewers at OUP for suggestions and comments.
1 For some of the literature on the ability to do otherwise, see Moore (1912); Berofsky
(2002); J. Campbell (2005); Perry (2004); Vihvelin (2004); Smith (2004); Fara (2008);
John M. Fischer (2008); Randolph Clarke (2009). The claim that moral responsibility
requires alternative possibilities has been disputed since Frankfurt (1969). However, it is
still widely contended that free will requires being able to do otherwise.
 Phenomenal Abilities 127

involves a phenomenology of being able to choose among alternative

possibilities or whether people take their agentive experiences to have
incompatibilist elements.2
This paper reports a series of experiments that investigates the phenom-
enology of agency. To anticipate, we found remarkably consistent results
across three sets of studies: participants regarded their experience of the
ability to do otherwise as incompatible with determinism. Now that we’ve
spoiled any suspense, we will locate the issue in the broader literature.

1.2. The Experience of the Ability to Do Otherwise

Let us characterize determinism as follows: a statement of the facts of the world
at an instant, together with a statement of the laws of nature, entail all truths
about the world, including those about future human actions.3 Granting that
we often feel that we have an ability to act other than we do, for present purposes
incompatibilists think that the experience as of an ability to do otherwise is
incompatible with determinism, while compatibilists think the opposite.
There are a number of influential “introspectors” on both sides of this
issue. John Searle is a representative incompatibilist:
[R]eflect very carefully on the character of the experiences you have as you engage in
normal, everyday human actions. You will sense the possibility of alternative courses
of action built into these experiences . . . that we could be doing something else right
here and now, that is, all other conditions remaining the same. This, I submit, is the
source of our own unshakeable conviction of our own free will. (1984: 95)
Similarly, Keith Lehrer has claimed that the incompatibilist “accurately
describes what I find by introspecting, and I cannot believe that others do not
find the same” (1960: 150). Even such a paradigmatic compatibilist as David
Hume (1960/1739) agrees with this sentiment when he writes, “There is a
false . . . experience . . . of the liberty of indifference” (Bk. II, Part III, }II).
The appeal to an incompatibilist phenomenology plays a particularly
important role in libertarianism. Many libertarians maintain both that we
experience our agency as incompatible with determinism, and that this
experience provides reason to think that our agency defies determinism.
C. A. Campbell writes:

2 One exception in the recent literature is a paper by Nahmias and colleagues (2004),
which we discuss in Section 1.3. Although we challenge their experimental results, we are
indebted to them for pioneering the investigation. We also draw on their scholarship in
setting out some of the historical statements below. See also Monroe and Malle (2010).
3 In the experiments below, we will present this idea in terms of causation to make it
more intuitive and accessible.
128 Oisı́n Deery, Matt Bedke, and Shaun Nichols

Everyone must make the introspective experiment for himself: but I may perhaps
venture to report . . . that I cannot help believing that it lies with me here and now,
quite absolutely, which of two genuinely open possibilities I adopt. (1951: 463)
Campbell goes on to argue that, unless we have good reason to doubt the
impression that “it lies with me” which of two possibilities I adopt, we
should accept the impression to reflect the truth. Timothy O’Connor
makes this move as well. First, O’Connor describes the character of the
experience of decision-making:
[T]he agency theory is appealing because it captures the way we experience our own
activity. It does not seem to me (at least ordinarily) that I am caused to act by the
reasons which favor doing so; it seems to be the case, rather, that I produce my
decision in view of those reasons, and could have, in an unconditional sense, decided
differently. (1995: 196)
Next, O’Connor says that we should take these experiences to reflect
something important about the nature of decision-making:
Such experiences could, of course, be wholly illusory, but do we not properly
assume, in the absence of strong countervailing reasons, that things are pretty
much the way they appear to us? . . . Skepticism about the veridicality of such
experiences has numerous isomorphs that, if accepted, appear to lead to a greatly
diminished assessment of our knowledge of the world, an assessment that most
philosophers would resist. (1995: 196–7)
A number of compatibilists have challenged the basic phenomenological
claim. These compatibilists deny that we experience our agency as incom-
patible with determinism. John Stuart Mill, for instance, writes,
Take any alternative: say to murder or not to murder. . . . If I elect to abstain: in what
sense am I conscious that I could have elected to commit the crime? Only if I had
desired to commit it with a desire stronger than my horror of murder; not with one
less strong. When we think of ourselves hypothetically as having acted otherwise
than we did, we always suppose a difference in the antecedents: we picture ourselves
as having known something that we did not know, or not known something that we
did know; which is a difference in the external motives; or as having desired
something, or disliked something, more or less than we did; which is a difference
in the internal motives. (1865: 285)
On Mill’s view, the feeling of the ability to do otherwise is always contin-
gent on our supposing that the situation prior to the decision was somehow
different. Adolf Grünbaum repudiates any incompatibilist element with
equal vigor:
Let us carefully examine the content of the feeling that on a certain occasion we
could have acted other than the way we did. . . . Does the feeling we have inform us
 Phenomenal Abilities 129

that we could have acted otherwise under exactly the same external and internal
motivational conditions? No, . . . this feeling simply discloses that we were able to act
in accord with our strongest desire at that time, and that we could indeed have acted
otherwise if a different motive had prevailed at the time. (1952: 672)
Grünbaum’s last sentence here gestures at the payoff of denying the phenom-
enological claim of incompatibilist agency: if Mill and Grünbaum are right,
then the feeling of being able to do otherwise is consistent with determinism,
and this would undercut a crucial motivation for libertarianism.
This situation might seem to be a dialectical stalemate (cf. Fischer 1994:
84). However, these philosophers are making general claims about the
nature of our experience of agency. These are empirical claims, and they
can be illuminated by taking up empirical methods.

1.3. Previous Work on the Phenomenology of Free Will

We are not the first to recommend a more systematic investigation that is
partly empirical. Nahmias and colleagues (2004) suggest that we find out
how people actually tend to describe their agentive experience (what they
call the phenomenology of free will), including their experience as of being
able to do otherwise:
Taking a cue from recent empirical work on “folk intuitions”, we think the best way
to understand the phenomenology of free will—if there is one—is to find out what
ordinary people’s experiences are like. If this is not possible, philosophers’ compet-
ing introspective descriptions will remain in yet another free-will stalemate. (164)
Nahmias and colleagues undertook this task in survey studies. Their studies
appear to lend some support to the idea that the phenomenology of agency
is compatibilist. However, we think the studies have significant shortcom-
ings, so let us briefly describe one of those studies, and then identify what
we find lacking.
In one study, Nahmias and colleagues pitted compatibilism and incom-
patibilism against each other directly. The study was based on “competing
libertarian and compatibilist accounts of our experience of the ability to
choose otherwise” (174). Their survey asked participants to imagine (or
recall) an experience of making a difficult choice:
Imagine you’ve made a tough decision between two alternatives. You’ve chosen one
of them and you think to yourself, “I could have chosen otherwise” (it may help if
you can remember a particular example of such a decision you’ve recently made).
Which of these statements best describes what you have in mind when you think,
“I could have chosen otherwise”?
A. “I could have chosen to do otherwise even if everything at the moment of
choice had been exactly the same.”
130 Oisı́n Deery, Matt Bedke, and Shaun Nichols

B. “I could have chosen to do otherwise only if something had been different

(for instance, different considerations had come to mind as I deliberated or
I had experienced different desires at the time).”
C. Neither of the above describes what I mean. (2004: 175–6)
The majority of the participants gave the response that fits with compati-
bilism (i.e. B).
While this study is clearly focused on an issue that divides compatibilists
and incompatibilists, there are a number of limitations to the study. First,
participants are told to think of a decision and then told to think something
else about the decision: that they could have done otherwise. It is thus unclear
whether their initial recollection actually carried with it a sense of an ability
to do otherwise. So if people make compatibilist judgments about these
decisions, it might be because they are considering cases in which the
phenomenology of the ability to do otherwise is absent.
Second, the key question is about experiences sometime in the past,
rather than present-focused experiences where the phenomenology of
agency is actually present and thus presumably more accessible.
Third, Nahmias and colleagues asked participants about difficult decisions,
and this presents the opportunity to interpret “could have done otherwise” in
confounding ways. Consider Martin Luther’s decision to renounce his
writings or be declared an outlaw and heretic. Legend has it that, after
praying and consulting with advisors for a day, he said, “Here I stand.
I can do no other,” thereby reaffirming his writings. Luther might have
chosen B in Nahmias’s survey. But if he did, we should not conclude that
there is no sense of “could have done otherwise” that captures some aspect of
Luther’s phenomenology and that is incompatible with determinism. For
Luther could have responded as he did to express his commitment to his
cause, a commitment that would only change if the considerations before
him and his reasons for breaking with the Roman Catholic Church presented
themselves differently. This commitment-expressive meaning of “could not
have done otherwise” is consistent with other senses of “could have done
otherwise”—consider whether Luther thought it was up to him whether to
renounce his views—that might or might not be incompatible with deter-
minism. Difficult decisions are subject to confounds like this, so the above
survey does not cleanly address the question whether there is some aspect of
the phenomenology of agency that is in tension with determinism.
Fourth and last, it is not clear whether the participants really understand
the intended meaning of “even if everything at the moment of choice had
been exactly the same” or “only if something had been different.”
We wanted to run more comprehensive studies that fix these short-
comings. The result was the following three studies, which share a common
 Phenomenal Abilities 131

structure. First, participants were asked whether they had an experience as

of the ability to do otherwise when faced with a simple decision. Next, they
were given a description of determinism. Of course, we did not use the term
“determinism”, since that might have conjured up unwanted associations
in participants. Rather, we used a technical term—“causal completeness”.
To address concerns about comprehension of the materials, we used the
familiar psychological technique of training to criterion, where we asked a
series of questions that tested and, if necessary, corrected, the participant’s
understanding of determinism. Participants who passed the training were
asked about the compatibility of their experience with determinism. In
study 1, this question focused on both a first-person, present-focused
experience in a hypothetical deliberative context and a past-focused judg-
ment about such a situation. In study 2, we explored the phenomenology
of actual rather than imagined choices. In study 3, we tested whether
epistemic phenomenology—the phenomenology of uncertainty—feels
incompatible with determinism.

2. S T UD Y 1

2.1 Overview
In our first study, we had participants imagine a decision about whether to
go left or right on a sled. In one condition, the sledding scenario was set in
the future; in the other condition, the scenario was set in the past. After
reading the scenario, participants in condition 1 were asked whether they
had a feeling of an ability to do otherwise; participants in condition 2 were
asked for a retrospective judgment about whether they could have done
otherwise. Participants who affirmed feeling (or having) an ability to do
otherwise were directed to the training section in which causal complete-
ness (i.e. determinism) was explained to them. Participants who passed the
training were reminded of their affirmation regarding the ability to do
otherwise and asked about consistency with causal completeness.
Our prediction was that when asked about the phenomenology of
imagined decision-making, participants would tend to affirm a feeling of
an ability to do otherwise and also regard this feeling as incompatible with
determinism; but when asked for a retrospective judgment about the ability
to do otherwise, we predicted that participants would be less likely to treat
the ability to do otherwise in an incompatibilist way.4

4 See e.g. van Inwagen 1983 (8–13) for an overview of other uses of “can.”
132 Oisı́n Deery, Matt Bedke, and Shaun Nichols

Method:
Participants:
Eighty-four participants were initially recruited online through the Mech-
anical Turk (MTurk) website.5 The survey itself was conducted using
SurveyMonkey. Two participants did not complete the survey. They were
excluded from the analysis.

Materials:
Each condition had three parts.

Part 1: The ability to do otherwise Participants were presented with a

vignette and a question about the ability to do otherwise. For condition 1,
this went as follows:
Please read the following passage, and answer the questions that follow as best you
can:
Imagine that you are sledding down a snowy path on a mountainside. Your sled has
a steering mechanism that allows you to control the direction of the sled. Below you
is a fork in the path with snow built up in the middle, and you can tell that, if you
don’t direct your sled one way or the other, the contours of the mountain will
channel you and your sled either to the left or to the right.
Ability Question
Consider how things seem to you as you approach the fork in the path. In particular,
consider what it’s like to decide which way the sled will go.
Please indicate your level of agreement with the following statement:
When deciding which way the sled will go, it feels like I can either go to the left or
go to the right.
Participants were asked to indicate their agreement with this statement on a
7-point scale (1 = disagree completely; 7 = agree completely).
For condition 2, the vignette was the same except that participants were
asked to imagine that the sledding episode occurred many years ago. And
instead of getting a response regarding a phenomenology of the ability
to do otherwise, we asked them to indicate agreement with a statement
about a past ability to have done otherwise: “I could have gone right
instead of left.”

5 MTurk is a website supported by Amazon.com <https://requester.mturk.com/

mturk/welcome> that provides users the opportunity to fill out surveys for modest
compensation. Recent work indicates that the data gathered through MTurk is at least
as reliable as that gathered through standard psychology pools composed of undergradu-
ates (see Buhrmester et al.).
 Phenomenal Abilities 133

Part 2: Training on determinism We wanted to focus on participants who

had a phenomenology of the ability to do otherwise, so only participants
who indicated a positive level of agreement to the first questions (5 or
higher) were directed to the training section. Here, participants were given
a detailed explanation of causal completeness, summed up as follows:
“According to causal completeness, everything that happens is fully caused
by what happened before it. This is true from the very beginning of the
universe, so what happened in the beginning of the universe fully caused
what happened next, and so on right up until the present. Causal com-
pleteness holds that everything is fully caused in this way, including
people’s decisions.”
Participants were then given two kinds of cases to illustrate the phenom-
enon. In one case, they were told that an earthquake fully caused the
volcanic eruption at Mt St Helens,6 and they were then told, “According
to causal completeness, if we could somehow replay the entire past right up
until St Helens erupted on May 18, 1980, then St Helens would once again
erupt at that time. Another way to put this is to say that all the events
leading up to the eruption made it so that the eruption had to happen.” In
another case past events, feelings and beliefs led to Obama’s decision to
pick Joe Biden as his running mate, and participants were told “According
to causal completeness, if we replayed the past right up until Obama’s
decision—including everything that was going through Obama’s mind—
then Obama would once again make exactly the same decision. That is, all
the events leading up to Obama’s decision (including everything that was
going through Obama’s mind), made it so that it had to happen that
Obama would pick Biden.”7
We then tested comprehension of causal completeness. We first asked
participants to indicate whether the following was true or false:
According to causal completeness, St Helens would have erupted on May 18, 1980
even if there had been no earthquake.
Participants who answered “True” (the incorrect answer) were corrected,
and given an explanation of the right answer. These participants were then

6 This is an oversimplification of the geological facts which we adopted to ease the

load on participants.
7 In defining determinism—our causal completeness (CC)—as meaning “everything
that happens is fully caused by what happened before it,” some might think this
consistent with certain indeterminist conceptions of causation. But to say that events
are fully caused is meant to avoid this reading—being fully caused suggests that nothing
extra-causal is needed to help settle events. Our examples aid the preferred interpretation.
We say, e.g. that under CC, “it had to happen that Obama would pick Biden.”
134 Oisı́n Deery, Matt Bedke, and Shaun Nichols

given a similar question to see if they had absorbed the training. If they
answered incorrectly yet again, they did not move on to answer the
compatibility question, as they were deemed to have insufficient compre-
hension of causal completeness.
Participants who passed this first kind of question either on the first or
second try were given another true/false question to test for comprehension:
According to causal completeness, if a week from now Barack Obama decides to
have soda with dinner, all the events leading up to that decision will make it the case
that he has to decide to have a soda with dinner.
Our objective here was to test for and correct overly weak interpretations of
causal completeness. Those who answered “False” (the incorrect answer)
were corrected and given another chance at a similar question. If they
answered incorrectly yet again they failed the training and did not answer
the compatibility question. Participants who passed both kinds of questions
either on the first or second try were deemed to have adequate comprehen-
sion of determinism, and these participants moved on to the third part of
the study, the compatibility question.

Part 3: Consistency After successful completion of the training, in both

conditions participants were told to recall their agreement with the state-
ment regarding the ability to do otherwise (from Part 1 of the survey). For
example, in condition 1, they were told:
Now, recall that you previously agreed with the following statement:
When deciding which way the sled will go, it feels like I can either go to the left or
go to the right.
Following this, they were asked the compatibility question. In condition 1,
this read as follows:
Compatibility Question
Considering this previous statement and your understanding of causal complete-
ness, please indicate your level of agreement with the following:
Even though it felt like I could either go to the left or go to the right, if causal
completeness is true there is something mistaken about how that decision felt to me.
In condition 2, the compatibility statement was:
Even though I said I could have gone right instead of left, if causal completeness is
true there is something mistaken about what I said.
Agreement was indicated on the same 7-point Likert scale as was used for
the Ability Question, and an answer above 4 was taken to be an incom-
patibilist answer.
 Phenomenal Abilities 135

Results:
Of the thirty-four participants who started condition 1, thirty-three com-
pleted it. Of these, thirty-one indicated a phenomenology of an ability to
choose among possibilities and all but four of them passed the training
section.8 The remaining twenty-seven participants gave a mean response
of 4.93 on the compatibility question, which differed significantly from
the midpoint of the scale, t(26) = 2.65, p = .014. That is, participants
tended to interpret their agentive experience as being incompatible with
determinism.
In condition 2, of the fifty participants who started the survey, forty-nine
completed it. Of these, forty-seven indicated an ability to choose among
possibilities and all but two of them passed the training section.9 The
remaining forty-five participants gave a mean response of 5.24 on the
compatibility question, which differed significantly from the midpoint of
the scale, t(44) = 5.05, p < .001.
A t-test comparing conditions 1 and 2 showed no significant difference,
p = .448. So participants tended to be just as incompatibilist about
retrospective judgments of their ability to do otherwise as they are
about their current experience as of being able to do otherwise. This
first study provides evidence that people do indeed judge that their
experience of deciding is inconsistent with determinism, in the sense
that the experience is somehow mistaken or nonveridical if determinism
is true. It also suggests that the effect is robust across retrospective and
present-focused cases.

3. S T UD Y 2

3.1 Overview
One major limitation of study 1 is that it involved merely imagined
choices. This inserts a distance between the actual phenomenology of
decision-making and judgments about that phenomenology. As a result,
in study 2 we introduce conditions in which agents actually make decisions.
In addition, study 1 focused on decisions that have no moral weight. We

8 Nine participants required correction, and ultimately passed the training section.
The responses of those who required correction did not differ from those who answered
correctly without training (p >.2).
9 Seven participants required correction, and ultimately passed the training section.
Again, there were no differences between those who required correction and those who
didn’t.
136 Oisı́n Deery, Matt Bedke, and Shaun Nichols

thus added a condition in study 2 in which the decision does have a moral
element. So this study comprises three conditions to test for any effect from
actual choices or from morally salient choices. We also introduced two
innovations to the study’s design.
First, we wanted to make our vignette more “choicey”. It struck us that in
situations such as sledding down a hill often we don’t have a salient
experience as of deciding which way to go. We just go one way or the
other. Second, we wanted to address a potential worry about our use of the
word “mistaken” in the compatibility question. For example, in condition
1 from our first study, we asked:
Even though it felt like I could either go to the left or go to the right, if causal
completeness is true there is something mistaken about how that decision felt to me.
(Emphasis added.)
One worry about this wording was that participants might misinterpret
it as asking whether they were mistaken in thinking that their experiences
felt a certain way, rather than as asking whether there would be something
mistaken about the content of their felt experiences.10 Our solution was to
replace the above wording with a wording of the following form:
Even though it felt like I could either choose to X or choose to Y, if causal
completeness is true then I couldn’t really have chosen differently than I did.11
With these modifications, condition 1 presented participants with an
imagined choice among two very similar charities, condition 2 presented
participants with an actual choice among two similar charities, and condi-
tion 3 presented participants with an actual morally salient choice among
two charities, one for endangered trees, another for children’s cancer
treatments.

10 Thanks to Lucas Thorpe for this objection.

11 Two reviewers worried that, with causal completeness in mind (earlier described in
terms of events that “had to happen”) participants fix on one reading of the modal
“couldn’t really have chosen differently” and on that reading they give an “incompatibi-
list” response, whereas the description of their phenomenology might invoke a different
reading of the modal that would not merit an incompatibility response. Of course, the
key issue for us is whether participants feel their phenomenology wouldn’t be veridical if
CC were true. The question in study 3 is designed to first refer to the participants’ reports
on their phenomenology—that it felt like they could choose X or choose Y—and we
think this helps subjects focus on that modal content and whether it would be
veridical if CC were true. Further, in study 1 we ask the compatibility question using
different language that avoids this worry. We get the same incompatibilist results there.
 Phenomenal Abilities 137

Method:
Participants:
One hundred and fifty-five participants were initially recruited online
through the Mechanical Turk (MTurk) website. The survey itself was
conducted using SurveyMonkey. Twenty-one participants did not com-
plete the survey, or indicated that they had recently taken a “very similar”
survey.12 They were excluded from the analysis.

Materials:
As in study 1, this study had three parts.

Part 1: The ability to do otherwise For condition 1 of this study, we asked

participants to imagine deciding between two charities for endangered trees.
Imagine that you have $0.50 to donate. You have two options:
Donate to a foundation that protects the endangered tree Castanea Dentata.
OR
Donate to a foundation that protects the endangered tree Ulmus Dentata.13
These are your only two options.
Condition 2 was similar except that participants were given an actual
choice. Participants were told that they had $0.50 to donate to one of the
two tree charities. We informed participants (truly) that we would actually
donate this money to whichever charity they chose. Participants read:
You have $0.50 to donate. We, the researchers, will actually donate this money for
you whichever way you decide.
Participants were then presented with the same option language as in the
imagined condition, and each option appeared as a radio button at the
bottom of the page.
Finally, in condition 3 we presented participants with a morally salient choice
between a foundation that protects the tree Castanea Dentata and The Child-
hood Cancer Foundation,14 on the assumption that people tend to think that
saving dying children has greater moral weight than saving endangered trees.

12 Studies 2 and 3 were run after study 1, and some of the conditions in studies 2 and
3 were run serially, so we excluded participants who indicated they had taken a very
similar survey to minimize the influence of having previously taken one of our surveys.
13 Castanea Dentata and Ulmus Dentata are the names of the American Chestnut and
the American Elm, respectively. They are endangered species in North America. The
charities we used were The American Chestnut Foundation <http://www.acf.org/>, and
Trees Winnipeg: Coalition to Save the Elms <http://www.savetheelms.mb.ca/>.
14 <http://www.candlelighters.ca>.
138 Oisı́n Deery, Matt Bedke, and Shaun Nichols

In all conditions, after being given the imagined or actual choice,

participants were asked a question about the ability to do otherwise. For
instance, in condition 3, participants were asked to indicate their level of
agreement (on a 7-point scale) with the following statement:
When deciding which option to choose, it feels like I can either choose to donate to
the endangered tree Castanea Dentata or choose to donate to the Childhood Cancer
Foundation.15
(In conditions 2 and 3, participants were subsequently required to make a
choice between the charities.) As in study 1, only participants who agreed
with the ability-to-do-otherwise statement proceeded to the training.

Part 2: Training on determinism The training section was the same as that
used in study 1, and once again only those who passed the training
proceeded to the compatibility question.

Part 3: Consistency The compatibility question was adapted for the new
cases. For example, in conditions 1 and 2, participants were asked to
indicate agreement (on a 7-point scale) with this statement:
Even though it felt like I could either choose to donate to Castanea Dentata or
choose to donate to Ulmus Dentata, if causal completeness is true then I couldn’t
really have chosen differently than I did.

Results:
Of the fifty participants who started condition 1, forty-two completed it
and had not recently taken a very similar survey (three had). Of these,
thirty-eight indicated a phenomenology of an ability to choose among
possibilities and all but three of them passed the training section.16 The
remaining thirty-five participants gave a mean response of 5.60 on the
compatibility question, which differed significantly from the midpoint of

15 We want to forestall a potential concern about the phrasing here. Suppose I am

determined to choose p. It follows that I can choose p. And you might think it further
follows that I can choose p or choose q, for this follows from the simple logical principle
of disjunction introduction. In that case the ability to choose p or choose q is clearly
compatible with determinism. However, participants report an incompatibilist phenom-
enology as of an ability to choose p or q, which suggests that they are not reading “can
choose p or can choose q” in this compatibilist way.
16 Twelve participants required correction and successfully passed the training section.
There was a significant difference in responses between those who required correction and
those who didn’t. Those who required correction reported that their phenomenology was
incompatible with causal completeness (M = 4.82) but to a lesser degree than those who
answered these questions correctly the first time (M = 5.96), p = .038.
 Phenomenal Abilities 139

the scale, t(34) = 6.08, p < .001. The results of an imagined choice are
consistent with the results of condition 1, study 1, if not stronger by virtue
of the more “choicey” vignette.
In condition 2, of the forty-eight participants who started the survey,
forty-two completed it and had not recently taken a very similar survey (four
had). Of these, thirty-nine indicated a phenomenology of an ability to choose
among possibilities and all but two of them passed the training section.17
The remaining thirty-seven participants gave a mean response of 5.78 on
the compatibility question, which differed significantly from the midpoint
of the scale, t(36) = 6.85, p < .001. That is, participants were again
incompatibilist about the phenomenology, this time of an actual choice.
In condition 3, of the fifty-seven participants who started the survey, fifty
completed it and had not recently taken a very similar survey (three had).
Of these, forty-three indicated a phenomenology of an ability to choose
among possibilities and all but three of them passed the training section.18
Most of the remaining forty participants (90 percent) opted to donate to
the Childhood Cancer Foundation, as we expected on the assumption that
the cancer charity would be regarded as more morally salient. The forty
participants gave a mean response of 5.85 on the compatibility question,
which differed significantly from the midpoint of the scale, t(39) = 7.66,
p < .001. Once again we find incompatibilist phenomenology—this time
with a morally salient choice.
ANOVA testing showed no overall effect of condition among conditions
1, 2, and 3, F(2, 111) = .254, p = .776. So there appears to be no effect
produced by making the condition an actual choice, or by making the
choice morally salient.
The results of study 2 show that people report incompatibilist phenom-
enology of agency for actual choices. Indeed, whether the decision is set up
as an imagined one or an actual one does not affect the degree to which
participants interpret their agentive experience as being incompatible with
determinism. The results also show that whether or not the decision is
morally salient doesn’t affect the degree to which participants interpret
their agentive experience as being incompatible with determinism. So the
results of previous studies seem to extend to the moral domain, where issues
of responsibility loom large.

17 Six participants required correction and successfully passed the training section.
There was no difference between the responses of those who required correction and
those who didn’t (p > .2).
18 Six participants required correction and successfully passed the training section.
Again, we found no difference between the responses of those who required correction
and those who didn’t (p > .2).
140 Oisı́n Deery, Matt Bedke, and Shaun Nichols

4. S T UD Y 3

4.1 Overview
One possible concern with our previous studies stems from the way we have
phrased the key compatibility question. Notice our use of an “even though”
locution in the following:

Even though it felt like I could either choose to donate to Castanea Dentata or
choose to donate to Ulmus Dentata, if causal completeness is true then I couldn’t
really have chosen differently than I did. (Emphasis added.)
Although we took this to be a natural phrasing of the question, one might
think that “even though” primes the participant to agree with the state-
ment, which in this case is an incompatibilist response. Our final study
drops this potentially troublesome phrase and also tests whether the phe-
nomenology of epistemic uncertainty differs from the phenomenology of
being able to do otherwise in terms of compatibility with determinism. In
condition 1, we once again presented participants with an actual choice
among two options and tested whether they would continue to report
having an incompatibilist phenomenology as of being able to do otherwise.
In condition 2, we focused on epistemic phenomenology.

Method:
Participants:
One hundred and six participants were initially recruited online through
the Mechanical Turk (MTurk) website. The survey itself was conducted
using SurveyMonkey. Fifteen participants did not complete the survey, or
indicated that they had recently taken a “very similar” survey. They were
excluded from the analysis.

Materials:
The vignette and first question for condition 1 read as follows.

Part 1: The ability to do otherwise In both conditions, participants were

told that they would have a chance to win 5 cents if they picked the right
button. The text went as follows:
At the bottom of this page, there are two buttons, labeled H and V. Each option is
currently available for you to choose. In a moment, we’ll ask you to choose just one of
them. For this survey, only one of the buttons will give you an extra $0.05 (as bonus
payment on MTurk) if you choose it. But we won’t tell you which button it is—you’ll
have to make a choice and find out.
But don’t decide just yet.
 Phenomenal Abilities 141

First, consider how things seem to you as you face your decision. In particular,
consider what it’s like to decide which option to choose.
In condition 1, participants were asked to indicate agreement (on a 1–7
scale) with the following statement:
When deciding which option to choose, it feels like I can either choose H or
choose V.
Condition 2 was the same except that we dropped the modal “can” and
asked participants to “consider what it’s like to wonder which option you’ll
choose.” Participants were then asked to indicate their level of agreement
with a statement about epistemic phenomenology:
When wondering which option I’ll choose, it feels like I don’t know for sure before
I select a button which button is the bonus button.
As in study 1, only participants who agreed with the initial statement
proceeded to the training.
The two available options—H and V—appeared at the bottom of the
screen, with a radio button representing each option. Participants were not
told whether they had chosen the bonus button (H) until after they had
answered the compatibility question.

Part 2: Training on determinism The training section was the same as that
used in study 1, and again participants only proceeded to the compatibility
question if they passed the training.

Part 3: Consistency The compatibility question was adjusted for the new
cases. In condition 1, participants were told:
Now, recall the button-choosing situation. You previously agreed with the
following statement:
When deciding which option to choose, it feels like I can either choose H or choose V.
Considering this previous statement about how things felt to you before your choice
and your understanding of causal completeness, please indicate your level of
agreement with the following:
If causal completeness is true, then I couldn’t really have chosen differently than
I did.
In condition 2, participants were reminded that they agreed with this
statement:
When wondering which option I’ll choose, it feels like I don’t know for sure before
I select a button which button is the bonus button.
142 Oisı́n Deery, Matt Bedke, and Shaun Nichols

They were then asked to indicate their level of agreement with the
following:
If causal completeness is true, then I knew for sure before I selected a button which
button was the bonus button.
Our aim was to test whether participants distinguish the sort of alternative
possibilities they reported themselves as experiencing in other conditions
from clearly compatibilist alternative possibilities, which have to do simply
with our ignorance of the future.

Results:
Of the fifty-three participants who started condition 1, forty-seven com-
pleted it and had not recently taken a very similar survey (two had). Of
these, forty-four indicated a phenomenology as of there being alternative
possibilities in the situation and all but three of them passed the training
section.19 The remaining forty-one participants gave a mean response of
5.34 on the compatibility question, which differed significantly from the
midpoint of the scale, t(40) = 4.54, p < .001. That is, participants once
again demonstrated a strong tendency to interpret their agentive experience
as being incompatible with determinism.
In condition 2, of the fifty-three participants who started the survey,
forty-four completed it and had not recently taken a very similar survey
(eight had). Of these, thirty-nine indicated a phenomenology of an ability
to choose among possibilities and all but one of them passed the training
section.20 The remaining thirty-eight participants gave a mean response of
2.66 on the compatibility question, which differed significantly from the
midpoint of the scale, t(37) = -5.23, p < .001. That is, participants tended
to regard their phenomenology of uncertainty about the future as
compatible with determinism. A t-test between conditions 1 and 2 showed
that results differed significantly between these two conditions, t(76) = 6.85,
p < .001.
This final study provides yet further evidence that people do indeed
judge that their experience of deciding is inconsistent with determinism, in
the sense that the experience is nonveridical if determinism is true. At the

19 Eight participants required correction, and passed the training section. There was a
significant difference in responses between those who required correction and those who
didn’t. Those who required correction reported that their phenomenology was incom-
patible with CC (M = 4.25) but to a lesser degree than those who answered these
questions correctly the first time (M = 5.60), p = .057.
20 Nine participants required correction, and passed the training section. There were
no statistically significant differences in responses between those who required extra
training and those who didn’t (p = .2).
 Phenomenal Abilities 143

same time, people tend to think that the feeling of not knowing what will
happen is perfectly consistent with determinism. This suggests an appro-
priate sensitivity to the fact that ignorance is not incompatible with
determinism.

5. GENERAL DISCUSSION

5.1 Incompatibilism
Our results have implications for several issues concerning free will. Per-
haps most importantly, our studies seem to vindicate the incompatibilist
descriptions of our experience as of being able to do otherwise suggested by
Campbell, O’Connor, and Searle. By the same token, our results run
counter to the compatibilist descriptions of our experience suggested by
Mill, Grünbaum, and Nahmias and colleagues. The design of our studies
left it open for participants to describe their experience as involving the
ability to do otherwise, while allowing them to interpret this ability
however they wished. The results indicate that the people in the population
we tested tended to judge that their experience was incompatible with
determinism.
The results also address a concern that has plagued recent work
on intuitions about free will. Nahmias and Murray (2011) contend
that people give incompatibilist responses in previous experiments simply
because people misunderstand determinism. This is an important concern.
But rather than merely testing to see whether people misunderstand
determinism, we attacked the comprehension issue directly by exploiting
the familiar technique of training to criterion. And we did not find any
widespread confusion of determinism and bypassing. Part 1 of our training
controls for confusion between determinism and fatalism, and the majority
of our participants reported that the accuracy of their experience as of being
able to do otherwise is inconsistent with determinism, correctly understood.
Across all our studies, the percentage of participants who didn’t make it to
the compatibility question due to failing the training section was small,
at 6.15 percent. When we look at those participants who answered part 1
of the training incorrectly—that is, at those who did initially confuse
determinism with fatalism, and who were directed to the follow-up
training question—the percentage was small compared with Nahmias
and Murray’s results: only 20.68 percent of participants initially made
this mistake. Of those who initially made the mistake, 85.71 percent
answered the follow-up training question correctly. Thus, fewer than 3
percent of participants continued to confuse determinism and fatalism after
144 Oisı́n Deery, Matt Bedke, and Shaun Nichols

training. And those who required correction did not respond in any
significant way differently from those who didn’t.21

5.2. The Ability to Do Otherwise

Much of the free-will debate, since at least Hobbes, has been about an
ability to do otherwise. One influential compatibilist thought is that the
notion of the ability to do otherwise should be understood in contrast to
constraint or coercion. The idea is that an agent is able to do otherwise just
in case, if she had chosen, or wanted, or tried to do otherwise, then she
would have done so (cf. Moore 1912). There are also recent versions of
such a “conditional analysis” of the ability to do otherwise. According to
Kadri Vihvelin (2004), for instance, an ability to act (or not to act, which is
simply to be able to act in another way) is analyzable along something like
the following lines: an agent can Ф at t1 (say, raise her hand at t1) just in
case were she to choose to Ф at t2, and her body stayed working normally
and nothing interfered with her, she would Ф at t2.22 In other words,
Vihvelin holds that “persons have abilities by having intrinsic properties that
are the causal basis of the ability” (2004: 438). So Vihvelin thinks that an
ability to act is a disposition, or a bundle of dispositions. And, as she points
out, “no one denies that dispositions are compatible with determinism”
(2004: 429). After all, even if determinism is true, glass is still fragile—i.e. it
has the disposition to break if struck.23
Other compatibilists embrace an epistemic reading of “can do other-
wise.” On this view, to maintain that I can go left or right is simply to note
that it is epistemically open whether I will go left or right. J. J. C. Smart
argues that this is a natural way to interpret the expression “could have done
otherwise” even outside the sphere of action. When I say, “the plate fell,
and it could have broken,” I am not, says Smart, committing myself to any
claim about determinism. Rather, what I am saying is that, before the plate
completed its fall, for all I knew, the plate would break (1961: 298).
Similarly, perhaps when I say that Oswald could have done otherwise, all

21 Again, with the exception of study 2, condition 1, and study 3, condition 1. (See
footnotes 16 and 19.)
22 Vihvelin’s exact formulation is as follows: “S has the ability at time t to do X iff, for
some intrinsic property or set of properties B that S has at t, for some time t´ after t, if S
chose (decided, intended, or tried) at t to do X, and S were to retain B until t´, S’s
choosing (deciding, intending, or trying) to do X and S’s having of B would jointly be an
S-complete cause of S’s doing X ” (2004: 438).
23 For similar accounts, see Smith (2004) and Fara (2008). Questions persist (see e.g.
Clarke 2009) about whether any “dispositionalist” account is an adequate analysis of the
ability to act, and thus of the ability to act otherwise.
 Phenomenal Abilities 145

I’m saying is that, before Oswald pulled the trigger, for all anyone knew, he
wouldn’t pull the trigger. If I’m merely making a claim about epistemic
possibilities, then there is no conflict with determinism.
By contrast, incompatibilists think being able to do otherwise (in the
relevant contexts) means being able to do something other than what one
does, all prior conditions (including one’s desires) remaining the same.
This ability is presumed to be a matter of fact, not something about our
epistemic access to facts.
At least insofar as the relevant notion of the ability to do otherwise is
reflected in the experience as of being able to do otherwise, our results
suggest that the compatibilist accounts fail. Across three studies, partici-
pants tended to interpret their agentive experience in terms of an ability to
do otherwise, and they interpreted that ability incompatibilistically. Con-
cerning the traditional compatibilist analysis, our results equally undercut
old and new versions. After all, we allowed participants to describe their
experiences as involving the ability to do otherwise or not, where they were
free to interpret this ability however they wished. Participants then judged
that this ability—the one they had been allowed to interpret however they
wished—was incompatible with determinism. The epistemic compatibilist
account is also undermined by these results. Participants gave compatibilist
judgments about the case of ignorance about the future (study 3, condition 2),
indicating that they do have an appreciation that the feeling of uncertainty
is consistent with determinism.
We conclude that the notion of “can do otherwise,” at least with respect
to one’s decisions, is naturally interpreted in ways that contravene the most
familiar compatibilist approaches in the philosophical literature. When
participants attend to their experience while they consider future events,
their usage of “can” tends to reflect a sense of metaphysical openness that is
incompatible with determinism.

5.3. Misinterpreting One’s Agentive Experience

Obviously, the fact that people interpret their agentive experience as
incompatibilist doesn’t show that people actually have an incompatibilist
ability to do otherwise. Terry Horgan argues that people might be mistaken
in their interpretation of their own phenomenology. He allows that people
might regard their agentive experience as incompatibilist:
When one attends introspectively to one’s agentive phenomenology, with its . . .
[representational] . . . aspects of freedom . . . and when one simultaneously asks
reflectively whether the veridicality of this phenomenology is compatible with causal
146 Oisı́n Deery, Matt Bedke, and Shaun Nichols

determinism . . . , one feels some tendency to judge that the answer to such compati-
bility questions is No. (Forthcoming)24
But Horgan notes that we must distinguish between the content of our
experience and the content of judgments. The former kind of content
Horgan dubs “presentational content,” and it
. . . is the kind that accrues to phenomenology directly—apart from whether or not
one has the capacity to articulate this content linguistically and understand what one
is thus articulating, and apart from whether or not one has the kind of sophisticated
conceptual repertoire that would be required to understand such an articulation.
(forthcoming)
By contrast, “judgmental content” is the kind of content associated with
linguistic articulations. Of course, we make judgments about our phenom-
enology, and so we can have judgmental content that aims to capture our
presentational content. The key point here is that it is possible for our
judgments about the (presentational) content of our experience to go awry.
That said, those judgments are at least prima facie evidence of the nature
of the presentational-cum-phenomenal content, we maintain, so we would
need some positive reason to think that participants have systematically
misinterpreted the nature of their phenomenology. Further, even if we
grant arguendo that the presentational content of agentive experience is (in
the first instance) compatible with determinism, and that reports to the
contrary count as mistaken interpretations, still, the fact that people judge
the experience incompatibilist would be significant. For one thing, when
considering how best to understand the notion of the “ability to do
otherwise,” in many cases what will be of primary importance is how
people think about their ability to do otherwise, and that is clearly judg-
mental. Second and more interestingly, judgmental content can feed back
into presentational content. It is well known that what one judges about a
situation can affect one’s perception of the situation. Horgan recognizes
this, and he notes that the distinction between presentational and judg-
mental content isn’t always sharp: “it may very well be that the two kinds of
content can interpenetrate to a substantial extent” (forthcoming). As a
result, even if the presentational content of agentive experience is, in the
first instance, compatibilist, that doesn’t mean that the presentational con-
tent remains compatibilist. It might be that the incompatibilist judgment
shapes the presentational content.25

24 For Horgan, this representational “aspect of freedom” is what we have been calling
the experience as of being able to do otherwise.
25 Note that Horgan’s notion of “presentational content” is not simply “raw feels”
with no propositional content. For everyone would concede that insofar as we have
 Phenomenal Abilities 147

5.4 Deliberation Compatibilism

A final issue that might be illuminated by our results is the debate over the
presuppositions of deliberation. Some philosophers have maintained that
deliberation carries with it a presumption of genuinely open possibilities
of an incompatibilist variety. Richard Taylor writes, “I cannot deliberate
about what to do, even though I may not know what I am going to do,
unless I believe that it is up to me what I am going to do” (1983: 38–9).
And this “up to me” is incompatible with determinism. Peter van Inwagen
makes a similar point: “[I]f someone deliberates about whether to do A or
to do B, it follows that his behavior manifests a belief that it is possible for
him to do A—that he can do A, that he has it within his power to do A—
and a belief that it is possible for him to do B” (1983: 155).
On the other side, we find “deliberation compatibilists,” who maintain
that deliberation contains no such presuppositions. Tomis Kapitan begins
his paper (which would become the locus classicus for deliberation compa-
tibilism) thus:
By deliberation we understand practical reasoning with an end in view of choosing
some course of action. Integral to it is the agent’s sense of alternative possibilities,
that is, of two or more courses of action he presumes are open for him to undertake
or not. (1986: 230)
Kapitan goes on to argue that the presumption of openness does not require
metaphysical openness, but only epistemic openness.26 A number of
philosophers have followed Kapitan in developing compatibilist accounts
of the presuppositions behind deliberation (e.g. Nelkin 2004, Pereboom
2008).
Insofar as deliberation compatibilism claims that deliberation is not as a
matter of fact experienced as having incompatibilist presuppositions, our

incompatibilist phenomenology, it must be presented at a level with greater conceptual

sophistication than is provided by raw feels. Horgan is explicit about the possibility of
rich conceptual resources being implicated in presentational content: “It is plausible . . .
that humans can have presentational contents the possession of which require (at least
causally) a fairly rich repertoire of background concepts that can figure in judgmental
states.” For instance, “One can have presentational experiences, for instance, as-of
computers, automobiles, airplanes, train stations” (Horgan, forthcoming).
26 According to Kapitan and other deliberation compatibilists, there are other condi-
tions, too. In particular, Kapitan maintains that deliberation carries a presupposition of
efficacy, which he characterizes roughly as follows: “an agent presumes that his Ф-ing is an
open alternative for him only if he presumes that he would Ф if and only if he
were to choose to Ф” (1986: 234). See also Pereboom (2008: 288). We leave this
complication aside since it doesn’t affect our point.
148 Oisı́n Deery, Matt Bedke, and Shaun Nichols

studies indicate this position is mistaken. This does not decide the dispute
concerning deliberation compatibilism, but it does show that we need to
distinguish three versions of deliberation-compatibilism:
(1) People’s beliefs about their current deliberations are compatible with
determinism;
(2) People’s beliefs about their current deliberations are not compatible
with determinism, but they can be adjusted to be compatible;
(3) People’s beliefs about their current deliberations are not, and cannot be
adjusted to be, compatible with determinism, but we can conceive of a
rational being whose beliefs about deliberation are compatible with
determinism.
Our results suggest that the first version of deliberation compatibilism is
false. People’s beliefs about their deliberations are incompatibilist. The
second version—that our actual experiences are incompatibilist but revis-
able—is an interesting possibility, but it remains an open question whether
it is possible to revise this aspect of our experience. Until we know more
about what generates the incompatibilist experience, it is hard to know
whether it can be modified. One possibility is that the incompatibilist
experience is generated in a way that is not cognitively penetrable (see
e.g. Bayne 2011). That is, it might be that even if we form the explicit high-
level belief that deliberation is theoretically compatible with determinism,
this will not eradicate our experience of our deliberation as incompatibilist.
The third version of deliberation compatibilism—that we can conceive of
rational creatures who deliberate as determinists—is not under any threat
from our results. But if it turns out to be impossible for us to be such
rational animals, that might undercut some of the interest of deliberation
compatibilism.

6 . C O N C LU S I O N

The experience as of being able to do otherwise has long been central to

debates about agency and free will. Libertarians appeal to this experience as
evidence that determinism is false; compatibilists reject the libertarian
accounts of the character of the experience. Despite the pivotal role of
experience in these arguments, the experience itself has received scant
attention. Our studies are an attempt to advance the issue. We find
consistently incompatibilist judgments about the nature of the experience
as of being able to do otherwise. This lends support to the phenomeno-
logical claim of libertarians, though we ourselves are not inclined to take the
 Phenomenal Abilities 149

phenomenology of indeterminism as evidence that agency isn’t deter-

mined. Our results also suggest that an incompatibilist interpretation of
the notion of “ability to do otherwise” is the best interpretation of that
notion, at least insofar as that notion is supposed to reflect our experience as
of being able to do otherwise. Finally, our results also speak to the presup-
positions of deliberation. What our studies indicate is that as a matter of fact
our experience of deliberation features metaphysical openness (that is
inconsistent with determinism). While this does not decide the dispute
between deliberation compatibilists and deliberation incompatibilists, it
does make salient the possibility that deliberation compatibilism requires
an account of deliberation that is explicitly revisionist with respect to our
actual experience of deliberation.

REFERENCES

Bayne, Tim (2011). “The Sense of Agency.” In F. Macpherson (ed.) The Senses.
(Oxford: Oxford University Press), 355–74.
Berofsky, Bernard (2002). “Ifs, Cans, and Free Will: The Issues.” In R. Kane (ed.) The
Oxford Handbook of Free Will. (New York: Oxford University Press), 181–201.
Buhrmester, M., T. Kwang, and S. Gosling. (2011). “Amazon’s Mechanical Turk:
A New Source of Inexpensive, Yet High-Quality, Data?” Perspectives on Psycho-
logical Science 6: 3–5.
Campbell, C. A. (1951). “Is ‘Freewill’ a Pseudo-problem?” Mind 60/240: 441–65.
Campbell, Joseph (2005). “Compatibilist Alternatives.” Canadian Journal of
Philosophy 35: 387–406.
Clarke, Randolph (2009). “Dispositions, Abilities to Act, and Free Will: The New
Dispositionalism.” Mind 118: 323–51.
Fara, Michael (2008). “Masked Abilities and Compatibilism.” Mind 117/468:
843–65.
Fischer, John M. (1994). The Metaphysics of Free Will. (Oxford: Blackwell
Publishing).
—— (2008). “Freedom, Foreknowledge, and Frankfurt: A Reply to Vihvelin.”
Canadian Journal of Philosophy 38/3: 327–42.
Frankfurt, Harry (2003/1969). “Alternate Possibilities and Moral Responsibility.”
In G. Watson (ed.) Free Will. (Oxford: Oxford University Press), 167–76.
Grünbaum, Adolf (1952). “Causality and the Science of Human Behavior.” Ameri-
can Scientist 40(4): 665–76.
Horgan, Terry (Forthcoming). “Causal Compatibilism about Agentive Phenomen-
ology.” For a festschrift for J. Kim, eds., M. Sabates, D. Sosa, and T. Horgan.
Hume, David (1960/1739). A Treatise on Human Nature, L. A. Selby-Bigge (ed.)
(Oxford: Clarendon Press).
Kapitan, Tomis (1986). “Deliberation and the Presumption of Open Alternatives.”
The Philosophical Quarterly 36/143: 230–51.
150 Oisı́n Deery, Matt Bedke, and Shaun Nichols

Lehrer, Keith (1960). “Can We Know that We have Free Will by Introspection?”
The Journal of Philosophy 57/5: 145–57.
Mill, John Stuart (1865). An Examination of William Hamilton’s Philosophy.
(Boston: William V. Spencer).
Monroe, Andrew E. and Malle, Bertram F. (2010). “From Uncaused Will to
Conscious Choice: The Need to Study, Not Speculate about People’s Folk
Concept of Free Will.” Review of Philosophy and Psychology 9: 211–24.
Moore, G. E. (1912). “Free Will.” From Ethics. (Oxford: Oxford University Press),
122–37.
Nahmias, Eddy, Stephen Morris, Thomas Nadelhoffer, and Jason Turner. (2004).
“The Phenomenology of Free Will.” Journal of Consciousness Studies 11/7–8:
162–79.
Nahmias, Eddy and Dylan Murray (2011). “Experimental Philosophy on Free Will:
An Error Theory for Incompatibilist Intuitions.” In New Waves in Philosophy of
Action, Aguilar, Jesús, Andrei Buckareff and Keith Frankish (eds.) (Basingstoke,
UK: Palgrave Macmillan), 189–216.
Nelkin, Dana. (2004). “Deliberative Alternatives.” Philosophical Topics 32: 215–40.
O’Connor, Timothy (1995). “Agent Causation.” In T. O’Connor (ed.), Agents,
Causes, and Events: Essays on Indeterminism and Free Will. (New York: Oxford
University Press), 173–200.
Pereboom, Derk (2008). “A Compatibilist Account of the Epistemic Conditions on
Rational Deliberation,” The Journal of Ethics 12/3: 287–306.
Perry, John (2004). “Compatibilist Options.” In J. Campbell, M. O’Rourke,
and D. Shier (eds.) Freedom and Determinism. (Cambridge, MA: MIT Press),
231–54.
Searle, John (1984). Minds, Brains, and Science. (Cambridge, MA: Harvard Uni-
versity Press).
Smart, J. J. C. (1961). “Free Will, Praise and Blame.” Mind 70: 291–306.
Smith, Michael (2004). “Rational Capacities.” In M. Smith, Ethics and the A Priori:
Selected Essays on Moral Psychology and Meta-Ethics. (New York: Cambridge
University Press), 114–35.
Taylor, Richard (1983). Metaphysics, 3rd edn. (Englewood Cliffs, NJ: Prentice
Hall).
Van Inwagen, Peter (1983). An Essay on Free Will. (Oxford: Oxford University
Press).
Vihvelin, Kadri (2004). “Free Will Demystified: A Dispositional Account.”
Philosophical Topics 32: 427–50.
 6
Reasons-Responsiveness, Agents,
and Mechanisms1
Michael McKenna

Many philosophers are convinced by Harry Frankfurt’s (1969) controver-

sial argument that moral responsibility does not require the ability to do
otherwise. According to Frankfurt, it is possible to construct an example in
which an agent acts freely and is morally responsible for what she does but
is unable to do otherwise. Frankfurt’s example featured Jones, who shot
Smith for his own reasons. As it happened, Black wanted Jones to shoot
Smith on his own, but covertly arranged things so that if Jones were about
to fail to do so, Black would manipulate Jones so that he (Jones) would
shoot Smith. Since Jones shot Smith on his own, Black remained dormant;
he played no role in Jones’s act. Jones, it seems, acted freely and was morally
responsible for doing so. Yet because of Black’s presence, Jones could not
have done otherwise. Hence, as the example involving Jones illustrates,
moral responsibility does not require the ability to do otherwise.2

1 This paper was written for the 2011 New Orleans Workshop on Agency and
Responsibility (NOWAR), organized by David Shoemaker. I would like to thank him
for arranging this excellent event. I delivered an earlier version of this paper to the
University of Texas at Austin Philosophy Department. I am indebted to various
members of the audience for their instructive comments and criticisms. Notable
among them were Sinan Dogramaci, Jeremy Evans, Alex Grzankowski, Bob Kane,
Chris Simpson, Nicole Smith, David Sosa, and Michael Tye. At NOWAR, I profited
from discussion with many in attendance, especially Michael Bratman, David Brink,
Sarah Buss, Stephen Kearns, Rahul Kumar, Dana Nelkin, Paul Russell, Tim Scanlon,
Josh Sheppard, David Shoemaker, (yet again) Nicole Smith, Jada Strabbing, and Gary
Watson. For further helpful advice, I would also like to thank John Martin Fischer, Ish
Haji, Terry Horgan, Al Mele, Shaun Nichols, Derk Pereboom, Carolina Sartorio, and
Brandon Warmke. Finally, I would like to thank two anonymous referees for Oxford
University Press.
2 For an examination of the further details offered in support of Frankfurt’s argu-
ment, see the introduction to Widerker and McKenna (2003), and also my (2008).
Among those who defend some variation of Frankfurt’s argument are Fischer (1994),
152 Michael McKenna

Granting that moral responsibility requires some kind of freedom, the

lesson that those convinced by Frankfurt’s argument should take from it is
that the kind of freedom exercised by Jones, whatever it comes to, is not to
be accounted for in terms of the ability to do otherwise. Hence, it is not to
be understood in terms of alternatives to the way an agent acts. It is, instead,
to be understood in terms of the source of her actions. It requires, as John
Martin Fischer (1994) has put it, attention to the actual-sequence of events
leading to action. Call this kind of freedom source freedom, and the kind
that is understood in terms of alternatives leeway freedom.3
Despite what many assume, Frankfurt’s conclusion is not the exclusive
domain of compatibilists. Source incompatibilists have accepted the conclusion
to Frankfurt’s argument and set out to develop an actual-sequence account of
free action, one that requires the falsity of determinism.4 Nevertheless,
compatibilists stand to gain more should Frankfurt’s argument turn out to
be sound. This is because incompatibilists have a powerful argument for the
conclusion that determinism is incompatible with the ability to do otherwise.
This argument, the consequence argument, strongly suggests that if free will
and moral responsibility require the ability to do otherwise, then determinism
is incompatible with both.5 Armed with Frankfurt’s argument, source
compatibilists can grant that determinism is incompatible with the ability
to do otherwise. But then they can argue that, even if it is, this is not
relevant to whether moral responsibility and (all of) the freedom required
for it are compatible with determinism. As a source compatibilist, this is how
I proceed.6
The project for the source compatibilist who relies upon the soundness
of Frankfurt’s argument is to offer a plausible actual-sequence account of
source freedom that satisfies two desiderata. First, it is of a sort that can be

Fischer and Ravizza (1998), Haji (1998), Hunt (2000), McKenna (2003), Mele (2006),
Mele and Robb (1998), Pereboom (2001), Stump (1996), Widerker (2006), and
Zagzebski (2000).
3 I write in terms of “source freedom,” while Fischer (1994), and Fischer and Ravizza
(1998) use the term “guidance control.” I regard these terms as amounting to the same
thing. Likewise, I prefer the expression “freedom to do otherwise” or “leeway freedom,”
whereas Fischer and Ravizza write in terms of “regulative control.”
4 For example, see Hunt (2000), Pereboom (2001), Stump (1996), Widerker (2006),
and Zagzebski (2000).
5 The consequence argument was first developed by Ginet (1966), and then
subsequently refined by van Inwagen (1975), and Wiggins (1973).
6 The person who deserves the credit for this way of framing the dialectic is Fischer
(1994). Fischer coins a very similar view “semicompatibilism.” Where I differ from him
is that, on the basis of (some version of) the consequence argument, I am prepared to
grant to the incompatibilist the incompatibility of the ability to do otherwise. Fischer
instead wishes to remain uncommitted on this point.
 Reasons-Responsiveness, Agents, and Mechanisms 153

present in a Frankfurt example; second, it can be shown to be compatible

with determinism. Although many source compatibilists are attracted to a
mesh or hierarchical theory of the sort that Frankfurt himself endorses, in
my estimation, such theories face insurmountable problems.7 In what
follows, I shall instead explore the prospects of accounting for source
freedom in terms of responsiveness to reasons. In so doing, I will pay
special attention to a distinctive approach thoughtfully developed by
Fischer and Mark Ravizza (1998) which focuses on the mechanisms of an
agent’s actions. As I shall explain, a mechanism-based approach appears to
be necessary in order to fit a reasons-responsive theory for Frankfurt
examples. Unfortunately, this approach comes at a high cost, since there
are serious problems with analyzing exercises of source freedom in terms of
an agent’s mechanisms of action. Ultimately, I shall seek an account of
reasons-responsiveness that avoids these difficulties.

1 . T H E A P PE A L O F A R E A S O N S - R E S P O N S I V E T H E O R Y

Reasons-responsive theories have an ancient lineage that arguably can be

found in Aristotle.8 They are appealing irrespective of the compatibilism-
incompatibilism debate for at least two reasons. First, they target a feature
of agency distinctive of persons, a feature that marks persons as rational
animals. Second, they offer elegant explanations of the conditions in which
various excuses and exemptions apply—the persons in question, it is
contended, are not suitably responsive to a proper range of reasons. For
instance, a plausible way to distinguish between the person who, weak
willed, freely drinks the beer from the person who is compelled by her
addiction to do so is that the former is responsive to a wider range of
reasons for not drinking than is the latter.
Compatibilists have found powerful reasons to develop a reasons-
responsive theory quite apart from the unique dialectical burdens of source
compatibilism. Reasons-responsiveness seems especially well suited for a
compatibilist analysis since in explaining sensitivity to reasons, we must
consider how an agent would respond to a range of reasons and not just
those present to an agent when she performs an action in the actual
circumstances that she is in. This fits naturally with attending to a range
of dispositional or modal properties of agency, and there is little reason to

7 See Appendix I for my source compatibilist objection to these theories.

8 See Terrence Irwin’s (1980).
154 Michael McKenna

think that determinism is incompatible with the possession of these sorts of

properties generally.
To appreciate how a reasons-responsive theory might be developed,
begin with the following two preliminary points:
First, when an agent acts who is suitably reasons-responsive, the most
important factor for the source compatibilist in accounting for her free-
dom is that the etiology of the act which she actually performed involved
springs that were sensitive to reasons. For now, let us think of those
springs in terms of the agent herself as the cause of her acts, where
this is not meant to commit to thinking of agents as distinct, irreducible
substance causes. Different reasons, understood as different inputs, would
have yielded different outputs, understood as alterations in modes of
conduct. And what this shows is that the agent’s response to the actual
“inputs” played a role that was itself sensitive to, or responsive to, the
actual conditions in which the agent acted. To illustrate, consider a simple
example of the sensitivity of a primitive gizmo, a thermostat. Suppose a
thermostat is set at 76 degrees (Fahrenheit) and the room the thermostat
is in turns out to be 76 degrees. One might wonder if the thermostat’s
setting accounts for the temperature of the room. After all, it might be
disconnected and so merely a fluke that its setting and the room’s
temperature are 76 degrees. When we learn that the room would come
up to 78 were the thermostat set to 78, or would come down to 74 were
the thermostat set to 74, and so on for numerous other values high and
low of 76, we do not just learn something about the way the thermostat
would behave; we also learn about how, in the actual scenario when it is
set to 76, it does behave; it plays a certain causal role from reliable and
suitably sensitive resources.
The same can be claimed about a person who is responsive to reasons.
Consider by contrast a clear case of reasons-responsive failure, a case in
which a person fails to take reasons as inputs in a manner that yields proper
results (like a defunct thermostat). Imagine a compulsive hand-washer
washing her hands after handling some trash. Not knowing her condition,
we might initially think that her freedom consists in part in her responding
rationally to conditions that would warrant hand washing. But now sup-
pose that we learn that she’ll wash her hands when they brush up against a
bottle of bleach, or when she touches her own shirt, or if the sun shines on
them, or if the wind blows, or if Don Knots is on TV, or there is a
Christmas carol playing on the radio. This casts doubt on whether the
reason she is washing her hands in the actual world is because she was
handling some trash. Or, even if her washing her hands is to be accounted
for in terms of her handling some trash, her insensitivity to other reasons
suggests that the manner in which she caused her action was not suitably
 Reasons-Responsiveness, Agents, and Mechanisms 155

sensitive to the significance of trash-handling, as in contrast with, say, the

wind blowing.
Second, responsiveness to reasons comes in degrees. This raises ques-
tions about what amounts to an adequate degree of responsiveness and
what sorts of failures are tolerable. Consider again the thermostat. Sup-
pose the thermostat functioned as described for uses in a normal home
environment. It would not impugn the adequacy of the thermostat to
learn that it is not super-responsive, and so not responsive to temperatures
of 2 degrees at one end and 150 at the other. Alternatively, the thermostat
might be only very minimally responsive, functioning properly between
settings of 75 through 77, but yielding whacky results at any other
settings. Furthermore, failures of responsiveness can be localized. Imagine
that the thermostat that is set at 76 would not have responded differently
to any settings of between 74 and 78, but would have responded properly
to settings of lower than 74 or higher than 78. Here, we could still not
determine what role setting the thermostat at 76 played in the room’s
being 76 degrees, since the room would have remained at 76 degrees with
settings of as low as 74 or as high as 78. Even if we were to conclude that
setting the thermostat at 76 caused the room to become 76 degrees, we
would be right to think that the mode of its causing the change in
temperature was too fluky, so much so that all we could feel confident
in claiming as a credible explanation is that it reliably caused the room’s
temperature to be some temperature anywhere between 74 and 78, and as
chance would have it, it happened to be 76.
The same points apply to persons and their responsiveness to reasons.
A person who is suitably reasons-responsive in a way that is sufficient for
free action need not be able to respond to a range of reasons that meets
exceedingly high standards, otherwise almost no one would turn out to
act freely. And a person who is only responsive to a very slim range of
reasons will fall short of acting freely while nevertheless being responsive
to some reasons. Take our compulsive hand-washer. It seems that she will
wash her hands come what may. But it might be that a very limited range
of incentives would result in her not washing her hands: if her child were
on fire, or if her favorite Mozart concerto were playing and she instead
wanted to dance, and so on. While such cases might show her to be
responsive to reasons, she would only be minimally so, and this would not
be enough to regard her as acting freely in washing her hands as she did.
Furthermore, as with simple gizmos like thermostats, so too with persons,
localized failures are possible. Someone who is as competent as anyone
might fall to pieces in the presence of an abusive spouse or a domineering
partner. Localized foibles are part of the package deal for most of us
imperfect humans.
156 Michael McKenna

2. FISCHER AND RAVIZZA’S ACCOUNT OF

MODERATE REASONS-RESPONSIVENESS

Now, in light of the two preceding points, let us examine Fischer and
Ravizza’s carefully developed reasons-responsive theory.9 According to
them, at one end of the spectrum is strong reasons-responsiveness (SRR).
An agent who is SRR with respect to an act is such that, if there were
sufficient reason for her to do otherwise, she would do otherwise (Fischer
and Ravizza, 1998: 41). At the other end of the spectrum is weak reasons-
responsiveness (WRR). An agent who is weakly reasons-responsive with
respect to an act is such that there is at least one (nonactual) scenario in
which there is sufficient reason to do otherwise and the agent does
otherwise (44). As should be clear, and in keeping with the second point
developed above (in Section 1), SRR is too strong; if a condition of free
agency required SRR, it would turn out that almost no one acts freely and is
morally responsible for what she does. On the other hand, WRR is too
weak; if a condition of free agency required no more than WRR, it would
turn out that many severely impaired agents would fully satisfy the freedom
conditions for moral responsibility. Our compulsive hand-washer was
WRR in washing her hands since she’d have not washed them if her
children were on fire. So, what is needed, and what Fischer and Ravizza
develop in admirable detail, is a middle-of-the-spectrum degree of
responsiveness, moderate reasons-responsiveness (MRR), that is suited for
sane, competent, albeit imperfect, moral agency—the sort of agency of
which most psychologically healthy adults are capable while functioning in
normal practical contexts of deliberation and action (69–84).
So how do Fischer and Ravizza develop MRR? Understanding their
proposal requires careful attention to what is involved in being responsive
to reasons. Fischer and Ravizza distinguish between a receptivity component
and a reactivity component (1998: 41). Being reasons-receptive is a matter
of being able to recognize what reasons there are, and in particular, being
able to recognize what reasons count as sufficient for action. Being reasons-
reactive is a matter of being able to react to the reasons one recognizes as
sufficient by choosing and acting as needed. According to Fischer and
Ravizza, MRR requires moderate reasons-receptivity. For a person to be
morally responsible for what she does, she must be receptive to reasons in

9 In examining Fischer and Ravizza’s view in this section, I shall depart from their
doing so in terms of mechanisms that are reasons-responsive and that are owned by the
agent. I shall instead just write in terms of agents being reasons-responsive. In the next
section, I’ll recalibrate to fit their mechanism-based formulation.
 Reasons-Responsiveness, Agents, and Mechanisms 157

a manner that displays a rich pattern of recognition whereby reasons can be

placed on a scale with a continuum involving stronger and weaker incen-
tives. For instance, if an agent regarded an incentive of $100 as sufficient for
acting in a certain way, she would take higher values as also sufficient, while
being open to taking lower values as insufficient. Furthermore, the agent’s
pattern of reasons recognition must display a sane appreciation for reasons
that are grounded in reality, and the agent must also be able to recognize
moral reasons and be able to grasp that sometimes they are sufficient for
acting in ways that morality prescribes. All of this, of course, is consistent
with being an agent who is not receptive to the full spectrum of reasons for
how she ought to act, including the full spectrum of moral reasons.
As regards reasons-reactivity, Fischer and Ravizza allow for a striking asym-
metry. An agent who is MRR need only be weakly reasons-reactive. It is enough,
they contend, that among the worlds in which an MRR agent is receptive to
sufficient reasons to do otherwise, she react to only one of those reasons. Here,
one might wonder why weak reactivity would be enough. Wouldn’t this show
that in the wide spectrum of cases in which the agent recognizes sufficient
reasons but does not act upon them that the agent cannot act upon them and so
is impaired for morally responsible agency? Fischer and Ravizza do not
think so. They contend that “reactivity is all of a piece” (1998: 73). As they
see it, the fact that there exists just one world in which an agent reacts differently
to a sufficient reason to do otherwise is sufficient to establish that in each
world in which an agent recognizes sufficient reasons to do otherwise that she is
able to react to those reasons, even if at those worlds she does not.
Critics have pressed Fischer and Ravizza on various details of their
proposal regarding the spectrum of MRR. One has to do with their
requirement of weak reasons-reactivity. As I have argued elsewhere
(2005), weak reactivity looks to be too weak. The fact that an agent
would react to only one reason to do otherwise but not a constellation of
other similarly related reasons calls into question her agency. Hence, I have
proposed a variation on MRR in terms of weaker reactivity, not weak
reactivity.10 The spectrum of reasons to which an agent must be reactive
can be weaker than the spectrum of reasons to which she is receptive11, but

10 Fischer (2006: 328) has granted this point in response to Mele (2006b: 290), who
presses a similar worry.
11 This is needed in order to allow for cases in which an agent freely and knowingly
acts contrary to what she has sufficient reason to do. Note, however, that I have only
claimed here that the spectrum of reactivity can be weaker than that of receptivity.
Elsewhere (2005), I have made the stronger claim—that their view requires this asym-
metry. Dana Nelkin and David Brink have convinced me that this is a mistake. The
requirements of morally responsible agency do not, strictly speaking, require any asym-
metry. Nevertheless I suspect for nearly all actual persons who are morally responsible
158 Michael McKenna

the spectrum of reactivity still needs to display a sane, stable pattern along
the lines of moderate receptivity. There are further details that we might
add here to supplement Fischer and Ravizza’s proposal, or instead various
details of their thesis with which we might take issue, but for present
purposes the preceding discussion will do.
While it is important to make clear the delicate effort Fischer and
Ravizza make to get right the degree of responsiveness needed to account
for MRR, the most important factor in their account of freedom is that the
etiology of an agent’s act involved springs that were themselves sensitive to
reasons. This shows Fischer and Ravizza’s deference to the first of the two
preliminary points set out above (in Section 1). I pause here to emphasize
this since it is easy to lose sight of. The salient point to note is that by
demonstrating that when an agent acted she was MRR, one does not just
show how that agent would respond to other reasons if those reasons were
present to her. One also shows that in acting as she did, the agent was
responsive to the conditions she was actually in. Her actually-operative
springs of action were functioning in ways that themselves were responses to
the conditions in which she actually found herself.
To elaborate on the preceding point: It is easy when reflecting upon
Fischer and Ravizza’s proposal to misunderstand the theoretical purpose to
which certain counterfactuals are put. Consider a MRR agent who in
deciding to remain home and work on an article would be receptive and
reactive to the reason that the legendary blues guitarist Buddy Guy is
putting on a concert that night and she could easily go. Were she presented
with this reason, she would not work on her article. Instead, she’d go to the
show. Don’t be snookered here into thinking that what this counterfactual
is meant to establish is that, in acting as the agent did, she could have done
otherwise. That is, the point of this counterfactual, as well as other counter-
factuals involving other reasons to which this agent might be both receptive
and reactive, is not meant to establish that the agent had a freedom that
concerned alternatives to what she did do—leeway freedom. It is instead
meant to establish that in acting as she did, she was sensitive to reasons in
such a way that her actual bringing about of her act of remaining home
and working on her article was itself suitably sensitive to rational consider-
ations. Some compatibilists, such as those classical conditionalists like Ayer
(1954), Davidson (1973), Hobart (1934), and Moore (1912), attempted to
rely upon similar counterfactuals in order to account for a compatibilist
theory of the freedom to do otherwise. But Fischer and Ravizza do not

agents, there will be some asymmetry of this sort; most persons are not moved to act by
the full range of sufficient reasons for action to which they are receptive.
 Reasons-Responsiveness, Agents, and Mechanisms 159

understand the worlds in which an agent does otherwise in response to

differing reasons as worlds that, as they put it, are accessible to the agent in
the actual world in which she does act (1998: 53). So, as they see it, it
cannot be that by virtue of such worlds an agent in the actual world is able
to do otherwise. Hence, Fischer and Ravizza do not mean for these
counterfactuals to aid in underwriting an account of the freedom to do
otherwise.

3. SHIFTING TO A MECHANISM-BASED THEORY

Given the preceding treatment, can we fit a reasons-responsive theory into

the actual-sequence constraints of Frankfurt’s argument? Regrettably, on
first appearance, these two compatibilist-friendly theses do not appear to
make comfortable bedfellows. Recall Jones from the Frankfurt example
presented earlier. Jones, says the source compatibilist, acted freely. But due
to Black’s presence, whatever reasons might have been put to Jones, it is not
the case that he would have done other than as he did. This is because Black
was playing the role of a “counterfactual intervener” and stood prepared to
ensure that Jones would shoot Smith no matter what. Thus, if an agent
acts freely in a Frankfurt example, it seems that her freedom cannot be
explained in terms of her responsiveness to reasons. Due to the presence of a
counterfactual intervener, the agent would not act otherwise even if differ-
ent reasons were present (1998: 38). So is reasons-responsiveness ill-suited
for a source compatibilist theory?
According to Fischer and Ravizza, we can fit a reasons-responsive theory
to the kind of freedom present in a Frankfurt example by shifting from an
agent-based to a mechanism-based theory. The agent in a Frankfurt example,
they contend, is not responsive to reasons, but the mechanism from which
she acts, which is her own mechanism of action, is reasons-responsive. To
appreciate their intriguing proposal we need to answer two questions. First,
what is a mechanism of action as they understand it? Second, how is it that
focus upon an agent’s own mechanism of action ensures immunity to the
problem that seems to arise for an agent’s reasons-responsiveness in a
Frankfurt example? Why does the shift to mechanisms help?
As regards the first question, the mechanism of action, as Fischer and
Ravizza understand it, is meant to pick out “nothing over and above the
process that leads to the relevant upshot” (1998: 38). To make clear that
they do not mean to reify the notion of mechanism or to suggest that it is
something like a natural kind, they remark that rather than use the term, we
could instead just speak in terms of the “way the action came about” (38).
160 Michael McKenna

Clearly, in this sense, the mechanism of action at work in the Frankfurt

example involving Jones’s acting on his own is very different from the
mechanism that would have operated were Black to have intervened and
caused Jones to shoot Smith. That much, it can be safely granted, is clear
and uncontroversial. Furthermore, one can also infer that among the full
constellation of events, states and processes constituting Jones, or Jones’s
complete psychic condition at the time, not all of the elements of that
constellation were causally implicated in Jones’s shooting Smith. Jones, for
instance, might have a deep love of his mother, a fondness for kittens, and
be especially skilled at solving calculus problems. But these, we can safely
assume, can be “filtered out” of the complex of Jones’s psychic events,
states, and processes that were implicated in the process leading to has act of
shooting Smith. Thus, an agent’s mechanism of action, however it is to be
construed, carves out a subset of the full set of psychological elements
constituting the agent’s psychology.
As regards the second question, Fischer and Ravizza argue that to test
whether in a Frankfurt example an agent’s own actually-operative mechan-
ism of action was reasons-responsive, we need to test it for sensitivity to
various reasons in conditions in which it operates uninhibited, and to do
this we have a license to “go to worlds”, as the expression goes, in which
Black’s presence is subtracted. So, were Black not present, and were Jones
presented with pertinent reasons for not shooting Smith, Jones, by way of
his mechanism of action, would react otherwise and not shoot Smith.
According to Fischer and Ravizza, in order for it to be the case that Jones
exercises adequate source freedom in shooting Smith, his mechanism of
action must be such that the pattern of reasons to which it is sensitive would
be MRR.

4 . PR O B LE M S W I T H M E C H A N I S M S

Various critics have objected to Fischer and Ravizza’s mechanism-based

approach. R. Jay Wallace has remarked that by shifting from an agent-
based to a mechanism-based view, the “intuitive locus of responsibility, the
person, drops out of view” (1997: 159). He claims that instead we should
tend to the normative competence of persons. Gary Watson has expressed
his concern that characterizing a mechanism of action in terms that are no
more specific than “ ‘a process resulting in behavior’ is too amorphous to do
the refined work needed by the theory” (2001 as appearing in Watson,
2004: 299). And Carl Ginet has claimed that the postulation of
mechanisms is not doing any useful work in their theory (2006: 235–6).
 Reasons-Responsiveness, Agents, and Mechanisms 161

According to him, Fischer and Ravizza want it to be that an agent would

respond differently to certain reasons by virtue of a mechanism. Their
theory, Ginet contends, could be preserved by cutting out the middle
term and just theorizing in terms of an agent being reasons-responsive.12
I too have registered some concerns with their appeal to mechanisms.
However, my criticisms have been more limited. I have only argued that
Fischer and Ravizza need to provide further support for their account of
mechanisms in order to avoid various objections.
In my estimation (2001), Fischer and Ravizza face difficulty with indi-
viduating the mechanisms operative in exercises of free action. They resist
pressure to provide any principled basis for mechanism individuation and
contend that they can make do with an intuitive sense of what counts as
the same mechanism and when, in shifting between contexts, we move to
different mechanisms (1998: 40).13 Clearly, there is a different mechanism
of action at work when an intervener such as Black causes Jones to act as
in contrast to when Jones acts on his own. Thus, on their view, just as
we are able to reidentify the same face or smile, or the same house or car,
without a particular theory or general principle, so too we can recognize
same mechanism of action operative in different contexts. The problem,
however, is that there are some contexts crucial to the development of their
theory where, absent some principled bases for mechanism individuation
and differentiation, it is hard to assess their broader claims. Obviously they
are in no danger when distinguishing between actual and alternative
mechanisms in Frankfurt examples. But it is an indispensible part of their
theory that we are to test how the same mechanism behaves in response to
different reasons. How are we to settle whether, when different reasons are
put to an agent, the same mechanism is operative? We need some purchase
on what it is that we are holding fixed when we hold fixed “the same
mechanism” while testing its degree of responsiveness.
The problem stems from Fischer and Ravizza’s lean characterization of a
mechanism of action. “The process that leads to action” or “the way the
action came about” allows for further specification by way of extremely
narrow or instead extremely wide descriptions. Picking out the process that

12 For Fischer’s replies, see his (2004: 169–71) reply to Watson, and his (2006: 333–4)
reply to Ginet. He replies to Wallace in his (2011, ch. 9: 144–62).
13 They do offer one theoretical constraint. The descriptive content by which the
mechanism is identified must be “temporally intrinsic” (1998: 46–7). That is, it must
pick out the mechanism in such a way that it does not involve reference to later times.
Otherwise, the mechanism would require certain outcomes, and thus would not permit
variability in the face of different reasons. Hence, the temporally extrinsic mechanism
“deliberation prior to donating to a charity” is ruled out since it entails acting in a certain
way.
162 Michael McKenna

leads to action seems most naturally to involve the entire complex of

proximal antecedent states and events figuring into a correct explanation
of the pertinent action.14 Suppose that we limit the items in this complex
just to the agent-involving ones—an obviously permissible restriction.
Still, is every agent-involving feature of the complex to be included in
the description? What salient features might one chose to whittle the
description down? At one level of description, sameness of mechanism
right down to microdetails would be required. But, as I have argued
elsewhere (2001: 97), this would likely give rise to serious problems.
Suppose that the blameworthy reason upon which an agent acts in the
actual world—or rather the set of psychological ingredients involved in
recognition of this reason—is identical with or supervenes upon some
neurological state of the brain. If sameness of mechanism requires
sameness of microdetails, when that mechanism is tested to learn how it
would respond to different reasons, it would have to respond to different
reason by resources that shared the same microdetails with the brain states
involved in recognition of the blameworthy reason in the actual world. This
is nomically unlikely, albeit, admittedly not metaphysically impossible. It
thus exposes Fischer and Ravizza’s theory to the rather likely prospect of
empirical refutation. So it seems that a very narrow description needs to be
ruled out.
Of course, Fischer and Ravizza would be on solid ground in ruling out
such narrow descriptions.15 Typically, when identifying causal processes or
ways events are brought about, we do not mean to focus upon microdetails.
For instance, identifying the event of the flood as the cause of the erosion
needn’t require that we focus on the microdetails of just how it was that the
water molecules were arranged. In such contexts it would be highly
unreasonable to demand such a level of specificity. Fischer and Ravizza
are certainly entitled to a similar reply here as regards the sorts of processes
leading to (putatively) free acts.
But then, how wide, and so permissive, should the needed description
be? Because Fischer and Ravizza offer no principled basis for mechanism
individuation, we are left to settle the matter exclusively by appeal to our
intuitive reactions to varying cases. Unfortunately, intuitions in these
kinds of highly theoretically charged contexts can vary widely. The neuro-
physiologists’ modes of parsimony, the cognitive scientists’, or instead
the cognitive therapists’, are going to be very different from those
employed in normal folk-psychological discourse in which typical claims

14 I am indebted to Carl Ginet for this point. See also his discussion of it (2006: 233).
15 I am indebted to Derk Pereboom, who emphasized this point.
 Reasons-Responsiveness, Agents, and Mechanisms 163

of blameworthiness and praiseworthiness have their natural homes. And

among philosophers differentially committed as regards the free will
dispute, it is easy to see that some will be inclined to a hyperrestricted
description that will narrow in on requiring same past and laws. Whereas
others will be open to descriptions that are far more permissive. Of
course, this would just be to redraw old lines in new sand. So that
won’t help. What is needed, it seems, is some rationale with theoretically
independent appeal that will allow us to assess claims of same or different
mechanism.16

5. A DEEP PROBLEM FOR ANY MECHANISM-BASED

APPROACH

The preceding considerations strongly suggest that if a mechanism-based

version of a reasons-responsive theory is to survive, there must be some
principled way to get an independent purchase on what counts as the same
mechanism across relatively diverse contexts. In my earlier work (e.g. 2001,
2011), I have often expressed confidence that there is a way to supplement
Fischer and Ravizza’s theory so as to offer the right kind of rationale for
mechanism individuation. Unfortunately, my considered opinion now is
that there is a structural problem which stands in the way of offering any
more specific content to the individuation conditions for mechanisms of
action. To explain, it will be useful to begin by considering an objection
developed by Ginet (2006: 234–5).17 Fischer and Ravizza take up the
example of a person driving a car from a mechanism of unreflective habit
(1998: 86). This person takes an exit she often takes, thinking nothing of it.
Were it blocked, she’d respond properly to the pertinent reason without
any deliberation at all. She’d just head to the next exit. But as Ginet plainly
points out, while the case Fischer and Ravizza consider is readily handled by
their treatment, there is a slight variation on it which is problematic.
Imagine instead a person who, upon seeing the exit blocked, is unsure of
what other exit she ought to use. Here, what would be most natural would
be for her to deliberate, and it would seem that a reasons-responsive theory
should seek to accommodate such a case. But Fischer and Ravizza hold
fixed the mechanism of unreflective habit. Hence, they cannot permit that

16 For an illustration of one particular point in which, in responding to an incompatibi-

list concern, Fischer and Ravizza cannot make do without a principled basis for mechanism
individuation, see Appendix II. There, I also discuss Fischer’s (2004: 166–71) thoughtful
reply to my criticism.
17 Watson has raised a similar worry (2001 as appearing in 2004: 298).
164 Michael McKenna

the mechanism of unreflective habit from which this agent acts is plastic
enough to allow reflection when conditions call for aborting reliance on
habit alone.
Ginet’s insightful criticism is a specific instance of a more general problem
that is bound to plague any attempt to give more content to what might count
as the conditions for individuating mechanisms. The general problem, as I see
it, is that any complex system will have “subsystems” that are designed to
function precisely by shutting down or by permitting other systems to
override in some contexts but not others. Assuming a person functioning as
a practical agent can be understood as a complex system, or at least as
relevantly analogous to one, any attempt to whittle down which agential
elements are the ones implicated in “the” mechanism of her action is bound to
restrict the agent’s flexibility as regards reasons-responsiveness.
The postulation of an agent’s mechanism of action as narrower than the
full person herself looks innocent enough when one is invited to consider a
case like Jones and note that Jones’s ability to do calculus is irrelevant to the
agent-involving factors causally contributing to the shooting of Smith.
There are likely lots of these ingredients in Jones’s complete psychic consti-
tution that could be whittled away: his love of his mother, his fondness for
kittens, and so on. But once we seek to put more meat on a restriction of a
pertinent mechanism of action, we will face problems like the one Ginet
has identified. If we back off and allow for the description of the mechan-
ism to be “fatter” so as to allow for shifts between processes like unreflective
and then reflective habit, we pretty much have what is the functional
equivalent of the agent herself as the mechanism, minus various outlier
traits like the ability to do calculus.
Reflection on complex mechanisms like automobiles or computers, things
that are built up out of smaller mechanisms, helps to illustrate the point.
Suppose that one of a computer’s programs runs unimpeded, but the
computer is designed to divert that program and prioritize other operations
if the system is under stress, or uploading new software, or about to lose
power, or what have you. The same applies to an automobile. A car will allow
things like unimpeded gas flow through the fuel injection system unless
another part of the system recognizes problems with the fuel mix or some-
thing of the sort. If we were to evaluate the degree of sensitivity of such a
system by “holding fixed the actually operative sub-mechanism” we’d ham-
string the system for a variety of conditions to which, without holding these
fixed, the larger system as a whole would be able to respond quite easily.18

18 An anonymous referee for OUP has astutely asked about the following way of
defending Fischer and Ravizza’s appeal to mechanisms. To illustrate, consider Ginet’s
example of the driver acting from the mechanism of unreflective habit. Why is it a
 Reasons-Responsiveness, Agents, and Mechanisms 165

To give just one action-theoretic example of this sort of thing, consider

Alfred Mele’s (1995, 2006) work on free agency (sometimes he writes in
terms of autonomy). One element in Mele’s approach is his account of
both weakness and strength of will. To account for these phenomena, Mele
distinguishes between the processes by which an agent judges what it is best
to do from the motivational ingredients involved in directing her actions.
Often, when a person acts, her evaluations of the objects of her desires are
aligned with the strength of the desires themselves. In these cases, when she
acts, the resources by which she might be able to exercise strength of will
and avoid temptation are, so to speak, off line. They’re causally inert, and
so could not be counted as anything like part of the actual causal process
issuing in action. But the agent might be such that she has a kind of
monitoring system, and were it that conditions would present themselves
(were certain reasons to arise) that would involve the agent desiring more
something incompatible with what she judges it best to do, the agent would
be able to rely upon these agential resources in such a way as to act with
strength of will and thereby act contrary to what she most desires (Mele,
1995: 27). Mele’s model here is a perfect example of the kind of dynamic
relation between “subsystems” within the overall complex of an agent as a
whole. If we want to include in “the mechanism of action” this entire
complex, we will have to open the descriptions for “same mechanism” to
include more than just what is, narrowly construed, causally implicated in
the process leading to action. Here, it seems, “the mechanism” will become
so inclusive that we might as well simply identify the entire person with the
mechanism.19

problem for Fischer and Ravizza that in some scenarios the agent would have to “shift” to
a different mechanism by beginning to deliberate? Isn’t it enough that the mechanism of
unreflective habit is responsive enough to certain reasons to allow for it to, so to speak,
shut down and allow another to come on line? Why isn’t this enough to say that the
mechanism is suitably reasons-responsive? In response, I acknowledge that this is after all
a thoughtful way to attempt to defend Fischer and Ravizza. But the upshot is to do so by
claiming that the full spectrum of reasons to which an agent (by way of mechanisms or
otherwise) ought to be responsive for (something like MRR) requires the postulation of a
plurality of mechanisms. At this point, it appears that it is not a matter of holding fixed a
mechanism, and now one wonders how close we are getting to just straightforwardly
talking of the responsiveness of persons.
19 In correspondence, David Shoemaker has thoughtfully expressed some skepticism
about this last point—that a more inclusive notion of a mechanism of action will be so
inclusive that we might as well identify it with the person, or with the functional
equivalent of the person. In conversation, Shaun Nichols has as well, citing the work
by Fodor on the modularity of the mind. Here, I have no problem leaving it as an open
question whether there is some richer notion of mechanisms that would do the work that
needs to be done for a mechanism-based view like Fischer and Ravizza’s. But I strongly
suspect that even if such a kind of mechanism were identified, and even if it turned out to
166 Michael McKenna

6. RETURNING TO AN AGENT-BASED THEORY AND

THE REJECTION OF SOURCE COMPATIBILISM?

So, are we back to the incompatibility of a reasons-responsive theory and

source compatibilism? According to Fischer and Ravizza, in order for an
agent to be reasons-responsive, it must be that if different reasons were
presented to the agent, then for at least some range of reasons, the agent
would react otherwise. In a Frankfurt example, they contend, due to Black’s
presence, the agent is not able to react otherwise.
This result—the incompatibility of reasons-responsiveness and source
compatibilism—would come as welcome news to those contemporary
compatibilists who reject Frankfurt’s argument and defend the traditional
association between free action, moral responsibility, and the ability to do
otherwise. Call these traditionalists leeway compatibilists. Amongst these
leeway compatibilists Michael Fara (2008), Michael Smith (2003), and
Kadri Vihvelin (2004) have each independently built upon recent
developments regarding the nature of dispositions to account for the
ability to do otherwise. In a Frankfurt example, each argues, an agent
retains the ability to do otherwise. Randolph Clarke (2009) has labeled
the thesis these philosophers advance the new dispositionalism. It is open to
the new dispositionalists to argue that in a Frankfurt example, an agent is
reasons-responsive, because, in being able to do otherwise, she is able to
react to different reasons.
While there are differences in how they execute their arguments, the
new dispositionalists are united in proposing that the free-will ability
is somehow to be accounted for in terms of dispositions. Following the
efforts of philosophers like David Lewis (1997) and C. B. Martin (1994),
these leeway compatibilists all pay heed to accounts of dispositions
that incorporate lessons learned from the way dispositional properties can
be masked or finked. A disposition is masked when its manifestation is
concealed in some way. A piece of salt placed in water remains soluble even
if, when placed in water, it does not dissolve because it is encased in wax.
A disposition is finked when, in just those conditions that would otherwise
trigger manifestation, it is altered so as not to have the disposition. A glass
vase sitting on a shelf undisturbed possesses the disposition of fragility even
if, were it knocked over, a wizard would turn it to stone before striking the
ground, rendering it not fragile.

be narrower than the full person qua agent, it would be much richer in content than
anything like something restricted just to what is actually involved in the causal
generation of action.
 Reasons-Responsiveness, Agents, and Mechanisms 167

Cases of masking and finking show that dispositions cannot be analyzed

in terms of simple conditionals such as, “if the vase were toppled, it would
break” or “if the salt were placed in water, it would dissolve.” The vase
mentioned above was fragile, but because it was finked, it would not break
when toppled. The salt was soluble, but because it was masked, it would not
dissolve if placed in water. In slightly different ways, each of the new
dispositionalists propose more complex counterfactuals the truth of
which confirms the possession of pertinent dispositions. Roughly, their
strategy is of the following variety:
SC: If this vase were toppled, and if it retained during the relevant
duration of time its intrinsic properties P1-Pn, and if it were not inter-
fered with in a way that would impede the causal efficacy of those
intrinsic properties, then it would break.
Now, SC is no more than a crude illustration of what a carefully worked out
formulation should look like, the sort that, for instance, both Fara and
Vihvelin consider.20 But for present purposes it will suffice. It is by virtue
of counterfactuals like SC that the new dispositionalists propose a more
sophisticated treatment of dispositions. If abilities are then to be explained
(Fara, 2008), or, more vigorously, fully analyzed (Vihvelin, 2004), in terms of
dispositions, then counterfactuals like SC provide a template for an
explication or analysis of the free-will ability. An especially important factor
in any such proposal is that the deeper explanation of the possession of the
dispositional properties by an object concerns the relevant intrinsic properties
possessed by the object (e.g. P1-Pn as mentioned in SC). It is these that give
the object the kinds of causal resources to have the effects characteristic of a
disposition’s typical manifestation(s). The same, then, can be said on such an
account for a person, qua agent, and her abilities; the deeper explanation of
her abilities, and most notably the ones constituting her free-will ability, will
be by virtue of certain intrinsic properties possessed by the agent.
It is open to the new dispositionalists to develop their respective accounts
in terms of responsiveness to reason, as both Vihvelin and Smith appear to
do (though neither explicitly use the label “reasons-responsive theory”). For
a relevant range of reasons, counterfactuals like SC could be constructed to
help underwrite claims about the spectrum of responsiveness that Fischer
and Ravizza attempted with their MRR—in particular, the spectrum of
reasons to which an agent is both receptive and reactive. So, take Jones, and

20 Smith’s (2003) essay is written in a more general, programmatic style. He does not
develop the kind of detailed proposal that either Fara or Vihvelin does, but it is clear
from his exposition that he is open to the development of his view in ways are amenable
to Fara’s or Vihvelin’s efforts.
168 Michael McKenna

for the moment set aside his acting in a Frankfurt example. Suppose for now
he is not in a Frankfurt example and that he shot Smith on his own and for
his own reasons. Suppose also that one of the reasons to which Jones would
have been both receptive and reactive in a way that would have resulted in his
not shooting Smith is if he were to have learned that Smith’s child was with
him at the time. Were Jones to have learned of this, he would not have shot
Smith. But as it happens in the actual world, he learns of no such reason and
proceeds to shoot Smith. Here is a counterfactual, built by modeling it on
SC, that shows how it is that Jones is responsive to this reason, call it R1:
RR1: If Jones were to become aware of R1, and if Jones retained during
the relevant duration of time intrinsic agential properties P1-Pn, and if
Jones were not interfered with in a way that would impede the casual
efficacy of those properties, then Jones would not shoot Smith.
By compiling a collection of counterfactuals like RR1, say RR1 through
RRn, for a spectrum of reasons R1 through Rn, the new dispositionalists
could establish that an agent like Jones is MRR in just the fashion that
Fischer and Ravizza have been careful to work out.
Thus far, I have tried to show, in admittedly only broad brushstrokes,
how it is that by relying upon counterfactuals like RR1 through RRn, the
new dispositionalists could make use of their account of abilities to develop
a theory of reasons-responsiveness. As I have explained, it is open to them
to build the theory in such a way that it is very much like Fischer and
Ravizza’s proposal for MRR but for the fact that their account would be
agent-based, not mechanism-based. But how is it that on their view the
truth of counterfactuals RR1 through RRn also underwrites the ability to do
otherwise in a way that shows how an agent retains this ability even in a
Frankfurt example? This is what is needed for them to put the nail in the
coffin of source compatibilists. Here their rationale is simple. Take Jones
whom we pulled from the context of a Frankfurt example, and take RR1
through RRn. Now place Jones back in a Frankfurt example. All of RR1
through RRn, the new dispositionalists will tell us, remain true. Like Fischer
and Ravizza, to assess the truth of RR1 through RRn they too must go to
worlds in which P1-Pn are allowed to operate unfettered, and this will also
require factoring out the presence of a counterfactual intervener like Black.
But according to them, this is perfectly consistent with Jones’s possession of
the ability to do otherwise when he acts on his own. How so? When Jones
acts in a Frankfurt example, according to the new dispositionalists, Black
plays the role of a fink. When Black remains inactive, because he in no way
alters or impedes the efficacy of the pertinent agential properties such as
P1-Pn, Jones retains the ability to do otherwise—just like the vase sitting on
the shelf remains fragile despite that fact that, if toppled, a sorcerer would
 Reasons-Responsiveness, Agents, and Mechanisms 169

turn it to stone. Of course, were Black to intervene, Jones would lose that
ability. But Black remains passive (like the sorcerer). So, when Jones acts
freely, according to the new dispositionalists, and contra Fischer and
Ravizza, he retains the ability to do otherwise.
Is the new dispositionalists’ effort to undermine Frankfurt’s argument
successful? No. As Clarke (2009: 339–42) is careful to point out, the
problem with the new dispositionalists’ reply to source theorists is not
that they fail to identify an ability that an agent like Jones retains in a
Frankfurt example. Indeed, because when Jones acts on his own, none of
his relevant dispositions are disturbed, there clearly is a kind of ability that
he retains. But this ability can be described more carefully as a general
capacity to do otherwise, or instead as a general ability to do otherwise that
can, for instance, be exercised outside the context of Frankfurt examples, or
when the agent is not asleep, or when she is not tied up, and so on.21
Nevertheless, as Clarke remarks:
[T]here apparently are abilities that Jones lacks, because of Black’s readiness to
intervene. Though Jones might have the capacity to act otherwise, the circumstances
are not friendly to his exercising that capacity, and it may fairly be said that it is not
up to him whether he exercises it, or that he does not have a choice about whether he
does so. (2009: 340)
Furthermore, Clarke notes (341), Fischer, in articulating Frankfurt’s argu-
ment, is careful to point out that what is in dispute is whether under the
circumstances an agent can chose and do otherwise (Fischer, 2002: 304).
The salient point is that the new dispositionalists are only able to claim
victory in refuting Frankfurt’s argument if the ability that they contend
remains for an agent in a Frankfurt example is the same ability that is in
dispute in debates about free will and moral responsibility—and it is this
ability at which Frankfurt’s argument is aimed.
Precisely the same point applies regarding the question of whether
an agent can be reasons-responsive while acting within the context of a
Frankfurt example. If, as Fischer and Ravizza suppose, reasons-responsiveness
requires that an agent be able to react otherwise in response to sufficient
reasons to do otherwise, then what is in question is whether under the
circumstances of a Frankfurt example, an agent is able to react as required.
Recall (from Section 3) Fischer and Ravizza’s qualification regarding the
counterfactuals in virtue of which MRR is confirmed; they do not involve
worlds that are accessible to the agent (1998: 53). In the context of a Frankfurt
example, it is not accessible to an agent like Jones to react otherwise in
response to different reasons and, while in the presence of Black, not shoot
Smith.

21 Ann Whittle makes a similar point in her insightful (2010).

170 Michael McKenna

7 . A N E W P R O P O S A L : A N A G E N T - B A SE D ,
R E A S O N S - R E S PO N S I V E , S O U R C E
COMPATIBILIST THEORY

I have argued that a mechanism-based account of reasons-responsiveness is

not a viable option because it requires a principled basis for mechanism
individuation. Regrettably, this leads to insuperable problems fitting mech-
anisms for the degree of plasticity called for by proposals like MRR. But if
the reasons-responsive theorist is forced to return to an agent-based theory,
is it possible for her to fit such a thesis into the constraints of Frankfurt
examples, and thereby account for source freedom? According to Fischer
and Ravizza it is not; Frankfurt examples rule out the ability to do
otherwise, and agent-based reasons-responsive theories require that agents
who are suitably reasons-responsive be able to react, and so do, otherwise.
As I have explained, in light of this assumed conflict, the new disposition-
alists are positioned to claim that the winner is reasons-responsiveness.
According to them, what must go is a commitment to Frankfurt’s argu-
ment and the presumption that source freedom is freedom enough. But,
the new dispositionalists can reason, this is all to the good, since their
analysis of the free-will ability in terms of dispositions is, quite independ-
ently of any commitments to a reasons-responsive theory, sufficient to show
that in a Frankfurt example an agent is able to do otherwise. The problem
with this rejoinder is that the new dispositionalists have it wrong; they have
not refuted Frankfurt’s argument. What is at stake in a Frankfurt example is
whether—while in the context of a Frankfurt example—an agent is able to
exercise an ability to do other than as she does. And nothing in the new
dispositionalists’ playbook speaks to that issue.
So, assuming a commitment to Frankfurt’s argument and a focus on
source freedom, is it possible to develop a reasons-responsive theory with-
out relying upon the problematic notion of mechanisms? I think it is. The
single proposition standing in the way of fitting an agent-based reasons-
responsive theory to the contours of source freedom is that an agent in a
Frankfurt example is not reasons-responsive because, being unable to do
otherwise, she is thereby unable to react otherwise when give sufficient
reason to do so. I think this proposition should be rejected. I now wish to
propose a simple solution to this puzzle. An agent who acts freely in a
Frankfurt example, I shall argue, is suitably reasons-responsive despite
being unable to react otherwise when given sufficient reason so to do.
I offer two points in support of my claim.
First, the fact that an agent in a Frankfurt example is not able to react
otherwise in response to sufficient reasons does not exhaust the resources
 Reasons-Responsiveness, Agents, and Mechanisms 171

available to show that, when she acts unimpeded, she, the agent, is reactive
to sufficient reasons to do otherwise. Being reactive to sufficient reasons to
do otherwise, I now propose, is not the same as being able to react otherwise
in response to sufficient reason to do otherwise. In a Frankfurt example, an
agent who is suitably reasons-responsive is reactive to sufficient reasons to
do otherwise even if, due to the presence of a character like Black, she is
unable to react otherwise.
To illustrate this point, consider again Jones outside the context of a
Frankfurt example. Again suppose he shoots Smith for his own reasons. In
this case, suppose that his actual reason for shooting Smith is revenge.
Smith harmed Jones’s family, and now, as Jones sees it, it’s payback time.
Let us also assign Jones a pattern of receptivity and reactivity that, we can
stipulate, satisfies the conditions of MRR. Jones is receptive to reasons R1
through Rz, and is reactive to reasons R1 through Rn. Assume that the latter
is a subset of the former in such a way that Jones’s degree of reactivity is
weaker than his degree of receptivity. Recall that reason R1 as mentioned
above is that Smith’s child is with him, and were Jones to learn of this, he’d
be receptive to this as a sufficient reason not to shoot Smith, despite his
actual reason of revenge, and he would react accordingly, not shooting
Smith. And so it would be for the entire spectrum of reasons R1 through
Rn, but not for the reasons ranging Rn + 1 through Rz. For this latter range,
Jones would be receptive to these reasons as sufficient for not shooting
Smith, but he’d shoot Smith regardless. Imagine, for instance, that reason
Rn + 1 is that shooting Smith would cause Jones’s mother to be disap-
pointed with her son. While Jones would recognize this as a sufficient
reason for not shooting Smith, sorry son that he is, he’d not be reactive to it;
he’d shoot Smith for his own reasons of revenge, despite the grief he’d
knowingly cause his dear mum. Jones, given this set up, for the range of
reasons R1 through Rn is both reactive to sufficient reasons to do otherwise,
and would react otherwise in light of such reasons.
Now reinsert Black. Note that Black need not interfere with Jones were
Jones to face up to any of the reasons within the spectrum of Rn + 1 through
Rz. Jones is not reactive to these reasons in the sense that they would not
deter him from acting on his own reasons for shooting Smith. So Black is
able to leave Jones to function on his own. Should Jones consider the reason
Rn + 1, that his mother would be disappointed with him if he shot Smith,
he’d shoot Smith on his own anyway. Hence, Black can leave well enough
alone. But as for the reasons R1 through Rn, were any of these reasons to
become relevant to Jones’s practical context, Black would intervene. In
being prepared to do so, clearly Black makes it the case that Jones is unable
to react otherwise in response to these reasons. But it is nevertheless true
that Jones is reactive to this spectrum of reasons R1 through Rn in just this
172 Michael McKenna

sense: for each reason, were it to become salient for Jones, it is not the case
that, given his own intrinsic agential condition, he would act on his own
reasons of revenge for shooting Jones. He is at least reactive to reasons in
this manner. In reacting to this spectrum of reasons (at relevant possible
worlds), he, by virtue of his own agency and his own reasons of revenge, is
not the cause of shooting Smith. This is so despite that fact that he is unable
to act otherwise and thereby avoid shooting Smith. While acting within
the context of a Frankfurt example, Jones is reactive to this spectrum of
sufficient reasons to do otherwise, R1 through Rn, despite the fact that he is
not able to react otherwise. His being reactive in this context consists in his
not persisting in acting on his own reasons of revenge to shoot Smith in
light of reasons such an R1 (Smith’s child is with him).
Drawing upon the metaphysics of causation, Carolina Sartorio has made
important contributions to our understanding of source compatibilist
accounts of freedom, one of which is relevant to my current proposal. In
threat-cancelation scenarios, Sartorio (forthcoming) points out, causation
is not transitive along the path of a single causal chain. And the contrary to
fact scenarios in Frankfurt examples—that is, the scenarios in which the
intervener becomes activated—are set up as threat-cancelation scenarios.
To explain: Jones, given his own reasons for shooting Smith, is such that, in
the presence of certain reasons, such as R1 (Smith’s child is with him), he
would not shoot Smith. Suppose that, indeed, Jones is in the presence of
this reason, R1. This creates a threat to the event of his shooting Smith; it’s
the sort of thing that’s liable to lead to Jones not shooting Smith. The
presence of this threat, as a link in a causal chain, then causes the intervener
to bring it about that Jones shoots Smith. Hence, there is a causal chain
from Jones and his relation to R1 through to his shooting Smith. But
because of the way this causal chain unfolds, Jones, given the way he is and
his own reasons prior to the intervention, does not cause his shooting of
Smith. Why? Because, the way he is disposed to respond to reasons like R1
creates a threat to his shooting Smith. So, despite the fact that there is a
causal chain from Jones to the event of his shooting Smith via Black’s
manipulation, Jones—just as he is as an agent, given his intrinsic proper-
ties—is not the cause of shooting Smith. Thus, as Sartorio makes clear, the
difference between actual and counterfactual cases in Frankfurt examples is
a difference in the causal role played by agents such as Jones. This,
I maintain, helps cast light on the point I am at pains to emphasize here.
The point has to do with the way Jones is reactive to the different kind
of reasons that might bear on exercises of his agency. Some reasons, such as
Rn + 1, are such that, in their presence, Jones persists in being the right kind
of agential cause of his action (of shooting Smith). Some, such as R1, are
 Reasons-Responsiveness, Agents, and Mechanisms 173

such that, in their presence, he does not persist in being the right kind of
agential cause of his action.
The first point I am offering here bears some similarity to earlier failed
efforts to defend PAP against Frankfurt’s argument by distinguishing
acting on one’s own and not acting on one’s own, which seems to be a
kind of alternative that cannot be expunged from Frankfurt examples (e.g.
see Naylor, 1984). This strategy for defending PAP was handily rejected
because the alternative of acting or not acting on one’s own in suitably
structured Frankfurt examples were not alternatives that were themselves
within the voluntary control of the agent (Fischer, 1994). Why doesn’t this
apply to my current proposal?22 Here’s why: My focus on the distinction
between an agent’s acting upon her own reasons in an actual scenario and
her not acting upon her own reasons in alternative scenarios is not being
used to underwrite any claim about an agent’s freedom to do otherwise. It is
only being used to call attention to a fact about the mode of an agent’s
acting in the actual-world scenario in which she acts on her own. The only
reason that in the alternative scenario the counterfactual intervener causes
Jones to act as he does is because, given Jones’s state, just as he is as an agent,
he would not act on his own reasons of revenge for shooting Smith were
these other reasons also salient. Granted, in the alternative scenario in
which the intervener forces him to act, Jones does not do anything
intentional to make it so that he does not act on his own reasons to shoot
Smith; this scenario is not within the scope of his voluntary control. But
there is no claim here that the degree of reactivity displayed by Jones in this
range of counterfactuals affords Jones anything like a freedom to do
otherwise. These counterfactual scenarios are not alleged to be accessible
to Jones from the actual world. But, I say, they still display that, given the
way he is in the actual world, he is in some manner reactive to pertinent
reasons to do otherwise.
Perhaps some will find my argument here too thin. They might demand
that being reactive to sufficient reasons to do otherwise requires that, in the
presence of the relevant range of reasons, the agent’s own responsiveness-
grounding resources have to be directly causally involved in the mode of
differential reactivity. And when Black the intervener is in the driver’s
seat—that is, when Black has taken over and is generating the action—
that’s just not what is going on. In fairness, I grant that this is a reasonable
source of resistance. However, in my own estimation, those inclined to
press it in this context are fixed upon an alternative possibilities model of

22 Thanks to both Derk Pereboom and Brandon Warmke for raising this worry.
174 Michael McKenna

freedom, which the current proposal is meant to reject. Nevertheless, there

is a distinct way of arguing that an agent in a Frankfurt example is suitably
reactive to reasons, which I shall now explore.
Second, return to the complex counterfactuals which the new disposi-
tionalists exploit for the purpose of advancing a compatibilist theory of the
ability to do otherwise. Recall this one, where “R1” names the reason that
Smith’s child is with him:
RR1: If Jones were to become aware of R1, and if Jones retained during
the relevant duration of time intrinsic agential properties P1-Pn, and if
Jones were not interfered with in a way that would impede the casual
efficacy of those properties, then Jones would not shoot Smith.
As noted, to test for the truth of RR1, we have license to go to worlds in
which a counterfactual intervener like Black is missing. This is similar to
the way that Fischer and Ravizza test counterfactuals regarding the disposi-
tional properties of the mechanisms of action. The reason that factoring out
Black in these contexts is felicitous is because it is what is needed to test the
dispositional properties up for consideration. Now, the new dispositional-
ists might be wrong to think that counterfactuals such as RR1 can be used
in the service of giving an adequate account of the dialectically relevant
ability to do otherwise. And Fischer and Ravizza might be wrong to think
that we should draw on similar counterfactuals to account for the disposi-
tional properties of mechanism rather than agents. But the counterfactuals
themselves, such as RR1, still tell us something important about the agents
in question. They call to our attention dispositional or modal properties
that these agents really do have—even in the context of Frankfurt examples.
Jones really is disposed to respond to a reason such as R1 in certain ways.
Were he left unfettered to act as he wished, in relevant contexts, he would
act accordingly.23
Why is the preceding point so important? The challenge presently before
us is to account for an agent’s reactivity to reasons in contexts in which she
is not able to react otherwise, due to the presence of a counterfactual
intervener like Black. The point currently on offer is that counterfactuals
such as RR1, and in particular a sequence of them, say in a patterned form
of RR1 through RRn help to make what is in essence the same basic point
I was at pains to emphasize early on (Section 2). These counterfactuals aid
in demonstrating that when an agent like Jones acts as he does on his own,

23 Ginet (2006: 235–6) makes a similar point when advising Fischer and Ravizza to
give up the mechanism component of their theory and instead just consider the reasons-
responsiveness of agents.
 Reasons-Responsiveness, Agents, and Mechanisms 175

the act he actually does perform is itself a reaction or a response to the

actual conditions in which he acts. Furthermore, what these counterfactuals
collectively help to underscore is not merely that Jones was indeed the cause
of his so acting, but that the manner of his causing his act was sufficiently
sensitive to reasons.24 In this way, when an agent in a Frankfurt example
acts, she is, so to speak, being suitably reactive in acting as she does, and so is
exercising a kind of source freedom, a kind that can be characterized in
terms of responsiveness to reasons.
One residual worry about the second point I offer here concerns cases in
which, intuitively, an agent acts freely, but in which there are conditions
intrinsic to her own agency such that these conditions play the role of a fink
or a mask analogous to the role played by an extrinsic counterfactual
intervener like Black. For instance, suppose that if Jones were not about
to shoot Smith for the very particular sane reasons he had, then rather than
Black intervening, some psychotic episode would unfold, leading to his
shooting Smith anyway. Or we could instead run the case in terms of some
latent phobia that would arise in the case of certain reasons but not others.
If we hold fixed all of the intrinsic properties constituting Jones, we’d have
to include these psychotic-constituting or phobia-constituting properties as
well. In such cases, the relevant counterfactuals would come out false as
applied to the agents on the proposal currently on offer. But if one were
able to identify just certain features of the agent, as mechanisms, and hold
these fixed, one might still be able to account for responsiveness. Clearly,
these kinds of cases push back in the direction of a mechanism-based
view.25

24 An anonymous referee for OUP has raised the following insightful objection: Here
I advocate an account of an agent’s reactivity by claiming that the pertinent counter-
factuals help to reveal the manner of the agent’s causing her action. But, as I myself have
noted above, Fischer and Ravizza themselves contend that they do not mean to reify
the notion of a mechanism, and all they mean by appeal to an agent’s mechanism is the
“manner that an action was caused.” Why, the referee asks, is my proposal any different
from Fischer and Ravizza’s? This is an especially keen question, but the answer, to my
mind, is telling in a way that speaks against Fischer and Ravizza and on behalf of my
agent-based proposal. Even if Fischer and Ravizza mean to pick out no more than “the
way an action is brought about,” they are at least committed to the existential claim of
there being such a way, and to holding that fixed when testing it for responsiveness to
different reasons. My proposal is that what gets held fixed for the pertinent testing is the
agent. Now, either Fischer and Ravizza will grant that they mean the same thing, in which
case their appeal to mechanisms really is the functional equivalent of the notion of agency
itself and they might as well drop talk of mechanisms altogether, or they will insist that
they mean to fix on something narrower than the agent. In that case we have just the very
difference I have been at pains to bring out in this essay.
25 Stephen Kearns, Nicole Smith, David Sosa, and Jada Strabbing each pressed this
challenging concern in slightly different ways.
176 Michael McKenna

Previously, it was precisely because of cases like these that I had been
convinced that reasons-responsive theorists could not recover an agent-
based theory while remaining committed to source compatibilism. But
upon reflection, I do not think that these sorts of cases should carry the
day. Admittedly, they are problematic. However, there are two quite
different strategies for resisting their threat to the current agent-based
proposal for explaining reasons-reactivity. One is to resist what seems
intuitive in such cases. It’s not entirely clear that in such cases the agents
really do act freely even when the pertinent psychosis or the phobia is inert.
The agent’s “success” in acting uninfluenced by the troubling latent condi-
tion seems too fluky. The agent, after all, seems highly constrained in
responding to just the reasons she does, so much so that her own agential
resources impede her from responding to relevant patterns of reasons.
A very different strategy would be to consider whether, when specifying
the agent-constituting intrinsic properties that are to be held fixed in
counterfactuals like R1, there is a principled way to rule out those consti-
tuting the psychosis or the phobia. The rough idea would be to treat these
conditions as in some way alien or distant from those ingredients consti-
tuting the agent’s identity, or her real self.26 This is a promising avenue
worth pursuing, though I’ll not do so here. It is enough, it seems to me, just
to make clear that cases such as these pose a threat to the current proposal,
but that there are avenues for addressing them which allow preservation of
this basic claim. When a range of counterfactuals like RR1 through RRn are
true of an agent, these do help to show how it is that when she acts as she
does, she is reacting to reasons, even if she cannot react otherwise.
For a long while now, at least since the appearance of Fischer and Ravizza’s
excellent and highly influential Responsibility and Control (1998), it has been
assumed that the only way to marry a reasons-responsive theory of freedom
to the lesson learned from Frankfurt’s argument is to forgo an agent-based
theory. The way to account for source freedom in terms of reasons-
responsiveness, I had previously thought, is by shifting to a mechanism-
based theory. In light of the two preceding points, combined with the
difficulties involved in accounting for mechanism individuation, I believe
that this is mistaken. I propose an agent-based, reasons-responsive compati-
bilist theory of source freedom.

26 I am indebted to Michael Bratman for this point. He drew upon it to point out that
this gives us reason to consider joining the resources of a reasons-responsive strategy with
those of mesh theories like Frankfurt’s. The latter seek to make sense of something like
the boundaries of the “real self ” as narrower than the full spectrum on an agent’s
psychological profile. (Though Bratman then raised the question of whether such a
joint venture should be regarded as a friendly or instead a hostile takeover.)
 Reasons-Responsiveness, Agents, and Mechanisms 177

APPENDIX I: WHY NOT A MESH THEORY RATHER

T H A N A R E A S O N S - R E S P O N S I V E T HE O R Y ?

In the wider philosophical community, perhaps the best known source compatibi-
list account of freedom is not a reasons-responsive theory. It is Frankfurt’s own,
which he expresses in terms of acting freely and of one’s own free will (1971).
Frankfurt developed his position by attending to a fitting relation between higher-
order and lower-order desires. An agent acts of her own free will when the first-order
desire moving her all the way to action is the one that, at a higher order, she
identifies with and so most wants to act upon. Here, the further details do not
interest me. What does is the general strategy. Frankfurt’s is one version of a mesh
theory because it accounts for freedom in terms of a harmonious mesh between
different sub-systems within an agent’s overall mental economy.27 On Frankfurt’s
approach, an agent’s freedom consists in the well-functioning relation between
different orders of desire. When these different elements “line up,” the agent acts
unencumbered, and so she acts freely. But when, for instance, what Sally most wants
is for her desire not to drink the beer to win out, and when, despite that, her first-
order desire to drink the beer leads her all the way to action, Sally acts from an
unharmonious mesh. She is, in a sense, impeded or encumbered by her own
psychological constitution and so, according to Frankfurt, is not free. The system
issuing in her actions is out of alignment with the system that at a higher order
constitutes what she most wants.
When reflecting upon cases in which an agent acts from a harmonious mesh,
it is easy to see how that agent’s freedom, so construed, could easily be slipped
into a Frankfurt example. So long as her action does actually issue from the
mesh, there is no reason to be concerned about alternatives to what an agent
does do. All that matters in such a case is that her well-functioning mesh
operated unfettered. Thus, mesh theories such as Frankfurt’s fit seamlessly into
the actual-sequence demands of Frankfurt examples. Why then explore the
prospects of a reasons-responsive theory at all? Why not opt for some variation
on a mesh theory?
In my estimation, all mesh theories, not just Frankfurt’s, face a deep problem for
those committed to source compatibilism, one that I suspect is insurmountable. Mesh
theories might comfortably capture conditions sufficient for acting freely: When
an agent acts from a harmonious mesh, she acts freely. But in many cases in
which an agent acts from an unharmonious mesh, mesh theories either generate
the wrong results, or are instead unsatisfactorily silent—at least this is so for
those committed to source compatibilism. Mesh theories generate the wrong
results when they treat acting from a harmonious mesh as also necessary for

27 For other efforts to develop a mesh theory, see Dworkin (1970), and Watson
(1975). Bratman (2007) has explored similar ideas. For a more detailed critical
discussion of mesh theories, see McKenna (2011).
178 Michael McKenna

acting freely; they are unsatisfactorily silent when they refrain from accounting
for an agent’s freedom in such cases.
To explain, consider the case of Sally above. Grant that on Frankfurt’s view,
acting from a harmonious mesh is not only sufficient but also necessary for free
action. Maybe Sally acts from a freedom-compromising compulsion in drinking
that beer; suppose she is a full-blown, hopeless alcoholic. If so, there is no threat to
Frankfurt’s mesh theory; it yields the result that Sally does not act freely insofar as
she fails to satisfy the necessary condition at issue. But maybe instead Sally freely
acts upon her lower-order desire as in opposition to the higher-order preferences
with which she identifies. Suppose she also judges it best not to drink the beer.
This is a classic case of weakness of will, understood in terms of freely acting
contrary to one’s better judgment. A mesh theory such as Frankfurt’s generates the
wrong result in a case like this. It has it that Sally does not act freely when in fact
she does.
A viable option for the mesh theorist is to claim that the necessary condition
which she proposes is merely an ability to act from a harmonious mesh. This
would allow the mesh theorist to handle the distinction between exercises of free,
weak-willed agency and failures that are due to freedom-undermining compulsion.
The weak-willed agent, it can be argued, did not act from a harmonious mesh,
but it remains true that she had the ability to do so; she could have done otherwise.
Obviously, this reply will not work for any mesh theorist who, like Frankfurt, is
also committed to source compatibilism, since it relies on claims about leeway
freedom.
It is open to the mesh theorist who is committed to source compatibilism to
retreat by weakening the reach of her theory. She could give up any proffered
necessity condition and claim that she is only offering sufficient conditions for free
action in those cases involving acting from a harmonious mesh. These conditions
can be shown to be compatible with determinism without in any way relying
upon any assumptions about leeway freedom. After all, this is all that is strictly
required for her to answer the metaphysical challenge of the incompatibilist: the
incompatibilist contends that determinism is incompatible with free action;
she has produced cases of free action from a harmonious mesh that are compatible
with determinism. Case closed. The objection to mesh theories currently
under consideration concerns a refutation of the necessary conditions they pro-
pose. Why can’t the mesh theorist simply retreat by relinquishing this necessary
condition?
Such a retreat strikes me as an unpromising dialectical tactic. A mesh theory that
remained silent with respect to acting freely from an unharmonious mesh would
leave unaccounted for a wide swath of (putatively) free actions. A credible theory of
freedom should be able to account for these sorts of mundane cases of agency.
And so a theoretical lacuna would invite further attempts to account for freedom
in these cases. But then, once these kinds of cases are accounted for in ways that
would be satisfying to compatibilists, it will only be natural to consider whether
the proper account can also explain the freedom present in the cases involving
a harmonious mesh. If so, then the work of accounting for free action will be
 Reasons-Responsiveness, Agents, and Mechanisms 179

done more systematically with these alternative resources. Hence, the work
done by the mesh theory will be rendered otiose. This is, in fact, how I see the
explanatory resources of a reasons-responsive theory of freedom.28

APPENDIX II: ONE POINT WHERE FISCHER

AND RAVIZZA REQUIRE AN ACCOUNT OF
MECHANISMS

To illustrate a place where, without an independent rationale for doing so, Fischer
and Ravizza are too quick to allow sameness of mechanism to fall their way, consider
their contention that reactivity is all of a piece (1998: 73). They offer this proposal
in response to the incompatibilist worry that in the actual world, assuming a
deterministic context, when a blameworthy agent fails to act on the moral reason
to do otherwise, the agent cannot react to this reason. After all, holding fixed
the past and the laws, when that reason is present in the nearest possible world
(the actual one), the agent does not react to it. Fischer and Ravizza attempt to
ward off this objection by arguing that an agent’s reacting differently to a
different reason to do otherwise in another possible world is sufficient to
establish that she is able to react differently to any reason to do otherwise. Of
course, if this were so, she would be able to react differently to the moral reasons
that were present in the actual scenario, even though she did not. It is in this
context that Fischer and Ravizza (1998: 73–4) consider the following
incompatibilist-friendly challenge to their ‘reactivity is all of a piece’ thesis
(which I paraphrase here):
Is it not possible for the same mechanism to get more energy or focus from different incentives?
If it were, then the fact that an agent would react to some reasons to do otherwise, does not
mean that she is able to respond to all reasons to do otherwise, and so it does not mean that she
is able to react differently to the actual moral reason to do otherwise that was present to her at
the time of action.
If mechanisms were to behave this way—that is, in a way that disconfirms that
reactivity is all of a piece—then Fischer and Ravizza would be forced into a difficult
dialectical corner. They would be forced to take head on the incompatibilist worry
that, in a deterministic context, a blameworthy agent in the actual situation in which
she acts is not able to react to the moral reasons that were in fact present at the time.

28 I do, however, think that there is a way for mesh theories and reasons-responsive
theories to form a mutually beneficial alliance. The basic idea would be that a mesh
theory could be used to account for the nature of agency more generally (rather than free
agency). In particular, on the proposal I am considering, only agents able to adopt
higher-order attitudes towards their motivations can even be regarded as candidates for
having the kind of complexity for free agency as specified by something like MRR. Note
this might very well afford reasons-responsive theorists the resources to handle the sorts
of problems alluded to above. (See especially note 26).
180 Michael McKenna

They would have to show directly that, contrary to what the incompatibilist
claims, such an agent is able to react differently to the particular moral reasons
that were present, or they would have to commit to the claim that an agent could
be blameworthy for failing to act on moral reasons that she was not able to act
upon.29
To this challenge, they counter that in those cases in which an agent reacts
differently to some incentives only because she acquires more energy or focus, it is
natural to say that it is because a different mechanism of action is operative (74). This
is one particular point where, I contend, Fischer and Ravizza are too quick to allow
sameness of mechanism—or, rather, in this case, difference of mechanism—to fall
their way (McKenna, 2001: 98–9). They claim that it is natural to think of the
pertinent mechanisms as they propose. But it is hard to see how it is natural. It’s not
clear in the first place what a mechanism of action is. It’s not a term of folk psychology
that has accrued enough use to allow for much in the way of intuitive baggage. Why
can’t a mechanism of action be reactive under, say, extreme pressures, but fail in
normal contexts? After all, literal mechanisms, like thermostats, behave this way all the
time, especially when they begin to fail. Indeed, what is the exception is to find an
actual mechanism—a manmade gizmo of some sort—that does not respond
differentially under extreme pressures, that does not have foibles and limits, and so on.
Fischer has responded to my objections by likening the problem I raise here to
generality problems in other areas of philosophy. A certain level of generality, and
thus lack of specification, is acceptable in other domains of inquiry, such as ethical
theory. So too, Fischer argues, is a degree of generality acceptable when theorizing in
terms of mechanisms of agency (Fischer, 2004: 169). Fischer also points out that it
is unreasonable to demand of a successful theory that all of its elements are fully
analyzed (168). Thus, he resists the burden of offering any “purely ‘principled’
account of mechanism individuation—an account that did not at some level appeal
to intuition” (167–8). He thus remains committed to relying exclusively on appeal
to intuitions in response to difference cases.
While I sympathize with Fischer’s thoughtful reply, it is not enough to put the
objection to rest. Requesting some principled basis of mechanism individuation is
not the same as demanding a full analysis of mechanisms—one that is “purely”
principled. And a degree of generality is of course acceptable in theorizing in areas
such as this one. But when the degree of generality and an exclusive reliance on
intuition will not help adjudicate differences between cases crucial to assessing the
theory, then demanding some principled basis for settling the dispute is only
dialectically fair and reasonable. This is fully consistent with accepting Fischer’s
contention that appeal to intuitions ought to enter the scene at some level.

29 I have defended the former of these two options (2005), while Pereboom (2006)
has advised Fischer and Ravizza to opt for the latter. Watson has also commented on this
issue (2001, as appearing in 2004: 300–1). He suggests that on pain of consistency,
Fischer and Ravizza should opt for the latter option.
 Reasons-Responsiveness, Agents, and Mechanisms 181

REFERENCES

Ayer, A. J. (1954). “Freedom and Necessity.” In his Philosophical Essays. (New York:
St Martin’s Press), 3–20.
Bratman, Michael (2007). Structures of Agency. (New York: Oxford University
Press).
Buss, Sarah and Lee Overton (2002). Contours of Agency: Essays on Themes from
Harry Frankfurt. (Cambridge, MA: MIT Press).
Clarke, Randolph (2009). “Dispositions, Abilities to Act, and Free Will: The New
Dispositionalism.” Mind 118: 323–51.
Davidson, Donald (1973). “Freedom to Act.” In Honderich (ed.), 1973: 67–86.
Dworkin, Gerald (1970). “Acting Freely.” Noûs 4: 367–83.
Fara, Michael (2008). “Masked Abilities and Compatibilism.” Mind 117 (468):
843–65.
Fischer, John Martin (1994). The Metaphysics of Free Will. (Oxford: Blackwell
Publishers).
—— (2002). “Frankfurt Style Compatibilism.” In Buss and Overton (eds.), 2002:
1–26.
—— (2004). “Responsibility and Manipulation.” Journal of Ethics 8. 2: 145–77.
—— (2006). “The Free Will Revolution (Continued).” Journal of Ethics 10, No. 3:
315–45.
—— (2011). Deep Control. (New York: Oxford University Press).
—— and Mark Ravizza (1998). Responsibility and Control: An Essay on Moral
Responsibility. (Cambridge: Cambridge University Press).
Frankfurt, Harry (1969). “Alternate Possibilities and Moral Responsibility.” Journal
of Philosophy 66: 829–39.
—— (1971). “Freedom of the Will and the Concept of a Person.” Journal of
Philosophy 68: 5–20.
Ginet, Carl (1966). “Might We Have No Choice?” In Lehrer, 1966: 87–104.
—— (2006). “Working With Fischer and Ravizza’s Account of Moral Responsi-
bility.” Journal of Ethics 10: 229–53.
Haji, Ishtiyaque (1998). Moral Appraisability. (New York: Oxford University
Press).
Hobart, R. E. (1934). “Free Will as Involving Indeterminism and Inconceivable
Without It.” Mind 43: 1–27.
Honderich, Ted (ed.) (1973). Essays on Freedom and Action. (London: Routledge &
Kegan Paul).
Howard-Snyder, Daniel and Jeff Jordan (eds.) (1996). Faith, Freedom, and Ration-
ality. (Lanham, MD: Rowman and Littlefield).
Hunt, David P. (2000). “Moral Responsibility and Unavoidable Action.” Philo-
sophical Studies 97: 195–227.
Irwin, T. H. (1980). “Reason and Responsibility in Aristotle.” In A.O. Rorty (ed.),
1980: 117–55.
Lewis, David (1997). “Finkish Dispositions.” Philosophical Quarterly 47: 143–58.
182 Michael McKenna

McKenna, Michael (2001). “Review of John Martin Fischer and Mark Ravizza’s
Responsibility & Control.” Journal of Philosophy. XCVIII, no. 2: 93–100.
—— (2003). “Robustness, Control, and the Demand for Morally Significant
Alternatives.” In Moral Responsibility and Alternative Possibilities (eds.), Widerker
and McKenna. 2003.
—— (2005). “Reasons Reactivity & Incompatibilist Intuitions.” Philosophical
Explorations 8. 2: 131–43.
—— (2008). “Frankfurt’s Argument against Alternative Possibilities: Looking
Beyond the Examples,” Noûs 42: 770–93.
—— (2011). “Contemporary Compatibilism: Mesh Theories and Reasons-
Responsive Theories.” In R. Kane (ed.), 2011, Oxford Handbook of Free Will,
2nd edn. (New York: Oxford University Press), 175–98.
Martin, C. B. (1994). “Dispositions and Conditionals.” Philosophical Quarterly 44:
1–8.
Mele, Alfred (1995). Autonomous Agents. (New York: Oxford University Press).
—— (2006a). Free Will and Luck. (New York: Oxford University Press).
—— (2006b). “Fischer and Ravizza on Moral Responsibility.” Journal of Ethics 10.
3: 283–94.
—— and David Robb (1998). “Rescuing Frankfurt-Style Cases.” Philosophical
Review 107: 97–112.
Moore, G. E. (1912). Ethics. (Oxford: Oxford University Press).
Naylor, Marjory (1984). “Frankfurt on the Principle of Alternate Possibilities.”
Philosophical Studies 46: 249–58.
Pereboom, Derek (2001). Living Without Free Will. (Cambridge: Cambridge
University Press).
—— (2006). “Reasons Responsiveness, Alternative Possibilities, and Manipulation
Arguments Against Compatibilism: Reflections on John Martin Fischer’s My
Way.” Philosophical Books 47: 198–212.
Sartorio, Carolina (forthcoming). “Actuality and Responsibility.” Mind.
Smith, Michael (2003). “Rational Capacities, or: How to Distinguish Recklessness,
Weakness, and Compulsion.” In Stroud and Tappolet (eds.), 2003: 17–38.
Stroud, Sarah and Christine Tappolet (eds.) (2003). Weakness of Will and Practical
Irrationality. (New York: Oxford University Press).
Stump, Eleonore (1996). “Libertarian Freedom and the Principle of Alternative
Possibilities.” In Howard-Snyder and Jordan, 1996: 73–88.
Van Inwagen, Peter (1975). “The Incompatibility of Free Will and Determinism.”
Philosophical Studies. 27:185–99.
Vihvelin, Kadri (2004). “Free Will Demystified: A Dispositional Account.” In
Philosophical Topics 32: 427–50.
Wallace, R. Jay (1997). “Review of John Martin Fischer’s The Metaphysics of Free
Will.” Journal of Philosophy 94. 3: 156–59.
Watson, Gary (1975). “Free Agency.” Journal of Philosophy 72: 205–20.
—— (2001). “Reason and Responsibility.” Ethics 111: 374–94.
—— (2004). Agency and Answerability. (New York: Oxford University Press).
Whittle, Ann (2010). “Dispositional Abilities.” Philosophers’ Imprint 10. 12: 1–23.
 Reasons-Responsiveness, Agents, and Mechanisms 183

Widerker, David (2006). “Libertarianism and the Philosophical Significance of

Frankfurt Scenarios.” Journal of Philosophy 103: 163–87.
—— and Michael McKenna (eds.) (2003). Moral Responsibility and Alternative
Possibilities. (Aldershot, UK: Ashgate Press).
Wiggins, David (1973). “Towards a Reasonable Libertarianism.” In Honderich
1973: 31–62.
Zagzebski, Linda (2000). “Does Libertarian Freedom Require Alternate Possibil-
ities?” Philosophical Perspectives 14: 231–48.
 7
Responsibility, Naturalism, and
“The Morality System”*
Paul Russell

Theory typically uses the assumption that we probably have too many
ethical ideas . . . Our major problem now is actually that we have not
too many but too few, and we need to cherish as many as we can.
Bernard Williams, Ethics and the Limits of Philosophy
Lying at the heart of P. F. Strawson’s core strategy in “Freedom and
Resentment” is his effort to direct his naturalist claims and observations
against not only the philosophical extravagance of a general skepticism about
moral responsibility but also against all nonskeptical attempts to provide
responsibility with some form of external rational justification (Strawson
1962: 23). According to Strawson, efforts of this kind are not only misguided
and unconvincing in themselves, when they fail they encourage a general
skepticism about moral responsibility. The alternative strategy that Strawson
pursues is one that places the foundations of responsibility on our natural,
universal emotional propensities and dispositions relating to moral
sentiments or reactive attitudes. This naturalistic turn invites us to focus
our attention on familiar facts about human moral psychology, and to drop
our focus on the analysis of the concept of “freedom” as a way of dealing with
the threat of determinism. Beyond these general features, however, the details
of Strawson’s strategy become both complex and layered. As a consequence
of this, interpretations and assessments of his arguments differ greatly. There

* I am grateful to David Shoemaker and other participants at the New Orleans

Workshop on Agency and Responsibility (November 2011) for their very helpful
comments and criticism. Versions of this paper were also presented at a Workshop on
Moral Agency, Deliberative Awareness, and Conscious Control (Erasmus University,
Rotterdam, October 2010) and at the University of Arizona (February 2012). Thanks
also to those who were present on those occasions for their comments and suggestions,
and especially to Michael McKenna, Maureen Sie, and Andras Szigeti.
 Responsibility, Naturalism, and “The Morality System” 185

is, nevertheless, a general consensus among both followers and critics alike
that there are significant strands in Strawson’s specific naturalistic arguments
that are implausible and unconvincing and that some “retreat” from the
original strong naturalist program that he advanced is required.
In this paper I take up two closely related issues arising out of this overall
problem. First, I want to consider if the right way to amend and modify the
naturalistic program is to adopt a “narrow” construal of our moral reactive
attitudes along the lines proposed by R. Jay Wallace, one of Strawson’s
most prominent followers on this subject. The narrower approach, as I will
explain, involves a substantial retreat from Strawson’s original naturalistic
program and has significant implications for Strawson’s core claim that our
commitment to responsibility requires no external rational justification.
The second issue that will be addressed is whether or not we should
interpret moral responsibility entirely within the confines of what Bernard
Williams has described as “the morality system.” (Williams 1985: ch.1).
The narrow construal of responsibility requires that we understand moral
responsibility within the conceptual resources provided by the morality
system, making notions of obligation, wrongness, and blame essential
to the analysis of moral responsibility. There is, therefore, an intimate
connection between these two issues. With respect to both these issues
I will argue that the narrow approach, while it has legitimate criticisms to
make of Strawson’s original strategy, nevertheless takes us in the wrong
direction and involves an unacceptable distortion and truncation of moral
responsibility.

1. TWO MODES OF NATURALISM

Strawson’s naturalistic account of moral responsibility insists that a proper

understanding of this matter must begin with a description of what is
involved in holding a person responsible. The key to his analysis is the
role that reactive attitudes play in this sphere, where this is understood in
terms of our natural emotional responses to the attitudes and intentions
that we manifest to each other. Strawson’s naturalistic strategy, as based on
this general observation, has two aspects or dimensions that need to be
carefully distinguished—although this is not done in his own presentation.
There is, in the first place, a strong form of naturalistic argument that
involves the claim that even if we had some theoretical reason to abandon
entirely or altogether suspend our reactive attitudes (e.g. as required by
skeptical arguments based on considerations of determinism), it would be
psychologically impossible for us to do this (Strawson 1962: 9–12,18).
A systematic repudiation of all reactive attitudes of this kind, he argues,
186 Paul Russell

would result in “a thoroughgoing objectivity of attitude” to others, with the

resulting loss of all our interpersonal personal relations (1962: 12–3).
While the objective attitude may be appropriate when we are dealing
with the abnormal and incapacitated, and may even on occasion be
available to us when dealing with normal adults, we cannot “do this for
long, or altogether” (1962: 9–10). Armed with these claims, Strawson
suggests an easy way to deal with the skeptic. No matter what theoretical
arguments may be presented, our human nature is such that we will
continue to feel or entertain tokens of reactive attitude. According to this
view, skeptical arguments will inevitably fail to dislodge or wholly eradicate
these attitudes and responses. Let us call this strong line of reply token-
naturalism, since it turns on the claim that our tokens of reactive attitudes
cannot be systematically eliminated or abandoned whatever (philosophical)
arguments may be advanced against them.
Most philosophers have found Strawson’s token-naturalism too strong
and unconvincing. There are two basic objections to be made against
it.1 First, the psychological claim that it makes is doubtful in point of fact.
It is not at all obvious that we are constitutionally incapable of entirely
ceasing to entertain tokens of reactive attitude should we be persuaded
that they are systematically unjustified. Moreover, even if the
psychological claims were true, this does nothing to remove or discredit
the objections put forward by the skeptic, since these claims do not
address the justificatory issue that concerns us. It would, in fact, be
disturbing to discover that we will naturally continue to entertain
tokens of reactive attitude in face of well-founded grounds for
discrediting these emotional responses to others.
While Strawson’s token-naturalism may be judged too strong, there is a
weaker form of naturalism that is more convincing and has attracted greater
interest. What is crucial to naturalism, on this account, is that we are all
liable or disposed to reactive attitudes. It is from within the framework of
this weaker form of naturalism that Strawson develops his general rationale
of excuses and exemptions. While we may all have a natural liability to
reactive attitudes, particular tokens can be discredited by reference to
excuses and exemptions. Excuses operate by way of showing that the agent’s
conduct was consistent with the underlying general demand for “some
degree of goodwill or regard” (1962: 7). In cases of this kind, involving
an accident, ignorance and other such factors, the conduct in question does
not manifest any ill-will or malicious intent, even if some injury has

1
For a more detailed development of this analysis of Strawson’s naturalism see
Russell (1992).
 Responsibility, Naturalism, and “The Morality System” 187

occurred. With respect to exemptions, however, we are asked to view the

individual as an altogether inappropriate target of our reactive attitudes on
the general ground that we are dealing with someone for whom we cannot
make the moral demand due to an abnormality or incapacity of some
relevant kind. Strawson employs this two-level account of excuses and
exemptions to show that the thesis of determinism fails to engage any of
these recognized considerations and cannot, therefore, constitute a basis for
systematically discrediting all our reactive attitudes associated with moral
responsibility.
This weaker naturalist approach may be described as a form of type-
naturalism, where this is understood in terms of our natural disposition to
reactive attitudes (just as we have a natural liability to love, fear, joy and
other basic emotions). What is crucial, however, is that unlike token-
naturalism, type-naturalism offers no easy way of discrediting the skeptic.
On the contrary, it is essential, on this approach, that a plausible account of
excusing and exempting conditions is provided consistent with compatibi-
list commitments. At this level, concerning our natural liability to reactive
attitudes, we must still engage in arguments that counter the skeptical
challenge. At the same time, type-naturalism does insist on the “internal”
nature of these replies to the skeptic (1962: 23). While it is possible that
our token reactive attitudes could be systematically discredited from
within, there remains no need or possibility of providing an external,
rational justification of a more general kind for our (natural) propensity
to these emotions (any more than we need to do this for our similar liability
to love, fear, etc.). Although justificatory issues remain with us, and cannot
be evaded by way of token-naturalist claims, these justificatory require-
ments do not take the form of a demand for general or external rational
justifications.

2. REACTIVE ATTITUDES AND NARROW

R E S PO N S I BI L I T Y

Having established a distinction between token- and type-naturalism, let us

now consider Wallace’s amended account of the Strawsonian project, as
developed on the basis of his narrow construal of the reactive attitudes.
Wallace’s compatibilist account weaves together two distinct strands of
thought. The first is a broadly Strawsonian description of holding people
responsible, interpreted in terms of our reactive attitudes (Wallace 1994:
8–12). The other strand is his Kantian theory of reflective self-control or
moral agency (1994: 12–15). Taken together, these two strands constitute
188 Paul Russell

what Wallace calls his “normative interpretation” of responsibility, which

maintains that the correct way to understand what it is to be a morally
responsible agent is by way of describing those conditions under which it is
fair to hold an agent responsible (1994: 5,15, 64). Our stance of holding a
person responsible must itself be understood in terms of the mutual
dependence between expectations and reactive attitudes (1994: 20–5). To
hold someone to an expectation, or a demand of some kind, is to be
susceptible to reactive attitudes when the expectation is violated (1994:
21). We are susceptible to a reactive attitude if we either feel this emotion or
believe that it would be appropriate to feel it in these circumstances (1994:
23, 62). While these moves are generally consistent with Strawson’s original
approach, Wallace aims to substantially modify and amend this approach
by providing a narrower and more fine-grained account of moral reactive
attitudes.
According to Wallace, Strawson’s account of reactive attitudes is too
“inclusive” and needs to be refined into several different categories relating
the various emotions we are concerned with (1994: 25–40). On Wallace’s
analysis we need to draw two overlapping distinctions (1994: 33). The first
is between moral and nonmoral reactive attitudes. Both forms of reactive
attitudes depend on their reciprocal relationship with a system of expect-
ations. In the case of moral reactive attitudes the relevant set of expectations
are justified with reference to moral reasons, and the expectations they
justify are obligations (1994: 35–6). However, not all expectations are
backed by moral reasons, as we find in the case of etiquette, where a breach
may generate resentment even though no distinctive moral claim has been
violated (1994: 36–7). According to Wallace, the central cases of reactive
attitudes are the emotions of resentment, indignation, and guilt (1994:
29–30), which are all negative in character. This is a feature of our reactive
attitudes, he suggests, that explains their “special connection to the negative
responses of blame and moral sanction” (1994: 62, 71).
Wallace draws out several significant points from this narrow account
of the reactive attitudes. The first is that reactive attitudes are not coexten-
sive with the emotions we feel towards people with whom we have
interpersonal relationships, since reactive attitudes must be identified
with reference to their “constitutive connections with expectations”
(1994: 31). There are many interpersonal emotions we may experi-
ence—such as attachment, friendship, sympathy, love, and so on—that
cannot be counted as reactive attitudes since they do not have any relevant
connection with expectations. It follows from this that we must reject
Strawson’s claim that a life without reactive attitudes would commit us to
an (impossible) universal adoption of the “objective attitude.” People may
continue to entertain various other forms of interpersonal emotions and
 Responsibility, Naturalism, and “The Morality System” 189

relations even in the complete absence of reactive attitudes narrowly con-

ceived. Moreover, against Strawson, Wallace argues that even if some
form of interpersonal relations are an inescapable feature of human life, it
does not follow that the reactive attitudes are “similarly inevitable” (1994:
31). This point leads Wallace to his second distinction relating to his
account of the reactive attitudes.
We also need to draw a distinction, Wallace says, between moral reactive
attitudes and other kinds of moral sentiment. Not all moral sentiments take
the form of moral reactive attitudes, with some identifiable tie to expect-
ations and obligations. Among the examples of nonreactive moral senti-
ments Wallace cites are shame, gratitude, and admiration; all of which
involve different kinds of “modalities of moral value” (1994: 37–8). This
distinction allows us to acknowledge that there are other cultures with
forms of ethical life that do not have any commitment to reactive attitudes
but may have other kinds of moral sentiment, as we find in shame cultures
(1994: 31, 37–40). Evidently, then, on Wallace’s narrow interpretation of
reactive attitudes, Strawson’s naturalism is excessive not only at the token
level, but also at the type level, since it is entirely conceivable that there are
cultures where members are not subject or liable to reactive attitudes at all.
For Wallace, abandoning naturalism at both the token and type levels is a
price well worth paying, as it is the only way to avoid an overly inclusive
account of reactive attitudes and a false dichotomy between the reactive
attitudes and the objective attitude. The narrow construal, Wallace main-
tains, is more faithful to the relevant psychological and sociological facts
and also permits us to identify more accurately the justificatory issues that
arise with respect to issues of moral responsibility.
Wallace is well aware that his narrow construal of the reactive attitudes
commits him to an interpretation of moral responsibility understood
entirely in terms of the conceptual resources of “the morality system.”2
The morality system is understood as a particular form of ethical life,
associated with our modern, Western, Christian culture. Its central
normative concepts are obligation and blame, along with related notions
of wrongness and voluntariness. These are all the same key elements that
feature in Wallace’s narrow construal of moral responsibility. The narrow
account renders moral responsibility, so interpreted, as a local and
contingent cultural achievement, involving a legalistic, neo-Kantian view
of morality. Understanding moral responsibility in narrow terms presents
its adherents with their own set of difficulties. Some of these difficulties are
anticipated in Wallace’s discussion.

2
Wallace 1994: 39–40, 64–66; and Williams 1985: esp. ch. 10.
190 Paul Russell

3. THE C OST S OF GOIN G N ARR OW

One of the most obvious costs of analyzing moral responsibility in terms of

a narrow interpretation of our reactive attitudes is that it commits us to an
“asymmetrical” account with respect to “worthy and unworthy actions”
(Wallace 1994: 71). Since our moral reactive attitudes are, on the narrow
interpretation, aroused only when expectations that we endorse are
violated, it follows that any moral sentiments we experience that are
positive responses to other “modalities of moral value” are not strictly
reactive attitudes—and so no part of moral responsibility. While we may
feel gratitude or admiration for a morally worthy act, the moral emotions
involved are not to be accounted for in terms of the specific structure of
beliefs about the violation of moral obligations (1994: 37–8). Wallace is
unapologetic about this asymmetrical feature of the narrow view on the
ground that it accounts for the “special connection” that exists between
holding people responsible and our retributive practices involving blame
and punishment (1994: 71). Beyond this, Wallace also claims that holding
a person responsible for a worthy action “does not seem presumptively
connected to any positive emotion in particular” (1994: 71).
Wallace’s defense of the asymmetrical features of the narrow account of
moral responsibility is unconvincing for several reasons. First, while we may
agree that there is a close connection between moral reactive attitudes and
our retributive practices, this does not imply that we need to interpret our
reactive attitudes as exclusively negative in character, and as always connected
with blame and moral sanctions. If we allow that there are reactive attitudes
of a positive kind, based on beliefs concerning worthy actions and admirable
character traits, then these too may be connected with positive retributive
dispositions such as praise and rewards. Second, it is not obvious that our
emotional resources with respect to our responses to morally worthy actions
and traits are any more impoverished or limited than in the case of our
negative reactive attitudes. As Wallace’s own observations suggest, we not
only have “thin” ethical concepts such as approval and praise (correlates to
disapproval and blame), we also have many “thicker” concepts, including
gratitude and admiration. Finally, and most importantly, any asymmetrical
account of the kind advanced by the narrow view inevitably truncates and
distorts our experience of moral life and the various ways in which our
reactive attitudes are grounded and directed. A one-sided view that is exclu-
sively concerned with negative reactive attitudes, focusing entirely on their
connection with blame and retribution, offers us an impoverished and
unbalanced interpretation of responsibility and fails to properly accommo-
date the constructive role of reactive attitudes in endorsing and supporting
morally worthy or admirable actions and traits.
 Responsibility, Naturalism, and “The Morality System” 191

Another objection to the narrow interpretation, which Wallace also

anticipates and tries to fend off, is that it presents an account of moral
responsibility that has a “local” bias toward (our own) modern, Western
Christianized culture—i.e. toward “the morality system.” It follows from
the narrow account that the ancients Greeks and shame cultures, among
others, have practices that are at best “analogous” to ours, based as they are
on different moral beliefs with distinct patterns of moral response (1994:
65–6) Considered in terms of the narrow interpretation, these alien forms
of ethical life do not share our understanding of moral responsibility, not
simply in the sense that they have a different understanding but rather that
they have no commitment to moral responsibility (since their ethical
responses cannot be understood in terms of the narrow account of reactive
attitudes). This view of things renders moral reactive attitudes and moral
responsibility as both local and contingent, and thereby places a conceptual
barrier between ourselves (modern, Western, etc.) and alien forms of ethical
life that are removed from us in historical time and geographical space.
From one point of view this narrow account oversimplifies our own
(modern) attitudes and practices, which are not perfectly or purely repre-
sented by “the morality system” and evidently involve dimensions of
holding people responsible that cannot be compressed into the narrow
and rigid framework provided. From another point of view, it denies us the
critical apparatus and resources to question and challenge the way we
(moderns) have (locally) arranged and structured our own attitudes and
practices relating to responsibility. When we are confronted with other
cultures and forms of ethical life outside “the morality system” they must,
according to the narrow account, be set aside as—by definition—no longer
possessing any conception of responsibility. Even if it is granted that we
moderns are straightforwardly committed to “the morality system” and its
narrow construal of responsibility, this still puts unnecessary and excessive
(conceptual) distance between ourselves and these alternative forms of
ethical life. More specifically, it erects a conceptual barrier to any genuine
critical exchange and confrontation on the subject or question of responsi-
bility itself—since there is, on this account, no shared or common ethical
life with respect to the attitudes and practices that are actually constitutive
of moral responsibility.3

3
Clearly confrontation between ethical cultures that are removed from each other in
historical time or geographical space may be, as Williams observes, either “notional” or
“real” (Williams 1985: ch. 9). Be this as it may, historical sensitivity about the
contingency of our own “local” commitments naturally puts pressure on reflective
confidence in our own attitudes and practices. To this extent our awareness of other
modes and forms of ethical life, less attached to the rigidities of “the morality system,”
192 Paul Russell

Finally, there is a further objection to the narrow construal of moral

responsibility, which raises difficulties that Wallace does not anticipate or
directly address. Granted that moral responsibility narrowly interpreted in
terms of the concepts provided by the morality system is both local and
contingent, it follows that we must reject Strawson’s original claims con-
cerning our type-naturalist commitments to the reactive attitudes. If this is
the case, then a fundamental plank of Strawson’s original naturalistic
strategy must be abandoned: namely, the claim that we do not need and
cannot provide any external rational justification for moral responsibility. If
we accept the narrow construal, then there is no natural, universal liability
to reactive attitudes and they do not serve as a natural foundation for all
recognizably human forms of ethical life. Since the framework of moral
responsibility is erected around culturally local forms of moral emotion,
confrontations with other cultures and forms of ethical life will place us in
the position of needing external rational justification for the entire frame-
work of moral responsibility so conceived. Internal justifications, provided
in terms of a rationale of excuses and exemptions, will not serve this
purpose even if it is successful in fending off the (internal) skeptical
challenge based on worries about determinism. To this extent, the skeptical
threat remains with us not just at the token level but also at the type level.
Nor is it an option to retreat back to other “analogous” forms of moral
emotion since, if the narrow account is correct, this will not secure or
preserve responsibility properly understood. The crucial point remains that
if we embrace the narrow construal then, contrary to Strawson’s core
original view, we are faced with the task of providing the whole edifice of
moral responsibility with an external rational justification.

4 . T Y P E - N A T U R A L I S M AN D B R O A D R E S P O N S I B I L I T Y

Whatever difficulties may be found for the narrow construal of moral

responsibility it is important to begin with a full appreciation of Wallace’s
critique of the original Strawsonian strategy, much of which is justified.
There are four particular features of Wallace’s critique that should be
endorsed as clearly justified.
(1) Token-naturalism is, as we have noted, psychologically implausible
and fails, in any case, to discredit the justificatory issues advanced by

may bring us to question whether our confidence in “our modern concept of

responsibility” is altogether well-founded.
 Responsibility, Naturalism, and “The Morality System” 193

the skeptical challenge (e.g. as based on worries about the implications

of determinism).
(2) We do require a more fine-grained and less inclusive account of the
reactive attitudes. In particular, it is essential that we exclude interper-
sonal emotions that lack any relevant cognitive element containing
ethical content, as in the case of emotions such as love, sympathy,
friendly feeling, and so on.
(3) It is also essential that any plausible naturalistic strategy is one that is
historically or genealogically sensitive, displaying an awareness of the
considerable variation in human ethical life and the range of moral
emotions this may involve. In particular, we must avoid any crude
form of naturalism that projects our (local) sentiments onto (alien)
others.
(4) Finally, it is entirely correct to argue, as Wallace does, that we need a
theory of moral capacity to provide the basis for an account of exemp-
tions, in order to answer the (internal) skeptical challenge that deter-
minism would somehow render us all morally incapacitated and thus
inappropriate targets of reactive attitudes. This last issue is, however,
not itself part of the revised Strawsonian analysis of holding people
responsible and is not, therefore, a matter for our present concern.4
Our concerns rest with the issues arising out of the first three items on the
list above. In the case of all three of these items, I will argue, Wallace’s
narrow view construal of responsibility involves a series of unnecessary and
misleading oppositions. It is possible for us to avoid the weaknesses
identified in Strawson’s original strategy without collapsing into the exces-
sively narrow view of moral responsibility that reflects the meager resources
of “the morality system.”
We may begin by noting that we can readily reject token-naturalism
without rejecting type-naturalism. If we adopt this approach then, it is true,
we will be denied the sort of easy way with skepticism that token-natural-
ism encourages. Moreover, as already explained, the type-naturalist
approach, building on our natural liability to reactive attitudes, still leaves
us needing a theory of excuses and exemptions that is consistent with
compatibilist commitments, if we are to defeat the skeptical challenge.
Although the skeptical effort to discredit all tokens of reactive attitudes is
one that we must take seriously, if we accept type-naturalism, we do not

4
I have argued elsewhere that there is a more intimate relationship between our
capacity for holding agents responsible and our capacity for reflective self-control than
(Kantian) theories such as Wallace’s acknowledge. See, in particular, Russell 2004 and
Russell 2011.
194 Paul Russell

need any external rational justification for reactive attitudes (i.e. justifica-
tion at the foundational level). Whether this further claim is acceptable or
not will depend on whether we accept the narrow construal of our reactive
attitudes. Clearly if we go narrow, then type-naturalism and the associated
claim regarding the dispensability of external rational justifications must be
dropped. The resolution of this issue depends, therefore, on our interpret-
ation of reactive attitudes.
As I have argued, although we do need a narrower account of our reactive
attitudes, we need to make sure we do not go too narrow, as otherwise we
will generate some of the difficulties that have already been noted. Wallace’s
narrow view places considerable and appropriate emphasis on the “propos-
itional content” involved in the beliefs that serve to delineate our reactive
attitudes (1994: 11,19,74).5 The narrow view would restrict the contents in
question to the limited range provided by the conceptual resources of “the
morality system.” It is these limits that result in the problems of asymmetry
and localism, as we have described them. We need to find, therefore, a
middle path that avoids the inclusiveness of Strawson, on one side, and the
excessively narrow approach of Wallace on the other. To put ourselves back
on the right track we may turn again to Bernard Williams’s critique of “the
morality system.”
The narrow view, as we have seen, presents ethical considerations in highly
restricted terms, specifically with reference to obligation and blame, which is
appropriate when obligations are violated. Williams identifies these features
as central to the morality system (1985: ch. 10). This tendency to reduce and
simplify is also manifest in ethical theory, a philosophical project which is
itself intimately linked to the assumptions and prejudices of the morality
system (1985: ch. 1). One aim of ethical theory is to provide an account of
morality that will provide an exact boundary between ethical and nonethical
considerations. This is done primarily by reducing the diversity of ethical
(and nonethical) considerations, with a view to identifying a narrow and
strict range of ethical considerations that may serve as moral reasons available
to all rational agents—the universal constituency. The most notable features
of moral theory are its simplicity, reductionism, and systemization of our
ethical concepts and claims. Williams’s critique of the morality system
involves challenging and rejecting these assumptions. In the first place,
while our ethical considerations certainly include obligations, under some

5
Wallace claims that Strawson’s account of reactive attitudes does not manage to
clearly connect them with any propositional content (1994: 39). This charge seems
unfair to Strawson since he is careful to ground reactive attitudes in our beliefs about the
attitudes and intentions of other human beings and “the very great importance” that we
attach to them (Strawson 1962: 5).
 Responsibility, Naturalism, and “The Morality System” 195

interpretation, they extend well beyond this. The scope of the ethical relates
more broadly to “the demands, needs, claims, desires, and, generally the lives
of other people, and it is helpful to preserve this conception in what we are
prepared to call an ethical consideration” (1985: 12; see also 1985: 153,
where Williams mentions our need to share a social world in relation to these
various ethical considerations). What is required, from this perspective, is an
account of ethical considerations that also includes forward-looking concerns
relating to welfarism and utilitarianism, as well as ethical considerations that
relate to our ideals and self-conceptions that mark out actions that fail the
standards and boundaries that we may set for ourselves (e.g. considerations of
what we regard as demeaning, base, dishonorable, etc.). When we interpret
ethical considerations in this broader manner we find that these interests are
plural and varied in their nature and secure no sharp boundary between
ethical and nonethical considerations. Vagueness, conflict, and diversity—
contrary to the demands of “theory” and the prejudices of “the morality
system”—are of the essence of human ethical life.
Our own ethical reactive attitudes must be understood in these broader
and vaguer terms. One of the implications that Williams draws from this is
that we should be skeptical of the effort to understand reactive attitudes in
the reductive, thin language of binary judgments; approval and disap-
proval, guilt and innocence, and so on (1985: 37, 177, 192). If we recon-
figure our ethical reactive attitudes in terms of a broader construal of the
ethical considerations that ground them and serve as their propositional
content, then we may acquire a very different understanding of the scope
and content of moral responsibility, as based upon these emotions. Our
ethical qualities are manifest in the “deliberative priority” and “import-
ance” that we give to ethical considerations as expressed in our conduct and
character. So interpreted, ethical reactive attitudes may be construed as
reactive ethical value, where this is understood as emotional responses to the
weight and value given by an agent or person to ethical considerations
widely conceived (i.e. in terms of our human needs, interests, welfare,
claims, and the requirements of social cooperation). Ethical reactive atti-
tudes involve coming to see a person in a certain ethical “light” based on
these lower-order evaluations of their ethical qualities. Clearly we have
varied and diverse ethical norms and standards that serve as the relevant
basis for evaluating an agent’s ethical qualities understood in these terms.
These evaluations of agents based on their ethical qualities serve to generate
or arouse a myriad of ethical reactive attitudes which may be either
“positive” or “negative” in nature. As Williams argues, our ethical and
emotional language here is not at all “thin” (e.g. praise and blame etc.) but
is “thick” and varied, involving notions such as “being creepy,” or a “cad,”
and so on—all of which are responses loaded with ethical significance.
196 Paul Russell

Different cultures and different forms of ethical life will not only have
different lower-order ethical norms, they also deploy a different or variable
set of ethical reactive attitudes (reflecting their variable propositional content).
The significance of this criticism of “the morality system,” along with the
style of ethical theory associated with it, for our understanding of ethical
reactive attitudes should be clear. The revised account is broad enough to
accommodate positive ethical reactive attitudes (e.g. gratitude, admiration,
etc.) as well as “alien” reactive attitudes (e.g. shame) all under the umbrella
of those ethical considerations that serve to ground or justify them. This
avoids the costs of going too narrow, by way of relying on the limited and
restricted resources of the morality system. The broader construal is,
nevertheless, controlled and focused enough to exclude elements that do
not relate at all to ethical considerations and ethical qualities (e.g. friendly
feeling, sympathy, romantic love, etc. do not count as ethical reactive
attitudes because they are not reactive to ethical qualities as such). On
this analysis, we should not be surprised or disappointed to find that
there is no sharp or clear boundary between ethical reactive attitudes and
other emotional responses to qualities and features of those with whom we
are dealing. No such sharp boundary should be expected if we want an
accurate understanding of the nature of ethical life and the way in which it
“bleeds” into human life in general.6
Taking this broader approach to ethical reactive attitudes has other signifi-
cant advantages as well. It avoids, for example, the “legalism” of the narrow
account, which turns moral responsibility into a model of legal responsi-
bility—eliminating the more nuanced and complex set of responses we have
outside legal contexts. We also avoid the failings of what Williams refers to as
“progressivism,” the assumption that we moderns alone have access to a full
and complete concept of moral responsibility and are “better off ” than those
who lack our own understanding.7 It is Williams’s view that not only should
we be open to the possibility that we might learn from the ancients, this is in
fact our situation. Learning from the ancients is possible—and desirable—
precisely because we share a concept of moral responsibility with them,
however differently we may interpret various key elements associated with
it (1993: 55).
For our present purposes, our concern is not to present a worked-out
alternative to the narrow model of reactive attitudes—not the least because,

6
This is, of course, a recurrent theme throughout Williams’s writings.
7
Williams 1985: 32 n. 2; and, more generally, Williams 1993: esp. ch. 1. Williams’s
view is, of course, the opposite of this, since he holds that we would be better off without
the morality system (1985: 174), just as we don’t need ethical theory and should
abandon its aims (1985: 17, 74).
 Responsibility, Naturalism, and “The Morality System” 197

for reasons given, this may itself be a problematic ambition driven by the
aims of “ethical theory.” What is important, however, is to insist on finding
some middle ground that can accommodate ethical reactive attitudes
broadly conceived without expanding this set to include interpersonal
emotions that have no relevant ethical content (i.e. which do not involve
our emotional reactions to a person’s ethical qualities). Wallace’s own
observations suggest that this can readily be done, since he allows “analo-
gous” forms of responsibility and also speaks of “responsibility for worthy
acts” (Wallace 1994: 38–40, 64–6, 71). To see, in a particular case, how
this middle ground between an excessively narrow and overly inclusive view
may be found let us consider shame. Shame may be based on standards and
norms that have no ethical content, as in the case of concern about one’s
physical appearance (e.g. my frail constitution) or economic status (e.g.
my family’s poverty). In other cases, however, the relevant standards and
norms may move into the territory of our ethical qualities and
characteristics, such as feeling shame about being lazy or being vulgar.
Whether a response is an ethical reactive attitude or not will depend on
the nature of the quality or consideration it is a reaction to. There are,
moreover, clear cases of ethical shame that cannot be analyzed or
understood in terms of the apparatus of obligation and doing wrong.
We may, for example, feel ashamed of failing to live up to our own ethical
ideals and standards, even when we are well aware that we have not failed
to comply with any obligations and cannot be blamed for our conduct.
An example of this is provided in Joseph Conrad’s Lord Jim, where Jim is
ashamed of himself because he fails to act heroically and is, therefore,
disappointed in himself in these ethical terms but not in terms of any
recognizable requirements of “the morality system.”8 The general point
here is that there is a wide range of ethical reactive attitudes lying outside
the theoretical schema of the narrow interpretation that, nevertheless, do
not collapse into an overly “inclusive” set of interpersonal emotions
lacking any relevant ethical content. Some cases of shame will be cases
of ethical reactive attitudes and others will not. What will settle this issue
will be the specific content and target of what we are ashamed of.
Moreover, the fact that there are no sharp or precise boundaries to draw
here is a failing only if we assume the prejudices of the morality system
and the forms of “theorizing” associated with it.
We have noted that it is essential that the naturalist approach to moral
responsibility should be sensitive to historical and cultural variations with
regard to our understanding of moral responsibility and the specific and

8
For an illuminating discussion of this example see Doris 2002: 160–4.
198 Paul Russell

various forms which our ethical reactive attitudes may take. It should be
pointed out that Strawson is himself alive to these concerns. Speaking of
our increased “historical and anthropological awareness of the great variety
of forms which these human attitudes may take at different times and in
different cultures” Strawson says:
This makes one rightly chary of claiming as essential features of the concept of
morality in general, forms of these attitudes which may have a local and temporary
prominence. No doubt to some extent my own descriptions of human attitudes
have reflected local and temporary features of our own culture. But an awareness of
variety of forms should not prevent us from acknowledging also that in the absence
of any forms of these attitudes it is doubtful whether we should have anything that
we could find intelligible as a system of human relationships, as human society.
(1962: 24–5. Strawson’s emphasis)
It may be argued, along the lines of Wallace’s criticisms of Strawson’s
claims about the objective attitude, that Strawson’s remarks in this passage
run together two distinct issues. One claim is that we could not recognize a
society as truly human without any ethical reactive attitudes. Clearly this
need not be the case, so long as we do not overly expand the class of ethical
reactive attitudes to include all interpersonal emotions. Nevertheless, the
general point that Strawson is primarily concerned to make in this context
still stands: namely, without some form of ethical reactive attitudes we could
not recognize or find intelligible a system of human relationships that
would qualify as a human ethical life. In other words, an ethical life devoid
of all forms of ethical reactive attitudes is not recognizable or intelligible to
us as a form of human ethical life.
It should be clear, in light of Strawson’s observations, that we do not
need to choose between naturalism and genealogy, where this is understood
in terms of sensitivity to historical and cultural variation and diversity. On
the broad construal, ethical norms and the ethical considerations to which
they give weight, may vary greatly from one culture and historical period to
another. With these variations we will also find variations in the particular
forms of ethical reactive attitudes that are adopted and endorsed. One form
these ethical reactive attitudes may take is the narrow form encouraged
by the morality system—which makes obligation and blame its central
features. While this form of ethical reactive attitude may be local and
contingent, it does not follow that ethical reactive attitudes broadly con-
strued are local and contingent, unless we take the local form to be the sole
legitimate representative form of moral responsibility. Since the broad
construal neither interprets ethical reactive attitudes nor moral responsi-
bility in these restrictive terms, it is able to acknowledge that the ancient
Greeks, among others, have ethical reactive attitudes that are recognizably
 Responsibility, Naturalism, and “The Morality System” 199

continuous with our own conception of moral responsibility. This can be

done without in any way denying that there are significant differences
between their culture and our own with regard to the attitudes and practices
involved (e.g. with respect to issues of voluntariness and intention). The
irony about this situation is that it is the narrow construal, which rejects
(type) naturalism, which cannot accommodate genealogical sensitivity to
cultural and historical variation. Given that the narrow construal insists
that moral responsibility be understood in terms of the concepts of the
morality system, and that this interpretation alone constitutes genuine or
real moral responsibility, it is compelled to exclude all other understandings
(i.e. as based on a broader construal of ethical reactive attitudes) as falling
outside the parameters of moral responsibility. It is, therefore, the narrow
construal that is insensitive to genealogical variation and diversity as it arises
within the framework of moral responsibility. For reasons that have been
explained, this is not simply a verbal issue, as it involves and encourages a
truncated and distorted understanding of moral responsibility and the way
in which it is naturally rooted in human ethical life.
It has been shown that “asymmetry” and “localism” are unnecessary and
unacceptable costs of the narrow approach. What, then, about the further
issue relating to type-naturalism and external rational justifications? If we
adopt the broad construal then no external rational justification of our
liability to ethical reactive attitudes is required, where ambitions of this
kind involve what Strawson describes as the tendency to “over-intellectual-
ize the facts” concerning the natural foundations of moral responsibility
(Strawson 1962: 23). In contrast with this, the narrow construal needs to
provide some relevant external rational justification since, per hypothesis, it
is a local (modern, Western) achievement. Clearly the broad view can
accept that the local forms of ethical reactive attitudes (e.g. as based on
the requirements of the morality system) may come and go. Considered
from this perspective, the morality system and its associated narrow
interpretation of moral responsibility may be judged vulnerable to
extinction for several related reasons. First, as already argued, it suggests
a truncated and distorted account of our own ethical concerns (i.e. even
from a modern, Western perspective). Second, it generates (unnecessary)
problems of asymmetry and localism as we have described them. Third, the
narrow account is also especially vulnerable to internal skeptical collapse
due to worries about determinism. (Although Wallace believes these
internal skeptical objections can be defeated, not all those who endorse
the morality system believe this can be done, much less done on the basis of
compatibilist commitments.) It follows from all this that there is a real
prospect of this local form of ethical reactive attitudes collapsing under both
internal and external skeptical pressure. What does not follow from this,
200 Paul Russell

however, on the broad construal, is the total collapse of moral responsibility

in any recognizable form. We may still have available other forms of ethical
reactive attitudes that are not similarly vulnerable in any of these
dimensions. Clearly it would be a mistake, therefore, to present the
collapse of the local form of moral responsibility associated with the
morality system as putting us in the predicament of having to adopt
some form of utilitarian, forward-looking approach that has no place for
ethical reactive attitudes of any kind. Alternative forms of reactive attitudes
remain available and viable within the structure of ethical life that is still
recognizably human and intelligible to us.9
The question we must now turn to is what is the relationship between
skepticism and type-naturalism as understood on the broad approach?
Skepticism may take the form of aiming to discredit local understandings
of our ethical reactive attitudes. This does not, as has been argued, show
that all forms of ethical reactive attitude are thereby discredited or that
moral responsibility, as such, cannot be vindicated. It remains true, never-
theless, that whatever local forms our ethical reactive attitudes may take,
they (all) remain vulnerable, in principle, to internal, global skeptical
challenge at the token level. That is to say, our commitment at the type
level, even on a broad construal, does not secure any general immunity
from potential global skepticism with respect to all tokens of our ethical
reactive attitudes.10 The crucial point that needs to be emphasized,
however, is that even in these circumstances, the skeptic remains
committed to ethical reactive attitudes at the type level (unless, of course,
the capacity to feel and experience ethical reactive attitudes is itself
damaged). In other words, although the skeptic may systematically
disengage from all tokens of ethical reactive attitude from “the inside,” she
cannot abandon the propensity or liability to these attitudes. That is a
project that, from one point of view, would require radical intervention
with her own nature (e.g. by way of genetic engineering or medical surgery)

9
Wallace refers to the utilitarian approach as “the economy of threats” model (1994:
54–61). It is crucial to his critique of this model that it lacks “depth,” where depth is
provided by the “attitudinal” features of blame and retribution (1994: 56, 75). On a
broader construal, however, “depth” can be found in other forms of ethical reactive
attitude, such as shame and anger—a point that Wallace comes close to endorsing in
some passages. See, e.g. 1994: 89.
10
This may well be regarded as highly unlikely or even incredible—but it is not
inconceivable. Imagine, for example, the spread of some terrible disease or genetic
mutation that affected us all by damaging our most basic and universal moral capacities
in such a manner that exemptions applied universally (however broadly interpreted).
 Responsibility, Naturalism, and “The Morality System” 201

and, from another point of view, would place her outside the recognizable
human ethical community.11
We have already noted that whereas the narrow construal would place
the ancient Greeks (and other shame cultures) outside the framework and
fabric of moral responsibility, the broad construal does not. What, then,
about the skeptic? In contrast with individuals who are engaged partici-
pants in forms of ethical life involving ethical reactive attitudes, the skeptic
has systematically disengaged from all such participation or involvement.
Disengagement of this kind requires (internal) doubts about the justifica-
tion for any proposed tokens of ethical reactive attitude. So described, the
skeptic cannot evade the challenge of providing some account of the
excusing and exempting considerations that apply universally in such a
manner that all tokens of ethical reactive attitude are discredited. If the
skeptic fails or refuses to provide any such rationale for her (disengaged)
stance then her skeptical stance has not been vindicated or justified.
Contrast the skeptic with another distinct character, who we may call the
“Vulcan.”12 Vulcans are understood to be entirely rational but incapable of
human emotion. As such, Vulcans may rationally understand (human)
ethical norms but are incapable of feeling or entertaining ethical reactive
attitudes (or similar moral emotions with an attitudinal aspect). Vulcans
have, in other words, no type-naturalist commitments with regard to
ethical reactive attitudes. It is, for this reason, a mistake to assimilate the
skeptic to a Vulcan, as plainly the skeptic is not a Vulcan. The Vulcan faces
no skeptical problem with respect to ethical reactive attitudes. They have no
token commitment because they have no type commitment to this range of
(ethical) emotion. For the (human) skeptic, however, the skeptical
challenge is real because their type propensities require something to be
said with respect to disengaging all tokens of these reactions to the ethical
qualities of others in their community. In this way, both the skeptic and
the anti-skeptic, in contrast with the Vulcan, can accept Strawson’s type-
naturalism and dispense with the search for external rational justifications.
What divides them is the issue of whether or not a theory of excuses and
exemptions can handle relevant internal skeptical worries (e.g. as based on
the implications of determinism).

11
An individual who lacks any type commitment to ethical reactive attitudes would
not be recognizably human, not because she is a systematic skeptic with respect to these
attitudes but because the skeptical issue does not arise for her with respect to these
attitudes, since she is constitutionally incapable of experiencing or entertaining such
attitudes.
12
Vulcans are aliens from the planet Vulcan, as described in Star Trek. The character
of Mr Spock was half Vulcan and half human. Wallace refers to this example in a related
context at Wallace 1994: 78 n.41. See also Russell 2011: esp. 212–14.
202 Paul Russell

5 . A G AI N ST T H E N A R R O W C O N S T R U A L
O F M O R AL R E S PO N S I B I LI T Y

It has been argued that the narrow construal of reactive attitudes and its
associated account of moral responsibility has unacceptable costs. While it is
true that there are significant failings in Strawson’s original naturalistic project
that need to be addressed and corrected (e.g. we should reject token-natural-
ism), we should retain the core feature of type-naturalism. In order to do this
we need to provide a broader account of ethical reactive attitudes that extends
beyond the constraints and limits of “the morality system” and its conceptual
structures. It is only by taking this route that the difficulties we have described
relating to “asymmetry” and “localism,” as well as the fruitless and misguided
search for external rational justifications, can be avoided. We may summarize
the significance of these observations in the following points.
(1) The narrow construal of moral responsibility, as developed on the basis
of the morality system, both distorts and truncates our understanding of
human ethical life as it relates to moral responsibility. In particular, it
makes it impossible to accommodate both positive ethical reactive attitudes
and alien ethical reactive attitudes as they may arise from outside our
(modern, Western) culture. Even our own local understanding of moral
responsibility is not fully or adequately captured by this narrow construal.
(2) It is the broad construal, along with its commitment to type-naturalism,
which is able to accommodate genealogical sensitivity to historical and
cultural variation in relation to our understanding of moral responsibility.
The narrow construal excludes all alternative forms that do not fall into the
constraints imposed by “the morality system” as mere analogues or proto-
types of moral responsibility. As such, the narrow construal constitutes a
form of conceptual imperialism with regard to (real, true) moral responsi-
bility and also commits us to an implausible “progressivism” concerning
our own (modern, Western) views. In contrast with this, the broad
approach recognizes the variation in modes and forms of ethical reactive
attitude within a wider understanding and appreciation of the emotional
fabric of moral responsibility.
(3) Type-naturalism, as understood on the broad construal, provides no
easy way of dealing with a potential internal skeptical challenge (i.e. in
contrast with the aims of token-naturalism). Even allowing for our natural
liability to ethical reactive attitudes, on a broad construal, we must still
formulate some relevant schema of excuses and exemptions. From this
perspective it is always conceivable that a systematic or global skepticism
 Responsibility, Naturalism, and “The Morality System” 203

could be generated from “the inside” (i.e. extending to all our token ethical
reactive attitudes). This possibility does not, however, license a search for
external rational justifications, since our liability or propensity to such
emotions is natural and not rationally grounded. The skeptic remains
committed to ethical reactive attitudes at this level, even if she has entirely
abandoned or disengaged any commitment to tokens of these attitudes (in
light of internal skeptical pressures of some kind).
(4) Much of the motivation behind Wallace’s narrow construal of the
reactive attitudes is to find a satisfactory compatibilist account of moral
responsibility consistent with the core requirements and constraints of the
morality system. From this perspective the internal skeptical challenge is
especially acute, since it is targeted on the notions of wrongness, blame,
desert, and retribution that are central to moral responsibility as the
morality system interprets it. It is evident, however, that the broad con-
strual of ethical reactive attitudes, along the lines that has been sketched,
significantly deflates these (internal) skeptical pressures. The reason for this
is that a broader and more liberal conception of ethical reactive attitudes
does not place such heavy weight or emphasis on the very elements of the
morality system that have proved especially vulnerable to skeptical criticism
(i.e. desert, blame, etc., along with their apparent dependence on ultimate
or absolute agency). Even if—contrary to what Wallace argues—it proves
impossible to vindicate this local interpretation of moral responsibility, as
understood on the narrow construal, it does not follow, given a broad
interpretation of ethical reactive attitudes, that global skepticism results. All
that follows from the success of the skeptical challenge, so described, is that
the local understanding of moral responsibility encouraged by the morality
system cannot survive critical reflection.13

13
It is true, of course, that many skeptics about moral responsibility are concerned to
discredit the local conceptions of moral responsibility associated with the morality
system. However, for reasons that have been discussed, skepticism of this kind does
not in itself constitute global skepticism—since it does not discredit, and may not even
aim to discredit, alternative forms of ethical reactive attitudes. Having said this, it is
important to note that many skeptical projects of this kind either explicitly or tacitly
endorse the narrow construal and its assumption that alternative accounts of ethical
reactive attitudes somehow fail the standard of real or genuine forms of moral responsi-
bility. When this assumption is made, the critique of our local conception of moral
responsibility framed in terms of the requirements of the morality system is (mistakenly)
inflated into a form of global skepticism about moral responsibility. Suffice it to say that
much of the contemporary free-will debate, along with its associated worries about the
skeptical threat, proceeds on this assumption of the narrow construal and the morality
system.
204 Paul Russell

In sum, we may contrast the relative strengths and weaknesses of the broad
and narrow accounts in these terms. The narrow construal not only
generates a partial and incomplete account of moral responsibility, it also
leaves the entire edifice of moral responsibility, so understood, vulnerable
to both internal and external skeptical threat. The broad construal not only
avoids the significant difficulties that the narrow construal encounters (e.g.
asymmetry), it provides for the complexity, variation and nuance that we
find in this sphere. Moreover, the broad construal, by moving away from
the rigidities and (peculiar) demands of the morality system, deflates the
internal skeptical threat and eliminates all worries relating to the misguided
ambition of providing a satisfactory external rational justification. These
are fundamental points relating to moral responsibility and the defects of
the morality system that the discussions of both Strawson and Williams
converge on.

REFERENCES

Doris, John (2002). Lack of Character: Personality and Moral Behavior. (Cambridge &
New York: Cambridge University Press).
Russell, Paul (1992). “Strawson’s Way of Naturalizing Responsibility.” Ethics 102. 2.
—— (2004). “Responsibility and the Condition of Moral Sense.” Philosophical
Topics, 32, 1; 2, 287–305.
—— (2011). “Moral Sense and the Foundations of Responsibility.” In Free Will,
2nd edn., ed. Robert Kane (New York: Oxford University Press).
Strawson, P. F. (1962). “Freedom and Resentment,” reprinted in P. F. Strawson,
Freedom and Resentment and other essays. (London, New York and Oxford:
Methuen. 1974).
Wallace, R. Jay (1994). Responsibility and the Moral Sentiments. (Cambridge, MA:
Harvard University Press).
Williams, Bernard (1985). Ethics and the Limits of Philosophy. (London: Fontana).
—— (1993). Shame and Necessity. (Berkeley, CA: University of California Press.).
 8
The Three-Fold Significance
of the Blaming Emotions
Zac Cogley

1. INTRODUCTION

Many philosophers working on moral responsibility follow P. F. Strawson

(1982) in understanding claims about someone’s moral responsibility or
the phenomenon of holding people morally responsible in terms of the
appropriateness of a certain class of emotions (Bennett 1980; Watson 1993;
Wallace 1994; Fischer and Ravizza 1998; McKenna 1998; Macnamara
2009). But even those who would not follow Strawson in identifying moral
responsibility attributions with the appropriateness of emotions hold that
emotions do play a role in our moral responsibility practices (Scanlon
2008, 143). In spite of this, the significance of the blaming emotions for
moral responsibility has been under-theorized. (I am concerned here with
people’s moral responsibility for their actions or omissions, rather than, for
example, whether in general someone is a morally responsible person.)
In order to fully appreciate the import of the blaming emotions for moral
responsibility we need a more adequate moral psychology. As an initial step,
in this paper I appeal to recent work in psychology of emotion to argue that
the blaming emotions—anger, resentment, and indignation—are significant
for our moral responsibility practices in three different ways.1 They are
important to moral responsibility in appraising people as acting wrongfully,
in communicating the appraisal to perceived wrongdoers, and in sanctioning
people who are appraised as wrongful.2 I also investigate the conditions of
appropriateness of the blaming emotions. My methodology is inspired by

1 While there are positive emotions that are connected to moral responsibility, I focus
on the blaming emotions as they have received much more philosophical and psycho-
logical discussion than have candidate positive emotions like gratitude.
2 An anonymous reviewer points out that these also correspond to three broad
categories of response to wrongdoing. I agree—in fact, I think we categorize responses
206 Zac Cogley

recent philosophical attention to reasons for attitudes: for example, the

reasons in favor of believing a proposition (Shah 2003) or blaming another
person (Hieronymi 2004). There has also been some attention—though not
nearly as much—to the reasons that bear on emotions (D’Arms and Jacobson
2000). As I will demonstrate, the three ways in which the blaming emotions
are significant for our moral responsibility practices are associated with very
different kinds of appropriateness considerations.
My work is also inspired by the fact that although there has been
significant recent attention to the concept of moral responsibility, there is
little agreement about it. Indeed, in one recent attempt to clear the concep-
tual territory, John Martin Fischer and Neal Tognazzini argue that there are
up to thirteen different analytical or conceptual “stages” of moral responsi-
bility attributions, organized (roughly) into two broad categories: attribut-
ibility and accountability (2010). Here they are inspired by Gary Watson’s
(1996) distinction between these two concepts, but urge that conceptual
clarity about moral responsibility requires far more distinctions.3
I am deeply sympathetic to the project of achieving clarity about our
conception of moral responsibility as it is central to making progress on
some of our most vexing issues about moral responsibility, including
whether moral responsibility is compatible with determinism. However,
I fear that some recent attempts to introduce clarity risk further confusion
because they have not paid sufficient attention to the moral psychology
of the blaming emotions. Not only, then, do I try to enrich our moral
psychological picture of the blaming emotions, but I also link appraisal,
communication, and sanction to representative accounts of moral responsi-
bility. I suggest that each kind of account is inspired by a different way in
which the blaming emotions are significant, and thus each account impli-
citly emphasizes a different consideration of emotional appropriateness.
Fittingness accounts of moral responsibility are linked to appraisal, moral
address accounts correspond to the communicative dimension of the
blaming emotions, and desert accounts of moral responsibility are inspired
by the blaming emotions’ sanctioning role. If I am right, part of the reason
debates about moral responsibility have been so intractable is that many
theorists share the assumption that appropriate blaming emotions are a
reliable indicator of a person’s moral responsibility, while inappropriate
blaming emotions are evidence of a lack of moral responsibility. This makes
it appear as if all parties to the debate are operating with the same

to wrongdoing as appraisals, communications, and sanctions in virtue of their connec-

tion with the blaming emotions. Space precludes making that argument here.
3 Fischer and Tognazzini’s analysis places consideration of the blaming emotions
squarely into the accountability category. My analysis here complicates that categorization.
 Significance of the Blaming Emotions 207

conception of moral responsibility in mind. However, because different

accounts are implicitly linked to different kinds of appropriateness, the
wide agreement that the appropriateness of the blaming emotions is
revealing of moral responsibility obscures significant disagreements about
the concept and the conditions for its application that emerge with a more
refined focus.4
While discussion of all of the blaming emotions is common, theorists
often emphasize one or two to the exclusion of others. For example,
R. J. Wallace speaks of indignation and resentment (Wallace 1994) as
does Tamler Sommers (2007), while Derk Pereboom has remarked that
“of all the attitudes associated with moral responsibility, it is anger that
seems most closely connected with it” (Pereboom 2001, 208).5 In what
follows, I assume that from a psychological standpoint, resentment and
indignation are ways of being angry.

2 . A P PR AI SA L

An important strand of contemporary psychological research on emotion

seeks to determine characteristic appraisals that are assumed to elicit distinct-
ive emotions.6 “Appraisal” refers to a person’s evaluation or interpretation
of a situation. According to this research, different emotions are caused
by distinct appraisals. For example, Richard Lazarus claims that anger is
produced by a person’s appraisal of a “personal slight or demeaning offense”
(1991, 223), while in a later collaboration with Craig Smith (1993), both
believe that anger is caused by an appraisal of “other-blame,” which they
claim can be broken into three separate components: motivational relevance
(the situation is personally relevant), motivational incongruence (the
situation is inconsistent with what is desired), and other-accountability (the
emotion is directed at someone else). Philosophers have roughly concurred.

4 In his recent paper, “Attributability, Answerability, and Accountability: Toward a

Wider Theory of Moral Responsibility” David Shoemaker offers a similar argument that
distinguishes among moral responsibility concepts, though without a focus on the
blaming emotions (2011).
5 Most commonly, it appears that these different terms mark a difference in whether
the object of the emotion is second or third-personal. For example, see (P. F. Strawson
1982; Wallace 1994; Sommers 2007; Pereboom 2009).
6 While there has been much debate over whether or not the relevant appraisals are
cognitive, beginning with (Zajonc 1980; Lazarus 1982) and continued in (Zajonc 1984;
Lazarus 1984), that debate is orthogonal to my concerns. For an excellent recent
discussion of this issue, see (Prinz 2004, 21–51).
208 Zac Cogley

For example, Jesse Prinz and Shaun Nichols claim that “Anger arises when
people violate autonomy norms, which are norms prohibiting harms against
persons” (2010, 122). If we make the plausible assumption that slights and
offenses both involve the violation of norms, we can see all these authors
offering a roughly similar account of the appraisal involved with the blaming
emotions, though they do disagree about how best to capture it.
While I agree that the blaming emotions have a characteristic appraisal,
the above accounts make two errors regarding it. First, these accounts fail to
pinpoint the characteristic appraisal of the blaming emotions. The early
Lazarus, as well as Prinz and Nichols, construe the appraisal too narrowly.
For example, blaming emotions are commonly elicited by harms against
nonhuman animals, violations of religious commandments, the nonharm-
ful breaking of promises and many other situations that go beyond slights
and harms against persons.7 On the other hand, the account from Lazarus
and Smith is too broad; adding up their three appraisal components (an
emotion directed toward a personally relevant situation that is inconsistent
with what is desired) does not give us the characteristic appraisal of the
blaming emotions. Such an appraisal is also compatible with sadness. We do
better if we follow James Averill, who argues that “the typical instigation to
anger is a value judgment. More than anything else, anger is an attribution
of blame” (1983, 1150) or Shaver et al., who hold that the eliciting
appraisal is that “the situation is illegitimate, wrong, unfair, contrary to
what ought to be” (1987, 1078).
I propose, then, that the way a person feeling a blaming emotion
appraises her situation is best captured as an appraisal of wrongful conduct.
This is the core appraisal of the blaming emotions, but we can break it into
constituent parts as follows:
If a person, A, feels a blaming emotion, she evaluates her situation as containing:
(i) a person, B,8 whose
(ii) action or omission
(iii) transgresses a norm on proper conduct (including, but not limited to,
moral norms, though the norm need not be codifiable by a rule)
(iv) because B is motivated by ill will or has shown insufficient concern,
(v) and A glosses B’s action as bad.9

7 Surprisingly, Prinz and Nichols themselves note the connection between blaming
emotions and harms against nonhuman animals (2010, 130).
8 In some situations A and B will be the same person.
9 I have in mind here the fact that anger has a distinctive unpleasant phenomenology
that might be glossed as “feeling ready to explode” (Roseman, Wiest, and Swartz 1994).
 Significance of the Blaming Emotions 209

This treatment of the appraisal dimension of the blaming emotions handles

the fact that we often feel the blaming emotions in response to violations
that don’t harm persons, as well as the variety of situations where we feel
blaming emotions because a person’s action violates an autonomy norm, or
is a demeaning offense or personal slight.
This is an improvement, but there is another error in the above treat-
ments of the appraisal involved in the blaming emotions. All of the above
treatments construe the relation between the appraisal and the blaming
emotions as a causal relation. That is, the appraisal is what brings about the
blaming emotion. This is the second mistake in the literature about the
relation between the blaming emotions and their characteristic appraisal.
Not all psychologists believe that appraisals always precede blaming emo-
tions or are necessary for them; indeed there is not clear evidence that
appraisals always cause episodes of the blaming emotions, though there is
no question they often do (Berkowitz and Harmon-Jones 2004a; Berkowitz
and Harmon-Jones 2004b; Parkinson 1999).
I think we better understand the psychology of a person feeling a
blaming emotion if we hold that the blaming emotions need not be caused
by their characteristic appraisal (though they often are). However they are
caused, the blaming emotions are an appraisal of conduct as wrongful.
Consider, by analogy, a particular belief: my belief that it is sunny outside.
While my belief that it is sunny outside might be caused by present sun
outdoors (if I was just outside and noticed the weather), that belief might be
caused in a number of other ways. I might come to believe it is sunny
outside based on your testimony or by inferring today’s weather based on
what the weather was yesterday. In these cases my belief that it is sunny
outside isn’t caused by occurrent sun. A similar point applies to the
blaming emotions. While in many cases they are caused by their character-
istic appraisal, their link to appraisal is better understood as conceptual
(Parkinson 1997).
This analogy between beliefs and the blaming emotions is also relevant
because it relates to our practice of taking our blaming emotions to be
appropriate or inappropriate, depending on their aptness for the situation.
The blaming emotions are not unique in this respect. As Justin D’Arms and
Dan Jacobson have pointed out, we commonly argue about whether or not
things are sad, enviable, shameful, or worthy of pride or resentment. Our
practice of considering these issues of emotional appropriateness presup-
poses that we can make sense of whether or not an emotion’s characteristic
appraisal is accurate, or to use their terminology, fitting (2000). When a
blaming emotion is fitting, it accurately presents its object as having the
features contained by its appraisal; the fittingness of a blaming emotion is
analogous to the epistemic relation that obtains between the world and a
210 Zac Cogley

true belief. Anger, resentment, and indignation are fitting to feel when, for
example, someone intentionally wrongs you out of ill will.
Thus, the blaming emotions are fitting when they are felt in response to a
person who satisfies conditions (i)-(v), above. The lack of any one of the
five conditions means that a blaming emotion is unfitting.10 We can also
distinguish between “degrees” of fit between a blaming emotion and the
situation it appraises. I should be angrier with someone who tries to ruin
my career than a neighbor who thoughtlessly mows his lawn at 8 a.m. on a
Sunday morning. And you should be more upset with the driver who
intentionally tries to run you over while you are out for a walk than you
should be with a person who somewhat carelessly backs his car into your
path. Thus, the seriousness of the wrong in question and the person’s
relation to the wrong both help to determine the amount of anger fitting
for the situation.

3. COMMUNICATION

There is much psychological evidence to suggest that the blaming emotions

not only appraise the conduct of others, but also play a role in communi-
cation. Specific speech patterns (including rate of articulation, intensity,
and frequency of vocal fold vibrations) appear to be associated with
different emotions, particularly anger (Scherer 1986; Scherer et al. 1991).
Psychologists have also found that different bodily movements and
postures are associated with different emotions (Wallbott 1998). Perhaps
most probatively, the blaming emotions, just like many other emotions, are
associated with characteristic facial expressions (Ekman 1999). Relevantly,
while people commonly interpret the emotional facial expressions of others
as signifying a person’s appraisal of her situation, anger expressions are
more likely than the expression of other emotions to be interpreted as
conveying intentions or requests (Horstmann 2003). Also notably, the
characteristic facial expressions of different emotions appear to be highly
associated with interpersonal interaction. For example, winners on the
medal stand at the Olympic games are more likely to smile during
interactions with other people than during the rest of the ceremony

10 By distinguishing between the five conditions on the fittingness of a blaming

emotion, I call our attention to conceptual distinctions. However, I allow that these
different aspects of a blaming emotion’s appraisal may often, or even always, affect each
other in interesting ways. For example, it may be that someone’s act motivated by ill
will—even if she does something that I know has no chance of actually harming
anyone—itself transgresses a norm on proper conduct and is therefore bad.
 Significance of the Blaming Emotions 211

(Fernández-Dols and Ruiz-Belda 1995) and bowlers are less likely to smile
when they first roll a strike than when they turn to face others watching at
the end of the alley (Kraut and Johnston 1979).
In a typical interpersonal episode of a blaming emotion, various bodily,
vocal, and facial responses communicate that the person feeling the emo-
tion is angry and thereby give the person who is the target of the blaming
emotion information about the way her conduct is being appraised. In
many situations, this information is not contained in spoken words but is
transmitted instead by the overall emotional demeanor of the person feeling
the blaming emotion. These communicative aspects of a person’s emo-
tional demeanor are observed, responded to, or ignored by others. The
responses of others—or the lack of a response—are then an opportunity for
continued emotional engagement and transformation. Thus, in most inter-
personal interactions, a person who feels a blaming emotion not only
appraises the conduct of another as wrongful, she also communicates to
the target of the blaming emotion that she construes his behavior as
wrongful.11
In human psychological response, these communicative aspects are very
closely connected to the having of the blaming emotion itself. It turns out to
be almost impossible not to register your emotional state on your face in
some way, even if briefly, and it is hardest to mask negative emotions (Porter
and Brinke 2008). Since others are extremely attentive to such displays, it is
often better to try not to have the emotion than to allow the emotion to run
its course while attempting to mask what you feel. Thus, I want to suggest
that the communicative significance of the blaming emotions can sometimes
result in their being inappropriate to feel even when they fittingly appraise a
target. That is, there are additional considerations that bear on whether to
have a blaming emotion than merely considerations of fit. In characterizing
communicative considerations that bear on emotional appropriateness, I am
inclined to follow Angela Smith in distinguishing between considerations of
standing, the degree of fault displayed in the wrongful action, and the
response that the blamed person takes (or will, or could take) to the person
who feels the blaming emotion.12
Whether or not someone has proper standing, or authority, to feel what
would be a fitting blaming emotion toward a wrongdoer has much to do

11 To be clear, I am not arguing that the communication of this information is

intentional on the part of the person feeling the blaming emotion, only that the infor-
mation is “there for the taking.” Along similar lines, the signals that indicate which play
is being called by a team may transmit the same information to the opposing team if the
opposing team has attended to which signals are reliable signs of particular plays.
12 My discussion of these three issues is indebted to (A. M. Smith 2007, 478–83).
212 Zac Cogley

with her relationship to the wrongdoer and those who are wronged. So, for
example, even though I may observe what I regard as condescending and
obnoxious behavior from one of two parties having a public argument,
I may be inappropriately angry with the offending party if I have no social
connection to either one of them. We can see this more clearly when we
consider that it would not be uncommon for the victim of the obnoxious
behavior to become angry with me when my indignation on his behalf
becomes known.
My own fault and hypocrisy can also affect my standing to react to
another’s conduct with a blaming emotion. Thus, if my awareness of its
health effects has not moderated my long-time smoking habit, my friend
with a drinking problem will regard as out-of-line my fitting anger at him
for neglecting his health when he succumbs to the temptation to drink. My
friend’s failing, just as my own, may be a legitimately moral one, but the
fact that I am unable or unwilling to similarly guide my own behavior
makes it inappropriate to feel angry toward him even though anger is fitting
for what he does.
Beside the fact that the communicative appropriateness of the blaming
emotions can be undermined by my standing to have such responses, it can
also be affected by the relative significance of the fault a person displays in her
conduct. Thus, while a blaming emotion would fittingly appraise a student
who fails to keep his scheduled appointment with me, the degree of fault
(simple forgetfulness) and the degree of harm to me (a very mild inconveni-
ence) prohibit me from feeling any resentment toward him. We can see this
interact with another communicatively salient factor—the agent’s own
response—if we suppose that the student rushes over to my office, apologiz-
ing profusely even as he walks in the door. The student’s self-reproach
indicates that he understands he did wrong and is committed to doing
what he can to prevent it from occurring in the future. While being indignant
would fittingly appraise his faulty conduct, I should not feel indignant
toward him because it would be communicatively inappropriate for me to
target him with a blaming emotion given his indication that he understands
his error and the importance of keeping appointments.13 (Of course, things
might be different if this same student has routinely missed appointments
even while protesting that it will never happen again.)
In such a case, we again have a communicative reason against feeling a
blaming emotion, even though the blaming emotions fittingly appraise the
actions of the person in the situation.

13 For an analysis of the communicative dimension of emotions supporting this

claim, see (Macnamara forthcoming).
 Significance of the Blaming Emotions 213

4. SANCTION

Another way in which the blaming emotions are significant for our moral
responsibility practices relates to communication but is ultimately distinct,
namely, affecting the behavior of others by imposing costs.14 While this is
sometimes accomplished via the communication of a message, at other
times it is accomplished simply through changing the costs and benefits of
another person’s possible actions. Though the relationship between the
blaming emotions, deliberation, and action-aimed-at-sanction is complex,
in this section I will highlight some of the relevant psychological findings to
demonstrate the connection of the blaming emotions to sanction.
Anger has important effects on deliberation and social perception that play
a role in determining the behavior of someone feeling a blaming emotion
(though the effects vary across individuals, depending on their level of
awareness and cognitive skills). Angry people tend to have a sense of signifi-
cant control (Lerner and Keltner 2000) that leads them to be optimistic about
the success of their probable actions (Lerner and Keltner 2001). They are also
“eager to make decisions and are unlikely to stop and ponder or carefully
analyze” (Lerner and Tiedens 2006, 132), which likely leads them to take
actions that have a low probability of succeeding but high payoffs (Leith and
Baumeister 1996). When they act, angry people tend to be more punitive
toward those they blame (Lerner, Goldberg, and Tetlock 1998).15
We can see these deliberative effects demonstrated in experimental work
on altruistic punishment and ultimatum games. In altruistic punishment,
the punisher receives no material benefit but imposes a cost on the party
punished. Thus, people are willing to punish free-riders even when it is
costly for them to do so and they cannot expect future benefits from
punishing (Fehr and Gächter 2000). Jonathan Haidt has argued that a
paradigm feature of human morality is this third-party enforcement of
moral norms (2001). In one significant study, Ernst Fehr and Simon
Gächter (2002) demonstrated that free-riding on a common good is less
prevalent when altruistic punishment of free-riders is possible. When such
punishment occurs, it is reported by punishers to express their anger and
those who are punished perceive their punishers as angry.16 A similar

14 I am not arguing that such costs are always intended by the person feeling the
blaming emotion, though they certainly sometimes are.
15 For an excellent overview of recent empirical study of anger’s effects on judgment
and decision-making, see (Litvak et al. 2010).
16 Notably, angry sanctions lead to positive behavior change even when free-riders
interact with a new group of people that does not include their previous punishers.
214 Zac Cogley

finding concerns experimental work on ultimatum games. In such two-

person games, one party (suppose it’s me) controls some resources (say,
$10) and makes an offer to another to split the resources in a particular
fashion with another party ($8 for me, $2 for you). You have the opportunity
to accept or reject the offer. Although game theory would predict that the
splits offered should heavily favor the person controlling the resources and
that all offers should be accepted, people tend to offer more than 40 percent
of the resources and 15 to 20 percent of offers are rejected (Ochs and Roth
1989). The most perspicuous explanation of this behavior is that people
expect the splits to be fair; if they are not, the split is angrily rejected even
though the rejection leaves the rejector worse off than had she accepted
(Pillutla and Murnighan 1996). Because ‘offerers’ anticipate the possibility
of sanctioning reactions, they tend to offer more equitable splits.
I’ve been using sanction to specifically demarcate the cost-imposing
function of the blaming emotions from the communicative aspect. As
I argued above, one function of the blaming emotions is to communicate
appraisals to others. While one way to bring about a change in another
person’s behavior is to successfully communicate an appraisal to them, the
communication will ultimately only be successful if the other is willing to
appraise herself as acting wrongfully and see her wrongness as a reason for
change, apology, or restitution. However, even if the other person is
unwilling or unable to see herself as acting wrongfully, placing a cost on
particular ways she might behave can impact her chosen course of action.
The threat of sanction can lead someone to refrain from a wrongful action
not yet performed, not repeat a wrong he already performed, or not copy
the successful wrongdoing of others. Importantly, I also assume that
expressions of the blaming emotions, themselves, are experienced as sanc-
tions by the emotion’s target. Not only is it unpleasant in its own right to be
the target of another person’s blaming emotion, but Baumeister et al.
suggest that one function of the blaming emotions may be to stimulate
guilt in the person who is the target of the emotion (2007, 189). Thus,
people tend to avoid actions that they know would lead to being the target
of the blaming emotions of others in order to avoid psychological and
physical sanctions.17
There are a number of considerations that bear on the appropriateness of
the blaming emotions qua sanction, some of which we have touched on

17 Again, note that while the sanctioning effects of the blaming emotions are at least
sometimes directly intended by people who feel a blaming emotion (Fehr and Gächter
2000), they need not always be. However, even if people feeling blaming emotions do
not intend that their emotions be experienced as harms by the targets of the blaming
emotion, that doesn’t mean they are not experienced as such by the targets.
 Significance of the Blaming Emotions 215

already. Again, the lesson is that the question of whether a blaming emotion
is appropriate, all things considered, is not solely determined by whether or
not the blaming emotion fittingly appraises someone’s conduct. For
example, the propriety of sanctioning someone with a blaming emotion
can be affected by the seriousness of the wrong done. Just as with commu-
nicative appropriateness, some extremely minor wrongs ought not be
responded to, while the propriety of sanctioning a wrongdoer may increase
with the seriousness of the wrong.18 The person’s own repentance, or lack
thereof, is relevant because sanctioning an already repentant person may be
unnecessary to affect his future conduct.
As each of us has only limited motivational and actional resources, the
appropriateness of a blaming emotion qua sanction can also be affected by
what other wrongs you might plausibly respond to. For example, you do
better to get angry with the perpetrators of wrongs when you might
successfully undo the wrong or positively influence the perpetrator. If,
hypothetically, you became aware of two different wrongs committed by
two different people that were approximately as severe, your blaming
emotions would be better directed toward a wrongdoer who would be
more swayed by the communicative and motivational significance of your
blaming emotions. Thus, there is something to the phenomenon where
people are more likely to feel a blaming emotion in response to a wrong
that directly affects them or someone they know well, rather than a wrong
that affects persons with whom they have little contact. Other things equal,
your motivational resources are more likely to lead to beneficial outcomes if
you address concerns with which you can profitably engage.19
A related, but distinct, issue concerns what I will term the fairness of
sanctioning a wrongdoer with a blaming emotion. We can see this notion
displayed first by returning to the phenomenon of hypocrisy, earlier raised
in reference to the communicative appropriateness of a blaming emotion. If
you habitually commit a certain type of wrong, your right to sanction
others with the blaming emotions for similar wrongs will be called into
question—especially if you protest the sanction of the blaming emotions
when it is applied to you. Similarly, if two people jointly undertake to
commit a wrong (say, robbing someone’s home) but one of them plays

18 I say “may” increase because I am doubtful that the relative deservingness of a

wrongdoer, itself, is a sufficient reason to license a sanction for her. Explaining why
would require that I develop a theory of desert and deservingness, which is a topic for
elsewhere. For reasons why I am skeptical that desert can be a sufficient justification for
sanctions, see (Dolinko 1991a; Dolinko 1991b).
19 Technological advances that give you knowledge of wrongs done to little-known
people far away don’t contradict this point, though they do complicate it considerably.
216 Zac Cogley

more of a role in planning and executing the deed than the other, it is
appropriate to sanction the “mastermind” to a greater degree with a
blaming emotion. Thus, if you must choose where to direct your blaming
emotions, fairness considerations speak to you blaming the mastermind
more than the accomplice. I believe the notion of the unfairness of blaming
emotions qua sanction also accounts for our appropriate reluctance to
blame the victims of wrongs or injustices, even if the victims of such wrongs
are complicit in, or partially responsible for, the wrongdoing. In such a
situation, targeting the person who was wronged with a blaming emotion
amounts to piling another bad thing on top of whatever misfortune the
person has already suffered.

5. LINKING ACCOUNTS OF MORAL RESPONSIBILITY

TO THE THREE FUNCTIONS

To this point, I have argued that the blaming emotions are significant for
our moral responsibility practices as appraisals, communications, and
sanctions. I have also argued that the appropriateness of a blaming emotion
in one sense does not guarantee that the blaming emotion is appropriate in
others. I have particularly focused on situations where a blaming emotion
fittingly appraises another’s conduct but is at the same time communi-
catively inappropriate or is inappropriate as a sanction. I do not take myself
to have exhausted all possible appropriateness considerations that bear on
these three functions, though I do hope to have captured many of the most
interesting and relevant appropriateness considerations for our practices of
moral responsibility. I now want to discuss several prominent accounts of
moral responsibility to suggest that differing accounts of moral responsi-
bility are motivated by attention to different ways in which the blaming
emotions are significant for moral responsibility.20
The notion that the blaming emotions involve appraisals of conduct as
wrongful is implicit in a number of theories of moral responsibility. For
example, John Martin Fischer and Mark Ravizza hold that someone is
morally responsible for her conduct to the extent that she is an “appropriate
candidate for at least some of the reactive attitudes on the basis of that
behavior” (1998, 6). Fischer and Ravizza admit that “in some contexts it

20 While this project seeks to locate the source of agreements and disputes about the
concept of moral responsibility, my own view is that considerations of fit are the only
appropriateness conditions of the blaming emotions that bear on a person’s moral
responsibility because these considerations circumscribe the concept of blameworthiness.
Unfortunately, I don’t have space to make that case here.
 Significance of the Blaming Emotions 217

may not be justified or appropriate, all things considered, actually to have

any reactive attitude to a particular agent” who is nonetheless morally
responsible for her conduct (1998, 7).21 Thus, their theory of moral
responsibility needs a sense of appropriateness for the blaming emotions
that can be apt, even if feeling a blaming emotion is not, all things
considered, appropriate. I believe that the fittingness of a blaming
emotion’s appraisal is the notion they are searching for.
Fittingness also allows us to make sense of R. J. Wallace’s idea that
resentment, indignation, and anger share a distinct propositional object,
namely, that “an expectation to which one holds a person has been
breached” (1994, 12). The blaming emotions, Wallace believes, are
fundamental to understanding the nature of moral responsibility. Coleen
Macnamara has also recently noted the connection between the blaming
emotions, appraisal and moral responsibility. When a person resents
her brother for not helping her move as he had promised, Macnamara
notes, her resentment is “a particularly deep form of moral appraisal” that
responds to the meaning of the brother’s insensitive action (2009, 89).
This sense of appraisal is, on her view, one face of holding others morally
responsible. Angela Smith has also recently urged that the fundamental
question of responsibility is whether an action can be attributed to a
person “in a way that makes moral appraisal, in principle, appropriate”
(2007, 470) and that negative moral appraisal, in terms of a judgment
of culpability, is entailed by feeling a blaming emotion like anger,
resentment, or indignation toward someone on the basis of her conduct
(2007, 467).
These views about the nature of moral responsibility are all unified in
taking the fittingness of the blaming emotions to be connected to ascrip-
tions of moral responsibility. Some, like Smith, hold that the question of
whether a person is morally responsible for her conduct just is the question
of whether an appraisal like that of the blaming emotions is fitting for
someone on the basis of her conduct.22 Others, like Macnamara, hold that
this is one significant aspect of our practice of holding others responsible,
but that it is not the only one. Macnamara argues that another important
“face” of moral responsibility is found in communicative acts (2009, 90).
Several theorists beside Macnamara have developed accounts of moral
responsibility that exploit the communicative function of the blaming
emotions. Thus, Gary Watson urges “the negative reactive attitudes express

21 On their terminology, the blaming emotions are clearly reactive attitudes.

22 Thus, as I read Smith’s account, the question of someone’s moral responsibility is a
question of the fittingness of certain appraisals, but is not a question of the fittingness of
emotions.
218 Zac Cogley

a moral demand, a demand for reasonable regard . . . the reactive attitudes

are incipiently forms of communication” (1993, 264). Taking up and
extending Watson’s idea, Michael McKenna claims that
holding another morally responsible for doing morally wrong is a manner of
communicating with her. In particular, it is a manner of responding to what she
had done as on analogy with a conversation in which the blameworthy person’s
conduct has a significance . . . for her to be blameworthy is for her to be a fitting
target of this manner of response, for her to be one with whom the relevant sort of
communication is called for. (2004, 187–8)
Stephen Darwall concurs that the blaming emotions are communicative.
He appeals to the idea that when I feel a blaming emotion in response to
you stepping on—and then continuing to stand on—my foot, my feeling
the blaming emotion is in part a demand that you remove your foot from
mine (Darwall 2006, 17). Even theorists like Angela Smith, who do not
regard the blaming emotions as fundamental to ascriptions of moral
responsibility, hold that the appraisal of another’s conduct as wrongful
has a communicative dimension. She writes, “Moral criticism, by its very
nature, seems to address a demand to its target. It calls upon the agent to
explain or justify her rational activity in some area, and to acknowledge
fault if such a justification cannot be provided” (2006, 381).
Finally, some other accounts of moral responsibility bring the sanction-
ing function to the fore. So, for example, Galen Strawson holds that in
asking whether people are morally responsible, we are asking if they are
responsible for their actions in such a way that they are, without any sort of
qualification, morally deserving of praise or blame or punishment or reward for
them. (2002, 441)
Note how Strawson thinks the question of moral responsibility just is the
question of whether a person is deserving of sanctions like punishment, and
that he must have the sanctioning function of blame (as an attitude) in
mind if he is to think that whether someone deserves blame or punishment
amounts to the same question.
Echoing Strawson’s claim that the question of moral responsibility is a
question of deservingness “without qualification,” Derk Pereboom claims
that
For an agent to be morally responsible for an action is for it to belong to her in such
a way that she would deserve blame if she understood that it was morally wrong, and
she would deserve credit or perhaps praise if she understood that it was morally
exemplary. The desert at issue here is basic in the sense that the agent, to be morally
responsible, would deserve the blame or credit just because she has performed the
 Significance of the Blaming Emotions 219

action (given that she understands its moral status), and not by virtue of conse-
quentialist considerations. (2007, 86).23
Also placing the sanctioning function at the fore of their analyses are
Tamler Sommers and Neil Levy who, like Galen Strawson, are skeptics
about desert-entailing moral responsibility.24 Sommers claims that “we feel
resentment when we feel that people have wronged us, and that they
deserve blame (and perhaps punishment) for what they did” (2007, 327).
Levy agrees that moral responsibility has a link to desert, but that the
connection between moral responsibility and deserved sanction is less
direct than Strawson and Sommers believe. On Levy’s view, the thought
that someone is morally responsible for wrongful acts she performs
amounts to the claim that such a person no longer deserves the full
protection of a right they would otherwise be entitled to: “a right against
having their interests discounted in consequentialist calculations” (2011, 3).
To have one’s interests discounted in such calculations is to incur a cost on
acting wrongfully: a sanction.

6 . C O N C LU S I O N

In this paper, I’ve argued that the blaming emotions relate to our practices
of holding people morally responsible in three different ways: appraisal,
communication, and sanction. I’ve shown that these ways in which the
blaming emotions are significant for our moral responsibility practices are
themselves associated with distinct considerations of appropriateness (fit-
tingness, communicative appropriateness, and appropriateness qua sanc-
tion) and that these different considerations can come apart from one
another. A blaming emotion can be fitting, for example, but inappropriate
qua communicative considerations or fitting yet inappropriate as a sanc-
tion. Finally, I’ve suggested that the different functions of the blaming
emotions and their characteristic conditions of appropriateness are quite

23 Pereboom has been consistently advocating this account of the conditions of

appropriateness for moral responsibility since the publication of Living Without Free
Will (2001). I interpret Pereboom as being concerned with a sanction-based understand-
ing of blame due to his emphasis on the potentially harmful effects of anger (Pereboom
2011).
24 This “accountability” face of moral responsibility has also been emphasized by
Gary Watson (1996). Watson also characterizes another, aretaic, face of responsibility
that involves beliefs or judgments about where someone’s conduct falls against some
standard. His discussion bears some similarity to my account of the appraisal function of
the blaming emotions. However, I am uncertain whether our analyses perfectly line up.
220 Zac Cogley

naturally seen as inspiring corresponding accounts of moral responsibility,

itself.
I think this goes some distance toward accounting for the fact that there
is so little agreement about the nature of moral responsibility, even after so
much attention to it. In my view, all the accounts I’ve discussed get
something right about the moral psychology of moral responsibility and
its associated conditions of appropriateness, but they also ignore other
important features of our moral responsibility practices. My analysis sug-
gests moral responsibility may be best captured as a prototype concept:
when we hold people morally responsible in normal interpersonal inter-
actions, our responses typically conform to archetypal emotional reactions
that involve all three aspects of the blaming emotions. If we aim to have a
psychologically realistic picture of our moral responsibility practices,
I believe we must have a tripartite theory. Thus, in response to the
voluminous literature defending fittingness, communication, or sanction
as the correct account of moral responsibility, my analysis suggests that
such attempts require additional argument. There is little to be gained for
supporting one account over another as the correct account of moral
responsibility without some attention to our purpose in raising the question
of someone’s moral responsibility in a given context.
It’s also no surprise on my analysis that theorists who emphasize different
emotional functions in their characterization of moral responsibility dis-
agree about the conditions under which people are morally responsible for
what they do. In particular, we find a ready division between incompatibi-
lists, who tend to emphasize the sanctioning function, and compatibilists,
who tend to emphasize the appraisal and communicative functions.
Moving forward with respect to points of dispute between the camps
may be aided by further attention to the moral psychology of the blaming
emotions and their conditions of appropriateness. In particular, I want to
suggest that if we often implicitly attend to the different conditions of
emotional appropriateness in our moral lives, our intuitive judgments
about when people are morally responsible will be influenced by those
considerations. Further, I expect that this influence will also be present
when we think about people’s moral responsibility in the context of
philosophical thought experiments. This is a speculative claim, but it
bears investigation, particularly in light of recent work on implicit and
affective psychological process.25

25 For an overview of some of the relevant empirical data, see (Bargh and Chartrand
1999).
 Significance of the Blaming Emotions 221

There is no question that P. F. Strawson’s “Freedom and Resentment”

looms large over investigations into moral responsibility. One way in which
Strawson influenced current debates is by motivating the idea that attribu-
tions of moral responsibility can be helpfully understood as “natural human
reactions to the good or ill will or indifference of others towards us”
(P. F. Strawson 1982, 53). I have here restricted my attention to just one
set of natural human reactions, namely, the blaming emotions. If my
account is plausible, these reactions to the quality of other’s wills are a
matter of significant complexity. Theorizing about moral responsibility
must respect these intricacies if we are to progress.26

REFERENCES

Averill, James R. (1983). “Studies on Anger and Aggression: Implications for

Theories of Emotion.” American Psychologist 38 (11), 1145–60.
Bargh, John A, and Tanya L Chartrand (1999). “The Unbearable Automaticity of
Being.” American Psychologist 54 (7): 462–79. doi:10.1037/0003-066X.54.7.462.
Baumeister, Roy, K. D. Vohs, and C. Nathan DeWall (2007). “How Emotion
Shapes Behavior: Feedback, Anticipation, and Reflection, Rather Than Direct
Causation.” Personality and Social Psychology Review 11 (2): 167.
Bennett, Jonathan (1980). “Accountability.” In Philosophical Subjects: Essays Pre-
sented to P.F. Strawson, ed. Zak van Straaten. (Oxford: Clarendon).
Berkowitz, Leonard, and Eddie Harmon-Jones (2004a). “More Thoughts About
Anger Determinants.” Emotion 4 (2): 151–5.
—— (2004b). “Toward an Understanding of the Determinants of Anger.” Emotion
4 (2): 107–30.
D’Arms, Justin and Daniel Jacobson (2000). “The Moralistic Fallacy: On the
‘Appropriateness’ of Emotions.” Philosophy and Phenomenological Research 61
(1) (July): 65–90. doi:10.2307/2653403.
Darwall, Stephen (2006). The Second-Person Standpoint: Morality, Respect, and
Accountability. (Cambridge, MA: Harvard University Press).
Dolinko, David (1991a). “Three Mistakes of Retributivism.” UCLA Law Review
39: 1623.
—— (1991b). “Some Thoughts About Retributivism.” Ethics 101 (3) (April):
537–59.
Ekman, Paul (1999). “Basic Emotions.” In Handbook of Cognition and Emotion,
eds. Tim Dalgleish and Mick Power, (New York: Wiley), 45–60.

26 This paper has been much improved by helpful discussion with a friendly audience
at the New Orleans Workshop on Agency and Responsibility. Antony Aumann, Andrea
Scarpino, and David Shoemaker also provided particularly beneficial constructive com-
ments and discussion. Finally, two anonymous referees from Oxford University Press
supplied instructive comments.
222 Zac Cogley

Fehr, Ernst and Simon Gächter (2000). “Cooperation and Punishment in Public
Goods Experiments.” The American Economic Review 90 (4): 980–94.
——.(2002). “Altruistic Punishment in Humans.” Nature 415, 137–40.
Fernández-Dols, José-Miguel, and Marı́a-Angeles Ruiz-Belda.(1995). “Are Smiles a Sign
of Happiness? Gold Medal Winners at the Olympic Games.” Journal of Personality
and Social Psychology 69 (6): 1113–19. doi.10.1037/0022-3514.69.6.1113.
Fischer, John Martin and Mark Ravizza (1998). Responsibility and Control: A Theory
of Moral Responsibility. (Cambridge, Cambridge University Press).
—— and Neal A. Tognazzini (2010). “The Physiognomy of Responsibility.” Philosophy
and Phenomenological Research (December). doi:10.1111/j.1933-1592.2010.00458.x.
Haidt, J. (2001). “The Emotional Dog and Its Rational Tail: A Social Intuitionist
Approach to Moral Judgment.” Psychological Review 108 (4): 814.
Hieronymi, Pamela (2004). “The Force and Fairness of Blame.” Philosophical
Perspectives 18 (1): 115–148. doi:10.1111/j.1520-8583.2004.00023.x.
Horstmann, Gernot (2003). “What Do Facial Expressions Convey: Feeling States,
Behavioral Intentions, or Actions Requests?” Emotion 3 (2): 150–66. doi:10.1037/
1528-3542.3.2.150.
Kraut, Robert E, and Robert E. Johnston (1979). “Social and Emotional Messages
of Smiling: An Ethological Approach.” Journal of Personality and Social Psychology
37 (9): 1539–53. doi:10.1037/0022-3514.37.9.1539.
Lazarus, R. S. (1982). “Thoughts on the Relations Between Emotion and Cogni-
tion.” American Psychologist 37 (9): 1019–24.
—— (1984). “On the Primacy of Cognition.” American Psychologist 39 (2): 124–9.
—— (1991). Emotion and Adaptation. (New York: Oxford University Press).
Leith, K. P, and Roy Baumeister (1996). “Why Do Bad Moods Increase Self-
defeating Behavior? Emotion, Risk Tasking, and Self-regulation.” Journal of
Personality and Social Psychology 71 (6): 1250–67.
Lerner, Jennifer S., and D. Keltner (2000). “Beyond Valence: Toward a Model of
Emotion-specific Influences on Judgement and Choice.” Cognition & Emotion 14
(4): 473–93.
—— —— (2001). “Fear, Anger, and Risk.” Journal of Personality and Social
Psychology 81 (1): 146–59.
Lerner, Jennifer S., and Larissa Z Tiedens (2006). “Portrait of the Angry Decision
Maker: How Appraisal Tendencies Shape Anger’s Influence on Cognition.”
Journal of Behavioral Decision Making 19 (2) (April 1): 115–37. doi:10.1002/
bdm.515.
Lerner, Jennifer S., Julie H. Goldberg, and Philip. E. Tetlock (1998). “Sober
Second Thought: The Effects of Accountability, Anger, and Authoritarianism
on Attributions of Responsibility.” Personality and Social Psychology Bulletin 24 (6)
(June 1): 563–74. doi:10.1177/0146167298246001.
Levy, Neil (2011). Hard Luck. (New York: Oxford University Press).
Litvak, P. M, J. S Lerner, L. Z Tiedens, and K. Shonk. 2010. “Fuel in the Fire: How
Anger Impacts Judgment and Decision-Making.” International Handbook of
Anger: 287–310.
 Significance of the Blaming Emotions 223

McKenna, Michael (1998). “The Limits of Evil and the Role of Moral Address:
A Defense of Strawsonian Compatibilism.” The Journal of Ethics 2 (2) (June 1):
123–42. doi:10.1023/A:1009754626801.
—— (2004). “Responsibility and Globally Manipulated Agents.” Philosophical
Topics 32 (1/2): 169.
Macnamara, Coleen (forthcoming). “ ‘Screw You!’ & ‘Thank You’.” Philosophical
Studies: 1–22. doi:10.1007/s11098-012-9995-3.
—— (2009). “Holding Others Responsible.” Philosophical Studies 152 (October
23): 81–102. doi:10.1007/s11098-009-9464-9.
Ochs, Jack and Alvin E. Roth (1989). “An Experimental Study of Sequential
Bargaining.” The American Economic Review 79 (3) (June 1): 355–84.
Parkinson, Brian (1997). “Untangling the Appraisal-emotion Connection.” Person-
ality and Social Psychology Review 1 (1): 62–79.
——— (1999). “Relations and Dissociations Between Appraisal and Emotion
Ratings of Reasonable and Unreasonable Anger and Guilt.” Cognition & Emotion
13 (4): 347–85.
Pereboom, Derk (2001). Living Without Free Will. (Cambridge, MA: Cambridge
University Press).
—— (2007). “Hard Incompatibilism.” Four Views on Free Will: 85–125.
—— (2009). “Free Will, Love, And Anger.” Ideas y Valores: Revista Colombiana De
Filosofı́a (141): 169–89.
—— (2011). “Free Will Skepticism and Meaning in Life.” In The Oxford Hand-
book of Free Will, ed. Robert Kane. 2nd edn. (New York: Oxford University
Press).
Pillutla, Madan M., and J. Keith Murnighan (1996). “Unfairness, Anger, and Spite:
Emotional Rejections of Ultimatum Offers.” Organizational Behavior and
Human Decision Processes 68 (3) (December): 208–24. doi:06/obhd.1996.0100.
Porter, Stephen and Leanne ten Brinke (2008). “Reading Between the Lies.” Psycho-
logical Science 19 (5) (May 1): 508–14. doi:10.1111/j.1467-9280.2008.02116.x.
Prinz, Jesse (2004). Gut Reactions: A Perceptual Theory of Emotion. (New York:
Oxford University Press).
—— and Shaun Nichols (2010). “Moral Emotions.” In The Moral Psychology
Handbook, ed. John Doris (Oxford: Oxford University Press), 111–48.
Roseman, Ira J., Cynthia Wiest, and Tamara S Swartz (1994). “Phenomenology,
Behaviors, and Goals Differentiate Discrete Emotions.” Journal of Personality and
Social Psychology 67 (2): 206–21. doi:10.1037/0022-3514.67.2.206.
Scanlon, T. M. (2008). Moral Dimensions: Permissibility, Meaning, Blame. (Cam-
bridge, MA: Belknap, Harvard University Press).
Scherer, K. R. (1986). “Vocal Affect Expression: A Review and a Model for Future
Research.” Psychological Bulletin 99 (2): 143–65.
Shah, Nishi. (2003). “How Truth Governs Belief.” The Philosophical Review
112 (4) : 447.
—— R. Banse, H. G Wallbott, and T. Goldbeck. (1991). “Vocal Cues in Emotion
Encoding and Decoding.” Motivation and Emotion 15 (2): 123–48.
224 Zac Cogley

Shaver, P., J. Schwartz, D. Kirson, and C. O’Connor (1987). “Emotion Know-

ledge: Further Exploration of a Prototype Approach.” Journal of Personality and
Social Psychology 52 (6): 1061–86.
Shoemaker, David (2011). “Attributability, Answerability, and Accountability:
Toward a Wider Theory of Moral Responsibility.” Ethics 121 (3): 602–32.
Smith, Angela M. (2006). “Control, Responsibility, and Moral Assessment.” Philo-
sophical Studies 138 (December 15): 367–92. doi:10.1007/s11098-006-9048-x.
—— (2007). “On Being Responsible and Holding Responsible.” The Journal of
Ethics 11 (January 4): 465–84. doi:10.1007/s10892-005-7989-5.
Smith, C. A, and R. S Lazarus (1993). “Appraisal Components, Core Relational
Themes, and the Emotions.” Cognition & Emotion 7 (3): 233–69.
Sommers, Tamler (2007). “The Objective Attitude.” The Philosophical Quarterly 57
(July): 321–41. doi:10.1111/j.1467-9213.2007.487.x.
Strawson, Galen (2002). “The Bounds of Freedom.” In The Oxford Handbook of
Free Will, ed. Robert Kane. (New York: Oxford University Press).
Strawson, P. F. (1982). “Freedom and Resentment.” In Free Will, ed. Gary Watson.
(Oxford, Oxford University Press).
Wallace, R. J. (1994). Responsibility and the Moral Sentiments. (Cambridge, MA:
Harvard University Press).
Wallbott, Harald (1998). “Bodily Expression of Emotion.” European Journal of
Social Psychology 28 (6) (November 1): 879–96. doi:10.1002/(SICI)1099-0992
(1998110)28:6<879::AID-EJSP901>3.0.CO;2-W.
Watson, Gary (1993). “Responsibility and the Limits of Evil: Variations on a
Strawsonian Theme.” In Perspectives on Moral Responsibility, eds. John Martin
Fischer and Mark Ravizza. (Ithaca, NY: Cornell University Press) 119–50.
—— (1996). “Two Faces of Responsibility.” Philosophical Topics 24 (2): 227–48.
Zajonc, R. B. (1980). “Feeling and Thinking: Preferences Need No Inferences.”
American Psychologist 35 (2): 151–75. doi:10.1037/0003-066X.35.2.151.
—— (1984). “On the Primacy of Affect.” American Psychologist 39 (2): 117–23.
 9
Unwitting Wrongdoers and the Role
of Moral Disagreement in Blame1
Matthew Talbert

In “Culpable Ignorance,” Holly Smith says that “[i]gnorance of the nature

of one’s act is the pre-eminent example of an excuse that forestalls blame”
(543). So, to adapt an example of Smith’s, a doctor who gives a patient the
wrong treatment might avoid blame if she thought she was providing her
patient with the best care she could. Whether the doctor avoids blame in
this way will depend, in part, on whether she was blameworthy for being
mistaken about what treatment was best. If the doctor prescribed the wrong
treatment because of an earlier “benighting act” that impaired (or failed to
improve) her cognitive position, then, if we take the doctor to be culpable for
the benighting act, we may also regard her as blameworthy for her ignorance
and for the consequences of her ignorance (547).
Smith’s focus is on agents whose ignorance is circumstantial rather than
moral: the doctor in the above example is unaware that the treatment she
prescribes is not best for her patient, but she presumably is aware that she
morally ought to give her patient the best treatment.2 More recently,
Gideon Rosen has applied an approach like Smith’s to cases of both

1 I would like to thank the participants at the 2011 New Orleans Workshop on
Agency and Responsibility for their questions and comments on this paper, particularly
Sarah Buss, T. M. Scanlon, Sarah Stroud, and Gary Watson. I am also very grateful to
David Shoemaker and to two anonymous referees for Oxford University Press for their
written comments.
2 It is true that the doctor suffers from normative ignorance since she does not know
that she has prescribed an inappropriate medication, but this is not what I mean by
“moral ignorance.” A doctor who suffers from moral ignorance (as I shall use the phrase)
would not know that it is wrong to intentionally prescribe a patient the wrong medica-
tion. The doctor in the example suffers from circumstantial ignorance because she is not
aware that, given her circumstances, her action will lead to bad consequences for her
patient.
226 Matthew Talbert

circumstantial and moral ignorance (Rosen 2004).3 Rosen assumes that for
an agent to be blameworthy for actions that issue from moral or
circumstantial ignorance, the agent must be culpable for her ignorance.
This assumption plays a crucial role for Rosen in a skeptical argument that
calls into question many intuitive judgments of moral blameworthiness.
Though I reject Rosen’s conclusion, I find his skeptical argument
powerful. Indeed, and as I attempt to show, the argument is sufficiently
strong that a relatively conservative approach to overturning it does not
succeed.4 Instead, if we are to avoid the skeptical conclusion, we must reject
the plausible sounding assumption that unwitting wrongdoers are
blameworthy only if they are culpable for their ignorance. To this end,
I argue that while ignorance of the circumstances and consequences of one’s
actions often undermines blame, moral ignorance typically does not do so.
For example, while I might not be blameworthy for injuring you if I was
unaware that my action would have that result, I likely would be
blameworthy if I were simply unaware that injuring you is impermissible.
I will argue, moreover, that a morally ignorant wrongdoer can be
blameworthy even if it is not her fault that she is ignorant of the moral
status of her behavior, and even if it would be unreasonable to expect her to
be aware of its status. In the context of making these points, I also try to
shed light on the role that moral disagreement plays in our judgments of
blameworthiness.

1 . S K E P T I C I S M A B O U T M OR A L R E S P ON S I B I L I T Y

Gideon Rosen has advanced a skeptical perspective on moral responsibility

based on the assumption that a wrongdoer is excused if she is nonculpably
unaware that she does wrong. For Rosen, moral responsibility (for bad
actions) amounts to being liable to the “sanctions” associated with moral
blame; so, for Rosen’s purposes, “we may simply identify moral responsi-
bility with culpability or blameworthiness” (2004: 296). As Rosen puts it,
“[s]kepticism about moral responsibility is thus the thesis that confident
positive judgments of blameworthiness are never justified ” (296).
Like many philosophers working on the subject, Rosen takes moral
blame to involve certain emotional responses. For example, “you blame
X for doing A, when you resent him or feel indignant towards him for

3 Also see Rosen (2003). Michael Zimmerman (1997) offers a view similar to
Rosen’s; also see Zimmerman (2008).
4 The conservative approach I have in mind is the one pursued by William FitzPatrick
(2008). I discuss FitzPatrick’s view in Section 2.
 The Role of Moral Disagreement in Blame 227

having done it” (296). Thus, if we judge that X is blameworthy, this means
“that X is liable [in our judgment] to a negative emotional response of this
sort for having done A, or equivalently, that some such response would be
appropriate or fitting” (297). Rosen’s skepticism amounts, then, to skepti-
cism about the aptness of negative reactive attitudes like resentment. In
what follows, I will understand moral responsibility in just the way Rosen
suggests: the primary question will be about whether judgments of blame-
worthiness, and responses like resentment, are apt.
Rosen’s skeptical argument proceeds as follows. Wrongdoing is either
witting or unwitting. While a knowing wrongdoer may be directly blame-
worthy for her behavior, an unwitting wrongdoer will be blameworthy for
her wrongdoing only derivatively or indirectly. This means that an ignorant
wrongdoer will be morally responsible, in the sense of being open to blame,
only if she is culpable for her ignorance.5 However, according to Rosen,
“[i]gnorance is culpable only if it derives from culpable recklessness or
negligence in the management of one’s opinion” and this prior recklessness
will be culpable only if it was knowing or was itself the product of knowing
mismanagement of one’s opinions (302). So, as Rosen puts it, ignorance is
culpable—and unwitting wrongdoing blameworthy—only if it results, at
some point, from an akratic act in which the agent knowingly violated an
epistemic or moral duty:
One is responsible for the act done from ignorance only if one is independently
responsible for something else . . . this entails that the only possible locus of original
responsibility is an akratic act. . . . Our first sin must be a knowing sin—a sin done in
full knowledge of every pertinent fact or principle. (307)6
Rosen applies this perspective to cases of circumstantial ignorance (like that
of the doctor in the introduction), but also to cases of moral ignorance. For
example, Rosen’s view suggests that an “ancient slaveholder who . . . believes
that it is morally permissible for him to buy and sell” slaves is blameworthy
for his behavior “only if he is culpable for the moral ignorance from which
he acts” (304).
The skeptical force of this position emerges when we consider the
possibility that many unwitting wrongdoers have never committed the
kind of akratic act that would make them culpable for their ignorance.

5 Following the authors I discuss, I use “culpable” and “blameworthy” interchangeably.

6 Zimmerman’s view is similar: if “one is culpable for ignorant behavior, then one is
culpable for the ignorance to which this behavior may be traced” (Zimmerman 1997:
418). However, “one is never in direct control of whether one is ignorant,” so culpability
for ignorant behavior must be indirect, which “presupposes direct culpability for
something else. . . . Hence all culpability can be traced to culpability that involves lack
of ignorance” (418).
228 Matthew Talbert

Imagine an “ambitious capitalist who is mistaken about where to draw the

line between permissibly aggressive business practices and reprehensibly
ruthless business practices” (305). Rosen supposes (and I agree) that it is
not hard to imagine a case in which the capitalist does something wrong
without recognizing that his behavior is wrong and that he has been as
reflective and careful as it is reasonable to expect him to be. Perhaps the
capitalist’s moral education was deficient, or perhaps the case he considers is
“just a hard case and after thinking about it for a decent interval he has simply
arrived at the wrong answer” (305). If we agree that “in reaching his [moral]
conclusion our capitalist has not been reckless or negligent,” Rosen thinks we
should also agree “that his moral ignorance is not his fault” and that it would
therefore “be a mistake to blame him for the wrong he does” (305).
Another example features Bill, who decides to tell a self-serving lie to his
wife even though he “knows that it’s just plain wrong to lie to your wife”
(305). Again, it is possible that Bill had a defective moral education that led
him to believe, “through no fault of his own, that while moral consider-
ations have some weight, they are not in general decisive” (305).7 Bill may
thus be “blamelessly (though mistakenly) convinced that the balance of
reasons comes down in favor of lying” (306). Rosen argues that in this case
it would be unreasonable to blame Bill for telling the lie. Rosen asks: “Does
it make sense to subject someone who blamelessly believes that he should
do A, and then does it, to moral sanctions—to recrimination, resentment,
righteous anger, contempt?” (306). Rosen thinks this doesn’t make sense.
Instead, Bill’s wife should respond this way:
Poor Bill. Through no fault of his own he found himself believing that all things
considered, he should lie. Given that he found himself in that state, I can hardly
fault him for lying. Holding the judgment fixed, the lie itself was a perfect
manifestation of practical rationality. I can fault him for the lie only if I can fault
him for believing that in the circumstances, his selfish interests were more important
than my moral interests. Since by hypothesis, it is not his fault that he held this
view, I have no option but to conclude that he is not properly culpable for his bad
action. (306)8
Rosen’s official skeptical conclusion is that “it would be unreasonable to
repose much confidence in any particular positive judgment of responsibility”
(308). This follows from the claim that an unwitting wrongdoer is blame-
worthy only if she has an akrasia-involving “inculpating history” (309)
together with the claim that “it is almost always unreasonable to place

7 Unlike the ambitious capitalist, Bill is mistaken about the force of moral consider-
ations rather than about a moral principle.
8 For discussion of a related example, see Rosen (2008: 605–9).
 The Role of Moral Disagreement in Blame 229

significant confidence in” the judgment that an agent has such a history
(308). In a recent response to Rosen, William FitzPatrick has argued that in
fact we often have good grounds for attributing akrasia to ourselves and
others (FitzPatrick 2008: 593–9). However, as FitzPatrick notes, this leaves
untouched Rosen’s conclusion that blameworthy wrongdoing is always
either knowing or the result of knowing wrongdoing (599). We are left,
then, with a striking skeptical challenge, for very many ordinary wrong-
doers may not have a relevant instance of akrasia in their past and so are not
morally responsible for their wrongdoing.9 Below, I consider Fitzpatrick’s
attempt to meet this skeptical challenge.

2. FITZPATRICK’S REPLY TO ROSEN

FitzPatrick accepts the “intuition at the core of [Rosen’s] argument, that it

is unfair to blame someone for an action done out of ignorance that he
cannot fairly be blamed for having” (2008: 601). However, FitzPatrick
rejects Rosen’s claim that unwitting wrongdoers are culpable for their
ignorance only if it resulted from knowing wrongdoing. According to
FitzPatrick, an unwitting wrongdoer will also be culpable for her ignorance
if she “could reasonably have been expected to take measures that would
have corrected or avoided it” (609).
To make his case, FitzPatrick expands on Rosen’s “ambitious capitalist”
example and considers Mr Potter, the ruthless businessman in Frank
Capra’s 1964 film, It’s a Wonderful Life. Potter is callous, vindictive, greedy,
and dishonest, but let us suppose that he wrongly believes that his behavior
is permissible. FitzPatrick argues that even if Potter did not akratically
acquire his false moral beliefs, there are three central factors “that make
most of us confident that Potter’s moral ignorance is culpable” (605). First
of all, Potter never engaged in the kind of moral reflection that might have
helped him correct his moral ignorance, yet “[t]here were no relevant
limitations in his social context or in his capabilities” that would have
made such reflection unreasonably difficult (605). This suggests that “[t]he
failure of adequate reflection” on Potter’s part “was instead the result of
voluntary exercises of vices such as overconfidence, arrogance, dismissive-
ness, laziness, dogmaticism, incuriosity, self-indulgence, contempt, and so
on” (605). According to FitzPatrick, Potter “could thus reasonably have
been expected to take steps that would have eliminated that [moral]
ignorance, by refraining from exercising those vices and instead taking

9 This is close to the skeptical conclusion advanced by Zimmerman (1997: 425).

230 Matthew Talbert

advantage of ” available opportunities for moral and epistemic improve-

ment (605).
So on FitzPatrick’s view, given Potter’s general capacities, and the fact
that “the opportunity for improved normative understanding was clearly
present in his social context,” it is reasonable to hold Potter to the expect-
ation that he correct his moral ignorance (603–4). Thus, Potter is culpable
for his ignorance and blameworthy for behavior that flows from it.
I agree with FitzPatrick that Potter is blameworthy for his bad behavior,
but I am not convinced that it is reasonable to expect Potter to correct his
moral ignorance. As FitzPatrick sees it, explaining Potter’s moral ignorance
in terms of his vices helps us see that this moral expectation is reasonable:
Potter’s failure to engage in reflection wasn’t forced on him by circum-
stances or inability, it resulted from vicious choices that he was free to omit.
I contend, however, that this emphasis on Potter’s vices is unhelpful for the
point FitzPatrick wants to make.
FitzPatrick characterizes Potter’s failure to correct his moral ignorance as
“the result of voluntary exercises of vices” (605). But what does Potter do
voluntarily? He does not voluntarily behave in ways that he regards as
impermissibly vicious—FitzPatrick’s point hangs on Potter not being a
knowing wrongdoer. Rather, Potter voluntarily exercises his vices in the
sense that he acts voluntarily and his vices shape his practical judgments in
characteristic ways. Now if Potter’s vicious actions are voluntary in this
sense, then he might have omitted them if he judged himself to have reason
to do so, but one symptom of Potter having his vices is that he sees little in
favor of such an omission. If Potter is afflicted with all the vices FitzPatrick
mentions, he is not likely to see much in favor of exploring the opportun-
ities for moral improvement available to him, so perhaps it is unreasonable
to expect him to do so.
Neil Levy argues similarly against FitzPatrick that it is not reasonable to
expect Potter to correct his vices because it is likely not subjectively rational
for Potter to do so. When we think about what can reasonably be expected
of a person, Levy says we should consider what that person can do by way of
a rational reasoning procedure, and what an agent can do in this way is a
function of her internal reasons, her “actual representations and proatti-
tudes” (Levy 2009: 736). As Levy notes, even if Potter’s failure to subject
his values to scrutiny is a manifestation of epistemic and moral vices,
by his lights, Potter governs his normative views adequately. He gives competing
views the attention he takes them to deserve. . . . But if Potter does not see that he is
managing his moral views badly, he has no (internal) reason to manage them any
differently. Potter exhibits epistemic vices aplenty, but because he does not conceive
of them as vices, he has no internal reason to refrain from so doing. (737)
 The Role of Moral Disagreement in Blame 231

Levy concludes that since Potter “could not rationally have taken advantage
of the opportunities for moral improvement” with which he was presented,
“we cannot reasonably expect him to do so” (735).10
There is also the question of how Potter acquired his vices. As FitzPatrick
notes, it may seem “problematic that the vices Potter exhibits in his
epistemically debilitating choices may trace back to his childhood and
may be largely a result of moral (bad) luck” (607). FitzPatrick deflects
this worry by noting that for most people “character traits are not merely
given but are formed, reformed and continuously shaped by our choices
from the point of moral maturity onward” (608). But this is not a helpful
response to the problem of constitutive moral luck since it simply pushes
the problem back to earlier stages of Potter’s development. Apparently,
Potter made poor choices as he shaped his character. But why did he make
such poor choices? By hypothesis, he did not knowingly make poor charac-
ter-forming choices, so perhaps Potter’s tendency to make “epistemically
debilitating” choices is explained by a tendency to see poor self-forming
choices as choice-worthy. But in this case, Potter’s self-forming choices
would seem to be shaped by incipient versions of the vices that his self-
formation is invoked to explain. And given the presence of these incipient
vices, why should we expect Potter to make the right choice when it comes
to choosing whether to act in a way that will strengthen his vices?
Finally, it is worth noting that FitzPatrick says that “cultural and
historical contexts” may make it unreasonable to expect a wrongdoer to
know better because “the relevant [moral] knowledge isn’t reasonably
available” (612). Aristotle, for example, may not be blameworthy for

10 This discussion is taken up again in (Levy 2011: 124–8). A referee for Oxford
University Press points out that we use the word “expect” in normative and descriptive
senses (to use the referee’s terminology). The descriptive sense presumably has to do with
what we anticipate from an agent in view of his capacities and the context in which he
acts, whereas the normative sense has to do with standards against which we measure
agents. When I say, “I expect it to rain tomorrow,” I use “expect” in the descriptive sense.
I anticipate rain; it seems like a safe bet. Neither Levy nor FitzPatrick expect moral
reform from Potter in this sense; neither regards it as a safe bet. (Though FitzPatrick is
keen to draw attention to Potter’s abilities and his social context and the way that these
make reform at least possible; he regards such reform as a live possibility even if not as a
likely outcome.) By contrast, both Levy and FitzPatrick agree that Potter falls short of
our expectations in at least one normative sense: they both regard Potter’s behavior as
morally subpar. The question is whether this subpar behavior opens Potter up to moral
blame. Here we might identify a slightly different sense in which “expect” can be
normative, the sense in which we talk about holding someone to an expectation.
Fitzpatrick thinks it is reasonable to hold Potter to our moral expectations because it is
possible for him to live up to these expectations given his abilities and the context in
which he acts. Levy thinks it is unfair to hold Potter to our moral expectations because he
can fulfill these only by behaving irrationally.
232 Matthew Talbert

thinking slavery permissible because slavery would have presented “a genu-

inely hard case for someone in Aristotle’s circumstances” (FitzPatrick: 600
n. 24). Since social and cultural factors can turn a question that is easy for
us—like “Is slavery wrong?”—into a difficult question for someone in
ancient Greece, this undermines the culpability of the morally ignorant
Greek. But Potter’s vices seem to turn apparently easy moral questions into
difficult ones, so perhaps these internal obstacles to moral knowledge
should (at least on FitzPatrick’s account) have the same excusing force as
cultural impediments to moral knowledge.

3 . R E J E C T I N G T H E SK E P T I C A L AR GU M E N T

I agree with Levy that it is unreasonable to expect Potter to recognize and

correct his moral ignorance. If this is right, and if it is also true that Potter
never committed a relevant act of knowing wrongdoing, then he would not
be culpable for his moral ignorance on either FitzPatricks or Rosen’s
account of culpable ignorance. What goes for Potter presumably goes for
many other unwitting wrongdoers, so if we accept the assumption—
endorsed by Rosen, FitzPatrick, and Levy—that unwitting wrongdoers
are blameworthy only if they are culpable for their ignorance, then we
should conclude that many (perhaps very many) ordinary wrongdoers are
not blameworthy for their bad behavior. Fortunately, we need not accept
this conclusion because there is good reason to reject the assumption that
unwitting wrongdoers are blameworthy only if they are culpable for their
ignorance.
The central instance of wrongdoing in It’s a Wonderful Life occurs when
Mr Potter keeps $8000 that George Bailey’s uncle Billy misplaced. Potter
keeps the money, believing (and hoping) that this will cause the foreclosure
of Bailey Building and Loan, as well as the prosecution of George Bailey for
bank fraud, and that this will leave the town of Bedford Falls ripe for
economic exploitation. Recall that we are concerned here with whether
someone like Potter is morally responsible for his behavior in the sense of
being an appropriate target for blaming attitudes like resentment. Plaus-
ibly, if George Bailey were to resent Potter, this resentment would be
provoked by the fact that Potter willingly acted so as to cause a bad
outcome for George, and that he did so because of a desire to injure George
and to extract an economic benefit from George’s misfortune. These
aspects of Potter’s behavior make George’s resentment natural and appro-
priate. If Potter’s actions had lacked these features—if he had been coerced
or had thought he was acting in George’s best interests—then it would be
 The Role of Moral Disagreement in Blame 233

inappropriate for George to resent Potter. But if Potter deliberately injured

George, and did so for self-serving reasons, then George can quite reason-
ably point to these facts to explain and justify his resentment even if Potter
happens to regard his behavior as permissible.11
The features of Potter’s behavior just mentioned make George’s blaming
responses appropriate because of the way these features are tied up with
Potter’s expression of contemptuous judgments and attitudes towards
George.12 For example, if Potter deliberately injured George for self-
serving reasons, then this behavior expresses the implicit judgment that
George’s welfare is unimportant in comparison with whether Potter
achieves his aims. (In fact, it’s not just that George’s welfare is
unimportant for Potter, he regards the possibility of injuring George as a
reason for acting.) Given the contemptuous judgments that inform Potter’s
behavior, George can reasonably regard Potter’s behavior as unjustifiable
and morally offensive—rather than as merely harmful or unwelcome—and
thus as proper grounds for resentment.
Similar points apply to Rosen’s example of Bill. Recall that, through no
fault of his own, Bill believes that his selfish reasons for lying to his wife
trump the moral reasons against doing so. Rosen says that
[a]nyone who bears our principles [about culpable ignorance] in mind and none-
theless judges that Bill is responsible for lying to his wife, is thereby committed to
the view that somewhere in the story of that lie there exists a full-blown episode of
altogether knowledgeable wrongdoing. (2004: 307–8)
“Responsibility” is an elastic term, so we can interpret the claim, “Bill is
responsible for lying to his wife,” in different ways. By hypothesis, Bill is
not at fault for thinking that he has decisive reason to lie, so Bill is not
responsible for lying to his wife in the sense of having played a particular
sort of causal role in the process that led him to think that lying is the thing
to do. But whether Bill is responsible for lying in this sense is not what is at
issue. Again, the issue is whether Bill is morally responsible in the sense that
attitudes like resentment would be a fitting response to him. The central
question about Bill is whether it makes “sense to subject someone [like Bill]
who blamelessly believes that he should do A, and then does it, to moral
sanctions—to recrimination, resentment, righteous anger, contempt?”
(306).

11 In Section 5, I address the concern that Potter’s nonculpable moral ignorance

makes it unfair to blame him.
12 For other applications of this kind of approach to blameworthiness, see
T. M. Scanlon (1998) and (2008), Angela Smith (2005) and (2008), and my own
(2008) and (2012a).
234 Matthew Talbert

I take it that Rosen believes that Bill is responsible for lying, in the sense
of being open to attitudes like resentment, only if he is responsible for
lying in the sense of having played the right sort of role in bringing about
his own tendency to favor lying. But Bill’s case is a good example of why
moral responsibility, in the sense of blameworthiness, does not require
that agents have this sort “inculpating history.” It is reasonable for Bill’s
wife to blame him because of the way his lying expresses Bill’s morally
faulty judgments and attitudes. Bill has these faults, and they contribute
to his behavior, regardless of whether he is at fault for having them, and
regardless of whether Bill genuinely believes that lying to his wife is the
thing to do.
One of Bill’s faults has to do with how his wife’s interests rate when he is
trying to figure out what to do: Bill is willing to overlook his wife’s interests
when they conflict with Bill avoiding trouble. It is reasonable, then, to
attribute to Bill the judgment that his wife’s interest in not being lied to can
be overlooked, if that is how Bill can get what he wants. If Bill’s wife were to
find out how her interests rate with her husband, that he lied to her and the
basis on which he did so, then she would have good grounds for blame.
That is, it would be appropriate for her to be offended and hurt by what
Bill’s action expresses, for her to protest that her interests ought to rate
more highly with Bill, to resent him for his callousness, to insist that he
change his ways, and so on.
The general perspective developed above is as follows. Even if a wrongdoer
is ignorant of the fact that her behavior is wrong, and even if this ignorance is
not her fault, her actions may still express the contemptuous judgment that
certain others do not merit consideration, that their interests do not matter,
and that their objections can be overlooked. If one is injured by a wrongdoer
who is moved by such judgments, then the attitudes and responses involved
in moral blame are reasonable regardless of what the wrongdoer thinks about
the moral status of her behavior. I will develop this perspective below, but
there is enough here to see one way of rejecting the skeptical perspective on
moral responsibility with which this paper began. Rosen’s skepticism
depends on assuming that unwitting wrongdoers are open to blame only if
they are culpable for their ignorance. However, as I have argued, certain
features of an unwitting wrongdoer’s behavior can qualify her for blame
regardless of whether she is culpable for her ignorance.

4. MORAL IGNORANCE AND MORAL DISAGREEMENT

Angela Smith has applied a view like the one I just outlined to show that
agents are often morally responsible for things that are not under their
 The Role of Moral Disagreement in Blame 235

direct control: their desires and emotions, their advertences and inadver-
tences, and so on. We are responsible for these things, on Smith’s view,
because of the evaluative judgments they express and the importance of
these judgments for our interpersonal relations.
Smith gives special attention to the fact that “we often take what a person
notices and neglects to have an enormous amount of expressive signifi-
cance” (2005: 242). What a person notices attracts our attention because
we assume a connection between what one notices and what one values.
According to Smith, “if one judges some thing or person to be important or
significant,” this should “have an influence on one’s tendency to notice
factors which pertain to the existence, welfare, or flourishing of that thing
or person” (244). And if one fails to notice such factors, this “is at least
some indication that she does not accept this evaluative judgment [about
the thing’s importance]” (244). As Smith says, I may fail to “notice when
my music is too loud,” or that “my advice is unwelcome,” or that “my
assistance might be helpful to others,” and even if these failures are involun-
tary, they may still indicate “that I do not judge your needs and interests to
be important, or at least that I do not take them very seriously” (244).
Though she does not describe them this way, the examples Smith cites
are instances of circumstantial ignorance. On Smith’s view, then, judg-
ments about a circumstantially ignorant wrongdoer’s moral responsibility
should track the plausibility of associating her behavior with interpersonally
significant evaluative judgments—particularly judgments about the nor-
mative status of the needs and interests of those affected by the wrongdoer’s
actions and omissions.
As should be clear from the previous section, I am largely in agreement
with Smith, but it is worth emphasizing that we cannot always infer that an
agent does not care about something from the fact that she fails to notice
how her actions (or omissions) will affect it. Smith is aware of this; she
notes that in some cases of inadvertence, “the person in question may be
extremely tired or under a lot of stress” and this may “block the normal
inference from what a person notices to what she cares about” (244 n. 14).
Indeed, we sometimes hesitate to make the inference from what a person
notices to what she cares about even when we cannot point to stressors that
intuitively explain an agent’s inadvertence. Sometimes, and for no obvious
reason, we just forget things, or fail to notice them, yet we may have as
much concern for these things as we ought to have. This can be true even in
cases in which an inadvertence has horrible consequences, such as when a
parent mistakenly leaves a child in a hot car. In some of these cases, it is no
doubt correct to infer that the parent has a condemnable lack of concern for
the child’s welfare. But in other cases, it is difficult to read the testimony of
the parents involved and come away with the thought that their forgetting
236 Matthew Talbert

is best explained by a morally deficient degree of parental concern. Many of

these parents seem to have been as concerned with their child’s welfare as
morality requires; yet, they left them behind all the same.13
We may often be unsure how to assess the sorts of cases just described,
but my general point is uncontroversial: we should be cautious before we
conclude that an unwitting wrongdoer’s actions or omissions express
evaluative judgments that might ground blame. More controversially,
I would argue that such caution is particularly appropriate in cases in
which the unwitting wrongdoer is ignorant of features of the context in
which she acts, or of the likely consequences of her behavior. There is
correspondingly less reason to be cautious about attributing blame-
grounding judgments in cases of moral ignorance in which an agent is
aware of the consequences of his behavior, but is unaware that it is wrong to
bring about those consequences.
One consideration in favor of this last claim is obvious: if an agent is
aware that act A will have consequence C, then, when she A’s, it is usually
reasonable to attribute to her at least an implicit judgment about how the
prospect of C bears on the question of whether to A. Of course, the fact that
an agent knows that her action will cause C does not entail that she formed
an objectionable judgment about the significance of C. The reason we
should expect the actions of morally ignorant wrongdoers like Mr Potter
to express objectionable judgments is that their moral ignorance is—unlike
a lot of circumstantial ignorance—often a manifestation of a normative
disagreement the agent has with those who object to his behavior. More
specifically, the moral ignorance of such an agent is often tied up with a
perspective that regards as unobjectionable the very thing that his victims
take to make his behavior objectionable.
Mr Potter’s moral ignorance, for example, is partly constituted by the
fact that he thinks he assigns George’s interests their proper weight in his
practical deliberations. Of course, we disagree with him about the norma-
tive significance of George’s interests, and since we take our position to be
the right one, we regard Potter as lacking moral knowledge. In such a
case—where a wrongdoer’s moral ignorance is part and parcel of a pro-
found normative disagreement between us and him—it is no surprise that
his behavior expresses judgments and attitudes that are, by our lights,
objectionable.
Of course, our normative judgments may conflict more or less pro-
foundly with the judgments we attribute to morally ignorant wrongdoers,

13 Cf. Levy (2011: 182). Gene Weingarten’s (2008) Pulitzer Prize winning article,
“Fatal Distraction,” offers a detailed and affecting account of some of these cases.
 The Role of Moral Disagreement in Blame 237

and as the severity of this disagreement decreases, so too may the intensity
of our blame. Take the case of Robert E. Lee. Those interested in burnish-
ing Lee’s reputation often note that he joined the secessionist cause in the
US Civil War out of loyalty to the state of Virginia. Lee did wrong,
I assume, in leading Confederate troops against the Union, but he thought
he was doing what duty and loyalty required. Suppose, however, that Lee
was motivated to support the Confederacy solely by racial hatred and a
desire to see slavery preserved. In this case, Lee would still be a morally
ignorant wrongdoer, but his defenders would face a much more difficult
task in convincing us to excuse him. This is because Lee’s actions would
have been associated with judgments and attitudes that we find deeply
objectionable. An excuse like, “Lee thought he was doing the right thing,”
would, I submit, have little influence on us if we found the judgments that
informed his choices thoroughly repugnant. And if we do have some
tendency to accept this excuse, I suspect it is because the judgments
about the value and requirements of loyalty that purportedly guided Lee
are not utterly foreign or repellant to us.

5. INTERNAL REASONS AND

THE FAIRNESS OF BLAME

In this section, I respond to the claim that it is unfair to blame an unwitting

wrongdoer who can’t reasonably be expected to omit her bad behavior.
Along the way, I develop the suggestion I made at the end of the last section
that moral blame often rests on the recognition of a moral disagreement
between ourselves and the one we blame. I will argue that what matters
most for judgments of blameworthiness are the considerations that count as
reasons for us (as issuers of blame), and not whether these considerations
could have counted as reasons for those we blame.
Let us return to Neil Levy’s discussion of Mr Potter’s blameworthiness.
According to Levy,
[i]t is not reasonable to blame agents for actions they cannot (intentionally) omit by
way of some reasoning procedure; we cannot hold them responsible for failing to
do things they could do only by chance or through a glitch in their agency. (Levy
2009: 739)
For Levy, part of the problem with blaming Potter is that his actual
proattitudes do not rationalize a decision to reassess his values and epi-
stemic practices, and this means that he lacks a fair opportunity to avoid
the actions for which we would blame him. As Levy sees it, “agents have a
fair opportunity to avoid performing actions . . . only if they could have
238 Matthew Talbert

rationally chosen to omit the action, and what agents can do rationally is a
function of their internalist reasons alone” (739).
I believe that the sense in which Potter has trouble avoiding wrong
actions does not make it unfair to blame him. Let us agree that Potter
has no internal reason to avoid a morally bad action, and that if he had
avoided the action, this would have been because of a “glitch” in his agency,
as Levy puts it. Importantly, this does not mean that Potter’s actual bad
behavior is the result of a glitch, mistake, compulsion, or anything similar.
If someone is subject to a compulsion or to “glitchy” agency, then some of
her actions may be unavoidable and she will have a claim to being excused
for this behavior. But the basis of excuse here—the reason the compulsive
agent is not appropriately targeted with resentment—is that her behavior is
not under her control in such a way that it can express the objectionable
judgments that would make blame appropriate. Usually, an agent’s inability
to avoid an action goes together with that action not being under the agent’s
control in this way, but cases of moral ignorance like Potter’s are an excep-
tion. Even if Potter does not have rational access to doing the right thing, his
bad behavior may be a manifestation of perfect reflective self-control and
subjective practical rationality: he may be acting just as he likes and for
reasons that really do speak in favor of so acting, given his proattitudes. But if
Potter’s bad actions are guided in this way by his judgment about reasons,
then they can express the kinds of attitudes and evaluative judgments that
I have argued make blaming responses appropriate.14
In addition, Potter presumably has the capacity to avoid wrongdoing in
the sense that, for any wrong action, he would have refrained from per-
forming that action if he had taken himself to have a decisive reason to do
so. Of course, if Potter’s actual constitution entails that he will see no such
reason, then he will not rationally avoid his wrongdoing, and it is this lack
of rational access to avoiding wrongdoing that Levy thinks makes blame
unfair. But if Potter’s bad actions are guided by his judgments about how to
behave, and if he would have acted differently if he had recognized a reason
to do so, then his wrongdoing is unavoidable mainly in the sense that an
action that is wrong (by our lights) is bound to seem choice-worthy to
Potter. Since this sort of unavoidability does nothing to make Potter’s
behavior less knowing, deliberate, and dismissive of the interests of those
his actions affect, I do not see why it should make blame unfair.
So, while I agree with Levy that it is not reasonable to demand that
Potter behave contrary to his internal reasons, what I think this tells us is

14 There is an important relation here with the theme of Harry Frankfurt (1969).
I develop this point in (2012a).
 The Role of Moral Disagreement in Blame 239

that having subjective rational access to avoiding an action is not a require-

ment on being properly blamed for that action. In other words, an agent
may be blameworthy for performing an action that she had no internal
reason to avoid. In my view, approaches like Levy’s focus too much on the
perspective of the blamed agent, on what can count as a reason for him, and
what can be expected of him. In fact, I think that our judgments about
blameworthiness are often formed—and rightly so—on the basis of the
normative considerations that we recognize, and not on whether those we
blame could have assigned the same normative weight to these consider-
ations that we do.
Suppose we believe that Potter unjustifiably injured George by a certain
action, and that while Potter willingly acted with the intent of injuring
George, he could not have been expected to omit this action because,
through no fault of his own, he had no internal reason to do so. If we
regard Potter’s treatment of George as unjustified, we at least think that
Potter does not agree with us about the status of things that are for us
important normative considerations. George, in particular, may think that
his welfare is valuable, and that a person of good will who is appropriately
sensitive to value will see a reason to avoid actions that are contrary to his
welfare. From George’s perspective, then, since Potter has failed to see his
welfare as reason-giving, this suggests that he is not disposed toward George
as a person of good will would be.
Because of George’s judgment about the importance of his own welfare,
Potter’s action appears unjustifiable and offensive to George. Potter, of
course, will see things differently: he thinks his treatment of George is
entirely appropriate and that he has no reason to refrain from it. This may
mean that Potter has no internal reason to omit his action, but regardless of
whether this is the case, it is still true that Potter does not view the prospect
of George’s injury as a reason to refrain from his action. Thus, Potter’s
action expresses the offensive judgment that George’s welfare is not par-
ticularly valuable.15 And even if George agrees that Potter had no (internal)
reason to refrain from his action, it is inappropriate to demand that George
regard Potter’s action as unobjectionable because this asks George to
concede that his welfare is not normatively significant. Even if George

15 What if a wild animal injured George? Would we say that the animal’s behavior
expresses a blame-grounding rejection of the significance of George’s welfare? I think
not. The behavior of animals is not morally significant in this way because a judgment
like, “George’s injuries don’t matter,” is not meaningfully attributed to them. I don’t
mean that nonhuman animals can never be described as being sensitive to reasons but
rather that their behavior does not have the same significance for us as that of beings
whose behavior is more richly and generally informed by evaluative judgments.
240 Matthew Talbert

agrees that, in the internalist sense, Potter has no reason to care how he
fares, it is still appropriate for George to insist that his welfare is valuable
and to see Potter’s rejection of this claim as a manifestation of ill will.
If we agree with George that his welfare is valuable, and that a person of
good will would see his welfare as a source of reasons, we should conclude
that Potter’s judgment about the significance of George’s welfare reason-
ably elicits blaming responses on George’s part. By contrast, if we think that
George’s welfare has little normative significance, we are likely to find
George’s blame inappropriate. However, this conclusion would stem not
from the thought that Potter can’t reasonably be expected to recognize the
significance of George’s welfare, but rather from our disagreement with
George about this significance.
It may be thought that Potter can reject the normative status of George’s
welfare in a morally relevant way only if he had rational access to an
accurate judgment about this status.16 I don’t see why this should be so.
What reason is there to think that the expressive significance, for George
and for us, of Potter’s behavior should hang on whether Potter might
rationally have made different judgments about reasons? After all (and as
I pointed out above), Potter’s actual behavior, and his actual failure to agree
with George about the significance of his welfare, need not have been
caused by a glitch, but may be a thoroughly deliberate and controlled
exercise of his agency. And suppose that Potter did have some internal
reason to take George’s welfare as a constraint on his behavior, but that he
still unjustifiably and deliberately injured him. If we did not already think
that Potter’s action was offensive in a blame-grounding way, I do not see
how adding this element to the story would make blame appropriate. What
matters is that George believes (and we believe) that he has standing that
makes Potter’s treatment of him illegitimate. Regardless of whether Potter
could have been rationally moved to accept our view about the treatment to
which George is entitled, his knowing and willing behavior demonstrates
that he rejects this view.
In one of his discussions of internal reasons, Bernard Williams charac-
terizes blame as functioning like advice given after the fact—it tells wrong-
doers what they ought to have done. As Williams says, “if ‘ought to have’ is
appropriate afterwards in the modality of blame, then (roughly) ‘ought to’
was appropriate at the time in the modality of advice” (Williams 1995: 40).
However,

16 This would be similar to Gary Watson’s claim that a wrongdoer flouts a moral
demand in a way that justifies resentment only if he is capable of recognizing the validity
of the demand (Watson 2004 : 234).
 The Role of Moral Disagreement in Blame 241

‘ought to’ in the modality of advice implies ‘can,’ because advice aims to offer
something as a candidate for a deliberative conclusion. If ç-ing is not available to
the agent, ‘You ought to ç’ cannot function as a piece of advice about what he
should now do; when it is a matter of what I am to do, manifestly ‘I cannot’ acts as a
stopper. (40)
Similarly, we might worry that if a wrongdoer has no internal reason to
refrain from an action, this is a “stopper” on blame because there is no
point to the advice that is supposedly implicit in blame.
I agree that when we blame a person, we are typically committed to the
claim that she ought not to have done what she did. But the “ought” of
blame is not always, or at least not solely, in the mode of advice. It can also
be an “ought” that points to an ideal or to a moral fact: the fact, for
example, that George deserves better treatment from Potter, and that a
person of good will would not have disregarded George’s interests and
welfare. This sort of “ought” does not imply “can.” Of course, as a form of
advice, it may be futile for George to insist to Potter that his welfare
matters. But as a form of protest this insistence is a natural way of expressing
the moral offense and resentment involved in blame.17 The reasons that
matter most for blame are our reasons—the considerations recognized by
the victim and by those who sympathize with her. Whether these
considerations can also be reasons for the blamed party is, I think, of
secondary importance.18

17 I argue that moral blame is sometimes best construed as a form of moral protest in
(2012a).
18 A referee for Oxford University Press suggests that I might note the affinity
between what I say in this section and, e.g., the sort of view once defended by Philippa
Foot (1972). There certainly seem to me to be affinities here but I am not sure how to
characterize them; I will simply note that I am very attracted to the sort of view Foot
outlines. Another referee suggests that I can in fact say that it is reasonable to expect
wrongdoers like Mr Potter to correct their moral ignorance. This suggestion relies on my
characterization of these wrongdoers as expressing ill will, the fact that expressions of ill
will (arguably) flout the demand to treat others with reasonable regard, and the fact that
this demand is connected to our normative expectations of other agents. I don’t doubt
that this proposal is workable but (in this and previous work) I prefer to grant to
opponents that certain moral demands and expectations are off the table in Potter’s
case (and other related cases). I think this helps to make the debate more clear than it
would be otherwise. However, the referee’s comment prompts me to note that my
disagreement with FitzPatrick is in some ways less stark than I presented it in
Section 2. FitzPatrick and I agree that Potter can reasonably be held to our moral
standards in the sense that we regard him as a proper target of moral blame. (I suspect
this is the sense in which the referee believes I can say that we properly expect better
things of Potter.) However, for FitzPatrick, the legitimacy of blaming Potter depends on
his having been able to avoid his moral ignorance. On my view, Potter’s blameworthiness
does not depend on this but only on the actual judgments that inform his behavior.
242 Matthew Talbert

6. CIRCUMSTANTIAL KNOWLEDGE AND BLAME

On the account I have presented, when it comes to thinking about how

knowledge relates to blameworthiness, we should consider what a wrong-
doer needs to know in order for her actions to express the attitudes and
judgments that make blame appropriate. What does an agent need to know
before we attribute to her a judgment like, “your objections and interests
can be overlooked when I am deciding how to act”? As I have argued,
awareness that one’s actions are wrong is not necessary for the presence or
expression of such judgments. Much more relevant in this context is
knowledge of the effects that one’s actions will have on others. If a person
knows that her action will injure someone who objects to this result, then it
is reasonable to attribute to her the judgment that the other’s injuries and
objections are not a decisive reason to refrain from the action. Judgments
like this help to ground blaming responses because of the way they call into
question the standing of the other to raise objections to certain forms of
treatment and to cite her interests as normative considerations.
Of course, we will not always think that an agent is open to blame when
her actions express the judgment that another’s interests and objections
may be overlooked. As I suggested with my example of Robert E. Lee in
Section 4, our views about an agent’s blameworthiness vary with the degree
to which we accept the judgments that move her. Most people believe that
it is permissible to use violence in self-defense against an aggressor. If we
take an instance of self-defense to be justified, then we likely think that
whatever objections the aggressor might raise to the force used against him
are rightly overlooked. Thus, we would be unlikely to blame the person
who defended herself even if she acted on the judgment that, in the relevant
instance, the aggressor’s welfare can be overlooked. The injured aggressor
may resent the one who injured him in self-defense, but by our lights, and
by those of the one who acted in self-defense, this blame will appear inapt
and unreasonable.
Similarly, if I uproot a plant, my action may express an implicit judg-
ment that plants do not have standing to raise objections to the way I treat
them. But I am not blameworthy here because this judgment is accurate;
treating plants this way is unobjectionable. Whatever objections there may
be to uprooting a plant are not grounded in the fact that plants are, in
themselves, the sort of things that can be treated objectionably.
Suppose, however, that I am wrong about this, and that plants can suffer,
and that they have a perspective from which to raise objections to my
treatment of them. In the context of criticizing views like the one defended
in this paper, Neil Levy proposes just this case:
 The Role of Moral Disagreement in Blame 243

Suppose that there is a kind of harm that is objectively morally relevant, but of
which we are ignorant. Suppose, for instance, that plants can be harmed, and that
this harm is a moral reason against killing or treading on them. In that case, many of
us are causally responsible for a great many moral harms. Are we morally responsible
for them? Do we flout a moral requirement, and challenge plants’ standing as
objects to which some moral consideration is owed? No to all these questions: If
we do not grasp the moral requirement, and this ignorance is not culpable, we do
nothing blameworthy. (2005: 9)19
Now I agree that we are not blameworthy for stepping on plants in Levy’s
example, but the example is offered as an instance of exculpation by way of
nonculpable moral ignorance and this is not what it shows. Levy’s example
fails to illustrate the exculpatory power of moral ignorance because what
explains why we are not blameworthy in the example is that we lack crucial
information about how walking on plants affects them. Since I do not
know that plants can be harmed, my stepping on one does not express a
denial of the significance of its being harmed. So, on the view I advocate,
stepping on a plant does not express a judgment that could properly ground
blame.
The problem is that Levy’s example does not involve the relevant sort of
moral ignorance. There is, of course, a kind of moral ignorance in the
example: we lack knowledge of the moral status of stepping on plants
(because we are ignorant of their capacity for suffering). But this moral
ignorance derives from circumstantial ignorance and not from ignorance of
a moral principle. In the example, I presumably know that I generally have
moral reasons to not cause pain; I just don’t know that stepping on plants
causes them pain. However, the moral ignorance of someone like Mr Potter
is just the reverse: he knows that his actions will cause George Bailey and
others pain, but he does not know (because he does not believe) that this
counts decisively against performing these actions.
For Levy’s example to help us see that Potter’s sort of moral ignorance
counts as an excuse, the example would have to feature Potter’s sort of
ignorance. The example would need to be one in which I know, say, that
stepping on a plant will cause it pain and I (nonculpably, but wrongly)
believe that this pain does not matter. Now if we think that I am blameless
in this revised version of Levy’s example, we might draw a conclusion about
the exculpatory significance of nonculpable ignorance of moral principles.
Of course, I would not agree that I am blameless in this revised case. If

19 I should note that Levy is considering psychopaths (rather than relatively normal
wrongdoers like Mr Potter) when he offers this example, but I don’t think this affects its
usefulness here. David Shoemaker (2011) treats a similar example; I reply to Shoemaker
in (2012b).
244 Matthew Talbert

I knowingly (and unjustifiably) cause a plant severe pain, then I am open to

resentment and indignation because my action dismisses the normative
significance of the plant’s pain, and this is something to which the plant (or
at least one who is concerned for its welfare) could reasonably object.

CONCLUSION

I have argued against the assumption that morally ignorant wrongdoers are
open to moral blame only if they are culpable for their ignorance. Thus,
I reject skepticism about moral responsibility that depends on this assump-
tion. On the view I have defended, the attitudes involved in moral blame
are responses to the features of an action that make it objectionable from
the perspective of the one who issues the blame. One important way that an
action can appear objectionable to us is that it expresses a judgment with
which we disagree about the significance of the needs and interests of those
affected by the action. Whether a wrongdoer’s action has this feature
depends more on whether she is aware of the consequences of her behavior
than on whether she regards her behavior as wrong.

REFERENCES

FitzPatrick, William J. (2008). “Moral Responsibility and Normative Ignorance:

Answering a New Skeptical Challenge.” Ethics 118: 589–613.
Foot, Philippa (1972). “Morality as a System of Hypothetical Imperatives.” The
Philosophical Review 81: 305–16.
Frankfurt, Harry (1969). “Alternate Possibilities and Moral Responsibility.” Journal
of Philosophy 66: 829–39.
Levy, Neil (2005). “The Good, the Bad, and the Blameworthy.” Journal of Ethics
and Social Philosophy 1: 1–16.
—— (2009). “Culpable Ignorance and Moral Responsibility: A Reply to FitzPa-
trick.” Ethics 119: 729–41.
—— (2011). Hard Luck: How Luck Undermines Free Will and Moral Responsibility.
(New York: Oxford University Press).
Rosen, Gideon (2003). “Culpability and Ignorance.” Proceedings of the Aristotelian
Society 103: 61–84.
—— (2004). “Skepticism about Moral Responsibility.” Philosophical Perspectives
18: 295–313.
—— (2008). “Kleinbart the Oblivious and Other Tales of Ignorance and Responsi-
bility.” The Journal of Philosophy 105: 591–610.
Scanlon, T. M. (1998). What We Owe to Each Other. (Cambridge, MA: Harvard
University Press).
 The Role of Moral Disagreement in Blame 245

—— (2008). Moral Dimensions: Permissibility, Meaning, Blame. (Cambridge, MA:

Harvard University Press).
Shoemaker, David (2011). “Attributability, Answerability, and Accountability:
Toward a Wider Theory of Moral Responsibility.” Ethics 121: 602–32.
Smith, Angela (2005). “Responsibility for Attitudes: Activity and Passivity in
Mental Life.” Ethics 115: 236–71.
—— (2008). “Control, Responsibility, and Moral Assessment.” Philosophical Stud-
ies 138: 367–92.
Smith, Holly (1983). “Culpable Ignorance.” The Philosophical Review 92: 543–71.
Talbert, Matthew (2008). “Blame and Responsiveness to Moral Reasons: Are
Psychopaths Blameworthy?” Pacific Philosophical Quarterly 89: 516–35.
—— (2012a). “Moral Competence, Moral Blame, and Protest.” The Journal of
Ethics 16: 89–109.
——(2012b). “Aliens, Accountability, and Psychopaths: A Reply to Shoemaker.”
Ethics 122: 562–74.
Watson, Gary (2004). “Responsibility and the Limits of Evil: Variations on
a Strawsonian Theme.” In Agency and Answerability. (New York: Oxford
University Press), 219–59.
Weingarten, Gene (2008). “Fatal Distraction.” The Washington Post, March 8, W8.
Williams, Bernard (1995). “Internal Reasons and the Obscurity of Blame.” In
Making Sense of Humanity, 35–45. (Cambridge: Cambridge University Press
(1995)), 35–45.
Zimmerman, Michael (1997). “Moral Responsibility and Ignorance.” Ethics 107:
410–26.
—— (2008). Living with Uncertainty. (Cambridge: Cambridge University Press).
 10
Partial Desert
Tamler Sommers

1. INTRODUCTION

Moral luck occurs when agents are morally evaluated for things that are
beyond their control (Williams, 1981; Nagel, 1979). A particularly clear
kind of moral luck is “resultant luck,” or luck in the way things turn out.
Attempted murderers are judged less harshly than successful ones though
both intended to kill their victims. Drunk drivers who have an accident
resulting in the death of a pedestrian may be convicted for manslaughter or
worse; drunk drivers who are caught without incident get nothing more
than a suspended driver’s license or compulsory participation in an online
seminar. Some philosophers, most notably Aristotle, accepted moral luck as
a fact of life. Contemporary philosophers, however, tend to regard moral
luck as a serious problem or even a paradox because they employ a Kantian
concept of moral desert according to which agents can be justly blamed or
praised only for aspects of actions that are within their control.
The ongoing controversy over victim impact statements (VIS) shows
that this conception of moral desert is pervasive within criminal justice
systems as well, at least in the West. VIS are statements that express the grief
and suffering of the victims of crimes as well as their views on what would
be a suitable punishment for the offender. In the well known 1987
Supreme Court case Booth v. Maryland, the Justices overturned a lower
court’s use of a VIS in a capital crime. Writing for the majority decision,
Justice Lewis Powell explains:
The focus of a VIS . . . is not on the defendant, but on the character and reputation
of the victim and the effect on his family. These factors may be wholly unrelated to the
blameworthiness of a particular defendant. As our cases have shown, the defendant
often will not know the victim, and therefore will have no knowledge about the
existence or characteristics of the victim’s family. Moreover, defendants rarely select
their victims based on whether the murder will have an effect on anyone other than
the person murdered. Allowing the jury to rely on a VIS therefore could result in
 Partial Desert 247

imposing the death sentence because of factors about which the defendant was unaware,
and that were irrelevant to the decision to kill. This evidence thus could divert the
jury’s attention away from the defendant’s background and record, and the circum-
stances of the crime. ((Booth v. Maryland, 1987, p. 253; my italics)
The worry about VIS noted in the majority decision is precisely that it
introduces an unacceptable degree of resultant luck in the sentencing
process. According to Powell, punishment should be tied exclusively to
the “personal responsibility and moral guilt” of the offender. The VIS
contains information that the offender could not possibly have known or
foreseen at the time of the crime. The amount of grief a particular family
will feel, how vindictive or forgiving their natures happen to be—all of this
is beyond the control of the offender. It has nothing to do with the
offender’s “decision to kill” and, according to Powell, is therefore unrelated
to his moral guilt and personal responsibility.
Though the courts may try to minimize its effects in certain cases, it is
clear that moral luck pervades our legal system and everyday lives. It seems
right that a drunk driver who accidentally kills a pedestrian deserves more
punishment than one who made it home safely, even if it is perhaps unfair
for the difference to be so extreme. How can we account for our intuitions
in such cases? One alternative would be to claim that although their
intentions were the same, the drivers performed two different actions.
The first driver is judged for the act of killing a pedestrian while driving
drunk, the second for a normal DUI. Another would be to claim that the
two drivers are equally deserving of blame, but that judgments about the
proper punishment must take harm into account.1 Finally, one may claim
that the drivers deserve the same amount of punishment, but that for
various consequentialist reasons we have to punish the second driver
more harshly.
In this chapter, I defend a fourth alternative. I argue that we should not
understand desert as impartial or “blind,” connected only to the personal
culpability of the agent. Rather, we should instead adopt a “partial” account
according to which desert judgments are properly sensitive to the feelings,
desires, and behavior of those most closely affected by the wrongdoing.
Section 2 outlines in a little more detail the conception of moral desert that
I wish to challenge. Sections 3 and 4 present several cases and variations that
appear to undermine the impartial view and offers a new account of desert
that can better account for our judgments in the cases. Section 5 introduces a

1
This alternative may just push the problem back a step, however, since it does not
explain why it is fair to punish wrongdoers for aspects of their behavior that are beyond
their control.
248 Tamler Sommers

relevant distinction in penal philosophy about the relationship between desert

and proportionality. The final sections defend the partial account against
common objections and offer reasons to prefer to it alternative accounts.

2. THE ACCEPTED FRAMEWORK FOR

D ES E R T J UD G M E N T S

Theories of moral desert—both compatibilist and incompatibilist—devote

most of their attention to identifying general conditions or criteria that
have to be met in order for agents to be blameworthy for their behavior.2
The conditions differ depending on the theory, but they all focus
exclusively on facts about the agent. This is true for both compatibilist
conditions (e.g. reasons-responsiveness, attributability) and incompatibilist
conditions (e.g. ultimate responsibility or the ability to do otherwise).3
Only agents who meet these conditions are eligible to deserve blame and/or
punishment for their actions.
Skeptics about desert can stop here since the conditions for moral
responsibility in their accounts cannot be met. Nonskeptical theories,
however, must also offer a way to determine how much blame or punish-
ment an agent deserves. Although this aspect of the debate does not receive
much attention, the formula seems to go as follows. Agent-centered facts
determine if the agent is morally responsible and perhaps to what degree.4
This judgment is then coupled with judgment about the gravity of the
offense to determine the amount of blame or punishment the agent
deserves. For the purposes of this paper, I will refer to judgments that
combine (1) the severity of the wrongdoing and (2) the agent’s moral
responsibility for performing it as judgments about the agent’s personal
culpability. The accepted framework—or what I will sometimes refer to as
the impartial conception of desert—regards personal culpability as
determinative of how much blame or punishment the agent deserves.
For some, these remarks may seem so obvious as to be hardly worth
mentioning. We are, after all, assessing what the agent deserves, so of course
we look to facts about agents and their actions. And certainly, when

2
This chapter focuses only on desert for morally wrong or bad actions.
3
See e.g. Fischer and Ravizza (1998), Kane (1996), van Inwagen (1983). Exceptions
to this rule, depending on one’s interpretation, may include Strawson (1962), Wallace
(1994), and Scanlon (2008).
4
It is surprisingly difficult to find discussions of the degrees of moral responsibility in
the philosophical literature. Perhaps this is because theories of moral responsibility tend
to be framed in terms of necessary and sufficient conditions.
 Partial Desert 249

considered in abstract terms this approach to desert is intuitively compel-

ling. It is also consistent with certain intuitions about fairness and our
desire to be master of our own fate. When applied to particular cases,
however, the framework can produce counterintuitive results and may
therefore require revision. The modification I propose accepts that facts
about personal culpability are necessary for making desert judgments, but
denies that they are the whole story. Desert judgments, I argue, must in
addition consider certain facts that are independent of the agent and the
action. I refer to this as the “partial conception” of desert because it takes
into account facts about particular individual victims—their behavior,
desires, and attitudes—all of which can be beyond the offender’s control.5
In my revised framework, the agent’s personal culpability sets a spectrum for
how much blame and punishment can be deserved. But additional facts are
required to make more precise determinations within that spectrum. On
my account, then, agents who are equally culpable may deserve different
amounts of blame and punishment depending on these facts.

3. PARTIAL DESERT

The following two cases offer some support for my proposal. They begin
the same way:
John is a 33-year-old graduate student at the University of Utah. He goes
to a football game and gets very drunk. He plans to leave his car at the
stadium and get a ride home from a friend, but there is miscommunica-
tion and his friend leaves without him. In general, John is morally
opposed to drunk driving and almost never does. But it is almost impos-
sible to get a cab, so John reluctantly drives home in his intoxicated state.
Just before he reaches his house, he has an accident causing him to swerve
into a driveway where a young girl was playing. The girl is killed instantly.
Panicked, still drunk, not thinking clearly, he leaves the scene and goes
home. As he sobers up, he is overcome with remorse. He considers
turning himself in but is terrified of going to jail and decides against it.
Now the story splits into two directions.

First Scenario
The police track down John, arrest him, and put him on trial. Perhaps
because the death involved a child, as well as a hit and run, the DA

5
As should be clear, I use “partial” here to contrast with “impartial” rather than with
“wholly” or “fully.”
250 Tamler Sommers

manages to convict John for homicide and the judge sentences him to
the death penalty. Since this takes place in Utah, John dies at the hands
of a firing squad.
Judgment: Most, I imagine, would call this an unjust verdict. The killing,
after all, was completely unintentional. John showed no ill will whatsoever
towards his victim or anyone else. True, he made the decision to drive drunk
but there were many people in far worse condition than John who drove
home from the game and were lucky enough to avoid this tragedy. It is truly
a case of terrible moral luck that his accident resulted in the death of a child.
To be sure, John deserves a harsh sentence for his crime, John himself would
likely agree with that. But few would say that he deserves to die for it.

Second Scenario
The police are unable to discover who caused the accident. The parents
of the child are grief stricken, completely distraught. Their daughter
meant everything to them. Since the police are overtaxed and the case has
gone cold, they vow to find the culprit themselves. They cash out their
retirement funds, sell their house, and hire the best private investigators.
Eventually they discover that it was John who caused the death of their
daughter. The father goes to John’s house, taking his gun. When he sees
John, he is overwhelmed with anger and grief. The image of his daughter
playing in their driveway flashes through his head. He takes out his gun
and shoots John in the heart, killing him.
Judgment: In this scenario, by contrast, it seems far more plausible that John
gets what he deserves. At the very least, John seems significantly more
deserving of his fate in the second scenario than the first. (If “John gets
what he deserves” were on a Likert scale, I imagine people would be much
closer to “agree” in the second scenario than the first.) Furthermore, John
himself would likely feel the same way—I certainly would in his shoes.
Facing the firing squad, John might be furious at the injustice and the
unlawfulness of the verdict. But looking down the barrel of the father’s gun,
he may think: “I ended the life of this man’s child. If he wants to shoot me,
that’s his right. I have this coming, it’s what I deserve.”
I should emphasize we are not evaluating the morality of the father’s
action, but rather whether John receives what he deserves. These are
separate matters. To take a grisly example, imagine that a gang of rapists
coincidentally choose as their victim a man who is a serial rapist himself.
We might say that gang acted immorally but nevertheless that the serial
rapist got precisely what he deserved. My claim, then, is not that the father
acted rightly in shooting John. It is that John seems to deserve his fate
(being shot in the heart) more when it is the father, rather than State, who
 Partial Desert 251

carries it out. Yet the accepted framework cannot account for this judg-
ment. In both cases, John’s personal culpability and his punishment (being
shot in the heart) are identical. Our judgments about what John deserves,
then, do not seem to be based entirely on facts about the wrongdoing and
John’s responsibility for performing it.6
For readers who lack my intuitions about these cases, my argument will
likely not have much force—at least not yet. Those who share my intuitions
but still wish to preserve the impartial conception of desert must explain
why we come to different judgments in the two cases. One might appeal to
consequentialist considerations, but desert—as a backwards-looking con-
cept—is essentially nonconsequentialist in nature; it would be surprising if
the difference in intuitions were sensitive to such factors. More import-
antly, it is not clear that the consequences are better in the second case than
in the first. Indeed, they may be worse, since the father will likely be
imprisoned himself, causing even more suffering for himself and his wife.
One might object that our intuitions in the first case are responding to
the legal injustice of the verdict. After all, involuntary vehicular manslaugh-
ter is not a capital crime. The DA would probably have to fudge the
evidence or mislead the jury to get the conviction. Perhaps we are feeling
more lenient towards John in the first case because of the legally unfounded
conviction for homicide. I agree that there seems to be a legal injustice in
the first case, but I do not think it can account for the difference in
intuitions. After all, the law is not being respected in the second scenario
either. Federal or State law does not allow for parents of victims to take the
law into their own hands. And if it is unjust in principle to issue a capital
sentence when there is no mens rea on the criminal’s part, then it should be
equally unjust for the father to carry out the killing himself.
A more promising strategy might appeal to our natural sympathy for the
father. We may feel that the father’s actions were understandable in a way
that the State’s was not. We may even believe that the father deserves his
vengeance, which then affects our judgment about what John deserves.
I agree that our judgments may be sensitive to our sympathy for the father,
but it is not clear that this is a distorting influence rather than an appropri-
ate one. My claim is that our desert judgments should be sensitive to our
sympathy for the particular victims of wrongdoing. Again, imagine the case

6
We may also imagine an analogous set of cases in which John receives what seems
like too lenient a sentence. In the first case, the judge gives him probation and no jail
time. In the second, the father finds John, sees that he feels tremendous guilt, that he is
horrified by what happened, and decides not to turn him in to the police. Again, I would
suggest that John seems more deserving of the lighter sentence in the second case than the
first.
252 Tamler Sommers

from John’s perspective. If I were John, I would feel enormous sympathy for
the child’s father, and my sympathy might lead me to think it is in large part
up to him as an individual to determine what I deserve. Unless we are already
committed to the impartial conception, I see no reason why we should regard
this sympathy as a distortion of John’s judgment. One might try to turn this
reply into another objection to the partial conception. Perhaps the difference
in judgments can be traced to what happens when we take the agent’s
perspective. But again, there is no reason to think that this is a distorting
influence—why shouldn’t we take the agent’s perspective into account? The
reply: “Because the agents’ subjective perspective is irrelevant to objective
judgments about what they deserve” begs the question. It is true that agents
are likely to be biased in their own favor, or at times feel excessive unwar-
ranted guilt. But this just means we should be careful about how we interpret
the agent’s perspective, not that we should ignore it entirely.7

4. THE COMPLEXITY OF DESERT

Two more cases may help to illustrate the relevance of the victims’ feelings
to the offender’s deservingness. Both are variations of the second case in
which the father shoots John. The variations focus on the father’s feelings
after the shooting.8

Third Scenario
The father recognizes that he acted in a moment of blind rage and
despair, and regrets the shooting immediately. He calls for an ambulance
but it arrives too late, John is dead. Although there is a small glimmer of
satisfaction that John will not get away unpunished, on balance he feels

7
As an anonymous referee notes, this case has an additional complication, namely that
the victim is dead and her wishes regarding the punishment are unknown. (Or she may be
too young to have well-considered feelings about John’s punishment.) This raises the
question of how desert might be affected if her parents or close relatives had different
wishes regarding the punishment—for example, if the mother felt more retributive and the
father more merciful. I agree that this is a difficult and important question, one that my
“partial” view of desert must address. For the purposes of this more programmatic paper,
however, it’s enough to point out that such factors (agreement or disagreement among the
relatives or those closely connected to the victim) actually matter, even if we cannot yet
specify how much. On the impartial conception of desert, these factors would be irrelevant.
8
These variations are inspired by Chandra Sripada and his comments on the Flickers
of Freedom blog. I am grateful for his contributions as well as many others on that post.
See: <http://agencyandresponsibility.typepad.com/flickers-of-freedom/2010/08/can-there-
be-partial-as-opposed-to-impartial-desert.html>.
 Partial Desert 253

worse than before. John did not mean to hurt his daughter, people drive
drunk all the time. As the father looks down at John’s lifeless body, the
senselessness of his revenge seems tangible. The only thing he seems to
have accomplished is the waste of another life. The father wishes desper-
ately that he had simply turned John into the police.
Judgment. Before learning about the father’s feelings, it seemed that John
deserved his fate (or at least that he was more deserving than in the first
scenario). But the father’s regret seems to undermine this intuition. Now
my reaction resembles when John was executed by the State—the punish-
ment seems excessive and undeserved.

Scenario 4
The father recognizes that he acted in a moment of blind rage. Still, upon
reflection, he feels that justice was done. He recognizes that this act will
not bring his daughter back, and that nothing will alleviate the suffering
he feels in her absence. But at least he has paid his debt to her and did not
allow the person who killed her to get away with it. He feels a strange
sense of peace, although his grief is just as acute. The father calls 911
right away and confesses to the crime. He accepts responsibility for his
action, and waits for the police to come and arrest him.
Judgment: Now my initial intuitions that John deserved his fate are, if
anything, even stronger than in the second case. The father has performed
the act, owns up to it, and even feels a small degree of satisfaction. He has
risked and sacrificed a great deal to bring about the punishment. He accepts
responsibility and will now go to prison. Perhaps if we were in the father’s
place, we would not feel or act this way. But there is a significant sense in
which it is not up to us, because we did not suffer from the offense. Again, if
I can inhabit John’s perspective (now from beyond the grave), I would
accept that I had received what was coming to me.
It may seem that I am edging (or hurtling) towards a reductio of my own
position. Judgments of John’s deservingness are supposed to be sensitive to
the father’s feelings about his act of revenge after performing it? To how
much the father risked and sacrificed to make it happen? To his willingness
to accept responsibility and punishment? In the remainder of this chapter
I hope to minimize the incredulity that accompanies such questions and
argue that the answer to all of them is a simple “yes.”

5. CARDINAL AND ORDINAL PROPORTIONALITY

I mentioned earlier that when considered abstractly, there is a good deal of

intuitive plausibility to the impartial conception of desert. This is due in
254 Tamler Sommers

large part to a long-standing (but insufficiently analyzed) idea that the

punishment should fit the crime. The proportionality principle is a hall-
mark of retributive or “just-desert” theories of criminal justice, and indeed
one of the main objections to rival utilitarian theories is that it would allow
for disproportionate punishments. The criminologist Andrew von Hirsch
calls the principle a “basic requirement of fairness” and describes it as
follows:
(1) [T]he principle of proportionality concerns how much punishment one
deserves; (2) deserved punishment should be commensurate to the degree of
blameworthiness of the conduct; and (3) blameworthiness depends both on the
harmfulness of the conduct and on the degree of culpability of the actor blame-
worthiness depends both on the harmfulness of the conduct and on the degree of
culpability of the actor.” (Von Hirsch 1978: 622)
But the proportionality principle has some well-known difficulties as well.
The primary problem concerns our inability to determine what kind of
punishment is commensurate with a given crime. What is the deserved
punishment for armed robbery? Ten years in prison? Fifteen years? Two
years and probation? Flogging? As Von Hirsch recognizes, desert theorists
have been notoriously unsuccessful at offering principled answers to these
questions. Von Hirsch presents the problem in the form of a dilemma. If
we assume that a particular crime warrants a specific quantum of punish-
ment, then we must presuppose “a heroic kind of intuitionism: that if one
only reflects enough, one will ‘see’ the deserved quanta of punishment for
various crimes.” (Von Hirsch 1992: 76) Unfortunately, no one seems to
have such illuminating intuitions and it is implausible to think that more
moral reflection will remedy this. The other option is to employ a “range-
only” view of desert according to which a criminal’s personal culpability
determines only the upper and lower limits of deserved punishment—a
view defended by Norval Morris (Morris 1982). Morris’s view allows for a
wide range of deserved punishments for a particular crime. It is only when
punishments fall outside this range that our intuitions give us a clear sense
that the punishment is undeserved. A heroic form of intutionism, then, is
not required for the range-only view.
According to Von Hirsch, however, Morris’s account is open to what he
calls “a fundamental objection”: it would allow two offenders who are
equally culpable to receive different punishments. And this is just the sort
of unfair outcome the proportionality principle is supposed to rule out.
Von Hirsch’s proposed solution to this dilemma employs a distinction
between cardinal and ordinal proportionality. Cardinal proportionality is
absolute: it “anchors” the severity of the punishment to the culpability of
the criminal (which includes the harmfulness of the crime). Cardinal
 Partial Desert 255

proportionality must remain “range only,” issuing upper and lower limits
where punishments would obviously be either too severe or too light. Ordinal
proportionality is relative. It has two aspects. The first is parity: like crimes
must be treated alike. If two criminals are equally culpable then they should
receive the same punishment. Second, punishments must be proportionate
relative to one another. If one crime is twice as serious as another, the
punishment should be twice as serious as well. The leeway that cardinal
proportionality allows in deciding the anchoring points of the scale explains
why we cannot perceive a single right or fitting penalty for a particular
criminal. Once the anchoring points of the scale have been fixed, however,
the more restrictive requirements of ordinal proportionality begin to apply.
In practical terms, the idea would be roughly as follows. We do not know
precisely what the punishment should be for, say, car theft. There are a
range of punishments that might be proportionate for this crime and in
absolute terms, proportionality just requires that we stay within this range.
We do know, however, that car theft is a less serious crime than armed
robbery. So the proportionality principle requires that (a) two equally
culpable car thieves receive the same punishment, and (b) armed robbers
receive a more severe punishment than car thieves. Von Hirsch offers
university grading practices as an analogy. The standards for “A” papers
and “B” papers and so forth are real but indeterminate (cardinal propor-
tionality) and may depend on nonmerit based factors about the university.
But once those standards are set, fairness requires that we give papers of
equal merit the same grade (ordinal proportionality).
The initial intuitive resistance to the idea of partial desert is rooted in our
commitment to ordinal proportionality. But the depth of this commitment
is open to question. In fact, outside of the context of criminal justice, it’s
not clear that we are committed to ordinal proportionality at all. Imagine
that a woman decides to leave her philandering husband and he replies:
“I understand you’re angry, but fairness requires that you don’t leave me.
Bill’s wife stayed with him and he’s had several more affairs than I have.”
Would the wife be moved by this consideration? Should she be? Everyday
life is filled with cases like this—acts of infidelity, betrayals of trust,
insulting or offensive remarks, and many others. We do not imagine that
there is a correct response or punishment, one that is tied only to the agent’s
personal culpability. Nor do we cry foul when people who are equally
culpable do not receive the same amount of blame or punishment. We
leave it up to the relevant parties to determine the right response, within
certain boundaries.
What does survive outside the context of criminal justice is our commit-
ment to cardinal proportionality. We maintain that there is a range of
appropriate blame or punishment responses and that responses outside of
256 Tamler Sommers

this range would be undeserved. Whether the betrayed spouse asks for a trial
separation, files for divorce, gives the partner another chance is largely up to
her. All of these are proportionate responses. But imprisoning the spouse or
killing him or even cutting off all access to the children would be dispropor-
tionate. The husband (as well as third parties) might legitimately complain
that the treatment is undeserved. It is significant that von Hirsch employs the
practice of essay grading to illustrate the importance of ordinal proportion-
ality. Certainly, it is a desert-based practice, but when a student writes a bad
paper, there is no victim.9 Offenses or crimes, by contrast, have identifiable
victims who have suffered at the hands of the offender. The presence of
victims is a morally relevant factor that affects our understanding of
proportionality and desert. Exactly how is the topic of the next section.

6. THE DESERT SPECTRUM

On my account, whenever a desert-based practice involves victims as well as

agents, personal culpability can only set a spectrum for how much blame
and punishment the offender deserves. For more precise determinations,
we must take facts about particular victims into account. The impartial
conception regards the feelings and desires of the offended parties to be
irrelevant to desert. By contrast, I see them as essential for determining an
appropriate response within the spectrum of deserved responses. How
narrow or broad is the desert spectrum set by personal culpability? This is
a tough question. It seems to vary depending on the kind of case that we are
judging. My hunch is that the spectrum is broadest in cases where the
offense is accidental or the result of negligence. This is why there is such a
wide range of deserved outcomes in the John cases. At one end of the
spectrum is the outcome of the second case—John being shot in the heart.
The other end might include outcomes that allow John to go free. Imagine
that the parents track John down and confront him. They express their
anger and grief and sense of loss. They see that John is consumed with
remorse and has been since the accident. The meeting gives the parents a
sense of peace and closure. They see that John would not do well in prison,
and they decide, after some tortured reflection, not to turn him into the
police. Many might call this a just outcome. Others might disagree and
claim that the parents were admirable in showing mercy, but that John was
clearly getting less than he deserved. There is room for reasonable disagree-
ment on this question. But compare this outcome to one in which John,

9
Aside from the instructors who have to comment on them.
 Partial Desert 257

through a plea bargaining agreement, is offered a suspended sentence—in

spite of the protests of the child’s parents. It seems undeniable that John is
more deserving of his freedom when the parents bestow it on him. I am not
claiming that the parents’ wishes and attitudes determine John’s deserving-
ness entirely. John could not deserve a month long cruise to the Galapagos
Islands no matter what the parents’ wanted. Nor could he deserve to be
tortured for twenty consecutive years. Again, we may reasonably disagree
about the end points of the spectrum. But the accidental nature of the
crime does seem to yield a strikingly broad range of deserved punishments.
By contrast, if John had deliberately killed the young child in cold blood,
the range might be significantly compressed.

7 . P H I L O S O PH I C A L B U S Y B O D I E S

The philosophical temperament may rebel against the looseness of this

account. To some, it will appear arbitrary, irrational, unsystematic, and
perhaps even antithetical to the project of providing a principled basis for
desert assignments. Certainly, this has been the reaction of the legal
academy to the rapidly growing victim’s right movement that involves
the victims in the sentencing process. Yet as I have noted, the demand for
impartiality and rational consistency is completely at odds with our every-
day practices, where victim involvement is expected and welcomed. Indeed,
the partial view is probably most intuitive when punishment is not at issue,
and the question concerns how much blame to assign to the offender.
Should Sarah blame her sister Emma for not remembering her birthday
because she was stressed about her job? That is up to Sarah—no theory
should tell her the right or rational response. Certainly, impartial consider-
ations should play some role in desert assignments, especially in more
severe cases of wrongdoing. But the partial account allows for this by
maintaining the commitment to cardinal or range-only proportionality.
Why should we aspire to more precision than this? The common assump-
tion that theories should dictate to people exactly how much they should
blame the person who wronged them deserves scrutiny.
The increasing popularity of restorative or restitutionary movements in
penal philosophy is relevant here. These movements have emerged out of
the increasing dissatisfaction with the depersonalized, process-oriented,
excessively rationalistic nature of our current criminal justice system.10

10
See e.g. Barnett (1977), Strang and Braithwaite (2000), Van Ness (1993), and
Zedner (1994).
258 Tamler Sommers

According to Lucia Zedner, the system “has transformed the drama and
emotion of social interaction and strife into technical categories which can
be subjected to the ordering practices of the criminal process.” (Zedner
1994: 231) Proponents of restorative justice argue that the blend of
retributive (desert-based) and utilitarian principles of our current system
is unjustly one-sided in its focus on the criminal. The victim is just a faceless
vessel for wrongdoing—like a poor essay that justifies a low grade. This has
the effect of alienating the victims and diminishing their self-respect even
further.
The criminologist Nils Christie has famously accused the criminal justice
system of “stealing conflicts” from their rightful owners. Lawyers, he
claims, are particular good at this form of larceny:
Lawyers are . . . trained into agreement on what is relevant in a case. But that means a
trained incapacity in letting the parties decide what they think is relevant. If the
offender is well educated, ought he then to suffer more, or maybe less, for his sins?
Or if he is black, or if he is young, or if the other party is an insurance company, or if
his wife has just left him, or if his factory will break down if he has to go to jail, or if
his daughter will lose her fiancé, or if he was drunk, or if he was sad, or if he was
mad? There is no end to it. And maybe there ought to be none. (Christie 1977:.8)
This extraordinary passage should trouble more than just lawyers. For if we
replace “lawyers” with “philosophers” or “desert theorists” in Christie’s
remarks, we expose some of the dubious assumptions and aspirations in our
current approach to moral desert. Our ever-more refined accounts of blame
and punishment—accounts that are supposed to apply across the board, no
matter what the relevant parties might think—may have the effect of
stealing conflicts from particular individuals. This whole approach exhibits
a “trained incapacity” to let individuals decide which factors are relevant
and how much. Zedner’s accusation seems apt as well. By focusing entirely
on impartial conditions of criminal culpability, our theories transform the
drama and emotion of social interaction and strife into technical (often
metaphysical) categories which can be subjected to a systematic ordering
process of desert attribution.11
Let me conclude this section by describing a real criminal case that
occurred recently in Grand Junction, Colorado, one that resembles
the hypothetical example I introduced earlier. A woman was driving in
the early morning, drunk and high on methamphetamines. On the way, she
had an accident and hit a sanitation worker. The worker’s legs were

11
One might interpret Strawson (1962) as making a similar point about the moral
responsibility skeptic who believes that “blame is metaphysical.” “The metaphysics,”
Strawson writes, “is in the eye of the metaphysician.” (p. 24).
 Partial Desert 259

shattered, causing him to endure ten surgeries. During her trial, the worker
testified and asked the judge to give the woman a lenient sentence. He
explained that he could relate to her predicament, that he had been in a
dark place once, and he hoped she could eventually get to a better place.
The prosecutors and judge took his desires into account, dismissed the
most serious charges, and the woman received the minimum sentence
allowable—still over five years in jail.
Here we have a clear violation of ordinal proportionality. Many people
who are equally culpable in Colorado have been given much higher
sentences and many will again. But did the judge and prosecutor violate a
“basic requirement of fairness”? Is this case clearly unjust? The victim of the
offense is satisfied, more than he would be if his wishes had not been
considered. The woman is still receiving a punishment within a reasonable
cardinal range. Is it our business as philosophers to complain about the
verdict, those of us who have not suffered in any way from this crime?
Doing so, in my view, would make us “philosophical busybodies” sticking
our collective nose where it doesn’t belong.

8 . C O N C LU S I O N

For all the insights it provides, there is a crucial difference between the
restorative justice critique and my own. Proponents of restorative justice
regard it as an alternative to retributivist or desert-based approaches to
criminal justice. In my view, we should incorporate facts about individual
victims into the way we understand desert for wrongdoing, criminal and
noncriminal. My account, then, is not an alternative but rather a revised
version of desert theory or retributivism. In order to make a judgment
about what John deserves in our original case, we have to know more about
what John and the parents want and believe. For partial desert judgments, it
matters whether the parents are vindictive or forgiving. And it matters what
John himself feels when he looks into their eyes.
Defenders of our current criminal justice system like to think that as an
enlightened society, we have transcended the revenge feelings and practices
of our barbarous past and replaced them with “justice,” which is rational
and not subject to emotional bias. But the retributivist project in the West
has struggled to develop a coherent notion of “just-deserts” that does not
appeal in any way to our natural disposition for vengeance. In my view, the
fears of allowing emotions into the equation are way overblown. No one is
advocating for a return to the days of endless tribal warfare. There is a
middle ground, one that allows individual victims to influence our desert
judgments under certain defined parameters, but not to determine them.
260 Tamler Sommers

These remarks lead to what may be the strongest objection to my

argument. One might claim that I am conflating two distinct concepts:
(1) what the offender deserves and (2) the just outcome of the crime.
A critic might concede that to determine the just outcome, we must take
other factors besides the deservingness of the offender into account—the
costs of the punishment and perhaps even the victims’ interests and facts
about what they deserve.12 Our judgments about the John cases, then,
reflect our intuitions about the just outcome of the crime rather than
intuitions about what John deserves.13
In response, let me first distinguish consequentialist factors from other
justice-related factors independent of the just-deserts of the offender.
I agree that we can distinguish desert judgments from “all-things-con-
sidered” judgments about blame and punishment that take consequentialist
considerations into account. Recall, however, that the different desert
judgments in the John cases could not be traced to such considerations.
I agree as well that judgments about John’s moral responsibility for the
offense are distinct from judgments about what John deserves for having
performed it. Moral responsibility on this account is constituted by agent-
centered factors in a way that desert is not.14 The objection, then, must be
that we need to distinguish John’s deservingness from other justice-related
judgments about his case. Interpreted in this manner, however, the
objection just begs the question by assuming a conception of desert that
is tied only to the personal culpability of the offender. If we do not employ
this conception from the outset, then there is no reason to think the justice-
related judgments are distinct.
The partial conception has some significant advantages as well. Since it
lacks the commitment to generality and objective precision, it is less
vulnerable to the endless array of counterexamples and theoretical difficul-
ties that have plagued desert theory to date. The partial conception may
also offer a new way of addressing the “paradox” of moral luck that
I described at the outset of this chapter. The reason the drunk driver who

12
In penal philosophy, this type of account is known as the “hybrid view” developed
by Paul Robinson (1987) among others.
13
Once again, this objection is inspired by comments on my Flickers of Freedom
post. A related objection, raised by an anonymous referee, is that I am conflating the
notions of what John deserves and those of what “serves him right.” In response, let me
first say that I’m not sure there is a substantive difference between the notions of desert
and those like “serves him right” and “had it coming to him”—although I recognize that
many or most philosophers will disagree with me on this point. Second, even if there is
an important difference between those notions, it’s not clear that the distinctions explain
our different intuitions regarding desert in the John cases.
14
Thanks to Tim Scanlon and Sarah Buss for convincing me on this point.
 Partial Desert 261

has an accident involving a pedestrian deserves more blame and punish-

ment than the drunk driver who makes it home without incident is that
there are victims in the former case. Although the drivers had the same
degree of control over the offense, desert judgments must take the harm
and interests of victims into account as well. Since there are no victims for
the second driver, he deserves significantly less blame and punishment. One
can apply a similar strategy to other kinds of moral luck, including the most
pervasive—constitutive luck. If our method for settling on a desert concept
is reflective equilibrium, the ability of the partial account to address a
problem as ancient and intractable as this one should count as a consider-
able virtue in its favor.15

REFERENCES

Barnett, Randy E. (1977). “Restitution: A New Paradigm of Criminal Justice.”

Ethics 87.4 : 279.
Christie, Nils (1977). “Conflict as Property.” British Journal of Criminology 17.1:
1–15.
Fischer, John Martin, and Mark Ravizza (1998). Responsibility and Control:
A Theory of Moral Responsibility. (Cambridge: Cambridge University Press).
Kane, Robert (1996). The Significance of Free Will. (New York: Oxford University
Press).
Morris, Norval (1982). Madness and the Criminal Law. (Chicago: University of
Chicago Press).
Nagel, Thomas (1979). Mortal Questions. (Cambridge: Cambridge University
Press).
Robinson, Paul (1987). “Hybrid Principles for the Distribution of Criminal
Sanctions.” Northwestern University Law Review 82, 19–42.
Scanlon, Thomas (2008). Moral Dimensions: Permissibility, Meaning, Blame.
(Cambridge, MA: Belknap, Harvard University Press).
Strang, Heather, and John Braithwaite (2000). Restorative Justice: Philosophy to
Practice. (Aldershot,: Ashgate).
Strawson, P. F. (1962). “Freedom and Resentment.” Proceedings of the British
Academy 48 : 1–25.
Van Inwagen, Peter (1983). An Essay on Free Will. (Oxford: Clarendon Press).
Van Ness, Daniel W. (1993). “New Wine and Old Wineskins: Four Challenges of
Restorative Justice.” Criminal Law Forum 4.2: 251–76.

15
Thanks to Sarah Buss, Dave Shoemaker, Chandra Sripada, two anonymous refer-
ees, and the commentators at the Flickers of Freedom blog for valuable comments. This
project has benefited immensely from the discussion at the 2011 New Orleans Work-
shop on Agency and Responsibility organized by Dave Shoemaker.
262 Tamler Sommers

Von Hirsch, Andrew (1978). “Proportionality and Desert: Reply to Bedau.” Journal
and Philosophy 75.11: 622–4.
—— (1992). “Proportionality in the Philosophy of Punishment.” Crime and Justice
16: 55–98.
—— (1993). Censure and Sanctions. (Oxford: Clarendon Press).
Wallace, R. Jay (1994). Responsibility and the Moral Sentiments. (Cambridge, MA:
Harvard University Press).
Williams, Bernard Arthur Owen (1981). Moral Luck: Philosophical Papers, 1973–1980.
(Cambridge Cambridge University Press).
Zedner, Lucia (1994). “Reparation and Retribution.” Modern Law Review 57. 2:
228–50.
 11
Values, Sanity, and Responsibility*
Heidi L. Maibom

1. SANITY AND MORAL INTRANSIGENCE

According to one influential view of responsibility, due to Harry Frankfurt

(2003), what is central to responsibility is a person’s ability to evaluate her
own will and identify with some volitions and distance herself from others.
If she is free to have the will she wants, she is responsible. Whether this
lucky situation is the result of someone else’s will is beside the point.
Someone who is free to do what she wants and free to want what she
wants to want has “all the freedom it is possible to desire or to conceive”
(2003: 333). The view has been criticized for not accommodating our
intuitions that an insane person is neither legally nor morally responsible.
For Susan Wolf, there is one more freedom required for responsibility. For
us to be responsible, we must be able to:
(a) evaluate ourselves sensibly and accurately, and
(b) transform ourselves insofar as our evaluation tells us to do so. (385)
This is the freedom of sanity.1
Frankfurt was interested in cases like addictions and compulsions, where
having the right kind of second-order volitions might absolve one from
moral blame, even if one is not free to have the will one wants. We are
inclined not to hold an addict responsible for taking drugs, because we
think he cannot help himself. Nevertheless, though he may have little
choice in the matter of wanting to take the drug of his addiction, he can
at least resist. He can want to not want to take the drug. We can distinguish

* Thanks to audiences at the first New Orleans Workshop on Responsibility, and to

Vanessa Davis, Annie Larivée, Melanie Mortensen, Darien Shanske, David Shoemaker,
and two anonymous referees for helpful comments and suggestions.
1
I focus on the critique of Frankfurt, though Wolf ’s critique affects all so-called
Deep Self Views, including those of Gary Watson and Charles Taylor (Watson 1975,
Taylor 1976).
264 Heidi L. Maibom

between addicts on the basis of their so-called deep selves; one might be a
willing addict, the other unwilling. Wolf thinks Frankfurt’s view cannot
accommodate our intuitions about the insane because they have the will
they want to have, yet we hesitate to hold them (fully) morally responsible.
The trouble is that the insane are not able to know “the difference between
right and wrong” (382) and they cannot but possess “values that are
unavoidably mistaken” (383).
Wolf uses the example of the son of an evil dictator to illustrate her
point. JoJo’s father habitually sent people to prison, torture chamber, or
death on a whim. Having been under his father’s educational regimen, JoJo
grows up performing just the kinds of actions that his father did. He does so
willingly; just like a willing addict satisfies his addiction. He has the first-
order volitions that he wants to have and the freedom to act on them.
Contrary to what Frankfurt claimed, this is not all the freedom anyone
could want, Wolf argues. For it seems unavoidable that JoJo should have
the values—second-order volitions—that he does, and therefore it is not
right to blame him:
These are people, we imagine, who falsely believe that the ways in which they are
acting are morally acceptable, and so, we may assume, their behavior is expressive of
or at least in accordance with these agents’ deep selves. But their false beliefs in the
moral permissibility of their actions and the false values from which these beliefs
derived may have been inevitable, given the social circumstances in which they
developed. If we think that the agents could not help but be mistaken about their
values we do not blame them for the actions those values inspired. (382)
Wolf ’s point is not that we must be the authors of our second-order
volitions. The inevitability of which she speaks does not refer to the origin
of such values, but to their continued existence as values that characterize an
agent’s deep self. What is central to sanity is the ability to transform one’s
values, and thereby one’s deeper self. Seriously wrongheaded values trap a
subject, in a manner of speaking, by making her incapable of evaluating
them realistically. Since the insane are stuck with their second-order voli-
tions, but we are not, they are not responsible whereas we are. That JoJo
espouses the values that he does shows that he is insane:
Sanity, remember, involves the ability to know the difference between right and
wrong, and a person who, even on reflection, cannot see that having someone tortured
because he failed to salute you is wrong plainly lacks the requisite ability. (382)
This suggests that in order to possess the abilities required for sanity and
responsibility mentioned above (a and b), we must first have:
(c) the ability cognitively and normatively to recognize and appreciate the
world for what it is. (383)
 Values, Sanity, and Responsibility 265

At least three conditions violate (c) according to Wolf: mental illness,

seriously deviant upbringings, and membership of groups that espouse
significantly different moral values than ours. In essence, any person who
has values that contrast sufficiently with ours, and who holds on to such
values doggedly, suffers from essentially the same deficit as the insane,
barring laziness, distraction, and greed. The consequence is that people
with significantly different moral values are not responsible, or at least not
fully responsible. This includes male chauvinists of our father’s generation,
Ancient Greek slave-owners, and Nazis. In effect, we should get used to the
idea that those who do great evil, by our standards, are not sane and
therefore not responsible (386–7).
The concerns raised here are typically associated with what is sometimes
known as the epistemic condition on responsibility, i.e. the condition that
an agent must be aware of what she is doing. And there certainly seems to
be many ways in which an agent may not be entirely aware of what she is
doing. What Wolf seems to do is to lump many such cases together under
the heading “insanity.” There are many reasons to think this is infelicitous.
Terminology aside, there are deeper issues with characterizing people who
are insane and agents who are not, but who hold morally intransigent and
divergent views, as suffering from the same epistemic deficit. Insanity and
what I shall call moral intransigence are psychologically, legally, and
morally quite distinct. On the one hand, people who are insane typically
suffer from hallucinations and delusions, but do not possess values that
differ significantly from our own. On the other, in many, if not most, cases
of divergent and intransigent moral values, there is an inferential route
from a person’s values and beliefs, or from values and beliefs accessible to
her, to the recognition that what they are doing is wrong. Such cases are
therefore not generally ones in which the agent is unable to see that what
they are doing is wrong. This is presumably why we can be held responsible
although we no doubt also have a number of wrongheaded values. But if we
can be held morally responsible, so can people from different cultures
(though not, perhaps, from all other cultures). Conversely, if they are not
responsible, so neither are we. The way to avoid this consequence is to
maintain that only those who do wrong while fully knowing it are blame-
worthy. This, however, deconstructs our notion of blameworthiness in an
unacceptable way.
In what follows, I shall use criminal responsibility as a guide to moral
responsibility. This is not because I think the former is prior to, or more
important than, the latter. Rather, I think the conditions under which we
hold people criminally responsible reveal some of our deep intuitions about
when someone is morally responsible. This is particularly true when it
comes to the question of sanity.
266 Heidi L. Maibom

2 . I N SA N I T Y I N L AW

The widely used McNaughtan Rule stipulates that in order to be insane the
defendant must suffer from a mental illness which is causally implicated in
the criminal action that the defendant is on trial for: (Moran 1981: 169)2
To establish a defence on the ground of insanity it must be clearly proved that, at the
time of committing the act, the party accused was labouring under such a defect of
reason from disease of the mind, as not to know the nature and quality of the act he
was doing, or if he did know it, that he did not know that what he was doing was
wrong.
If the accused was conscious that the act was one which he ought not to do, and if
that act was at the same time contrary to the law of the land, he is punishable.
Contrary to what is sometimes thought, mental illness is not synonymous
with insanity. A mental illness does not simply deprive a person of responsi-
bility for her actions, but it can sometimes overwhelm her, and when it
does, she cannot be held responsible for the actions performed under its
influence. The law acknowledges that mental illness can affect everyone
alike, and though it may affect a person profoundly, it rarely blots out her
preexisting character or moral sensibilities. As such, a crime committed by a
person suffering from a mental illness may not, in any interesting sense, be
the result of that person’s illness, but of their deficient moral character.
A person can be both bad and mad.
Most of those judged to be not guilty by reason of insanity3 suffer from
paranoia.4 The majority of successful insanity pleas in Canada (Rice and

2
If not directly used, the Rule often serves as the basis of acts concerning the
responsibility of persons judged to be insane. In Canada, the Criminal Code section 16
specifies that “(1) No person is criminally responsible for an act committed or an
omission made while suffering from a mental disorder that rendered the person incap-
able of appreciating the nature and quality of the act or omission or of knowing that it
was wrong.” In Britain, the corresponding act concerns diminished responsibility.
According to section 52 of the Coroners and Justice Act 2009 (the Act), “a person
who kills or is a party to the killing of another is not to be convicted of murder if they
were suffering from an abnormality of mental functioning which: (a) arose from a
recognised medical condition, (b) substantially impaired their ability to do one or
more of the following: understand the nature of their contact, form a rational judgement,
or exercise self control, and (c) provides an explanation for [their] acts and omissions in
doing or being a party to the killing. An abnormality of mental functioning provides an
explanation for the conduct if it causes, or is a significant contributory factor in causing,
the defendant to carry out that conduct.”
3
In Canada, “not criminally responsible on account of mental disorder” has replaced
“not guilty by reason of insanity” (Criminal Code section 672.34).
4
The DSM-IV lists three different types: paranoid personality disorder, paranoid
subtype of schizophrenia, and the persecutory type of delusional disorder.
 Values, Sanity, and Responsibility 267

Harris 1990), New York (Steadman et al. 1983), Colorado (Jeffrey et al.
1988), and Oregon (Rogers et al. 1984) involve defendants who have been
diagnosed with psychosis, mostly commonly schizophrenia (often 80
percent or higher). Psychosis is characterized by delusions, hallucinations,
and disorganized thought and speech. Consequently, people deemed insane
are usually disturbed individuals who have profound problems with reality
generally, not just moral reality if such a thing exists. For instance, Daniel
McNaughtan, who gave the name to the famous insanity rule, killed
Edward Drummond because he mistook him for the British Prime
Minister Robert Peel, whose secretary he was. He believed himself to be
persecuted by Peel and his political party to such an extent that his life was
becoming unbearable. He also thought they were planning his demise.
Andrea Yates was suffering from severe postpartum psychosis and believed
that, were her children to live, they would eventually face eternal
damnation. To kill them was therefore an act of mercy in her mind. So
she drowned her five children in a bathtub, one after the other. James
Hadfield thought he was something of a new Messiah whose lawful
execution or death would bring about the second coming of Christ,
thereby saving humanity from much unnecessary suffering. He shot at
King George III—missing by a wide margin—because he knew that
attempted regicide was punishable by death.5 Henry Maudsley (1898)
tells the story of a father who believed that he was fighting a fierce snake
only to find that he had killed his infant son.
In typical successful insanity pleas it is the subjects’ delusions that cause
the problem, not their moral compass. Yates and Hadfield clearly thought
they were acting for the greater good, and at the expense of their own well-
being. Had Yates and Hadfield been right about their mistaken beliefs,
their actions would not have been culpable. The so-called As-if rule stipu-
lates that someone who acts on beliefs about the world that are due to
delusions or hallucinations has an excuse on the condition that had their
beliefs been true, their action would have been morally acceptable (Reznik
1997). The complimentary If-only rule specifies that in the absence of
mental disorder, the subject would not have performed the action in
question. And, in fact, Yates or Hadfield would probably not have done
what they did had they not been in the grip of strong delusional beliefs. The
As-if and the Only-if rules typically characterize successful insanity pleas. In
other words, it is not usually the case that a person judged to be not guilty
by reason of insanity is someone whose values are foreign to us or are such
that we cannot identify with them, in some sense.

5
He could not kill himself as he believed suicide was not permitted by God.
268 Heidi L. Maibom

Some, like Lawrie Reznik (1997), think that the notion of a good
character is central to insanity pleas. For the defendant to have an excuse,
in cases of insanity, he must be shown to have acted out of character, in some
sense. His actions may be the result of his mental illness, for instance.
Character, for Reznik, is similar to the deeper self that Frankfurt refers to.6
We are responsible for actions performed “in character” even if we are not
responsible for our character or, to be more precise, our moral character.7
If, however, “a person changes from a good character to an evil one,
commits an offense, then changes back again, the good character has an
excuse” (227). Mental illness and brain injury can result in such character
change. If the disorder is irreversible—a progressive brain disease, for
instance—the individual cannot be held responsible, and if the disorder is
reversible, we cannot blame the good character should we succeed in
restoring it. Lack of prior convictions, due regard for the well-being of
others, etc., are all signs of a good character.
The Yates case helps illustrate the importance of character. Yates was
diagnosed with postpartum psychosis. As already mentioned, psychotic
subjects appear to inhabit, at least part of the time, a different reality
from the rest of us, and usually experience great difficulties functioning
normally. Parents suffering from postpartum depression or psychosis often
fantasize about killing their infants. Prior to the murders, Yates had been
catatonic and unable to breastfeed her youngest child properly. She had a
history of postpartum depression and attempted suicides, but no criminal
record. Yates’s husband testified to her declining condition and insisted
that she had undergone a personality change. She had expressed doubts
about her ability to be a good mother, but ultimately claimed to have
drowned her children to save them from an eternity of suffering in Hell.
Her history of relatively blameless behavior suggests that an otherwise
morally upstanding citizen—to put it somewhat primly—had been over-
come by mental illness.
People judged not guilty by reason of insanity are not always so benevo-
lent, of course. Where Yates and Hadfield thought of themselves as min-
imizing harm by their actions, subjects of command hallucinations believe
they are simply obeying orders. Following orders, however, is usually not

6
Reznik explicitly rejects Frankfurt’s notion of personhood. This, however, seems
based on a misunderstanding of Frankfurt’s philosophical position.
7
A person’s moral character consists in “the set of dispositions that explain his ethical
beliefs, his moral sentiments, and ethical conduct.” (Reznik, 223) Talking about moral
character is required at least for legal purposes. If someone has a temporary change in
character, but it does not affect their moral outlook or their moral dispositions, it cannot
excuse any crime that they may commit in their altered state.
 Values, Sanity, and Responsibility 269

an excusing condition in civilian life when the action is illegal. Yet, people
who kill because they hallucinate God commanding them to do so are
sometimes deemed insane even though their actions would not have been
legally or morally acceptable had their beliefs been true (at least in a secular
society). Kim John, together with Francis Philip, bludgeoned a Catholic
nun to death and set a priest on fire at the Cathedral of the Immaculate
Conception in St Lucia. He was subsequently diagnosed with delusional
disorder, paranoid type.
Kim John claims to have been under a divine injunction to target the
Catholic Church, which he also believed was persecuting him. He blamed
the Church for his own troubles and for “child abuse, molesting, stealing
the tithes and offerings, buying off the land, poisoning the water and food,
robbing and taking funds” (Larbey 2007). Unlike McNaughten, Kim did
not think that his life was in danger. He did not, therefore, have the excuse
that had his beliefs been true, his act would have been reasonable.
Nevertheless, his delusions were judged to be of such a nature as to make
him unfit for punishment (Francis Philip & Kim John v. The Queen). He
claims to have had visions from an early age of “[s]elling of human beings,
children like cargo, women raped and murdered, tortured, castrations. The
falling of the wicked in flames, seeing them burn up, skeletons, ashes,
smoke, fire, weeping and wailing” (Larbey 2007). He also thought he
could do magic. The pervasiveness of Kim’s delusions made the Appellate
Court think it useless to try to assess his moral character, which,
presumably, they thought obliterated by, or buried under, his mental
illness.
It is notable that not all command hallucinations excuse criminal behav-
ior. Ronald Lafferty claimed to have had a divine revelation that he was to
“remove” certain individuals that stood in God’s way. Conveniently, they
were all people he blamed for encouraging his wife to leave him with their
children. He talked his brother Dan into killing their other brother’s wife
Brenda and their 15-month-old daughter. Ron received the death sentence
in 1985. In the much publicized 1996 retrial, his lawyer entered an insanity
plea, arguing that Ron’s religious ideas and “revelations” were signs of a
delusional mind. The prosecution countered that irrational and strange
ideas—including the idea that one receives revelations from God—are a
stable of religion. But we cannot simply assume that all religious people are
insane. That would render much of the population and the political
establishment in large parts of the world insane.8 Instead, witness for the

8
The DSM-IV specifies that “culturally sanctioned” responses (p. xxxi) or political,
religious, or sexual deviant behavior are not signs of mental illness unless they are
symptoms of a dysfunction of the individual associated with present distress, disability,
270 Heidi L. Maibom

prosecution Dr Stephen Golding maintained that Ron was a zealot, and

zealots “are [not] mentally ill, per se.” “A zealot is simply someone who has
an extreme, fervently held belief ” and is willing to go “to great lengths to
impose those beliefs, act on those beliefs.” (Krakauer 2003: 305).
Ron’s revelations took place within a community of fellow Mormons
who all thought they were receiving revelations from God and were in the
habit of discussing them. The communal nature of the revelations and
other religious beliefs suggest that Ron was not mentally ill. Delusions
caused by mental illness are typically delusions of a single individual, not
ideas encouraged and nurtured by a group of people. The religious nature
of Ron’s ideas clearly did not explain his actions either because the group as
a whole refused to act on Ron’s “removal” revelation. Several individuals
were quite disturbed by the violent nature of his revelations. This suggests
that the violence of the revelations were a reflection of Ron’s fantasies and
urges. In short, they reflected his bad character. Did he even believe that
they were revelations, one might ask, given how convenient they were given
his aims and goals? Ron certainly had odd beliefs, but were they delusional?
He stopped working and paying taxes because he thought he should not
have to do either; he mistreated his wife and children for not obeying him
unquestioningly, and so on. This is more suggestive of an immature and
entitled attitude than of a subject suffering from delusions.
Was Ron mentally ill? Dr Golding suggested that perhaps he suffered
from Narcissistic Personality Disorder. However, as we have seen, mental
illness is only a necessary, not a sufficient, condition for insanity. What is
crucial is that it be plausible the criminal act was the result of the illness.
Dr Golding argued that there was no reason to think that Ron’s narcissism
caused him to believe that he was receiving revelations from God or to
orchestrate the killings of others. Few narcissists engage in the sort of
harmful activities that Ron did, even if they are grandiose, self-centered,
and lacking in empathy for others. By contrast to Ron, Kim John was
diagnosed with delusional disorder, persecutory type, which is associated
with anger and violence (American Psychiatric Association 2000). A person
who suffers from persecutory delusions believes “he or she is being
tormented, followed, tricked, spied on, or ridiculed.” (2000: 299).9 This

or an increased risk of suffering “death, pain, disability, or an important loss of freedom”

(p. xxxi).
9
Command hallucinations are not always violent. When they are, people are less
likely to comply with them (My Lee et al. 2004). Those that do are more than twice as
likely as those who do not to have a recent history of violence. Furthermore, a subject’s
beliefs about the effects of the action, how socially acceptable it is, and so on, affect
whether or not she will comply with the command (Beck-Sander, Birchwood, and
 Values, Sanity, and Responsibility 271

makes it rather likely that the disorder played an important role in Kim’s
beliefs about the Church’s responsibility in his misfortunes. Compare such
delusions with Ron Lafferty again. Ron blamed a group of people for his
wife leaving him and he was, in part, right. Clearly he failed to take into
account his own role in the split, but these individuals did encourage her to
leave and take the kids. In short, Ron had the quite reasonable belief that a
number of individuals encouraged his wife to abandon him. His ensuing
belief that they should be killed was therefore not the result of delusions
that these individuals were (partly) responsible for his wife’s departure.
Quite likely, it was the result of his desire for revenge.

3. LESSONS OF INSANITY

We are now in a better position to consider whether those who have

significantly different values suffer from the same epistemic deficit as
those who are insane. Wolf appears to have used JoJo as an example of
how moral intransigence collapses into insanity. So let us begin by thinking
a bit more about that example. JoJo has perpetrated a great number of
crimes over many years, including murder and torture. He has done so
freely, with a callous disregard for the well-being of others, and without
remorse. Our main evidence for him having a mental disorder is his twisted
values. His is not a generally decent moral character overwhelmed by
disease. There is little question of his actions being compelled. Could he
have evaluated his twisted values objectively and, finding them lacking,
would he have been able to change them? Well, imagine JoJo becoming a
subject and someone else the dictator ruling as he and his father before him
had ruled. Would he come to regard torturing or killing someone on a
whim as wrong? My hunch is that, yes, he would be able to see that such
actions were wrong, and he would cease to embrace such values once he was
at the wrong end of them, as it were. So it certainly seems as if JoJo has the
ability to evaluate and change his values.10
Would JoJo’s actions have been acceptable had his beliefs been true (the As-if
rule) and his actions the direct result of his illness (the Only-if rule)? Would
torturing and killing people on a whim have been morally permi-
ssible if torturing and killing people on a whim were morally permissible?

Chadwick 1997). Having command hallucinations to harm someone are most likely not
sufficient to cause people to comply in the absence of other deluded beliefs (McNiel,
Eisner, and Binder 2000). Kim John, however, seemed to have no shortage of delusions.
10
According to David Faraci and David Shoemaker (2010), most ordinary subjects
also judge JoJo to be blameworthy for his actions.
272 Heidi L. Maibom

Obviously. But this cannot be the right reading of the As-if rule since
any conviction could pass the rule. Rather, it is its acceptability within
our moral and legal system that is at issue, and within those torturing or
killing people on a whim is not permissible. This means that the posses-
sion of divergent moral or legal values is not typically an excusing condi-
tion even if such values are the result of mental disorder. As mentioned
above, command hallucinations are an exception. Here the defendant’s
actions may not have been justified had their beliefs been true. In these
cases, it is usually the pervasiveness of delusions that excuses, as in the case
of Kim John. As in other cases of insanity, it is not the subject’s moral
values that are the primary issue. Indeed, if the Church were responsible
for the actions John thought it was, it would be culpable. This fact
shows just how much overlap there is between John’s moral values and
ours. The main problem with JoJo, however, is his values.
This brings us to the If-only rule, which JoJo also does not satisfy. He
does not suffer from any currently recognized mental illness. Someone
might argue that had JoJo been raised otherwise, he would not have
performed the actions that he did, wherefore we cannot hold him respon-
sible. It is instructive to note that this is not Wolf ’s argument. How people
come to have the values they do is irrelevant. What matters are the values
with which they end up. If we are to believe Wolf, most values that diverge
from ours in significant respects render the agents who possess them unable
to evaluate themselves “sensibly and accurately” and change themselves
accordingly.
JoJo is, of course, just one example. The fact that JoJo is not unable,
rather than merely unwilling, to change his values does not show that there
are not people who are incapable of comprehending that certain types of
actions are morally or legally impermissible, such as torture, murder, rape,
or persecution of others. What would it take, one wonders, to be incapable
of such a thing? Two interpretations present themselves. Either the agent’s
ability to evaluate and change values is directly affected, or it is intact, but
her environment does not yield opportunities for such evaluation and for
changing her values to be more aligned with values that we now regard to be
right. In the latter case, in particular, we need to distinguish those who have
sufficient access to values and information for us to say that they hold on to
their values from those whose environment and values are such that we
cannot expect them to be able to reach values similar to those we now
adopt. But first we must explore what knowing that what one is doing is
wrong amounts to.
The law typically does not care whether you are ignorant of the law or
whether you agree with it. Nevertheless, some readings of the McNaughtan
Rule demand that the defendant know that what he was doing was
 Values, Sanity, and Responsibility 273

morally wrong. In other words, the person must understand that the
action committed was not merely illegal (malum prohibitum), but
also bad in itself (malum in se). It is not just the ability to appreciate
that, say, killing in general is malum in se that can be required for legal
responsibility, for most people judged not guilty by reason of insanity
understand that. What must be meant is that, at the time of the crime, the
defendant was able to understand that committing this action was morally
impermissible.
Being able to understand that an action is morally wrong cannot
translate to truly believing, at the time, that what you are doing is wrong.
For, according to Roy Baumeister, “[m]ost people who perpetrate evil do
not see what they are doing as being evil.” (1997: 1).11 Our assessment of a
criminal act is usually different from that of the perpetrator. Perpetrators tend
to trivialize their transgressions and think that, given the circumstances, they
could not have helped doing what they did. By contrast, most victims magnify
the wrongness of the event and the wickedness of perpetrator’s intentions
(Baumeister 1997). In the laboratory, too, people minimize their own wrong
doings while they magnify wrongs perpetrated against them (Baumeister,
Stillwell, and Wotman 1990). Not surprisingly, therefore, there is a good
correlation between a person’s access to justification for violence and the
degree to which they engage in it (Berkowitz and Powers 1979; Calvete
2008, Schwartz, O’Leary, and Kendziora 1997, Zelli et al. 1999).
Crimes like violent assault or homicide are often the result of an escalated
process of reciprocal provocation (Berkowitz 1978, Luckenbill 1977,
Berkowitz 1978). It is not uncommon for the victim to throw the first
punch in such confrontations (Wolfgang 1958). Consequently, the person
who ends up killing another often sees himself as a victim. He was, after all
provoked, threatened, or attacked, and his action was justified under the
circumstances (Katz 1988). In such escalations, one person sees his act of
retaliation as equitable whereas the other usually regards it as an unjustified
escalation. As a result, both end up feeling victimized (Stillwell, Baumeister,
and Del Priore 2008). Clearly, therefore, many convicted criminals did not
see their actions as impermissible at the time.
The above makes it clear that we cannot simply demand that the subject
understands that there is some general moral injunction against the action
at hand. The average killer, like the average insane person, does not disagree
with the injunction against killing. Typically, he thinks that this instance of

11
Philosophers will be familiar with the idea from The Meno where Socrates argues
that nobody who “recognizes evils for what they are” desires them (77C).
274 Heidi L. Maibom

killing is not wrong or at least that it is not that wrong. The circumstances
were sufficiently mitigating—he insulted me, she was cheating on me, he
attacked me first, she is a blot on my honor—that I should not be held
responsible for murder (but perhaps for some lesser crime). Though it is
quite possible that the criminal is right in some of these cases, in most cases
the court judges that he should have realized that the situation did not call
for that degree of violence. The standard here is usually that of a “reason-
able man.” So perhaps this is a better way of thinking about the require-
ment that one understands that one’s action is malum in se: if there is an
inferential link from beliefs and values that the person has at the time of
the crime to the recognition that the action is morally impermissible, then
the person is responsible and culpable should they nevertheless perform the
action (cf. Williams 1981).
The condition is too weak, however. A person might defend herself by
pointing out that she did not embrace the relevant values at the time. If she
does not believe killing, or this form of killing, to be wrong there is no
inferential route to her appreciating that this killing was wrong. But clearly
we would still hold her responsible. Being on the fence about, or disagree-
ing with, a moral prohibition does not seem to excuse. Consider honor
killing in the West.
People who commit honor killings believe that killing to preserve or
reinstate honor is morally permissible by contrast to other forms of
killing, which they agree are impermissible. This is causing something
of a stir in North America. In Canada, Mohammad Shafia, his second
wife, and their son were recently found guilty of killing Shafia’s first wife
and the couple’s three daughters. Apparently, the three daughters
objected to wearing the hijab, wanted to spend time with their friends
after school, have boyfriends, and so on. The story is very similar to
another recent honor killing in Toronto where Aqsa Parvez was murdered
by her father, Muhammad, and her brother, Waqas, for bringing “insult”
to the family for similar reasons (CBC News, June 15, 2010). The two
Waqas’s received life sentences.
Both Mohammad Shafia and Muhammad Pavez and their families are
recent immigrants from, respectively, Afghanistan and Pakistan, where
values differ significantly from those of the West. In both countries
honor killings are disturbingly common, and though prohibited by law,
are rarely punished. Pakistan has a long tradition of karo-kari, the practice
of killing women for their perceived immoral behavior, e.g. adultery,
having been raped, refusing an arranged marriage, or wanting a divorce.
Often, girls and women are killed on mere suspicion of some sort of
affiliation with a male outside the family. Honor killers doubtlessly regard
their actions as justified. The Shafias and Pavezs evidently acted in
 Values, Sanity, and Responsibility 275

accordance with their native morality.12 They do not accept that killing to
save honor is wrong.
In the case of honor killings, it is fairly clear that failure to accept or
adopt a moral or legal norm does not deprive someone of responsibility.
Honor killings are cases of moral intransigence, not of moral inability.
They are no different from typical homicides. Neither killer believes they
are truly culpable. However, this lack of recognition is not usually thought
to be mitigating. By contrast, it is often thought to make the offender more
culpable. For their action is not the result of a mistake or an accident, it is
the result of adopting values that we find despicable. It is their beliefs about
the moral value of Jews, in particular, that make Nazis so despised. It is the
belief that women are chattel and have no value other than to serve men and
bear male children that makes many of the practices in the Middle East so
horrific. The inferential link we require for responsibility, therefore, must
be to the current values of the person in question or to values that we judge
that she ought to have had under the circumstances.
If we assume that culture does not inhibit people’s ability to evaluate or
change their values—I shall argue for this shortly—then what is at issue
with honor killers is whether they hold on to their values or whether they do
not have access to relevant values and information that would enable them
to change them in the right way. Wolf suggests that if one is brought up in a
particular moral milieu, which allows, perhaps encourages, certain wrongs,
one is not really in a position to accurately evaluate or change those values.
It remains obscure, however, why we have the capacity to change our
wrongheaded values. My hunch is that Wolf thinks that there is a pretty
straightforward inferential route from our current moral values to future,
improved, ones. For instance, our indifference to the death and suffering of
nonhuman animals, particularly the ones that we eat, is quite plausibly
culpable. We have the capacity to arrive at valuing the life and well-being of
nonhuman animals because of values and beliefs that we either possess or
that are readily available in our environment. We believe that if an action or
policy creates unnecessary or avoidable suffering, then we have a prima facie
reason not to perform or institute it. We also know—or if we do not
actually know, we could easily come to know—that factory farming creates
a great amount of suffering, and that we do not need to consume as much
meat as we do for proper nutrition. From those beliefs, there is a relatively
straightforward inferential route to the belief that factory farming is wrong
and that we ought to oppose policies that permit it. As we shall see, this line

12
This is not to say that many people living in those countries do not regard honor
killings with horror, and would never engage in, nor condone such acts.
276 Heidi L. Maibom

of reasoning shows not only that we are responsible, but also that there
usually is little question of someone’s culture inhibiting her responsibility.
If we are responsible for our wrongdoings, so are the slaveholders of Ancient
Greece and the male chauvinists of our father’s generation. If they are not,
neither are we.
Michele Moody-Adams (1994) has argued that people in cultures that
permit, or encourage, practices that we condemn, such as slavery, are
typically exercising so-called affected ignorance. That is, they chose not to
question, seek information or otherwise know about these practices of
wrongdoing.13 I suspect that there is another form of affected ignorance
that derives from the degree of difficulty involved in endorsing values that
significantly diverge from the culture at large, not to mention advocating a
societal change of standards. It requires some imagination to envisage a
different moral order. It is, perhaps, an affected lack of imagination. Wolf
maintains that the ancient Greeks cannot be held responsible for their
attitudes towards slavery. The question, however, is whether there is a not
too onerous inferential route from beliefs and values that the ancients
possessed that leads to the recognition that slavery is morally wrong.
At the time of its practice, slavery was widely regarded as a terrible fate.
When Andromache bewails the death of Hector in The Illiad, she decries
the fate of the citizens of Troy: “all who will soon be carried off in the
hollow ships and I with them—And you, my child, will follow me to labor,
somewhere, at harsh, degrading, work, slaving under some heartless
master’s eye” (Book 24, line 860 ff.).14 In Xenophon’s Symposium,
Antisthenes suggests that enslaving others is a crime: “Want prompts a
thousand crimes, you must admit. Why do men steal? why break
burglariously into houses? why hale men and women captive and make
slaves of them? Is it not from want?” (Xenophon 2008a: }27) In Hellenica,
Xenophon not only talks about the great lengths that people will go to, to
avoid being “reduced to” slavery, but also recounts of the Spartan general
Callicratidas who refused to enslave the Methymnaeans because they were
fellow Hellenes (Xenophon 2008b). In the Politics, Aristotle refers to
people who “affirm that the rule of a master over slaves is contrary to
nature, and that the distinction between slave and freeman exists by law

13
Another feature of affected ignorance is the rephrasing of wrongs in relatively
inoffensive sounding language. For instance, the message on the Rwandan radio encour-
aging the Hutus to kill the Tutsis was “cut down the tall trees” (Dallaire 2004).
14
The sentiment appears to have been typical in all times where slavery was a
predictable result of conquest. Thus, in Beowulf “[a] Geat woman too sang out in
grief; with hair bound up, she unburdened herself of her worst fears, a wild litany of
nightmare and lament: her nation invaded, enemies on the rampage, bodies in piles,
slavery and abasement.” (line 3150 ff.)
 Values, Sanity, and Responsibility 277

only, and not by nature; and being an interference with nature is therefore
unjust.” (Politics, Book I, Part III) In Rhetorica ad Alexandrum,15 the Sophist
Alcidamas says “The deity gave liberty to all men and nature created no one a
slave” in reference to the Thebans freeing the Messenians, who had been
taken slaves by the Spartans (Garlan 1988: 125).
The passages above suggest that it was not unthinkable to the Ancient
Greeks that slavery was wrong. Clearly some people thought slavery was
unjust. Furthermore, the elements for a realization of the moral wrong-
ness of slavery were there. Every free person wanted to remain free and
regarded slavery as degrading and awful (1988). They recognized that
slaves were fellow human beings, that they were capable of suffering,
that suffering was morally relevant, that they themselves might be in
danger of enslavement by others, and so on. All the talk of the baseness
and stupidity of slaves that we also find in the extant literature seems
designed to protect an affected ignorance of the wrongness of the insti-
tution. After all, as long as you are not a slave but others are your slaves, it is
to your advantage to embrace a norm that permits slavery. If we were to
make a comparison to current culture, we might point out that we, too,
are in the possession of everything we need to know to recognize that
factory farming, for instance, is an immoral practice. Far from being
incapable of recognizing that this is so, we are choosing to ignore it
because it is difficult to imagine not eating meat or eating meat much
more rarely, to imagine a societal change, to eat differently from every-
body else, and so on, not to mention the economic difficulties that would
be involved. Despite these difficulties, however, we are hardly unable to
recognize the wrongness of the practice.
There are other features of cultural value systems that speak against them
having the power to deprive agents of responsibility. It is no secret that to
talk of a value system of a particular culture is something of an idealization.
Though there may be agreement about the most serious forms of transgres-
sions, a society is not characterized by complete agreement about moral
norms. Furthermore, such values are subject to change. And, as a matter of
historical fact, values do change. Such change may be a relatively fluid affair,
or be more cataclysmic. Now, people instantiate or embody norms (Moody-
Adams 1994). For change in values to be possible, people must be able to
change their norms: evaluate them, adopt them, defend them, relinquish
them, overthrow them, and so on. The point is obvious on reflection. If
it were true that people brought up in societies where slavery was

15
Written around the same time as Rhetoric, it was traditionally attributed to
Aristotle, but might have been written by Anaximander.
278 Heidi L. Maibom

sanctioned by law and common morality were thereby incapable of

thinking it was wrong, then we should expect no moral change. But such
change did happen; not in Ancient Greece, but in Europe and the
Americas. Therefore, adoption of seriously wrongheaded norms does not,
by itself, deprive a subject of responsibility (or reduce said responsibility).
The fact that slavery was abolished suggests two things. First, there is
an inferential route from values and beliefs the subject did have or values
and beliefs that she was capable of gaining access to (without unreason-
able hardship) to holding the belief/adopting the value that slavery was
wrong. Second, a person who is capable of embracing one set of norms is
capable of evaluating and changing said norms. The capacity for self-
evaluation and change is, after all, a ubiquitous feature of our abilities.
When I was a child, I believed in God, now I do not. More pertinently,
perhaps, I once thought abortion was wrong, now I do not. This says
something about my belief and value formation abilities. Possessing skills,
tastes, habits, and values do not prevent change. To the contrary, it reveals
the capacity to acquire, evaluate, update, and change such skills, tastes,
habits, and values. Unless the subject has been exposed to some physio-
logical or psychological insult, if she possesses values, she has the capacity
to think about them, consider their worth, and change them if required.
Consequently, people from other cultures typically have the capacity to
self-evaluate and change their values accordingly. Furthermore, in many,
if not most, instances their environment is sufficiently rich to contain
values and information sufficient for them to change their values to what
we now take to be the right ones. Consequently, they can be held respon-
sible for their wrongdoing because their adoption of wrongheaded values
was based on affected ignorance.
None of this shows that all cases of cultural differences in value are due
to affected ignorance, nor that there are no agents who are genuinely
incapable of comprehending that some of their values are questionable.
But we have seen that affected ignorance characterizes many such cases.
And where it does not, it is not the capacity to evaluate or change one’s
values that is at issue, but whether the person’s value and environment
would have allowed her to reach values close enough to what we think are
the right ones. The slavery example suggests that affected ignorance is
the main culprit behind divergent moral values. But to make this argu-
ment requires a more thoroughgoing exploration. Suffice it to say that
affected ignorance is an important factor behind the holding of intransi-
gent and divergent values.
The usual examples of cultural differences in values are, at any rate,
quite different from the prototypical cases of insanity. The insane rarely
possess values that differ from ours, nor do they suffer from specifically
 Values, Sanity, and Responsibility 279

moral deficits.16 They typically suffer from delusions and hallucinations

that affect many different areas of their lives, not just their values.17 These
delusions or hallucinations create the disturbance that gives rise to the
criminal action. For instance, an insane person might think that he is
facing a creature very different from the one he is actually struggling
with, like the infanticidal father thinking he’s fighting a snake; he might
think that the person has designs on his life and take himself to be acting in
self-defense, as apparently McNaughtan did; or he could believe he is
committing a harm only to avoid a greater future harm, as in the case of
Yates. Instead of espousing substantially different values, the insane are
typically wrong about nonmoral matters of fact. In other cases, their
madness creates lacunas in their moral outlook. A person suffering from
command hallucinations might think that she ought to carry out an
otherwise prohibited action, for instance kill someone. But even in these
cases there is rarely a radical change in their moral compass.
When it comes to insanity, then, determinations of someone’s responsi-
bility depend in no small measure on whether their belief formation is
subject to undue influences as a result of mental disorder (or physiological
insult), to what extent, and how it affects their values. Depression is not
typically an excusing condition, for though the subject’s thoughts are
affected by the depression—her life seems lackluster and meaningless—it
is unlikely to affect her in such a way that she becomes unable to tell right
from wrong or become incapable of acting in accordance with her values.18
At the other extreme, someone in the grip of psychosis whose vision of the
world has been distorted may kill her child in the mistaken belief that she is
preventing future greater harm. The prototype of insanity looks little like
the prototype of intransigent values that differ from ours.

16
This is why I think psychopaths are not insane (Maibom 2008).
17
One might argue that people from different cultures possess a whole range of false
beliefs, which play a distorting role similar to that of delusions or hallucinations, and that
therefore the analogy between insanity and culturally induced values holds. Not so. First,
people who have looked for such differences in beliefs that would be relevant to the
morally divergent views have had difficulties finding them (Brink 1989; Doris and
Plakias 2008). Second, by contrast to ordinary beliefs, subjects tend to be deeply
convinced by the truth of their delusions although they are often bizarre. A delusion is
not justified by the available evidence—it is isolated from other relevant beliefs—and is
often held despite overwhelming reasons not to believe it. And hallucinations involve as-
if perceptions, which is not characteristic of false beliefs generally. Third, insofar as we
are all subject to culturally induced beliefs, that are similar to delusions, either we are as
little responsible for holding values derived from them as are people from different times
or cultures, or we are all responsible.
18
Or rather, we may excuse her for small failures that seem to flow from her
anhedonic state, but should she kill someone we will be skeptical that “the depression
made her do it.”
280 Heidi L. Maibom

4 . C O N C LU S I O N

If there is an inferential route from someone’s beliefs, values, etc. to know-

ledge that what she is doing is wrong, and it was not too onerous to follow,
then we can hold her responsible. Moreover, if she does not know what she is
doing is wrong and she does not possess beliefs or values from which there is
an inferential route to such knowledge, but she could acquire such know-
ledge, beliefs, or values without too much hardship given her environment
and her mental condition, she can also be held responsible. This is, I have
argued, the best interpretation of the epistemic condition on responsibility, at
least when it comes to moral intransigence. Choosing to ignore that what one
does is wrong or simply disagreeing with one’s community about what is
morally right or wrong cannot be held up as an excuse.
Wolf is right to point out that if a person is unable to objectively
evaluate or appropriately change her values, then she should be excused
ceteris paribus. She cannot be held responsible for performing actions that
she was unable to know were wrong. However, culture does not, in general,
inhibit moral change, nor do divergent upbringings. The very fact that we
possess values suggests that we are capable of changing them, barring
madness or brain damage. Whether a person can change her values so
that they are more in accord with what we now believe are the right ones
depends, of course, on her values and beliefs and the information that her
environment affords. I argued that ancient Greek slaveholders can be held
responsible, and I see little reason to think male chauvinists of our father’s
generation cannot also be blamed. It may turn out that most cases of
culturally divergent values are due not to inability, but to inexpediency.
Unless we simply assume that one can only be held responsible if one thinks
of one’s action as wrong, we should not suppose that people we usually
judge to be the most evil—e.g. perpetrators of genocides, slaveholders, and
pedophile rapists—are the least responsible. Such a view reduces all culp-
ability to moral incontinence. But as we have seen, most who do wrong do
not take themselves to do so. These wrongs are not typically perpetrated by
individuals who are unable to see the error of their ways, but by individuals
who are unwilling or are neglecting to do so; they engage in affected
ignorance. In the case of cultural differences, there is often an inferential
route from values and beliefs that are held by the person to what we take to
be the right values. For instance, there is little question that the Hutu killers
were able to recognize the humanity of their Tutsi neighbors, that they
recognized the prohibition on killing, etc., etc. It was, however, expedient
to ignore this, so those who were not coerced into engaging in the genocide
chose to ignore the wrongness of their actions. Consequently, they did not
think of their actions as wrong. Yet, we can surely hold them responsible.
 Values, Sanity, and Responsibility 281

REFERENCES

American Psychiatric Association (2000). Diagnostic and Statistical Manual of

Mental Disorders, 4th edn. Text Revision (DSM-IV). (Washington DC: American
Psychiatric Association).
Aristotle (1984). “The Politics.” Trans. Jowett. In: The Complete Works of Aristotle.
ed. J. Barnes. (Princeton, NJ: Princeton University Press).
Baumeister, R. (1997). Evil: Inside Human Violence and Cruelty. (New York: Henry
Holt).
——Stillwell, A., and Wotman, S. (1990). “Victims and perpetrator accounts of
interpersonal conflict: Autobiographical narratives about anger.” Journal of Per-
sonality and Social Psychology 59, 994–1005.
Beck-Sander, A., Birchwood, M., and Chadwick, P. (1997). “Acting on command
hallucinations: A cognitive approach.” British Journal of Clinical Psychology 36,
139–48.
Beowulf. (2000). Trans. Seamus Heaney. (London: W. W. Norton & Co).
Berkowitz, L. (1978). “Is criminal violence normative behavior? Hostile and instru-
mental aggression in violent incidents.” Journal of Research in Crime and Delin-
quency 15, 148–61.
——and Powers, P. (1979). “Effects of timing and justification of witnessed
aggression on the observers’ punitiveness.” Journal of Research in Personality 13,
71–80.
Brandt, R. B. (1959). Ethical Theory: The Problems of Normative and Critical Ethics.
(Englewood Cliffs, NJ: Prentice-Hall).
Brink, D. (1989). Moral Realism and the Foundation For Ethics. (Cambridge:
Cambridge Uuniversity Press).
Calvete, E. (2008). “Justification of violence and grandiosity schemas as predictors
of antisocial behavior in adolescents.” Journal of Abnormal Child Psychology 36,
1083–95.
CBC News, June 15, 2010. Father, son plead guilty to Aqsa Parvez murder. <http://
www.cbc.ca/news/canada/toronto/story/2010/06/15/parvez-guilty-plea.html>.
Dallaire, R. (2004). Shake Hands with the Devil: The Failure of Humanity in
Rwanda. (Toronto: Vintage Canada).
Doris, J. and Plakias, A. (2008). “How to argue about disagreement: Evaluative
diversity and moral realism.” In W. Sinnott-Armstrong (ed.), Moral Psychology,
vol. 2. The Cognitive Science of Morality: Intuition and Diversity. (Cambridge,
MA: MIT Press), 303–31.
Faraci, D. and Shoemaker, D. (2010). “Insanity, Deep Selves, and Moral Responsi-
bility: The Case of JoJo.” Review of Philosophy & Psychology 1, 319–32.
Frankfurt, H. (2003). “Freedom of the will and the concept of a person.” In
G. Watson (ed.), Free Will, 2nd edn. (New York: Oxford University Press),
322–36.
Garlan, Y. (1988). Slavery in Ancient Greece. (Cornell: Cornell University Press).
The Iliad. (1990). Trans..Robert Fagles. (Bath: The Bath Press).
282 Heidi L. Maibom

Jeffrey, R., Pasewark, R., and Bieber, S. (1988). “Insanity pleas: predicting Not
Guilty by Reason of Insanity adjudications.” Bulletin of the American Academy of
Psychiatry and Law 16, 35–9.
Katz, J. (1988). Seductions of Crime: Moral and Sensual Attractions in Doing Evil.
(New York: Basic Books).
Krakauer, J. (2003). Under the Banner of Heaven: A Story of Violent Faith.
(New York: Anchor Books).
Larbey, C. (2007). “The secret lives of Kim John and Francis Philip. What do we
really know about the Cathedral killers?” St. Lucia Star, May 18, 2007.
Luckenbill, D. (1977). “Criminal homicide as a situated transaction.” Social
Problems 25, 176–86.
McNiel, D., Eisner, J., and Binder, R. (2000). “The relationship between command
hallucinations and violence.” Psychiatric Services 51, 1288–92.
Maibom, H. (2008). “The mad, the bad, and the psychopath.” Neuroethics 1, 167–84.
Maudsley, H. (1898). Responsibility in Mental Disease. (New York: D. Appleton and
Company).
Moody-Adams, M. (1994). “Culture, responsibility, and affected ignorance.” Ethics
104, 291–309.
Moran, R. (1981). Knowing Right From Wrong. (New York: The Free Press).
MY Lee, T., Chong, S., Chan, Y., Sathyadevan, G. (2004). “Command hallucin-
ations among Asian patients with schizophrenia.” Canadian Journal of Psychiatry
49, 838–42.
Reznik, L. (1997). Evil or Ill: Defending the Insanity Defence. (New York: Routledge).
Rice, M. and Harris, G. (1990). “The predictors of insanity acquittal.” International
Journal of Law and Psychiatry 13, 217–24.
Rogers, J., Bloom, J., and Manson, S. (1984). “Insanity defences: contested or
conceded?” American Journal of Psychiatry 141, 885–8.
Schwartz, M., O’Leary, S., and Kendziora, K. (1997). “Dating aggression among
high school students.” Violence and Victims, 295–305.
Steadman, H., Keitner, L., Braff, J., and Arranites, T. (1983). “Factors associated
with a successful insanity plea.” American Journal of Psychiatry 140, 401–405.
Stillwell, A., Baumeister, R., and Del Priore, R. (2008). “We’re all victims here:
Towards a psychology of revenge.” Basic and Applied Social Psychology 30, 253–63.
Taylor, C. (1976). “Responsibility for self.” In A. E. Rorty (ed.) The Identities of
Persons. (Berkeley, CA: University of California Press), 281–99.
Watson, G. (1975). Free agency. Journal of Philosophy, LXII, 205–20.
Williams, B. (1981). “Internal and external reasons.” In his Moral Luck. (Cambridge:
Cambridge University Press), 101–13.
Wolf, S. (2003). “Sanity and the metaphysics of responsibility.” In G. Watson (ed.),
Free Will, 2nd edn. (New York: Oxford University Press), 372–87.
Wolfgang, M. (1958). Patterns in Criminal Homicide. (Philadelphia, PA: University
of Pennsylvania Press).
Xenophon. (2008a). Symposium. Trans. H. G. Dakyns. Project Guttenberg
(accessed January 14, 2012).
 Values, Sanity, and Responsibility 283

Xenophon. (2008b). Hellenica. Trans. H. G. Dakyns. Project Guttenberg (accessed

January 14, 2012).
Zelli, A., Dodge, K., Laird, R., Lochman, J., and Conduct Problems Prevention
Research Group. (1999). “The distinction between beliefs legitimizing aggression
and deviant processing of social cues: Testing measurement validity and the
hypothesis that biased processing mediates the effects of beliefs on aggression.”
Journal of Personality and Social Psychology 77, 150–66.
 12
Fairness and the Architecture
of Responsibility1
David O. Brink and Dana K. Nelkin

In this essay, we explore a conception of the nature and structure of

responsibility that draws on ideas about moral and criminal responsibility.
Though the two sorts of responsibility are not the same, the criminal
law reflects central assumptions about moral responsibility, and the two
concepts of responsibility have very similar structure. Our conception of
responsibility draws on work of philosophers in the compatibilist tradition
who focus on the choices of agents who are reasons-responsive and work
in criminal jurisprudence that understands responsibility in terms of
the choices of agents who have capacities for practical reason and whose
situation affords them the fair opportunity to avoid wrongdoing.2 We
treat these two perspectives as potentially complementary and argue that
each can learn things from the other. Specifically, we think that criminal
jurisprudence needs a more systematic conception of the capacities for
normative competence and that ideas from the reasons-responsive literature

1
This essay is fully collaborative. The authors are listed in alphabetical order. The
ideas for this essay grew out of a graduate seminar that we taught together on the topic of
partial responsibility in 2008 and were refined in a seminar on responsibility that DB
taught in 2011. Versions of this material were presented at the University of Illinois, the
University of Calgary, the Murphy Institute at Tulane University, Cornell University,
the University of Western Ontario, and the New Orleans Workshop on Agency and
Responsibility. We would like to thank audiences on those occasions for helpful
feedback. We owe special thanks to input from Craig Agule, Sarah Aikin, Amy Berg,
Mitch Berman, Michael McKenna, Per Milam, Richard Miller, Michael Moore,
Stephen Morse, Derk Pereboom, Erick Ramirez, Sam Rickless, Tim Scanlon, David
Shoemaker, Jada Twedt Strabbing, Sarah Stroud, Matt Talbert, Michael Tiboris, and
Gary Watson.
2
The philosophical work in the reasons-responsive wing of the compatibilist trad-
ition on which we draw includes Fischer and Ravizza 1998, Wallace 1994, Wolf 1990,
and Nelkin 2011. The criminal jurisprudence work on which we draw includes Hart
1957 and 1961, Moore 1997, and Morse 1994, 2002, and 2003.
 Fairness and Architecture of Responsibility 285

on moral responsibility, some familiar and some novel, can fill this need.
However, we think that moral philosophers tend to focus on the capacities
involved in responsibility and so tend to ignore the situational element in
responsibility recognized in the criminal law literature. Our conception of
responsibility brings together the dimensions of normative competence
and situational control, and we factor normative competence into
cognitive and volitional capacities, which we treat as equally important
to normative competence and, ultimately, responsibility. Moreover,
we argue that normative competence and situational control can and
should be understood as expressing a common concern that blame and
punishment presuppose that the agent had a fair opportunity to avoid
wrongdoing. Thus, we treat the value that criminal law theorists associate
with the situational element of responsibility as the umbrella concept for
our conception of responsibility, one that explains the distinctive
architecture of responsibility.
This essay aims to motivate and articulate this sort of fair opportunity
conception of the architecture of responsibility. It is part of a larger project
that develops this conception and applies it to issues of partial responsi-
bility, involving insanity and psychopathy, immaturity, addiction, provo-
cation, and duress. The details and applications of the fair opportunity
conception of responsibility are interesting and important, and we hope to
address them more fully elsewhere. But the framework itself is important
and requires articulation.

1 . R E S P O N S I B I L I T Y , BL A M E , A N D T H E
REACTIVE ATTITUDES

P. F. Strawson famously highlighted the link between ascriptions of

responsibility and the reactive attitudes (Strawson 1962). The reactive
attitudes involve emotional responses directed at oneself or another in
response to that person’s conduct. Reactive attitudes include hate, love,
pride, gratitude, anger, regret, resentment, indignation, and forgiveness. So
understood, the reactive attitudes form a large and heterogeneous class.
Some of these reactive attitudes have little direct connection with moral
praise and blame and responsibility. Consider the difference between
anger and resentment. Anger need not have moral content. I might be
momentarily angry or upset with a very young child who has carelessly
damaged a treasured keepsake of mine. But resentment would seem to be
out of order. Resentment seems to involve a kind of anger or upset that
presupposes that one has been mistreated or wronged by another in some
286 David O. Brink and Dana K. Nelkin

way. This kind of moral judgment does not seem to apply to a very young
child. The idea that assumptions about responsibility are embedded in the
reactive attitudes makes most sense if we focus on this narrower class of
reactive attitudes that are moralized (cf. Wallace 1994: ch. 2).
In particular, we want to focus on the attitudes and practices of praise
and blame, especially as they reflect assumptions about responsibility. Some
negative reactive attitudes, such as regret, don’t seem to implicate responsi-
bility at all. Bernard Williams describes the case of a truck driver who,
through no fault of his own, hits and kills a child who has darted into the
street (1976: 28). As Williams claims, it is appropriate for the driver to feel
a kind of agent-regret at being the instrument of the child’s death, which is
distinct both from the regret or horror that bystanders might feel and from
guilt for having been responsible for wrongdoing.
Normally, blame only makes sense if the agent is responsible for some
kind of wrong. Some philosophers distinguish between two kinds of blame
and responsibility. For instance, Gary Watson distinguishes between
responsibility as attributability and as accountability (1996). An agent is
responsible in the attributive sense, roughly speaking, when her actions
reflect the quality of her will in the right way. Some kinds of blame can be a
fitting response to the quality of the agent’s will. For instance, A might have
hard feelings toward B if B injures A through malice, recklessness, or
negligence. Here, our reactive attitudes track the insufficient regard that
B shows A’s interests and rights. But attributability does not guarantee
accountability. Consider a situation in which we find out that though B’s
actions exhibit malice, in no relevant sense did B have an opportunity to do
otherwise, perhaps because he suffers from a serious mental illness and is
not a competent decision-maker as a result. In these cases, we are likely to
think that B was not at fault or culpable and so not accountable for the harm
he did. Although hard feelings may remain and be perfectly appropriate in
such cases, reactive attitudes involving resentment and indignation cease to
be appropriate and tend to dissipate. Attributability is necessary but not
sufficient for accountability. In this essay, we are especially interested in
responsibility as accountability and its connection with reactive practices and
attitudes involving blame, and we rely on an intuitive understanding of the
reactive attitudes that seem to be especially responsive to accountability.3 It is
this sense of blame and responsibility that we take to be most relevant to the
sort of responsibility required for punishment and to capture what is
common to both moral and legal responsibility.

3
Here, we are in agreement with Watson (1996: 276, 2011). By contrast,
T. M. Scanlon develops a conception of blame that seems to presuppose only
attributability, not accountability (2008: ch. 4, esp. 202).
 Fairness and Architecture of Responsibility 287

Thus, our focus in what follows will be restricted to attitudes of praise

and blame involving accountability, with special attention to attitudes and
practices of blame, rather than praise. We do so because our aim is to
combine insights from criminal jurisprudence, which focuses on criminal
acts that involve wrongdoing, with those from moral theory. But we believe
that there are natural ways of extending what we say here about attitudes of
blame and blameworthy actions to praise and praiseworthy actions.
Strawson links responsibility and reactive attitudes, such as those of
resentment and indignation, in a biconditional fashion.
Reactive attitudes involving blame and praise are appropriate just in case the targets
of these attitudes are responsible.
Call this biconditional claim Strawson’s thesis. Strawson’s thesis can be
interpreted in two very different ways, depending on which half of the
biconditional has explanatory priority.
According to the first interpretation, there is no external, or response-
independent justification of our attributions of responsibility. This reading
fits with Strawson’s view that our reactive attitudes and ascriptions of
responsibility, as a whole, do not admit of external justification. Particular
expressions of a reactive attitude might be corrigible as inconsistent with a
pattern of response, but the patterns of response are not themselves corri-
gible in light of any other standard. Similarly, particular ascriptions of
responsibility might be corrigible in light of patterns in our ascriptions of
responsibility, but the patterns themselves are not corrigible in light of any
other standards. Responsibility judgments simply reflect those dispositions
to respond to others that are constitutive of various kinds of interpersonal
relationships. This is a response-dependent interpretation of Strawson’s
thesis.
This response-dependent interpretation of Strawson’s thesis is probably
the right interpretation of Strawson.4 But as a systematic, rather than an
interpretive, matter, we favor an alternative interpretation of Strawson’s
thesis that is realist, rather than response-dependent. This interpretation
stresses the way that the reactive attitudes make sense in light of and so
presuppose responsibility. As such, the reactive attitudes are evidence about
when to hold people responsible, but not something that constitutes them
being responsible. It’s true that the reactive attitudes are appropriate if and
only if the targets are responsible, but it’s the responsibility of the targets

4
Watson defends this response-dependent interpretation of Strawson’s thesis, at least
on interpretive grounds (1987: esp. 222). Wallace defends this interpretation of Strawson’s
thesis in its own right (2004: 19), though we think other elements in Wallace’s account fit
better with an alternative realist reading.
288 David O. Brink and Dana K. Nelkin

that makes the reactive attitudes toward them fitting or appropriate. In

the biconditional relationship between responsibility and the reactive
attitudes, it is responsibility that is explanatorily prior, according to this
realist interpretation. Strawson points out that the limits of our reactive
attitudes are indicated by our practices of exemption and excuse. Because
the realist believes that the reactive attitudes presuppose responsibility,
she can appeal to our practices of exemption and excuse to help understand
the conditions under which we are responsible. This will be a response-
independent conception of responsibility.
A response-independent conception of responsibility is hostage to
traditional worries about freedom of the will. The problem of free will
is the problem of reconciling responsibility with determinism, because
responsibility may seem to presuppose freedom of the will, and freedom
of the will may seem incompatible with determinism. Our realist
approach to responsibility and the reactive attitudes is best articulated
as a version of compatibilism that denies that responsibility requires a
form of freedom that would be undermined by the truth of determinism.
In particular, because our practices of exemption and excuse track forms
of normative competence and situational control, rather than the truth of
determinism, they promise to ground a compatibilist conception of
responsibility. Though we will articulate this compatibilist interpretation
of our project, we cannot defend it here (though we return to these issues
briefly in Section 7 below).5
Although Strawson focuses primarily on the relation between the reactive
attitudes and responsibility, his thesis fits well with a particular approach to
punishment and criminal responsibility. In particular, the realist interpret-
ation of Strawson’s thesis fits with a broadly retributive approach to blame
and punishment, precisely because the retributivist thinks that the reactive
attitudes and our practices of blame and punishment can be appropriate
responses to culpable wrongdoing, where culpable wrongdoing is wrong-
doing for which the agent is responsible. To see this, it will be useful for us
to say more about both blame and punishment.

5
Because we think that reactive attitudes involving praise and blame presuppose that
the targets of these attitudes are responsible, we accept the need to provide a response-
independent conception of responsibility and to answer skeptical doubts about responsi-
bility. Consequently, we see response-dependent conceptions of responsibility as offering
skeptical solutions to skeptical worries (cf. Kripke 1982: 66–7). We view skeptical
solutions to skeptical problems as, at best, a kind of fallback solution to be entertained
only after straight solutions have clearly failed. For a fuller exploration of the
compatibilist aspects of this conception of responsibility, see Nelkin 2011. We believe
that at least some of what we say here can be accepted by incompatibilists who accept
further conditions on responsibility, beyond that of indeterminism.
 Fairness and Architecture of Responsibility 289

When agents are responsible (accountable) for doing wrong blame is

appropriate. Blame typically involves both censure and sanction. When we
blame someone, we not only censure her conduct but also censure the agent
herself for engaging in that conduct. Parents are often warned to disapprove
the bad conduct of their children but not to blame them. This is because
blame involves finding fault in the agent and that seems to assume that the
agent is responsible (accountable) and could have avoided the conduct.
Being blameworthy licenses various kinds of sanction, often informal and
sometimes formal. Blame itself can involve overt reproach, which is a kind
of sanction, whether directed at another or at oneself. Sometimes reproach
is the only appropriate sanction. But sometimes blameworthiness licenses
other informal sanctions, such as public rebuke or social distancing. And in
other cases, blameworthiness might license various kinds of punishment,
whether personal, social, or legal. To be blameworthy is to be a fitting
object of blame, censure, and sanction. It is to be deserving of these
attitudes and responses. No doubt, where sanctions are appropriate, they
have to be proportionate, and there may be cases in which one is blame-
worthy and yet it is not on balance appropriate to blame or sanction. But,
presumably, even in these cases there is a pro tanto case for blame and at
least some informal sanction, if only self-reproach, as a fitting response to
culpable wrongdoing.
On this view, punishment is a species of blame for culpable wrongdoing.
On a broadly retributive view of the criminal law, this is true of legal
punishment as well. We understand criminal punishment as the authorized
deprivation of an agent’s normal rights and privileges, because he or she has
been found guilty of a criminal act.6 Punishment is a form of blame, and like
other kinds of blame, presupposes culpable wrongdoing. Legal retributivism,
as we understand it, is the claim that legal punishment is justified on the basis
of culpable legal wrongdoing. This claim can take positive or negative forms.
According to positive retributivism, culpable wrongdoing is both necessary
and sufficient for justifying punishment. The sufficiency claim admits
of both strong and weak interpretations. According to strong sufficiency,
culpable wrongdoing is a sufficient condition of justified proportional
blame and punishment, whereas, according to weak sufficiency, culpable
wrongdoing is sufficient for a pro tanto case for proportional blame and
punishment. Weak sufficiency allows for the pro tanto case for retributive
blame and punishment to be overridden in particular cases by nonculpability
moral considerations, such as forgiveness or mercy. By contrast, according to

6
Cf. Bedau and Kelly 2010. We aim for a normatively neutral and ecumenical
definition of punishment and one that identifies punishment as involving deprivations
of certain sorts, but not essentially involving the imposition of pain or suffering.
290 David O. Brink and Dana K. Nelkin

negative retributivism, culpable wrongdoing is necessary, but not sufficient,

for justified punishment.7
Legal retributivism (in either version) has the virtue of explaining well
the two principal forms of affirmative defense in the criminal law.
According to the retributivist, justified punishment aspires to track culpable
wrongdoing.8 Wrongdoing and culpability are independent variables.
Affirmative defenses, whose success justifies acquittal, deny either
wrongdoing or culpability. Justifications, such as the necessity defense,
deny wrongdoing, insisting that behavior that would otherwise be wrong
is not in fact wrong in these circumstances. Excuses, such as the insanity
defense, deny culpability or responsibility, claiming that the agent acted
wrongly but was not responsible for her wrongdoing.
Here, the criminal law reflects the moral landscape well. Moral retribu-
tivism could be understood as the claim that moral blame (that presupposes
accountability) and informal sanction are appropriate only as a response to
culpable moral wrongdoing. It too has the virtue of explaining the two
principal ways of avoiding blame—justifying and excusing conduct. Justi-
fication denies wrongdoing, and excuse denies responsibility for wrong-
doing. Insofar as moral retributivism says that moral blame ought to track
desert, where desert is the product of the two independent variables of
wrongdoing and responsibility, it fits our moral defenses like a glove.
In this way, the realist interpretation of Strawson’s biconditional can
appeal to our understanding of excuses to provide a window on to the
nature of responsibility.9 An analysis of criminal law doctrines of excuse can
be a part of this investigation. In this context, it is worth addressing the
relationship between excuses and exemptions. The prototypical case of an
exemption is a case in which an actor is not responsible for what he did
because of quite general impairments of his agency. So, for instance,
insanity and immaturity are sometimes described as exemptions. By
contrast, excuses are sometimes claimed to be prototypically case-specific
in which the agent is otherwise normal and responsible but acted

7
Cf. Duff 2008. The view that we have called “negative retributivism” is sometimes
called a “mixed theory” of punishment, because it requires more than one type of
justificatory reason, typically, both retributivist and consequentialist. It is also worth
noting that our definition of retributivism does not commit retributivists to endorsing
the thesis that punishment is intrinsically good, as some retributivists claim.
8
Precisely for this reason, a skeptic about moral responsibility will deny that any
retributivist view of punishment can be correct. For a skeptical view and its relation to
punishment, see Pereboom 2012.
9
Moore describes excuse as the “royal road” to responsibility (1997: 548). Whereas
the realist regards our practices of excuse as potential evidence of a response-independent
conception of responsibility, a response-dependent conception will understand our
practices of excuse as constitutive of responsibility.
 Fairness and Architecture of Responsibility 291

inadvertently or was subject to coercion in a specific situation. Despite the

existence of these two different kinds of prototypical cases, we think that it
is a mistake to treat exemptions and excuses as disjoint classes. First, while
Strawson and others include insanity among the exemptions, the criminal
law treats insanity as an excuse. In fact, the criminal law includes all claims
to less than full culpability in the single category of excuse. So there is some
reason not to assume that exemptions cannot be excuses. Second, the
prototypical cases are not exhaustive of the possibilities, as Strawson
himself recognized (1962: 79). Strawson’s partition is into cases in which
the reactive attitudes are generally disabled in regard to a particular agent
and cases in which they are selectively disabled due to inadvertence or
compulsion. But there are at least three different dimensions on which
these paradigm cases can be distinguished: scope, duration, and the location
of the obstacle to culpability. Immaturity, for example, or even more
temporary conditions, such as depression or even dementia due to
dehydration, might undermine responsibility for all sorts of actions
during the episode in question, and so have wide scope. In contrast, a
particular perceptual deficit, or a compulsive disorder narrowly confined to
one area, like kleptomania, might have a relatively narrow scope. Paradigm
cases can also be distinguished on the basis of duration. A phobia, for
example, might affect one’s choices in a narrow area, but be lifelong, in
contrast to a short spell of dementia caused by dehydration. The third
dimension is the location of the obstacle as either within or outside of the
agent. Immaturity is an example of the former, and low lighting conditions
that prevent one from seeing someone else in need is an example of the
latter. All three of these are separable in principle, but in the original
paradigm cases, go together. For example, childhood is long-lasting, has a
wide scope (though narrowing as one ages), and seems to be explained by
the agent’s own capacities. Not realizing one is stepping on another’s toes,
in contrast, is typically short-lived, narrow in scope, and explainable by
something about the particular situation rather than one’s capacities.
Recognizing that considerations that mitigate culpability can fall in a
variety of places along all three dimensions suggests to us that it would be
most useful to consider all of the cases as ones involving potential excuses
with varying degrees of scope and duration and with varying locations
between the agent and the situation. On the proposal that we favor,
exemptions are best understood as comparatively global or standing
excuses.
The challenge for the realist interpretation of Strawson’s thesis is to use
the reactive attitudes as evidence to uncover an independent conception of
responsibility that can support the reactive attitudes. If we study responsi-
bility by studying excuses, we find that excuses factor into two main kinds
292 David O. Brink and Dana K. Nelkin

on the location dimension. Some excuses reflect compromised psycho-

logical capacities of agents. We will conceptualize these as failures of
normative competence. Insanity is the most familiar excuse of this type.
But some excuses reflect no failure of normative competence. Instead,
they reflect a lack of normal situational control. In such situations, though
the agent is normatively competent, factors external to her deprive her of
the fair opportunity to avoid wrongdoing. Coercion and duress provide
excuses of this type. We think that attention to these two kinds of excuse
provides the key to understanding the architecture of responsibility.

2 . N O R M A T I VE CO M P E T E N C E

If someone is to be culpable or responsible for her wrongdoing, then she

must be a responsible agent. So we need to distinguish between responsible
and nonresponsible agents. Our paradigms of responsible agents are
normal mature adults with certain sorts of capacities. We do not treat
brutes or small children as responsible agents. Brutes and small children
both act intentionally, but they act on their strongest desires or, if they
exercise deliberation and impulse control, it is primarily instrumental
reasoning in the service of fixed aims. By contrast, we suppose, responsible
agents must be normatively competent. They must not simply act on their
strongest desires, but be capable of stepping back from their desires,
evaluating them, and acting for good reasons. This requires responsible
agents to be able to recognize and respond to reasons for action. If so,
normative competence involves reasons-responsiveness, which itself involves
both cognitive capacities to distinguish right from wrong and volitional
capacities to conform one’s conduct to that normative knowledge.10
It is important to frame this approach to responsibility in terms of
normative competence and the possession of these capacities for reasons-
responsiveness. In particular, responsibility must be predicated on the
possession, rather than the use, of such capacities. We do excuse for lack
of competence. We do not excuse for failures to exercise these capacities
properly. Provided they had the relevant cognitive and volitional capacities,
we do not excuse the weak-willed or the willful wrongdoer for failing to

10
In framing our approach to the internal dimension of responsibility this way, we
draw on previous work in the compatibilist tradition that emphasizes normative compe-
tence (Wolf 1990, Wallace 1994) and reasons-responsiveness (Wolf 1990, Wallace
1994, Fischer and Ravizza 1998, and Nelkin 2011) and distinguishes cognitive and
volitional dimensions of reasons-responsiveness (Wallace 1994, Fischer and Ravizza
1998).
 Fairness and Architecture of Responsibility 293

recognize or respond appropriately to reasons. If responsibility were predi-

cated on the proper use of these capacities, we could not hold weak-willed
and willful wrongdoers responsible for their wrongdoing. It is a condition
of our holding them responsible that they possessed the relevant
capacities.11
Normative competence, on this conception, involves two forms of
reasons-responsiveness: an ability to recognize wrongdoing and an ability
to conform one’s will to this normative understanding. Both dimensions of
normative competence involve norm-responsiveness. As a first approxima-
tion, we can distinguish moral and criminal responsibility at least in part
based on the kinds of norms to which agents must be responsive. Moral
responsibility requires capacities to recognize and conform to moral norms,
including norms of moral wrongdoing, whereas criminal responsibility
requires capacities to recognize and conform to norms of the criminal
law, including norms of criminal wrongdoing.
Reasons-responsiveness is clearly a modal notion and admits of degrees;
one might be more or less responsive. This raises the question how respon-
sive someone needs to be to be responsible. This is an important and
difficult issue, deserving more careful discussion than we can give it here.
We make some preliminary remarks here, which we will refine in later
sections. We might begin by distinguishing different grades of responsive-
ness. Here, we adapt some ideas from John Fischer and Mark Ravizza in
their book Responsibility and Control about the responsiveness of the
mechanisms on which agents act to our issue about how reasons-responsive
the agents themselves are.12 We propose to specify the degree to which an
agent is responsive to reasons in terms of counterfactuals about how she
would believe or react in situations in which there was sufficient reason for
her to do otherwise.13 An agent is more or less responsive to reason

11
Sidgwick famously objects to Kant’s conception of autonomy as conformity to
principles of practical reason that this would prevent us from holding criminals respon-
sible and would allow us to recognize only morally upright behavior as responsible
(Sidgwick 1907: 511–16). The solution to this problem is for Kant to define autonomy
in terms of capacities for conformity to principles of practical reason.
12
The conception of reasons-responsiveness that Fischer and Ravizza defend is
mechanism-based, rather than agent-based (1998: 38). By contrast, we favor a version
of reasons-responsiveness that is agent-based, rather than mechanism-based, precisely
because we think that responsibility and excuse track the agent’s capacities, rather than
the capacities of her mechanisms. For defense of the agent-based approach, see Nelkin
2011: 64–79 and McKenna 2012.
13
For present purposes, in specifying an agent’s capacities in terms of such counter-
factuals, we can remain agnostic about whether capacities or counterfactuals have
explanatory priority, in particular, whether capacities ground the counterfactuals or
whether the capacities just consist in the truth of such counterfactuals.
294 David O. Brink and Dana K. Nelkin

depending on how well her judgments about what she ought to do and her
choices would track her reasons for action.
We could begin this process by distinguishing two extreme degrees of
responsiveness.
 Strong Responsiveness: Whenever there is sufficient reason for the agent to
act, she recognizes the reason and conforms her behavior to it.
 Weak Responsiveness: There is at least one situation in which there is
a sufficient reason to act, and the agent recognizes that reason and
conforms her behavior to it.
However, it does not seem plausible to model normative competence in
terms of either strong or weak responsiveness. Strong responsiveness is too
strong for the same reason we gave for focusing on competence, rather
than performance. We do not require that people actually act for sufficient
reasons to do otherwise; it is the capacities with which they act that matter.
The weak-willed are, at least typically, responsible for their poor choices.
Moreover, weak responsiveness seems too weak. It treats someone as respon-
sive in the actual situation even if she did not respond in the actual situation
and there is only one extreme circumstance in which she would recognize and
respond to reasons for action. The Goldilocks standard of responsiveness
evidently lies somewhere between these extremes. Of course, there is consid-
erable space between the extremes—the gap between always and once.
We might stake out an intermediate form of responsiveness in something
like the following terms.
 Moderate Responsiveness: Where there is sufficient reason for the agent to
act, she regularly recognizes the reason and conforms her behavior to it.
Moderate responsiveness is deliberately vague; it specifies a range or space
of counterfactuals that must be true for the agent to be responsive. Ideally,
we would be able to specify a preferred form of moderate responsiveness
more precisely. But what is important for present purposes is that reasons-
responsiveness is a matter of degree and that the right threshold for
responsibility is probably some form of moderate responsiveness.
So far, this conception of responsiveness is coarse-grained in ways that
might prove problematic. For one thing, it lumps together cognitive and
volitional dimensions of responsiveness. But if they are independent aspects
of normative competence, then we may need to assess responsiveness along
these two dimensions separately. Moreover, it is at least conceivable that we
might require different degrees of responsiveness in cognitive and volitional
dimensions of competence. For instance, Fischer and Ravizza distinguish the
cognitive and volitional dimensions of reasons-responsiveness in terms of
“reasons-receptivity” and “reasons-reactivity” (respectively). Their conception
 Fairness and Architecture of Responsibility 295

of reasons-responsiveness is mixed, because it treats receptivity and reactivity

asymmetrically. They combine moderate receptivity and weak reactivity (1998:
81–2). We ultimately reject this asymmetry, but it represents a conception of
responsiveness worth considering.
Furthermore, this initial formulation of responsiveness assumes that we
consider all situations in which there is sufficient reason to act together. But
we may find it more informative to partition possibilities into groups,
depending on the kinds of reasons at stake and other aspects of the
situations in which agents finds themselves. For instance, in deciding
whether an agent had sufficient volitional capacity to overcome fears that
stood in the way of her performing her duty, we may think it best to restrict
our attention to those counterfactuals in which she faced threats or fears of
comparable kind or magnitude.
For these reasons, we may need to make our assessments of the degree of
an agent’s responsiveness more fine-grained in several ways. We address
some of these complications below.

3. THE COGNITIVE DIMENSION OF

NORMATIVE COMPETENCE

Normative competence requires the cognitive capacity to make suitable

normative discriminations, in particular, to recognize wrongdoing. If respon-
sibility requires normative competence, and normative competence requires
this cognitive capacity, then we can readily understand one aspect of the
criminal law insanity defense. A full account of the elements of insanity is
controversial, as we will see. But most plausible versions of the insanity
defense include a cognitive dimension, first articulated in the M’Naghten
rule that excuses if the agent lacked the capacity to discriminate right from
wrong at the time of action.14
Here is one place it might be important to distinguish between the
demands of moral and criminal responsibility. Presumably, moral responsi-
bility requires the ability to recognize moral norms, including norms that
specify moral wrongdoing, whereas criminal responsibility requires the
ability to recognize criminal norms, including norms that specify criminal
wrongdoing.15 The cognitive abilities to recognize these two different kinds

14
M’Naghten’s Case, 10 Cl. & F. 200, 8 Eng. Rep. 718 (1843).
15
There is a debate about whether the cognitive dimension of the insanity test,
expressed in M’Naghten’s rule, should be formulated in terms of capacities for recogniz-
ing criminal or moral wrongdoing. British criminal law has focused on criminal wrong-
doing, and American jurisdictions remain divided.
296 David O. Brink and Dana K. Nelkin

of norms might be different, with the result that it might be possible to be

criminally responsible without being morally responsible and vice versa.
It is common to contrast reason and emotion. This common contrast
might lead one to suppose that the cognitive dimension of normative
competence is purely cognitive and does not involve emotion or affect.
But this conclusion would be misguided. Emotional or affective deficits
may block normative competence by compromising cognitive capacity. For
instance, lack of empathy may make it impossible or very difficult to
recognize actions as injurious and, hence, legally or morally wrong. There
is also evidence that congenital damage to the amygdala, which is thought
to be the part of the brain responsible for emotional learning and memory,
may prevent the formation of normative or, at least, moral concepts. There
is emerging research that shows that psychopathy involves both abnormal-
ities in the amygdala and empathy deficits (Blair et al. 2005) Moreover,
psychopaths have been thought to have trouble with a psychological test
used to discriminate between moral norms and conventional norms.16
These findings raise questions about whether psychopaths have moral
concepts and so whether they have the cognitive capacity to distinguish
moral right from wrong. Even if they lack cognitive moral competence,
it doesn’t follow that they lack the capacity to recognize legal wrongdoing.
It is at least possible that some psychopaths might be criminally responsible
without being morally responsible.17 These are complicated issues that
deserve fuller examination, but they illustrate ways in which emotion
and affect can have a bearing on the cognitive dimension of normative
competence. Here, emotional capacities may be upstream from normative
cognition.

4. THE VOLITIONAL DIMENSION OF

NORMATIVE COMPETENCE

But there is more to normative competence than this cognitive capacity.

We assume that intentional action is the product of informational states,
such as beliefs, and motivational states, such as desires and intentions.
Though our beliefs about what is best can influence our desires, producing
optimizing desires, our desires are not always optimizing. Sometimes they
are good-dependent but not optimizing, when they are directed at lesser

16
See Blair et al. 2005: 57–9. For some skepticism, see Aharoni et al. 2012.
17
We suspect that severe psychopathy impairs, but does not eliminate, reasons-
responsiveness and that it may make for a better moral excuse than a criminal excuse.
Contrast Fine and Kennett 2004.
 Fairness and Architecture of Responsibility 297

goods, and sometimes they are completely good-independent. This is

reflected in cases of weakness of will in which we have beliefs about what
is best (and perhaps optimizing desires) but in which we act instead on the
basis of independent nonoptimizing passions and desires. This psycho-
logical picture suggests that being a responsible agent is not merely having
the capacity to tell right from wrong but also requires the capacity to
regulate one’s actions in accordance with this normative knowledge. This
kind of volitional capacity requires emotional and appetitive capacities to
enable one to form intentions based on one’s optimizing judgments and
execute these intentions over time, despite distraction and temptation.
Here, emotion and appetite are downstream from cognition and play a
separate, volitional or executive role. If one’s emotions and appetites are
sufficiently disordered and outside one’s control, this might compromise
volitional capacities necessary for normative competence. Consider the
following obstacles to volitional competence.
 Irresistible desires or paralyzing fears that are neither conquerable nor cir-
cumventable, as perhaps in some cases of genuine agoraphobia or addiction.18
 Clinical depression that produces systematic weakness of will in the form
of listlessness or apathy.
 Acquired or late onset damage to the prefrontal cortex in which agents
have considerable difficulty conforming to their own judgments about
what they ought to do, as in the famous case of Phineas Gage.19
Each of these cases involves significant volitional impairment in which
agents experience considerable difficulty implementing or conforming to
the normative judgments they form.
Notice that recognition of a volitional dimension of normative compe-
tence argues against purely cognitive conceptions of insanity, such as the

18
Mele understands a desire as conquerable when one can resist it and as circumven-
table when one can perform an action that makes acting on the desire difficult or
impossible (1990). The alcoholic who simply resists cravings conquers his impulses,
whereas the alcoholic who throws out his liquor and stops associating with former
drinking partners or won’t meet them at places that serve alcohol circumvents his
impulses. Conquerability is mostly a matter of will power, whereas circumventability is
mostly a matter of foresight and strategy.
19
Phineas Gage was a nineteenth-century railway worker who was laying tracks in
Vermont and accidentally used his tamping iron to tamp down a live explosive charge,
which detonated and shot the iron bar up and through his skull. Though he did not lose
consciousness, over time his character was altered. Whereas he had been described as
someone possessing an “iron will” before the accident, afterward he had considerable
difficulty conforming his behavior to his own judgments about what he ought to do. The
story of Phineas Gage is related, and its larger significance explored, in Damasio 1994.
298 David O. Brink and Dana K. Nelkin

M’Naghten test, which recognizes only cognitive deficits as the basis for
insanity, and in favor of the more inclusive Model Penal Code conception.
Mental Disease or Defect Excluding Responsibility: (1) A person is not responsible
for criminal conduct if at the time of such conduct as the result of a mental disease
or defect he lacks substantial capacity either to appreciate the criminality [wrongful-
ness] of his conduct or to conform his conduct to the requirements of law. (2) [T]he
terms “mental disease or defect” do not include an abnormality manifested only by
repeated criminal or otherwise anti-social conduct.20
The Model Penal Code conception of insanity is an important advance on
the M’Naghten conception, precisely because it recognizes an independent
volitional dimension to sanity and so recognizes a wider conception of
insanity as involving significant impairment of either cognitive or volitional
competence.
Recognizing the volitional dimension of normative competence may
require revising the rationality or practical reason conceptions of responsi-
bility employed by criminal law theorists such as Michael Moore and
Stephen Morse. Strictly speaking, rationality conceptions of normative
competence need not reject the volitional dimension of normative compe-
tence. There might be more to rationality than correct belief or knowledge.
For instance, one might not count as practically rational unless one’s
appetites and passions are sufficiently under control to enable one to
conform one’s will to one’s normative judgment.
As far as we can tell, Moore is noncommittal on this issue and could
agree with these claims about the importance of the volitional dimension of
normative competence, folding them into claims about rational capacities.
However, Morse is skeptical about the volitional dimension of normative
competence. In part because his skepticism finds echoes in Fischer and
Ravizza’s treatment of reasons-reactivity, it is worth considering his com-
plaints about the volitional dimension in some detail.
In his essay “Uncontrollable Urges and Irrational People,” Morse critic-
ally discusses proposals to treat wrongdoers with irresistible impulses as

20
American Law Institute, Model Penal Code }4.01, emphasis added. The Model
Penal Code is a model statutory text of fundamental provisions of the criminal law, first
developed by the American Law Institute in 1962 and subsequently updated in 1981.
The MPC was intended to serve as a model for local jurisdictions drafting and revising
their criminal codes. Notice three differences between MPC and M’Naghten: (a) unlike
M’Naghten, MPC includes volitional, as well as cognitive, capacities in its conception of
insanity; (b) whereas M’Naghten makes complete incapacity a condition of insanity,
MPC makes substantial incapacity a condition of insanity; and (c) whereas M’Naghten
requires only capacity for normative recognition for sanity, MPC requires capacity for
normative appreciation. Here, we focus only on (a), but all three points of contrast
between MPC and M’Naghten are potentially significant.
 Fairness and Architecture of Responsibility 299

excused for lack of control. He claims, not implausibly, that many with
emotional or appetitive disorders are nonetheless responsible, because they
retain sufficient capacity for rationality (2002: 1040). In discussing excuses
that appeal to uncontrollable urges, he makes clear that his conception of
rationality excludes volitional components.
This . . . Essay claims that our ambivalence about control problems is caused by a
confused understanding of the nature of those problems and argues that control or
volitional problems should be abandoned as legal criteria [for excuse] (2002: 1054).
But why should we abandon a volitional dimension to normative compe-
tence and control? Morse focuses on the alleged threat posed by irresistible
urges and makes several (incompatible) claims about them: (1) we cannot
make sense of irresistible urges, (2) we cannot distinguish between genu-
inely irresistible urges and urges not resisted, (3) there are no irresistible
urges, because under sufficient threat of sanction we can resist any strong
urge.
Morse focuses on irresistible urges. This is already problematic, because
it ignores the varieties of volitional impairment, which include not just
irresistible urges but also paralyzing fears, depression, and systematic weak-
ness caused by damage to the prefrontal cortex.
But consider what Morse does say about irresistible urges. He argues
against the claim made by the majority in Kansas v. Crane that civil detention
be limited to those who are dangerous to themselves or others on account of
control problems that are the result of mental abnormality.21 Morse plausibly
claims that mental disease or abnormality, as such, is irrelevant to excuse
(2002: 1034, 1040). All that mental abnormality signals is something about
the cause of urges; by itself, it does not signify anything about the agent’s
capacities, and so cannot serve as an excuse (2002: 1040). That is surely right,
but the Court in Crane did not say that mental abnormality was sufficient
for excuse, but at most that it was necessary.22 What was critical, the Court
claimed, was whether the urges were sufficiently irresistible to present a
control problem. A control problem can be understood as a lack of
relevant volitional capacities. So the Court is just not making the fallacious
argument that Morse rightly criticizes. Demonstrating that abnormality does
not imply incapacity does not show that responsibility does not require
volitional capacity. So Morse’s criticism of the abnormal cause requirement
does not support a rationality conception of agency that eschews volitional
capacities.

21
Kansas v. Crane, 534 US 407 (2002).
22
Insofar as the Court is requiring a mental abnormality, perhaps defined on the
disease model, we disagree. It is neither a necessary nor sufficient condition for excuse.
300 David O. Brink and Dana K. Nelkin

Morse goes on to claim that the idea of irresistible urges is not coherent
and that, even if it was, we could not distinguish between irresistible urges
and urges not resisted (1994: 1601, 2002: 1062). This is the problem of
distinguishing between can’t and won’t. Finally, he asserts that even if we
could distinguish between irresistible urges and urges not resisted, we
would find that in actual cases the urges in question would be resistible.
In discussing whether an addict’s cravings are irresistible, Morse argues that
they are not because if you hold a gun to the addict’s head and tell him that
you’ll shoot him if he gives in, he can resist (2002: 1057–8, 1070). This is
reminiscent of the sort of weak reactivity that Fischer and Ravizza defend
and that Kant requires in The Critique of Practical Reason.
Suppose that someone says his lust is irresistible when the desired object and
opportunity are present. Ask him whether he would not control his passion if, in
front of the house where he had this opportunity, a gallows were erected on which
he would be hanged immediately after gratifying his lust. We do not have to guess
very long what his answer would be. (Kant 1788: 30)
But these different complaints about irresistible urges are all resistible.
First, there seems to be no conceptual problem with irresistible urges.
We can conceive of paralyzing emotions or irresistible desires, as Mele does,
as emotional states or appetites that stand in the way of implementing the
verdicts of practical reason that are virtually unconquerable and uncircum-
ventable (1990). Resistibility is a modal notion. There is a question about
how unconquerable or uncircumventable impulses must be to be excusing,
and there may be evidential or pragmatic problems about identifying
desires that are genuinely irresistible. But the concept of irresistible desires
does not seem especially problematic.
Second, consider the worry that we cannot reliably distinguish between
an inability to overcome and a failure to overcome such obstacles. First of
all, this is an evidentiary problem, not a claim about the ingredients of
normative competence. Moreover, this evidentiary problem seems no
worse than the one for the cognitive dimension of normative competence,
which requires us to distinguish between a genuine inability to recognize
something as wrong and a failure to form correct normative beliefs or
attend to normative information at hand. Making the distinction between
can’t and won’t is a challenge, but not an insurmountable one, in either the
cognitive or volitional case. For instance, there are neurophysiological tests
for various forms of affective, as well as cognitive, sensitivity, such as
electrodermal tests of empathetic responsiveness (Blair et al. 2005: 49–50).
Finally, consider Morse’s claim that volitional capacity is easily demon-
strated insofar as agents can always resist desires and temptations under
sufficient threat. Morse’s position here bears comparison with that of
 Fairness and Architecture of Responsibility 301

Fischer and Ravizza. As we saw earlier, while they defend moderate reasons-
receptivity, they require only weak reasons-reactivity.23 In defense of weak
reactivity, Fischer and Ravizza claim that reactivity is “all of a piece”—if
you can conform in some cases, even one case, that shows that you can
conform in any case. (1998: 73). Kant and Morse seem to agree. There are
two problems here. First, they want to recognize an asymmetry between
cognitive and volitional capacities. Yet, if reactivity were “all of a piece,”
then why not say the same thing about receptivity? If one can recognize
some moral reasons, one can recognize any. Or if one can recognize them
under some circumstances, then one can recognize them under any. This
would be to accept weak reasons-receptivity, which both Morse and Fischer
and Ravizza reject. Second, they are committed to claiming, at least about
reasons-reactivity, that one can’t have weak responsiveness without having
moderate responsiveness. Anyone who can resist an urge in one extreme
situation can resist it in others. But we see no reason to accept this
psychological stipulation. An agoraphobe might have such a paralyzing
fear of public spaces that she would be induced to leave her home only
under imminent threat of death. There’s no reason to assume that we
cannot have weak reactivity without moderate reactivity (cf. McKenna
2001, Watson 2001, Pereboom 2006, and Todd and Tognazzini 2008).
Our own view is that weak reactivity is simply implausible as a general
reactivity condition on responsibility. Cases in which a person would only
react differently under a threat of imminent death, because of a paralyzing
fear or compulsion, for example, seem to be cases in which we should
excuse.24 If a desire is really only resistible in this one counterfactual case,
then we think that the agent is not responsible, or at least not fully
responsible, in the actual case. That doesn’t mean that we can’t detain
him if he is dangerous to himself or others, but it would mean that it would
be inappropriate to blame and punish him.
On closer inspection, it seems Morse is really ambivalent between two
different kinds of skepticism about the volitional dimension of normative
competence and its significance. In some moments, he denies that there is
any separate volitional dimension to normative competence, beyond the
cognitive dimension. At other times, he recognizes the need for a separate

23
It is worth noting that Fischer’s and Ravizza’s view is doubly asymmetric insofar as
they require receptivity to at least some moral reasons, but require reactivity only to
reasons in general, not necessarily moral ones (1998: 79). We reject this sort of
asymmetry, as well.
24
Mele’s example of the agoraphobe, who will not leave his house, even for his
daughter’s wedding, but would leave it if it were on fire, seems coherently described as
one in which someone is weakly reactive, but nevertheless, not responsible.
302 David O. Brink and Dana K. Nelkin

volitional dimension but claims that it is easily satisfied because volitional

conformity to what one judges right and wrong is “all of a piece.” We hope
to have shown that neither form of skepticism is especially promising.

5. SITUATIONAL CONTROL

An important part of an agent’s being responsible for wrongdoing that she

chose and intended consists in her being a responsible agent. This we have
conceptualized in terms of normative competence and analyzed into cog-
nitive and volitional capacities. Evidence for this view is that one seems to
have an excuse, whether complete or partial, if one’s normative competence
is compromised in significant ways. The most familiar kinds of excuse—
insanity, immaturity, and uncontrollable urges—all involve compromised
normative competence.
But there is more to an agent being culpable or responsible for her
wrongdoing than her being responsible and having intentionally engaged
in wrongdoing. Moreover, excuse is not exhausted by denials of normative
competence. Among the factors that may interfere with our reactive atti-
tudes, including blame and punishment, are external or situational factors.
In particular, coercion and duress may lead the agent into wrongdoing in a
way that nonetheless provides an excuse, whether full or partial. The
paradigm situational excuse is coercion by another agent, as when one is
threatened with physical harm to oneself or a loved one if one doesn’t assist
in some kind of wrongdoing, for instance, driving the getaway car in a
robbery. Though criminal law doctrine focuses on threats that come from
another’s agency, hard choice posed by natural forces seems similarly
exculpatory, as in Aristotle’s famous example of the captain of the ship
who must jettison valuable cargo in dangerous seas caused by an unex-
pected storm (NE 1110a9–12). Situational duress does not compromise
the wrongdoer’s status as a responsible agent and does not challenge her
normative competence, but it does challenge whether she is responsible for
her wrongdoing.
The details of duress are tricky. Some situational pressures, such as the
need to choose the lesser of two evils, may actually justify the agent’s
conduct, as is recognized in necessity defenses. If the balance of evils is
such that the evil threatened to the agent is worse than the evil involved in
her wrongdoing, then compliance with the threat is justified. But in an
important range of cases, coercion and duress seem not to justify conduct
(remove the wrongdoing) but rather to excuse wrongdoing, in whole or in
part. In such cases, where the evil threatened is substantial but less than that
contained in the wrongdoing, the agent’s wrongdoing should be excused
 Fairness and Architecture of Responsibility 303

because the threat or pressure was more than a person could or should be
expected to resist.25 The Model Penal Code adopts a reasonable person
version of the conditions under which a threat excuses, namely, when a
person of reasonable firmness would have been unable to resist, provided
the actor was not himself responsible for being subject to duress (section
2.09).
Whereas the situational aspect of responsibility was recognized by clas-
sical writers, such as Aristotle, Hobbes, and Locke, it has been less promin-
ent in more recent philosophical discussions of responsibility. Perhaps
because of case law and doctrine involving duress, criminal theorists, such
as Moore, and Morse, have clearly recognized the importance of the
situational component of responsibility (Moore 1997: 554, 560–1,
Morse 1994: 1605, 1617, and Morse 2002: 1058). They explain the
rationale for this situational component and the associated excuse of
duress in terms of the fair opportunity to avoid wrongdoing. The idea is
that normatively competent agents, through no fault of their own, due to
external threat or hard choice may lack the fair opportunity to avoid
wrongdoing. In normal cases, this opportunity may just blend into the
background, taken for granted. But in cases of duress it is absent. This not
only explains why duress should be excusing but also alerts us to the
importance of this opportunity in the normal case, where duress is absent.

6 . T W O M O D E L S O F N O R M A T I VE C O M P E T E N C E
AND SITUATIONAL CONTROL

We think that this emerging picture of the architecture of responsibility in

which normative competence and situational control are the two main
elements of responsibility is quite attractive. Others have thought so too.

25
Exactly when duress justifies and when it excuses is an interesting and difficult
question. Much will depend on how the necessity and balance of evils doctrines are
understood. Suppose A threatens to rape B’s loved one if B doesn’t kill C, who is
innocent. On one interpretation, this case fails the balance of evils test (murder is
worse than rape), so it tends to excuse, rather than justify. But this may be less clear if
the balance of evils test is performed using a moral balance employing agent-centered
prerogatives. We can’t get a clear handle on the difference between duress justifications
and duress excuses until we fix the moral conception employed in the balance of evils test.
But we think that it is safe to assume that however exactly the lines are drawn on these
issues about interpreting the balance of evils test there will be some duress excuses. That
is especially plausible, we think, when we recognize that duress and excuse can be scalar.
That is sufficient to justify our architectural assumption that there should be a separate
wing for situational control, whether or not that wing is densely populated.
304 David O. Brink and Dana K. Nelkin

For example, in “Negligence, Mens Rea, and Criminal Responsibility”

H. L. A Hart seems to accept such a view.
What is crucial is that those whom we punish should have had, when they acted, the
normal capacities, physical and mental, for abstaining from what it [the law]
forbids, and a fair opportunity to exercise these capacities. (Hart 1961: 152)
And Moore endorses this idea.
Hart thus subdivides the ability presupposed by his sense of ‘could’ into two
components. One relates to the equipment of the actor: does he have sufficient
choosing capacity to be responsible? The other relates to the situation in which the
actor finds himself: does that situation present him with a fair chance to use his
capacities for choice so as to give effect to his decision? (Moore 1997: 554)
We see two different conceptions of how these two factors relate to
responsibility.
On one conception, normative competence and situational control are
individually necessary and jointly sufficient but independent factors in
responsibility. On this conception, there is an appropriate degree of com-
petence and an appropriate degree of situational control that can be fixed
independently of each other and which are both necessary for responsi-
bility, such that falling short in either dimension is excusing. On this
picture, we assess an agent in each area separately. We figure out whether
she had the relevant capacities (e.g. were they “normal” or “sufficient”), and
then we figure out whether she had the fair opportunity to exercise them.
This has been the conception of the architecture of responsibility that we
have articulated so far. However, an alternative conception of normative
competence and situational control is possible that treats them as individu-
ally necessary and jointly sufficient but at least sometimes interacting. On
this picture, how much and what sort of capacities one needs can vary
according to situational features. So, for example, there might be situations
in which the wrongdoing in question was especially clear, such as a murder
or an assault, and in which there was no significant provocation, duress, or
other hard choice. We might think that culpable wrongdoing, in such cases,
requires less in the form of cognitive or volitional capacities than in cases in
which the normative issues are less clear or in which there is substantial
provocation or duress. Or hold constant the wrongdoing in question and
compare the interaction of situational factors and competence in different
individuals. It’s plausible to suppose that normative competence requires
an ability to make one’s own normative judgments and hold to them
despite temptation, distraction, and peer pressure. It’s also plausible to
suppose that adolescents have less independence of judgment and ability to
resist peer pressure than their adult counterparts. But then we might be
 Fairness and Architecture of Responsibility 305

more excusing of adolescent wrongdoing committed in groups than of

comparable adult behavior committed in groups. If so, differential compe-
tence may explain why the same level of situational pressure may be
excusing for some and not for others (Brink 2004). But then we might
prefer a conception of normative competence and situational control, in
which they are potentially interacting, rather than independent, dimensions
of responsibility. Such a conception would also imply that the requisite
levels of normative competence and situational control are not invariant,
but rather context-dependent.

7 . R E S P O N SI B I L I T Y A S T H E F A I R O P P O R T U N I T Y
TO AVOID WRONGDOING

So far, the conception of responsibility emerging from our discussion is a

two-factor model twice over. We factor responsibility into normative
competence and situational control, and we factor normative competence
into partially independent cognitive and volitional capacities, which we
treat as equally important. This kind of two-factor model seems plausible,
in significant part because it promises to fit our practices of excuse in both
moral assessment and the criminal law pretty well. Perhaps this is adequate
justification. But it would be nice if there were some unifying element to its
structure.
One possible umbrella concept is control (the concept Fischer and
Ravizza associate with reasons-responsiveness). Freedom from coercion
and duress, cognitive competence, and volitional competence all seem to
be aspects of an agent’s ability to control her actions. But control seems
important, at least in part, because it seems unfair to blame agents for
outcomes that are outside their control.
But this suggests that the umbrella concept should perhaps be fairness, in
particular, the fair opportunity to avoid wrongdoing (the concept that
Moore and Morse associate with the situational component), because
failure of either normative competence or situational control violates the
norm that blame and punishment be reserved for those who had a fair
opportunity to avoid wrongdoing.
A related proposal would be to treat the umbrella concept as the fair
opportunity to avoid blame. For, we might also say that fairness requires
that we not blame those who lacked normative competence or situational
control. While the opportunity to avoid blame is no doubt a by-product of
the fair opportunity to avoid wrongdoing, we think that it is the latter
concept that should be primary in our understanding of responsibility. For
306 David O. Brink and Dana K. Nelkin

one thing, if we understand responsibility (as accountability) as the condi-

tion that justifies blame, then appeal to the fair opportunity to avoid blame
threatens to introduce circularity into our account. As we would then be
explaining what makes actions potentially blameworthy in terms of the
fair opportunity to avoid blame. That seems like too small an explanatory
circle. Moreover, making the primary focus on blame seems potentially
narcissistic—too focused on the agent’s moral ledger. It may be that one can
avoid blame if one has the fair opportunity to avoid wrongdoing, but being
responsible does not seem to be primarily about the agent avoiding blame.
Rather, the ability to avoid blame seems consequential on responsibility. We
can explain this, without focusing unduly on the responses of others, if we
treat the fair opportunity to avoid wrongdoing as the primary concept.
If we treat the fair opportunity to avoid wrongdoing as the key to responsi-
bility, we get the following picture of the architecture of responsibility.

Fair
Opportunity to
Avoid
Wrongdoing

Normative Situational
Competence Control

Cognitive Volitional
Competence Competence

One way to see the importance of the fair opportunity to avoid wrong-
doing is to think about why strict liability is problematic. Strict liability
offenses are those for which wrongdoing is sufficient and culpability is not
required. While tort law recognizes some strict liability offenses, it is
significant that the penalties for such torts are civil monetary damages.
Strict liability is extremely problematic within the criminal law, where guilt
results in blame and imprisonment. Indeed, the main strict liability crime is
 Fairness and Architecture of Responsibility 307

statutory rape (consensual intercourse by an adult male with an underage

female), where a reasonable belief that the female was not a minor is not
exculpatory. But this conception of statutory rape is anomalous. Statutory
rape is not treated as a strict liability crime in all jurisdictions, and where it
is so treated, it is widely viewed as morally problematic. Indeed, the Model
Penal Code condemns criminal offenses for which conviction does not
require culpability (section 2.02).26 Why exactly are strict liability crimes
so problematic?
In “Legal Responsibility and Excuses” H. L. A. Hart suggests that the
criminal law conditions liability on culpability out of respect for “the
efficacy of the individual’s informed and considered choice in determining
the future” (1957: 46). A corollary of this concern with individual auton-
omy is the demand for the fair opportunity to avoid wrongdoing. This
principle is at work in support of the fundamental legal principle of legality.
Legality is the doctrine that there should be no punishment in the absence
of public notice of a legal requirement. The principle of legality is usually
defended as part of fair notice. Ex post facto or retroactive criminal law
would be unfair, because it would punish those for failing to conform to
behavioral expectations of which they had not been apprised in advance. Ex
post facto law thus threatens individual autonomy and its demand of fair
opportunity to avoid wrongdoing. But a similar rationale is at work against
strict liability crimes. Just as it would be unfair to convict actors for failing
to conform to standards that had not been promulgated in advance, so too
it would be unfair to convict actors for failing to conform to standards
(promulgated in advance) that they did everything within their power to
obey. Conviction without culpability denies the fair opportunity to avoid
wrongdoing.
Some have suggested that fairness requires normative competence
as a condition of responsibility.27 Others have suggested that the fair

26
For a brief discussion of the anomalous position of strict liability crimes within the
criminal law, see Dressler 2009: ch. 11.
27
See Wallace 1994: 109, 116, 161–6. The substance of our conception of
responsibility agrees with Wallace on many points, including the idea that
responsibility requires normative competence, that normative competence has both
cognitive and volitional dimensions, and that normative requirement is a condition of
the fairness of blame and punishment. But there are important differences. Wallace’s
account of excusing and exempting conditions (chs. 5–6) fails at many points to
distinguish properly between justification and excuse. Also, like Fischer and Ravizza
and other philosophers, Wallace tends to focus on the kind of normative competence
that fairness requires for responsibility, thus ignoring or at least underestimating the
importance of the kind of situational control that fairness also requires for responsibility.
Finally, and perhaps most importantly, Wallace insists on adopting a response-
dependent interpretation of this conception of responsibility insofar as he claims that
308 David O. Brink and Dana K. Nelkin

opportunity to avoid wrongdoing requires situational control. We put these

ideas together and conclude that the guiding principle underlying our
conception of responsibility is the fair opportunity to avoid wrongdoing.
Because the fair opportunity to avoid wrongdoing is the central value in
our conception of responsibility, it is perhaps worth addressing skepticism
about the fair opportunity to avoid wrongdoing. There is a familiar kind of
skepticism about responsibility that appeals to incompatibilism and deter-
minism. One form of skepticism alleges that responsibility presupposes the
Principle of Alternative Possibilities (PAP), requiring that one could have
done other than one did, which determinism shows to be false. Our
conception of responsibility is not immune to these sorts of skeptical
threats, because we understand responsibility in terms of the fair opportun-
ity to avoid wrongdoing and we understand the fair opportunity to avoid
wrongdoing to require alternative possibilities. To have the fair opportun-
ity to avoid wrongdoing, it must be true that when one commits wrong,
one could have done otherwise. There are two kinds of challenge to
PAP. First, PAP may seem to require indeterminism. Second, alternative
possibilities may seem unnecessary for responsibility because of Frankfurt-
style counterexamples in which an agent chooses to act in a normal and
responsible way and in which someone else (the counterfactual intervener)
stands ready to intervene to cause her to act that way in case she should
waver (Frankfurt 1969).
These are familiar philosophical issues about responsibility that are quite
complicated, and this is not the place to defend compatibilism in any detail.
But we can and should indicate the lines along which we would like
to defend our commitments. Our response is to understand the “could
have done otherwise” clause in PAP as consistent with both the existence
of typical Frankfurt scenarios and with determinism. This requires a
compatibilist understanding of the ability to do otherwise. A variety of
such compatibilist accounts have recently been defended.28 The account we
favor is one according to which one has the general ability to do otherwise,
manifested in various cognitive and volitional capacities and opportunities,

we cannot give an account of the conditions under which an agent is responsible

independently of our reactive attitudes and practices of excuse (1994: 19). But while
the reactive attitudes and our practices of excuse may be evidence about the conditions of
responsibility, on our conception, they point to a conception of responsibility grounded
in the fair opportunity to avoid wrongdoing, which itself requires normative competence
and situational control. On this conception, it is possible to give a response-independent
account of the conditions under which the reactive attitudes are justified.
28
For example, there have been several attempts to reconcile PAP with determinism
by understanding the ability to do otherwise in terms of the having of a disposition. See,
for example, Smith 2003, Vihvelin 2004, Fara 2008. For criticism, see Clarke 2009.
 Fairness and Architecture of Responsibility 309

and, at the time of the action, nothing impairs these capacities and
opportunities in a relevant way. Such a condition can be satisfied in
Frankfurt-style cases, and, we claim, in deterministic settings.29 There is,
to be sure, another sense of the ability to do otherwise, in which the agents
in Frankfurt-style cases lack such an ability, but this is not the sense we
believe is required by the fair opportunity to avoid wrongdoing. The
legitimate threats to the opportunity to avoid wrongdoing and, hence,
responsibility come not from the fact that our actions are caused but
rather from specific impairments of cognitive and volitional capacities
and specific kinds of external interference with our ability to conform to
the relevant norms.

8 . PA R T I AL R E S PO N S I B I LI T Y AN D EX C US E

An important feature of our conception of responsibility is that the com-

ponents of normative competence and situational control are scalar con-
cepts that admit of degree. This implies that responsibility is itself scalar
and that partial responsibility is a real phenomenon.30 Insanity and
immaturity are clearly scalar concepts. But so are the ideas of paralyzing
emotions, irresistible urges, and a disabling depression. So too is the
hardness of a choice, posed by coercion or natural forces, and the
question of whether it is more than a reasonable person could resist.
Partial responsibility is reasonably well recognized within our assessments
of moral responsibility. But the recognition of partial responsibility
suggests a basis for criminal jurisprudence reforms.
Criminal law trials are bifurcated into two phases—the guilt phase and, if
guilty, the sentencing phase. Verdicts at the guilt phase are generally bivalent—
guilty or not guilty. There are various possible reasons for a finding of not
guilty. The prosecution may fail to prove some element of actus reus or mens
rea beyond a reasonable doubt, or the defense may provide adequate evidence
of justification, by demonstrating necessity, or excuse, by demonstrating
duress or insanity. But the factors determining both justification and excuse
would appear to be scalar. One might conclude that in an ideal world guilt

29
The skeptic might reply that a determined past prevents one from exercising one’s
capacities. But, as Nelkin has argued (2011: 64–79), the fact that the past is deterministic
does not make it interference or prevention in the relevant sense any more than is the
counterfactual intervener in a Frankfurt scenario. See Wolf (1990: 94–116) for a
different kind of defense of the conclusion.
30
One recent discussion of the scalar character of reasons-responsiveness is Coates
and Swenson 2012. For a defense of the scalar character of immaturity and the resulting
partial responsibility appropriate for juvenile justice, see Brink 2004.
310 David O. Brink and Dana K. Nelkin

verdicts should be multivalent reflecting the degree of justification or excuse

the defendant possessed. Insofar as our bivalent system recognizes only
perfect necessity as a justification and complete or substantial lack of norma-
tive competence as an excuse, its conception of guilt and punishment is over-
inclusive and punishment is not proportional to just deserts. Perhaps there
are legitimate pragmatic reasons for opposing an infinitely multivalent
system. But even a trivalent system that recognized partial guilt, as well as
guilty and not-guilty verdicts, would provide more retributive justice.
Justifications do not bear on responsibility, as excuses do. So we focus on
partial excuse, rather than partial justification. While some European
criminal justice systems recognize partial excuse, American criminal law
does not. Instead, we recognize only full excuses, relegating considerations
that are relevant to excuse but fall short of full excuse to the sentencing
phase of a trial.31 Some commentators have called for the adoption of a
trivalent system for the guilt phase in which the accused can be found
guilty, not guilty, or guilty but partially responsible.32
A trivalent system might be superior to our current bivalent one. Sup-
plementing a bivalent system with sentence mitigation is an imperfect
proxy for recognizing partial excuses. Though one can mitigate sentences
to respond to various nonculpability factors (e.g. mercy or forgiveness),
mitigation is a poor response to reduced culpability. First, mitigation is not
always possible where there are mandatory sentencing rules. Second, miti-
gation, where possible, is largely discretionary. But it shouldn’t be discre-
tionary whether to hear and evaluate considerations that would tend to
excuse. If complete incompetence is relevant to the guilt phase and a full
excuse, then partial competence is also relevant to the guilt phase and a
partial excuse, because it appeals to the very same factors that a full excuse
appeals to, only to a reduced degree. In effect, the claim is that excuses,
whether full or partial, belong to the guilt phase of a trial and that sentence
mitigation should be confined to nonculpability considerations (or, per-
haps, in a trivalent system, to those culpability considerations that fall
below the threshold necessary for partial excuse).
Our conception of the architecture of responsibility recognizes the
scalar nature of the components of responsibility and so justifies findings
of partial responsibility. If, as the positive retributivist believes, punish-
ment and the reactive attitudes should be proportional to culpable

31
The one exception to the generalization that American criminal law does not
recognize partial excuse is the provocation defense, under which intentional homicides
committed with adequate provocation reduce to an offense of voluntary manslaughter.
32
See Morse 2003. Morse’s position is especially interesting, because this defense of
partial excuse is a reversal of his previous skepticism about partial excuse (Morse 1984).
 Fairness and Architecture of Responsibility 311

wrongdoing, then just deserts require that we recognize partial excuses.

Otherwise, we punish more than just deserts. As we have seen, not all
retributivists are of this sort. Yet all retributivists, positive and negative,
believe that punishment and the reactive attitudes should not be out of
proportion to culpable wrongdoing. If this is correct, then justice requires
that we recognize partial excuses. Presumably, our moral assessments and
practices of moral blame can and do already reflect recognition of partial
excuses, albeit imperfectly. The conception of responsibility we have
sketched suggests that we should recognize a doctrine of generic partial
excuse in the criminal law as well and abandon that doctrine only if and
when it proves administratively unfeasible or to have worse retributive
outcomes overall.

REFERENCES

Aharoni, E., Sinnott-Armstrong, W., and Kiehler, K. (2012). “Can Psychopathic

Offenders Discern Moral Wrongs? A New Look at the Moral/Conventional
Distinction.” Journal of Abnormal Psychology 121: 484–97.
American Law Institute. (1981). Model Penal Code, ed. M. Dubber (New York:
Foundation Press, 2002).
Aristotle. (NE) Nicomachean Ethics. Trans. T. Irwin (Indianapolis, IN: Hackett,
1985).
Bedau, H. and Kelly, E. (2010). “Punishment.” The Stanford Encyclopedia of
Philosophy <http://plato.stanford.edu/archives/spr2010/entries/punishment>.
Blair, J., Mitchell, D., and Blair, K. (2005). The Psychopath: Emotion and the Brain
(Oxford: Blackwell).
Brink, D. (2004). “Immaturity, Normative Competence, and Juvenile Transfer: How
(Not) to Punish Minors for Major Crimes.” Texas Law Review 82: 1555–85.
Clarke, R. (2009). “Dispositions, Abilities to Act, and Free Will: The New
Dispositionalism.” Mind 118: 323–51.
Coates, D.J. and Swenson, P. (forthcoming). “Reasons-Responsiveness and Degrees
of Responsibility.” Philosophical Studies.
Damasio, A. (1994). Descartes’ Error: Emotion, Reason, and the Human Brain
(New York: Putnam).
Dressler, J. (2009). Understanding Criminal Law, 5th edn. (Boston: LexisNexis).
Duff, A. (2008). “Legal Punishment.” The Stanford Encyclopedia of Philosophy
<http://plato.stanford.edu/archives/fall2008/entries/legal-punishment>.
Fara, M. (2008). “Masked Abilities and Compatibilism.” Mind 117: 843–65.
Fine, C. and Kennett, J. (2004). “Mental Impairment, Moral Understanding,
and Criminal Responsibility.” International Journal of Law and Psychiatry 27:
425–43.
Fischer, J. and Ravizza, M. (1998). Responsibility and Control (New York:
Cambridge University Press).
312 David O. Brink and Dana K. Nelkin

Frankfurt, H. (1969). “Alternate Possibilities and Moral Responsibility,” reprinted

in Frankfurt (1988).
—— (1988). The Importance of What We Care About (New York: Cambridge
University Press).
Hart, H L. A. (1957). “Legal Responsibility and Excuses,” reprinted in Hart (1968).
—— (1961). “Negligence, Mens Rea, and Criminal Responsibility,” reprinted in
Hart (1968).
—— (1968). Punishment and Responsibility (Oxford: Clarendon Press).
Kant, I. (1788). Critique of Practical Reason. Trans. L.W. Beck (Indianapolis, IN:
Bobbs-Merrill, 1956) (Prussian Academy pagination).
Kripke, S. (1982). Wittgenstein on Rules and Private Language (Cambridge, MA:
Harvard University Press).
McKenna, M. (2001). “Review of Fischer and Ravizza, Responsibility and Control .”
Journal of Philosophy 98: 93–100.
—— (2012). “Reasons-Responsiveness, Agents, and Mechanisms” (this volume).
Mele, A. (1990). “Irresistible Desires.” Noûs 24: 455–72.
Moore, M. (1997). Placing Blame (Oxford: Clarendon Press).
Morse, S. (1984). “Undiminished Confusion in Diminished Capacity.” The Journal
of Criminal Law and Criminology 75: 1–55.
—— (1994). “Culpability and Control.” University of Pennsylvania Law Review
142: 158–660.
—— (2002). “Uncontrollable Urges and Irrational People.” Virginia Law Review
88: 1025–78.
—— (2003). “Diminished Rationality, Diminished Responsibility.” Ohio State
Journal of Criminal Law 1: 289–308.
Nelkin, D. (2011). Making Sense of Freedom and Responsibility (Oxford: Clarendon
Press).
Pereboom, D. (2006). “Reasons-Responsiveness, Alternative Possibilities, and
Manipulation Arguments against Compatibilism.” Philosophical Books 47: 198–
212.
—— (2012). “Freedom and Punishment.” In The Future of Punishment, ed.
T. Nadelhoffer (Oxford: Clarendon Press).
Scanlon, T. M. (2008). Moral Dimensions: Permissibility, Meaning, and Blame
(Cambridge, MA: Harvard University Press).
Sidgwick, H. (1907). The Methods of Ethics, 7th edn. (London: Macmillan).
Smith, M. (2003). “Rational Capacities, or How to Distinguish Recklessness,
Weakness, and Compulsion.” In Weakness of Will and Practical Irrationality,
eds. S. Stroud and C. Tappolet (Oxford: Clarendon Press).
Strawson, P. F. (1962). “Freedom and Resentment,” reprinted in Watson (1982).
Todd, P. and Tognazzini, N. (2008). “A Problem for Guidance Control.” Philo-
sophical Quarterly 58: 685–92.
Vihvelin, K. (2004). “Free Will Demystified: A Dispositional Account.” Philosoph-
ical Topics 32, 427–50.
Wallace, R. J. (1994). Responsibility and the Moral Sentiments. (Cambridge, MA:
Harvard University Press).
 Fairness and Architecture of Responsibility 313

Watson, G., (ed.) (1982). Free Will (New York: Oxford University Press).
—— (1987). “Responsibility and the Limits of Evil,” reprinted in Watson (2004).
—— (1996). “Two Faces of Responsibility,” reprinted in Watson (2004).
—— (2001). “Reasons and Responsibility,” reprinted in Watson (2004).
—— (2004). Agency and Answerability (New York: Oxford University Press).
—— (2011). “The Trouble with Psychopaths.” In Reasons and Recognition: Essays
on the Philosophy of T. M. Scanlon (New York: Oxford University Press).
Williams, B. (1976). “Moral Luck,” reprinted in Williams (1981).
—— (1981). Moral Luck (Cambridge: Cambridge University Press).
Wolf, S. (1990). Freedom Within Reason (New York: Oxford University Press).
This page intentionally left blank
 Index

ability to do otherwise 144–5, 151–2, appropriateness of 227, 228, 233,

166, 168–70, 172–3 238–41, 242, 243, 265, 289
the experience of 127–31 fair opportunity to avoid 237–8,
action(s) 305–6
intentional 22, 24, 292, 296–7 modality of 240–1
relation to feeling 98 blameworthiness 289
relation to intention 83–90 judgments of 226–7, 228–9, 239, 289
addiction 263–4 blaming emotions 205–21
agency anger 205, 207, 210–14, 217,
goal-directed 49, 76–7 219 n.23
phenomenology of 126–7, 145–6 appropriateness 205–7, 209–12,
planning 47–8, 50–4 214–20
shared 55–6 degrees of appropriateness 210
temporally extended 54–5 fittingness; see appropriateness
agent(s) indignation 207, 210, 212, 217
goal-directed 48 resentment 205, 207, 209–10, 212,
moral and prudential 2 217, 232, 233, 238, 285–6; see also
nonrational 12, 15, 17, 32 judgments of blameworthiness;
rational 12, 27, 43 reactive attitudes
responsible 292 Bok, Hilary 26
agent-centered facts 248, 260 Booth v. Maryland 246–7
agent-dismay 116–17 Boyle, Matt 71
agent-regret 95–8, 100, 105, 106–10, Brewer, Talbot 25
115, 117 Buridan’s ass 50
akrasia 227–9; see also weakness of will
appraisal 205–6, 208–9, 211–12, Campbell, C.A. 127–8, 143
214–19 causation
communication of 205–6, 210–20 metaphysics of 172
Aristotle 104, 153 censure 289
As-if rule 267, 271–2 Chamberlain, Neville 96, 101
attitude(s) Churchill, Winston 96, 101
aiming 74–9, 89 Clarke, Randolph 166, 169
executive 73–4 cognitivism
practical 72–3 cognitivist theory of the
Averill, James 208 emotions 96, 113
problems for 97–8
Baier, Annette 71 compatibilism 5, 126–7, 144–5, 193,
Baron, Marcia 108–9 203, 220, 288, 308–9
Baumeister, Roy 214 leeway compatibilism 166
behavior source compatibilism 152–4, 166,
goal-directed 13–16, 27, 33, 43 168, 172, 176–8
instance of 12 control 305; see also volitional
instrumentally rational 13, 16, 22, dimensions of reasons-
25, 43–4 responsiveness
voluntary 14 criminal jurisprudence 309–11
blame 8, 10, 185, 189, 190, 194, 197, criminal responsibility 293; see also
203, 226–7, 289 responsibility
316 Index

criminal responsibility (cont.) Frankfurt, Harry 2, 16, 62, 65, 151–2,

as a guide to moral responsibility 265, 166, 168–70, 172, 175, 177–8,
284–5, 295–6 263–4, 308–9
culpability Frankfurt-style cases 308–9
considerations that mitigate 291 free action; see freedom
freedom 152, 154–6, 159, 161–2, 166,
D’Arms, Justin 209 169, 176–8, 184, 263–5
Darwall, Stephen 218 leeway freedom 152, 158, 178
deep self 263–4 of sanity 263
deliberation 53, 61–3, 147–9, 213 source freedom 152–3, 170, 176
delusions 265, 267, 269, 279 free will 152, 163, 165, 167, 169–70,
persecutory 270–1 177, 288
depression 279
postpartum, see postpartum psychosis Gachter, Simon 213
desert 290 Gibbard, Allan 61
desires 20–2, 30–4, 37, 72–3; see also Gilbert, Margaret 60
impulses Ginet, Carl 160–1, 163–4
higher order 177–8 Grünbaum, Adolf 128–9, 143
irresistible 298–302 guilt 99–100, 107, 111–12, 115,
lower order 177–8 119, 214
noninstrumental 27, 29
nonrational 14, 44 Hadfield, James 267
objects of 70 Haidt, Jonathan 213
determinism 4–5, 127, 143–6, 152, hallucinations 265, 267, 279
178–9, 184, 187, 192, 199, 206, command 268–70
288, 308 Hart, H. L. A. 304, 307
as causal completeness 131, 133–4 holding responsible 185, 188, 190–1,
dispositions 166–7, 169–70 193 n.4, 205, 217, 220
as finked 166–8 honor killings 274–5
duress 302–3 Horgan, Terry 145–6
Hume, David 101, 127
emotion(s) 187–8, 195–6, 205, 220
interpersonal 188, 193, 197–8 If-only rule 267, 271–2
moral 188, 190, 193, 201 ignorance
and normative competence 296, 297 affected 276–7, 278, 280
as not admirable 113–15 circumstantial 225–6, 227, 235, 243
as unfitting 113 culpable 227, 229–30, 232,
as unreasonable 113–14 234, 243
evaluative judgments 101 moral 225–6, 227, 229–30, 232, 236,
expressed implicitly through 238, 243
behavior 233, 234–6, 238–40, imagination
241 n.18, 243 affected lack of 276
excuse(s) 186, 192–3, 201–2, 290–2, impulses; see also desires
302–3 behavior-determining 15
mental abnormality as 291, 299 nonrational 14–15, 37
partial 310–11 practical 12, 27, 29, 42–3
exemptions 186–7, 192–3, 201–2 incompatibilism 127–9, 143, 145–9,
220, 308
Fara, Michael 166–7 source incompatibilism 152, 178–9
fatalism 143 indeterminism 149, 308
Fehr, Ernst 213 insanity 297–8; see also delusions;
Fischer, John Martin 152, 156–63, hallucinations; mental abnormality
166–70, 174–6, 179–80, 206, 216, as excuse; psychosis; mental illness
293, 294, 301 and religion 269–70
 Index 317

intentions 50–1, 71–2, 79–83, 90–3, partial conception of 249, 252, 257,
297 259, 260–1
attributions of 70–1, 90 moral disagreement 236–7, 239–40;
future-directed 17 see also group/cultural differences in
present-directed 17 values
shared 56–60, 71 moral intransigence 265, 271, 275, 278,
internal reasons 230–1, 237–41 279, 280
It’s a Wonderful Life 229, 232 moral luck 231, 246, 260–1
moral responsibility 6–8, 151–2, 166,
Jacobson, Dan 209 168–70, 172, 175, 177–8, 184–5,
John, Kim 269, 270–1 187, 189–93, 195–207, 213,
JoJo 264, 271–2 216–21, 260, 293; see also
justification 290, 302 responsibility
as accountability 206, 219 n.24
Kansas v. Crane 299 as attributability 206, 217
Kant, Immanuel 30 as blameworthiness 226, 233–4
Kapitan, Tomis 147 as playing a certain causal role 233–4
Korsgaard, Christine 29, 31–2, 34–6, 43 rational justification of 184–5, 187,
192, 194, 199–204
Lafferty, Ronald 269–71 skepticism about 184, 186–7, 192–3,
Lavin, Doug 71 199, 201–4, 226, 228–9, 234,
Lazarus, Richard 207–8 290 n.8, 308
legality 307 moral sentiments 189
Lehrer, Keith 127 Morris, Norval 254
Levy, Neil 219, 230–1, 237–8, 242–4 Morse, Stephen 298–302, 303
Lewis, David 166
libertarianism 4–5, 126–7, 129, 148 Nahmias, Eddy 129–30, 143
Locke, John 53, 63 naturalism 184–7, 189, 197–8, 201–2
token naturalism 186–9, 193, 200–2
M’Naghten test; see McNaughtan Rule type naturalism 187, 189, 192–4,
McKenna, Michael 218 199–202
Macnamara, Coleen 217 new dispositionalism 166–70, 174
McNaughtan, Daniel 267 Nichols, Shaun 208
McNaughtan Rule 266, 272–3, 295, Nils, Christie 258
297–8 normative competence 288, 292–5,
Mele, Alfred 165 304–5
mental disorder; see mental illness normative judgments 17–18, 25, 33, 35
mental illness 265, 266, 268, 279; see also
delusions; hallucinations; insanity; O’Connor, Timothy 128, 143
mental abnormality as excuse; objective attitude 186
psychosis obligation(s) 185, 190, 194, 197
distinct from insanity 266 conflicting 99, 114
Mill, John Stuart 128–9, 143 obsessive-compulsive disorder 19
mitigation 310
Model Penal Code 298, 303 Pereboom, Derk 207, 218–19
Moody-Adams, Michelle 276 personal culpability 248
Moore, Michael 298, 303, 304 relationship to desert 248–9, 251, 255,
moral character 266, 268 256–7
moral desert personal identity 53, 63
compatibilist theories of 248 Principle of Alternative Possibilities
impartial conception of 248–9, 251, (PAP) 308–9
252, 258 Prinz, Jesse 208
incompatibilist theories of 248 psychological realism 98, 120
318 Index

psychological states 12, 22 agency principle 95, 106, 112,

psychopathy 279 n.16, 296 113, 116
psychosis 267, 279 agent-; see agent-regret
postpartum 268 as about loss 110, 118–19
punishment 218, 286, 289 constitutive thought of 95
altruistic 213 defined 104
proportionality of 253–6, 259 distinguished from dismay 116–17,
122
Railton, Peter 26 distinguished from guilt 99–100
rationality expressions of 100–1
norms of 54 as motivating 105
Ravizza, Mark 156–63, 166–70, 174–6, policy change 104–5, 111, 115,
179–80, 216, 293, 294, 301 117, 119
Raz, Joseph 34 rationality of 97–100, 106, 112–17,
reactive attitudes 184–90, 192–203, 217 118, 121
n.21, 227, 232, 285–8, 291; see also residue principle 98–9, 112, 117–18,
judgments of blameworthiness; 121–2
resentment as sentiment 100, 102, 104, 110–11
appropriateness of 286, 287–8 stable recalcitrance 103, 111, 117
moral vs. nonmoral 188 standing 100
reasoning responsibility 263–5; see also moral
instrumental 26 responsibility; criminal responsibility
reasons as accountability 286–7, 289
mind-independent 15 as attributability 286
practical 28–30, 34, 37, 43 epistemic condition on 265
reasons-responsiveness 153–5, 158–9, inhibited by cultural values 275–8,
161, 163, 166–71, 173, 176–7, 179, 280
292, 305 partial 309–11
agent-based theories 166, 168, restorative justice 257–8, 259
170, 176 retributivism 259, 288, 310–11
cognitive dimensions of 292, 294–6 legal 289–90
degrees of 293–5 moral 290
mechanism-based theories 159–63, Reznick, Lawrie 268
168, 170, 175–6, 179–80 Rosen, Gideon 225–9, 233–4
mechanism individuation 161–4,
170, 180 sanction 205–6, 213–16, 218–20, 289;
mesh theory 177–9 see also punishment
moderate 156–8, 160, 167–71, 294 deserved 206–7, 215 n.18, 218–19
norm-responsiveness 293 fairness of 215–16
receptivity and reactivity 156–7, Sartorio, Carolina 172
171–2, 294; see also cognitive Sartre, Jean Paul 49, 52, 62
dimensions of reasons- Searle, John 60, 127, 143
responsiveness; volitional second-order volitions 263, 264
dimensions of reasons- self-alienation 14, 17, 24, 27, 37, 42
responsiveness self-governance 60–1
strong 156, 294 sentiments 102
volitional dimensions of 292, 294–5, distinguished from emotion 103, 116
296–302 Setiya, Kieran 71
weak 156, 294 shame 197, 200 n.9
regret Shaver, P. 208
acting without thinking 103–4, 111, situational control 288, 302–3, 304–5
117 slavery 276–7
 Index 319

Smart, J. J. C. 64, 144 group/cultural differences in 265, 278,

Smith, Angela 211, 217–18, 234–5 279 n.17, 280; see also moral
Smith, Craig 207–8 disagreement
Smith, Holly 225 inferentially available 274, 275, 278,
Smith, Michael 22–3, 25, 166–7 280
Sommers, Tamler 207, 219 Van Inwagen, Peter 147
Stocker, Michael 98–9, 102, 105–6, 110, Velleman, David 26, 37–41, 61
113, 118–22 vice 230–1
Strawson, Galen 218–19 victim impact
Strawson, P. F. 6–7, 14–16, 188–9, relevance for judgments of
192–4, 196, 198–9, 204–5, 221, desert 251–2, 253, 258–9
285–6, 287–8, 291 victim impact statements (VIS) 246–7
Strawson’s thesis 287–92 Vihvelin, Kadri 144, 166–7
strict liability 306–7 Von Hirsch, Andrew 254–6

Taylor, Gabrielle 108 Wallace, R. Jay 7, 17, 160, 207, 217,

Taylor, Richard 147 307 n.27
Thompson, Michael 55, 71 Watson, Gary 160, 206, 217,
Tognazzini, Neal 206 219 n.24
Tomasello, Michael 55 welfarism 195
will
ultimatum games 213–14 good 186
uncertainty ill 186, 208, 210 n.10, 221
phenomenology of 131, 140, 142, 145 Williams, Bernard 7, 30, 32, 95–122,
utilitarianism 195, 200 n.9 184–5, 191 n.3, 194–6, 204, 240–1,
286
value Wolf, Susan 263–5, 271, 272, 275–6
conflicts of 99, 114, 117–22 wrongdoing
monism vs. pluralism 122 n.30 fair opportunity to avoid 303, 305–9
values knowing 229, 230, 231–2, 292–3
capacity for evaluating and changing unwitting 226, 227, 228, 229, 232,
them 271, 278, 280 234, 236, 237; see also ignorance
cultural 277–8; see also responsibility
inhibited by cultural values Yates, Andrea 267, 268
divergent values as excusing
condition 265, 267, 272, 274–5 Zedner, Lucia 258