Scientific Approaches to the Philosophy of Religion

Palgrave Frontiers in Philosophy of Religion

Series Editors: Yujin Nagasawa and Erik J. Wielenberg

Titles Include:
Yujin Nagasawa (editor)
SCIENTIFIC APPROACHES TO THE PHILOSOPHY OF RELIGION

Forthcoming titles

Gregory Dawes and James Maclaurin (editors)

COGNITIVE SCIENCE AND RELIGION
Trent Dougherty
THE PROBLEM OF ANIMAL PAIN:
A Theodicy for All Creatures Great and Small
Aaron Rizzieri
PRAGMATIC ENCROACHMENT, RELIGIOUS BELIEF AND PRACTICE

Also by Yujin Nagasawa

THE EXISTENCE OF GOD:
A Philosophical Introduction
GOD AND PHENOMENAL CONSCIOUSNESS:
A Novel Approach to Knowledge Arguments
NEW WAVES IN PHILOSOPHY OF RELIGION (co-editor with Erik Wielenberg)
THERE'S SOMETHING ABOUT MARY:
Essays on Phenomenal Consciousness and Frank Jackson's Knowledge Argument
(co-editor with Peter Ludlow and Daniel Stoljar)
Scientific Appr oache s to
the Philo sophy of Religion
Edited by

Yujin Nagasawa
University ofBirmingham, UK

Palgrave
macmillan
*
Selection and editorial matter © Yujin Nagasawa 2012
Chapters © their individual authors 2012
Softcover reprint of the hardcover 1st edition 2012
All rights reserved. No reproduction, copy or transmission of this
publication may be made without written permission.
No portion of this publication may be reproduced, copied or transmitted
save with written permission or in accordance with the provisions of the
Copyright, Designs and Patents Act 1988, or under the terms of any licence
permitting limited copying issued by the Copyright Licensing Agency,
Saffron House, 6-10 Kirby Street, London EC1N 8TS.
Any person who does any unauthorized act in relation to this publication
may be liable to criminal prosecution and civil claims for damages.
The authors have asserted their rights to be identified as the authors of this
work in accordance with the Copyright, Designs and Patents Act 1988.
First published 2012 by
PALGRAVE MACMILLAN
Palgrave Macmillan in the UK is an imprint of Macmillan Publishers Limited,
registered in England, company number 785998, of Houndmills, Basingstoke,
Hampshire RG21 6XS.
Palgrave Macmillan in the US is a division of St Martin's Press LLC,
175 Fifth Avenue, New York, NY 10010.
Palgrave Macmillan is the global academic imprint of the above companies
and has companies and representatives throughout the world.
Palgrave® and Macmillan® are registered trademarks in the United States,
the United Kingdom, Europe and other countries.
ISBN 978-1-349-33187-1 ISBN 978-1-137-02601-9 (eBook)
DOI 10.1007/978-1-137-02601-9
This book is printed on paper suitable for recycling and made from fully
managed and sustained forest sources. Logging. pulping and manufacturing
processes are expected to conform to the environmental regulations of the
country of origin.
A catalogue record for this book is available from the British Library.
A catalog record for this book is available from the Library of Congress.
10 9 8 7 6 5 4 3 2 1
21 20 19 18 17 16 15 14 13 12
Contents

Series Preface vii

Acknowledgements viii
Notes on Contributors ix

Introduction 1
Yujin Nagasawa

Part I Divine Attributes

1 The Necessity of God and the Psychology of
Counterfactual Thinking 11
Robin Le Poidevin

2 Why Would Anyone Believe in a Timeless God?

Two Types of Theology 28
Benjamin Murphy

Part II God, Creation and Evolution

3 Darwin's Argument from Evil 49
Paul Draper

4 Attributing Agency:
Fast and Frugal or All Things Considered? 71
Graham Wood

Part III God and the Universe

S On Non-Singular Space-times and the
Beginning of the Universe 95
William Lane Craig and fames D. Sinclair

6 The Theistic Multiverse:

Problems and Prospects 143
Klaas f. Kraay

v
vi Contents

Part IV Religious Beliefs

7 How Relevant Is the Cognitive Science of Religion to
Philosophy of Religion? 165
David Leech and Aku Visala
8 The Rationality of Classical Theism and Its Demographics 184
T.f. Mawson

Part V Religious Tolerance and Disagreement

9 Coercion, Consequence and Salvation 205
Steve Clarke
10 Polarized Yet Warranted Christian Belief 224
David Efird

Part VI The Compatibility of Science and Religion

11 Freedom, Science and Religion 237
Katherin A. Rogers
12 The Compatibility of Science and Religion:
Why the Warfare Thesis Is False 255
Michael Ruse

Index 275
Series Preface

The philosophy of religion has experienced a welcome re-vitalization

over the last fifty years or so and is now thriving. Our hope with the
Palgrave Frontiers in Philosophy of Religion series is to contribute to the
continued vitality of the philosophy of religion by producing works that
truly break new ground in the field.
Accordingly, each book in this series advances some debate in the phi-
losophy of religion by offering a novel argument to establish a strikingly
original thesis or approaching an ongoing dispute from a radically new
point of view. Each book accomplishes this by utilising recent devel-
opments in empirical sciences or cutting-edge research in foundational
areas of philosophy, or by adopting historically neglected approaches.
We expect the series to enrich debates within the philosophy of reli-
gion both by expanding the range of positions and arguments on offer
and establishing important links between the philosophy of religion
and other fields, including not only other areas of philosophy but the
empirical sciences as well.
Our ultimate aim, then, is to produce a series of exciting books that
explore and expand the frontiers of the philosophy of religion and con-
nect it with other areas of inquiry. We are grateful to Palgrave Macmillan
for taking on this project as well as to the authors of the books in the
series.

Yujin Nagasawa
Erik]. Wielenberg

vii
Acknowledgements

This volume originated in my project 'Anselmian Perfect-Being Theol-

ogy and the Cognitive Science of Religion', which was undertaken as
part of the Cognition, Religion and Theology Project at Oxford Univer-
sity, funded by the John Templeton Foundation. I would like to thank
them for their generous financial and academic support. It should be
noted, however, that the views expressed in this volume are not neces-
sarily those of the Cognition, Religion and Theology Project, Oxford
University or the John Templeton Foundation. I would also like to
thank Priyanka Gibbons, Melanie Blair and other staff of Palgrave
Macmillan for their impeccable editorial support. I am also grateful
to three anonymous reviewers for helpful comments and constructive
suggestions.

Yujin Nagasawa

viii
Notes on Contributors

Steve Clarke is James Martin Fellow in the Institute for Ethics and
Science, Oxford Martin School and Faculty of Philosophy, University
of Oxford, UK, and Senior Research Fellow in the Centre for Applied
Philosophy and Public Ethics, Charles Sturt University, Australia.

William Lane Craig is Research Professor of Philosophy at Talbot School

of Theology, USA.

Paul Draper is Professor of Philosophy at Purdue University, USA.

David Efird is Senior Lecturer in Philosophy at the University of York,

UK.

Klaas J. Kraay is Associate Professor of Philosophy at Ryerson University,

Canada.

Robin Le Poidevin is Professor of Metaphysics at the University of

Leeds, UK.

David Leech is Postdoctoral Research Fellow at the Universities of Paris

10 and Blaise Pascal University, France.

T. J. Mawson is Fellow and Tutor in Philosophy at St Peter's College,

University of Oxford, UK.

Benjamin Murphy is Associate Professor of Philosophy and Religious

Studies at the Florida State University, Panama.

Yujin Nagasawa is Professor of Philosophy of Religion and Co-Director

of the John Hick Centre for Philosophy of Religion at the University of
Birmingham, UK.

Katherin A. Rogers is Professor of Philosophy at the University of

Delaware, USA.

ix
x Notes on Contributors

Michael Ruse is the Lucyle T. Werkmeister Professor of Philosophy and

Director of the Program in the History and Philosophy of Science at
Florida State University, USA.

james D. Sinclair is a physicist and Senior Combat Analyst for the

United States Navy, USA.

Aku Visala is Research Fellow at the Academy of Finland and the Ian
Ramsey Centre for Science and Religion at the University of Oxford,
UK.

Graham Wood is Lecturer in Philosophy at the University of Tasmania,

Australia.
Introduction
Yujin Nagasawa

1. The God of Abraham, the God of the Philosophers

and the God of the Scientists
The terms the 'God of Abraham' and the 'God of the philosophers' cor-
respond to two traditional approaches to religion. In the first approach,
call it the 'supernatural approach', religious believers try to comprehend
the existence and nature of God primarily through supernatural means,
such as mystical experiences and biblical revelations. In the second
approach, call it the 'conceptual approach', philosophers and theolo-
gians try to comprehend the existence and nature of God primarily
through rational and analytical thinking.
Relatively recently, however, a third approach has emerged. This
approach might be called the 'scientific approach' and the concept
of God understood through this approach the 'God of the scientists'.
On this approach we comprehend the existence and nature of God by
appealing to relevant empirical research and scientific studies. This is
an interesting combination of the above two traditional approaches.
On the one hand, like the supernatural approach, it relies (at least partly)
on empirical observations rather than pure conceptual analysis. On the
other hand, like the conceptual approach, it adopts rational, analytical
thinking. The scientific approach contributes to debates in the phi-
losophy of religion and philosophical theology by addressing relevant
research in natural, behavioural and mathematical sciences.
The present volume collects essays that consider central problems in
the philosophy of religion and philosophical theology through various
forms of the scientific approach (often combined with the conceptual
approach). Each chapter aims to advance a debate in the philoso-
phy of religion by referring to specific findings in recent scientific

1
2 Yujin Nagasawa

research. By utilizing theories and discoveries in cognitive science,

anthropology, developmental psychology, decision theory, biology,
physics and cosmology the present volume tackles several important
problems in the philosophy of religion, such as divine attributes; God,
creation and evolution; God and the universe; religious beliefs; reli-
gious tolerance and disagreement; and the compatibility of science and
religion.

2. Divine Attributes
We typically ascribe many impressive attributes - such as omniscience,
omnipotence, omnibenevolence, timelessness, incorporeality and nec-
essary existence - to God. But why do we do that? There are two
traditional responses to this question. The first such response is that reli-
gious experiences and biblical revelations teach us that God manifests
these impressive attributes. This response corresponds to the supernat-
ural approach mentioned above. The other response is that God is, by
definition, the being than which no greater can be thought and this
definition entails that God has these specific attributes. This response
corresponds to the conceptual approach mentioned above. Over the
last few years, however, researchers in cognitive science, anthropology
and developmental psychology have defended an alternative account.
According to them, people ascribe these impressive attributes to God
because of cognitive and evolutionary facts about human nature. Part
I of this volume addresses the debate on divine attributes by referring to
these scientific studies.
In Chapter 1 Robin Le Poidevin focuses on God's necessary existence.
God's necessary existence is normally interpreted in terms of de dicta or
de re reading of it but each of the interpretations faces difficulties of its
own. Le Poidevin, hence, proposes a third interpretation, a projectivist
interpretation, of understanding it that is free from these difficulties. He
explains, by appealing to the psychology of counterfactual situations,
why the assumption of God's necessary existence is such a natural one
for theists to make.
Cognitive scientists have studied people's ascription of superlative
attributes to specific agents. For example, they have done empirical
research on children's ascription of omnipotence and omniscience to
adults (Barrett, 2004). In Chapter 2, Benjamin Murphy focuses on
another superlative attribute that is commonly ascribed to God, namely,
timelessness. Reviewing relevant empirical research he provides two
distinct reasons for believing in a timeless God. Murphy argues that
 Introduction 3

these reasons are rooted in two distinct types of theological thinking,

which he calls technical theology and experience-enhancing theology,
respectively.

3. God, Creation and Evolution

The dispute over evolution and creation is perhaps the most widely dis-
cussed topic concerning religion and science. William Paley published
his seminal work Natural Theology in 1802, wherein he defended the
design argument for the existence of God. This argument purports to
derive the existence of God as an intelligent designer by referring to cer-
tain features, mainly biological features, of the world. Paley's argument
was well received because there was no fully satisfactory naturalis-
tic account of biological diversity and complexity in his time. Thus
even Richard Dawkins, one of the fiercest contemporary opponents
of creationism, remarks that 'Paley's argument is made with passion-
ate sincerity and is informed by the best biological scholarship of his
day' (Dawkins, 1985, p. 5). However, with the publication of Charles
Darwin's On the Origin of Species in 1859, Paley's theory was no longer
tenable. Darwin succeeded in establishing a very powerful naturalistic
account of the origin of the biological diversity and complexity. Never-
theless, disputes over evolution among theists and atheists continue.
In response to the evolution-creation debate, some theists have
argued that the theory of evolution is compatible with theism. That
is, they argue that theists can consistently hold the claim that while
the theory of evolution is correct God is still the ultimate creator of
the world. However, in Chapter 3, Paul Draper opposes such a claim. He
contends that Darwin's theory of evolution motivates a form of the argu-
ment from evil against the existence of God. Draper examines Darwin's
autobiography and other writings, and maintains that this argument is
strongly suggested by Darwin himself. The argument purports to show
that the theory of natural selection is more compatible with atheism
than it is with theism. Draper strengthens the argument by utilizing
Darwin's ideas, his own ideas and findings in evolutionary biology.
In Chapter 4 Graham Wood approaches the notion of design from
another perspective. He focuses on the attribution of agency, which
seems to be central to religious belief. For example, when theists claim
that God created the universe they effectively attribute agency to the
cause of the universe. Wood offers, relying on evolutionary psychology
and the philosophy of mind, a characterization of the cognitive process
that explains how and why we attribute agency to entities, including
4 Yujin Nagasawa

religious entities. Using Daniel Dennett's insight on the intentional

stance, Wood introduces the attribution of agency in religious contexts,
which he calls the 'spiritual stance'.

4. God and the Universe

Among arguments for the existence of God, the design argument has
attracted the most attention over the last few years. However, there
are many other arguments discussed by philosophers of religion. The
cosmological argument is one of them. Some theists claim that con-
temporary cosmology reinforces the cosmological argument and the
credibility of the existence of God as a creator of the universe. Address-
ing the big bang theory, Pope Pius XII once said, 'true science discovers
God in an ever increasing degree - as though God were waiting behind
every door' (Time, 3 December 1951). Since he made this remark in the
1950s, cosmologists have developed much more advanced theories and
hypotheses concerning the origin of the universe.
In Chapter 5 William Lane Craig and James D. Sinclair focus on the
Kalam cosmological argument. The Kalam argument is a version of
the cosmological argument originally developed by medieval Islamic
philosophers as a conceptual argument for the existence of God as
the first cause of the universe. Craig has revived the argument and
attempted to strengthen it by appealing to contemporary cosmology.
Chapter 5 is a thorough discussion of the cosmological underpinning of
the Kalam argument. Craig and Sinclair claim that space-time, featuring
an initial cosmological singularity, is likely to have a beginning, which
is good news for the Kalam argument. However, they claim, that does
not mean that the argument relies on the truth of the singular nature
of space-time. With reference to relevant scientific findings and theories
they argue that the Kalam argument can be defended even if space-time
is non-singular.
In Chapter 6 Klaas J. Kraay addresses the multiverse hypothesis,
another currently disputed topic in cosmology and theoretical physics.
According to this hypothesis, physical reality consists of multiple uni-
verses, including our own universe. Kraay relates scientists' discussion
of the multiverse hypothesis with the idea in the philosophy of reli-
gion concerning God's actualization of possible universes. He assesses
the claim that multiverse theories can help theists in responding to the
problem of evil and concludes that it reframes the problem of evil, rather
than strengthening theists' or atheists' position on the problem of evil.
 Introduction 5

5. Religious Beliefs
One of the most important questions in the philosophy of religion,
at least in the Judaeo-Christian-Islamic tradition, is the existence of
God. Philosophers of religion have developed a number of arguments
for and against the existence of God over the last few hundreds of
years. Since the 1960s, however, questions concerning the justifica-
tion of, as opposed to arguments for and against, theism seem to have
attracted more attention. That is, philosophers of religion have intensi-
fied their interest in considering whether theists are justified in believing
that God exists, rather than whether they can demonstrate that God
exists. In recent years cognitive scientists and psychologists have con-
ducted empirical studies concerning the naturalistic origin of beliefs
in God and supernatural entities. Based on such studies some have
concluded that people, especially small children, are 'intuitive theists'
because they are naturally predisposed to hold beliefs about supernatu-
ral nonhuman designers (Kelemen, 2004). A challenge for theists here
is to justify theistic beliefs in spite of naturalistic explanations of their
origins.
In Chapter 7, David Leech and Aku Visala focus on the cognitive
science of religion. Atheists often appeal to findings in the cognitive
science of religion to undermine theism while theists try to show the
contrary - some theists even claim that the cognitive science of reli-
gion motivates theism. Leech and Visala examine some implications of
the cognitive science in this context. They argue that it is reasonable to
conclude that the cognitive science of religion supports neither theism
nor atheism and, thus, it should not make any significant impact on the
dispute between theists and atheists.
In Chapter 8, T. J. Mawson focuses on the demographics of theism,
that is, the geographical and cultural distribution of the population that
subscribes to theism. He first raises some worries about the methods
used to determine the distribution. He then considers its implications to
theism. He examines a version of the argument from evil that appeals to
the demographics of theism. Roughly speaking, the argument says that
the actual, uneven distribution of belief in God spread throughout the
world differs from what one would expect were a perfect God to exist.
Mawson explains why this argument fails and concludes that theists
must see the demographics of theism not as a reason to think that we
do not know that God exists, let alone as a reason to think that God
does not exist.
6 Yujin Nagasawa

6. Religious Tolerance and Disagreement

Religion has been a source of conflict throughout human history.
As society grows more diverse, religious conflicts appear to multiply
and intensify. Inter-religious relations and religious diversity have thus
attracted significant attention in the philosophy of religion in recent
years. Religious exclusivists hold that their own religions are correct
and that all others are incorrect. Religious inclusivists hold that while
their own religions are correct and privileged they are compatible with
at least some others. Religious pluralists hold that their own religions
are correct and that, moreover, there are other religions that are equally
correct. The cogency of these views has been a matter of controversy
among philosophers and theologians for a long time. Recent research in
psychology and decision theory has provided useful tools with which to
address problems that arise from religious diversity.
In Chapter 9, Steve Clarke focuses on a specific form of exclusivism
called salvific exclusivism. A typical form of salvific exclusivism says that
those who fail to accept a particular set of religious beliefs are excluded
from the possibility of salvation. This view has been influential through-
out the history of Christianity and Islam. Salvific exclusivism is danger-
ous from the point of view of liberal societies if it is combined with a
commitment to the moral goal of maximizing the best consequences
for others. Clarke models salvific exclusivism, in particular the version
he calls interventionist salvific exclusivism, by utilizing contemporary
work on decision-making under risk and considers what the government
of the liberal state can do to influence leaders of salvific exclusivism.
In Chapter 10 David Efird applies psychological research on belief
polarization to religious disagreement. Belief polarization is a phe-
nomenon in which two people with conflicting views tend to become
more, rather than less, confident in their views when presented with
mixed evidence. In such a situation people harden, rather than soften,
their disagreements. It appears that belief polarization undermines the
warrant of religious belief as it seems to show that religious belief is not
the product of cognitive processes aimed at the truth. Efird, however,
provides a model of belief with which he aims to show that at least
Christian belief can be polarized yet warranted.

7. The Compatibility of Science and Religion

Chapters 1 through 10 are concerned with specific scientific findings
and their implications for specific issues in the philosophy of religion.
 Introduction 7

However, there is a broader issue concerning the compatibility of science

and religion. Some claim that science and religion are fundamentally
incompatible, so those who take science seriously cannot hold reli-
gious beliefs. Yet others have argued against such a view. The final two
chapters of this volume address this issue.
In Chapter 11, Katherin A. Rogers addresses research on free will in
experimental psychology. First, she expresses her concerns about a prac-
tical danger in denying human freedom by appealing to evidence found
in current research. She suggests ways in which future experiments
could be conducted and evaluated so as to avoid some of the worrisome
aspects. She predicts that, from a Christian perspective, while science
may uncover limitations regarding our ability to choose, it will not show
that we are unfree in a philosophically and theologically significant way.
In Chapter 12 Michael Ruse considers the compatibility of science
and religion from an even more general point of view. He examines the
so-called 'warfare' thesis, according to which science and religion are in
conflict and that if we hold to the one, we have to give up the other. He
argues against the warfare thesis and defends instead the 'independence'
thesis, according to which science and religion are independent from
each other and hence that they are not in contlict. He argues that what
science can grasp is necessarily limited and that it is legitimate for us to
try to answer questions reaching beyond the grasp of science.

References
Barrett, J. L. (2004) Why Would Anyone Believe in God? (Lanham, MD: Altamira
Press).
Darwin, C. (1998 [1859]) On the Origin of Species (Oxford: Oxford University
Press).
Dawkins, R. (1985) The Blind Watchmaker (London: W.W. Norton).
Kelemen, D. (2004) 'Are Children "Intuitive Theists"? Reasoning about Purpose
and Design in Nature', Psychological Science 15: 295-301.
Paley, W. (2006 [1802]) Natural Theology (Oxford: Oxford University Press).
Part I
Divine Attributes
1
The Necessity of God and the
Psychology of Counterfactual
Thinking
Robin Le Poidevin

And certainly this being so truly exists that it cannot be even

thought not to exist.
- Anselm, Proslogion

1. Introduction

That God is not a merely contingent or accidental being, one whose

birth and death are a matter of chance, is part of what, in orthodox
theistic thinking, separates him from us. His existence, unlike ours, is
not determined by anything external. While we wither and perish, he
is incorruptible. Moreover, independence and incorruptibility are them-
selves not accidental properties of God, but essential ones. These aspects
of his being, suggests john Findlay (as a prelude to an atheistic proof),
are required if God is to be truly worthy of worship:

The true object of religious reverence must not be one, merely, to

which no actual independent realities stand opposed: it must be one
to which such opposition is totally inconceivable. God mustn't merely
cover the territory of the actual, but also, with equal comprehensive-
ness, the territory of the possible. And not only must the existence of
other things be unthinkable without him, but his own non-existence
must be wholly unthinkable in any circumstances. (Findlay, 1955,
p. 52; emphasis original)

The necessity of God's existence also offers to play an explanatory

role, as indicated by a familiar modal form of the cosmological argu-
ment: All contingent beings have a cause of their existence, but to
avoid the infinite regress this would otherwise threaten, and a lack of

11
12 Robin Le Poidevin

explanation for the existence of the series of contingent causes as a

whole, we must posit a first cause whose existence is not contingent,
but necessary. This argument may not convince, of course, but even if
we do not agree that we are obliged to posit a necessary cause, its role
in the argument provides a reason for thinking that if God exists he
does so of necessity. But do we really understand what it would be for
a being to be necessary? The proposition that an object has such-and-
such a property as a matter of necessity can be interpreted as de dicto or
de re. That there are de re necessities- necessities that are out there in the
world - is a matter of controversy, and raises awkward epistemological
questions. Necessities de dicto, which are properties of statements, are
less controversial, but in the theistic context, the suggestion that 'God
exists' is necessarily de dicto leads to peculiar difficulties. What I propose
to do in this essay is to offer a way of understanding divine necessity
which avoids both kinds of difficulty.
In the next section I explain briefly why both of the obvious ways
of construing the proposition that God exists of necessity are problem-
atic. In section 3 I outline a projectivist account of necessity, along the
lines proposed by Simon Blackburn. Section 4 augments this projectivist
account by considering what natural tendencies of thought constrain
our imagining of counterfactual situations. Finally, I offer an explana-
tion, in terms of the psychology of ordinary counterfactual thinking, of
why the assumption of God's necessity is such a natural one for the-
ists to make. Given even a relatively basic concept of God, one which
does not explicitly include a modal element, it somehow goes against
the grain of counterfactual thought for the theist to contemplate the
proposition that God might not have existed.

2. Two Kinds of Necessity

The first of Quine's three grades of modal involvement consists of the
attaching of modal predicates to names of statements (Quine, 1953,
pp. 158-9):
Nee '9 > 5'
where 'Nee' is taken to stand for 'is necessary' or 'is necessarily true'. For
certain statements, their formal structure is sufficient to guarantee their
truth, and so they count as necessary: 'If both p and q, then p'; 'If either
p or q and Not-p, then q'. For another group, the meaning of their terms
guarantees it: 'All cubic objects have a shape'; 'A thing is smaller than
anything else that wholly contains it.' They are thus true of necessity,
but the source of necessity is a feature of language only. This is necessity
 The Necessity of God and the Psychology of Counterfactual Thinking 13

de dicto - of the statement. The statements in question are analytically

true: their negations would be self-contradictory. 1
To say that 'God exists' is necessary de dicto, therefore, is to say that
'God exists' is necessary simply by virtue of the meaning of 'God'.
In other words, some form of the ontological argument is sound.
Of course, this is a bold, even reckless, suggestion to make, and most
theists would not acknowledge that any form of the ontological argu-
ment is sound. But even if there were a sound ontological argument,
this would not be good news for theism. For statements that are neces-
sarily true de dicto are exceptions to the otherwise plausible thesis that
truths are made true by some part of external reality. Analytic statements
need no such truth-maker, for their truth is guaranteed by some inter-
nal feature of the sentence used to make the statement. So, ironically, if
'God exists' is necessary de dicto, then God himself would be de trop, not
needed for the truth of the statement. To spell this out more fully, if the
ontological argument, in some form, is successful, then 'God exists' is
analytic. But analytic statements (by themselves) tell us nothing about
the world itself. To the extent that they are informative, they tell us only
about linguistic conventions. So if the ontological argument succeeded
in establishing its conclusion, it would tell us nothing about the world.
But, surely, no theist will be prepared to assent to the proposition that
the truth of 'God exists' tells us nothing about the world!
If, then, God's existence is necessary, it must involve a different order
of necessity. We have to turn to Quine's third grade of modal involve-
ment, in which modal expressions are taken to be sentential operators
which are permitted to fall within the scope of the quantifier, so that
necessity is represented as a feature of the objects quantified over. 2 For
example:
(3x}nec(x > 5)
where 'nee' is to be interpreted 'it is necessarily the case that'. The kind
of necessity we encounter here is de re- that is, of the object. It is a step
too far for Quine, who warns that it 'leads us back into the metaphysi-
cal jungle of Aristotelian essentialism' (Quine, 1953, p. 176). But, given
the self-defeating nature of the ontological argument, if 'God exists' is
necessarily true, it must be so by virtue of some feature of God himself,
not of the word 'God'.
What feature, exactly? A tempting line of thought is that God's neces-
sity is a consequence of his relation to creation: all things are dependent
on him. So if anything exists, he does, for he is a precondition of the
existence of anything else. Could he nevertheless fail to exist? In such
14 Robin Le Poidevin

a world, nothing exists whatsoever: it would be a completely empty

world. Now if, as some have argued, it makes no sense to suppose that
there could be nothing at all, there is no such possible world. Every pos-
sible world, in that case, is a world in which something exists, so every
possible world is a world in which God exists.
Tempting though that line of thought is, however, it falls short of
offering an account of God's necessity. For it might be thought a con-
tingent matter that everything is dependent on God (much as it might
be thought a contingent matter that consciousness requires a physical
realization). So we have to build in a modal element to God's relation to
his creation: everything else is necessarily dependent on God. And if we
were puzzled by the idea of a being's having necessary existence, we will
be no less puzzled by the idea of creation being necessarily dependent
on some other being.
The problem, of course, is not specifically theological: it concerns
our understanding of modality in general. Simon Blackburn has sug-
gested that a dilemma faces anyone who takes what he terms a 'truth-
conditions' approach to modality, that is, anyone who thinks that the
difference between 'p' and 'Necessarily p' is to be explained in terms
of a difference in what in reality would make statements of these two
forms true. Suppose, on the one hand, we offer an account of what it
is for something to be necessary, and we do so in terms that do not
involve any modal concepts (i.e. in terms which make no explicit or
implicit reference to the necessary or the possible). Then the worry is
that we have explained away the modal, and there remains no sense in
which the state of affairs in question has to be so. On the other hand,
we might insist that any analysis of the modal must include the modal:
that 'Necessarily p' is true if and only if p is true in all possible worlds, or
that 'p' is contained within any complete and consistent description of
the world, and so on. Then, although we may have drawn connections
between different modal ideas in an informative way, we have, by stay-
ing within the sphere of the modal, simply postponed the question of
what is conveyed by modalizing (Blackburn, 1987, pp. 53-4).
This dilemma, in turn, is not unique to modality. One can equally
raise it about the moral: what is it to say that <P-ing is good? Or the uni-
versal: what is to say that x andy have the same properties? We want an
informative analysis, not a circular one. But a truly informative analysis
risks getting rid of the analysandum. A typical response to this kind of
dilemma is to say that some things are fundamental, and one's philo-
sophical outlook is in large part determined by what one considers to
be fundamental. Defenders of de re necessity may insist that no further
 The Necessity of God and the Psyclwlogy of Counterfactua/ Thinking 15

analysis of the modal is possible, because such necessities are ontological

bedrock.
So perhaps the difficulty is not so much that of providing a non-
circular analysis of the modal, but rather an epistemological one: just
how do we tell whether some state of affairs is necessary or not? We do
not have to be in the grip of a verificationist account of meaning to
feel the pull of the demand that our account of necessity be epistem-
ically responsible. We want to be able to say what difference necessity
makes. In the case of physical necessities (elemental sodium reacts with
water to form the hydroxide; a magnetized iron bar loses its magnetism
when heated etc.) we may not, as Hume pointed out, directly observe
any necessity. But we can make an inference to the best explanation:
there are these regularities because there is an underlying necessity link-
ing the properties involved. When it comes to metaphysical necessity,
however, this kind of inference is not obviously available. What obser-
vations could we make, for instance, which would justify the idea that
among your essential properties (as opposed to your properties, full stop)
is the property of belonging to the human race?
Perhaps the demand for empirical confirmation is misplaced, how-
ever, and we should look instead to largely a priori considerations. As we
noted above, modal forms of the cosmological argument provide a
reason for thinking of God as necessary (if he exists at all), and the
principles on which such arguments are based look more a priori than
empirical: there is an antecedent cause for every contingent state of
affairs; and there is no actual infinite in nature. The second of these
principles we might defend by appeal to parsimony and the availability
of a mind-dependence account of the infinite. But what is the basis of
the first? Is it that whatever allows us to discern contingency in nature
also intimates the need for a cause? Or is 'contingent', in this context,
just shorthand for 'requires a cause'? Either way, the problem arises how
we actually discern contingency (and necessity) in nature.
So the defender of divine necessity faces a dilemma: either the neces-
sity is interpreted de dicto, in which case we have a self-defeating
ontological argument, or it is interpreted de re, in which case we
face awkward questions concerning the epistemology of metaphysical
necessities. Is there a way of avoiding both horns of this dilemma?

3. A Projectivist Account of Modality

I think there is. The de re/de dicta distinction is a distinction between
objects of necessity, the items to which necessity is ascribed. And it is a
16 Robin Le Poidevin

natural assumption to make that this distinction coincides with a dis-

tinction between the sources of necessity. For when we ascribe necessity
to the sentence only, the source of necessary truth is linguistic. And
when we ascribe necessity to the world, the source of the necessity is
supposed to be some feature of the world. But whereas the distinction
between the objects (language and world) is exhaustive, the distinction
between these two sources, arguably, is not. A third possible source is
the mind. How so? Talking of the necessity that appears to accompany
the relation of cause and effect, Hume remarks,

The efficacy or energy of causes is neither plac'd in the causes them-

selves, nor in the deity, nor in the concurrence of these principles;
but belongs entirely to the soul, which considers the union of two
or more objects in all past instance. 'Tis here that the real power of
causes is plac'd, along with their connexion and necessity. (Hume,
1739-40, Book I, Part III, Section XIV; p. 166)

The constant conjunction we observe between causes and effects

(objects always falling when dropped, changes of speed or direction of
motion after collisions etc.) produces, explains Hume, a 'determination
of the mind' to pass from the idea of one to the idea of the other -
a phenomenon that would later be given the name of classical condi-
tioning. And this determination of the mind, this inevitable transition
from the idea of the cause to the idea of the anticipated effect, is then
attributed to the external events themselves, so that it appears that there
is some necessary connection between the events. Hume thus offers a
'projectivist' account of causal necessity: some feature of the mind is
projected, unconsciously, onto the objects themselves. The fact that this
projection is entirely automatic and unconscious explains why it should
appear to us that we perceive some necessity in the world. If we accept
Hume's account, are we obliged to say that our attribution of necessity
to causation is illusory? It is perhaps not possible to draw a very sharp
contrast between perception and projection, given that much percep-
tion is coloured by previous experience and conceptualizing, so there
is no need to say that our ordinary causal judgements rest on a mis-
take. But if anyone insists that causal necessity lies wholly in external
relations, then Hume is telling them that they are wrong.
Blackburn extends this projectivist account of causal necessity to
other kinds of necessity, even logical necessity. 3 The suggestion is this:
when we contemplate, say, arithmetical truths, such as 2 + 2 = 4, we can
make nothing of the proposition that they are false, that in fact 2 + 2 = 5.
 Tile Necessity of God and tile Psyclwlogy of Counterfactual Til inking 17

And this is not simply a failure of imagination. We cannot, for example,

imagine what a colour beyond the familiar spectrum would look like,
but we would not on that account suppose it impossible that one could
have an experience of such a colour. What prevents us from imagining it
are the limitations of our actual experience, limitations which are them-
selves a result of the structure of our visual sense organs. That we cannot
see beyond the actually visible spectrum we therefore pronounce con-
tingent. What inclines us to judgements of necessity, in contrast, is the
absence of a naturalistic explanation for our imaginative failure. And no
naturalistic explanation is in the offing for our failure to make anything
of the proposition that 2 + 2 = S. It is then that our imaginative block
allows us to project the necessity onto the proposition itself (Blackburn,
1987, pp. 69-70). When we turn to geometry, where we do have a natu-
ralistic explanation of our failure to visualize space as having more than
three dimensions, or as being curved, we can make something of the
idea that physical geometry is a contingent matter.
Moral judgements are an interesting case, standing as they do in
an intermediate position between arithmetical judgements and ones
reporting physical states of affairs. Once we accept that moral judge-
ments are inextricably bound up with our own emotional reaction to
situations, and not simply a cool reading-off of the moral properties
exhibited by those situations, we can contemplate the idea of our being
differently constituted, having different emotional responses to situa-
tions, and so arriving at different moral judgements. But this does not
entirely remove our resistance to the idea that, for instance, murder
is morally neutral, or positively praiseworthy. The imaginative block
is projected as moral necessity - this murder must be wrong - but the
necessity in question seems less inexorable than arithmetical necessity.
Metaphysical necessities, too, might perhaps occupy an intermediate
position between the logically necessary and the physically contingent.
The conviction that being human is one of your essential properties
could well be a projection of my failure to imagine you as anything other
than human - a giant beetle, for instance. But I can also be aware that
this failure on my part is tied up with the ways in which we ordinarily
identify people, and what is most important in our relations with them.
Imagining a giant beetle just gets in the way of thinking about you.
Consider another putative example of metaphysical necessity: the
impossibility of backwards causation. Causes never succeed their effects
in time. They can only precede or (perhaps) occur at the same time
as their effects. We try to imagine causes of backwards causation, of
the lighting of a match as being the cause of an earlier explosion, but
18 Robin Le Poidevin

find we cannot. The later event seems powerless to bring about what
has already happened. The past was what it was, whatever we do now.
What might explain this imaginative failure on our part? Here we might
plausibly appeal to the connection between perceiving temporal order
and acquiring the concept of causation. The fact that our perception
of an event is coloured by the memory of an immediately preceding
event is what makes us inclined to explain the former by the latter.
Once we see this, we might be willing to contemplate the notion of
backwards causation, but we will continue to wonder whether we are
really making sense of it, or just entertaining a form of words: 'the
cause came after the effect'. The necessity is therefore not of the same
order as logical necessity, but there is still a difference between the case
of backwards causation and the case of non-Euclidean geometry, for
instance.
So far, it looks as if the projectivist is providing a causal explanation
of the urge to modalize, and nothing more. But it is a short step from
this to a more explicit statement of the meaning of modal statements.
Recall what Blackburn calls the 'truth-conditions' approach to modal-
izing: an account of the content of modal assertions in terms of the
conditions which make them true - conditions which go beyond those
that make non-modal assertions true. The projectivist is able to propose
an alternative, 'non-cognitivist' account, according to which modal-
ized statements are not truth-evaluable, since they are purely expressive
rather than assertoric: they express a certain attitude towards a proposi-
tion, a commitment to it. The difference between 'necessarily p' and 'p'
is one of force, rather than truth-conditional content. Blackburn him-
self resists this non-cognitivist development of the projectivist position,
anxious as he is to avoid the famous 'Frege-Geach' objection to expres-
sivism (the objection is directed at moral expressivism, but it is easily
adapted to cover other modalities): if we understand '<P-ing is wrong'
as merely expressive of an attitude, rather than an assertion, we inval-
idate arguments in which that expression occurs as a component. The
solution is to adopt a 'quasi-realist' view, on which expressions of the
problematic kind (moral or modal) are treated as capable of truth and
falsity, but their function is still to express a certain state of mind, rather
than to draw attention to some extra bit of reality (Blackburn, 1984,
pp. 189-96).
That is just the bare bones of a projectivist account of necessity. And
it is not yet apparent how it is supposed to apply to the notion of a
necessary being, or whether it can apply to such a notion. More needs
to be said. Since the account makes appeal to naturalistic explanations
 The Necessity of God and the Psychology of Counterfactual Thinking 19

of our imaginative limitations, one way in which it can be augmented is

by considering psychological studies of the way in which we ordinarily
construct counterfactuals- which non-actual situations we are prepared
to contemplate as what might have happened, which not, and why.

4. The Psychology of Counterfactual Thinking

Thinking of what might have been is something that apparently comes
naturally to us. It is not simply an indulgence, but an important part
of learning a skill, cultivating our interpersonal relations, improving
historical understanding, avoiding economic mismanagement, and so
on. We learn from our mistakes, and the mistakes of others. We also
learn from our and others' successes. (What if he had not moved to
France that year? Would his business have prospered to the extent it
did?) We contemplate counterfactuals quite routinely, and when we do
so, we change - in imagination - certain aspects of the past (suppose
I had left the house ten minutes earlier), but keep other aspects con-
stant (I walked to the train station). So what governs our choice of what
to change and what to keep constant?
Naturally, we are reluctant to contemplate counterfactuals whose
antecedents are decidedly outre (if I had always lived on the moon,
could have transmitted my thoughts directly without using any commu-
nicative medium, had been a cactus ... ), but there are apparently some
quite ordinary-looking antecedents that are less likely to appear in our
counterfactual thought than others. We show certain kinds of bias.
One such bias is temporal. When thinking about a series of indepen-
dent events that jointly led to a certain outcome, and how those events
might have been different, we tend to alter the more recent events in
the series. In one study, subjects were presented with the following sit-
uation: john and Michael are playing a rudimentary card game for a
television show in which they take turns to choose cards which are
either red or black. If both pick the same colour card they each win
£1000. If they pick different colours, they win nothing. john starts, and
picks a red card. Michael (presumably in ignorance of John's choice)
then picks a black card, so neither wins anything. Subjects were then
asked how things could have turned out differently, and which of john
and Michael will feel more guilt. They revealed a tendency to think
'If only Michael had picked a red card, they would have won the cash
prize', and that Michael would feel more guilt. In other words, they were
more ready to contemplate the counterfactual where the later event
was changed, rather than the counterfactual where the earlier event was
20 Robin Le Poidevin

changed, that is, 'If only john had picked black, they would have won'
(Byrne et al., 2000). By varying the conditions, however, it is possible to
compensate for this natural bias. In one variant of the game, john starts
and picks a black card. But before Michael picks his, the game is inter-
rupted. After a pause, the game is started from scratch, and john goes
first once again, this time picking a red card. Michael follows, picking a
black card. Now when subjects are asked how things could have turned
out better for the participants, they are just as likely to think 'If only
john had picked black' as 'If only Michael had picked red', since the
first of these possibilities was made salient by the interrupted game.
In another study, subjects were asked to read a story about a man who
decides to stop for a beer on the way home from work, and in addition
is delayed by a series of traffic incidents (a lorry manoeuvring, a flock of
sheep crossing, a tree trunk blocking the road). He arrives home too late
to help his wife who has had a heart attack. On being invited to say how
things might have turned out differently in this story, subjects tended to
focus on the possibility of his not stopping for a beer, rather than his not
being held up in the car (Girotto et al., 1991). Why is stopping for a beer
the most natural event to imagine differently? It cannot simply be the
time bias operating again, because subjects were given different versions
of the story, in which the events occurred in different order. In some
versions, stopping for a beer occurred early in the narrative, yet that
remained the event most likely to be changed in subjects' counterfactual
thinking. The plausible explanation is that that event, unlike the others,
was readily controllable. So here we have evidence for another bias in
counterfactual thinking: a tendency to adjust events within our control,
rather than those outside it.
There may also be a moral dimension. Other studies suggest a bias
towards maintaining social or moral norms. That is, people are more
likely to contemplate alternatives to actual events where the actual
event was morally or socially unacceptable, than ones where the coun-
terfactual alternative involves the unacceptable (Walsh and Byrne, 2005,
pp. 71-2). In the narrative above, stopping for a beer broke the taboo
against drink-driving, so it is natural for subjects to imagine not having
done so.
It is likely that some more fundamental constraint underlies these
biases (tendency to change earlier events, controllable events, and those
that violate social norms). Our counterfactual thinking tends to be rel-
atively conservative: that is, we make relatively modest changes to the
past in contemplating a counterfactual alternative. The reason is not
hard to find: the more divergent the counterfactual situation from the
 The Necessity of God and the Psychology ofCounterfactual Thinking 21

one that actually obtained, the harder it is to work out the conse-
quences. Oust how exactly would your life have turned out if you had
been born in the seventeenth century?) In general, in constructing the
counterfactual situation, background conditions are kept as they in fact
were, and more local adjustments are made to the course of events. The
more revisionary the antecedent, the more elusive the consequent- and
the greater the demands on our information-processing capacities. That
natural conservativeness would explain the tendency to adjust later,
rather than earlier events. The earlier in time the adjustment, the more
revisionary we have to be, since adjustments have knock-on effects.
That, at any rate, is the principle, and although it may not apply in
every case, the disposition of mind that it captures will generalize. In the
original coloured-card case above, for example, it is actually no more
complicated to imagine John's picking black than to imagine Michael's
picking red. But the overall habit of the mind will still favour the latter
over the former.
Similarly, it is easier to imagine acting differently than it is to imagine
events beyond our control turning out differently, because we have a
clearer idea, in the former case, of what to change and what to keep
constant. In the case of morally salient circumstances, however, there
is a check on what actions we imagine differently, for contemplating
actions which break taboos involves greater revision than those that do
not. A willingness to ignore a particular injunction (e.g. not to jump
a red light) raises the question of what other injunctions one might
be willing to break. Once we discard certain conventions, or choose to
ignore them, much more is up for grabs. In contrast, when we imagine
not breaking the injunction that we did in fact break on that particular
occasion, we return things to normal, so to speak. What we imagine in
that case is more consistent with how we would ordinarily expect/hope
to behave.
What significance might any of this have for thoughts about the
nature of divine existence?

5. Tracing the Source of Divine Necessity

Consider a relatively basic concept of God, before it becomes subject to
demands for modal sophistication: God is the creator of all things, and
(in some way yet to be articulated) the basis of all goodness. Let us say
that we accept the actual existence of such a being. Is there anything
in the way of our contemplating the thought that such a being might
not have existed? Let us say that we do contemplate that possibility,
22 Robin Le Poidevin

and then make it the antecedent of a counterfactual: 'If God had not
existed .. .' . We might treat this as a historical counterfactual, involving
some adjustment to the past, and then articulating how what followed
would have been different (much as we might construct a counterfac-
tual with the antecedent 'Had the Treaty of Versailles contained no
reparation clauses .. .'). In assessing historical counterfactuals we sup-
pose things to be pretty much as they actually were, up until the time at
which the antecedent is supposed to obtain. The further back in time we
go, the more we have to adjust, but there is at least some background,
common to our actual situation, against which the counterfactual sit-
uation is cast. But when it comes to contemplating the non-existence
of God, this cannot be against any determinate background. For the
creator is there for the whole of history. So there is no state of affairs
prior to God's existence, against which we can cast the antecedent of
the counterfactual 'If God did not exist .. .'. In attempting to define the
background conditions against which we are to consider God's not exist-
ing, we therefore already have to take a view on what the consequent of
such a counterfactual would be. Would things have come into existence
spontaneously? What would their character have been? As theists, we
might be at a loss for an answer to such questions. Perhaps nothing
at all would have existed. Or perhaps what did exist would be entirely
chaotic. Or perhaps it would be as ordered, but as devoid of fundamental
purpose, as a machine whose parts are in constant and regular motion,
yet which produces nothing. So the proposition that God might not
have existed is, for the committed theist, curiously empty: it suggests no
very definite situation that we could examine. That does not mean that
theists are incapable of contemplating it, but rather that they can make
little of it, precisely because so much is bound up with God's existence.
It does not help much to point out that we are, after all, not deal-
ing with an historical counterfactual here. That, since God is outside
time, the antecedent 'If God had not existed .. .' is not inviting us to
imagine a different past, but the absence of some timeless fact (compare
'If the Inverse Square Law of gravitational attraction had been false .. .').
Again, we have no guide as to what the background conditions for the
antecedent are, how much we can, or cannot, take for granted. To imag-
ine how things would have been different in the absence of the creator
is akin to asking how Oliver Twist would have turned out if Dickens had
not written it.
Our reflections in the previous section suggested that counterfactual
thinking is naturally conservative when it comes to the amount of revi-
sion to the facts in which we are prepared to engage. We are similarly
 Tile Necessity of God and tile Psychology of Counterfactual Thinking 23

conservative when it comes to considerations of value: we prefer not to

contemplate too many violations of the moral order. But for any the-
ist who views God as the source of goodness, no violation of the moral
order could match that of the absence of God, for there would then
be no moral order. Contemplating the non-existence of God therefore
would put the theist at sea, in terms of any value system.
These considerations mirror the (modal) cosmological and moral
arguments for God. However, the purpose of those arguments is to
establish the existence of God starting from apparently neutral premises
about, respectively, the contingency of the universe and the objectivity
of moral values. The projectivist account of divine necessity is not an
attempt to resurrect those arguments. Rather it points out that, once we
adopt the theistic view, the links between God, the universe and value
put obstacles in the way of imagining God's possible non-existence.
Of course, for the atheist, things are quite otherwise. Contemplating
the non-existence of God creates no intellectual tension since the atheist
thinks this is how things are anyway. It is an interesting question, how-
ever, whether the quite general constraints on counterfactual thinking
we have noted would make it difficult for the atheist to contemplate, or
take seriously, the logical possibility of there being a God. Perhaps this
too would involve too great a violation of both fact and value for the
atheist to give it more than a rather abstract content. 4
I have suggested that a natural predisposition to counterfactual con-
servatism makes the thought of God's non-existence less readily avail-
able to the theist than thoughts about other kinds of counterfactual
situation. This difficulty in making much of the thought of God's non-
existence might then be projected onto God himself, giving rise to a
further thought that is expressed by the proposition that God's exis-
tence is necessary. In saying that God is necessary, we need then be
adding no further factual content to the non-modal assertion that God
exists; the difference, rather, is one of force. But to this line of thinking,
two objections could be raised. The first is that people's thinking about
God often shifts, from theism to atheism, or vice versa. Take, for exam-
ple, an atheist who, as a result of either religious experience, or careful
consideration of theistic arguments, comes to believe in God. Prior to
conversion, he or she had no difficulty in contemplating the possibility,
indeed the actuality, of God's non-existence. Is it plausible that, now a
believer, this thought should no longer be available? The second objec-
tion is that theists routinely defend their position by pointing to the
unattractive consequences of the opposing hypothesis, and this suggests
that they are able to entertain and assess the associated counterfactual
24 Robin Le Poidevin

'If God did not exist ... '. They must therefore be able to make something
of the possibility of God's non-existence.
On the first of these objections, we might note that thoughts we enter-
tain at some times can indeed become less accessible at others, as a result
of a shift in our outlook. We may well remember that we used to think
such-and-such, but be unable to capture the sense of its obviousness or
inevitability, and perhaps even struggle to make sense of it. Consider
shifts in moral thinking, for example. Suppose Helen, at the age of 20,
regards euthanasia as murder, pure and simple. However well motivated
it might be on some occasions, she cannot reconcile herself to the idea
that it is at all morally acceptable. But thirty years later, after a career in
health care, looking after the terminally ill, she sees it in quite a differ-
ent light. She knows that she once thought it wrong, but the conception
she now has of what it is for something to be right or wrong makes it a
struggle to conceive of how those acts of euthanasia that are performed
out of love and compassion for the suffering patient could possibly be
wrong. Similarly, the convert may view his or her earlier conviction of
God's non-existence as not merely wrong, but muddled.
On the second objection, to express the importance of a certain truth
in terms of the conditional, which says what would be the case if it
were not true, is not necessarily to concede the intelligibility of the
antecedent. Consider this conditional: 'If there were no space, noth-
ing would have mass.' That is one way of drawing attention to the
connection between the property of having mass and the property of
occupying space. The thought conveyed is perfectly intelligible. It does
not follow that a world without space is intelligible. Similarly, one might
say, 'If there were no irrational numbers, then Pythagoras's Theorem
would be false', meaning simply that Pythagoras's Theorem yields irra-
tional values for the hypotenuse, given certain values for the other two
sides. It implies neither that the absence of irrational numbers is a gen-
uine possibility, nor that we can clearly understand what it would be
for there to be no irrational numbers. What I want to suggest, then, is
that there is a difference between expressing certain thoughts by means
of counterfactuals, and engaging in genuine counterfactual thinking.
The latter requires, as the former does not, an understanding of the
situation represented by the antecedent as a fully intelligible state of
affairs, and working out what would be the case in a world in which
that state of affairs obtained. Similarly, a theist may express the deep
connection they see between God and the moral order by means of
the counterfactual 'If God had not existed, there would have been no
moral order', without thereby implying that the antecedent represents a
 The Necessity o{God and the Psychology o{Counterfactual Thinking 25

situation they can make perfect sense of. Counterfactual assertion does
not always betoken genuine counterfactual thinking.

6. Conclusion
In summary, then: Why is it so natural for the theist to say that God
exists of necessity? For the theist, the domains of fact and value are per-
vaded by God: nothing is fully intelligible without reference, at some
point, to God. This, together with a natural conservatism in counter-
factual thinking, a conservatism revealed by psychological experiments
which require subjects to imagine how things might have turned out
differently in certain contexts, makes it harder, more unnatural, for the
theist to imagine seriously a possible world in which God does not exist.
And this difficulty experienced by the theist in making much of the
notion of God's non-existence is then projected onto God, in the form
of the idea that God cannot fail to exist. As we noted earlier, however,
not all imaginative blocks are projected onto the world. Our inability to
imagine a fourth primary colour is not projected as the impossibility of
there being more than three primary colours, since there is a naturalistic
explanation of why we see colours as we do. But have we not just given
a naturalistic explanation of the theist's difficulties in imagining a world
without God? Well, we have, but there is still a difference between this
kind of explanation and the explanation of our imaginative limitations
over colour. Although we cannot imagine a fourth primary colour, we
can readily conceive of beings with differently constituted visual sys-
tems. It is not so straightforward for the theist to imagine a being who
is in a better position than he or she is to conceive of a world with-
out God. Belief in God is not simply accepting the existence of another
item in the landscape: it completely transforms that landscape. And that
transformed landscape presents itself as a vision of how things are, not
simply of how they appear from a certain perspective. The difficulty of
conceiving of God's non-existence appears more objective than the dif-
ficulty of conceiving of a fourth primary colour, as does conceiving of
alternative arithmetics. From the committed theist's point of view, any
rational thinker, however physiologically or neurologically constituted,
will struggle genuinely to understand the absence of God. 5
That provides a causal explanation of theological modalizing, but it
also offers something more: an account of what it is to ascribe necessity
to God. The centrality of God to the theist's conceptual framework, its
role in determining the character of that framework, gives rise to a com-
mitment that seems unavoidable. 'God necessarily exists' expresses that
26 Robin Le Poidevin

commitment. And, to borrow from Blackburn's account of modal think-

ing, we do not have to deny that the expression has assertoric force,
that it is evaluable as true or false. The sense of the inexorability of
commitment is what provides assertibility conditions for the statement.
Before leaving the topic, there is a final, methodological worry to be
addressed. I have appealed to natural biases in our construction of coun-
terfactuals to explain the psychological basis of the doctrine of divine
necessity. But, it will be urged, the philosophical analysis of divine
necessity surely transcends such natural dispositions. So whereas psy-
chology might cast some light on the aetiology of theological doctrine,
it can hardly illuminate the true meaning of that doctrine. In reply,
I would point out that philosophy (in this context) is in the service of
theology, not vice versa. And theology, in turn, is an attempt to articu-
late our deep-seated beliefs about the nature of God. Where those beliefs
come from is not irrelevant to understanding their meaning. Faith first,
theory second.
What has been offered here is a projectivist account of God's neces-
sity. It is not a projectivist theory of God. We have assumed, that is, that
there is a being onto whom necessity is projected, just as, according to
Hume, causal necessity is projected onto events in the world. The nat-
ural question this prompts, namely what other divine properties might
also be projected, I leave for the reader to ponder.

Notes
1. Quine himself does not characterize the first grade of modal involvement in
terms of analyticity, having elsewhere cast doubt on the analytic/synthetic
distinction (Quine, 1951).
2. The second grade of modal involvement, which we have skipped over here,
takes modal terms as operators attaching to statements:
nec(9 > S)
The inverted commas from the first grade have disappeared, implying that it
is the content of the statement that is modified by the modal expression. But
we do not yet have quantification outside the scope of the operator. As Quine
explains, the second grade can be understood as a way of articulating the rela-
tively harmless first grade, but it can also be a stepping stone to the anything
but harmless third grade (Quine, 1953, p. 176).
3. Why do we need a projectivist account of logical necessity? Is it not enough to
appeal to logical laws and linguistic conventions? Quinean scruples, however,
may make us wary of drawing a sharp de dicto/de re distinction, insofar as
the former is taken to be coincident with the analytic. And in the case of
arithmetical truths, we will not be able to combine a Platonist ontology (i.e.
one that holds that there really are numbers, independently of arithmetical
 The Necessity of God and the Psychology ofCounterfactual Thinking 27

thought and language) with the view that arithmetical necessity does not go
beyond the level of language.
4. This might seem to raise the spectre of incommensurability. If what theists
assert does not coincide with what atheists deny, the prospects for rational
debate in this area are dim. However, the inability of each side fully to enter
imaginatively into the other side's point of view does not mean that they
cannot articulate the propositions over which they disagree.
5. As Anselm put it, 'How indeed has he [the Fool] "said in his heart" what he
could not think ... in one sense a thing is thought when the word signifying
it is thought; in another sense when the very object which the thing is is
understood. In the first sense, then, God can be thought not to exist, but
not at all in the second sense' (Proslogion, Chapter IV; Charlesworth, 1965,
pp. 119-21).

References
Blackburn, Simon (1984) Spreading the Word: Groundings in the Philosophy of
Language (Oxford: Clarendon Press).
Blackburn, Simon (1987) 'Morals and Modals', reprinted in Blackburn, Essays in
Quasi-Realism (Oxford: Oxford University Press, 1993), pp. 52-74.
Byrne, R. M. ]., S. Segura, R. Culhane, A. Tasso and P. Berrocal (2000) 'The Tempo-
rality Effect in Counterfactual Thinking about What Might Have Been', Memory
and Cognition 28: 264-81.
Charlesworth, M. ]. (1965) Anselm's Proslogion (Oxford: Clarendon Press).
Findlay,]. N. (1955) 'Can God's Existence Be Disproved?', in Antony Flew and
Alasdair Macintyre (eds), New Essays in PIJilosopllical Theolo,gy (London: SCM
Press, 1955), pp. 47-56.
Girotto, V., P. Legrenzi and A. Rizzo (1991) 'Event Controllability in Counterfac-
tual Thinking', Acta Psychologica 78: 111-33.
Burne, David (1739-40) A Treatise ofHuman Nature, ed. L.A. Selby-Bigge, rev. P. H.
Nidditch (Oxford: Clarendon Press, 1978).
Quine, W. V. (1951) 'Two Dogmas of Empiricism', Philosophical Review 60: 20-43.
Quine, W. V. (1953) 'Three Grades of Modal Involvement', in idem, Ways of
Paradox (New York: Random House, 1966), pp. 158-76.
Walsh, Clare R., and Ruth M. J. Byrne (2005) 'The Mental Representation of What
Might Have Been', in David R. Mandel, Denis J. Hilton and Patrizia Catel-
lani (eds), The Psychology of Counterfactual Thinking (Abingdon: Routledge),
pp. 61-73.
2
Why Would Anyone Believe
in a Timeless God? Two Types
of Theology
Benjamin Murphy

1. Is Theology a Waste of Time?

God may have infinite time, but for the rest of us, time is a precious
commodity, to be used carefully. Consequently, it is somewhat disturb-
ing to contemplate the possibility that theology is a complete waste of
time, particularly if one is a professional or even an amateur theologian.
No doubt practitioners of other disciplines have such feelings on occa-
sion. The difference is that recently the proposition that theology is a
waste of time has been advanced as a thesis that has scientific support.
I am referring here to research that has been carried out by practition-
ers of that relatively new discipline, cognitive science of religion. The
overwhelming view within this field is that religion itself is a natural
human phenomenon. According to this view, a tendency to believe in
the supernatural is part and parcel of being an average human being.
It is not that religious belief is innate, but that, from an early age, chil-
dren are primed to accept religious beliefs. If cognitive scientists are to
be believed, religion has a healthy future. 1
However, along with good news for religion comes bad news for the-
ology. It appears that there is frequently a gap between the official
beliefs propounded by religious professionals, such as theologians, and
the beliefs that come naturally to most people. Furthermore, in day-
to-day thinking, even those who have had theological training tend to
revert to their natural religious beliefs, a phenomenon known as 'theo-
logical incorrectness'. In one study (Barrett and Keil, 1996) participants
were given a series of stories about God and asked to retell them. When
retelling the stories, people have a notable tendency to describe God

28
 Wily Would Anyone Believe in a Timeless God? 29

in more anthropomorphic terms. For example, theologians agree that

God can perceive everything, but a story in which God is listening to a
bird while an aeroplane lands became a story in which the sound of the
aeroplane landing interrupts God's enjoyment of the birdsong. God, if
he exists, is surely capable of multi-tasking, but a story in which God
answers listens to and answers two prayers becomes a story in which
God deals with one prayer at a time. When God is replaced by a specially
created fictional entity with super-abilities that include super-hearing or
multi-tasking, people do not make these errors.
These are perhaps trivial errors in a trivial task. Ever since God was
described as walking through the Garden of Eden in the cool of the
day, storytellers have been willing to spice up stories with a bit of
anthropomorphism. The anthropomorphisms that people resort to in
the experiment can be said to improve the story, since they give a ratio-
nale for otherwise irrelevant details: if the noise of the aeroplane wasn't
an interruption, why even mention it? As Chekhov pointed out, if you
show the audience a gun in Act One, it must be fired by the end of
the play. In other cases, theological incorrectness is not so trivial. For
example Theravada Buddhists may say that the Buddha was a normal
human being who attained a goal, enlightenment, that is open to any
human to achieve, but their refusal to ordain nuns now that the valid
lines of female ordination have been broken is best explained by the lin-
gering belief that there was something supernatural about the touch of
the Buddha's hands (Slone, 2004, pp. 81-3). The theologically incorrect
belief about the Buddha's supernatural touch motivates the politically
incorrect refusal to ordain women.
So, it seems that religious beliefs occur naturally without any help
from theologians. When theologians try to introduce ideas that con-
flict with these natural religious beliefs, the best they can hope for is
lip-service. So, why do we keep on trying? This is not only a source of
dismay for theologians, it is a puzzle for cognitive scientists. Why would
so many intelligent people persist, over many generations, in an activity
that is doomed to fail? Why spend so much time trying to teach difficult
theological systems when there is the easier option of indoctrinating
people with naturally attractive religious beliefs?
Of course, to say that theology is a failure because theologians do not
manage to win most people over to their beliefs presupposes that the
main goal of theologians is to attract as many adherents as possible.
Impact and popularity are factors that can be objectively measured, and
it is understandable that cognitive scientists of religion adopt such a
standard to measure the relative success or failure of religious writers.
30 Benjamin Murphy

Yet this misses the fact that theologians, like many academics, may not
be motivated by a desire to win popular support. Rather than judg-
ing theology a failure by standards imposed from outside, we should
attempt to understand what factors, other than the desire to persuade
large numbers of people, might motivate people to become theologians.
Joseph Bulbulia suggests that it is the fact that the study of theology
requires a lot of time and yet has no value outside a religious community
which makes it useful as a way of signalling a high level of commitment
to the community, and speculates that it may serve 'few utilities' apart
from this (Bulbulia, 2009, p. 52).
Bulbulia writes this because he believes in what Pascal Boyer refers to
as the 'tragedy of the theologian'. According to Boyer (2001, pp. 283-5),
religious professionals frequently organize a 'guild of priests' to protect
their interests. The guild claims a monopoly on knowledge of the super-
natural, and the agreement of all members of the guild on the contents
of such knowledge is evidence of their collective reliability. That level
of stability requires a system, thus theology comes into existence, but
teaching people systems requires repetition, and this induces tedium.
People are inevitably drawn to the new, the exciting, that is to say, to
the unorthodox. Theological orthodoxy will never prevail.
As Boyer points out, the distinction that he is employing between rou-
tine, bureaucratized religion and charismatic, revolutionary religion is
not new; it can be traced back to the work of Max Weber. However, Boyer
is relying on a recent proposal by the anthropologist Harvey Whitehouse
that is intended to provide a firm scientific basis for this familiar dis-
tinction (2001, p. 283). Whitehouse points out that any long-lasting
religious idea must be implanted somehow in our memory, and, given
what we now know about the way human memory functions, this can
happen in one of two ways. Either a formula is repeated again and again,
like learning multiplication tables or conjugating verbs, or else a lesson
is associated with some vivid, unusual and exciting event. Whitehouse
calls the former technique of teaching and learning a religion the doctri-
nal mode of religiosity. It is this mode of religiosity that is utilized by the
guild of priests, according to the theory, and tedium is an inevitable side-
effect. The latter technique of learning, that of creating exciting events,
is associated with Weber's charismatic religion, or, as Whitehouse calls
it, the imagistic mode of religiosity {Whitehouse, 2004, pp. 63-85).
Boyer and Bulbulia both write as though there were nothing more to
theology than the composing and memorizing of catechisms, as though
good theology is nothing more or less than orthodox theology. Once
again, this is an outsider's view of theology. A five year old reciting
 Why Would Anyone Believe in a Timeless God? 31

the four-times table may well imagine that, at university level, maths
students recite the seventeen-times table. In the same way, Boyer and
Bulbulia seem to picture theology as being simply a very advanced ver-
sion of the kind of Sunday-school classes that were satirized by Mark
Twain. This leaves out the creativity that characterizes theology at its
best - a creativity that is quite independent of orthodoxy. For exam-
ple, I have heard conservative Catholics admit that Hans Kiing and
Karl Rahner excelled in the field of theology, while regretting their lack
of orthodoxy. Indeed, one sometimes encounters the idea that exces-
sive zeal for theology tends to lead to heresy. One might try to defend
Boyer's thesis by classifying such renegade theologians with Weber's
charismatic prophets, breaking away from the monotony of doctrinal
religion. However, this would not be in keeping with the scientific basis
for the distinction between two modes of religiosity that Boyer takes
from Whitehouse. What is distinctive about the imagistic mode, which
corresponds to Weber's charismatic religion, is that memories are formed
by providing, on rare occasions, striking rituals that appeal to the senses.
Certainly, one could describe Karl Barth, for example, as a prophetic
voice and a charismatic writer, whose work is full of verbal fireworks.
Only the use of literal fireworks, however, would qualify Barth as a prac-
titioner of the imagistic mode of religion in Whitehouse's sense. I sug-
gest, then, that theology is a creative discipline with internal standards
of excellence that are independent of the standards by which orthodoxy
is judged. I will not argue this point at length, since I trust that readers
of this volume are capable of judging the truth of this for themselves.
This means that, from the perspective of cognitive science of religion,
the practice of theology remains an unsolved puzzle. There is one other
suggestion about the role of theology that comes not from a cogni-
tive scientist, but from a philosopher with a strong interest in cognitive
science of religion, Daniel Dennett. He states,

A milder and more constructive response to relentless skepticism

is the vigorous academic industry of theological discussion and
research, very respectfully inquiring into the possible interpretations
of the various creeds. This earnest intellectual exercise scratches the
skeptical itch of those few people who are uncomfortable with the
creeds they were taught as children, and is ignored by everybody else.
(Dennett, 2006, p. 208)

Dennett is not in the business of flattering the egos of theologians, but

he does at least acknowledge that theologians have genuine intellectual
32 Benjamin Murphy

motivations, beyond a desire for status within or control of a com-

munity. There is, after all, a certain kinship between philosophers and
theologians. It is not unknown for philosophers to advance abstruse
theories that defy common sense, and not usual for them to retract
such theories when they fail to attract a popular following. I think that
Dennett's description does fit much analytical philosophy of religion,
and analytical theology. However, as I hope to demonstrate, it does not
cover the whole of theology.
According to cognitive science of religion, popular religious concepts
stick in our mind because they are minimally counter-intuitive. The
term 'minimally counter-intuitive' was introduced by Justin Barrett,
based on the work of Pascal Boyer (Barrett, 2004, p. 30). A concept is
minimally counter-intuitive if it violates a single ontological category:
flying fish, flightless birds, virgin births and men who walk on water are
all examples. These concepts stick in the mind because they violate an
ontological category, and thus contain an element of surprise. However,
because they violate only one ontological category, we can still form
expectations based on our existing knowledge. A flightless bird still has
feathers and a beak, and a man who walks on water still has two legs
(Boyer, 2001, pp. 62-5). Theological concepts, unlike popular religious
concepts, violate many expectations and are thus harder to grasp. The
question is why anyone would even try.
In this essay I have picked one particular property that separates
the abstract and extremely counter-intuitive God of theology from
the anthropomorphic and minimally counter-intuitive God of popu-
lar religious belief: divine timelessness. I plan to consider other divine
attributes, such as impassibility, in future work. If a God who multi-tasks
is hard to understand, how much more so a God who is completely out-
side of time? What is more, even advocates of divine timelessness, such
as Paul Helm, admit that fidelity to the biblical tradition does not require
such a concept (Helm, 2001, pp. 31-32). In 1970, Nelson Pike concluded
his study of God and Timelessness with a question: 'What reason is there
for thinking that God's timelessness should have a place in a system of
Christian theology?' (Pike, 2002, p. 190).
Pike is anything but ignorant of the arguments that have been
advanced in favour of divine timelessness. His question comes after
190 pages of careful analysis. So his question is not about the jus-
tifications that have been advanced for believing in a timeless God,
but the underlying motivations. Frequently, when people construct
complicated philosophical argumeats, it is because the conclusion is
 Why Would Anyone Believe in a Timeless God? 33

intuitively attractive. What attraction might the doctrine of divine

timelessness have?

2. Timelessness and Foreknowledge: Theology

as a Technical Discipline
In the Summa Theologiae Iae, Q. 14, Art.ll, Aquinas considers the
question of God's knowledge of future contingents. To summarize his
reasoning: since God knows now what actions humans will perform,
there are truths about future choices, but since such future choices are
contingent, there cannot now be truths about them. The solution is
that God is timeless. Our future is part of God's present. God knows
everything, but his knowledge is simultaneous with what is known.
Anthony Kenny has pointed out the following problem. Simultane-
ity is a transitive relationship. If A is simultaneous with B, and B is
simultaneous with C, then A must be simultaneous with C. But in
that case, since God's knowledge of every free action performed by
humans is simultaneous with those actions, all human actions must
be simultaneous with each other (Kenny, 1979, pp. 38-9). A mini-
mally counterintuitive idea is still comprehensible enough to generate
a set of expectations and inferences that make sense. However, as
Kenny's argument shows, the idea of a timeless God generates infer-
ences that undermine all of our usual chronological expectations. It is
counter-intuitive, certainly, but far more than minimally so.
Both Aquinas's initial argument and Kenny's objection follow the
same pattern of reasoning, a methodology that can be traced back to
Aristotle, and remains common amongst analytical philosophers today.
Here is the methodology as explained by Aristotle in the Nicomachean
Ethics:

As in the other cases we must set out the appearances, and first of
all go through the puzzles. In this way, we must prove the common
beliefs about these ways of being affected - ideally all the common
beliefs, but if not all, then the most important. For if the objections
are solved and the common beliefs are left, it will be an adequate
proof. (114Sb, S; Aristotle, 1999, p. 100)

For Aquinas, the common beliefs are that human actions are con-
tingent and that God has infallible knowledge of them. The puzzle
is that this creates a contradiction, and then the doctrine of divine
34 Benjamin Murphy

timelessness - a rather uncommon belief - solves the problem. Kenny

then brings in another common belief - that simultaneity is a transi-
tive relationship - and creates another problem, and so it goes. It is, as
I have said, a method frequently employed by analytical philosophers.
To give just one example of this methodology as practiced within ana-
lytical philosophy, Donald Davidson reasons that common beliefs tell
us that human actions are unpredictable, and also that all events are,
in principle, predictable, and his theory of anomalous monism offers
a reconciliation (Davidson, 2001). So it is not surprising that Daniel
Dennett, an analytical philosopher, should look upon theology as an
exercise that follows this same pattern. The clash in the common beliefs
creates a sceptical itch, an awkward philosophical question, and the the-
ologian aims to provide an intellectually satisfying answer. Indeed, Ilkka
Pyysiainen has proposed this as a model for how theology is created
(Pyysiainen, 2004).
Pyysiainen points out that many cognitive theorists have offered a
dual-process view of the human mind, according to which we have
two ways of thinking, one of which has been labelled 'implicit', 'expe-
riential' and 'intuitive' and the other which has variously been called
'explicit', 'rational' and 'reflective'. Despite the different terminology
used by different theorists, it is clear that they are all referring to
the same distinction. To avoid unwarranted inferences that might
come from adopting any particular labels from his list, Pyysiainen
refers to implicit/experiential/intuitive thinking as the A-system, and
explicit/rational/reflective thinking as the B-system (Pyysiainen, 2004,
p. 142). As the reader has no doubt realized, Pyysiainen's A-system
is closely related to Whitehouse's imagistic mode of religion, and his
B-system to Whitehouse's doctrinal mode (Pyysiainen, 2004, pp. 127-8).
It is the A-system that supplies what I have been calling, following
Aristotle, common beliefs. The main finding of cognitive science of reli-
gion, as I have already noted, is that the A-system will naturally produce
religious beliefs. The Aristotelian task of sorting through these beliefs
and solving puzzles falls to the B-system.
Where Pyysiainen differs from Boyer is in his suggestion that cogni-
tive science needs an integrated model. As far as Boyer is concerned,
theology, which is doctrinal, is perpetually engaged in a losing bat-
tle with imagistic thinking. Pyysiainen does not see conflict, rather he
states that the A-system and B-system 'work together most of the time'
(Pyysiainen, 2004, p. 127) and on the basis of this integrated model, he
hopes to tackle the issue that he thinks Boyer avoids: the origin of the-
ological ideas (Pyysiainen, 2004, p. 143). So, according to Pyysiainen,
 Wily Would Anyone Believe in a Timeless God? 35

theology is what happens when the B-system attempts to rise to the

challenge of the minimally counter-intuitive ideas that the A-system
sends its way:

To the extent that counterintuitiveness is based on explicit modi-

fication of intuitive concepts, it is to be expected that theological
thinking is characterized by both counterintuitiveness and a ten-
dency to rationalize. Theologians, indeed, try to rework minimally
counterintuitive ideas into such abstract concepts that the apparent
contradictions seem to disappear. (Pyysiainen, 2004, p. 143)

If this is what theologians do, then successful theology becomes a

matter of carrying out the balancing act described by Aristotle. As many
of the common beliefs as possible must be preserved, provided the
puzzles can be solved to the satisfaction of awkward philosophers like
Kenny. Stump and Kretzmann's solution to Kenny's puzzle is to intro-
duce a highly counter-intuitive (but, they claim, logically consistent)
concept of simultaneity which is non-transitive (Stump and Kretzmann,
1981). Open Theism, on the other hand, sacrifices the common view
that God has infallible knowledge of all future actions in the interest of
maintaining a coherent system. I will refer to theology that follows this
pattern as 'technical theology'.
It is important to be clear that technical theology is never purely tech-
nical. The technical aspects - the work of the B-system - are meant to
support, as far as possible, the non-technical intuitions of the A-system.
This may be compared to the work of a software engineer. I use Microsoft
Windows on my computer, and as long as it runs without any problems,
I am happy. But sometimes problems do arise, and I have either to solve
them myself, or get in a software engineer to help. In the old days, fixing
problems sometimes required the use of MS-DOS, which was much less
user-friendly. I was aware that MS-DOS was there, and I knew how to
enter simple instructions, but as long as my computer was functioning
smoothly, I was happy to ignore its existence. Stump and Kretzmann's
theory of ET-Simultaneity is rather like MS-DOS. Only a tiny minority of
believers are even interested in trying to understand it. If those who do
understand it conclude that it is a coherent solution to the puzzle, then
they can reassure the majority of believers that there is a solution to
the problem of divine foreknowledge, the exact details of which are of
interest only to the technically minded. If the theological boffins reject
Stump and Kretzmann's theory, then they might have to resort to Open
Theism, which does interfere somewhat with the natural inclinations of
36 Benjamin Murphy

believers - as though a software engineer were to tell me that I can run

Windows on my computer, but only in a foreign language. Either way,
the technical stuff is not an end in itself: the B-system is used to orga-
nize, support and repair the indispensable raw material provided by the
A-system. Or, in more familiar theological terminology, technical theol-
ogy tends to serve an apologetic function: it defends and occasionally
modifies an existing structure of beliefs, but does not create a new one.
It will not have escaped attention that what I am calling 'technical theol-
ogy' is a common approach amongst analytical philosophers of religion.
Indeed, if I am right, Pyysiainen has provided a model that explains the
way a lot of philosophical and theological thinking functions.
However, I do not think that Pyysiainen's model is sufficient to
explain all theological thinking. When Aquinas was writing in the
thirteenth century theology was established as a rigorous academic dis-
cipline. But the doctrine of divine timelessness entered into Christian
thinking well before theology became part of the scholastic curriculum,
and it was not motivated purely by technical concerns. To under-
stand the real attraction of this idea, we must trace it back to the
pre-scholastic era.

3. Timelessness and Modes of Consciousness: Theology

as Experience-Enhancing
When Aquinas affirmed that God is timeless, he was reinforcing a tradi-
tion that had already been strongly established, Boethius's Consolations
of Philosophy clearly standing out as a particularly influential source.
Boethius, of course, did not invent the idea of a timeless God; it was
taken from Plotinus. Nor was he the first Christian theologian to make
use of the idea; it had already been integrated into Christian theol-
ogy by Augustine. But still, as john Marenbon states, the Consolations
has exerted a magnetic pull on readers because it tackles the theme
of human happiness within the most dramatic of contexts (Marenbon,
2003, p. 96). If we want to know why the idea of a timeless God has
been so popular for much of the history of Christian theology, Boethius
is surely the author to whom we should turn.
When Boethius wrote The Consolations of Philosophy, he was in prison
awaiting execution on a charge of treason. The drama of the book comes
not merely from his impending execution, but the fact that he fell so
suddenly from high political office. His unhappiness is increased by
the thought of everything he has lost, and at first, he gives way to
lamentation. He is interrupted by the arrival of a lady who personifies
 Why Would Anyone Believe in a Timeless God? 37

Philosophy. She banishes the Muses of poetry, who have encouraged

him to wallow in emotion, and reminds him that he took up the study
of philosophy because he aspired to true happiness, which he should
have known could not be attained from wealth and status that is here
today and gone tomorrow. Philosophy then assures him that God is true
happiness, and that we humans become happy insofar as we become
more like God (Boethius, Consolatio Philosophiae, Libra III, Prosa 10).
Later, Boethius gives his famous description of what God's life is like.
God is eternal, and eternity is 'complete and simultaneous possession of
endless life' (Boethius, Consolatio Philosophiae, Libra V, Prosa 6). 2 Since
God is true happiness, and we become happy insofar as we become like
God, it follows, one would expect, that the closer we come to a com-
plete and simultaneous perception of endless life, the happier we will be.
Perhaps we could have, if not a complete and simultaneous possession
of endless life, a nearly complete and simultaneous possession of (seg-
ments of) our own finite life. In that case, Boethius's high status would
no longer be perceived as something that once was his but now is lost.
But what would it mean to experience different times as present?
Here I turn to the work of Margaret Donaldson, one of the lead-
ing figures of developmental psychology. In her book Human Minds
(Donaldson, 1992) she traces the development of the modes of con-
sciousness we have available to us as we reach maturity. 3 Our first
experiences are in 'point mode', focusing on what is taking place now.
To say that things are 'taking place' is dynamic, not static: the 'point'
is 'not like the idealized extension-less point of the mathematician but
rather like a moving spot of light' (Donaldson, 1992, p. 42). As we grow
older, we start to make definite plans about the future, drawing upon our
knowledge of the past. This involves thinking in what Donaldson terms
the 'line mode', a process that begins at around eight to ten months
(Donaldson, 1992, p. 61). Line-mode thinking is important because it
enables us to become responsible. If I understand Donaldson's position
correctly, line-mode thinking is valuable because it brings to light errors
that we might otherwise miss. Even though right now I am safe and
comfortable, perhaps my actions of this morning are likely to have bad
consequences for me tomorrow, unless I rouse myself to take action now.
This explains why Donaldson associates the line-mode with negative
emotions. She describes it as 'the mode of "if only" and "what if" -of
resentment, guilt and fear' (Donaldson, 1992, p. 61). Like a car alarm,
the line-mode alerts us to hidden threats and keeps us awake at night.
The other two modes of consciousness are the construct mode,
in which we imagine what might happen in some unspecified time
38 Benjamin Murphy

(Donaldson, 1992, p. 81),and the transcendent mode in which our

thoughts are not bound to any place or time (Donaldson, 1992, p. 125).
If I imagine a mathematical problem involving two trains setting off
from two unspecified stations, that would involve the construct mode.
I am thinking of a concrete train travelling on a journey, but not a
particular train on a particular journey. If I then focus purely on the
mathematics, and forget about the train itself, I enter the transcen-
dent mode. This particular choice of example makes it sound as though
the construct and transcendent modes are purely intellectual, but this
is not necessarily the case. Donaldson states that the construct and
transcendent mode can be value-sensing rather than intellectual (1992,
pp. 150-9), where value-sensing involves arriving at judgements about
what is worth paying attention to ar:d what may be ignored (pp. 13-14).
Although Donaldson does not specifically mention Boethius, she is
well aware that the history of spirituality provides data that support
her theory about modes of consciousness. She points out, for example,
that the goal of Buddhist meditation seems to be the attempt to escape
from the tyranny of the line-mode and return to the point-mode, thus
having the sensation of transcending time (1992, pp. 222-3), and she is
aware of the connection to the idea of divine eternity (p. 15 7). Her reflec-
tions on the connections between modes of consciousness and religion
are deeply indebted to the work of Keith Ward, whom she cites in the
following passage:

Ward says that such a sense is best evoked not by science or phi-
losophy but rather by poetry or music; 'it is a major heresy of
post-Enlightenment rationalism to try to turn poetry into pseudo-
science, to turn the images of religion, whose function is to evoke
eternity, into mundane descriptions of improbable facts'. 4

There is a danger that Donaldson's gloss will obscure Ward's real point.
Her passage might be taken to suggest that music and poetry are inter-
changeable, but they are not. Pure music, without words, can evoke
moods and emotions and even, in its way, engage the intellect, but it
has no propositional content. To get propositional content, we need to
add lyrics - and if we detach the lyrics from the music, we are left with
poetry. Once we have words, we have the possibility of propositional
content. Theology is not pseudo-science, but nor is it pseudo-music: the-
ologians should not try to compete with musicians in the realm of pure
emotion. Theology, like poetry, does offer propositions, but propositions
that need to be understood within a particular emotional context.
 WIJy Would Anyone Believe in a Timeless God? 39

In the Consolations, we can, to a certain extent, track the move-

ment between imaginative and factual writing, because Boethius moves
between prose and poetry. The qualifying phrase, 'to a certain extent',
is important. The dividing line between prose and poetry is permeable.
After all, the very idea of a conversation with a personification of Philos-
ophy is clearly metaphorical. However, it is surely not a modern heresy
to say that some of the propositions that Boethius ascribes to Philosophy
really are intended as serious philosophical propositions.
The initial section in which Boethius gives way to sorrow is written
in verse. The arrival of Philosophy is marked by a shift to prose, and
her first act is to banish the Muses of poetry. From then on, although
the book continues to alternate between prose and poetry, it is clear
that the prose sections of the book, in which Boethius and Philosophy
engage in vigorous intellectual debate, are driving the narrative forward.
Poetry and imagination have their place, but they are subordinate to rea-
son, that is Philosophy, who provides the direction. Within the poems,
the gods of Greek and Roman mythology make frequent appearances.
Since the author is a Christian, we know that in this context stories of
Bacchus, Circe and Orpheus are not intended as descriptions of improb-
able facts. Within the prose sections, pagan deities are almost entirely
absent. There is indeed a reference to jove's two vessels, one containing
good and one evil, in Libra II, Prosa 2, but these are referred to as some-
thing that Boethius was surely told of when he was a young man, rather
than being referred to without qualification as really existing objects.
Indeed, I would go so far as to suggest that one function the poetry
serves is to emphasize, by means of contrast, the factual nature of the
philosophical propositions advanced in the prose sections.
When contrasting 'facts' with metaphors, we often refer to 'hard'
facts, or 'brute' facts. Facts are the things we will have to face up to
sooner or later whether we want to or not, and by describing them
as hard, we indicate that perhaps we may not want to. When Samuel
johnson attempted to refute Berkeley by kicking a stone, he completely
misunderstood the point of Berkeley's philosophy, but he did exhibit a
grasp of what most people take to be the distinguishing feature of real-
ity: it gets in your way. In Libra II, Prosa 1, when Philosophy reminds
Boethius that in the past he used 'manly words', her point is that he
should face his troubles with courage, head-on. If her consolation is
really to have the value that she claims for it, it must be such as to
enable Boethius to better understand his position on the basis of solid
facts, rather than merely distracting him with airy persiflage lacking in
any real content.
40 Benjamin Murphy

In emphasizing that the prose passages contain factual content I am

not trying to deny that there is also emotional content. My suggestion
is not that the switch from poetry to prose indicates a turning away
from emotion as such, rather, it indicates a shift in emotional register
as Boethius's writing adopts a suitably manly tone. The emotions are
those of a warrior, boldly staring the enemy down. The prose sections
are factual not because they are written from a neutral standpoint, but
because Boethius is determined not to flinch in the face of reality.
The application of Donaldson's modes of consciousness to the Conso-
lations should, by now, be apparent. Boethius's situation is a bad one,
and it is made all the worse by the fact that he is stuck in line-mode,
reviewing his past decisions and preparing for the future. Line-mode
thinking is useful when we are trying to take responsibility for our lives
and make plans, but Boethius has now lost the ability to control his own
life. Line-mode thinking only accentuates his frustration. There are no
more plans to be made, and so he must turn his mind to something else.
Initially, he turns to the value-sensing mode, as he explores the ques-
tion of what is truly valuable in life. However, his answer, that what has
always truly mattered to him is the pursuit of truth by means of philos-
ophy, leads him naturally into the intellectual modes, and by the final
book, where he considers the abstract issue of God's foreknowledge, he
becomes caught up in the intellectual transcendent mode. The intellec-
tual transcendent mode offers an escape from the line-mode, and thus
carries with it a sense that we are stepping outside of time. Boethius's
goal, of course, is to attain true happiness by becoming more like God,
a being whose state of mind - grasping everything all at once - is pre-
cisely that of someone in the intellectual transcendent mode, as all the
different pieces of the puzzle are fitted into their proper place. Ward's
statement about what it means to have a vision of God can here be
applied to Boethius; it is a vision 'of one's self, as transfigured by the
infinite' (Ward, 1987/1998, p. 153).
The use of intellectual problems as a means of coping with stressful
situations is not, of course, unique to theology. In 1908, the German
industrialist Paul Wolfskehl made plans to commit suicide at mid-
night precisely. With everything prepared, he decided to spend his
last remaining hours studying a paper that dealt with Fermat's infa-
mous last theorem. He noticed an error in the paper he was reading,
and became engrossed in finding a way to correct it. By dawn, he had
completed the correction and made the choice not to commit suicide.
Later, he established a prize for anyone who could definitively prove
the theorem (Singh, 1997, pp. 133-7). I cannot pretend to have any
 Why Would Anyone Believe in a Timeless God? 41

great mathematical ability, but I do always try to remember to take

with me a book of Sudoku puzzles when I fly, as an antidote to the
boredom of the departure lounge and a distraction from the terrors
of a bumpy flight. So there is nothing particularly special or unusual
about responding to stress by choosing to manipulate one's mode of
consciousness. However, when I focus on my Sudoku puzzle, I am aware
of the fact that in doing so, I am choosing to shut out the uncomfort-
able situation in which I find myself. I might say, afterwards, that I had
successfully generated the illusion that time had flown by more quickly.
Once the flight is over, I can put my Sudoku puzzles aside. They are
useful when I need to be distracted, but they are not weighty problems
that demand a resolution. Boethius and Wolfskehl found consolation by
turning to tasks that they considered important. In Boethius's case, there
is the added bonus that the conclusion he arrives at - that God experi-
ences all things simultaneously- implies that insofar as he has attained
something approaching a timeless perception, freeing himself from the
stresses of the line-mode, what he has attained is not a delusion, a refuge
from reality, but something closer to the most accurate way of perceiv-
ing reality, the way it is perceived by God. Aquinas, too, affirms that
God grants some people a share of his eternity, including the experience
of unchanging thoughts (Summa Theologiae, lae, Q. 10, Art. 11).
This, then, is what I mean by experience-enhancing theology, and it
is important to note that I mean by this something more than theol-
ogy that evokes certain experiences. Certainly, an essential feature of
the Consolations is that we follow Boethius on a journey through dif-
ferent states of consciousness, each of which we are able to share by a
sympathetic reading of his rich descriptions. However, as I have already
noted, if this were all that were claimed for theology, it would seem to
be merely an inferior substitute for music. What theology adds is an
evaluation of the cognitive value of the states of consciousness we are
going through, that is, an assessment as to whether they are helping us
see reality from a better perspective, or drawing us in to potentially dan-
gerous illusions. The intellectual element is not only part of the means
by which Boethius achieves a different state of consciousness; having
the backing of reason is necessary to generate a positive evaluation of
his new state of mind.

4. Conclusion
At the start of this essay, I said that if theology is a complete waste of
time, that is bad news for professional and amateur theologians alike.
42 Ben;amin Murphy

The explanations for the existence of theology provided by cognitive

scientists of religion (with the exception, perhaps, of Pyysiainen) are
really explanations for the existence of a professional class of religious
experts who provide a service to the religious community and receive
a reward (at the very least, a high status within the community) as a
return. They are able to reassure believers that difficult questions can be
solved (Dennett), and by ensuring a proper balance between A-system
and B-system thinking, they maintain the stability of religious thought
over time, providing a systematic ideological basis for religious commu-
nities (Whitehouse). These tasks may well be essential to the long-term
health of a religious community, but they can safely be delegated to a
few recognized experts who, by their investment of time and intellec-
tual effort, demonstrate a high level of commitment to the community
(Bulbulia). Experience-enhancing theology, on the other hand, is by its
nature an amateur activity insofar as the benefits attained from practic-
ing such a discipline accrue only to the practitioner. The term 'amateur'
is not, of course, meant as a term of disparagement, nor do I mean that
Boethius succeeded only in consoling himself. On the contrary, I am
sure that through the centuries, many people have found comfort in
reading the Consolations in time of crisis. However, some documents
bring comfort by their mere presence. I keep a copy of my insurance
policy in my car, and I am reassured by the knowledge that if I ever
need my car to be fixed, I will be able to call upon a qualified person
to do it. It is not written in my native language, and I have never felt
the necessity of reading the small print. Its presence alone is a comfort.
The Consolations is not like this: left to gather dust on the shelf it will
provide no comfort until the day when the owner is ready to pick it up
and read it.
One reason that cognitive scientists may have overlooked this form of
amateur theology is that they tend not to be interested in altered states
of consciousness, arguing that profound mystical experiences are not
sufficiently widespread amongst religious believers to account for the
prevalence of faith (Barrett, 2004, pp. 121-2). Indeed, it is somewhat
refreshing that cognitive scientists explicitly raise the point that, for
much of the time, the dominant state of mind for many participants in
religious rituals is one of boredom (Whitehouse, 2004, pp. 97-9). How-
ever, the point about the paucity of mystical experiences, even if it is
true, cannot be raised as an objection to the view I am proposing here.
In the first place, I am not attempting to offer an explanation for the
popularity of religious belief, which cognitive scientists are confident
 Wily Would Anyone Believe in a Timeless God? 43

they can explain, only for the theological beliefs of a minority, with
which cognitive scientists seem to have more trouble. Second, the main
goal of Donaldson's theory is not to explain highly unusual altered states
of consciousness, although her work might have some relevance for
such a study, but fluctuations in our state of mind that are a normal
part of human development. It is no part of my argument that Boethius
achieved some ineffable state of consciousness that is beyond the lim-
its of normal human experience. Rather I am proposing that a state of
mind that is not in itself unusual was given the seal of rational approval
because of the content of his theology.
So, perhaps cognitive scientists of religion have something to learn
from theologians about the motives for engaging in theology. If these
motives remain obscure, cognitive scientists are liable to wonder
whether theology is, in fact, a waste of time. The charge that this
is the case is one that theologians must attempt to refute, and in
attempting to meet this challenge, we must consider carefully the goals
of theological activity, and the ways we have of recognizing whether
those goals are being met. In other words, we must make explicit the
standards by which we judge contributions to theology. My own pro-
posal is that there are two sets of standards, which can, potentially,
conflict. Technical theology aims as far as possible to preserve a set
of intuitively attractive beliefs in the face of intellectual challenges.
Experience-enhancing theology aims to provide an intellectual frame-
work that reinforces some of the emotional resources that all of us can
draw on to deal with the challenges of life. In this essay I have attempted
to show how the existence of these two types of theology explains the
persistent attraction of divine timelessness, despite the fact that this
conflicts with popular religious ideas. Of course, to establish that this
is an accurate model of theology would require further work, focusing
on other divine attributes. That must be left for another occasion.
My concern in this essay has been to make explicit the reasons why
anyone would want to believe in a timeless God rather than with the
question of whether there is such a being. The best justification of time
spent on theology would be to demonstrate that theology is capable
of discovering truths. However, my goal was to look at implications of
cognitive science of religion for the practice of theology, and I do not
think that cognitive science can tell us anything one way or the other
about the truth of religious beliefs. But if cognitive science does not offer
a new source of truths about God, it can at least help theologians attain
a higher level of self-knowledge. 5
44 Benjamin Murphy

Notes
1. See Barrett, 2000 for a useful survey.
2. I take responsibility for this translation, although I make no claim to
originality.
3. Donaldson provides a useful summary of the various modes in 1992,
pp. 267-70.
4. The quotation appears in Donaldson, 1992, p. 155. It is taken from Ward,
1987/1998, p. 3. Donaldson cites Ward's book by its original title, Images of
Eternity. It was republished as Concepts of God.
5. This essay is based on a paper with the same title that I presented at the con-
ference on The Concept of God and the Cognitive Science of Religion, which
took place at the University of Birmingham, June 2009. I would like to thank
Florida State University- Panama's Faculty Development Committee for pro-
viding me with the funding to attend this conference, Yujin Nagasawa for
organizing the conferences and the conference sponsors, which were the John
Templeton Foundation and the Cognition, Religion and Theology Project at
the University of Oxford. I received a lot of useful feedback, which has enabled
me to improve the paper.

References
Aristotle (1999) Nicomacllean Ethics, trans. Terence Irwin (Indianapolis: Hackett
Publishing Company).
Barrett, J. (2000) 'Exploring the Natural Foundations of Religion', Trends In Cog-
nitive Sdence 4.1: 29-34.
Barrett, J. (2004) Why Would Anyone Believe in God? (Walnut Creek, CA: Altamira
Press).
Barrett, J., and F. Keil (1996) 'Conceptualizing a Nonnatural Entity: Anthropo-
morphism and God Concepts', Cognitive Psychology 31: 219-47.
Boethius (1990) Consolatio Philosophiae, ed. J. O'Donnell (Bryn Mawr, PA:
Bryn Mawr College). Available online at www9.georgetown.edu/faculty/jod/
boethius/boethius.html.
Boyer, P. (2001) Religion Explained (New York: Basic Books).
Bulbulia, J. (2009) 'Religiosity as Mental Time Travel: Cognitive Adaptations for
Religious Behavior', in]. Schloss and M. Murray (eds), The Believing Primate,
(New York: Oxford University Press), pp. 44-75.
Davidson, D. (2001) 'Mental Events', in idem, Essays on Actions and Events
(Oxford: Clarendon Press), pp. 207-27.
Dennett, D. (2006) Breaking The Spell (London: Penguin).
Donaldson, M. (1992) Human Minds (London: Penguin).
Helm, P. (2001) 'Divine Timeless Eternity', in Gary Ganssle (ed.), God and Time
(Downers Grove, IL: Inter Varsity Press), pp. 28-60.
Kenny, A. (1979) The God of the Philosophers (Oxford: Oxford University Press).
Marenbon, J. (2003) Boethius (New York: Oxford University Press).
Pike, N. (2002) God and Timelessness (Eugene, OR: Wipf & Stock).
Pyysiainen, I. (2004) 'Intuitive and Explicit in Religious Thought', Journal of
Cognition and Culture 4.1: 123-50.
 Wily Would Anyone Believe in a Timeless God? 45

Singh, S. (1997) Fennat's Last Theorem (London: Fourth Estate).

Slone, D. J. (2004) Theological Incorrectness - Wily Religious People Believe What
They Shouldn't (New York: Oxford University Press).
Stump, E., and N. Kretzmann (1981) 'Eternity', Journal of Philosophy 78: 429-57.
Ward, K. (1987) Images of Eternity (London: Darton, Longman and Todd);
republished as (1988) Concepts of God (Oxford: Oneworld).
Whitehouse, H. (2004) Modes of Religiosity (Walnut Creek, CA: Altamira Press).
Part II
God, Creation and Evolution
3
Darwin's Argument from Evil
Paul Draper

The position that evolutionary biology makes the problem of evil worse
for the theist is hardly new. Indeed, as we shall see, Darwin himself
seems to have held it, and its more recent defenders include philoso-
phers, scientists, historians and even theologians (though some of the
evangelical Christian theologians who hold this position infer from it
not that there is no God, but instead that the theory of evolution is
false!). Consider, for example, the following three frequently quoted
remarks. According to Bertrand Russell (1997, pp. 79-80},

Religion in our day, has accommodated itself to the doctrine of evo-

lution .... We are told that ... evolution is the unfolding of an idea
which has been in the mind of God throughout. It appears that
during those ages ... when animals were torturing each other with
ferocious horns and agonizing stings, Omnipotence was quietly wait-
ing for the ultimate emergence of man, with his still more exquisite
powers of torture and his far more widely diffused cruelty. Why the
Creator should have preferred to reach His goal by a process, instead
of going straight to it, these modern theologians do not tell us.

jacques Monod (1976) claims that

[Natural] selection is the blindest, and most cruel way of evolving

new species, and more and more complex and refined organisms ....
The struggle for life and elimination of the weakest is a horrible pro-
cess, against which our whole modern ethics revolts. An ideal society
is a non-selective society, one where the weak is protected; which is
exactly the reverse of the so-called natural law. I am surprised that a

49
SO Paul Draper

Christian would defend the idea that this is the process which God
more or less set up in order to have evolution.

Finally, David Hull (1991, p. 486) maintains that

The evolutionary process is rife with happenstance, contingency,

incredible waste, death, pain and horror.... the God implied by evo-
lutionary theory and the data of natural history ... is ... not a loving
God who cares about His productions. He is ... careless, wasteful,
indifferent, almost diabolical. He is certainly not the sort of God to
whom anyone would be inclined to pray.

These are rhetorically powerful statements, expressed with great con-

viction by highly respected thinkers; but at best they only hint at any
sort of serious argument against the existence of God. Indeed, fully devel-
oped Darwinian arguments from evil (let alone convincing ones) are not
easy to find. The politically correct stance nowadays, held by most athe-
ists and almost all theists who accept evolution, is to deny the relevance
of evolution to the issue of God's existence. 1 And some of those who dis-
sent (e.g. Ayala, 2007) actually believe that evolution helps to solve or
at least mitigate the problem of evil! Following a line of reasoning sug-
gested by Darwin, I will dissent in the opposite direction by proposing
a serious Darwinian argument from evil, an argument that is strongly
suggested by Darwin's discussion of the problem of evil in his autobi-
ography and elsewhere. My strategy (or, arguably, Darwin's strategy) is
to use Darwin's theory of natural selection as a sort of 'atheodicy'- an
explanation of a broad range of facts about good and evil that works
much better when Darwin's theory is combined with a serious atheis-
tic hypothesis than when it is combined with (orthodox) theism. I will
begin by getting clear on some crucial terminology. Then I will offer an
interpretation of Darwin's views about religion in general and about the
problem of evil in particular. Finally, after identifying what is arguably
Darwin's own argument from evil, I will try to develop and defend
the argument, using some of Darwin's ideas, some ideas taken from
contemporary evolutionary biology, and a few of my own ideas.

1. Terminology
By 'evolution' I mean the conjunction of two theses. The first asserts
that evolution did in fact occur - complex life did evolve from sim-
pler life. Specifically, it is the view that all multicellular organisms and
 Darwin's Argument from Evil 51

all (relatively) complex unicellular organisms on earth (both present

and past) are the (more or less) gradually modified descendants of a
small number of relatively simple unicellular organisms. The second
addresses the issue of how evolution occurred. It states that almost
all evolutionary change in populations of complex organisms occurs
because of trans-generational genetic change or, more broadly, trans-
generational change in nucleic acids. This thesis was not, of course,
established until after Darwin's death, but it has since become an essen-
tial part of our understanding of evolution. While this second thesis
is a claim about the mechanisms by which evolution takes place, it is
important to distinguish it from the more specific claim that natural
selection operating (indirectly) on random genetic mutation is the prin-
cipal mechanism driving evolutionary change (or at least the principal
mechanism driving the evolutionary change that results in increased
complexity). It is this claim that I call 'the theory of natural selection'
or sometimes just 'Darwin's theory', though Darwin of course did not
know about the role of genetic mutation in producing the inheritable
phenotypic traits that nature (directly) selects. In those parts of the essay
in which I discuss specifically Darwin's own views, the phrase 'operat-
ing on random genetic mutation' can be understood to be replaced by
'operating on random variation'. Finally, I use the term 'Darwinian evo-
lution' to refer to the conjunction of evolution and the theory of natural
selection.
Darwin's main contribution to the theory of natural selection is the
idea of natural selection. According to Darwin's understanding of this
idea, evolutionary change results from competition for survival, ran-
dom variation and heredity. A population of organisms will reproduce
at a rate that guarantees a greater number of offspring than parents. This
results in overpopulation, which is one of the main causes of competi-
tion for survival. This competition leads to evolutionary change because
of variation and heredity. Although the characteristics of organisms in
the population will vary only slightly and only randomly, such variation
will suffice to give some organisms an advantage in the struggle for sur-
vival, thereby making these organisms more likely to reproduce. Because
of heredity, these advantageous characteristics will often be passed on
to offspring, and those offspring will also be more likely to reproduce.
Thus, over generations the frequency of those characteristics in the pop-
ulation will gradually increase until eventually they become normal for
the population. At this point, it is correct to say that the population has
evolved. When enough of these small evolutionary changes accumulate
over very long periods of time, the result is a new species.
52 Paul Draper

According to Darwin, however, not all evolutionary change occurs

because a characteristic confers an advantage in the struggle to sur-
vive. Some change occurs because a characteristic, though not itself
advantageous, is correlated with characteristics that do confer such an
advantage. Darwin called this 'correlated variation'. (Today it is asso-
ciated with pleiotropy.) And still other change occurs, not because a
characteristic makes an organism more likely to survive, but because it
makes an organism more likely to reproduce, given that it survives long
enough to do so. Darwin called this 'sexual selection'. Although Darwin
restricted the term 'natural selection' to the first of these three mecha-
nisms (which is sometimes called 'survival selection'), I will understand
it to include all three (and others as well). Our contemporary under-
standing of natural selection is typically broadened to include a variety
of causes, though not of course every cause, of differential reproduction.

2. Darwin on Religion
Darwin spent a fair amount of time reflecting on religion and espe-
cially on the religious implications of his theory of natural selection.
In the part of his autobiography describing his views on religion and
how they developed over time, he first briefly explains hvw he came
to reject Christianity (1958a, pp. 71-2) and then begins a lengthier dis-
cussion of his reflections on the question of whether or not a personal
(though non-Christian) God exists. For the most part, his focus is on
the weaknesses of various arguments for God's existence, starting with
William Paley's famous design argument. He had read Paley's Natural
Theology (and two other books by Paley) while studying for a BA degree
at Cambridge. Although Paley's argument previously seemed conclusive
to him, he rejects the argument after discovering 'the law of natural
selection', since that law shows that intelligent design is not necessary
to account for the 'endless beautiful adaptations' in living things (1958a,
p. 73). Interestingly, he distinguishes the issue of how to account for
adaptations from the issue of whether 'the generally beneficent arrange-
ment of the world [can] be accounted for' (ibid.). He claims, however,
that natural selection can also account for this arrangement. I will return
to his argument for this claim in section S. After his discussion of the
problem of evil, to which I will return momentarily, he rejects a number
of other reasons for believing in God. He claims, however, that as late as
the time he wrote Origin ofSpecies he still could be considered a 'Theist'-
'Deist' would have been a better choice of words given his views on the
problem of evil - because at that time he still accepted some sort of
 Darwin's Argument from Evil 53

cosmological argument from the existence of an 'immense and wonder-

ful universe' to a 'First Cause having an intelligent mind in some degree
analogous to that of man' (19S8a, p. 77). He later came to believe, how-
ever, that 'The mystery of the beginning of all things is insoluble by us',
on the basis of which he declares himself to be an 'Agnostic' (1958a,
p. 78).
While Darwin's focus in the Autobiography is, as I said, on the absence
of good reasons to believe in God, he also thought that there are good
reasons to reject the omnipotent, omniscient and omnibenevolent God
of orthodox theism. He was, without question, an atheist about that sort
of God. 2 There is evidence of this both in his autobiography, especially
as we shall see in the passage on the problem of evil, and also in his cor-
respondence, especially his letters to the botanist (and orthodox theist)
Asa Gray, who was a leading defender of Darwin's theory in the USA,
but who wanted to reconcile it with teleology, divine design and even
natural theology.
One of the most significant passages in the Autobiography (1958a,
p. 75), especially for my purposes, is this one:

That there is much suffering in the world no one disputes. Some have
attempted to explain this in reference to man by imagining that it
serves for his moral improvement. But the number of men in the
world is as nothing when compared with that of all other sentient
beings, and these often suffer greatly with no moral improvement.
A being so powerful and so full of knowledge as a God who could cre-
ate the universe, is to our finite minds omnipotent and omniscient,
and it revolts our understanding to suppose that his benevolence is
not [sic] unbounded, for what advantage can there be in the suffering
of millions of the lower animals throughout almost endless time. This
very old argument from the existence of suffering against the exis-
tence of an intelligent first cause seems to me a strong one; whereas,
as just remarked, the presence of much suffering agrees well with the
view that all organic beings have been developed through variation
and natural selection.

This passage raises two questions. First, why does Darwin believe that
theism's failure to explain suffering is an argument against theism?
No theory (whether metaphysical or scientific) can explain even a small
portion of the facts that need explanation. Thus, so long as the hypoth-
esis that God exists is logically compatible with suffering, its failure
to explain it appears no more significant than its failure to explain
54 Paul Draper

the tides. Second, and even more importantly for my purposes, why
does Darwin compare theism to the theory of natural selection here?
Granted, that theory can explain many biological and psychological
facts, including facts about pain and pleasure, better than theism can;
but how is that supposed to strengthen the argument from evil against
theism? Atomic theory can explain many chemical facts much better
than theism, but that's no reason to doubt theism. After all, atomic
theory can also explain many facts better than naturalism or deism or
pantheism or any other alternative to theism that is not ad hoc.

3. A Preliminary Interpretation of Darwin's

Argument from Evil
One possible answer to both of these questions is that Darwin holds
that theism and the theory of natural selection are 'competing' or 'alter-
native' hypotheses in the sense that they are (known to be) mutually
exclusive, at least on our background knowledge. (Two hypotheses are
'mutually exclusive on our background knowledge' just in case our back-
ground knowledge entails that at most one of them is true.) If Darwin
believes this, then Darwin's answer both to the question of why the-
ists need a theistic explanation of suffering and to the question of how
his scientific theory of natural selection is relevant to the philosophical
problem of evil is that his theory provides a good atheistic explanation
of various facts about suffering (and pleasure). Chemists clearly cannot
make any similar claim about atomic theory, which explains various
chemical facts in a way that is, as far as we can tell, perfectly neutral
between theism and atheism. A good atheistic explanation of what we
know about suffering would threaten theism because, even though the-
istic belief is more often based on testimony (the testimony of one's
parents in most cases) than on theism's explanatory power, theism's
credibility is still challenged if a serious competing view, a competing
view that is neither ad hoc nor initially less plausible than theism,
explains certain facts better than theism does. Such a challenge would
be a good prima facie reason to believe that the alternative view is more
likely than theism to be true and so would be a good prima facie reason
to believe that theism is probably false. While this is an interesting and
plausible interpretation of Darwin's views, I intend to offer and defend
a different interpretation for two reasons.
First, while some of Darwin's contemporaries regarded his theory as
atheistic, 3 and while some of Darwin's remarks in various letters sug-
gest that he believes that his theory is incompatible with the sort of
 Darwin's Argument from Evil 55

providence implied by orthodox theism, I doubt this interpretation can

be correct simply because Darwin regards his view that natural selec-
tion drives most evolution as a highly confirmed theory, not as a mere
hypothesis or conjecture. Thus, if he really believed that his theory
entails atheism, then he would have no reason to appeal to an argument
from suffering in order to defend his disbelief in beneficent design. He
would simply infer the falsity of beneficent design and so of orthodox
theism directly from the truth of his theory. Granted, he does reject Asa
Gray's view that the (beneficial) variations selected by nature are 'inten-
tionally and specially guided', but it is arguable that what he objects
to here is the idea that God supplies these variations by miraculously
intervening in nature, which would mean that the variations do not
result from the operation of natural law. Since orthodox theism implies
the existence of an omnipotent and omniscient God, it seems to allow
for there being divine planning and remote causation (and so design)
without direct divine action in nature. Darwin himself emphasizes that
his theory is consistent with the view that all natural events result from
designed laws. To be sure, he also favours the view that whatever good
or ill results from the operation of these laws is a matter of chance, but
that, I believe, is an inference he draws at least in part from the suffering
in the world, not solely from the tenets of his theory.
This leads to or has already arrived at my second reason for reject-
ing any interpretation of Darwin's argument from evil that depends
on Darwin taking his theory of natural selection (or its conjunction
with background knowledge) to entail the falsity of orthodox theism.
To interpret Darwin's theory this way is to attribute a false or at least
poorly supported position to Darwin. The claim that most evolution-
ary change and most adaptations result from natural selection simply
does not, so far as we can tell, entail atheism - not by itself, not when
conjoined with what Darwin knew, not even when conjoined with cur-
rent biological knowledge. This is controversial. Some thinkers, both at
Darwin's time and today, believe that orthodox theism is incompatible
with the randomness entailed by natural selection. Thus, before I move
on to my preferred interpretation of Darwin's argument from evil, I will
examine three arguments for such an incompatibility.
The first argument proceeds as follows. If the universe was created
by a (temporal) omniscient being, then everything (with the possible
exception of causally undetermined events) is foreknown. And if this
omniscient being is also omnipotent and morally perfect, then anything
affecting the well-being of sentient creatures was presumably not just
foreknown but planned as well. But many particular aspects of nature,
56 Paul Draper

including much of the variation in populations of organisms, affect the

well-being of those organisms. Thus, orthodox theism entails that these
aspects were planned, and so is incompatible with Darwin's theory,
which entails that many of them were random and hence unplanned.
This argument fails because it equivocates on the meaning of the word
'random'. Natural selection entails that useful variations in a population
of organisms are random both in the sense that humans (because of our
ignorance) cannot predict them and in the sense that any changes in
the environment that make a particular variation useful do not directly
cause that variation. But being random in these senses is consistent with
being non-random in another sense. Specifically, as suggested above, it
is consistent with being planned and indirectly caused by an omnipo-
tent, omniscient and morally perfect person. If such a person created
the universe and the natural laws that govern it, then that person might
have planned, in accordance with those laws, exactly how the envi-
ronment would change, exactly which variations would occur, exactly
which traits would be selected for, and so on. Providence of this sort
would not involve the sort of miraculous intervention in the devel-
opmental history of organic beings that would be incompatible with
Darwin's theory. 4
A second reason for thinking that orthodox theism is incompatible
with the randomness entailed by the theory of natural selection is that
some genetic mutations, for example mutations caused by x-ray irradi-
ation, may depend on sub-atomic events governed by the probabilistic
laws of quantum physics. 5 If, as most physicists believe, such events
are causally undetermined, then some variations may be random in a
stronger sense than even Darwin ever imagined. Thus, if we assume
a temporal God and the logical impossibility of foreknowing causally
undetermined events, then some of the variations upon which natural
selection operates could not be planned even by an omnipotent and
omniscient God. However, as long as a temporal person's being omni-
scient at some time only requires having as much knowledge as it is
metaphysically possible to have at that time, this too is compatible
(at least so far as we can tell) with orthodox theism. For an omni-
scient God would still know at any given time which future variations
are physically possible, and so long as such a God had a providential
plan for every contingency, no conflict with God's omnibenevolence is
implied.
Finally, it might be argued that, because the variation involved in
natural selection is random, natural selection is an inherently waste-
ful process, and it is this wastefulness that is incompatible with theism.
 Darwin's Argument from Evil 57

Now it is true that random variations are often 'useless' in the sense
that they are not advantageous and may even be disadvantageous in
the struggle to survive. It is also true that natural selection entails com-
petition for survival and not all organisms can win such a competition.
As others have pointed out, however, it is a mistake to claim that an
omnipotent and omniscient person who produced such a process would
be guilty specifically of 'waste'; for waste is possible only when one's
resources are limited, and an omnipotent and omniscient being has
unlimited resources. Of course, to the extent that such waste harms sen-
tient beings, it may be incompatible with theism; but natural selection
does not entail that such harm occurs nor does it entail that God has no
good reason for causing or permitting that harm.
No doubt other possible reasons might be offered for believing that
orthodox theism and the theory of natural selection are mutually exclu-
sive given our background knowledge; but I am fairly confident that
these will be no more successful than the three I have considered. If the
literature on the logical problem of evil teaches us anything, it is that
deriving specific claims about the world from theism is very difficult and
so showing that theism is logically incompatible with our knowledge or
theories about the world is also very difficult. Thus, if we want to avoid
attributing to Darwin a false or at best poorly supported view about the
relationship between his theory and orthodox theism, then we must
reject this first interpretation of his argument from evil.

4. A Second Interpretation of Darwin's Argument

Fortunately, this doesn't return us to square one when it comes to
interpreting Darwin's argument. We can keep the basic structure of the
argument under the first interpretation, as long as we modify the argu-
ment's details enough to avoid the two objections to that interpretation
but not so much that it loses all plausibility as an interpretation of what
Darwin had in mind. Specifically, we can take as the hypothesis that
competes with orthodox theism another view that Darwin defended,
namely, that living things and their adaptations are not intentionally
designed. Let's call this the 'no-design hypothesis'. Shortly before the
passage in the Autobiography on the argument from suffering, he claims
that 'There seems to be no more design in the variability of organic
beings and in the action of natural selection, than in the course which
the wind blows' (1958a, p. 73). 6 Further, he repeatedly (both in the Auto-
biography and elsewhere) connects his view that living things are not
designed with examples involving suffering or death. For example, his
58 Paul Draper

most famous comment about the problem of evil is in reality a comment

about the absence of design in the world: 'I cannot persuade myself that
a beneficent and omnipotent God would have designedly created the Ich-
neumonidae with the express intention of their feeding within the living
bodies of caterpillars' (19S8b, p. 249; my italics). What all this demon-
strates is the plausibility of taking the no-design hypothesis to be the
alternative to orthodox theism that Darwin implicitly has in mind when
he discusses the problem of evil in the Autobiography.
The obvious advantage of this interpretation is that the no-design
hypothesis, unlike the theory of natural selection, really is incompatible
with orthodox theism. Of course, the purposes for which God designed
living things may be unknown to us and Paley may be wrong that
adaptations were designed in all cases for the purpose of benefiting the
organisms that possess those adaptations; but if an omnipotent, omni-
scient, and omnibenevolent God created the universe, then the living
world must be a providential one, and thus the ordered systems making
up the living world are foreseen (at least as possible outcomes of an inde-
terministic process), ordained (at least relative to God's consequent will)
and thus designed. Of course, the crucial question about the no-design
hypothesis is whether it can explain any facts about good and evil better
than orthodox theism. This is where Darwin's theory of natural selection
comes into play on this interpretation. That theory can serve as a good
'atheodicy': an explanation of various facts about good and evil that
works much better on the assumption that an alternative to theism- in
this case the no-design hypothesis - is true than on the assumption that
orthodox theism is true.
The disadvantage of this interpretation is that in the crucial passage
about the problem of evil in his autobiography quoted earlier, Darwin
doesn't mention the no-design hypothesis. He mentions his theory of
natural selection. Recall the two questions about that passage discussed
earlier. The first question was why Darwin thinks that theism needs
to explain known facts about suffering. Both interpretations answer
that question as follows: because a plausible atheistic hypothesis can
explain those facts. The advantage of the second interpretation over
the first is that it makes this answer true! The second question was
why Darwin believes that his theory of natural selection is relevant to
the problem of evil. The first interpretation can answer this question
more easily than the second because it takes natural selection to be the
crucial alternative hypothesis to orthodox theism. On the second inter-
pretation, the answer is more complicated. Darwin's theory becomes, in
effect, an auxiliary hypothesis that can successfully explain the suffering
 DaiWin's Argument from Evil 59

(and pleasure) in the world when it is combined with the no-design

hypothesis but not when it is combined with orthodox theism.
Of course, it is far from obvious that this answer to the second ques-
tion is true. In the remainder of this chapter, I will offer some pro tanto
reasons for thinking that it is true. I will divide my task into two parts.
First, I will examine what facts about good and evil the theory of natural
selection can explain. And second, I will argue that these explanations
are much more successful when Darwin's theory is combined with the
no-design hypothesis than when it is combined with orthodox theism.

5. Darwinian Explanations of Good and Evil

Let's begin with what Darwin says in the Autobiography. As I mentioned
in section 2, Darwin thinks that his theory explains the 'generally benef-
icent arrangement of the world'. By the 'beneficent arrangement of the
world', he means to refer to the alleged fact that, when it comes to the
quantity of pleasure and suffering in the world, happiness 'decidedly
prevails' over misery (1958a, pp. 73-4). He explains this alleged fact as
follows:

Every one who believes, as I do, that all the corporeal and mental
organs (excepting those which are [not advantageous] to the posses-
sor) of all beings have been developed through natural selection, or
the survival of the fittest, together with use or habit, will admit that
these organs have been formed so that their possessors may com-
pete successfully with other beings, and thus increase their number.
Now an animal may be led to pursue that course of action which is
the most beneficial to the species by suffering, such as pain, hunger,
thirst, and fear, - or by pleasure, as in eating and drinking and in
the propagation of the species, &c. or by both means combined, as
in the search for food. But pain or suffering of any kind, if long
continued, causes depression and lessens the power of action; yet
is well adapted to make a creature guard itself against any great or
sudden evil. Pleasurable sensations, on the other hand, may be long
continued without any depressing effect; on the contrary they stim-
ulate the whole system to increased action. Hence it has come to
pass that most or all sentient beings have been developed in such a
manner through natural selection, that pleasurable sensations serve
as their habitual guides. We see this in the pleasure from exertion,
even occasionally from great exertion of the body or mind, - in the
pleasure of our daily meals, and especially in the pleasure derived
60 Paul Draper

from sociability and from loving our families. The sum of such plea-
sures as these, which are habitual or frequently recurrent, give, as
I can hardly doubt, to most sentient beings an excess of happiness
over misery, although many occasionally suffer much. Such suffering,
is quite compatible with the belief in Natural Selection, which is not
perfect in its action, but tends only to render each species as success-
ful as possible in the battle for life with other species, in wonderfully
complex and changing circumstances. (1958a, pp. 74-S)

One problem here is that it's far from clear that pleasure really is pre-
dominant over suffering, and thus the alleged fact that Darwin is trying
to explain may not actually obtain. Darwin himself was unsure. Though
he says in the passage quoted above that he can 'hardly doubt' the
predominance of pleasure, he admits elsewhere that it 'would be very
difficult to prove' (1958a, p. 74). I believe that pleasure may be more
common than pain in sentient beings that are flourishing (and notice
too that Darwin's reasons for thinking the operation of natural selec-
tion would lead to the predominance of pleasure really only apply to
beings that are flourishing). But a state of flourishing is not the norm
for sentient beings in a Darwinian world, so Darwin's view is much too
optimistic. We can, however, make use of some parts of his explanation
of the alleged fact that pleasure predominates to explain the known fact
that both pain and pleasure are used to promote reproductive success.
Why not just suffering? Because excessive amounts of suffering tend
to weaken an organism in a variety of ways. Why not just pleasure?
Because pain works better when great or sudden dangers require a quick
or extreme response. Of course, another advantage of being motivated
by both suffering and pleasure, especially when it comes to biologically
important activities like eating, is that one sort of motivation can still
function if the other fails. For example, if an animal no longer derives
pleasure from the taste of food, hunger may still motivate that animal
to eat.
The theory of natural selection can also explain why sentient beings
still suffer or feel pleasure even when, because of physical limitations or
circumstances, such suffering and pleasure cannot causally contribute
to survival or reproduction. For example, why does a person hopelessly
trapped in a fire still feel pain? And why does a man still feel sexual
pleasure after a vasectomy? If Darwin's theory is true, then one would
not expect the mechanisms that produce pain and pleasure to be so fine-
tuned as to eliminate such biologically useless pain and pleasure. For
such fine-tuning would confer no advantage in the struggle for survival.
 Darwin's Argument from Evil 61

There are also a variety of more specific facts about suffering and
pleasure that can be explained by the theory of natural selection, for
example facts about the intensity of various sorts of suffering and
pleasure. It can also explain a variety of facts indirectly, for example
by explaining the anatomical causes of certain kinds of suffering. But
instead of discussing these, which won't in the end make much differ-
ence to the strength of my argument, I would like to turn my attention
to certain facts about goods and evils other than pleasure and suffer-
ing, specifically, the flourishing and languishing of sentient beings and
the mixture of moral goodness and moral badness that we find in the
human species.
Facts about flourishing and languishing are not reducible to facts
about pleasure and suffering, since non-sentient organisms can flour-
ish or languish and since a sentient organism might suffer much but
still flourish, while another might, though this is much less likely, suffer
little but still languish. Butler (1857, p. 253) wrote that

of the numerous seeds of vegetables and bodies of animals, which are

adapted and put in the way to improve to such a point or state of
natural maturity and perfection, we do not see perhaps that one in a
million actually does. Far the greatest part of them decay before they
are improved to it; and appear to be absolutely destroyed.

I don't know how Butler arrived at his figure of less than one in a mil-
lion, but he's certainly right that the vast majority of living things,
including sentient beings, never flourish, many more flourish for only
a very small portion of their lives, and almost none who live a full life
flourish for all of it. A Darwinian explanation of this sad fact is read-
ily available. If populations of organisms increase geometrically and
this leads to competition for the resources necessary to survive, then
inevitably a large percentage of all living things will not survive long
enough to thrive, many more will barely survive and thus languish for
all or almost all of their lives, and even those organisms that do flourish
for much of their lives will, if they live long enough, ultimately lan-
guish in old age. A Darwinian world is inevitably cruel, especially to the
young, the old and the genetically less fortunate.
Turning now to moral goods and evils, here we must proceed very
cautiously. Moral phenomena are extremely diverse and it is very hard
to separate out biological influences from cultural ones. The safest route
is to identify those moral characteristics that humans share with at least
some other animals, and then see if natural selection can explain these.
62 Paul Draper

The most obvious is that we are basically self-centred: our tendency to

behave in ways that promote our own welfare is typically much stronger
than any tendency to act for the good of others. A Darwinian explana-
tion of this is so obvious that it need not even be stated. But, like many
other animals, we will at times sacrifice our own interests for the sake
of others. Such altruistic behaviour can be divided into two types: kin
altruism and non-kin or 'social' altruism.
Darwin himself was unable to offer any convincing explanation of
how altruism of any sort arose. Thanks to the discovery of genes, how-
ever, we now have very plausible explanations of kin altruism. Because
I share as many as half my genes with my kin, characteristics like self-
centredness that help me survive and reproduce are not the only sorts
of characteristics that can increase the likelihood that my genes will
be passed down to future generations; characteristics like caring about
my family members, which promote their survival and reproduction
rather than my own, will also work. Thus, the theory of natural selec-
tion, when enriched by genetics, can account for kin altruism and the
various natural virtues that produce it.
The origin of social altruism is tougher to explain. Since it is found
only in humans and other higher mammals, perhaps it is a sort of
by-product of kin altruism and intelligence. More likely, most social
altruism is a form of reciprocal altruism, for which a variety of plau-
sible Darwinian explanations have been offered. Social altruism is much
weaker in most humans than kin altruism as well as more frequently
absent altogether, and this is not surprising on the theory of natural
selection. Notice also that it is typically very limited in its scope. The less
like Smith some other sentient being is, the less likely it is that Smith will
be concerned about that being's welfare. If, as Darwin thought, kin altru-
ism is the basis for other sorts of altruism, then once again this is not
surprising. Furthermore, it is clearly to be expected on Darwin's theory
that universal altruism, that is, the tendency to sacrifice one's own inter-
ests for the sake of the interests of any sentient being, no matter how
different from us and no matter how unlikely they are to reciprocate, is
very rare in humans and usually very weak in those humans that have it.
Having a strong dose of this characteristic would clearly not be advanta-
geous in the struggle to survive. So its rarity, while unfortunate, is easily
explained by the theory of natural selection. On the whole, then, the
mixture of a basic self-centredness, with limited altruistic tendencies,
can be explained quite plausibly by Darwin's theory.
Darwinian explanations have also been given for a variety of other
more specific moral facts, like the greater frequency of marital infidelity
 Darwin's Argument from Evil 63

among the male of the species or the fact that child abuse is more com-
mon in step-parents. Some of these explanations are a bit far-fetched,
but others are plausible despite their 'just-so' character. Of course, there
are also some moral facts, like facts about self-destructive behaviour, that
do not easily lend themselves to Darwinian explanations, but the the-
ory of natural selection doesn't entail that every characteristic possessed
by some members of a species (or even most members for that matter)
must make reproduction more likely.

6. The Darwinian No-Design Hypothesis versus

Darwinian Theism
The final and most important question that must be answered is why
these Darwinian explanations of good and evil work much better when
the theory of natural selection is combined with the no-design hypothe-
sis than when it is combined with theism. I will offer three reasons. The
first is that Darwin's theory is more likely on the no-design hypothesis
than on theism. The second is that Darwinian explanations of good and
evil are less complete when Darwin's theory is combined with theism
than when it is combined with the no-design hypothesis. And the third
is that, when combined with theism but not when combined with the
no-design hypothesis, Darwin's theory mystifies certain facts.
There is much more disagreement about how probable the theory of
natural selection is, either given the no-design hypothesis or given the-
ism, than there is about how probable evolution is. The evidence for
evolution is truly overwhelming. This is why almost all well-educated
theists are also evolutionists. But Darwin's theory is another matter
entirely. Few deny that natural selection accounts for some evolutionary
change, but, according to the theory of natural selection, natural selec-
tion operating on random genetic mutation is the principal mechanism
driving evolution. And many well-educated theistic evolutionists are
much less enthusiastic than their atheistic counterparts about this claim.
Not that there is any consensus among theistic evolutionists. Some
believe that Darwin's theory is highly probable on both the no-design
hypothesis and on theism; others believe it is very improbable on both
the no-design hypothesis and theism; and others think it more proba-
ble on theism than on the no-design hypothesis (because, they claim, it
could only work with divine assistance).
Still others agree with my position, which is that the theory of natural
selection is highly probable on the no-design hypothesis but not on the-
ism. This position entails that naturalistically inclined evolutionists who
64 Paul Draper

have great confidence in the truth of the theory of natural selection and
theistic evolutionists who have serious doubts about it are both right,
conditional on their respective metaphysical positions. This position is based
in part on my belief that the scientific case for the theory of natural
selection is impressive but far from complete. It is incomplete because
we don't know in any detail how most of the characteristics of living
things evolved by natural selection. Of course, if Darwin's theory is true,
we wouldn't expect to know all the details. And no one has shown that
any instance of adaptive complexity in present-day living things cannot
be accounted for by natural selection broadly understood. So these sorts
of gaps in our knowledge don't refute the theory of natural selection.
Nor do they prove that natural selection can work only with divine assis-
tance of some sort. But they do make the case for the theory incomplete.
If, however, one assumes that the no-design hypothesis is true, then
this incomplete case suffices to make the theory of natural selection
highly probable because there are no viable naturalistic alternatives to it.
Of course, one hears about scientific challenges to Darwinian evolution.
But the only plausible ones are either challenges to strict gradualism
or attempts to show that natural selection is not the only mechanism
driving evolution. The theory of natural selection, as I define it, does
not entail strict gradualism; so punctuated equilibriumism is compat-
ible with it. Nor does it entail that all evolutionary change is caused
by natural selection; so genetic drift, symbiosis and other mechanisms
may also be at work, especially when the change in question does not
involve a substantial increase in complexity. Thus, for the proponent
of the no-design hypothesis, the theory of natural selection as I broadly
define it really is the only game in town.
On theism, however, there are viable alternatives, and this means that
the gaps in our knowledge mentioned above make it unreasonable for
a theist to be confident in the truth of the theory of natural selection.
What are these alternatives? It's hard to know where to begin. For if
the universe was created by an omnipotent and omniscient being, then
so many logical possibilities are also real possibilities! For example, if
theism is true, then it is a real possibility that God has on billions of
occasions in the history of the development of life on earth miraculously
caused various mutations for the purpose of modifying a species in ways
that further God's goals. Another real possibility is that most traits that
arose by random variation have been artificially selected by God rather
than winning a fair competition in the struggle for life. Or perhaps God
never directly intervenes, but there is an as yet undiscovered hidden
mechanism that has caused most of the evolutionary change that has
 Darwin's Argument from Evil 65

taken place. This last scenario is, of course, also possible on the no-
design hypothesis, but on theism it is not just possible but plausible.
So one reason that Darwinian explanations of good and evil work bet-
ter for defenders of the no-design hypothesis than for theists is that
the theory of natural selection is much more probable on the no-design
hypothesis than on theism/
A second reason is that Darwinian explanations of facts about good
or evil will be less complete if theism is true than if the no-design
hypothesis is true. To see why, I will borrow a couple of definitions from
Richard Swinburne. Swinburne (2004) defines a full explanation as fol-
lows: 'An explanation of E by F is a full one if F includes both a cause, C,
and a reason, R, which together necessitated the occurrence of E' (p. 76).
A complete explanation is a special kind of full explanation. F is a com-
plete explanation of E if it is a full explanation and none of the factors
it cites can be explained by any contemporaneous factors (p. 78). The
key word here is 'contemporaneous'. A complete explanation need not
be an ultimate explanation. It need not be such that nothing prior to it
explains the factors it cites.
Darwinian explanations of good and evil are obviously not ultimate
explanations. And even if the no-design hypothesis is true, they are not
full since the factors they cite do not necessitate the facts they explain.
My claim, however, is that they are 'less complete' if theism is true, by
which I mean that they are not as close to being complete explanations.
This is so, because, if theism is true, then God is omnipotent and hence
the existence of the universe depends on God at any given time. God,
in other words, did not just create the universe, he sustains it. But that
means any complete explanation of facts about good and evil must, if
theism is true, include God's moral justification for allowing those facts
to obtain.
To put the point another way, Darwin's theory comes closer to solv-
ing the puzzle of good and evil faced by the proponent of the no-design
hypothesis than the puzzle of good and evil faced by the theist. If the
no-design hypothesis is true, then, the causes of good and evil are axio-
logically indifferent, and thus a naturalist will want to know why good
and evil are so abundant. Darwinian explanations, though not ultimate
explanations, enhance a naturalist worldview by providing intellectu-
ally satisfying explanations consonant with that worldview. Fill in some
more details and those explanations could be complete. The theory
of natural selection, however, does little to solve the theist's perplex-
ity about good and evil. For God wouldn't allow natural selection to
determine the pattern of good and evil observed in the biological world
66 Paul Draper

unless he had good moral reasons for doing so. And given the failure of
existing theodicies, we have no idea what those good reasons are. For the
theist, Darwinian explanations leave the biggest questions about good
and evil unanswered.
A third way of stating essentially the same point is that, whenever a
theist uses a Darwinian explanation of facts about good and evil, he or
she is in effect implicitly adopting a number of completely ad hoc and
very specific auxiliary hypotheses. If the no-design hypothesis is true,
no such auxiliary hypotheses are needed or presupposed. For example,
is there some greater good that, because of its logical connections to
suffering, requires that suffering be used to motivate animals to pur-
sue the biological goal of self-preservation? Does some moral end makes
it desirable for suffering to continue even when it serves no biologi-
cal purpose? Is there a moral justification for using natural selection to
produce a world in which most living things never or rarely flourish
because they compete with each other for survival? Or to produce a
species that possesses as little natural virtue and as much natural vice as
humans do? On the no-design hypothesis, Darwinian explanations of
good and evil work whether the answer to these and many similar ques-
tions is 'yes' or 'no'. But on theism, the answer must in every single case
be 'yes'. On theism, natural selection cannot drive evolution unless its
doing so coincides with the moral goals of an all-powerful, all-knowing
and morally perfect creator. And that's a really big coincideTlce that the
no-design hypothesis doesn't need.
A third reason that Darwinian explanations of good and evil work
better on the no-design hypothesis than on theism is that the theory of
natural selection, when combined with theism, mystifies certain facts.
Mystification is the opposite of explanation. A hypothesis 'mystifies' a
fact if it makes that fact more puzzling rather than less puzzling - if it
raises why-questions about that fact rather than answering them.
On Darwinian theism, harm to living things is not an unfortunate
side effect of a good creation, but rather is an integral part of the process,
started long before any human beings existed, by which God chose to
create various living things. Thus, if Darwin's theory is true, then harm
to living things is not, for example, the result of human beings or other
free agents rebelling against God and acting contrary to his purposes.
Rather, God has purposely used such harm as a means of achieving his
creative purposes. And this would require a very strong sort of moral jus-
tification. Thus, the theory of natural selection makes the evil in nature
even more shocking for theists than it would otherwise be.
Another problem that's exacerbated by adding Darwin's theory to the-
ism is the problem of animal suffering. Theists have for centuries tried
 Darwin's Argument from Evil 67

to avoid this problem by at least quasi-Cartesian means. While almost

no theists today deny, as Descartes did, the reality of animal suffering,
they still try to minimize its significance or justify the fact that their
theodicies ignore it by claiming that we know very little about what the
mental states of animals are like, that for all we know animal suffering
isn't much like human suffering and may even be of no moral signifi-
cance.8 If, however, Darwinian evolution is true, then we are the more or
less gradually modified descendants of (non-human) animals. This means
that, when humans and closely related animals behave very similarly in
very similar circumstances, it is likely that they share very similar mental
states. Thus, for example, when animals exhibit what overly cautious (or
overly Skinnerian) psychologists might describe as a 'fear-like response'
in circumstances that would cause fear in humans, it is likely that what
those animals feel is very similar to human fear. In general, then, it is
likely on Darwinian evolution that higher mammals suffer in many of
the same ways and just as intensely as we do. 9
So animal suffering is as real and as serious a problem for Darwinian
theists as human suffering. In fact, to make matters worse, Darwinian
theists are forced to admit that animal suffering is similar to human
suffering despite the fact that animals are not moral agents. This deals a
severe if not deadly blow to the otherwise plausible position that, if
God exists, then our being moral agents has something to do with why
God allows us to suffer to the extent we do. Thus, it is not just animal
suffering that is mystified by mixing the theory of natural selection with
theism. Human suffering becomes more puzzling, too.

7. Implications
We have, then, three very strong reasons for believing that the mar-
riage of Darwinian biology to theistic metaphysics is an unhappy one.
Although theism does not entail that Darwinian explanations of good
and evil are false, it does greatly weaken them. It weakens them because
natural selection is unlikely on theism, because ad hoc and specific aux-
iliary hypotheses are presupposed, and because other facts are mystified.
This would not be a problem for orthodox theism if powerful theodicies
were available, but no such theodicies exist. The few theodicies that do
offer decent explanations of some facts about evil fail to offset the force
of Darwin's argument from evil for two reasons. First, they adequately
account for relatively few facts about suffering or other evils. Second,
their ability to explain the few facts they do explain is purchased at the
price of making other facts even more mysterious. For example, if we
assume that God wants to use suffering to build moral character, then
68 Paul Draper

certain sorts of suffering and other conditions that are, quite predictably,
demoralizing become even more puzzling. Or, if it is free will that is so
important, then why are so many human beings robbed by a variety of
(natural) circumstances of any genuine freedom?
If, instead of focusing on a few isolated cases, one looks at the overall
pattern of evil in the world, being careful not to ignore the non-
human part of that world, then one cannot help but be struck by its
apparent moral randomness. The theory of natural selection, when com-
bined with the no-design hypothesis, offers a systematic explanation
of a broad range of facts about good and evil with which no theod-
icy I know of can even begin to compete. Therefore, since the theist
has no decent and systematic explanation of these facts - or at least
none that is likely to be true if theism is true - while the proponent of
the no-design hypothesis does, it follows that these facts are powerful
evidence supporting the no-design hypothesis over theism. Further, the
no-design hypothesis is both initially plausible and fits with much back-
ground knowledge better than theism. An evolutionary process that,
instead of being beneficently designed, is not designed at all is more
likely to take a very long time, to include imperfect adaptations, and to
involve adaptations that are harmful for individuals (even when they
are good for the genes those individuals host). And this is, in fact, what
we observe. Therefore, in the absence of some other evidence support-
ing orthodox theism over the no-design hypothesis, it follows that the
no-design hypothesis is much more probable than theism and hence,
other evidence held equal, theism is very probably false. 10

Notes
1. I use the term 'politically correct' here because of the broadly political moti-
vations that many have for holding that evolution is religiously neutral.
These motivations include the desire to prevent creationism or intelligent
design theory (ID) from being taught in science classes in public schools in
the USA. In his ruling against the constitutionality of teaching ID, Judge
John Jones (2005, p. 136) of the US District Court in Pennsylvania based his
ruling in part on an appeal to the view of 'scientific experts' who 'testified
that the theory of evolution ... in no way conflicts with, nor does it deny,
the existence of a divine creator'.
2. Darwin consistently denied being an atheist. For example, in a letter to one
J. Fordyee, dated 1879, he said, 'In my most extreme fluctuations I have
never been an Atheist in the sense of denying the existence of God' (1958b,
p. 59). There is good reason to believe, however, that 'God' is being used
here in a broad sense. It is telling, for example, that when a German stu-
dent inquired in the very same year about Darwin's religious views, a family
member wrote back that Darwin 'considers that the theory of Evolution is
 Darwin's Argument from Evil 69

quite compatible with the belief in a God; but that you must remember
that different persons have different definitions of what they mean by God'
(1958b, p. 61).
3. For example, Charles Hodge (1874, pp. 176-7) wrote in a book called What
is Darwinism?, 'We have thus arrived at the answer to our question. What is
Darwinism? It is atheism. This does not mean ... that Mr Darwin himself and
all who adopt his views are atheists; but it means that his theory is atheistic;
that the exclusion of design from nature ... is tantamount to atheism.'
4. Darwin (1958b, p. 249) himself appears to affirm this possible reconciliation
of his theory with divine design in one of his letters to Asa Gray.
5. Philip Kitcher (1982, p. 87) makes this point.
6. One of Darwin's favourite arguments for the conclusion that God does not
guide natural selection by directly causing certain variations is found (among
other places) at the end of his book on The Variation of Animals and Plants
under Domestication, as Francis Darwin notes in his Autobiography (1958a,
p. 73 n. 60). He says there that, since it is impossible to believe that God
intentionally provides variations for the sake of breeders, it is unreasonable
to believe that he intentionally provides the variations used in natural selec-
tion. Although Darwin claims that this argument has never been answered
(1958a, p. 73), it isn't difficult to provide such an answer. Given the less than
respectable goals of many breeders, I agree that it would be unreasonable to
believe that all of the variations used by breeders are intentionally provided
by a morally perfect God for the benefit of the breeders. But from the fact
that they were not intentionally provided for the benefit of the breeders, it
does not follow that they were not intentionally and specially provided, and
hence it provides no reason at all for thinking that the variations used in nat-
ural selection are not in at least some cases produced by God's miraculous
intervention.
7. It is worth noting that, if the theory of natural selection someday becomes
virtually certain on both the no-design hypothesis and theism, then the
truth of that theory would itself be evidence favouring the no-design
hypothesis over theism because natural selection would, for the reasons
mentioned above, still be antecedently more probable on the no-design
hypothesis than on theism.
8. See, for example, Murray, 2008.
9. For an excellent statement and defence of this sort of argument, see Rachels,
1991, ch. 4.
10. Sincere thanks to William Hasker, Yujin Nagasawa and Beth Seacord for help-
ful comments on earlier drafts. I am also grateful for generous support from
the Center for Philosophy of Religion at the University of Notre Dame, the
John Templeton Foundation and Purdue University.

References
Ayala, F. (2007) Darwin's Gift: To Science and Religion (Washington, DC: Joseph
Henry Press).
Butler,]. (1857) The Analogy of Religion, to the Constitution and Course of Nature
(Philadelphia: J B Lippincott & Co.).
70 Paul Draper

Darwin, C. (1958a) The AutobiographyofCharles Darwin, 1809-1882, ed. N. Barlow

(New York: W.W. Norton & Co.).
Darwin, C. (1958b) The Autobiography of Charles Darwin and Selected Letters, ed.
F. Darwin (New York: Dover Publications).
Hodge, C. (1874) What Is Darwinism? (New York: Scribner, Armstrong).
Hull, D. (1991) 'The God of the Galapagos', Nature 352 (8 August): 485-6.
jones, ]. (2005) 'Memorandum Opinion', Kitzmiller et al. v. Dover Area School
District et al. Available at www.pamd.uscourts.gov/kitzmiller/kitzmiller_
342.pdf.
Kitcher, P. (1982) Abusing Science: The Case against Creationism (Cambridge, MA:
The MIT Press).
Monod, ]. (1976) 'The Secret of Life', interview with Laurie John, Australian
Broadcasting Co. (10 June).
Murray, M. (2008) Nature Red in Tooth and Claw: Theism and the Problem ofAnimal
Suffering (New York: Oxford University Press).
Rachels, ]. (1991) Created from Animals: The Moral Implications of Darwinism
(New York: Oxford University Press).
Russell, B. (1997) Religion and Science (New York: Oxford University Press).
Swinburne, R. (2004) The Existence of God, 2nd edn (Oxford: Oxford University
Press).
4
Attributing Agency: Fast and Frugal
or All Things Considered?
Graham Wood

1. Introduction

We attribute agency to ourselves and other people. We also attribute

agency to other physical entities or systems, for example, our cat, com-
puter or car. We might also attribute agency more broadly: to a river,
an ecosystem or the universe as a whole. The attribution of agency in
this broader sense is central to religious belief, and is often character-
ized as the assertion of the existence of gods and spirits. For example,
Australian Indigenous peoples believe that spirits formed the landscape,
Lowland Maya people hold beliefs about forest spirits, and Christians,
Muslims and jews believe in the existence of the god of classical mono-
theism. All these peoples are attributing agency as a central feature of
their religion. This chapter explores this attribution of agency. Using
insights from contemporary cognitive science and philosophy of mind,
I offer a characterization of the cognitive process by which agency is
attributed to ourselves, and religious entities. Along the way I address
two questions. How do we attribute agency? And, why do we attribute
agency?
The characterization I offer adapts Dennett's intentional stance
(1987). I suggest that the intentional stance is a good way to understand
the attribution of agency to humans. However, I further suggest that
with one adaptation it can also be used to understand the attribution of
religious agency. I begin the chapter with a discussion of several inter-
related themes that inform the position I am advancing: evolutionary
psychology, the dual process theory of mind and the modularity theory
of mind. And, in the light of these themes, I consider the three stances
of Dennett: the physical, design and intentional stances. I then present
an adaptation of Dennett's intentional stance and use it to characterize

71
72 Grallam Wood

the attribution of agency in religious contexts, what I call 'taking the

spiritual stance'.
Now, for Dennett, if one is able accurately to predict the behaviour of
a particular physical system (for example, a human body), by attribut-
ing beliefs and desires to it, then that physical system has beliefs and
desires (i.e. intentional states) in what he calls a 'semi-realist' sense.
In other words, the test for whether a system has beliefs and desires is
whether one is able successfully to predict the behaviour of that system
by attributing beliefs and desires to it. Thus, the test is predictive suc-
cess. If a system passes this test, then the system has beliefs and desires,
in Dennett's semi-real sense. He calls these beliefs and desires 'real pat-
terns'. Using my adaptation of the intentional stance, I argue that the
beliefs and desires of religious entities can also be understood as 'real
patterns' in Dennett's semi-realist sense. The position I advance relies
on a conceptual relationship between the attribution of agency and the
attribution of beliefs and desires to a physical system as follows. If beliefs
and desires are attributed to a system, then agency is attributed to that
system.
Now for the record, Dennett's intentional stance and the concep-
tualization of beliefs and desires as real patterns is not the theory of
mind standardly endorsed by contemporary cognitive science. Dennett's
theory of mind can be understood as instrumentalist. Instrumentalist
theories endorse concepts (such as beliefs and desires) because they are
useful in predicting events (such as the behaviour of humans). Notice
how beliefs and desires get endorsed as 'real patterns' on Dennett's
account because they are predictively successful.
My characterization of the attribution of beliefs and desires in the
religious context is also instrumentalist. But it relies on a broader under-
standing of instrumentalism, as it is understood within the tradition
of classical philosophical pragmatism Oames, 1975). And in particular
I rely on this broader notion of instrumentalism to justify my adap-
tation of Dennett's intentional stance. My account of the attribution
of agency within religion is a pragmatist account. Philosophical prag-
matists, among other things, take concepts to function as instruments
(Hookway, 2010). To understand this approach to concepts, consider
William james on the concept of truth: 'The true is the name of what-
ever proves itself to be good in the way of belief, and good, too, for
definite assignable reasons' (1975, p. 42). '"The true", to put it very briefly,
is only the expedient in the way of our thinking [... ] Expedient in almost
any fashion; and expedient in the long run and on the whole, of course'
(1975, p. 106; emphasis original). Notice that on both characterizations
 Attributing Agency: Fast and Frugal or All Things Considered? 73

of the true above, James includes conditions. On one account, the true
must be good for 'definite assignable reasons.' And on the other, the
true must be 'expedient in the long run and on the whole.' Here there
is a parallel with the instrumentalism of Dennett's intentional stance.
Dennett's instrumentalism can be understood as a version of james's
pragmatism by simply stipulating that predictive success is the 'assignable
reason' or the relevant 'expediency'. However, this prompts the ques-
tion: Why should predictive success be the relevant expediency? In my
adaptation I change the 'expediency', or the 'definite assignable reason',
to reproductive success over evolutionary time.
In Dennett's original intentional stance, the test is predictive success.
If a system passes this test, then the system has beliefs and desires, in
Dennett's semi-real sense. My adaptation changes the test from 'predic-
tive success' to 'reproductive success over evolutionary time'. Using the
new test, if the attribution of beliefs and desires to a system leads to
the reproductive success (of the lineage) of the organism making the
attribution (over evolutionary time), then the system has beliefs and
desires. This adaptation is justified as follows. From an evolutionary per-
spective the 'function' of cognitive systems is not, first and foremost,
accurate prediction, but rather it is to keep the organism reproducing.
Accurate prediction has a role in reproductive success, but it is not the
foundational role. I suggest that the attribution of agency within reli-
gious belief has led to reproductive success over evolutionary time for
the lineage Homo sapiens. And thus my account helps explain why we
believe in gods and spirits. But, as I noted above, this chapter also
addresses the question of how we attribute beliefs and desires. Thus,
I will also explore the cognitive origins of Dennett's 'semi realist' beliefs
and desires.
Dennett describes beliefs and desires as real patterns. But for Dennett
the perception of these patterns is fundamentally dependent upon the
organism doing the attributing. I take this to imply that the attribut-
ing organism itself to some extent constructs the beliefs and desires
attributed to other systems. And taking this a little further, I suggest
that the concept of a 'belief' or 'desire' is an heuristic concept produced
by a module within the mind of the organism doing the attributing.
And once beliefs and desires are attributed to a system, a natural way
to think of this system is as an agent, thus agency is also attributed
to the system. Borrowing the distinction from the heuristics literature
between 'fast and frugal' thought and 'all things considered' thought
(Gigerenzer et al. 1999), I suggest, the concept of agency is not an
'all things considered' concept, but rather a 'fast and frugal' heuristic
74 Graham Wood

concept. Thinking in terms of agency is a way of thinking that increases

the reproductive success of organisms that employ the method, and so
is a way of thinking that has been retained over the evolutionary time.
The chapter ends with some speculation concerning how the attri-
bution of agency in religious contexts has furthered the reproductive
success of our species over evolutionary time. And so, now I begin the
substantive content of the chapter by discussing the assumptions I make
and the perspective I take throughout.

2. Background Themes and Assumptions

Evolutionary psychologists assume that the mind is an evolved system,
and use this to guide their choice of theories about the structure and
functioning of the mind. 'Evolutionary analysis' (Samuels, 2000, p. 24)
imposes constraints on the type of theory it is reasonable to accept. If a
theory proposes a structure and function of the mind that evolution
is very unlikely to have produced, then that counts against the the-
ory. Alternatively, if a theory proposes a structure and function of the
mind that evolution is very likely to have produced, then that counts
in favour of the theory. So this leads to the question: What structures
and functions is evolution likely to have produced? This question can be
answered by looking at other evolved structures. The structure and func-
tion of other organs in the body all have an important similarity. They
are specific solutions to particular survival problems. For example, the heart
pumps blood, it does not oxygenate or filter the blood, and the kidneys
filter the blood, they do not pump or oxygenate it. The lungs oxygenate
the blood, they do not pump or filter it. So evolution produces specific
solutions to specific survival challenges, rather than general solutions to
general survival challenges. This evolutionary analysis leads inductively
to the conclusion that cognitive systems are also specific solutions to
specific problems faced by our hominid ancestors (Samuels et al., 1999,
pp. 85-6).
The lesson evolutionary psychologists take from this analysis is that
it is wrong to think of the mind as one general information process-
ing system. Rather, the mind should be thought of as a set of distinct
information processing systems (Carruthers, 2006, pp. 12-28, cf. Simon,
1962). Now, evolutionary psychology conflicts with orthodox psychol-
ogy because the latter assumes that a set of general cognitive capacities,
including habituation, operant and classical conditioning, imitation
and the rules of logic and probabilistic reasoning, are responsible for
most of our cognitive abilities. However, the perspective of evolutionary
 Attributing Agency: Fast and Fmgal or All Things Considered? 75

psychology is gaining ground among both psychologists and biologists,

as illustrated by the observation that psychologists 'should be trying
to identify and understand these specialized circuits rather than pre-
tending that human behaviour can be derived from a few law-like
mechanistic principles' (Wilson, 2002, p. 26). If we assume that the
mind is the result of evolution, then, it is argued, we should think about
the mind differently. For example, consider the dual process hypothesis
and the modularity hypothesis.
I assume that the mind has (at least) two distinct modes of thought.
This has been called the dual process theory or two systems hypothesis. I will
not argue for this, but rather simply assume it in the discussion that
follows. Stanovich and West (2000) called these two systems System 1
and System 2. They are characterized as follows:

System 1 (intuition): fast; automatic; undemanding of cognitive capac-

ity; acquired by biology, exposure, and personal experience.
System 2 (reasoning): slow; controlled; demanding of cognitive
capacity; acquired by cultural and formal tuition. (Kahneman, 2002)

Kahneman (2002) characterizes System 1 as generating intuitive judge-

ments that 'occupy a position - perhaps corresponding to evolutionary
history- between the automatic operations of perception and the delib-
erate operations of reasoning' - leaving the deliberate reasoning to
System 2.
Another way to conceptualize the distinction drawn between System
1 and 2 is to use the notion of modularity. Fodor (1983) first proposed
the notion of mental modules. Fodorian modules (part of System 1) are
said to receive information coming into the mind, via, for example, the
sense organs, and manipulate this information before it is then made
available to consciousness (System 2). A standard example of a Fodorian
module at work is the process that leads to the Miiller-Lyer illusion.
In this illusion, involving two lines of equal length with chevrons
pointing in different directions on each of the two lines, the conscious

>.,..__..~~<

~ )
Figure 4.1 The Miiller-Lyer Illusion
76 Graham Wood

mind cannot help but see the lines as unequal (d. Segall et al., 1966).
Given that the lines are in fact the same length, presumably the infor-
mation received by the retina corresponds to lines of equal length. It is
assumed that this illusion is the result of a manipulation of the infor-
mation stream between the retina and the conscious mind such that
the lines of equal length are represented to consciousness as lines of
unequal length. And it is assumed that a module is responsible for this
manipulation.
For the purposes of this chapter it is important to note two features
of Fodorian modules: first, the outputs they produce are mandatory,
and second, the manipulations made inside modules are relatively
inaccessible to System 2. Fodor describes these as follows:

It is worth distinguishing the claim that input operations are manda-

tory (you can't but hear an utterance of a sentence as an utterance
of a sentence) from the claim that what might be called 'interlevels'
of input representation are, typically, relatively inaccessible to con-
sciousness. Not only must you hear an utterance of a sentence as
such, but, to a first approximation, you can hear it only that way.
(Fodor, 1983, p. SS)

In the case of the Muller-Lyer illusion, we must see the lines as having
different lengths, and we cannot see them any other way. If we were able
to see them as both the same length and different lengths at the same
time then our visual image would be incoherent.
While Fodor believed that modules existed only on the periphery
of the mind, others have suggested that the mind is massively modu-
lar. For example, Cosmides and Tooby hold that many (non-peripheral)
cognitive processes are modular:

our cognitive architecture resembles a confederation of hundreds or

thousands of functionally dedicated computers (often called mod-
ules) designed to solve adaptive problems endemic to our hunter-
gatherer ancestors. Each of these devices has its own agenda and
imposes its own exotic organization on different fragments of the
world. There are specialized systems for grammar induction, for face
recognition, for dead reckoning, for construing objects and for rec-
ognizing emotions from the face. There are mechanisms to detect
animacy, eye direction, and cheating. There is a 'theory of mind'
module ... a variety of social inference modules ... and a multitude
of other elegant machines. (Cosmides and Tooby, 1995, pp. xiii-xiv)
 Attributing Agency: Fast and Frugal or All Things Considered? 77

The modules that are said to exist in the massive modularity hypothesis
are not all Fodorian modules. For example, some modules may be less
mandatory, and central cognition may have more access to the inter-
nal workings of those modules. But the concept of a modular system
that is to a significant extent independent from tile operation of central cogni-
tion will be central to this chapter. And I assume that while System 2 is
concerned with the 'deliberate operations of reason', System 1 processes
are largely independent processes undertaken by modules in the mind.
Furthermore, I suggest that, to a large extent, System 1 actually provides
System 2 with the concepts with which it thinks. In other words, while Sys-
tem 2 might be responsible for putting all the pieces of the jigsaw (that
is, conscious thought) together, it is System 1 that provides System 2
with the pieces of the jigsaw. To illustrate this with a famous example,
System 1 may provide System 2 with certain pieces of the jigsaw, for
example, 'I' and 'am', and then System 2 puts them together: 'I think
therefore I am.'
Two relevant modules are the 'folk physics' module (Spelke, 1990) and
the 'folk psychology', or 'theory of mind' module (Baron-Cohen, 1995).
However, while in the original presentation each of these was taken to
be a single module, I think of these as a set of modules (cf. Gerrans,
2002). So I assume a set of modules generates the contents of both 'folk
physics' and 'folk psychology'. On the modular account of 'folk physics'
there is a dedicated part of the mind (a module, or set of modules) that
processes information about physical objects or systems. This module or
set of modules allows us quickly and accurately to predict the behaviour
of physical objects, so we can successfully interact with those objects.
On the modular account of 'folk psychology' there is a dedicated part of
the mind (again, a module, or set of modules) that processes informa-
tion about objects or systems that are taken to have intentional states,
roughly speaking, objects or systems that are taken to have beliefs or
desires, or agency (for example, humans). This module or set of modules
allows us efficiently to interact with these intentional systems.

3. System 1 Modules Generate Heuristic Concepts

Earlier I suggested that 'beliefs' and 'desires' can be understood as
'heuristic concepts'. I will now explore this idea further. System 1 pro-
duces thought that is fast, automatic and undemanding of cognitive
capacity. This characterization has much in common with the notion
of heuristic thought (Gigerenzer et al., 1999; Gilovich et al., 2002).
An heuristic is a 'rule or solution adopted to reduce the complexity of
78 Graham Wood

computational tasks, thereby reducing demands on resources, such as

time, memory, and attention' (Richardson, 1999, p. 379).
An example is the recognition heuristic. If you are not a German and
you are asked the following question, 'Which of Dortmund or Munich
is the larger city?', you may use the fact that you recognize the name
of one city and not the other as a fast and frugal way of deciding that
Munich was the larger city (Gigerenzer et al., 1999, p. 41). This works
because there is a correlation between the population of a city (the target
attribute) and the likelihood that you recognize the name of the city (the
heuristic attribute). The fact that one can consciously reflect on the way
this heuristic works shows that it is not modular, in the sense that the
process is inaccessible to conscious awareness, thus it is employed by
System 2 (deliberative reasoning).
But, I suggest, System 1 employs similar heuristic techniques (within
modules). And if the modular process is mandatory and the internal
workings are inaccessible to System 2, then the conscious awareness of
such an heuristic process would be very different to our experience of
using the recognition heuristic, say. It might seem more like our experi-
ence of the Miiller-Lyer illusion. In the case of the Miiller-Lyer illusion
it is assumed the module uses the opposite directions of the chevrons
as visual cues to indicate the relative distance of the two lines from the
viewer. The module (erroneously) interprets one line as closer than the
other and thus manipulates the lengths of the lines before presenting
the image of the lines to consciousness. So, the fact that the lines appear
as having different lengths is itself an heuristic 'perception' that System
1 has provided to System 2 to allow System 2 quickly to make choices
about the relative distance of objects.
And if we take massive modularity seriously then we should expect
to find other modules manipulating information. Furthermore if these
other modules are mandatory and the processes occurring within the
module are inaccessible to System 2, then again, just as is the case with
the Miiller-Lyer illusion, we need to be more careful about accepting
the veridical nature of the outputs of these modules. Take for exam-
ple the 'folk physics' module (or set of modules), which is said to
provide consciousness with information about how mid-sized physical
objects move.
As we learn more about the process by which visual information is
manipulated by the mind we find that there are specific parts (modules?)
of the visual cognition system that are responsible for edge perception,
and other parts (yet other modules?) that take perceptions of edges as
 Attributing Agency: Fast and Frugal or All Things Considered? 79

inputs and then manipulate this information to generate the perception

of a solid physical object. Thus, perhaps modules in the mind generate
the perception of a solid physical object. This, of course, leads to the
question of whether solid physical objects exist in the world indepen-
dently of the perception and cognition of such objects. Now I will not
explore that question in any detail here. The only point I wish to make
is that the module (or set of modules) responsible for our cognition of
folk physics produces the perception of mid-sized physical objects, and
hence the module or modules may also be the origin of the very concept of a
physical object itself. And if we accept current physics, then mid-sized
physical objects are mostly empty space. So it is reasonable to ask: Are
mid-sized physical objects really 'out there' at all?
If the suggestion that the mind creates the concept of a solid physical
object seems rather surprising, consider the perception and thence the
concept of hot and cold. The perceptions of hot and cold are very famil-
iar to us, but 'hot' and 'cold' don't exist 'out there' in the world. All that
exists 'out there' are the vibrations of molecules. I suggest that the vibra-
tion of molecules has only an arbitrary connection to the perceptions
of hot and cold. But the contingencies of the evolution of conscious-
ness were such that the perception of hot and cold now represents the
vibration of molecules in conscious awareness.
Now, given the arbitrary relation between molecular vibration and
the sensation of hot and cold, we might expect other arbitrary relations
to exist between our consciousness of the world and what is really 'out
there' independent from our experience. However, once a representa-
tion, like the experience of hot or cold, is in place, it seems to us, as
conscious entities, that things really are hot or cold 'out there'. Similarly,
things like solid physical objects appear to be 'out there' but perhaps the
concept of a solid physical object is just an heuristic concept provided by
a module that makes the flux of experience comprehensible. Consider a
similar point made by Quine:

Physical objects are conceptually imported into the situation as con-

venient intermediaries - not by definition in terms of experience,
but simply as irreducible posits comparable, epistemologically, to the
gods of Homer .... Both sorts of entities enter our conception only
as cultural posits. The myth of physical objects is epistemologically
superior to most in that it has proved more efficacious than other
myths as a device for working a manageable structure into the flux of
experience. (Quine, 1963, p. 44)
80 Graham Wood

And if this is the case for the concept of a physical object, it may also
be the case for the concept of an agent. If a folk psychology module (or
set of modules) does exist, then I suggest that it may produce the very
concept of 'an agent with beliefs and desires' in a similar way to the way
in which a visual module produces the Miiller-Lyer illusion. To unpack
this suggestion a little more, one module might produce the concept
of a belief and attribute it to a physical system. Then another module
might take the concept of a physical system having a belief as an input
together with other inputs (e.g. the concept of a physical system hav-
ing a desire) and output the concept of an agent with that belief-desire
pair. Further, consider the mandatoriness and relative inaccessibility of
the workings of a module. I suggest that thinking in terms of an agent
with beliefs and desires may be mandatory in the same way that seeing
the lines of different lengths is mandatory in the Miiller-Lyer illusion.
Perhaps we cannot help but think in terms of an agent with beliefs and
desires.
Some might say that agents with beliefs and desires are clearly 'out
there' in the world. But if the concept of 'an agent with beliefs and
desires' is the output of a module or set of modules, and furthermore, if
the output of that module, or set, is mandatory and the internal work-
ings of the module are inaccessible to conscious awareness, then the
perception that an agent is 'really out there' would be indistinguishable
from the output of such a module or set of modules. Thus, I suggest,
we must seriously consider the possibility that there are no agents 'out
there' in the world, but rather, the concept of 'an agent with beliefs and
desires' is produced within a module or set of modules in the mind.
Consider another analogy to illustrate the point: the concept of
'the average family' used by social policy makers to make real world
decisions. No one expects to find the average family 'out there' in the
world. The average family does not exist in the same way that any par-
ticular family exists. I suggest that the concept of 'the average family'
is similar to the concept of 'an agent with beliefs and desires'. just as
policy-makers can use the concept of the average family to make real
world decisions, a human can use the concept of 'an agent with beliefs
and desires' to make real world decisions (either about themselves or
others). The social policy-maker does not expect to find the average fam-
ily 'out there' in the world, and perhaps the human individual should
not expect to find 'agents with beliefs and desires' out there in the world
either.
If this is correct, then concepts of agents with beliefs and desires can
be thought of as heuristic concepts. Perhaps the very idea of an agent
 Attributing Agency: Fast and Frugal or All Things Considered? 81

with beliefs and desires is a fast and frugal way of making sense of a
large and complex set of data, rather than an entity that is 'out there'
in the world. Thus, I suggest, concepts that we take for granted (such as
the existence of agents) are in fact heuristic concepts that modules in
our mind have created to make it easier to process the large information
stream that we must process to make our way in the world.
Now I will consider this suggestion in the context of Dennett's work
on the physical, design and intentional stances.

4. Dennett's Three Stances

Sellars drew a distinction between two rival images, the manifest and
scientific, characterizing the manifest image as 'an "inadequate" but
pragmatically useful likeness of a reality which first finds its adequate
(in principle) likeness in the scientific image' (1964, p. 57). For Sellars,
science is in the process of constructing the scientific image. But the
image of the world created by these scientific theories is very different
from the common sense or 'folk' image of the world. The common sense
or 'folk' image of the world is Sellars's manifest image.
I suggest the distinction between the manifest and scientific images
of the world corresponds to the distinction between the images of the
world generated by System 1 and System 2 (where System 1 is under-
stood as the modular mind). And I take Dennett to hold a similar
position:

It is no accident that we have the manifest image that we do; our ner-
vous systems were designed to make the distinctions we need swiftly
and reliably, to bring under single sensory rubrics the relevant com-
mon features of our environment, and to ignore what we can usually
get away with ignoring .... The undeniable fact is that usually, espe-
cially in the dealings that are most important in our daily lives, folk
physics works. Thanks to folk physics we stay warm and well fed and
avoid collisions, and thanks to folk psychology we cooperate on mul-
tiperson projects, learn from each other, and enjoy periods of local
peace. (1987, p. 11)

Both Sellars and Dennett appear to endorse the idea that our mind struc-
tures our experience of the world. But I take Dennett's position to be
particularly amenable to a modular account. (I think of his 'sensory
rubrics' as heuristic outputs of modules.) So to Dennett I now turn, and
consider his three stances: the physical, design and intentional stances.
82 Graham Wood

I examine Dennett's three stances in two stages. The first stage intro-
duces them as methods by which our limited cognitive systems process
the large amount of information we need to process in order to survive.
Limited cognitive systems need to use cognitive resources frugally. Thus
I characterize Dennett's three stances as fast and frugal heuristics that
are produced by modules. This first stage of the examination makes no
claims about the existence or non-existence of the entities that these
heuristics use as 'content'. The entities may or may not exist, and this is
importantly independent of the practical application of these heuristics.
We can productively use these heuristics independently from whether
the entities referred to by them actually exist (in the very same way
that we can use the concept of the 'average family' independently from
whether the average family is 'out there' in the world). The second stage
addresses the ontological status of the content of the intentional stance
(Dennett's 'real patterns') and this will be relevant to the attribution of
agency in religious belief.
So I begin with the purely pragmatic use of the three stances as meth-
ods of getting by in the world. We are physical organisms and as such we
must expend energy to engage in cognition, but we have limited energy
so we must be frugal. By contrast, we can conceive of a conscious entity
that has unlimited cognitive resources, personified by Laplace's 'demon'.
Laplace's demon would see the scientific image of the world. Assum-
ing (incorrectly but for the sake of illustration) a deterministic universe,
Laplace's demon could use the scientific image to calculate the future of
the universe in all its detail, including the behaviour of humans. But we
are not Laplacian demons, so we must employ some cognitive shortcuts,
some 'fast and frugal' heuristics. Dennett's three stances are examples of
these heuristics.
These stances are practical ways of calculating the behaviour of a
range of physical systems, from simple to complex. The stances are
practical because without them some physical systems would be too
complex for humans to predict. The physical stance is the most basic.
This stance assumes all objects of experience are simply physical objects.
I take the physical stance to be the output of the so-called 'folk physics'
module (or set of modules).
All objects can be considered from the physical stance, however if
one's cognitive capacities are limited, it may not be possible for one
to understand the behaviour of a complex physical system using the
physical stance alone. Dennett illustrates this by contrasting the limited
cognitive capacities of a normal human with an omniscient physi-
cist (Laplace's demon) who could predict the behaviour of any physics
 Attributing Agency: Fast and Frugal or All Tllings Considered? 83

system 'assuming it to be ultimately governed by the laws of physics'

(Dennett, 1987, p. 23). 1
If a physical system is too complex to predict using the physical
stance, we can move to the design stance 'where one ignores the actual
(possibly messy) details of the physical constitution of an object, and, on
the assumption that it has a certain design, predicts that it will behave
as it is designed to behave' (Dennett, 1987, pp. 16-17). And if the phys-
ical system is yet more complex still, even the design stance will not
work. Then Dennett moves to the intentional stance. Here is Dennett
on taking the intentional stance:

first you decide to treat the object whose behaviour is to be predicted

as a rational agent; then you figure out what beliefs that agent ought
to have, given its place in the world and its purpose. Then you figure
out what desires it ought to have, on the same considerations, and
finally you predict that this rational agent will act to further its goals
in the light of its beliefs. (Dennett, 1987, p. 17)

Thus we can see how the three stances can be employed pragmatically
to help us, as limited cognitive systems to survive in a complex world. 2
But, setting aside the utility of these stances, the question remains
concerning the reality, or otherwise, of the contents of these stances.
For example, the intentional stance attributes beliefs and desires to cer-
tain physical systems. But do these systems really have these beliefs and
desires? Or are beliefs and desires simply heuristic concepts employed
by the modules that generate these stances (in the very same way that a
statistician might use the concept of the average family without expect-
ing to find such a family out there in the world)? It is to these questions
that I now turn.

S. The Ontological Status of Real Patterns

For Dennett a system 'really' has beliefs and desires if you can suc-
cessfully take the intentional stance toward that system. But what does
'successfully' mean here? For Dennett, if you are successful in predicting
the behaviour of the object (or system) then that object (or system) has
beliefs and desires:

Any system whose behaviour is well predicted by this strategy is

in the fullest sense of the word a believer. What it is to be a true
believer is to be an intentional system, a system whose behaviour
84 Graham Wood

is reliably and voluminously predictable via the intentional strategy.

(1987, p. 15)

So it is the successful predictions generated by taking the intentional

stance toward a system that justifies the system being called an inten-
tional system (a system with beliefs and desires).
As we saw above, Laplace's demon would not need to adopt the
intentional stance, simply because it could predict the behaviour of
the system from the laws of physics and the initial conditions of the
system. But Dennett stresses that if Laplace's demon (now personified
as a Martian) failed to take the intentional stance towards the system
it was predicting it would be failing to see a 'real pattern' that existed
'objectively' in the world:

Our imagined Martians might be able to predict the future of the

human race by Laplacean methods, but if they did not also see us
as intentional systems, they would be missing something perfectly
objective: the patterns in human behaviour that are describable from
the intentional stance, and only from that stance, and that support
generalizations and predictions. (1987, p. 25)
Where there are intelligent beings, the patterns must be there to be
described, whether or not we care to see them. It is important to
recognize the objective reality of the intentional patterns discernible
in the activities of intelligent creatures ... (p. 28)
I claim that the intentional stance provides a vantage point for dis-
cerning ... useful patterns. These patterns are objective- they are there
to be detected - but from our point of view they are not out there
entirely independent of us, since they are patterns composed partly
of our own 'subjective' reactions to what is out there ... It is easy for
us, constituted as we are, to perceive the patterns that are visible from
the intentional stance- and only from that stance. (p. 39)

Dennett concedes that these patterns are not 'out there' entirely inde-
pendent of us. So what is the nature of this objectivity? Dennett
describes these 'real patterns' as partly composed of our own 'sub-
jective' reactions to what is out there. This I think is an interesting
approach. Dennett is claiming that something about the nature of the
human condition (our ability to adopt the intentional stance) is central
to the reality of these patterns. Now I want to push Dennett's position
slightly to see where we can take it. In particular, I want to explore the
 Attributing Agency: Fast and Frugal or All T11ings Considered? 85

possibility of granting this semi-real status to spirits associated with nat-

ural systems that are common in religious belief. I am claiming that
something about the nature of the human condition (our ability to
adopt the spiritual stance) is central to the reality of gods and spirits.
And here we move toward the second of the questions at the heart of
this chapter: Why do we attribute religious agency?

6. Changing the Measure of Success

Dennett justifies the attribution of beliefs and desires to a system if we
can accurately predict the behaviour of the system using those attribu-
tions. But why is predictive success the appropriate justification? The
ability to predict successfully their environment is important to organ-
isms making their way in the world. But it is certainly not the only
important ability. The most important ability is reproductive success.
Indeed, it could be argued that the ability to predict successfully systems
in their environment is only important if it furthers the reproductive
success of the organism.
So, moving from Dennett's instrumentalism to my pragmatism,
I change the measure of success from predictive success to reproduc-
tive success over evolutionary time. Now with this new measure of
success in place, a system has beliefs and desires (still in Dennett's semi-
realist sense) if attributing beliefs and desires to that system leads to
reproductive success for the attributing population of organisms over
evolutionary time.
Now critics (perhaps Dennett included) might say this has little to
do with Dennett's original intentional stance. Indeed, Dennett warns of
radical interpretationalism (1987, p. 24) of which this may be a species.
But in response to such concerns I simply point out that, from an evolu-
tionary point of view, reproductive success is all that matters. Predictive
success may lead to reproductive success, but at the end of the day it
is reproductive success that determines the long-term existence of any
lineage of organisms. And, as is pointed out by Wilson and Lynn, this
has implications for what we take to be real:

Dozens of evolutionists have observed that insofar as beliefs are prod-

ucts of natural selection, either proximally or distally, then they
should be designed to enhance fitness, not to perceive the world as it
really is .... To begin with genetic evolution ... deception begins with
perception. All organisms perceive only the environmental stimuli
that matter to their fitness. Our species can see only a narrow slice
86 Graham Wood

of the sound and light spectrum, cannot sense electrical and gravita-
tional fields at all, and so on. We also distort what we can perceive,
for example, by turning the continuous light spectrum into discrete
colors. Perception might not qualify as belief, but if the former is so
prone to adaptive distortions, it would be surprising if the latter was
not prone as well. (Wilson and Lynn, 2009, p. 539)

Of course, Dennett may wish to distinguish 'real patterns' that are

realized by predictive success from 'real patterns' that are realized by
reproductive success, and it is reasonable to do so. So, to distinguish
this stance (with its different measure of success) I call it the spiritual
stance.
Importantly the only thing I have changed in comparison to the
intentional stance is the measure of success. Everything else remains
as Dennett characterized it. So the spiritual stance attributes beliefs and
desires to systems and if that attribution leads to reproductive success
over evolutionary time, then the system in question has beliefs and
desires in the semi-realist sense that Dennett describes. So a successful
application of the spiritual stance leads to the existence of real spiritual
patterns.

7. Why Would the Spiritual Stance Lead to

Reproductive Success?
The spiritual stance is a way of understanding the attribution of agency
in religious contexts. It is also a way of understanding the (semi-real)
existence of gods and spirits as real patterns. But all this presupposes that
the attribution of religious agency would lead to reproductive success
over evolutionary time. So, let me now turn to the question, Would the
spiritual stance lead to reproductive success?
A position gaining acceptance within cognitive science of religion
makes the following claims: {1) religious belief supports cooperative
behaviour in groups Oohnson anu Bering, 2009); and (2) cooperative
groups outcompete non-cooperative groups (via group selection); and
thus (3) religious belief was selected over evolutionary time (Wilson,
2002). So, it is claimed, religious belief has led to reproductive success.
This position is reasonably well established in the cognitive science of
religion literature, so I will not examine it in any detail here, but see
Wood (2011) for a review. However, I will say a few words about how
the spiritual stance fits into this account of religious belief.
The position outlined above focuses on belief in the existence of pun-
ishing gods and their role of in the facilitation of in-group cooperation.
 Attributing Agency: Fast and Frugal or All Things Considered? 87

At first sight there does not seem to be much in common with that posi-
tion and the spiritual stance, in which beliefs and desires (and agency)
are attributed to physical systems. However, with only a small change in
the content of the relevant belief system one can get from an account
involving a physical system to an account involving punishing gods.
For example, if the physical system toward which the spiritual stance is
taken is, say, the local physical environment (or alternatively, the uni-
verse as a whole), and the beliefs and desires attributed to it relate to
a belief about an individual cheating in a group task, and the desire to
punish the cheat, then it is only a small change in the relevant belief
system to get from the local physical environment to a local god, as
might be relevant to an ancient hunter-gather culture (or alternatively,
from the universe as a whole to a god 'outside' the universe, as might be
relevant to classical monotheism).
However, there are many religious beliefs that do not involve gods and
spirits that desire to punish, so now I suggest how the spiritual stance
might lead to reproductive success without relying on the notion of gods
that punish. The suggestion involves a particular concept of person-
hood, namely the concept of an agent with moral standing. And I suggest
that attributing personhood (in this particular sense) to certain natural
systems may lead to reproductive success over evolutionary time.
To clarify this concept of personhood consider two distinctions. The
first is between humans and agents. I take humans to be defined bio-
logically with reference to the fact that they are members of the species
Homo sapiens. I take an agent to be defined intentionally with reference
to the fact that the intentional or spiritual stance is successfully taken
toward them. Thus a human is not essentially an agent, and an agent is
not necessarily a human.
The second distinction is between agents and persons. Again while
agents are understood intentionally, persons are understood morally.
I take personhood essentially to involve moral standing being bestowed
upon an agent (human or otherwise). Thus an agent is not essentially
a person. Rather personhood is bestowed upon an agent (human or
otherwise), either by the agent themselves or other agents. Alterna-
tively, agents may choose not to bestow personhood on themselves
or other agents (or other systems). And importantly, personhood can
be bestowed on any physical system (assuming agency has also been
attributed to it).
So far, I have kept the concept of attributing agency and attribut-
ing personhood distinct. It is conceptually possible to attribute agency
without attributing personhood. But I suggest that there is a deep con-
nection between attributing agency and attributing moral standing to
88 Graham Wood

that system. I suggest it is natural to attribute moral standing to a system

when agency is attributed to that system.
I think that attributing personhood to a system is one of the most
basic cognitive attitudes that we can take towards a system with which
we are interacting. I also suggest that the process of attributing person-
hood to a system is closely related to the modular process described
above whereby a module attributes beliefs and desires to a physical
system. Just as System 1 may be responsible for attributing agency
to a physical system, it may also be responsible for attributing moral
standing to that system, and then representing that system as a person
(an agent with moral standing) to System 2.
Now consider the following question: Would the attribution of per-
sonhood to a particular natural system lead to the reproductive success
of the linage of humans that made that attribution? In certain cir-
cumstances, I suggest that it would, and here is why. If one attributes
personhood to a system then one may act differently toward that
system. For example, consider a population of humans that grant per-
sonhood (and hence moral standing) to the rivers or forests with which
they interact. This may change the way they interact with those systems.
They might, for example, refrain from using all the resources available
within that system. By doing so they may interact sustainably with the
system and thus these organisms may survive over evolutionary time.
Furthermore, imagine a number of different populations of organism
that attributed personhood to different systems. If a particular attribu-
tion of personhood to a system led to sustainable resource use, then the
population that made that attribution would persist over evolutionary
time, while a different population that did not make that attribution
may deplete their resources (even if they flourished for a short period
while they depleted them) and thus they may tend toward extinction.
Over evolutionary time populations that have a propensity to make
attributions of personhood that lead to reproductive success may dis-
place populations that have no such propensity. And thus populations
that afford moral status to physical systems upon which their existence
depends may flourish in the long term.

8. Conclusion
I have offered a characterization of the attribution of religious agency.
I have drawn from the insights of evolutionary psychology and philos-
ophy of mind. Evolutionary psychology suggests that rather than think
of the mind as a single cognitive system, we should think of the mind
 Attributing Agency: Fast and Fmgal or All Things Considered? 89

as many distinct systems that evolved in response to distinct survival

challenges. These distinct systems can be conceptualized in a number of
ways. One way is the dual process theory of mind where System 1 gen-
erates fast and frugal intuitions and System 2 generates all things con-
sidered judgements. Another way involves the existence of modules in
the mind. Furthermore, the operation of such modules may be manda-
tory and conscious thought may have limited access to the modular
processes. I characterized System 1 as generating jigsaw pieces that were
given to System 2. But System 2 does not necessarily understand that
System 1 itself constructed the jigsaw pieces. System 2 (the conscious
mind) then puts those pieces together to make sense of its world.
In addressing the question 'How do we attribute agency?' I introduced
Dennett's intentional stance- the attribution of beliefs and desires (and
hence agency) to physical systems - and suggested that this stance was
produced by a set of modules in the mind. Due to the mandatory func-
tion of the modules and the fact that conscious awareness has limited
access to the functioning of these modules, conscious awareness only
sees the output of this modular system and thus takes humans to have
belief, and desires (and agency). I also explored the ontological status of
beliefs and desires as 'real patterns'.
Then moving toward the question 'Why do we attribute agency?'
I adapted the intentional stance by changing the measurt of success
from 'predictive success' to 'reproductive success over evolutionary time'
and thereby I introduced the spiritual stance. But the spiritual stance
preserved all of the implications of the original intentional stance. Thus,
if the attribution of beliefs and desires (and hence agency) to a physi-
cal system led to reproductive success over evolutionary time, then that
physical system had beliefs and desires (and hence agency) in Dennett's
semi-realist sense. Drawing on insights from the cognitive science of reli-
gion, and using a concept of personhood that bestowed moral standing
on physical systems attributed with agency, I suggested that there could
be circumstances in which such personhood/agent attributions did lead
to reproductive success over evolutionary time. In conclusion, all this
is a way to understand both the attribution of agency in religious con-
texts and the resultant (semi-real) existence of gods and spirits as 'real
spiritual patterns'. 3

Notes
1. The full passage reads as follows: 'The Laplacean omniscient physicist could
predict the behaviour of a computer - or of a live human body, assuming
it to be ultimately governed by the laws of physics - without any need for
90 Graham Wood

the risky, short-cut methods of either the design or Intentional strategies'

(Dennett, 1987, p. 23). Now, this passage reads as if the physical stance Is
read directly off the 'laws of physics' and on this point my analysis differs
from Dennett's. I take the physical stance to be distinct from the scientific
image. On my account, the physical stance is the output of the 'folk physics'
module. But it seems that Dennett takes the physical stance to be read directly
off the scientific image. But I have set this difference aside for the purposes of
this chapter.
2. There is some difference between Dennett's account and my account here.
Dennett's account implies that we can choose to swap stances, but it is not
clear to me whether the content of the three stances are produced by central
cognition or by modules (what Dennett called the manifest image). On my
account the core content (what I have called jigsaw pieces) of all three stances
are produced by modules, so the content is given to central cognition (cf. the
'given' in Sellars). But central cognition might choose to ignore the output of
a particular module or combine the outputs of modules in various ways (what
I have called putting the jigsaw together).
3. This chapter is an output from a project undertaken as part of the Cognition,
Religion and Theology Project at the University of Oxford, funded by the john
Templeton Foundation. The views expressed are not necessarily those of the
Cognition, Religion and Theology Project, the University of Oxford or the
John Templeton Foundation.

References
Baron-Cohen, S. (1995) Mindblindness: An Essay on Autism and the Theory of Mind
(Cambridge, MA: MIT Press).
Carruthers, P. (2006) The Architecture of the Mind (Oxford: Clarendon Press).
Cosmides, L., and J. Tooby, 'Foreword', in S. Baron-Cohen (1997) Mindblind-
ness: An Essay on Autism and the Theory of Mind (Cambridge, MA: MIT Press),
pp. xi-xviii.
Dennett, D. (1987) The Intentional Stance (Cambridge, MA: MIT Press).
Fodor, J. (1983) The Modularity of Mind (Cambridge, MA: MIT Press).
Gerrans, P. (2002) 'The Theory of Mind Module in Evolutionary Psychology',
Biology and Philosophy 17: 305-21.
Gigerenzer, G., P. M. Todd et al. (1999) Simple Heuristics That Make Us Smart
(New York: Oxford University Press).
Gilovich, T., D. Griffin and D. Kahneman (2002) Heuristics and Biases:
The Psychology of Intuitive Judgment (Cambridge: Cambridge University
Press).
Hookway, C. (2010) 'Pragmatism', in E. N. Zalta (ed.), The Stanford Encyclopedia
of Philosophy. Available at http://plato.stanford.edu/archives/spr2010/entries/
pragmatism/ (accessed 20 October 2011).
james, W. (1975) Pragmatism and the Meaning of Truth (Cambridge, MA: Harvard
University Press).
johnson, D., and J. Bering (2009) 'Hand of God, Mind of Man: Punishment and
Cognition in the Evolution of Cooperation', in J. Schloss and M. Murray (eds),
The Believing Primate (Oxford: Oxford University Press), pp. 26-43.
 Attributing Agency: Fast and Frugal or All Things Considered? 91

Kahneman, D. (2002) 'Maps of Bounded Rationality: A Perspective on Intu-

itive Judgement and Choice', Nobel Prize Lecture. Available at http://nobelprize.
org/nobel_prizes/economics/laureates/2002/ kahnemann-lecture.pdf (accessed
28 March 2011).
Quine, W. V. (1963) From a Logical Point of View (New York: Harper & Row).
Richardson, R. (1999) 'Heuristic', in R. Audi (ed.), The Cambridge Dictionary of
Philosophy (Cambridge: Cambridge University Press, 1999), p. 379.
Samuels, R. (2000) 'Massively Modular Minds: Evolutionary Psychology and Cog-
nitive Architecture', in P. Carruthers and A. Chamberlain (eds), Evolution and the
Human Mind: Modularity, Language and Meta-Cognition (Cambridge: Cambridge
University Press), pp. 13-46.
Samuels, R., S. Stich and P. Tremoulet (1999) 'Rethinking Rationality: From Bleak
Implications to Darwinian Modules', in E. Lepore and Z. Pylyshyn (eds), What
Is Cognitive Science? (Oxford: Blackwell), pp. 74-120.
Segall, M., D. Campbell and M. J. Herskovits (1966) The Influence of Culture on
Visual Perception (New York: The Bobbs-Merill Company).
Sellars, W. (1964) 'Philosophy and the Scientific Image of Man', in R. Colodny
(ed.), Frontiers of Science and Philosophy (London: Allen & Unwin), pp. 35-78.
Simon, H. (1962) 'The Architecture of Complexity', Proceedings of the American
Philosophical Society 106: 467-82.
Spelke, E. S. (1990) 'Principles of Object Perception', Cognitive Science 14: 29-56.
Stanovich, K., and R. West (2000) 'Individual Differences in Reasoning: Implica-
tions for the Rationality Debate', Behavioural and Brain Sciences 23: 645-65.
Wilson, D. S. (2002) Darwin's Cathedral (Chicago: University of Chicago Press).
Wilson, D. S., and S. ]. Lynn (2009) 'Adaptive Misbeliefs Are Pervasive, but the
Case for Positive Illusion Is Weak', Behavioural and Brain Sciences 32: 539-40.
Wood, G. (2011) 'Cognitive Science and Religious Belief', Philosophy Compass
6.10: 734-45.
Part III
God and the Universe
5
On Non-Singular Space-times and
the Beginning of the Universe
William Lane Craig and James D. Sinclair

1. Introduction

The most widely discussed argument for the existence of God is the
so-called kalam cosmological argument, which originated in attempts
on the part of certain ancient philosophers to rebut Aristotle's doc-
trine of the past eternity of the universe. 1 The argument assumed major
importance in medieval Islamic theology, from which its name derives.
In his Kitab al-Iqtisad the medieval Muslim theologian al-Ghazali pre-
sented the following simple syllogism in support of the existence of a
Creator: 'Every being which begins has a cause for its beginning; now
the world is a being which begins; therefore, it possesses a cause for its
beginning' (1962, pp. 15-16). In defence of the second premise, Ghazali
offered various philosophical arguments to show the impossibility of an
infinite regress of temporal phenomena and, hence, of an infinite past.
The limit at which the finite past terminates Ghazali calls 'the Eternal'
(1963, p. 32), which he evidently takes to be a state of timelessness.
Given the truth of the first premise, the finite past must, therefore, 'stop
at an eternal being from which the first temporal being should have
originated' (1963, p. 33).

The argument, then, is disarmingly simple:

1. Everything that begins to exist has a cause.

2. The universe began to exist.
3. Therefore, the universe has a cause.

Conceptual analysis of what it means to be a cause of the universe

then seeks to explore the relevance of this conclusion for theism. What

95
96 William Lane Craig and James D. Sinclair

makes the kalam cosmological argument scientifically interesting is that

contemporary proponents of the argument have claimed support from
current astrophysical cosmogony in defence of premise 2. A number of
recent critics of the argument, however, have denied that contemporary
physical cosmology lends any significant support to premise 2 because
the singular origin of space-time predicted by the standard model is an
artefact of General Relativity (hereafter GTR) which may well be resolved
by quantum cosmological models. For example, Bradley Monton states,

The big bang hypothesis is true given the assumption that GTR is
true, but we don't know that the big bang hypothesis is true of
the actual universe. The big bang theory doesn't take into account
quantum theory, and that gives us reason not to believe the big
bang theory. . . . Because the physics doesn't tell us what happens
once we trace the history of the universe backwards in time to
these high energies, we don't even know if there's a big bang at all.
(forthcoming)

Monton appeals specifically to the Steinhardt-Turok cyclic ekpyrotic

model as an example of a cosmological model which 'is compatible with
the universe having been in existence forever' (forthcoming). Similarly,
Graham Oppy complains that we are 'far from having good reason'
to suppose that quantum gravitational replacements of the standard
model will feature an absolute beginning of the physical universe' (2006,
p. 146). Undoubtedly, the most extensively developed criticism of this
sort comes from j. Brian Pitts in his wide-ranging critique of what he
styles 'the singularity argument for theism' (2008, p. 677). His critique
therefore merits careful scrutiny.
It is a curious feature of Pitts's lengthy critique that he never actu-
ally provides an explicit statement of the target singularity argument
for theism. At first blush, one might think that the kalam cosmological
argument is just the singularity argument for theism. But a moment's
reflection shows that this cannot be right, for, at most, evidence for
an initial cosmological singularity might be adduced in support of
premise 2. It is immediately obvious, then, that 'the singularity argu-
ment for theism' is a misnomer, for the singularity argument is not an
argument for theism. Rather it is an argument for premise 2, which is
a straightforward, physical statement, the likes of which is discussed in
any text on astronomy and astrophysics and which does not posit any
supernatural entity. What is really at issue, then, is some sort of singular-
ity argument for the beginning of the universe. 2 The recognition of this
 On Non-Singular Space-times and the Beginning of the Universe 97

fact is highly significant, since it implies that all of Pitts's worries about
'God-of-the-gaps' reasoning and the impenetrability of natural theology
(2008, pp. 696-700) are simply irrelevant to the success of the so-called
singularity argument for the beginning of the universe and so may be
left aside. In what follows, then, we shall have very little to say of the-
ology. This is, we think, all for the better, for the question of whether
contemporary cosmology furnishes good grounds for thinking that the
universe began to exist is of more than sectarian interest.

2. The Singularity Argument Formulated

Our concern, then, is with some supposed singularity argument for the
beginning of the universe. In order to discern the alleged failings of that
argument, we need a clear formulation of the argument. This Pitts unfor-
tunately does not provide, leaving us to reconstruct it as best we can.
The key to reconstructing the so-called singularity argument appears to
be Pitts's summarizing statement:

Whether one is tolerant or intolerant toward singularities, it turns

out that there is no first moment (unless one is installed by hand),
because every moment is preceded by earlier moments. Thus in the
relevant sense for the Kalam argument to be valid and to make sense
for a sufficiently broad collection of physical theories, there is no
beginning implied by physics, and so premise 2 might be false, as far
as physics can show. In order for the Big Bang singularity to provide a
good theistic argument, the singularity must be well enough behaved
to be a real and intelligible part of space-time and badly enough
behaved that it cannot have a past. Satisfying both conditions seems
difficult and unlikely to be achieved. (2008, p. 689)

This passage might suggest the following reconstruction:

4. The universe began to exist if there was a first moment of time.

S. If the initial cosmological singularity was a real and intelligible part
of space-time, then there was a first moment of time.
6. The initial cosmological singularity was a real and intelligible part of
space-time.
7. Therefore, the universe began to exist.

The problem with this suggested reconstruction of the argument, how-

ever, is that the argument would then be a straw man which has been
defended by no one and would hardly require 21 pages of sustained
98 William Lane Craig and fames D. Sinclair

argumentation to refute. Singularities of general relativistic space-times

are not space-time points but, at most, points, whether real or ideal,
which can be attached to space-time as boundary points. As such, they
are not parts of space-time, as (6) asserts. Indeed, on the customary
approach to singularities, they are not even accorded that much reality.
Rather a space-time is said to be singular if it is essentially inextendible
and so geodesically incomplete (Earman, 1995, pp. 28-37). The question
Pitts raises is whether an initial cosmological singularity, either in the
sense of a past boundary point or of the incompleteness of past geodesic
half-curves, gives grounds for thinking that the universe began to exist.
Asking the question in this way renders intelligible Pitts's further worry
whether some as yet unknown theory might not resolve the singulari-
ties of classical space-time - a worry which would become superfluous
with respect to the above proposed reconstruction - for even if an ini-
tial cosmological singularity implies the beginning of the universe, the
question remains whether space-time really is singular.
Accordingly, we might offer the following reconstruction of the
supposed singularity argument for the beginning of the universe:

8. If space-time is singular, the universe began to exist.

9. Space-time is singular.
10. Therefore, the universe began to exist.

Pitts may then be understood to deny (8) on the grounds that, whether
the initial singularity is understood as a boundary point to the past or as
the past geodesic incompleteness of space-time, there is no first moment
of the universe's existence, which he takes to be a necessary condition
of the universe's beginning to exist, and to undercut (9) by appealing to
as yet undiscovered theories which remove the singularity. This seems
to us a sympathetic reconstruction of the alleged singularity argument
for the beginning of the universe which also makes good sense of Pitts's
two-pronged response (Pitts, 2008, p. 682).

3. The Singularity Argument Assessed

Even so, it must be said that the singularity argument for the beginning
of the universe is still a straw man. For the evidence that the universe
began to exist does not depend on evidence for a singular beginning of
space-time. We have elsewhere discussed a variety of non-singular cos-
mogonic models and have noted the difficulties in plausibly tracing out
an infinite past (Copan and Craig, 2004; Craig, 2006; Craig and Sinclair,
 On Non-Singular Space-times and tile Beginning of tile Universe 99

2009). We shall have more to say of this in the sequel, when examining
what support contemporary physical cosmology lends to premise 2.

3.1. Assessment of premise 8

Taking the argument on its own terms, however, let us first examine
Pitts's objection to (8). Despite his flirtation with metric convention-
alism, Pitts recognizes that our universe is characterized by an objec-
tive temporal metric. A Friedman-Robertson-Walker universe, being
bounded by an initial cosmological singularity, is metrically finite in
the past. Following Smith (1985), we can say plausibly that time begins
to exist if for any arbitrarily designated, non-zero, finite interval of time,
there are only a finite number of isochronous intervals earlier than it; or,
alternatively, time begins to exist if for some non-zero, finite temporal
interval there is no isochronous interval earlier than it. That condition is
satisfied in the standard model of space-time. So why is the metrical fini-
tude of the past not sufficient for time's beginning to exist? Pitts's answer
is that 'Within the Robertson-Walker cosmological space-time fort> 0
(which is to say, always), one can explain each moment in terms of an
earlier one. Thus there is no beginning required and premise 2 might be
false, as far as physics can tell' (2008, p. 688). Pitts's point is not that
each instantaneous slice of space-time can be immanently explained in
terms of a prior slice, thereby obviating the need for a transcendent
cause of the universe's beginning to exist; 3 rather his point is that since
t = 0 is at most a boundary point of space-time, there is no first instant
of time and thus no beginning of time, even if past time is finite, and,
hence, no beginning of the universe.
Pitts's objection presupposes that beginning to exist entails having a
beginning point. But why should we think that? Pitts offers three rea-
sons why construing the universe's beginning to exist in topological
rather than metrical terms is theologically advantageous for the biblical
doctrine of creation (2008, pp. 680-1). But such a concern is foreign,
not only to natural theology, but to the religiously neutral question of
whether the universe began to exist. Relevant philosophical considera-
tions are those that have purchase with philosophers and cosmologists
regardless of their religious persuasion. What is advantageous to the
biblical doctrine of creation does not and should not enter into the ques-
tion of what it is to begin to exist. 4 Indeed, given the above intuitively
plausible sufficient conditions for time's beginning to exist, Pitts's sce-
nario of a metrically finite past lacking a first instant may be taken to
provide good reason to conclude that beginning to exist does not entail
having a beginning point. 5
100 William Lane Craig and James D. Sinclair

Moreover, Pitts's entailment would commit us to the reality of points.

But whether space and time really are composed of an actual, nondenu-
merable infinity of points, rather than simply modelled as such in GTR,
is surely a question to be settled by argument and evidence, not merely
assumed, as Pitts does (2008, p. 680).
Furthermore, Pitts's own Cosmic Destroyer argument (2008,
pp. 692-3) seems to undermine his presumed entailment. For if begin-
ning to exist entails having a beginning point, then, pari passu, ceasing to
exist entails having an ending point. Since a black hole singularity (or a
terminal cosmological singularity) is at most a boundary point of space-
time rather than a part of space-time, any object which falls into a black
hole has no ending point of its existence. Nevertheless, it does cease to
exist. As Robert Geroch explains,

Recall there is no event available to [an observer headed toward a

singularity] on the singularity itself. In particular, there is no possi-
bility for a further extension of [the observer's] world-line after it hits
the singularity. What, then, does 'the world-line of [an observer] hits
the singularity' mean physically? Mathematically, what happens is
that this world-line just stops. Physically, this would mean that [the
observer] is 'snuffed out of existence'; after some finite time according
to himself, he ceases to exist in space-time. (Geroch, 1978, p. 194)

This isn't simply rhetorical gloss on Geroch's part. The observer would
indeed be 'snuffed out of existence' despite the lack of a genuine ending
point to the space-time. If, then, the universe ceases to exist in the Big
Crunch, parity requires that it began to exist in the Big Bang, QED. 6
Finally, it deserves to be pointed out that Pitts's insistence that the
crucial consideration as to whether the universe began to exist is the
necessity of a first instant lands him squarely in the ancient Greek
paradoxes of stopping and starting (see Sorabji, 1983, pp. 403-21).
Ancient thinkers like Parmenides argued that if an object 0 is at rest
at time t, it is impossible for 0 to begin to move, since for any time
t' > t, if 0 is in motion at t', then there is a time t < t- < t' at which
0 is already in motion. Hence, nothing can ever begin to move. If we
assume the continuity of time and space, the solution to Parmenides'
puzzle is that 0 can begin to move without there being a beginning
point of its motion. Pitts's demand for topological closure would force
upon us the absurd conclusion that nothing ever begins to move. But
if we allow that beginning to move does not entail having a beginning
point of motion, then, generalizing, neither should we demand that
 On Non-Singular Space-times and the Beginning of the Universe 101

in beginning to exist the universe must have a beginning point of its

existence.
We need, then, some compelling reason to think that the admit-
ted finitude of past time is not sufficient for time's having begun to
exist, for thinking that a first instant of time is a necessary condi-
tion as well. Pitts's relevant remarks in this regard come in response
to Copan and Craig's definitions which they provide in preparation for
their defence of premise 3* (the temporal series of past physical events
was not beginningless). They explain:

By a 'physical event,' we mean any change occurring within the

space-time universe. Since any change takes time, there are no instan-
taneous events. Neither could there be an infinitely slow event, since
such an 'event' would in reality be a changeless state. Therefore,
any event will have a finite, nonzero duration. In order that all the
events be of equal duration, we arbitrarily stipulate some event as our
standard.
Taking as our point of departure the present standard event, we
consider any series of such standard events ordered according to the
relation earlier than. The question is whether this series of events had
a beginning or not. By a 'beginning' one means a first standard event.
It is therefore not relevant whether the temporal series had a begin-
ning point (a first temporal instant). The question is whether there
was in the past an event occupying a nonzero, finite temporal interval
that was absolutely first, not preceded by any equal interval. (Copan
and Craig, 2004, p. 199; emphasis original)
Pitts's objection to this Ansatz is surprising. He writes,
The Bach-Weyl theory of gravity shows that a physical theory need
not even define the length of a curve. In scalar-tensor theories or
other theories with multiple metrics, age might be radically ambigu-
ous. Thus, only a topological notion of 'beginning' in terms of a first
moment is available if the Kalam argument is intended to yield a
necessarily non vacuous necessary truth ....
Copan and Craig perhaps take their Kalam argument to express
a necessary truth that applies nonvacuously to all possible physical
theories. However, their criterion for a beginning is meaningless for
a Bach-Weyl theory because 'equal duration' is meaningless in that
theory .... Given that neither existence nor uniqueness of a metric
(for timelike curves) holds necessarily, the natural move is to adopt
a topological rather than metrical notion of beginning. Thus a first
102 William Lane Craig and fames D. Sinclair

moment is the point that needs to be addressed. (2008, pp. 677, 682;
emphasis original)

Pitts's presumption seems to be that Copan and Craig take not only
the premises of the kalam cosmological argument, but more particularly
their proffered criterion for there being a beginning of the series of past
events to be metaphysically necessary truths. Since there are possible
worlds characterized by metric conventionalism - that is to say, worlds
in which there is no objective fact of the matter whether or not non-
nested temporal intervals are isochronous - the proffered criterion for
a beginning would be inapplicable to such worlds, for there just is no
fact of the matter in such worlds whether any other temporal interval is
equal to the duration of the arbitrarily stipulated, standard event. Pitts
recognizes, as mentioned, that the actual world is characterized by an
objective metric; but he insists that 'a doctrine of creation needs to be
modally rich enough to accommodate the possibility of God's creating
worlds with physical laws without a unique or preferred metrical struc-
ture to license an answer of 'finite' or 'infinite' age' (Pitts, 2008 p. 681).
Pitts's presumption is wrong. The proffered criterion was intended
only to apply to the actual world. There is no intention or need to seek
its application to other metaphysically possible words governed by dif-
ferent laws of nature. 7 Thus, the proffered criterion for a beginning of
the series of past events is adequate to its purpose. It states a sufficient
condition for time's having begun to exist.
In sum, Pitts's attempt to discredit an intuitively plausible sufficient
condition of time and the universe's having begun to exist in terms of
metrical past finitude rests upon an ungrounded presumption about the
modal status of the proffered condition and would substitute a topo-
logical condition which, if accepted, would have highly implausible
consequences. Hence, he presents no serious objection to (8).

3.2. Assessment of premise 9

What, then, about (9)? If (9) is true, then, as we have seen above, we do
have good reason to infer that (10) is true, that the universe began to
exist, which is the second premise of the kalam cosmological argument.
Unfortunately, Pitts seems to assume that the evidential support for
premise 2 depends crucially on the cogency of his singularity argument,
so that if the warrant for (9) is undercut, so is the warrant for (2). But
this assumption is mistaken. Premise 2 does not presuppose or require
the truth of (9) or the cogency of the singularity argument. Evidence of a
non-singular beginning would be quite sufficient to warrant belief in (2).
 On Non-Singular Space-times and the Beginning of the Universe 103

Of course, if (9) is unwarranted, then we must ask what support

physical cosmology lends to (2). Pitts is worried by the fact that

As one sees all the time in papers on quantum gravity, most people
who work on quantum gravity take for granted that the Big Bang
singularity is an artifact of incomplete physical understanding and
expect or hope that uniting gravity with quantum mechanics in some
kind of quantum gravity will resolve the singularity into some well-
defined situation that admits to still earlier times, ad infinitum. (2008,
p. 688)8

The crucial consideration here will not be Pitts's worry that a suc-
cessful theory of quantum gravity will resolve the initial cosmological
singularity featured in the standard model into a situation which is
physically well defined, but rather whether, as he puts it, that situ-
ation 'admits extrapolation to still earlier times, ad infinitum'. Since
it is that consideration, rather than the truth of (9), that is crucial
to the evidence for (2), let us turn to an examination of that con-
sideration. Do models featuring an era 'before the Big Bang' imply a
beginningless past?

4. Physical Evidence for Premise 2

We shall now consider some contemporary cosmogonies.9 The mate-
rial in sections 4.2-6 is not intended as an inductive attempt to show
that all contemporary cosmogonies may lend support to (2). Rather, it
is intended to support the view evinced in Craig and Sinclair, 2009, that
there are general principles regarding cosmogonies that yield a taxon-
omy of models based on the models' expansion behaviour which speaks
to the question of temporal origin. We will hold that there is a genuine
unique objective past (temporally ordered), with evolutionary continu-
ity to the present. This preserves our veridical experience of time 10 and
permits theories of evolution (including Darwin) elsewhere in physics
to be understood on a realist footing, as opposed to some type of illu-
sion. Thus, for example, if a particular theory denies temporality as a
'fundamental' feature of the universe, but preserves the notion as a
quasi-classical 'approximation', we will take the quasi-classical history
as real and suggest the more general formalism should be interpreted
instrumentally.
Further, as Pitts seems to assert without evidence that a pre-Big Bang
timeline would, of necessity, be past eternal, we shall content ourselves
104 William Lane Craig and James D. Sinclair

to providing a few counterexamples as opposed to attempting to prove

that all pre-Big Bang models fail to demonstrate past eternality.

4.1. Models, methodology, and their use

In the sections following this one, we shall consider the evidence in
favour of the proposition that the universe has a temporal origin. Since
the consideration of such evidence commands the bulk of our essay, we
have little space to defend our implicit presupposition of time's reality.
One might base an 'atemporalist' position on the idea that a successful
theory of quantum gravity will cease to regard time as a fundamental
characteristic of nature (Kiefer, 2008). Here we offer the following brief
comments on the atemporalist position:

1. The atemporalist position is uncritically literalist with regard to model

interpretation. Our interest is in which cosmogonic model most nearly
approximates reality. The mere existence of a useful mathematical
formalism does not imply that the entities in that theory correspond
with reality. The past (and present) usefulness of GTR, for example,
never meant that we were obligated to make an ontological commit-
ment to its view that gravity just is the curvature of an objectively
real space-time. Nor are we obligated to believe that time does not
exist because a 'pure' Wheeler-DeWitt formalism does not contain
it and the WKB surrogate is described as merely an 'approximation'.
The Wheeler-DeWitt quantum gravity approach describes both the
disappearance of absolute time that results from the attempt to quan-
tize Einstein's gravity as well as time's reappearance as a semi-classical
approximation. The fact that time is recoverable as an approxima-
tion, however, does not imply that there is either an elimination of
time ('pure' Wheeler-DeWitt) or that the superposition of all possi-
ble metrics necessarily requires an indeterminate meaning to time
in reality. Useful models exist at different levels of abstraction from
physical reality. We therefore feel perfectly comfortable invoking the
approach of Kiefer (and his sometime colleague H. D. Zeh) in our dis-
cussion of cyclic universes and a reversed arrow of time (see section
4.6) on the grounds that one need not attach ontological commit-
ment to their larger metaphysical claims regarding the elimination
of time.
The issue at hand is therefore not the narrow issue of whether
there are quantum gravity modelling schemes that are atemporalist.
Rather the question is what determines our ontological commit-
ments. Although contemporary philosophy of time has been deeply
 On Non-Singular Space-times and tile Beginning of tile Universe 105

imbued with the naturalized epistemology of W. V. Quine (1969),

which shuns so-called 'first philosophy' in favour of taking the
deliverances of the natural sciences as authoritative guides to real-
ity, such epistemological naturalism is by its very nature simply a
methodological disposition, which cannot itself be grounded either
scientifically or in any other way, to restrict one's basic sources of evi-
dence to the natural sciences (Rea, 2002, pp. 63-7; cf. pp. 1-7). Not
being so disposed, we see no reason that physics should subvert the
veridicality of our experience of tense and temporal becoming, much
less of time itself. Indeed, even the illusion of temporal becoming
entails the reality of temporal becoming in the contents of conscious-
ness. We find it odd that it is 'ordinary' quantum mechanics which
brings to the table a near Newtonian view of absolute time, yet this
is the ingredient that supposedly spoils the notion of temporality.
It is 'ordinary' quantum mechanics that enjoys the impressive exper-
imental verification. We should take care in transferring a borrowed
legitimacy to a quantum gravity theory that itself lacks consensus.
It seems to us appropriate to consider each and every scheme that
cosmologists bring to the table to throw light on the question of
which models approximate physical reality. We are not wedded to
the specific concept of metric time found in GTR. Authors of cos-
mogonies routinely attempt to address either the origins question
itself or, at least, questions regarding the universe's initial condi-
tions. Many quantum cosmogonies include claims of past eternality
(and thus presuppose the reality of time). Others discuss cosmic
beginnings ex nihilo. Surely the cosmologists who formulate these
theories think that they are doing work relevant to the actual uni-
verse and not merely mathematical exercises that have no basis in
reality. Hence, it seems to us a profitable and important exercise to
address such models.
2. Quantum Gravity models must replicate the observed behaviour that more
primitive models of the universe (read GTR) seek to model. Isaac Asi-
mov, in an article entitled 'The Relativity of Wrong' (Asimov, 1989),
pointed out the fallacy of believing that current 'primitive' models
lack usefulness because they are 'wrong', given that a future theory
is always just over the horizon to overturn the current one. There are
degrees of 'wrongness' and theories which are less wrong than their
predecessors are so because they take account of what is empirically
correct in their forebears. It is simply mistaken to think that phenom-
ena predicted and analysed by GTR (especially indirectly observable
phenomena such as black holes) via such things as singularity
106 William Lane Craig and James D. Sinclair

theorems cease to be relevant because a different mathematical for-

malism may someday come into general use. There may be no such
things as singularities per se in a future quantum gravity formalism,
but the phenomena that GTR incompletely strives to describe must
nonetheless be handled by the refined formalism, if that formalism
has the ambition of describing our universe. This can be seen in
the above example of the Wheeler-Dewitt approach as described by
Kiefer. Big Bang Friedmann universes with a meaningful time coordi-
nate are recoverable in the theory (as Kiefer shows) but are described
as a semi-classical WKB approximation which invokes a mechanism
called decoherence. They in fact must be recoverable because that is
what we see. If Big Bang Friedmann-like behaviour were not recover-
able, that in itself would be grounds for rejecting the larger theory.
3. Preservation of time as a fundamental property of the universe is pos-
sible within a quantum gravity approach (as examples, see Mersini-
Houghton, 2009; Smolin, 2009; Sorkin and Rideout, 1999; Carroll,
2008; Loll, 2008). Mersini-Houghton defends an arrow-less but
nonetheless global and beginningless temporal axis within a larger
multiverse. Carroll wants a global time with an arrow that points
toward greater entropy given any arbitrary point of departure within
the multiverse. Smolin defends a similar view in his postulation of
multiple universes birthed in a temporal sequence through black
hole incubators. His article 'The Unique Universe' is a full-fledged
attempt at refutation of a timeless multiverse. Yet his defence of
the concept of fundamental time is not a defence of GTR's metric
time. Rather it defends temporality within the context of background
independent (i.e. emergent space-time) quantum gravity theories.
An example of this approach is the work of Renate Loll which
we briefly discuss in section S. Another approach which includes
explicitly the property of tense and at least half of the doctrine of
presentism (i.e. the thesis that the future does not exist) is the 'causal
set theory' of Rafael Sorkin, David Rideout and other prominent
theorists, such as Faye Dowker.
In our view, these approaches have a leg up on timeless multiverse
quantum gravity approaches precisely because they are consistent
with the veridicality of our experience of time and tense. In order
to justify a claim so extraordinary as the unreality of time, evolution
and illusoriness of our experience thereof, the timeless multiverse
needs something beyond demonstration as a 'consistent' quantum
gravity formalism.
 On Non-Singular Space-times and the Beginning oftl!e Universe 107

4. The atemporalist position agrees with ours with regard to the fact that dif-
ferent universes within a multiverse are now incommensurate with respect
to a temporal measure. Lastly, the atemporalist position may actually
prove to be supportive of the fact of the finitude of the past, if its
attempt to eliminate time as a measure should turn out to succeed
only in part. Suppose, for example, that the concept of 'emergent
time' turns out to be meaningful along with the objectivity of tense
(we discuss this in section 4.6). The impact on supposed 'pre-Big
Bang' cosmogonies can be enormous, if one can no longer stretch
a meaningful time back 'through' the Big Bang 'singularity'. If uni-
verse phase A is incommensurate with respect to time with universe
phase B, then it is just false to suggest that phase A precedes phase B.
If we suppose that time is a meaningful local concept within A and B,
then the implication is that both A and B have temporal origins. This
is supportive of the finitude of the past and, hence, of the beginning
of the universe.

4.2. Expanding universe

Pitts seems to think that the person who believes that current physi-
cal cosmogony lends significant support to (2) must be what he calls a
'GTR exceptionalist', that is, someone who tolerates singularities (and,
perhaps, takes GTR to be the fundamental theory of gravity). A GTR
exceptionalist would be a person who holds that our universe obeys
the Hawking-Penrose singularity theorem (Hawking and Penrose, 1970).
Hawking and Penrose assumed that gravity is always attractive (that is
to say, their theorem dealt only with 'ordinary' types of matter). just
about 30 years ago, however, the idea of 'inflation' was introduced into
cosmology as a resolution to several of the anomalies in the standard
hot Big Bang picture, and a key characteristic of inflation is that it is
driven by a type of energy that violates the energy condition of the
Hawking-Penrose theorem. Thus it became a live question as to whether
inflationary cosmogonies might accommodate a beginningless past. Per-
haps we are living in a vacuum bubble that resides in a much larger
(perhaps past and future fractal) structure of nested bubbles.
On this view, a small volume of 'false vacuum' may decay into the
'true vacuum' in which we live. This process, first suggested by Sidney
Coleman and Frank De Luccia (Coleman and De Luccia, 1980) would
identify our Big Bang with one of these quantum tunnelling 'decay'
events. Thus the Big Bang would not represent an absolute beginning
of space and time. Rather there would be a pre-Big Bang past.
108 William Lane Craig and James D. Sinclair

What Pitts does not mention, however, is that singularity theorems

have also evolved in the intervening 30 years. There has been a lively
debate centred (mostly) around the role of energy conditions in these
theorems. In 2003 Arvind Borde, Alan Guth and Alexander Vilenkin
published a singularity theorem that was completely independent of
energy conditions. Vilenkin explains:

A remarkable thing about this theorem is its sweeping generality.

We made no assumptions about the material content of the uni-
verse. We did not even assume that gravity is described by Einstein's
equations. So, if Einstein's gravity requires some modification, our
conclusion will still hold. The only assumption that we made was
that the expansion rate of the universe never gets below some
nonzero value, no matter how small. This assumption should cer-
tainly be satisfied in the inflating false vacuum. The conclusion is
that past-eternal inflation without a beginning is impossible. (2006,
p. 175)

In their formal paper, Borde, Guth and Vilenkin state that 'Our argu-
ment shows that null and time-like geodesics are, in general, past-
incomplete in inflationary models, whether or not energy conditions
hold, provided only that the averagt:d expansion condition Hav > 0 holds
along these past-directed geodesics' (2003, p. 3).
Borde et al. suggest that timelike and null geodesics are redshifted in
energy as they are stretched out in an expanding space. Looking back-
wards in time, they are blueshifted. The blueshift becomes infinite in
a finite amount of proper time (or affine parameter for null geodesics).
In GTR, the infinite blueshift suggests a singular condition.
The BGV conclusion applies beyond the inflationary model class. Any
universe (including universes modelled by higher dimensicnal cosmol-
ogy, pre-Big Bang cosmology and so forth) which, on average, expands
has to connect, in a finite time, to a past boundary. 11 According to
Vilenkin, this result is independent of modifications to Einstein's equa-
tions (which could be the result of 'low energy' corrections coming from
a quantum gravity approach) or the particular characteristics of the pro-
posed cosmogony. If the universe (or multiverse) expands (on average),
then it has a beginning, period.
It is possible to object that the BGV theorem is a classical result lacking
the rigor of a full quantum gravity approach. The objection is that BGV
is based on metric tensors, which one could claim would not exist in a
'full' theory of quantum gravity.
 On Non-Singular Space-times and tile Beginning of tile Universe 109

As discussed in section 4.1, we believe the objection is without merit.

We add that Borde, Guth and Vilenkin are not silent on the topic of
quantum gravity. They indicate that their results demonstrate a bound-
ary where a quantum gravity theory will be necessary to describe further
the physics. Thus the classical dynamics governed by their theorem are
adequate to describe the universe at 'low' energy states, including its
past history roughly back to the Planck time. There is then a quantum
gravity reckoning:
What can lie beyond this boundary? Several possibilities have been
discussed, one being that the boundary of the inflating region cor-
responds to the beginning of the Universe in a quantum nucleation
event. The boundary is then a closed spacelike hypersurface which
can be determined from the appropriate instanton.
Whatever the possibilities for the boundary, it is clear that unless
the averaged expansion condition can somehow be avoided for all
past-directed geodesics, inflation alone is not sufficient to provide
a complete description of the Universe, and some new physics is
necessary in order to determine the correct conditions at the bound-
ary. This is the chief result of our paper. (Borde, Guth and Vilenkin,
2003, p. 4)
As briefly mentioned earlier, a Feynman path-integral-based cosmogony
such as the Hartle-Hawking 'no-boundary' (Hawking and Hartle 1983),
or the Vilenkin 'tunneling from nothing' approach (Vilenkin 1982) can
be interpreted as a creation-ex-nihilo 12 that would explain the boundary
suggested by the BGV theorem as well as lend support to kalam's second
premise.
We discuss this quantum reckoning in section 5 of our paper. So either
the boundary represents a quantum nucleation event or else (within
the context of Pitts's original objection) we are left with the options
discussed in sections 4.3-4.6 of this paper.
A possible further objection to the BGV theorem might draw upon
Pitts's suggestion (2008, p. 681) that the Bach-Weyl approach to GTR
does not assign lengths to curves. Demonstrating a past inextendible
geodesic (i.e. one of finite length) is precisely how BGV purports to show
that their boundary exists. Does the Bach-Weyl formalism give us, then,
reason to doubt that objective scales exist?
We question whether Bach-Weyl should be taken as a realist approach
to modelling. It may well be better understood instrumentally. 13
We note that some theorists seem to prefer a more modest use of
conformal invariance. Roger Penrose's Cyclic Conformal Cosmogony
110 William Lane Craig and James D. Sinclair

(see section 4.5), for example, uses exactly the Bach-Weyl formalism to
model the boundary between consecutive oscillations. But Bach-Weyl
requires all contents of the universe to be conformally invariant. Mass-
less particles such as the photon and the graviton meet this criterion.
In Bach-Weyl one splits a metric into a scale-free part and another part
that measures space and time intervals. But massless particles (only) do
not experience a passage of time. So if the universe contained only mass-
less particles, one could argue (as Pitts does) that one could drop the part
of the metric that represents intervals. But where mass exists, conformal
invariance does not. So for all phases of the universe where mass exists
(which in the real universe is pretty much every era after the Big Bang),
one needs the full metric. Penrose thus splits his model into two epochs:
one consistent with conformal invariance and one without it.
Similarly Gerard 't Hooft, an advocate of the Bach-Weyl approach,
describes other necessary parts of a cosmogony that are not scale invari-
ant and are thus problematic for a 'pure' Bach-Weyl methodology. In a
paper defending the existence of space-time (as an objective feature of
nature) at the expense of 'emergent time' theories he states,

Describing matter in a g,,u metric will still be possible as long as we

restrict ourselves to conformally invariant field theories, which may
perhaps be not such a bad constraint when describing physics at the
Planck scale. Of course that leaves us the question where Nature's
mass terms come from, but an even more urgent problem is to find the
equations for the gravitational field itself, considering the fact that New-
ton's constant GN is not scale-invariant at all. The Einstein-Hilbert action
is not scale-invariant. Here, we cannot use the Riemann curvature or
its Ricci components, but the Weyl component is independent of w,
so that may somehow have to be used. ('t Hooft, 2009, p. 8; emphasis
added)

Ultimately 't Hooft believes that scales are introduced in nature via the
fundamental processes of information transfer. He sums up his paper
by arguing, 'The density of this information flow may well define the
Planck length locally, and with that all scales in Nature' (2009, p. 8).
Thus he supports the idea that scales (time and space intervals) are
objectively real features of nature.
It therefore seems to us that the mere fact of the Bach-Weyl approach
(which is all that Pitts offers) proves little. It may describe possible
worlds that are fully conformally invariant, but we are interested in a
model that can represent the matter-filled universe we see around us.
 On Non-Singular Space-times and tile Beginning of tile Universe 111

What, then, has been the reception of the BGV theorem by the
physics community? It, as well as its implications, is largely uncon-
troversial. 14 Instead, a new round of model building has ensued based
on possible exceptions to this theorem. Four alternatives present them-
selves:

1. Infinitely contracting universe 'bouncing' into an expansion phase

(average expansion< 0; example: de Sitter cosmogony).
2. Asymptotically static universe (average expansion = 0; example:
Emergent model class).
3. Eternally cyclic universe (average expansion = 0; example: Penrose
conformal cyclic cosmogony).
4. Reversal of the arrow of time (example: Aguirre-Gratton model).

Leaving (4) aside for the moment, let us consider universes that do not,
on average, expand over their past histories.

4.3. Infinitely contracting universe

Is a pre-Big Bang infinite contraction (followed by a rebound at a near
singular condition into our present day expansion) as envisioned in (1) a
promising alternative? We suggest not. George Ellis comments on the
problems that bedevil such an approach:

The problems are related: first, initial conditions have to be set in an

extremely special way at the start of the collapse phase in order that it
is a Robertson-Walker universe collapsing; and these conditions have
to be set in an acausal way (in the infinite past). It is possible, but a
great deal of inexplicable fine tuning is taking place: how does the
matter in widely separated causally disconnected places at the start
of the universe know how to correlate its motions (and densities) so
that they will come together correctly in a spatially homogeneous
way in the future?
Secondly, if one gets that right, the collapse phase is unstable, with
perturbations increasing rapidly, so only a very fine-tuned collapse
phase remains close to Robertson-Walker even if it started off so, and
will be able to turn around as a whole (in general many black holes
will form locally and collapse to a singularity).
So, yes, it is possible, but who focused the collapse so well that it
turns around nicely? (pers. comm., 25 january 2006)

First, then, such models encounter the significant problem of acausal

fine-tuning. One asserts not just brute contingency but also a very
112 William Lane Craig and ]ames D. Sinclair

curious form of it. In the face of apparent fine-tuning, physicists usually

prefer to offer some type of explanation. For example, cosmologists are
avidly seeking an explanation for apparent fine-tuning of the parameters
of the standard model (such as force-coupling constants) in the form of
a multiverse or a superdeterministic Theory of Everything. Or one thinks
of Guth's inflationary resolution of the horizon problem (past thermo-
dynamic equilibrium). If we are going to give up explanation, then what
was wrong with leaving cosmology as it was prior to 1980, namely, the
standard hot Big Bang model (with the associated breakdown of physics
at the singularity)? 15
The second problem is that the collapse becomes chaotic as it
approaches the singularity. This will produce a pre-expansion start con-
dition that is known to be dramatically different from our actual Big
Bang. This phenomenon is referred to as 'BKL chaos' after its discover-
ers (Belinsky, Khalatnikov and Lifshitz, 1970). The same problem will
appear for all attempts at a past-eternal timeline that seek to introduce
a pre-Big Bang phase that 'bounces' into the present expansion. In fact,
the real implication of BKL may well be that it is physically impossible
to 'bounce' through a 'singularity'. 16 So option (1) is unpromising.

4.4. Asymptotically static universe

An asymptotically static space, as envisioned in option (2), is one in
which the average expansion rate of the universe over its history is
equal to zero, since the expansion rate of the universe 'at' past infinity
is zero (thus 'infinity' dominates any finite expansion phase, no mat-
ter how long). Hence, the universe, perhaps in the asymptotic past, is
in a static state (neither expanding nor contracting). This feature of the
model allows it to escape the BGV singularity theorem. Consider, for
example, the GTR-based 'Emergent' model class of George Ellis et al.
(Ellis, Murugan and Tsagas, 2004; Ellis and Maartens, 2004). It features
two stages: an Einstein Static State (ESS) and an inflationary phase that
leads to our present, dynamic, expanding universe.
The Einstein static universe itself was originally viewed as past eternal.
But there are obvious problems with this model. The reason Einstein
himself dropped it was its feature of unstable equilibrium. Although,
in pure non-quantum GTR, one can consider a static state with world-
lines that trace to negative infinite time, in reality we know that gravity
is a quantum force. As Vilenkin notes, 'Small fluctuations in the size
of the universe are inevitable according to the quantum theory, and
thus Einstein's universe cannot remain in balance for an infinite time'
(Vilenkin, 2006, p. 209). On the other hand, the current observable
 On Non-Singular Space-times and the Beginning of the Universe 113

universe is demonstrably not in a static state. So there would need to

be at least two stages in such a model, a primordial ESS followed by an
expansion phase. A quantum fluctuation would force a transition from
ESS to an expanding universe. But this very mechanism implies that the
initial state is not past eternal, since such a fluctuation will inevitably
occur within a finite time. 17
To preserve past eternality, one could claim that the static state is
only an 'ideal point' asymptotically approached at past infinity. Some
philosophers have expressed reservations, however, over the contrived
nature of the past infinity featured in such an interpretation. The cos-
mological (and thermodynamic) arrows of time are so weak that they
are indistinguishable from a 'timeless state'. For example, RUdiger Vaas,
characterizing the Emergent models as 'soft-bang/pseudo-beginning' in
nature, views the asymptotic approach toward ESS as a mathematical
artefact (Vaas, 2004, p. 18).
A further problem with these models is the implied fine-tuning of
the initial state. One way to look at this is to imagine starting in the
present and extrapolating the past history of the universe. Will the resul-
tant evolution, with high probability, produce an ESS state? No, such
an outcome is improbable. Ellis is sensitive to the fine-tuning problem
but thinks his approach worth pursuing in the absence of a quantum
gravitational resolution (Ellis and Maartens, 2004, p. 228).
Subsequent to his initial work on Emergent models, which were based
solely on GTR, Ellis along with his colleagues integrated a quantum
gravity approach into their models. Ellis's colleague David Mulryne
elaborates: 'The importance of the [equilibrium state as described by
quantum gravity] is that, in contrast to the [ESS] solution present in
GTR, slight perturbations do not result in an exponential divergence
from the static universe, but lead instead to oscillations about it' (Mul-
ryne et al., 2005, p. 6). The ESS is now viewed as a 'low-energy' solution
of loop quantum gravity (LQG) to make the Einstein state stable against
perturbations of a limited size. LQG theorist Martin Bojowald explains
how Mulryne et al. use the mechanism of perturbation to cause an
initially stable oscillation to escape to full-blown inflation:

Static solutions do not evolve, and so are clearly ill-suited as a

model for the Universe. But by introducing a perturbation to a static
solution, one can slightly change it and thereby start a more interesting
history. Unfortunately, the classical solution (ESS) is unstable: any
disturbance grows rapidly, leaving little of the initial state behind.
The insight of Mulryne and colleagues is that quantum effects could
114 William Lane Craig and James D. Sinclair

supply all the necessary ingredients where classical solutions do

not. Within the framework of loop quantum gravity, repulsion also
implies static solutions at small size, but these - in cor.trast to the
classical case - are stable. According to the authors' model, perturbing
such a state leads to small cycles of interchanging expansion and contrac-
tion. During this process, matter will evolve slowly, and the cycles
will gradually change their behavior. By itself, this perpetual recur-
rence and incremental change seems to lack the spark necessary for
so momentous an event as the birth of the Universe. And indeed,
Mulryne and colleagues identify one final theoretical ingredient that
lights this spark: mediated through repulsive effects, potential energy
is gradually pushed into the matter during its slow evolution. At the
point when potential energy starts to dominate kinetic energy, the mun-
dane cycling is broken by a sudden, dramatic inflationary explosion - the
emergent Universe. (Bojowald, 2005, pp. 920-1; our emphasis)

Should the past be judged infinite in such an approach? There is a claim

by its authors to possible past eternality: 'The universe undergoes a
series of non-singular oscillations in a (possibly) past-eternal phase with
the field evolving monotonically along the potential' (Mulryne et al.,
2005, p. 6).
Similar to the earlier Emergent models of Ellis et al., which proposed
an ideal ESS point 'reached' in the infinite past, the model of Mulryne
et al. proposes that any point which is close to the stable equilibrium
point could serve as an 'initial' state for the model. The size of the
universe, then, oscillates about this initial state, always maintaining a
non-zero positive value. This oscillation has an amplitude which is asso-
ciated with a maximum size for the universe (per oscillation) which we
will call amax1 (see Bojowald and Tavakol, 2008a, fig. 2, and Mulryne
et al., 2005, fig. 7). In the absence of the self-interaction property of the
scalar field !fJ, this condition is said to be perpetual. But the potential
energy in the system must grow, forcing amax to grow with each sub-
sequent cycle. This is said to introduce a cosmological arrow of time,
which is based on the evolutionary increase in amax. We will take the
equation amax(t) to be fundamental; the potential V(!(J) and scalar field
!(J(t) are reverse engineered to produce it (see Ellis, Murugan and Tsagas,
2004, section II text and fig. 1, and Mulryne et al., 2005, fig. 5). 18
In the limit t-+ -oo, it is desired that the change in amax (call it
Aaevoiution) between cycles asymptotes to zero. Note that the interval in
time between a cycle with maximum amplitude amax1 + Aa (with arbi-
trarily small ~a) and the present is finite; this is shown by the numerical
 On Non-Singular Space-times and the Beginning of the Universe 115

analysis in Mulryne et al., 2005. So if the initial process of reaching

amax1 +A a has a finite timeline, the full timeline to the present is finite.
Now, consider a quantum fluctuation of the scale factor, Aa11 uctuauon·
This is a separate, instantaneous, mechanism whereby the universe may
reach a larger size. For the universe at amax (Bojowald and Tavakol,
2008a, fig. 2 and equations 32 and 33), and in the past asymptotic
limit of an unsqueezed quantum state, 19 the fluctuation is roughly
proportional to amax. So looking to the infinite past, Aaouctuauon will
be>> Aaevolutlon·
This suggests two things. First, as one looks to the past, the (typical)
size of fluctuations of the scale factor eventually exceeds and domi-
nates the evolutionary increase (in amax) between cycles. As Vaas might
describe it, there is no cosmological arrow of time in this region. Second,
it does not seem reasonable to assert that the universe will need an infi-
nite amount of time to grow amax1 to amplitude amax1 + Aa (where
Aa can be arbitrarily small). The probability should be unity that it will
reach that goal by instantaneous, discrete transition in finite time, as
opposed to continuously evolutionary growth over infinite time.
Furthermore, since the early phase of this model features an oscillat-
ing universe, it also encounters an entropy objection to past eternality
(see section 4.5). It seems, then, that the past eternal nature of this
model is a mathematical artefact. One need not reject the model; one
merely notes that there must a temporal origin of some sort in the
finite past.
In sum, here is a counterexample to Pitts's notion that a past lack-
ing a singularity must of necessity be eternaJ.2° The Emergent models
that feature (only) a GTR based ESS are metastable. 21 An improvement
to the Emergent model resolves this instability by providing quantum-
based modifications to the Friedman equation. As the authors say, this
provides 'partial' ameliorization of the metastability objection and the
fine-tuning objection (Mulryne et al., 2005, abstract). But the nature of
quantum fluctuations demands that one cannot have an infinite time-
line during which an arbitrarily small, continuous evolution in scale
factor occurs. It seems to us this objection would apply, generally, to
asymptotically static constructions. 22
So far, then, we suggest that there are good arguments to believe that
contemporary pre-Big Bang universe models that contract, are static or
expand (on average) either imply a beginning or are untenable. The
expanding models obey a newly discovered singularity theorem (BGV)
and likely have a beginning of a quantum nature. The contracting mod-
els featured in alternative 1 seem to be unpromising and add nothing
116 William Lane Craig and fames D. Sinclair

to the standard hot Big Bang model from the standpoint of explanation.
The static models featured in alternative 2 imply a beginning but do not
necessarily describe how that state came to be.

4.5. Eternally cyclic universe

What, then, about cyclic models, as envisioned in option 3 above?
According to these models, the universe goes through a cycle in which
it grows from zero (or near-zero) size to a maximum and then con-
tracts back to its starting condition. The universe itself is periodic, in the
sense that it undergoes many such cycles, perhaps an infinite number.
The average expansion of the universe would be zero in a 'pure' cyclic
model, since cycle by cycle, the universe always undergoes precisely
equal amounts of expansion and contraction. Hence, a cyclic model
evades the BGV Theorem.
Cyclic models face a well-known thermodynamic problem. As
Vilenkin notes, 'A truly cyclic universe has a problem with entropy
increase: it should have reached thermodynamic equilibrium by now'
(pers. comm., 19 january 2008). Our observation of the present day
universe indicates that we are not at a condition of thermodynamic
equilibrium- a good thing for us, as life requires nonequilibrium condi-
tions in order to exist. As one looks into the past, the size of each cycle is
also thought to decrease (due to radiation effect on entropy; see Barrow
and Dabrowski, 1995). Eventually a 'Planck entropy' could be reached
(the minimum currency of entropy exchange), which would preclude
the existence of still earlier cycles. A problem similar to that encoun-
tered for an asymptotically static cosmogony can also occur. If one is
attempting a realist ontology, the backward evolution of a cyclic model
must stop once quantum fluctuations in volume are of the same size as
the volume expectation value. One can continue pursuit of the problem
(the classical background disappears) in a full quantum approach (pers.
comm., Martin Bojowald, 15 April 2010). If this means that time 'disap-
pears' or that the arrow of time reverses, the implications are discussed
in section 4.6.
Cosmologists Thomas Banks. and Willy Fischler contend that a con-
tracting space filled with quantum fields will have an 'ergodic' property
as the space shrinks. Its fields become highly excited as one approaches
the end of contraction, and these fields will produce chaotic fluctua-
tions. Spontaneously created matter with a different equation of state
will dominate the energy density. That, and the inhomogeneity of the
fluctuations, will prevent cycling. Banks and Fischler even suggest that
the fields will spontaneously produce a dense 'fluid' of black holes
 On Non-Singular Space-times and the Beginning of the Universe 117

leading to a condition they call a 'Black Crunch' (Banks and Fischler,

2002) for arbitrary states approaching full contraction. 23
Martin Bojowald and collaborators (Bojowald, Maartens and Singh,
2004; Bojowald, 2006; Bojowald and Tavakol, 2008a, 2008b) have done
a significant amount of work on building contemporary cyclic models
to address these difficulties. The LQG approach does seem to resolve
the singularity admirably in its initial tests against simple (homoge-
neous) models (Ashtekar, Pawlowski and Singh, 2006; Bojowald, 2007),
although it is unclear if this approach solves all the problems brought
up by opponents such as Banks and Fischler. 24
In 2006, Bojowald suggested the possibility that the recollapse phase
of each oscillation was entropy-reducing (see Craig and Sinclair, 2009,
p. 172). Perhaps the universe does not cumulatively gain entropy cycle
to cycle. But if so, this could lead to a different problem in that it is
possible to interpret an entropy reducing recollapse in terms of a reversed
arrow of time; thus the 'recollapse' is really another expanding universe.
We shall take up this question again in section 4.6.
In 2008, Martin Bojowald and Reza Tavakol entertained the oppo-
site suggestion; entropy is monotonically gained from cycle to cycle.
Quantum squeezing of the conjugate variable pair V (volume) and
P (momentum) of the universe was identified as an entropy con-
tributor without classical parallel. Interestingly, the quantum gravity
model now displays the same behaviour with respect to entropy gain
that Barrow and Dabrowski (1995) note for classical models in their
paper:

An interesting question is whether in a cyclic model one generically

expects to have a finite or an infinite number of past cycles. The
problem with the finite case is that it does not resolve the origin
question. In the emergent scenarios, as well as some other such mod-
els, the universe is assumed to have undergone an infinite number
of past cycles so as to remove the question of the origin. In that
case any given cycle would have an infinite number of precursors
and generically we therefore have to expect the current state to be
squeezed .... The question then is how the squeezing in a generic
cycle is determined. If each cycle produces the same amount of squeez-
ing, a generic cycle would have infinitely squeezed states, which could not
be semiclassical . .. For growing cycles, as in the emergent scenario, the
change in squeezing is initially small and approaches zero for cycles in
the infinitely distant past. Depending on the precise scenario, the sum
of all squeezing contributions may converge, such that a finite value
118 William Lane Craig and James D. Sinclair

results for a generic cycle. (Bojowald and Tavakol, 2008a, p. 8; our

emphasis)

Similarly Aguirre comments:

For an observer along that worldline to perceive an arrow of time

(AOT) it must see local net entropy generation. To the past, it sees
entropy destruction. In a finite neighborhood, there is finite entropy,
so going far enough 'to the past' the entropy must (a) asymptote to
a constant, (b) start to increase again, or (c) become ill-defined as a
singularity is encountered. (Aguirre, 2007, p. 30)

So avoiding a reversal of the AOT requires entropy to asymptote to

a constant in the infinite past. In a quantum model, the squeezing
of states is going to be a factor in accomplishing this. In their paper
Bojowald and Tavakol suggest both that a special 'unsqueezed' state rep-
resents the asymptote but also that it is problematic to assert that such a
state existed. Bojowald (pers. comm., 15 April 2010) has indicated to us
that there were further problems since discovered with his and Tavakol's
assumptions on monotonicity of variables. So, as of this writing, the
problem of structuring entropy to be infinitely subdividable remains.
Returning to the general discussion, suppose that total entropy can
be kept constant with cumulative cycles and that contracting phases are
not reinterpreted as having a reversed arrow of time. There remains the
issue of dark energy, which may have the potential to halt the cycling
and induce an open-ended expansion. The current empirically observed
dark energy effect, for example, appears adequate to produce an open-
ended, accelerated expansion. This result would be definitive if the dark
energy is of the form of a cosmological constant (that is to say, if its
value is independent of space and time (see Barrow and Dabrowski,
1995). Indeed, open-ended, accelerating expansion does appear to be
the fate of the present day universe (Overbye, 2006). If an entropy
gain (cycle-to-cycle) is denied, then the amplitude of the oscillation
(maximum scale factor experienced by the universe) is constant. But
then one could never have more than one 'cycle', for the cosmologi-
cal constant would lead to open-ended expansion the first time. Hence,
the initial cosmological singularity (our Big Bang) would represent an
absolute beginning.
There are some genuinely exotic cyclic models that purport to escape
these problems (see Craig and Sinclair, 2009 for an analysis of vari-
ous cyclic models current in contemporary cosmological discussions).
 On Non-Singular Space-times and the Beginning of the Universe 119

A fascinating attempt to counter the entropy and dark energy prob-

lems mentioned above is Roger Penrose's Cyclic Conformal Cosmogony
(Penrose, 2006). Penrose suggests a solution to the entropy problem by
saying that the initial'singularity' is the same thing as the open-ended de
Sitter-like expansion that our universe seems about to experience. Their
mathematical equivalence is demonstrated through an appropriate con-
formal transformation of one state into another. Penrose explains how
we are to think about this situation:

Physically, we may think that again in the very remote future, the
universe 'forgets' time in the sense that there is no way to build a
clock with just conformally invariant material. This is related to the
fact that massless particles, in relativity theory, do not experience any
passage of time .... With conformal invariance both in the remote
future and at the Big-Bang origin, we can try to argue that the two
situations are physically identical, so the remote future of one phase
of the universe becomes the Big Bang of the next. (Penrose, 2006,
p. 2761; emphasis original)

Penrose admits that his view is heterodox in that, for this envisioned
scenario to obtain, all massive fermions and massive, charged particles
must disappear to radiation, including, for example, free electrons. He
concedes that there is (currently) no justification for positing this.
The Cyclic Conformal Cosmogony is based on the Weyl Curvature
Hypothesis (WCH) and Paul Tad's implementation of this idea within
GTR (Tod, 2003). WCH is defined as follows:

Weyl curvature is the kind of curvature whose effect on matter is

of a distorting or tidal nature, rather than the volume-reducing one
of material sources .... The physical conjecture that I refer to as the
Weyl curvature hypothesis asserts that (in some appropriate sense)
the Weyl curvature is constrained to be zero (or at least very small)
at initial singularities, in the actual physical universe. (Penrose, 2005,
pp. 765, 768)

Penrose describes the mathematical technique necessary to stitch a

singularity to a maximally extended de Sitter expansion:

Tad's formulation of WCH is the hypothesis that we can adjoin

a (past-spacelike) hypersurface boundary to space-time in which
the conformal geometry can be mathematically extended smoothly
120 William Lane Craig and James D. Sinclair

through it, to the past side of this boundary. This amounts to

'stretching' the metric by a conformal factor Q which becomes infinite
at the Big Bang singularity, so that we get a smooth metric gab which
actually extends across this boundary. So far, we regard the conformal
'space-time' prior to the Big Bang as a mathematical fiction, intro-
duced solely in order to formulate WCH in a mathematically neat
way. However, my 'outrageous' proposal (4) is to take this mathe-
matical fiction seriously as something physically real. (Penrose, 2006,
p. 2761; emphasis original)

A potential failing of this approach is the supposed correspondence

between Weyl curvature and entropy. The correspondence seems clear
enough when one is considering the structure of the initial Big Bang
singularity, given its vanishingly small entropy state. But while the de
Sitter-like end state of the universe also minimizes Weyl curvature, its
entropy is maximized. Like a black hole obeying the Hawking-Bekenstein
entropy law, a de Sitter space has a cosmological horizon with entropy
proportional to its area. It is generally believed that this state represents
the maximum entropy that can fit within the horizon.
Penrose has recently chosen to regard the entropy of the cosmological
horizon as spurious and to invoke non-unitary loss of information in
black holes in order to equalize the (vanishingly small) entropy at the
boundary (Penrose, 2009, p. 15). Penrose attributes the large entropy at
late universe times (but before significant decay of the universe's black
hole population) to degrees of freedom internal to the black holes. He
then suggests that in CCC the universe's entropy is 'renormalized' so
that we can discount the entropy contribution from the horizon when
all black holes have evaporated.
On the other hand, it seems, Penrose notwithstanding, that the
entropy of the cosmological horizon must have physical meaning, since
the entropy of the de Sitter-like system outside a black hole must be
higher than that of the black hole itself. For if it were not, then the
physics of black hole decay (upon which Penrose's scenario depends)
would not work properly. Black hole decay is actually a dynamic sys-
tem which is the sum of the energy lost by Hawking radiation plus the
energy gained by absorption of local matter created by thermal fluctua-
tions due to the de Sitter Gibbons-Hawking temperature. The mere fact
of these thermal fluctuations suggests that the entropy of the cosmolog-
ical horizon is a real physical manifestation as opposed to an accounting
gimmick.
 On Non-Singular Space-times and the Beginning of the Universe 121

As a thought experiment, one can imagine what happens as more and

more external matter (over and above that added by external environ-
ment thermal fluctuations) is added to a black hole. Is there a maximum
size? As the black hole gets larger, its temperature drops, approaching
that of the cosmic background. In fact, at a certain point the black hole
itself becomes physically indistinguishable from a cosmic horizon and
can be identified as such. At no point does it become unreal. It therefore
seems unwarranted to embrace Penrose's latest (2009) position, and very
few physicists have been persuaded by Penrose's non-unitary brand of
quantum physics.
In sum, while Weyl curvature is the same between the two states that
Penrose wishes to say are identical, the entropy is not. It seems to us,
then, that the two states cannot be identical. Hence, we question CCC's
viability.
It seems, then, that cyclic models do not avert the conclusion that
time had a beginning. Thermodynamic objections to cyclic models,
'classical' anticipations of which were seen in Richard Tolman's early
work in the 1930s (Tolman, 1934), still seem to find purchase. Entropy
should increase from cycle to cycle, implying that (1) we should now
be in thermodynamic 'heat death' (which we are not), and (2) the maxi-
mum size (scale factor) of the universe should grow with each cycle. This
large scale 'classical' behaviour likely ends when (looking into the past)
the cycle size shrinks to Planckian physics. If entropy gain is denied,
then the empirical observation of a cosmological constant of sufficient
size to lead to an open-ended expansion rather than a future contraction
(this does seem to be the fate of our universe) also implies that there can
have been no previous cycle. If our future does include a return to a sin-
gular condition, then, as Banks asserts (recalling our earlier discussion
of BKL chaos), 'It seems silly to imagine that, even if (a future singu-
lar condition for the universe) is followed by a re-expansion, that one
would start that expansion with a low entropy initial state, or that one
had any control over the initial state at all' (pers. comm., 17 October
2008).

4.6. Reversing time's arrow

There is a longstanding debate in cosmological circles as to how to
interpret universe phases that would be entropy-reducing. There is no
necessity (in GTR) for the cosmological and thermodynamic arrows of
time to align with each another, although they seem to do so in the
present era of our universe. This debate has consequences for discussion
122 William Lane Craig and James D. Sinclair

of a cosmic origin. For example, in inflationary models, the BGV theo-

rem seems to bar a beginningless past. But suppose one takes a model
that features a contraction, a bounce through a near-singular state, and
then an expansion. Such a model faces fine-tuning objections such
as that discussed in section 4.1. Yet it is possible to (re)consider the
contracting phase as having a reversed arrow of time. The boundary
that formerly represented the 'bounce' will now bisect two symmetric,
expanding universes on either side.
On such a view space-time is extendible beyond the boundary for an
observer, looking backwards in time, from either side. Does this imply
past eternality? Cosmologists Anthony Aguirre and Steve Gratton think
so. They have employed the concept to create an 'eternal steady state
inflation model' (Aguirre and Gratton, 2002). To visualize their idea,
start with the standard foliation of de Sitter space as looking something
like an hourglass. The width of the glass represents the scale factor of the
universe. Vertical height (pointed upward) is the time. Thus the space
shrinks to a minimum radius and then rebounds into an expansion.
On this understanding, one cannot possibly have 'past eternal inflation'.
Borde, Guth and Vilenkin explain:

The intuitive reason why de Sitter inflation cannot be past-eternal

is that, in the full de Sitter space, exponential expansion is preceded
by exponential contraction. Such a contracting phase is not part of
standard inflationary models, and does not appear to be consistent
with the physics of inflation. If thermalized regions were able to form
all the way to past infinity in the contracting space-time, the whole
universe would have been thermalized before inflationary expansion
could begin. (Borde, Guth and Vilenkin, 2003, p. 1)

To obtain a different outcome, Aguirre and Gratton choose to inter-

pret de Sitter space according to a procedure first suggested by Erwin
Schrodinger in 1956. Utilizing 'elliptic de Sitter space', thev remap the
hourglass by splitting it in two along a line from the top left to the
bottom right. For all intents and purposes, these are two separate uni-
verses. Maulik Parikh explains how this looks to antipodal observers on
opposite sides of the divide:

The antipodal map, XI~ -XI, (where 'I' is a dimensional index)

changes the sign of the time coordinate of the embedding space, and
also that of the direction of time in de Sitter space. The resulting
quotient space, dS/Z2, is as a consequence not time-orientable: although
 On Non-Singular Space-times and tile Beginning of tile Universe 123

one can locally distinguish past and future, there is no global direction
of time. This fact clearly changes many standard notions about space
and time that we are accustomed to. For instance, it is impossible to
choose a Cauchy surface for elliptic de Sitter space that divides space-
time into a future and a past region. (Parikh et al., 2003, p. 6; our
emphasis)

Aguirre and Gratton explain the relationship between the two discon-
nected regions:

In essence, this construction partitions the full de Sitter space-time

into a self-consistent set of two noncommunicating (steady state)
universes. An observer in region I does not see anything in its past
light cone from an observer in region II because that other observer
cannot signal into its past, and vice-versa. Seen in this way the bound-
ary condition forbidding physical particles from following geodesics across
(the boundary) - into one universe is in no way strange or unreasonable,
as it follows directly from the forbidding of causalily violations in the other
universe. (One could similarly partition de Sitter space-time by any
non-timelike boundary B away from which time flows.) ...
Without the identification, the space-time manifold is time-
orientable in the mathematical sense that it is possible to contin-
uously divide non-spacelike vectors into two classes which can be
labeled 'future' and 'past.' In our construction these labels will only
correspond to the physical (arrow of time) in one of the two regions.
With the identification the space-time is still a manifold but is not
mathematically time-orientable. The physical (arrow of time) is, how-
ever, still well-defined and no physical observer will see it reverse.
(Aguirre and Gratton, 2002, p. 3; our emphasis)

As can be seen, neither region of the remapped space stands in a rela-

tion of earlier than to the other. There is no global definition of time.
As Parikh notes above, it is impossible to determine a future and past
region (when considering the whole hourglass). A matter of some impor-
tance is that there is no communication; no causalily by definition
from one region to the other. Thus one can say that there is no eter-
nal past that evolved into our present. So Aguirre and Gratton's use of
the term 'past eternal' seems to be idiosyncratic in that they construe
one's ability to draw 'complete' geodesics (meaning 'from one end of
the space-time to the other', even though no observer or causal chain
could traverse them) from end to end on a graph to be a sufficient
124 William Lane Craig and fames D. Sinclair

condition of 'eternal'. What one really has in this model are two separate
universes that trace their origins to a past boundary (the very boundary
that Borde, Guth and Vilenkin have demonstrated).
Aguirre and Gratton conclude: 'We suspect that a construction like
that proposed here may be necessary in any reasonable model for an
eternal universe that avoids a beginning of time' (Aguirre and Gratton,
2002, p. 6). We suggest, on the contrary, that if they are right, then a
beginning of time has been demonstrated. Rather than a past eternal
universe, one has a past finite multiverse. This model is a perfectly viable
example of a cosmological model that averts the singularity theorems
yet still 'begins to exist', despite a technical past extendibility through
the boundary.
We may resume here our discussion of cyclic models in section 4.5.
Martin Bojowald has recently argued that the bounce events of a cyclic
model actually represent a scenario similar to the Aguirre-Gratton 'dou-
ble Big Bang' described above. Measurements of a state of the universe
'after' a bounce cannot realistically be used to derive knowledge of
the state of the universe 'before' the bounce. Thus the model contains
'cosmic forgetfulness' as a salient feature. Similarly, an observer just
'before' a bounce would find that attempts at predicting the post-bounce
future suffer from cosmic forgetfulness as well. Bojowald interprets the
situation as follows:

The kind of cosmic forgetfulness realized in this model provides an

orientation of time, telling us not only which of the properties before
the Big Bang can be forgotten, but also what direction 'before the Big
Bang' is. An observer after the bounce would be unable to recon-
struct the full state before the bounce, but could easily predict the
future development toward larger volume. This arrow agrees with the
standard notion.
Now asking how an observer before the Big Bang would experience
the same situation, the answer is also clear: such an observer would be
unable to determine the precise state at larger values of <p (which is a
matter field used in the role of a clock) beyond the bounce, but could
easily extrapolate the state to smaller values of <p (the direction away
from the 'bounce'). The state at smaller values of <p can be predicted,
while the state at large values of <p is forgotten once the bounce is
penetrated. Since one cannot forget the future, such an observer must be
attributed a reversed arrow of time, pointing toward smaller <p. At the
bounce, two arrows would emerge pointing in opposite directions as far as rp
is concerned. In this sense, the model resembles [Aguirre-Gratton and
 On Non-Singular Space-times and the Beginning of the Universe 125

similar suggestions from two other modelling teams using different

methods] ...
Taking the simplest models of loop quantum cosmology at face
value is often seen as suggesting the big bang transition to be viewed
as a smooth bounce, as one further element not just in a long history
of the universe itself but also in a long history of bouncing cosmo-
logical models. Some indications, however, suggest otherwise. The
bloomy scenario of loop quantum cosmology may well be this: a uni-
verse whose time-reversed prehistory we cannot access but which we
grasp in the form of initial conditions it provides for our accessible
part; a pseudo-beginning; an orphan universe, shown the rear-end by
whatever preceded (and possibly created) it. (Bojowald, 2009, p. 15;
our emphasis)
Recall we had earlier discussed the breakdown of models featuring a
classical background plus quantum fluctuations. A hypothetical resolu-
tion is to go to a full quantum gravity model. Would this produce past
eternality? Bojowald suggests it produces a reversed arrow of time on
the other side of the boundary. But, as the moments on one side of the
boundary are in no sense earlier than the moments on the other side,
there would be no infinite past.
H. D. Zeh, in the same volume in which Bojowald's essay appears,
gives independent justification for accepting the interpretation of the
boundary as an origin as opposed to a 'bounce'. The Wheeler-DeWitt
(WOW) wave function (upon which loop quantum gravity is based)
would be symmetric on both sides of a 'bounce'. This suggests that what
one really has is a 'double Big Bang' coming from the same creation
event. Zeh also suggests that the semi-classical approximation which
defines time in a WOW approach cannot, with validity, be extended
'through' a bounce. Lastly, the low entropy initial state is common for
all branches of the wave function.
According to loop quantum cosmology, the Wheeler-DeWitt equa-
tion (in this theory replaced by a difference equation with respect
to (scale factor) a) can be continued through a= 0 to negative val-
ues of a. The configuration space of three-geometries is in this way
duplicated by letting the volume measure assume negative values
(turning space 'inside out' while going through a= 0). Since the
Hamiltonian does not depend on the newly invented sign of a, how-
ever, the Wheeler-DeWitt wave function must be expected to be
symmetric under this parity transformation, too. Its continuation
would then have to be interpreted as an added superposition of
126 William Lane Craig and fames D. Sinclair

other physically expanding universes. Since the WKB times, which

represent classical time, can not be continued through a= 0, the inter-
pretation of negative values of a as representing pre-big-bang times
is highly questionable. The fundamental arrow, including its conse-
quence of decoherence outside the validity of a WKB approximation,
must depend on some low entropy 'initial' condition in a for all other
('spacelike') degrees of freedom that occur as physical arguments of
the Wheeler-DeWitt wave function. It would be hard to understand
how the low entropy state at a = 0 could have been 'preceded' by an
even lower entropy at a < 0 in order to avoid a reversal of the ther-
modynamical arrow in the classical picture of an oscillating universe.
(Zeh, 2009, pp. 11-12; author's emphasis)

On this view, a loop quantum approach predicts an origin out of which a

multiverse arises. Continuation through the 'bounce' boundary actually
represents other physically expanding universes. Zeh and Claus Kiefer
arrive at a similar conclusion arguing directly from the Wheeler-DeWitt
equation within the formalism of quantum geometrodynamics.

One might wonder what happens in the case of models which

classically describe bouncing cosmologies: the Universe would then
undergo many, perhaps infinite, cycles of expansion and recollapse.
What would happen with the entropy in these cases? If the entropy
were indeed correlated with the scale factor, as the scenario discussed
above suggests, the arrow of time would not continue through a
turning point. The bouncing models would thus make no sense in
quantum cosmology; one would only have branches of the wave
function in which the arrow would point from small to large universe
and where time would end when approaching a classical turning
point. (Kiefer, 2009, p. 9)
The different quasiclassical branches of the wave function which are con-
nected by 'quantum scattering' at the turning point should rather be
interpreted as all representing different expanding universes, which dis-
appear at the turning point by means of destructive interference
(similar to their coming into existence as separate Everett branches
from a symmetric initial state at the big bang). (Kiefer and Zeh, 1994,
p. 4152, our emphasis)

Here, Wheeler-DeWitt physics (with a boundary condition of a vanish-

ing wavefunction) seems to predict the creation ex nihilo of an Everett
multiverse as opposed to a recollapse.
 On Non-Singular Space-times and the Beginning of the Universe 127

Recalling Penrose's model, suppose that we grant his reversal in

entropy as one looks to the universe's far future. One may just as well
interpret his model as having a reversed arrow of time when looking to
the 'past' of a Big Bang singularity, so as to obey a second law of ther-
modynamics. As such, it would display precisely the behaviour cited
above by the Aguirre-Gratton, Bojowald and Kiefer-Zeh approaches.
Physically, the question is whether, in our universe, various arrows
of time (thermodynamic, cosmological, electromagnetic, psychological)
align. Philosophically, the question is whether there is an underlying
metaphysical time which the physical times manifest in various ways. 25
Extendibility, by this understanding, would be a technical artefact rather
than an indication of past eternality.
Positing the beginning of classical time at a past boundary may have
a dramatic implication first pointed out by philosopher Quentin Smith.
Smith's comments were aimed at Hawking's interpretation of the imagi-
nary time featured in the Hartle-Hawking quantum gravity cosmogony.
Hawking was attempting to posit an 'imaginary' time phase for the
universe that transitioned into a later, 'normal' time phase. Says Smith,

such an interpretation is implicitly logically self-contradictory (in fea-

turing a 'time' axis that starts off just like a space dimension, but
then transforms to its familiar properties). The problem appears in
the statement that the four-dimensional space joins on to the real
(Lorentzian) space-time 'once' (i.e., after) the quantum smearing
effects subside: 'The question then arises as to the geometry of the
four-dimensional space which has to somehow smoothly join onto
the more familiar space-time once the smearing effects subside.' If the
four-dimensional space does not possess a real time value, how can it
stand in relation to the four dimensional space-time of being earlier
than it? If the four dimensional space is not in real (Lorentzian) time,
then it is not really earlier than, later than, or simultaneous with the
four dimensional space-time manifold. Accordingly, it is false that
the 4-sphere joins onto the familiar space-time once (i.e., after in real
time) the quantum effects dissipate. (1993, p. 318)

Smith later showed (1997, 2000) that these two (actually three; there is a
transition region) 'stages' of the H-H model could coherently coexist in
the same model only if they stood in a more primitive 'topological' rela-
tion, 26 rather than a temporal one. Thus, if two stages of a cosmogony
did not stand on one and the same time axis, then it would be false
to ascribe to them an earlier than or later than relation, even though
128 William Lane Craig and fames D. Sinclair

extendibility existed between the relevant universe phases. Recall, for

example, that Penrose denies a real time value to his boundary as a phys-
ical necessity. When Penrose says, 'the remote future of one phase of the
universe becomes the Big Bang of the next', his statement is arguably
incoherent in the context of his model.
It seems to us Smith's insight applies in general to cosmological mod-
els whether they are classical, general relativistic models (which thus
include singularities) or quantum gravity models (which include 'strong
quantum regions'). The 'first' phase (which, alternatively, could be a
'transition' boundary between different classical or semi-classical space-
times) by definition does not possess a time value. As Smith points out,
because a model phase that lacks a real time value is not before, simul-
taneous with, or after any other universe state (with respect to a phase
that does have real time values) it is not correct to say that the universe
evolved from some state 'at' the boundary into some state with a real
time value.
Very similar language is used by loop quantum theorists themselves. 27
On Smith's view (similarly, Ashtekar et al.), it seems correct to view this
era as a primitive manifold with topological relations only between the
'cycles'. The cycles in the 'earliest' era seem to be a set of disconnected
space-times which lack temporal ordering. One could just as well razor28
these out. One is left with a space-time with a finite past timeline and no
first moment, which grows into a macroscopic universe similar to ours.
Thus it seems 'singularities' are resolved, or not. If so, one has a pre-
Big Bang model and faces issues such as a disappearing arrow of time, a
reversed arrow of time, and/or entropy problems. If not (meaning time
'dissolves'), then there are no 'pre-Big Bang' phases, since those phases
are incommensurate with respect to our FRW-like universe.
One recalls Pitts's insistence on a first moment as a necessary con-
dition of a beginning of time (2008, p. 677). Precisely the opposite
situation may obtain, generically, in models of the types discussed,
despite the technical extendibility of geodesics through the boundary.
We suggest that this situation represents a creation ex nihilo just as surely
as the Big Bang singularity in standard cosmology itself represents an
absolute beginning. 29
In summary, then, it seems to us that a strong case for a beginning
of the universe can be made with respect to pre-Big Bang cosmogo-
nies (only some of which involve non-singular models), so that (9) is
dispensable so far as premise (2) of the kalam cosmological argument
is concerned, regardless of how critical it might be to Pitts's imagined
singularity argument.
 On Non-Singular Space-times and the Beginning of the Universe 129

5. The Beginning of the Universe

But what, we may ask, is the nature of the universe's beginning if it is
not a GTR-type singularity? The most widely discussed framework for
addressing this problem is the Vilenkin 'tunneling from nothing' or the
Hartle-Hawking 'no boundary' model. 30 Vilenkin observes,

Many people suspected that in order to understand what actually

happened in the beginning, we should treat the universe quantum-
mechanically and describe it by a wave function rather than by a
classical space-time. This quantum approach to cosmology was initi-
ated by DeWitt and Misner, and after a somewhat slow start received
wide recognition in the last two decades or so. The picture that has
emerged from this line of development is that a small closed universe can
spontaneously nucleate out of nothing, where by 'nothing' I mean a state
with no classical space and time. The cosmological wave function can
be used to calculate the probability distribution for the initial config-
urations of the nucleating universes. Once the universe nucleates, it is
expected to go through a period of inflation, driven by the energy of
a false vacuum. The vacuum energy is eventually thermalized, infla-
tion ends, and from then on the universe follows the standard hot
cosmological scenario. (2002, p. 2; our emphasis)

Hartle and Hawking also claim that their cosmogonic model can be
interpreted in such a way that on that interpretation the universe came
into being out of 'nothing':

One can interpret the functional integral over all compact four-
geometries bounded by a given three-geometry as giving the ampli-
tude for that three-geometry to arise from a zero three geometry; that
is, a single point. In other words, the ground state is the probability
for the Universe to appear from nothing. (1983, p. 2961)

On these models the universe clearly has a beginning; but what does
it mean to 'nucleate out of nothing?' It seems to us that there are four
possible ways of interpreting this notion:

1. The 'initial' null topology31 represents literal nothingness. One could

imagine that 'nothing' as employed by Hartle-Hawking has exactly
the relevant meaning for the kalam cosmological argument, namely,
the assertion of nonbeing. If this is what is meant here, then models
of this sort feature an unambiguous creation ex nihilo. The second
130 William Lane Craig and James D. Sinclair

premise of the kalam cosmological argument is upheld, and the

debate shifts to the first premise.
2. The 'initial' null topology is to be construed purely instrumentally.
Hawking himself seems to give good grounds for treating Hartle and
his proposal as a FAPP (i.e. 'for all practical purposes') approach only.
The H-H formalism considers a superposition of all possible quan-
tum geometries, with our universe emerging as the most probable,
and uses Richard Feynman's 'sum over histories' approach to quan-
tum mechanics to predict universe observables. This formalism treats
time, in the quantum gravity era, as a fourth spatial dimension. But
in his collaboration with Roger Penrose, The Nature of Space and Time,
Hawking employs the same mathematical approach (analytic con-
tinuation) to describe pair production of electron/positron pairs in
a strong electric field (1996, p. 54). This is a standard mathematical
technique sometimes used when complex analytic functions are bet-
ter behaved in a certain domain than their real counterparts. It does
not imply ontological commitment to the alternative description,
however. It seems to us that given the unintelligibility of the 'imagi-
nary time' region in these models, it is most reasonable to treat this
regime as a useful fiction. Thus, the model serves only to reinforce
the second premise of the kalam cosmological argument.
3. The 'initial' null topology is literally something. Attempts to take
the imaginary time interpretation realistically can run into problems
similar to those Smith pointed out (1993, 1997). In fact, more recent
work using computer simulations of the evolution of space-time sug-
gests that classical space-time cannot have emerged from a Euclidean
4-space such as is envisioned in the H-H model. Jerzy Jurkiewicz,
Renate Loll and Jan Ambjorn explain:

In our search for loopholes and loose ends in the Euclidean

approach, we finally hit on the crucial idea, the one ingredient
absolutely necessary to make the stir fry come out right: the uni-
verse must encode what physicists call causality. Causality means
that empty space-time has a structure that allows us to distinguish
unambiguously between cause and effect. It is an integral part of
the classical theories of special and GTR.
Euclidean quantum gravity does not build in a notion of causal-
ity. The term 'Euclidean' indicates that space and time are treated
equally. The universes that enter the Euclidean superposition have
four spatial directions instead of the usual one of time and three
of space. Because Euclidean universes have no distinct notion of
 On Non-Singular Space-times and the Beginning of the Universe 131

time, they have no structure to put events into a specific order;

people living in these universes would not have the words 'cause'
or 'effect' in their vocabulary. Hawking and others taking this
approach have said that 'time is imaginary,' in both a mathemati-
cal sense and a colloquial one. Their hope was that causality would
emerge as a large-scale property from microscopic quantum fluc-
tuations that individually carry no imprint of a causal structure.
But the computer simulations dashed that hope. (2008, p. 3)

In their Causal Dynamical Triangulations approach, causality is demon-

strated to be a necessary fUndamental feature for a cosmogony yielding
our observable universe. Lee Smolin explains the implication:

Some of the most widely believed ideas about quantum gravity are in
fact wrong. For example, Stephen Hawking and others used to argue
that causal structure was inessential, and that calculations could be
done in quantum gravity by ignoring the differences between time
and space- differences that exist even in relativity theory- and treat-
ing time as if it were another dimension of space (imaginary time) ....
Ambjorn and Lolle's results show that this idea is wrong ....
In particular it was shown ... that if no restriction respecting
causality is put in, no classical space-time geometry emerges ....
One of the rules that Loll and Ambjorn impose is that each quan-
tum space-time has to be seen as a sequence of possible spaces that
succeed one another, like the ticks of a universal clock. The time coor-
dinate, it is argued, is arbitrary, as in GTR, but the fact that the history
of the world can be seen as a succession of geometries that succeed
one another is not. (2006, pp. 242-3)

Interestingly, the trajectory of quantum gravity research is tending to

correct the 'imaginary time' approach and in so doing to affirm the
importance of consistent earlier than/later than relations and fundamen-
tal causality.

4. The 'initial' null topology is misconceived. }. Richard Gott and Li

Xin Li have criticized Vilenkin (and Hartle-Hawking's) creation ex
nihilo approach on two grounds. First, transitions in QM are always
between allowed classical states. But Vilenkin and Hartle-Hawking's
approach has a transition from a classically forbidden region to a
classically allowed region. Second, the Vilenkin and Hartle-Hawking
approaches should contain realistic energy fields (something closer
132 William Lane Craig and James D. Sinclair

to what we actually see in nature). If they did, then Heisenberg's

uncertainty principle would require that the initial state of their
models have a finite and nonzero energy. If that is the case, then
semi-classical quantum models like Vilenkin's and Hartle-Hawking's
actually start in a classically allowed, metastable state, rather than
'nothing'. Gott and Li elaborate:

The problem with this model (Vilenkin and Hartle-Hawking) is

that it ignores the 'zero-point energy.' If there is a conformal scalar
field <j>, then the 'energy' levels should be En= n + 1/2. Even for
n = 0 there is a 'zero-point-energy.' The potential makes the sys-
tem behave like a harmonic oscillator in the potential well near
a= 0. A harmonic oscillator cannot sit at the bottom of the poten-
tial well - the uncertainty principle would not allow it. There
must be some zero-point-energy and the particle must have some
momentum, as it oscillates within the potential well when the
field <1> is included. Thus, when the 'zero point-energy' is consid-
ered, we see that the initial state is not a point but a tiny oscillating
(0 <a < a 1 ) big bang universe, that oscillates between big bangs
and big crunches (though the singularities at the big bangs and big
crunches might be smeared by quantum effects). This is the initial
classical state from which the tunneling occurs. It is metastable,
so this oscillating universe could not have existed forever: after a finite
half-life, it is likely to decay. It reaches maximum radius a 1, and then
tunnels to a classical de Sitter state at minimum radius a 2 where
az < ao. (1998, p. 38; our emphasis)

Thus, we seem to have the same sort of situation that we encountered

with respect to the Emergent model with its associated metastable ESS.
The universe cannot be past eternal because the initial metastable state
can have had only a finite lifetime. The Gott-Li interpretation seems
to be a reasonable option for a realist interpretation of these models. 32
It employs known, meaningful interpretations of physical phenomena
from 'classical' quantum theory and extends them to the quantum grav-
ity models. One avoids the problems associated with the novelty of
asserting a zero-energy condition for the initial state (denied by the
Heisenberg uncertainty principle), the novelty of asserting a quantum
transition from a forbidden to a classically allowed state (normal quan-
tum theory includes only transitions over or through forbidden regions
from one allowed state to another), and it is consistent with more
realistic energy fields. This alternative is, of course, also consistent with
the second premise of the kalam cosmological argument.
 On Non-Singular Space-times and the Beginning of the Universe 133

6. Concluding Remarks
In sum, we have seen no good reason to deny that space-times
featuring an initial cosmological singularity have a beginning; but even
more fundamentally we have seen that the claim that the universe
began to exist does not depend upon the singular nature of space-
time. Moreover, our survey of contemporary cosmological thinking
reveals significant evidence that non-singular models may themselves
to be incompatible with an infinite past, and therefore singularity-
free quantum gravitational approaches can be quite supportive of the
kalam cosmological argument's second premise. So when Pitts worries,
'It is not implausible that some singularity-wielding theistic apolo-
gists will be tempted to resist scientific progress in the form of a
new quantum theory of gravity in order to maintain an apologetic
strategy in which they have invested' (2008, p. 19), one cannot help
but wonder who is the object of so uncharitable an allegation. For
we have seen that the second premise of the kalam cosmological
argument enjoys considerable empirical support even if space-time is
non-singular. 33

Notes
1. Writing in the Cambridge Companion to Atheism, Quentin Smith reports, 'a
count of the articles in the philosophy journals shows that more articles
have been published about ... the Kalam argument than have been published
about any other philosopher's contemporary formulation of an argument
for God's existence' (2007, p. 183). For the history of the argument, see
Craig, 1980; for literature and a recent defence of the argument, see Craig
and Sinclair, 2009.
2. In places Pitts recognizes this: 'The truth of the second premise [of the kalam
cosmological argument], or rather, the source of warrant for the second
premise if it is true, is the key question' (2008, p. 687; cf. p. 676).
3. So to argue (while avoiding the fallacy of composition) would be to chal-
lenge the truth, not of (2), but of (1) of the kalam cosmological argument.
Sometimes Pitts does conflate what are really challenges to (1) with puta-
tive objections to (2). For example, his mention of four-dimensionalism
as rendering space-time 'self-explanatory' (2008, p. 688) is a case in point.
Again, Pitts's Cosmic Destroyer argument is really an objection to (1): if
the coming-into-being of the universe must have a cause, then so must its
ceasing-to-be.
4. One cannot, in passing, help but wonder how imposing topological con-
straints on the doctrine of creation in time, as Pitts would do, is modally
enriching, as he claims, since then there are possible worlds (like, appar-
ently, the actual world) in which the universe lacks a first moment
and so cannot be said on Pitts's account to have been created in time
by God.
134 William Lane Craig and James D. Sinclair

5. It is noteworthy that Earman, whose objection Pitts echoes, does not in the
end deny the acceptability of our explication of 'begins to exist' but merely
advises that on such a reading premise (1) 'is not an obvious "metaphysi-
cal truth"'; in particular (1) does not follow from the principle that every
event has a cause, which is satisfied in general relativistic cosmology (1995,
p. 208).
6. NB that both Pitts's argument and our response are limited to a faux universe
where GTR is the fundamental theory of gravity.
7. We also note that there is no good reason to think that metric convention-
alism holds in the actual world. As has been argued elsewhere (Craig, 2001,
ch. 2), metric conventionalism is an implausible thesis for which no good
arguments exist.
8. It is at least worth noting that Pitts, not content with the promise of current
quantum cosmology, goes so far as to champion the rights of unborn and,
indeed, unconceived hypotheses as grounds for scepticism about (9): 'The rel-
evant set of competitors for GTR includes the set of theories that agree with
GTR on all experiments to date, whether already entertained on Earth or
not. This set might be infinite, might well be large, likely contains several
members, and almost certainly has at least one member, a quantum the-
ory of gravity. The set most likely has at least one member that resolves the
singularities of GTR. Thus, in the strong field regime it is not at all clear
why one should take GTR seriously' (2008, p. 696). Pitts's final sentence is a
non sequitur. As Timothy McGrew has remarked in personal conversation, in
order to mount a significant challenge to GTR it is not enough to show that it
is probable that an unknown theory of quantum gravity exists which shares
GTR's empirical adequacy to date and resolves the singularities of GTR. Pitts
also needs to show that this unknown competing theory has, by comparison
to GTR, a non-negligible prior probability. For there may be other overriding
factors that are relevant to its probability, for example, its want of simplicity
or elegance. Since these theories are unconceived, it is hard to see how Pitts
can know such a thing. In any case, as we remark in the text, the important
issue is not whether that unknown theory resolves the singularities of GTR
but whether the physical universe as described by the theory is infinitely
extrapolable to the past. What one wants from Pitts is some evidence or
argument that there is an unknown theory of quantum gravity meeting all
the desiderata.
9. Our intention is to provide a complementary update to the discussion in
Craig and Sinclair, 2009. Thus, for example, the material in the original essay
had significant content regarding string-based cosmogonies which will not
be repeated here.
10. It also permits us to regard records of the past, such as our own memories, as
being relics of a true reality as opposed to illusory. We will hold that starlight
that appears in our telescopes in the present actually came from a distant star
in a real past. So called 'top-down' interpretations of the Feynman path inte-
gral approach to quantum gravity (Hawking, 2003), or 'decoherent histories'
(see Hartle, 1998; Dowker and Kent, 1996), are examples of quantum gravity
approaches that would deny the objective reality of a unique past that evolved
into the present. This doesn't mean, necessarily that these approaches must
be rejected. Rather, we reject the interpretation of the formalism. Hawking, in
 On Non-Singular Space-times and the Beginning of the Universe 135

the above reference, says about his own theory that 'One can interpret this in
the bottom up picture as the spontaneous creation of an inflating universe
from nothing.' On this view, we have precisely the creation ex nihilo that
lends evidence towards kalam's second premise.
11. This implication was affirmed by Vilenkin in a personal communication on
4 March 2004. Note that this conclusion would apply to the well-known
Ekpyrotic cosmogony of Paul Steinhardt and Neil Turok mentioned by Mon-
ton. Their model has a genuine past boundary and thus accords with the
second premise of the ka/am cosmological argument. See Craig and Sinclair,
2009 for a more detailed explanation.
12. See also the collection of papers by Brett Mcinnis (2007a, 2007b, 2008).
Mcinnis provides an alternative argument for why inflation must have an
ultimate origin. He suggests that in string theory (for purely geometric rea-
sons) the arrow of time can only originate in a topological creation event
'from nothing' such as in the Vilenkin proposal (see section 5 of our paper).
Subsequent bubble universes, whether embedded or distinct disconnected
space-times (see Carroll and Chen, 2004), will have an arrow of time only
if they inherit it from their mother. Thus, given our observed arrow of time,
we know we have an ultimate 'mother' that came into being via creation ex
nih i/o.
13. To illustrate, one of the authors once participated in a similar exercise while
working in the defence industry. The author noted that a particular air-to-air
combat model explicitly included missile flyouts against air targets. A much
more elegant (and efficient) model was built by deriving a set of equations
whereby time and distance intervals could be dropped from the formalism.
The combat could be described by an impact sequence of missile salvos that
does not reference the time or space intervals between salvo arrivals. Now
would the mere fact of the existence of this model (and its superiority in
terms of simplicity, explanatory scope etc.) imply that time and space inter-
vals do not exist? This would be a ludicrous conclusion, as any experience as
a combat aviator would demonstrate. Air superiority is maintained by hav-
ing the superior range missile (a distance interval), or shortest flight (time
interval) such that one's missile gets there first. It is precisely the intervals
that produce salvo impact order and hence air superiority. Individual life
and death, even the rise and fall of empires, ride on this conclusion. One is
reminded of Einstein's advice that we should make our models as simple as
possible, but not too simple.
14. See Aguirre, 2007 for a detailed attempt to circumvent BGV in favour of a
time-reversed or 'emergent' non-singular cosmogony. These suggestions of
Aguirre's are considered in sections 4.4 and 4.6.
15. See, e.g., Earman and Mosterin, 1999, for a related argument.
16. See also Damour and Henneaux (2000, p. 923), whose results update B-K-L
to consider quantum string cosmogonies: 'our findings suggest that the
spatial inhomogeneity continuously increases toward a singularity, as all
quasi-uniform patches of space get broken up into smaller and smaller ones
by the chaotic oscillatory evolution. In other words, the space-time structure
tends to develop a kind of "turbulence".'
17. Ellis does indicate that a finite past is an acceptable possible form for an
Emergent model (Ellis and Maartens, 2004, section V, p. 5).
136 William Lane Craig and fames D. Sinclair

18. We take as real the scale factor behaviour, while seeing the function of scalar
field vs. time as instrumental. Otherwise, the scalar field could itself perform
the function of a clock.
19. The quantum state must not be infinitely squeezed to be semi-classical (see
section 4.5); the centre equilibrium point only exists if the semi-classical
assumption holds.
20. See Craig and Sinclair (2009) for an extensive discussion of another asymp-
totically static cosmogony, the string-based pre-Big Bang Inflation model of
Maurizio Gasperini and Gabrielle Veneziano. See Kaloper eta!. (1998) for a
similar criticism based on fluctuation of the dilaton field.
21. Also see Tavakol and Carneiro (2009). The argument is that the past is infi-
nite. But a suggested method of leaving the stabilized ESS so as to allow an
inflationary expansion invokes a quantum tunnelling from a local minimum
in potential (stabilized ESS) to a global minimum (open-ended inflation).
But this local minimum is a metastable state with a finite lifetime. Thus the
model is not past infinite.
22. Examples in the recent literature of asymptotically static non-singular cos-
mogonies include, e.g., Falciano and Pinto-Neto, 2009; Seahra and Bohmer,
2009; and Barrow and Tsagas, 2009. The first compares the predictions of
such models with quantum scalar perturbations observed in the cosmic back-
ground radiation via the Wilkinson Microwave Anisotropy Probe and casts
doubt on their ultimate viability. The second and third investigate the via-
bility of a past eternal ESS model. The Seahra and Bohmer paper shows that
an ESS is unstable to some type of perturbation for generic quantum grav-
ity theories and generic equation of state (that is, the type of matter/energy
residing in the universe) for f(R) gravity theories. The Barrow and Tsagas
paper exploits a stability region for ESS for 'phantom' or 'ghost' matter
universes. See Craig and Sinclair, 2009 for a discussion of some of these
'phantom bounce' cosmogonies.
23. Banks complains, 'I have a problem with ALL cyclic cosmogonies .... The
collapsing phase of these models always has a time dependent Hamiltonian
for the quantum field fluctuations around the classical background. Further-
more the classical backgrounds are becoming singular. This means that the
field theories will be excited to highrr and higher energy states (define energy
in some adiabatic fashion during the era when the cosmology is still fairly
slowly varying, and use this to classify the states, even though it is not con-
served). High energy states in field theory have the ergodic property - they
thermalize rapidly, in the sense that the system explores all of its states. Willy
Fischler and I proposed that in this situation you would again tend to maxi-
mize the entropy. We called this a Black Crunch and suggested the equation
of state of matter would again tend toward p = p. It seems sil!y to imagine
that, even if this is followed by a re-expansion, that one would start that
expansion with a low entropy initial state, or that one had any control over
the initial state at all' (pers. comm., 12 October 2008). An important caveat
is that a general description of entropy in the presence of gravity (either in
GTR or the still undiscovered Quantum Gravity) has never been achieved.
But there are some special cases that are thought reliable by most cosmol-
ogists (see Carroll et al., 2007, slides 11 and 12), such as the entropy of a
thermal gas, black holes and cosmic horizons (this last being related to a
 On Non-Singular Space-times and the Beginning of the Universe 13 7

positive cosmological constant). These three cases are thought to represent

the dominant entropy contributors to the universe for three different eras:
the early universe (thermal gas), the present (black holes) and the far future
(a 'pure' de Sitter-like cosmic horizon). If these special cases are accurate,
they demonstrate that entropy increases as the universe ages, thus providing
some proof of a generalized second law of thermodynamics.
24. See also Coule (2008), where a suggestion is made that maximum entropy
is capped by the holographic bound (cosmic horizon entropy for a de Sitter
universe), and that the entropy of a bounce violates this restriction.
25. A philosopher sympathetic with a tensed theory of time would have an
additional rationale for believing that such models really represent a multi-
verse propagating from a common origin. There would be a reason why the
cosmological, thermodynamic, electromagnetic and psychological arrows of
time would all align; they are physical manifestations of the same underlying
temporal becoming of metaphysical time.
26. Note that Smith and Pitts are interested in different topological features
of space-time. Pitts's concern is whether the universe has a first moment.
Smith's concern is whether the topology of a space-time manifold (e.g.
unbroken sheet, torus, figure eight etc.) is more fundamental than its
metrical aspects.
27. '[f]he question of whether the universe had a beginning at a finite time is
now "transcended". At first, the answer seems to be "no" in the sense that
the quantum evolution does not stop at the Big Bang. However, since space-
time geometry "dissolves" near the big-bang, there is no longer a notion of
time, or of "before" or nafter" in the familiar sense. Therefore, strictly, the ques-
tion is no longer meaningful. The paradigm has changed and meaningful
questions must now be phrased differently, without using notions tied to
classical space-times' (Ashtekar, Bojowald and Lewandowski, 2003, p. 263;
our emphasis).
28. 'And why would we not apply Occam's razor to the pre-history?' (Bojowald,
2009, p. 16)
29. A creation ex nihilo makes more sense than positing time's emergence 'from'
a pure quantum state. Augustine seems to have had a similar insight 1600
years ago: 'We can correctly say, "There was a time when Rome did not exist:
there was a time when Jerusalem, or Abraham, or man, or anything of this
kind did not exist". We can in fact say, "There was a time when the world
did not exist", if it is true that the world was created not at the beginning of
time, but some time after. But to say, "There was a time when time did not
exist", is as nonsensical as to say, "There was a man when no man existed",
or, "This world existed when the world was not"' (1984, City of God, Bk XII,
ch. 16).
30. Also see Mcinnis (2007a, 2007b, 2008) for a string-based model of this type.
31. 'Creation of a universe from nothing ... is a transition from the null topolog-
ical sector containing no universes at all to the sector with one universe of
topology S3 ' (Vilenkin, 1994, p. 23).
32. In Craig and Sinclair, 2009, we profile Gott and Li's own attempt at solv-
ing cosmology's origins problem based on the postulate of closed timelike
curves (a time machine). Unfortunately, they can avoid Stephen Hawk-
ing's Chronology Protection Conjecture only by asserting an 'initial' cosmic
138 William Lane Craig and James D. Sinclair

vacuum which is measure zero with respect to all possible configurations.

In other words, their model is infinitely fine-tuned.
33. We thank Martin Bojowald, Thomas Banks, George Ellis, Alexander Vilenkin,
Donald Page, Quentin Smith, Timothy McGrew, Rudiger Vaas and Christian
Bohmer for useful discussions and comments.

References
Aguirre, Anthony (2007) 'Eternal Inflation Past and Future', in R. Vaas (ed.),
Beyond the Big Bang (Heidelberg: Springer Verlag); preprint: http:/ /arxiv.org/abs/
0712.0571.
Aguirre, Anthony, and Steve Gratton (2002) 'Steady State Eternal Inflation', PIJys.
Rev. D 65, 083507; preprint: http:/ /arxiv.org/abs/astro-ph/0111191 v2.
Al-Ghazali (1962) Kitab ai-Iqtisad fi'I-Iqtiqad (Ankara: University of Ankara Press).
Al-Ghazali (1963) Tahafut ai-Falasifah. Trans. S. A. Kamali (Lahore: Pakistan
Philosophical Congress).
Ashtekar, A., M. Bojowald and J. Lewandowski (2003) 'Mathematical Structure of
Loop Quantum Cosmology', Adv.Theor.MatiJ.Phys. 7: 233-68; preprint: http://
arxiv.org/abs/ gr-qc/030407 4.
Ashtekar, A., Tomasz Pawlowski and Parampreet Singh (2006) 'Quantum Nature
of the Big Bang', Phys. Rev. D 74, 084003; preprint: http://arxiv.org/abs/gr-qc/
0602086v2.
Augustine (1984) City of God (London: Penguin Books).
Asimov, Isaac (1989) 'The Relativity of Wrong', The Skeptical Inquirer
14.1 (Fall); available online at http://chem.tufts.edu/AnswerslnScience/
RelativityofWrong.htm.
Banks, T., and W. Fischler (2002) 'Black Crunch'; preprint: http://arxiv.org/abs/
hep-th/0212113vl.
Barrow,]., and M. Dabrowski (1995) 'Oscillating Universes', Monthly Notices of the
Royal Astronomical Society 275: 850--62.
Barrow, ]., and C. Tsagas (2009) 'On the Stability of Static Ghost Cosmologies';
available online at http://arxiv.org/abs/0904.1340v1 (21 May 2009).
Belinsky, V. A., I. M. Khalatnikov and E. M. Lifshitz (1970) 'Oscillatory Approach
to a Singular Point in the Relativistic Cosmology', Advances in Physics 19:
525-73.
Bojowald, M. (2005) 'Original Questions', Nature 436: 920--1.
Bojowald, M. (2006) 'Universe Scenarios from Loop Quantum Gravity', Annalen
Phys. 15: 326-41; preprint: http://arxiv.org/abs/astro-ph/0511557vl.
Bojowald, M. (2007) 'Large Scale Effective Theory for Cosmological Bounces',
Phys. Rev. D 74, 081301; preprint: http://arxiv.org/abs/gr-qc/0608100v2.
Bojowald, M. (2009) 'A Momentous Arrow of Time', in L. Mersini-Houghton (ed.),
The Arrow of Time (Heidelberg: Springer Verlag); preprint http:/ /arxiv.org/abs/
0910.3200.
Bojowald, M., R. Maartens and P. Singh (2004) 'Loop Quantum Gravity and the
Cyclic Universe', Pllys. Rev. D 70, 083517; preprint: http://arxiv.org/abs/hep-th/
0407115.
Bojowald, M., G. Date and G. Hossain (2004) 'The Bianchi IX Model in Loop
Quantum Cosmology', Class. Quant. Grav. 21, 3541; preprint: http://arxiv.org/
abs/gr-qc/0404039.
 On Non-Singular Space-times and the Beginning of the Universe 139

Bojowald, M., and R. Tavakol (2008a) 'Recollapsing Quantum Cosmologies and

the Question of Entropy', Phys. Rev. D 78, 023515; preprint: http://arxiv.org/
abs/0803.4484v1.
Bojowald, M., and R. Tavakol (2008b) 'Loop Quantum Cosmology: Effective
Theories and Oscillating Universes', in R. Vaas (ed.), Beyond the Big Bang
(Heidelberg: Springer Verlag); preprint: http://arxiv.org/abs/0802.4274vl.
Borde, A., A. Guth and A. Vilenkin (2003) 'Inflationary Spacetimes Are not Past-
Complete', Physical Review Letters 90, 151301; preprint: http://arxiv.org/abs/
gr-qc/0110012.
Carroll, Sean (2008) 'What If Time Really Exists?', FXQI essay contest on the
nature of time, available online at www.fqxi.org/community/forum/topic/318.
Carroll, Sean et al. (2007) 'Why Is the Past Different from the Future: The
Origin of the Universe and the Arrow of Time', available online at http://
preposterousuniverse.com/talks/time-colloq-07/.
Carroll, Sean, and Jennifer Chen (2004) 'Spontaneous Inflation and the Origin of
the Arrow of Time'; preprint: http://arxiv.org/abs/hep-th/0410270vl.
Coleman, S., and F. De Luccia (1980) 'Gravitational Effects on and of Vacuum
Decay', Phys. Rev. D 21, 3305.
Copan, P., and W. L. Craig (2004) Creation out of Nothing (Grand Rapids, MI:
Baker).
Coule, D. H. (2008), 'Holography Constrains Quantum Bounce'; preprint: http://
arxiv.org/abs/0802.186 7.
Craig, W. L. (2006) 'Naturalism and Cosmology', in A. Corradini, S. Galvan and
E. J. Lowe (eds), Analytic Philosophy without Naturalism (Routledge Studies in
Contemporary Philosophy; New York: Routledge), pp. 97-133.
Craig, W. L. (1980) The Cosmological Argument from Plato to Leibniz (London:
Macmillan & Co.).
Craig, W. L. (2001) Time and the Metaphysics of Relativity (Philosophical Studies
Series, 84; Dordrecht: Kluwer Academic Publishers).
Craig, W. L., and J. Sinclair (2009) 'The Kalam Cosmological Argument', in W. L.
Craig and J.P. Moreland (eds), Blackwell Companion to Natural Theology (Oxford:
Blackwell), pp. 101-201.
Damour, T., and M. Henneaux (2000) 'Chaos in Superstring Cosmology', Physical
Review Letters 85: 920-3; preprint: http:l/arxiv.org/abs/hep-th/0003139.
Dowker, F., and A. Kent (1996) 'On the Consistent Histories Approach to Quan-
tum Mechanics', f. Statist. Phys. 82: 1575-1646; preprint: http://arxiv.org/abs/
gr-qc/9412067v2.
Earman, J. (1995) Bangs, Crunches, Whimpers, and Shrieks: Singularities and Acausal-
ities in Relativistic Spacetimes (New York: Oxford University Press).
Earman, J., and J. Mosterin (1999) 'A Critical Look at Inflationary Cosmology',
Philosophy of Science 66: 1-49.
Ellis, G. F. R., and R. Maartens (2004) 'The Emergent Universe: Inflationary Cos-
mology with No Singularity', Classical and Quantum Gravity 21: 223; preprint:
http://arxiv.org/abs/gr-qc/0211082.
Ellis, G. F. R., Jeff Murugan and Christos Tsagas (2004) 'The Emergent Universe:
An Explicit Construction', Class. Quant. Grav. 21: 233-50; preprint: http://arxiv.
org/abs/gr-qc/0307112.
Falciano, F. T., and N. Pinto-Neto (2009) 'Scalar Perturbations in Scalar Field
Quantum Cosmology', Phys. Rev. D 79, 023507; preprint: http://arxiv.org/abs/
0810.3542.
140 William Lane Craig and James D. Sinclair

Geroch, R. (1978) GR from A to B (Chicago and London: University of Chicago

Press).
Gott, J. R. III, and L.-X. Li (1998) 'Can the Universe Create Itself?', Phys. Rev. D
58: 2, 023501-1.
Hartle, J. (1998) 'Quantum Pasts and the Utility of History', Phys. Scripta T76: 67;
preprint: http://arxiv.org/abs/gr-qc/9712001 vl.
Hartle, J., and S. Hawking (1983) 'The Wave Function of the Universe', Phys. Rev.
D 28: 12, 2960-75.
Hawking, S. (2003) 'Cosmology from the Top Down', talk presented at the Davis
Inflation Meeting; preprint: http://arxiv.org/abs/astro-ph/0305562.
Hawking, S., and R. Penrose (1970) 'The Singularities of Gravitational Col-
lapse and Cosmology', Proceedings of the Royal Society of London A 314:
529-48.
Hawking, S., and R. Penrose (1996) The Nature of Space and Time (Princeton, NJ:
Princeton University Press).
Jurkiewicz, J., L. Loll and J. Ambjorn (2008) 'Using Causality to Solve the
Puzzle of Quantum Spacetime', Scientific American Ouly); available online
at www.scientificamerican.com/article.cfm ?id=the-self-organizing-quantum-
universe.
Kaloper, N., A. Linde and R. Bousso (1998) 'Pre-Big-Bang Requires the Universe
to Be Exponentially Large from the Very Beginning', Phys. Rev. D 59, 043508;
available online at http:/lhep-th/9801073.
Kiefer, Claus (2008) 'Does Time Exist in Quantum Gravity?', FXQI essay contest
on the Nature of Time; available online at www.fqxi.org/data/essay-contest-
files/Kiefer_fqx.pdf.
Kiefer, Claus (2009) 'Can the Arrow of Time Be Understood from Quantum Cos-
mology?', in L. Mersini-Houghton (ed.), The Arrow ofTime (Heidelberg: Springer
Verlag); preprint: http://arxiv.org/abs/0910.5836vl.
Kiefer, C., and H. D. Zeh (1995) 'Arrow of Time in a Recollapsing Quan-
tum Universe', Phys. Rev. D 51: 4145-53; preprint: http://arxiv.org/abs/gr-qc/
9402036v2.
Loll, R. (2008) 'The Emergence of Spacetime, or, Quantum Gravity on Your Desk-
top', Classical Quantum Gravity 25.11: 114006; available online at http://arxiv.
org/abs/0711.0273v2.
Mersini-Houghton, Laura (2009) 'Notes on Time's Enigma', FXQI conference on
time, Azores; available online at http://arxiv.org/abs/0909.2330vl.
Mcinnis, Brett (2007a) 'Arrow of Time in String Theory', Nucl. Phys. B 782: 1-25;
preprint: http://arxiv.org/abs/hep-th/0611088v3.
Mcinnis, Brett (2007b) 'The Arrow of Time in the Landscape', in R. Vaas (ed.),
Beyond the Big Bang (Springer: Heidelberg); preprint: http://arxiv.org/abs/0711.
1656v2.
Mcinnis, Brett (2008) 'Initial Conditions for Bubble Universes', Phys. Rev. D 77,
123530; preprint: http://arxiv.org/abs/0705.4141 v5.
Monton, Bradley (forthcoming) 'Prolegomena to Any Future Physics-Based Meta-
physics', Oxford Studies in Philosophy ofReligion 3.
Mulryne, David, Reza Tavakol, james E. Lidsey and George F. R. Ellis (2005)
'An Emergent Universe from a Loop', Phys. Rev. D 71, 123512; preprint: http://
arxiv.org/abs/astro-ph/0502589v1.
Oppy, G. (2006) Arguing about Gods (Cambridge: Cambridge University Press).
 On Non-Singular Space-times and the Beginning of the Universe 141

Overbye, D. (2006) '9 billion-year-old "dark energy" reported', The New York Times
(15 January).
Parikh, M., Ivo Savonije and Erik Verlinde (2003) 'Elliptic de Sitter Space', Phys.
Rev. D 67, 064005; preprint: http://arxiv.org/abs/hep-th/0209120v2.
Penrose, R. (2005) The Road to Reality (New York: Alfred A. Knopf).
Penrose, R. (2006) 'Before the Big Bang: An Outrageous New Perspective and Its
Implications for Particle Physics', in Proceedings of tile European Particle Accelera-
tor Conference (EPAC) 2006 (Edinburgh: EPS-AG), pp. 2759-62; available online
at http:/ /accelconf.web.cern.ch/Acce!Conf/e06/PAPERS/THESPAO 1.PD F.
Penrose, R. (2009) 'Black Holes, Quantum Theory and Cosmology', in f. Phys.:
Conf Ser. 174, 01; available online at http:!/iopscience.iop.org/1742-6596/174/
1/012001/pdf/1742-6596_17 4_1_012001.pdf.
Pitts, J. B. (2008) 'Why the Big Bang Singularity Does Not Help the Kalam Cosmo-
logical Argument for Theism', British Journal for tl1e Philosophy of Science 59.4:
675-708.
Quine, W. V. (1969) 'Epistemology Naturalized', in idem, Ontological Relativity and
Other Essays (New York: Columbia University Press), pp. 69-90.
Rea, Michael C. (2002) World without Design: Tile Ontological Consequences of
Naturalism (Oxford: Clarendon Press).
Seahra, Sanjeev S., and C. G. Bohmer (2009) 'Einstein Static Universes Are Unsta-
ble in Generic f(R) Models', Pllys. Rev. D 79, 064009; preprint: http://arxiv.org/
abs/0901.0892.
Smith, Q. (1985) 'On the Beginning of Time', Norls 19: 5 79-84.
Smith, Q. (1993) 'The Wave Function of a Godless Universe', in Quentin Smith
and William Lane Craig (eds), Theism, Atheism and Big Bang Cosmology (Oxford:
Clarendon Press, 1993), pp. 301-37.
Smith, Q. (1997) 'The Ontological Interpretation of the Wave Function of the
Universe', Tile Monist 80.1: 160-85.
Smith, Q. (2000) 'The Black Hole Origin Theory of the Universe: Frontiers of Spec-
ulative, Current Physical Cosmology'; available online at www.faculty.umb.
edu/gary _zabel/Courses/Parallelo/o20Universes/Texts/the_black_hole_origin_
theory_of_the_universe_frontiers_of_s.htm.
Smith, Q. (2007) 'Kalam Cosmological Arguments for Atheism', in M. Martin
(ed.), The Cambridge Companion to Atheism (Cambridge Companions to Philos-
ophy; Cambridge: Cambridge University Press), pp. 182-98.
Sorabji, R. (1983) Time, Creation, and tile Continuum (Ithaca, NY: Cornell Univer-
sity Press).
Sorkin, R. D., and D.P. Rideout (1999) 'Classical Sequential Growth Dynamics for
Causal Sets', Pllys. Rev. D 61, 024002; available online at http://arxiv.org/abs/
gr-qc/9904062v3.
Smolin, Lee (2006) Tile Trouble with Physics (New York: Houghton Mifflin).
Smolin, Lee (2009) 'The Unique Universe', PhysicsWorld Oune); available online
at http:/ /physicsworld.com/cws/article/print/39306.
Tavakol, R., and S. Carneiro (2009) 'Stability of the Einstein Static Universe in the
Presence of Vacuum Energy', Pllys. Rev. D 80, 043528; preprint: http://arxiv.
org/abs/0907.4795v2.
't Hooft, Gerard (2009) 'Quantum Gravity without Space-time Singularities or
Horizons', presented at the Eric Summerschool of Subnuclear Physics, available
online at http://arxiv.org/abs/0909 .3426.
142 William Lane Craig and James D. Sinclair

Tod, K. P. (2003) 'Isotropic Cosmological Singularities: Other Matter Models',

Class. Quantum Gravity 20: 521-34.
Tolman, R. C. (1934) Relativity, Thermodynamics and Cosmology (Oxford:
Clarendon Press).
Vaas, R. (2004) 'Time before Time: Classifications of Universes in Contemporary
Cosmology, and How to Avoid the Antinomy of the Beginning and Eternity of
the World', in W. Loffler and P. Weingartner (eds), Knowledge and Belief Papers
of the 26th International Wittgenstein Symposium (Kirchberg am Wechsel: Aus-
trian Ludwig Wittgenstein Society), pp. 351-3; preprint: http://arXiv.org/abs/
physics/0408111.
Veneziano, G., and M. Gasperini (2002) 'The Pre Big Bang Scenario in String Cos-
mology', Physics Reports 373: 1; preprint: http://arxiv.org/abs/hep-th/0207130.
Vilenkin, A. (1982) 'Creation of Universes from Nothing', Phys. Lett. 117B: 25-28.
Vilenkin, A. (1994) 'Approaches to Quantum Cosmology', Phys. Rev. D SO:
2581-94; preprint: http:/ /lanl.arxiv.org/abs/gr-qc/9403010vl.
Vilenkin, A. (2002) 'Quantum Cosmology and Eternal Inflation', in The Future
of Theoretical Physics and Cosmology: Proceedings of the Conference in Honor of
Stephen Hawking's 60th Birthday; preprint: http:/ /arxiv.org/abs/gr-qc/0204061.
Vi! en kin, A. (2006) Many Worlds in One (New York: Hill & Wang).
Zeh, II. D. (2009) 'Open Questions Regarding the Arrow of Time', in L. Mersini-
Houghton (ed.), The Arrow of Time (Heidelberg: Springer Verlag); preprint
http:/ /arxiv.org/abs/0908.3 780.
6
The Theistic Multiverse:
Problems and Prospects
Klaas J. Kraay

In recent decades, there has been astonishing growth in scientific

theorizing about multiverses. Once considered outre or absurd, multiple
universe theories appear to be gaining considerable scientific respectabil-
ity. There are, of course, many such theories, including (1) Everett's
(1957) many worlds interpretation of quantum mechanics, defended by
Deutsch (1997) and others; (2) Linde's (1986) eternal inflation view,
which suggests that universes form like bubbles in a chaotically inflating
sea; (3) Smolin's (1997) fecund universe theory, which proposes that uni-
verses are generated through black holes; (4) the cyclic model, recently
defended using string/M theory by Steinhardt and Turok (2007), which
holds that distinct universes are formed in a never-ending sequence
of Big Bangs and Big Crunches; and (5) Tegmark's {2007) 'Level IV'
multiverse, which contains many universes governed by distinct math-
ematical and scientific laws. While not all of these preclude each other,
the details and implications of each one are hotly contested. 1
In one area within the philosophy of religion (the debate concern-
ing the 'fine-tuning' argument), scientific multiverse theories are widely
held to be hostile to theism. This is because such theories appear to
account for the relevant data - the biophilic parameters of the uni-
verse we inhabit - without appeal to an intelligent designer. Yet, in
recent years, several philosophers2 and one physicist3 have offered rea-
sons for thinking that if theism is true, the actual world comprises (or
probably comprises) many universes. I first set out some requirements-
both scientific and otherwise - for such a theory. I then survey some
problems such theories are held to face, and some prospects they are
thought to have. Finally, I examine arguments both for and against the
claim that multiverse theories can help theists respond to the problem of
evil. I conclude that such theories advantage neither the theist nor the

143
144 Klaas f. Kraay

atheist in the debate about evil: they merely require reframing arguments
from evil.

1. Some Requirements for a Theistic Multiverse Theory

The claim that on theism the actual world is (or probably is) a multiverse
is not explicitly about God's activity with respect to the natural world.
But of course theism holds that God is the ultimate causal explanation
for the existence of contingent reality, including the natural realm we
inhabit. Such a view must, then, be compatible with what contempo-
rary science has to say about causation. Theists typically think of God's
causal role as an initial act of creation ex nihilo, followed by the contin-
ued conservation in existence of all the things there are. This requires
an account of causation consistent with the doctrine of an immate-
rial being bringing about physical effects, which can allow for such a
being to bring about effects ex nihilo. There are related issues about God
and time: for example, if God is thought to exist outside of time, an
account is needed which can allow for a timeless being to have tempo-
ral effects. A model of causation is needed which can make sense of the
traditional idea that God sustains things in existence. And famously, of
course, there are vexing questions about how best to understand tradi-
tional claims about God's causal interventions in the course of nature.
In what follows, however, I set all these issues aside, since they pertain
to many theistic accounts of the existence of contingent reality, and are
not peculiar to multiverse theories.
A complete theistic multiverse theory will offer an account, consis-
tent with the best science, of what a universe is. 4 But, of course, scientific
research has not yet fully answered this question: apart from some
very general points of consensus, different theories vary widely. 5 The-
istic multiverse theories will not, then, be complete until the science
is settled. But in the meantime, defenders of theistic multiverses can
develop their views in accordance with one or more scientific theo-
ries, recognizing that these are controversial, or they can remain neutral
between many or all such theories. Most theistic multiverse theories take
for granted that universes are independent, non-interacting, spatiotem-
porally interrelated objects. Some, however, appear to assume that all
universes are related, either temporally (Stewart, 1993, p. 61), or by
being embedded in a higher spatial dimension (Hudson, 2006). Some
philosophers are officially neutral on this issue (Draper, 2004, p. 316;
Forrest, 1996, p. 218). I take it that whether or not there are com-
pletely disconnected space-times is a question for science, not theology
 The Theistic Multiverse: Problems and Prospects 145

or philosophy, to settle. It's worth noting that if there are, this will falsify
philosophical theories, like David Lewis's modal realism, which preclude
this. 6 And this matters in the current context because at least one rival
to theistic multiverse theories depends upon Lewis's view that there can-
not be independent space-times/ So here is a clear case where science
can help to adjudicate between competing philosophical theories.
Some defenders of a theistic multiverse maintain that universes come
in different kinds or types, without giving a complete account of what
they mean. Stewart appears to think that universes are to be distin-
guished with respect to the divine 'values' and 'purposes' that they
express, but offers no details (1993, p. 61). O'Connor speaks of 'organic
kinds' and 'value types', without definitions or examples (2008, p. 120).8
Forrest (1996, p. 116) suggests that God will create 'universes of many
types', without saying what a type is. In an earlier paper, he suggests that
universes can be classified into 'kinds', according to the laws of nature
they exhibit (1981, p. SO). For this suggestion to be defended, it must be
shown compatible with scientific accounts of the laws of nature. More
generally, any view that relies on such a classification scheme for uni-
verses must defend it, and this involves ensuring that it is consistent
with the best contemporary science.
Some authors claim that there can be duplicate universes within a mul-
tiverse (Parfit, 1992, p. 423; Monton, forthcoming); but others deny
this, citing the Principle of Identity of Indiscernibles (McHarry, 1978,
p. 133; Turner, 2003, p. 150). Still others are silent or neutral on this
issue (Stewart, 1993; Forrest, 1981, 1996; Draper, 2004; Hudson, 2006;
O'Connor, 2008; Kraay, 2010a). Of course this principle is controver-
sial, so those who maintain that there cannot be duplicate universes
should either offer plausible arguments for it, or else should offer some
other reason for thinking that there cannot be duplicates. But it is also
worth noting that, even if the Principle of Identity of Indiscernibles is
false, in principle there could be scientific reasons which preclude there
being duplicate universes. So physics and cosmology will be relevant
here as well.
Some authors appear attracted to the view that, on theism, every uni-
verse will feature living creatures (Leslie, 1989, p. 102; Turner, 2003,
p. 147; O'Connor, 2008, p. 112). For Turner and O'Connor, this is
because a loving God would want to show love to creatures. While God
may want to have creatures to love, there may of course be other rea-
sons why God would create uninhabited universes. In any case, this is
another point at which scientific research places a constraint on the-
ological speculation: if science establishes that there are uninhabited
146 Klaas f. Kraay

universes, as all current scientific multiverse theories maintain, these

claims will need to be abandoned.
Explicitly or implicitly, all theories of God's role as creator and sus-
tainer involve God selecting, from an array of possible worlds, just one
for actualization. 9 And it is generally thought that these objects of God's
choice are proper subjects of axiological evaluation. So all theistic mul-
tiverse theories are committed to claims about the axiological status of
both possible worlds and universes. Regarding the former, I have argued
elsewhere that a good framework for this involves taking the axiological
status of worlds to depend upon which world-good-making properties
(WGMPs) and the world-bad-making properties (WBMPs) it exhibits,
and, for degreed properties, the degree to which they are instantiated. 10
But of course this is just a schema: a complete theory would fill out the
details of what these properties are, and just how they determine the
axiological status of worlds. Ideally, such a completed theory will reveal
whether there is a best or unsurpassable world (in various senses to be
distinguished below), and whether or two or more worlds can be tied in
axiological status.
Scientific research may constrain this axiological theorizing. Here are
three examples of how this might happen. First, someone might hold
that some or all axiological properties supervene on natural ones: if so,
an account of the relevant natural properties is needed to ground axio-
logical claims. Second, someone might hold with Leibniz that the best
world features simple natural laws that generate great diversity: if so, an
account of both laws and simplicity is needed. Third, someone might
hold that worlds in which everything is spatiotemporally interconnected
are, ceteris paribus, more valuable than those which feature disconnected
space-times.U But if such interconnected worlds are precluded by sci-
ence, then such claims about what is expectable on theism will have to
be adjusted.
Suppose that axiological theorizing leads to a broad consensus about
what sorts of properties are WGMPs and WBMPs. A further issue- which
has received insufficient attention - is the modal status of these proper-
ties.12 Most authors appear to assume necessitarianism: the view there is
only one common set of WGMPs and WBMPs, with reference to which
all possible worlds are to be evaluated. On the rival view, contingentism,
there is no one common set of axiological properties relevant to all
worlds: some worlds are properly evaluated with respect to one set of
WGMPs and WBMPs, while other worlds are properly evaluated with
respect to a distinct set of WGMPs and WBMPs. On necessitarianism, all
possible worlds are commensurable: this is just what it means to say that
 The T/Jeistic Multiverse: Problems and Prospects 147

they are all to be evaluated with respect to the common set of axiological
properties. On contingentism, however, there are failures of commensu-
rability between possible worlds. On contingentism, it is useful to define
a world-cluster to be a set of worlds which are commensurable. (Neces-
sitarianism, then, amounts to view that there is just one world-cluster.)
A complete axiological theory will settle whether necessitarianism or
contingentism is true. 13
On either view, however, there is a further complication. Although
all worlds within a given cluster are commensurable, they may not all be
comparable: there may be a pair of world-cluster-mates w1 and w2 such
that neither is better than the other, nor are they equal in axiological sta-
tus. If so, it will be useful to reserve the term world-hierarchy to refer to
a set of worlds (within a given cluster) which are both commensurable
and comparable. No two worlds belonging to different world-hierarchies
are comparable, although they are commensurable only if they are
cluster-mates. A complete axiological theory will settle whether there
are genuine failures of comparability between commensurable worlds.
These distinctions are important because they bear on the question of
whether or not there is a best possible world: they reveal the question
to be too simplistic. On necessitarianism, perhaps there is exactly one
world which is best. in the strong sense of being better than all others,
or perhaps there is one or more world which is unsurpassable. in virtue
of being surpassed by no other world. On contingentism, however, there
can be no best. world (simply because there are failures of commensura-
bility between worlds in different clusters), although in principle there
could be one or more unsurpassablea worlds. 14 A world is bestc when it
is better than all of its other cluster-mates, and a world is unsurpassablec
when no world in its cluster surpasses it. These terms add nothing new
on necessitarianism, since on that view, all worlds are cluster-mates. But
on contingentism, perhaps some clusters feature such worlds, while oth-
ers contain no such worlds. 15 Finally, within a given hierarchy (on either
necessitarianism or contingentism) all worlds are both commensurable
and comparable, so in principle there could be a world which is best, by
surpassing all of its other hierarchy-mates, or unsurpassable, in the sense
of being unsurpassed by all hierarchy-mates. A complete axiological
account will perhaps reveal whether there really are worlds answering
to these definitions.
The foregoing can be applied mutatis mutandis to universes. 16
Some authors think it obvious that there are unsurpassable universes
(McHarry, 1978); others think it obvious that there are not (O'Connor,
2008)Y But it is important to be careful about what this means.
148 Klaas f. Kraay

If necessitarianism about universes is true, then in principle there could

be a best. universe, or unsurpassablea universes. But if contingentism
about universes is true, no universe can be best., although in principle
there could be one or more unsurpassable. universes. On contingen-
tism, perhaps some clusters features a best, universe, or one or more
unsurpassable, universes, and perhaps other clusters feature neither kind.
Finally, within a given universe-hierarchy (on either necessitarianism or
contingentism), all universes are both commensurable and comparable,
so in principle there could be a universe which is best1, in the sense of
being better than all of its hierarchy-mates, or unsurpassable, in the sense
of being unsurpassed by all hierarchy-mates. Until a more complete axi-
ological of universes is given, it cannot be determined whether there are
best or unsurpassable universes in all these various senses.
The axiological evaluation of universes is particularly important for
theistic multiverse theories, since they all posit that there is an objective
axiological threshold above which all universes are worthy of inclusion
in a multiverse created by God, and at or below which no universe is
worthy. Some say little about where this threshold lies. Hudson (2006),
for example, asserts that God will create every universe 'worth having'
(p. 167): all those which 'satisfy a certain minimal criterion of value'
(p. 170). O'Connor simply says that there is an objective threshold,
and adds that it may be vague (2008, p. 158). And I have said that
God will include 'all and only those universes worth creating and sus-
taining' (Kraay, 2010a, p. 363). Others offer more detail. Turner (2003)
suggests 'a favourable balance of good over evil' (p. 149), and Manton
(forthcoming) offers various construals of what this might mean. But
such a threshold is too simplistic: a universe might meet this condition,
while nevertheless containing some feature that makes it unworthy of
inclusion in a divinely furnished multiverse. Perhaps sensing this, some
philosophers have proposed additional requirements. Parfit says there
should be no injustice (1991, p. 5), and that each individual's life must
be worth living (1992, p. 423). Forrest first says that every individual
must have a life in which good outweighs evil (1981, p. 53), and later
adds two further restrictions: each creature who suffers must at least
virtually consent to it, and must receive ample recompense afterwards
(1996, pp. 225-7). Draper says that no individual's life may be bad over-
all, and that God must be a benefactor to all creatures (pp. 319-20).
These criteria may well be plausible, but of course they are only partial
specifications of a threshold, since they only pertain to universes inhab-
ited by creatures like ourselves. A more complete account of threshold
is needed.
 The Theistic Multiverse: Problems and Prospects 149

Most defenders of the theistic multiverse maintain that God will cre-
ate every universe above the threshold. 18 (Surprisingly, few explicitly
add the plausible restriction that God will create only these.) O'Connor,
however, denies that God will create every universe above the thresh-
old (2008, p. 119), and others are silent or neutral on this issue. 19
O'Connor's denial that God will create every universe above the thresh-
old has been persuasively criticized (Almeida, 2010, p. 304 and Manton,
forthcoming). At the very least, one who denies that God will create
all universes above the threshold bears the burden of undermining or
defeating the natural presumption that, all else equal, creating another
good thing makes the ensemble better overall. 20
Finally, while most theistic multiverse accounts in the literature con-
centrate on God's creation of universes, it is important also to incorporate
to the traditional idea that God sustains universes in existence. 21 In what
follows, then, 'TM' will refer to that theistic multiverse which includes
all and only those universes worthy of being created and sustained by
God. While a complete theory will develop this claim in the various ways
identified in this section, some important philosophical issues concern-
ing this theory can be discussed prior to these matters being settled.
I turn to these now.

2. Problems and Prospects for TM

In this section, I briefly survey some important objections to TM,
and some prospects it is thought to have. The first four threaten the
coherence of TM.

1. As we have seen, all theistic multiverse accounts presuppose that

there is an objective axiological threshold above which universes
are worthy of inclusion, and at or below which they are unworthy.
But here is an objection familiar from the literature on the Prob-
lem of No Best World: perhaps, for any such threshold one might
posit, there is a defensible superior threshold, in which case no puta-
tive threshold is good enough for an unsurpassable being. 22 Theistic
multiverse theories must respond to this objection.
2. Tom Talbott and Peter van Inwagen have both complained to me
that TM cannot contain all universes worthy of being created and
sustained, using the following thought experiment. Consider a per-
son S who exists in some universe U in TM. Assuming trans-world
(and trans-universe) identity, that person also exists in universes con-
tained in other worlds, and it is not unreasonable to suppose that at
150 Klaas f. Kraay

least one of these is worthy of being created and sustained. But then
the theistic multiverse cannot include all worthy universes. 23
3. Gale and Pruss (2003, p. xxvi) anticipate, and Monton (forthcoming)
offers, a different argument for the claim that that no multiverse can
include all universes worthy of creating and sustaining. Rejecting the
Principle of Identity of Indiscernibles, Monton says that God could
create duplicates of worthy universes. Since for any number of dupli-
cates God could create, God could have created more, Monton says,
it is impossible for God to create all worthy universes. 24
4. Pruss has suggested that whether a universe is worthy of creation and
sustenance may depend upon which other universes are worthy. 25
Consider, for example, a pair of universes Ul and U2 which are both
very close to the threshold. Suppose that thousands of philosophers
and scientists in Ul justifiably believe that U2 does not exist, and
that in U2 thousands of philosophers and scientists justifiably believe
that Ul does not exist. Suppose, further, that if the relevant belief in
Ul is true, the value added thereby to Ul suffices to bring it above
the threshold, in which case U2 is below the threshold - and vice
versa. On this view, it can seem an arbitrary matter which universe is
included. 26

Apart from those arguments which claim that TM is incoherent, three

other arguments against TM are worth briefly noting.

S. Appealing to Ockham's razor, someone might hold that multiverses

objectionably inflate our ontology. While this charge might be a
legitimate complaint against an ad hoc appeal to many indepen-
dent universes, defenders of the theistic multiverse can reply that
their model is not ad hoc: it is defended by arguments about what
a perfectly good creative agent would doY Moreover, the recent
growth in scientific theorizing <.bout multiverses should diminish,
if not entirely remove, the initial appeal of this objection.
6. Theists have traditionally held that God's decision to create is a free
one, but if one thinks that TM is the unique best of all possible
worlds, it might seem that God cannot do otherwise than to select
it, and that this in turn compromises God's freedom. 28 And if God's
choice is unfree, one might further urge, God is unworthy of thanks
and praise for creating. 29
7. Relatedly, some have held that if TM is the only world that God can
choose, it is in fact the only possible world simpliciter- a consequence
held to be absurd. 30
 The Theistic Multiverse: Problems and Prospects 151

These, then, are some of the problems that TM faces. But it is also
thought to have some important prospects in the philosophy of religion:

1. As mentioned in the introduction, scientific multiverse theories are

thought to undermine the fine-tuning argument. But if philosophers
can show that iftheism is true, the actual world is (probably) a multiverse,
this objection will be more difficult to sustain.
2. If TM really is, as some of its proponents claim, the best of all pos-
sible worlds, then an important argument for atheism is evaded: the
Problem of No Best WorldY
3. While the problem of actual evil holds that evil in the actual world
disconfirms theism, the problem of possible evil holds that the exis-
tence of bad possible worlds disconfirms theism. But if TM is both
good and the only possible world, then there simply are no bad
possible worlds to which the defender of this argument can appeal. 32

These three prospects have seen relatively little discussion in the liter-
ature. In contrast, rather more attention has been paid to a different
putative prospect: the claim that theistic multiverse theories can help
theists to respond to the problem of evil. In section 3, I outline (and crit-
icize) what defenders of this position have said. In section 4, however,
I take issue with some of their critics. In section 5, I argue that multi-
verse theories do not advantage either side in this debate: they merely
require reframing certain arguments from evil.

3. Defenders of Multiverse-Based Responses

to Arguments from Evil
Of course, there is no such thing as the problem of evil. Instead, there are
many different arguments for atheism, having different logical forms,
which involve one or more premises about evil. Authors who think that
a multiverse will help theists typically have one or both of the following
argument schemas in mind. (These come in both deductive and induc-
tive variants, and the second premise can be supported in various ways.)

Argument/

1. If theism is true, the actual world is the best of all possible

worlds.
2. (Probably,) the actual world is not the best of all possible worlds.
:. 3. (Probably,) theism is false.
152 Klaas f. Kraay

Argument II

1. If theism is true, there is no gratuitous evil.

2. (Probably,) evil e is gratuitous.
:. 3. (Probably,) theism is false.

McHarry asserts that his argument would, if sound, 'dissolve' the

problem of evil, which he takes to be Argument I (1978, p. 134). He
begins by identifying the best of all possible worlds with a multiverse
comprised of all (and, we should add, only) the threshold-surpassing
universes. McHarry grants premise 1, and so wants to deny premise
2. One might expect him to offer evidence for thinking that the actual
world is such a multiverse,but he does no such thing. 33 More mod-
estly, one might expect McHarry to argue that our universe is (probably)
above the threshold - or, at least, to block reasons for thinking that it is
not (probably) above the threshold. McHarry does neither. So McHarry
does not 'dissolve' even Argument I, and certainly not 'the' problem of
evii.34
Turner (2003) advertises his multiverse theory as a 'solution' (p. 143)
to the problem of evil, which he takes to be Argument I. But he later
moderates this claim, calling it a 'partial answer' and a 'partial solution'
(p. 157). Like McHarry, Turner thinks that his multiverse is the best of
all possible worlds, and he appears to concede premise 1 - so he presum-
ably wants to block premise 2. Turner says that if God were to create all
(and, we should add, only) those universes which have a favourable bal-
ance of good over evil, 'it should not be surprising if it seems to us that
we are in a [universe] that could be better than it is' (p. 157). But this
move only blocks one reason for thinking that our universe should not
be included in the multiverse. Perhaps there are others. Someone might
defend premise 2, not by saying that our universe is surpassable, but
by urging that it contains some feature which God would not permit.
Turner continues: '[o]nly if this were a [universe] with a preponderance
of evil over good should we conclude that it was a [universe] a benevo-
lent God would not have created' (p. 157). But this is not so. To see why,
recall the additional restrictions proposed by Parfit, Forrest and Draper
on a universe being worthy of creation and sustenance. If any such con-
dition is plausible, a universe with equal amounts of good and evil (or
indeed with more good than evil), and in which such a condition is not
satisfied, would be unworthy of inclusion in a theistic multiverse. So, con-
tra Turner, there being a preponderance of evil over good is not the only
way for a universe to be deemed unworthy. At most, then, Turner has
 The Theistic Multiverse: Problems and Prospects 153

provided a theistic response to just one defence of one premise of one

argument from evil.
Hudson (2006) thinks that his plenitudinous hyperspace can be dep-
loyed against both arguments for atheism displayed above. 35 Hudson
thinks that his multiverse is the best possible world, but like McHarry
and Turner, offers no reason for thinking that our universe is indeed
above the threshold he posits: 'worth creating'. So Hudson does not, in
this sense, defeat the first argument for atheism. He does, however, sug-
gest a way to use his hyperspace to block a certain defence of premise 2
in a deductive version of Argument II. That defence appeals to consid-
erations about suffering experienced by non-human animals that is not
caused by agents. Since these considerations can also be used to defend
premise 2 of a deductive version of Argument I, Hudson's move is best
seen as a response to both.
Hudson takes himself to have an undercutting defeater for such a
defence of premise 2 of both arguments. In the spirit of St Augustine,
he believes it is metaphysically possible that no such evil in our uni-
verse is gratuitous, since every instance is required to bring about some
otherwise-unobtainable aesthetic good(s) which outweigh(s) the evil in
question (2006, pp. 172-81). This response has been criticized in some
mistaken ways. Gilmore (2006) and Rea (2008) note that Hudson's move
does not require hyperspace: one might instead posit lower-dimensional
aesthetic properties as possible justifications for natural evil. This is per-
haps true, but of course Hudson did not devise his hyperspace simply
to reply to arguments from evil. He takes himself to have independent
a priori grounds for the claim that, on theism, God would create such
a hyperspace. Rea offers another criticism. He points out that all of the
allegedly gratuitous evils in 3-space are also contained in 4-space, so
'every reason we have for thinking that our 3-space isn't the best possible
will also be a reason for thinking that our 4-space isn't the best possible'
(p. 450). But this misses the point of Hudson's argument. Hudson wants
to concede that our 3-space may be surpassable, while insisting that this
fails to show that the actual world is surpassable.
A better criticism is anticipated by Hudson (2006, p. 179), and
explicitly levelled by Mouton (forthcoming): 'an omnibenevolent being
would not value aesthetic properties over preventing an innocent crea-
ture from pointlessly suffering'. Mouton does not argue for this claim,
thinking it obvious. My sympathies are with Mouton here: it is difficult
to see how a perfectly good being could permit creatures to suffer for the
sake of aesthetic goods which they cannot appreciate. Now it's true that
Hudson merely wants to establish a metaphysical possibility, in response
154 Klaas f. Kraay

to deductive variants of these arguments. So even if one grants Hudson

this much, and grants that this is sufficient to undermine this particular
defence of premise 2 in both arguments, it should be stressed that this
is only a very limited contribution to debate about evil. 36

4. Critics of Multiverse-Based Responses

to Arguments from Evil
So far, we have seen that theistic multiverses should not be thought
to 'solve' 'the' problem of evil. In this section, I turn to the critics
of multiverse-based responses to arguments from evi!Y The first two
deny that a multiverse containing an infinity of universes above some
threshold can be improved by adding another such universe.
Perkins (1980) follows McHarry (1978) in assuming that there are
'optimal' universes as well as 'non-optimal' ones. Perkins assumes that
our universe is of the latter sort. But if there are infinitely many non-
optimal universes worth creating, he says, 'it is difficult to understand in
what sense of "better" the [world] is better for including our [universe]'
(p. 170). He doubts that it would be meaningful to say that a world fea-
turing our universe has more value than a world without it (pp. 170-1).
In his response to O'Connor (2008), Almeida (2010) considers whether
God would actualize a super-universe SU00 containing infinitely-many
universes above threshold 1:. Almeida points out that if such a world
has infinite value, then removing some or even many universes from
it will not diminish its value: 'even if there were infinitely many uni-
verses is SU00 that included the immense suffering and disvalue found
in the actual world, we could remove all of those universes without
diminishing the overall infinitely positive value of SU00 ' (p. 306). So,
although he appears to grant that our universe has 'on-balance positive
value' (p. 306), he thinks that God, on O'Connor's model, would not
create it. 38
Perkins and Almeida thus both believe that God can create an
infinitely-valuable universe without including our universe, and that
accordingly, God should not include it. But there are reasons for resisting
this inference, even if the premise is accepted. First, theists sometimes
maintain that God is infinitely valuable, in which case the world in
which God creates nothing might well be thought infinitely valuable.
This is the 'bare world': it consists of God, whatever other necessary exis-
tents there are, and whatever uncreated contingent beings it contains.
Since most theists maintain that at least some worlds in which God
creates and sustains contingent entities are better than the bare world,
 Tile Theistic Multiverse: Problems and Prospects 155

there is some motivation for thinking that infinitely valuable states of

affairs can be bettered. 39
Almeida himself offers a different reason for resisting this inference
(2008, p. 156). Following Vallentyne and Kagan (1997), he imagines a
world w1 which comprises infinitely-many temporal locations, each of
which has axiological status 10. Surely, Almeida says, w1 is on balance
better than w2 , a world which comprises infinitely many temporal loca-
tions having axiological status 1. Almeida doesn't say what it is for a
temporal location to have an axiological status, but this needn't detain
us. His point is presumably intended to apply to multiverses as fol-
lows: surely it is intuitive to think that multiverse m1 is far better than
multiverse m2 when all the universes in the former are (say) ten times
more valuable than all the universes in the latter - even though both
multiverses have infinite value.
Both of these reasons motivate searching for a principled way to dis-
tinguish the relative value of infinitely good worlds. And, if such a way
can be found, this can be used in a response to Perkins (1980). Sev-
eral such ways are available. First, as Almeida himself notes, 'There are
of course nonstandard mathematical representations of infinite value
according to which addition and subtraction of infinite numbers is well
defined' (2008, p. 156). Here is a second approach. Manton, inspired by
Almeida (2008) and Vallantyne and Kagan (1997), suggests the following
principle:

If world w1 has all the locations that w2 has, but w1 has more loca-
tions as well, and if the values of all the shared locations are the same,
and the values of the non-shared locations in w1 sum to a positive
number, then w1 is better than w2.

As Manton rightly observes, a defect of this principle is that it would -

implausibly - license God to create some unworthy universes in w1 • But
this principle can easily be modified to prevent this unwanted outcome.
Consider:

If multiverse ml includes infinitely-many threshold-surpassing uni-

verses (and no other universes), and multiverse m2 includes all the
universes that m1 includes, and also includes threshold-surpassing
universes that m1 lacks, (and no other universes) then, ceteris
paribus, m2 is better than ml.

This captures the plausible intuition needed to defeat the objection

expressed in Perkins (1980) and Almeida (2010).
156 Klaas f. Kraay

But suppose that one rejects these moves, or even despairs of finding a
principled way to distinguish the rtlative axiological status of such mul-
tiverses. A different approach might concede that (at least some) pairs
of infinitely-membered multi verses ;:re equal in value, while denying that
the value of the two relevant world-actualizing actions is equivalent. For
example, one might hold that, in such cases, world-actualizing action
Al is better than world-actualizing action A2 just in case the world that
results from Al contains one or more threshold-surpassing universes
that the world resulting from action A2 lacks, and no other universes.
While Perkins (1980) and Almeida (2010) deny that a mu1tiverse con-
taining an infinity of universes above some threshold can be improved
by adding another such universe, Manton (forthcoming) takes the
opposite tack: he holds that God can improve such a universe. Man-
ton posits that our universe contains gratuitous evil, but nevertheless
is above his axiological threshold: it has a favourable balance of good
over evil. 40 One might expect Manton to say that since our universe is
above this threshold, it is worthy of inclusion in the multiverse, and that
since including it would add to the goodness of reality, then God should
include it. Surprisingly, however, this is not what Manton thinks. He
says that God could add to the goodness of reality even more by creating
duplicates of better threshold-surpassing universes: those lacking gratu-
itous evil. And so, Manton thinks, God would never 'feel compelled' to
create any universe like our own. But this is implausible. If, as he explic-
itly claims, every threshold-surpassing universe adds to the goodness of
reality, God has sufficient reason to create each one. 41

5. The Way Forward: Reframing Arguments from Evil

In sections 3 and 4, I objected to claims made by both defenders and
critics of multiverse-based responses to the problem of evil. I now argue
that multiverse theories are best thought of as requiring defenders of
certain arguments from evil to reframe their arguments. I begin with an
illustrative exchange between O'Connor and Oppy.
O'Connor is more modest than the authors considered in section 3:
he merely says that his multiverse 'has some relevance to' the problem
of evil (2008, p. 122; 2010a, p. 271). Specifically, he thinks that

God will in fact have compelling reasons to create a universe in

which significant suffering is permitted to occur even if the goods that
require suffering are not the greatest goods, or if the universe in which
they occur does not belong to a class of supremely valuable realms. All
 The Theistic Multiverse: Problems and Prospects 157

that is required is that the suffering-risking universe satisfy a mini-

mal threshold of goodness. (2008, p. 123; emphasis in original, and
see 2010b, p. 316)

It's not clear what the point of O'Connor's first italicized suggestion
is: I know of no arguments from evil which claim that God can per-
mit suffering only for the sake of 'the greatest' goods. That aside,
O'Connor seems to be saying that those who offer arguments from
evil against multiverse-theism bear the burden of showing that our uni-
verse fails to surpass the relevant threshold of goodness. Oppy (2008)
expresses sympathy for O'Connor's project, but offers the following
caution:

it is perhaps worth pointing out that O'Connor's theory clearly

doesn't help to overturn the thought that a morally perfect being
would not permit a person to suffer horrendously unless it was in
the interests of that very person to undergo the horrendous suffering
in question. Appearances generated by O'Connor's discussion of the
value of universes to the contrary, considerations about minimum
thresholds of goodness in suffering-risking universes cannot float free
of these kinds of deontological concerns (or so it seems to me).

Oppy is challenging O'Connor to say more about the threshold: in

particular, he is asking O'Connor to add a restriction similar in spirit
to those offered by Parfit, Forrest and Draper. O'Connor could do so,
and then, if he wished to develop his argument further, he could offer
reasons for thinking that this restriction is (probably) satisfied in our
universe, or at least could try to block reasons for thinking that it is
(probably) not.
What this exchange illustrates is that defenders and critics of
multiverse-theism can work together to find a mutually agreeable
threshold above which universes should be deemed worthy of inclu-
sion in a theistic multiverse. They may well subsequently differ on the
a posteriori question of whether our universe surpasses the agreed upon
threshold. Considerations about the existence, variety, magnitude, dura-
tion, scope, distribution, types or intensity of evil in our universe could
be appealed to in defending the claim that our universe is (probably) not
worthy of inclusion. And defenders of multi verse-theism could either try
to show that our universe (probably) is worthy of inclusion, or - more
modestly- they could try to defeat or undermine arguments to the con-
trary. In short, the typical moves in the debate concerning the problem
158 Klaas f. Kraay

of evil can easily be reframed to apply to multiverse-theism. But this, by

itself, will not furnish an advantage to either side. 42

6. Conclusion
The remarkable recent development of scientific multiverse theories
should be taken into account by philosophers. Those who wish to
argue that if theism is true, the actual world (probably) is a multiverse
should take care to ensure that their account is consistent with the best
contemporary physics and cosmology. An adequate axiological theory
for both worlds and universes is needed, and scientific considerations
may constrain this in various ways. Theistic multiverse theories face
several important problems, but also offer significant prospects for var-
ious issues in the philosophy of religion. Solving the problem of evil,
however, is not among these prospects.

Notes
1. For surveys of these and other theories, see: Leslie, 1989; Rees, 2001; Kaku,
2005; Vilenkin, 2006; Carr, 2007; Gribbin, 2009; and Greene, 2011.
2. McHarry, 1978; Parfit, 1991, 1992; Stewart, 1993; Forrest, 1981, 1996; Leslie,
1989; Turner, 2003; Draper, 2004; Hudson, 2006; Collins, 2007; O'Connor,
2008; and Kraay, 2010a and 2011a.
3. Page, 2010.
4. In some theistic multiverse theories, universes are insufficiently distin-
guished from possible worlds. See, for example, Vallicella's criticisms (1994)
of Stewart, 1993, and Almeida's criticisms (2008, pp. 146-8) of Turner, 2003.
5. See Leslie, 1989, pp. 66-9.
6. Interestingly, Lewis concedes in two places that he would prefer not to deny
that there can be disconnected space-times (1986, pp. 71, 74). Bricker (2001)
seeks to adapt Lewisian modal realism to allow this.
7. Michael Almeida's theistic modal realism (2008, 2011).
8. He does define a universe type a few pages earlier, but this does not seem to
be a value type (p. 116).
9. For more on this, see Kraay, 2008.
10. See Kraay, 2010a and 2011b.
11. Gale and Pruss mention this claim (2003, p. xxvii).
12. I discuss this in Kraay, 2011b.
13. In what follows, I assume that every world belongs to a cluster, and that no
world can belong to more than one cluster: for every world, there is one
unique set of WGMPs and WBMPs which constitutes the appropriate criteria
for assessing that world. I make the same assumption concerning universes
and their axiological properties.
14. If, on contingentism, there is a world which is is commensurable with
no other worlds, it must be deemed (trivially) unsurpassable•. The same
 The Theistic Multiverse: Problems and Prospects 159

consequence applies, mutatis mutandis, to the other senses of unsurpassable

distinguished below.
15. Even if there are failures of comparability within a given cluster, this by itself
does not preclude there being a best, world in that cluster. Conversations
with Graham Oppy, Ed Wierenga and Yujin Nagasawa helped me to see this.
16. Most theistic multi verse theories appear to assume that all universes are both
commensurable and comparable. One exception is O'Connor (2008), who
says that universes belonging to different kinds or types are 'incommen-
surate' (p. 120), or 'likely not fully commensurate' (p. 117). It's not clear,
however, whether O'Connor has in mind (using my terminology) failures of
commensurability or of comparability.
17. O'Connor offers a brief argument for this claim (2008, pp. 117-18), but does
not defend key aspects of it, including his distinctions between intensive,
aggregative and organic values, and his claim that God alone can possess
infinite intensive and organic value.
18. McHarry, 1978, p. 134; Parfit, 1991, p. 5; 1992, p. 423; Turner, 2003, p. 149;
Hudson, 2006, p. 167; Kraay, 2010a, pp. 361-3.
19. Stewart, 1993; Forrest, 1981, 1996; Draper, 2004.
20. I say more on this in section 4.
21. I discuss this in Kraay, 2010a.
22. I discuss this issue in the context of worlds in Kraay, 2010b.
23. It's worth noting that this objection could be expressed in counterpart-
theoretic terms. My tentative reply is developed in Kraay, 2011a: there is
reason to think that on theism, TM is the only possible world, in which case
a fortiori no individual can inhabit multiple worlds, and no individual has
an other-worldly counterpart.
24. Monton's view is puzzling: since many proponents of theistic multiverses
already concede that there may be infinitely many universes worthy of being
created and sustained, it's not clear what his appeal to duplicates adds to
the discussion. Perhaps, then, his real worry is that there can be no actual
concrete infinities.
25. http:/ I alexanderpruss .blogspot .com/2008/02 /complication -for -multiverse-
theories_04.html.
26. One might dispute the coherence of the objection in two ways, both of
which require defence. First, one might deny that there could be no other
reason for including (or failing to include) such universes. Second, one
might deny that many peoples' justified belief being true can raise the overall
axiological status of a world in the stipulated way.
27. O'Connor, 2008, p. 122, also makes this point.
28. Almeida (2008, p. 162) levels this charge against Hudson (2006). Those who
maintain that God needn't create every world above the threshold may pre-
serve some scope for divine freedom here, but at a cost: God's choices on this
view have been alleged to be arbitrary (Mawson, 2009).
29. I discuss both issues in Kraay, 2008.
30. Almeida (2008) levels this charge against Turner (2003) and Hudson (2006);
Monton (forthcoming) responds. I concede the charge in Kraay, 201la, while
tentatively arguing that it may be less serious than it appears.
31. I survey the literature on this argument in Kraay, 2010b.
32. I attempt such an argument, and survey the costs, in Kraay, 2011a.
160 Klaas f. Kraay

33. In fact, he says that to demand such evidence is to assume that his proposal
is not meaningful unless it can be verified (McHarry, 1978, p. 134). But this
is a mistake; one can easily grant that a multiverse model is meaningful,
and still ask whether there are good reasons to believe that it corresponds to
reality.
34. Similar issues arise for Parfit, who says that that his multiverse model is 'a
partial answer' to 'the problem of evil' (1991, p. 5, and see 1992, p. 423).
35. Almeida offers a succinct definition: 'A plenitudinous hyperspace is a col-
lection of many independent three-dimensional subregions in a connected
four-dimensional manifold' (2008, p. 158).
36. Megill (2011) offers a unique and interesting meta-argument: he claims that
the bare epistemic possibility of there being multiple universes can be shown
to defeat all arguments from evil. Space does not permit addressing it here;
I criticize it in Kraay, forthcoming.
37. Due to space constraints, I do not discuss Draper, 2004.
38. Walker (2009) defends what he calls 'the anthropic argument for atheism',
which is similar in structure to certain arguments from evil. He imagines a
multiverse-based response to his argument, and replies in the same vein as
Perkins and Almeida.
39. Coughlan (1987, pp. 17-18) mentions this move, and Turner (2003, p. 149)
appears to endorse it.
40. It is worth noting that many philosophers would deny that a universe fea-
turing gratuitous could be worthy of inclusion in a theistic multiverse, even
if it featured a preponderance of good over evil.
41. Of course, as we saw in section 2, Monton thinks it is not possible for God
to create all threshold-surpassing universes.
42. I am grateful to the Monash University philosophy department for hosting
me as a departmental visitor from February-July 2011. Special thanks are
due to Graham Oppy for illuminating discussions of key issues, and to Yujin
Nagasawa for helpful comments on an earlier draft. I am also grateful to the
Templeton Foundation and to Oxford University for supporting my research
in 2011-12.

References
Almeida, M. (2008) The Metaphysics of Perfect Beings (New York: Routledge).
Almeida, M. (2010) 'O'Connor's Permissive Universe', Philosophia Christi 12:
296-307.
Almeida, M. (2011) 'Theistic Modal Realism?', Oxford Studies in Philosophy of
Religion 3: 1-15.
Bricker, P. (2001) 'Island Universes and the Analysis of Modality', in G. Preyer
and F. Siebelt (eds), Reality and Humean Supervenience: Essays on the Philosophy of
David Lewis (Lanham: Rowman and Littlefield), pp. 27-56.
Carr, B. (ed.) (2007) Universe or Multiverse? (Cambridge: Cambridge University
Press).
Collins, R. (2007) 'The Multiverse Hypothesis: A Theistic Perspective', in
B. Carr (ed.), Universe or Multiverse? (Cambridge: Cambridge University Press),
pp. 459-80.
 The Theistic Multiverse: Problems and Prospects 161

Coughlan, M. ]. (1987) 'Must God Create Only the Best Possible World?', Sophia
26: 15-19.
Deutsch, D. (1997) The Fabric of Reality (Harmondsworth: Penguin).
Draper, P. (2004) 'Cosmic Fine-Tuning and Terrestrial Suffering: Parallel Problems
for Naturalism and Theism', American Philosophical Quarterly 41: 311-21.
Everett, H. (1957) '"Relative State" Formulations of Quantum Theory', Reviews of
Modem Physics 29: 454-62.
Forrest, P. (1981) 'The Problem of Evil: Two Neglected Defences', Sophia 20: 49-54.
Forrest, P. (1996) God without the Supernatural (Ithaca: Cornell University Press).
Gale, R., and A. Pruss (eds) (2003) The Existence of God (Aidershot: Ashgate).
Gilmore, C. (2006) 'Review of Hudson, The Metaphysics of Hyperspace', Notre Dame
Philosophical Reviews; available online at http://ndpr.nd.edu/review.cfm?id=
7783.
Greene, B. (2011) The Hidden Reality: Parallel Universes and the Deep Laws of the
Cosmos (New York: Knopf).
Gribbin, J. (2009) In Search of the Multiverse (London: Penguin).
Hudson, H. (2006) The Metaphysics of Hyperspace (Oxford: Oxford University
Press).
Kaku, M. (2005) Parallel Worlds: A Journey through Creation, Higher Dimensions, and
the Future of the Cosmos (New York: Random House).
Kraay, K. (2008) 'Creation, World-Actualization, and God's Choice among Possi-
ble Worlds', Philosophy Compass 3: 854-72.
Kraay, K. (2010a) 'Theism, Possible Worlds, and the Multiverse', Philosophical
Studies 147: 355-68.
Kraay, K. (2010b) 'The Problem of No Best World', in C. Taliaferro, P. Draper
and P. Quinn (eds), A Companion to Philosophy of Religion, 2nd edn (Oxford:
Wiley-Blackwell), pp. 481-91.
Kraay, K. (2011a) 'Theism and Modal Collapse', American Philosophical Quarterly
48:361-72.
Kraay, K. (2011b) 'Incommensurability, Incomparability, and God's Choice of a
World', International Joumal for Philosophy ofReligion69: 91-102.
Kraay, K. (forthcoming) 'Megill's Multiverse Meta-Argument', Intemational Journal
for Pl1ilosophy of Religion.
Leslie,]. (1989) Universes (New York: Routledge).
Lewis, D. (1986) On the Plurality of Worlds (Oxford: Blackwell).
Linde, A. D. (1986) 'Eternally Existing Self-Reproducing Chaotic Inflationary
Universe', Physics Letters B175: 395-400.
Mawson, T. (2009), Review of Timothy O'Connor's Theism and Ultimate Explana-
tion, Religious Studies 45: 237-41.
McHarry, J.D. (1978) 'A Theodicy', Analysis 38: 132-4.
Megill, J. (2011) 'Evil and the Many Universes Response', International Journal for
Philosophy of Religion 70: 127-38.
Monton, B. (forthcoming) 'Against Multiverse Theodicies', Philo.
O'Connor, T. (2008) Theism and Ultimate Explanation: The Necessary Shape of
Contingency (Melbourne: Wiley-Blackwell).
O'Connor, T. (2010a) 'Theism and the Ultimate Explanation: The Necessary
Shape of Consistency', Pllilosophia Christi 12: 265-72.
O'Connor, T. (2010b) 'Is God's Necessity Necessary? Replies to Senor, Oppy,
McCann, and Almeida', Philosophia Christi 12: 308-16.
162 Klaas f. Kraay

Oppy, G. (2008) 'Review of O'Connor, Theism and Ultimate Explanation', Notre

Dame Pl!ilosopl!ical Reviews; available online at http://ndpr.nd.edu/review.cfm?
id=13406.
Page, D. (2010) 'Does God So Love the Multiverse?', in M. Stewart (ed.), Science
and Religion in Dialogue: Volume One (Oxford: Blackwell), pp. 380-95.
Parfit, D. (1991) 'Why Does the Universe Exist?,' Harvard Review of Pl!ilosopl!y
(Spring): 4-5.
Parfit, D. (1992) 'The Puzzle of Reality: Why Does The Universe Exist?,' Times
Literary Supplement (3 July), reprinted in P. van Inwagen and D. Zimmerman
(eds), Metaphysics: Tile Big Questions (Oxford: Blackwell, 1998), pp. 418-27.
Perkins, R. K. (1980) 'McHarry's Theodicy: A Reply', Analysis 40: 168-71.
Rea, M. (2008) 'Hyperspace and the Best World Problem', Pl!ilosopl!y and Phe-
nomenological Research 76: 444-51.
Rees, M. (2001) Our Cosmic Habitat (Princeton: Princeton University Press).
Smolin, L. (1997) The Life of tile Cosmos (Oxford: Oxford University Press).
Steinhardt, P., and N. Turok (2007) Endless Universe: Beyond tile Big Bang
(New York: Doubleday).
Stewart, M. (1993) The Greater Good Defence: An Essay on the Rationality of Faitl!
(New York: StMartin's Press).
Tegmark, M. (2007) 'The Multiverse Hierarchy', in B. Carr (ed.), Universe or
Multiverse? (Cambridge: Cambridge University Press), pp. 99-125.
Turner, D. (2003) 'The Many-Universes Solution to the Problem of Evil', in R. Gale
and A. Pruss (eds), The Existence of God (Aldershot: Ashgate), pp. 1-17.
Vallentyne, P., and Kagan, S. (1997) 'Infinite Value and Finitely-Additive Value
Theory', Tile Journal ofPI!ilosopl!y 94: 5-26.
Vallicella, W. (1994) 'Review of M. Stewart, Tile Greater-Good Defence', Religious
Studies 30: 243-56.
Vilenkin, A. (2006) Many Worlds in One: The Search for Other Universes (New York:
Hill and Wang).
Walker, M. (2009) 'The Anthropic Argument against the Existence of God', Sophia
48:351-78.
Part IV
Religiou s Beliefs
7
How Relevant Is the Cognitive
Science of Religion to Philosophy
of Religion?
David Leech and Aku Visala

1. Introduction
On a standard view of the scope of philosophy of religion, philosophers
of religion are concerned with the meaning of religious claims and what
(if anything) they succeed in saying about the world. Naturalistic expla-
nations of religion, on the other hand, seek to causally explain how
religious ideas, beliefs and behaviours come about and persist in human
populations.
It is useful to distinguish three broad types of naturalistic explanations
of religion:

1. Evolutionary explanations that seek to explain the emergence of

religion in terms of evolutionary pressures.
2. Cognitive explanations that seek to explain why religious ideas are so
widespread in human populations in terms of cognitive systems that
process them.
3. Co-evolutionary explanations that seek to explain the dynamics of
religious traditions in terms of cultural evolution. 1

In this chapter we will be focusing on explanations along the lines of

(2) and (3) since researchers identifying with what is now called 'the
cognitive science of religion' (CSR) put the stress on cognitive factors
in religious belief formation and transmission. However, elements from
all three are de facto united (although with differing emphases) in the
theory building that goes on under the name of 'CSR'. 2
Are the results of the CSR relevant for philosophy of religion? As just
suggested, we should first make a preliminary distinction between

165
166 David Leech and Aku Visa/a

causes of beliefs and reasons for believing that some propositions are
true. Famously, philosophers caution against the danger of genetic fal-
lacy in these contexts: we cannot infer from the causes of our beliefs to
the truth or falsity of the propositions that are the objects of our belief. 3
In what follows we will be focusing on the truth-value and rationality
of belief in God and how CSR might be relevant for it. We acknowledge
that this does not exhaust the scope of the philosophy of religion and
that CSR might be relevant for other aspects of philosophy of religion
which we will not be discussing here.

2. An Argument for Atheism from the CSR

First let us recall the basic features of the CSR position. The starting
point of the approach is that the principal functions of the human
mind do not vary across cultures. This is due to the fact that human
minds emerge from similar biological foundations (brains) in basically
uniform natural environments. Further, our cognitive systems shape or
'constrain' the content of our beliefs and perceptions: our minds select
and transform information. Due to such processes, human cognition
produces recurrent patterns in human thinking and behaviour across
cultures by constraining and informing possible ways of thinking and
acting. According to CSR theorists, this is relevant to religion because
the most salient religious beliefs (e.g. beliefs about morally interested
gods and spirits) can be explained by evoking these systems.
It seems to us that two types of arguments against theistic belief
derived from this account have been proposed:

1. Arguments that infer from the existence of a sufficiently complete

causal explanation of religious belief (CSR) to the conclusion that
there is no God (no logical space arguments).
2. Arguments according to which the CSR shows that belief in God
is not a product of a belief-forming process that is sufficiently
truth-tracking and religious believers are, therefore, not rational in
assenting to theistic belief (unreliability arguments).

Notice that arguments of type 1 claim that God does not exist whereas
arguments of type 2 claim that it is irrational to believe in God.
Although it is often suggested that the CSR will make atheism more
probable than theism or even debunk theism in some profound way, the
arguments are seldom spelled out in any detail. Consider, for instance,
archaeologist Steven Mithen's position. It is a fact, Mithen notes, that
 How Relevant Is the Cognitive Science of Religion to Philosophy of Religion? 16 7

religiosity is a ubiquitous feature of human societies. This fact can be

explained in two different ways. (1) There exists a supernatural realm
inhabited by a supernatural agent (or agents) that has the power to act
in the natural world or even create it. If this explanation is true, then
religion is not natural but supernatural and God(s) exist(s). But there
is an alternative: (2) the pervasiveness of religiosity may just be due to
the fact that the human mind is prone to believe in the supernatural
and 'on-going activity of the universe and life are explained by entirely
natural processes'. 4 If this were the case, we could safely conclude that
'religion would be simply a curious invention of the human mind'. 5
Mithen concludes that

religious thought is uniquely associated with Homo sapiens and arose

as a consequence of cognitive fluidity, which was in turn a conse-
quence of the origin of language. In this regard, there appears to be
no need to invoke a moment of divine intervention that initiated the
start of a revelation. 6

This is but one instance of numerous tokens of reasoning that are on the
verge of committing the genetic fallacy. The argument seems to be that
because we now have a causal explanation for the emergence of reli-
gious ideas, we can exclude the possibility of supernaturalism being true
in general. The argument is that we now know why and how these puz-
zling phenomena have evolved and what biological and psychological
functions they serve, so we need not think that they represent reality
nor assent to their normative claims about how humans should live
their lives.
A deductive version of the argument against theism could go some-
thing along the following lines.

1. If we have a complete or sufficiently complete causal explanation of

how belief in God came about without any reference to the existence
of God, then God does not exist.
2. CSR can provide a complete or sufficiently complete causal account
of how belief in God came about without any reference to the
existence of God.
3. Therefore, God does not exist.

With respect to 2, the claim could be backed up as follows: for the pur-
poses of discovering whether theistic beliefs are justified, it does not
matter that we cannot yet explain more recent forms of theistic beliefs
168 David Leech and Aku Visa/a

and practices. It is enough if we have an adequate natural explanation

of their ancestor forms out of which the more recent forms have grown.
This assumes that all current theistic beliefs and practices in the world
have a causal ancestry directly involving earlier (indeed, prehistorical)
equivalents. At first sight this has a certain plausibility, since until very
recently all new religious movements have arisen within societies which
were already religious. So the atheist could maintain that it is sufficient
for her argument if she can explain naturalistically the prehistoric origin
of the basic religious belief disposition.
Let us say that the latter disposition is the disposition to believe that
there is at least one supernatural agent, and that we can be in relation
with it/them. Adapting a distinction in john Schellenberg, let us call this
basic supernaturalism, and distinguish it from elaborated supernaturalism.?
The atheist can then argue as follows: we have an explanation of basic
supernaturalism, and this is sufficient for establishing that disbelief con-
cerning theism is the most rational response. We do not need to have an
explanation of elaborated supernaturalisms, because they are causally
dependent on basic supernaturalism. In other words, they would not
have arisen if basic supernaturalism had not arisen. And CSR theory
provides a sufficient natural explanation of basic supernaturalism.
Now, what reasons do we have to think that 1 and 2 are true? Not
very many, we suggest. First of all, it seems that claim 1 commits the
genetic fallacy. We simply cannot infer God's nonexistence from the
mere fact that there exists a causal explanation for why we believe in
God that does not involve God. The causal explanation and God's exis-
tence could be compatible; if there is an in-principle incompatibility
between the two, this would require further argument. Claim 2 is also
quite problematic. The claim is open to what we will call the Inverted
God-of-the-gaps Objection. We will examine objections to the first premise
first and then move on to discuss the second.

3. Objection 1: Reasons and Causes

Most philosophers, both atheist and theist, are critical of clear-cut
debunking arguments of this sort, and they regard it as a mistake to
assume that simply by exposing the causal history of a set of beliefs
(even if complete and true) we get to the falsity of the propositions
involved. The reason why this way of arguing can be seen as problematic
is because we can assess the content of a belief (proposition) indepen-
dent of the causes which made people hold the proposition true (belief).
Every belief has a causal history of some kind and in this sense can to
 How Relevant Is the Cognitive Science of Religion to Philosophy of Religion? 169

some extent be explained causally, so the possibility of causal explana-

tion itself cannot count against the truth of a belief. The question is not
whether a belief, for instance, that God exists, has causes, but whether
those causes are of an appropriate kind. What kinds of causes are allowed
or not is a matter of considerable dispute in epistemology. We will return
to this point later. Suffice it to say that this dispute is ultimately about
the rationality of believing a proposition, not about the truth or falsity
of a proposition.8
For such reasons, most atheist and theist philosophers do not share
Mithen's faith in the power of causal explanations to settle philosophi-
cal disputes about the truth or falsity of religion.

4. Objection II: Inverted God-of-the-Gaps Objection

The next objection -the Inverted God-of-the-gaps objection (lGO) - has to
do with premise 2. In short, IGO consists of two claims. According to
the first, it is not true that CSR can provide us with a complete or suf-
ficiently complete causal explanation for belief in God. In other words,
the claim is, simply, that we have good reasons to think that the CSR
causal account of religious belief is not one which would allow us to
exclude the possibility of God being directly involved in the process. The
second claim included in IGO is that CSR only explains the emergence
of religion in a very generic way rather than explaining any particular
individual's belief in God.
First, the theist can claim that the CSR falls short of a complete,
or sufficiently complete, explanation. Let us consider the most trivial
point first. We simply cannot inspect the original Pleistocene context
in which religious behaviour supposedly evolved, and the evidence we
have by which we can reconstruct something about it is very slim.
We are reduced to informed speculation about what actually happened
at the point at which our primate ancestors gradually became us and
developed religiosity. So we need to have some way of knowing whether
the CSR account is the best speculation about how religious belief and
practice arose and became stabilized.
The theist defender of IGO can here argue that the mechanisms spec-
ified by CSR are not sufficient to have produced (and perhaps even
maintained) religious beliefs and practices in the Pleistocene, or at least
that we do not know whether they are sufficient. She can point out
that CSR fails to explain, for instance, why specific individuals have
the kind of beliefs they have about God. She can also point out that
CSR does not account for religious experiences and the details of more
170 David Leech and Aku Visala

sophisticated forms of religion. It can, therefore, be claimed that CSR

explanation concentrates on general tendencies in religious thinking
and behaviour and thus has scope-related, methodological limitations.
This generality is mainly a result of the CSR's commitment to a cultural
epidemiological approach to its subject matter inspired by Dan Sperber's
anthropological work. Sperber's epidemiology of representations model,
endorsed by CSR researchers, explains population-level tendencies to
religiosity rather than individual cases of religious beliefs or behaviours. 9
CSR consequently explains the 'cultural fitness' of representations or
their likelihood to spread, accounting for the memorable nature of reli-
gious representations by appeal to their fit with the properties of the
postulated nonconscious mental tools. 10
It follows that the CSR explanatory model underdetermines how
humans acquire and transmit religious beliefs and behaviours.U For
example, it is beyond the scope of CSR explanation to account for
why Trinitarian belief arose in communities of Christian theists, even
though it can predict generally that these communities would believe in
a supernatural agent or agents. Further, CSR cannot explain why given
individuals hold particular theistic beliefs, and whether those beliefs
are rationally justified. Particular theistic beliefs (along with all other
religious beliefs) are presumably supported by the cross-cultural human
intuitive biases and predilections specified by the CSR. However, they do
not all by themselves produce particular theistic beliefs. In other words,
it would be misconceived to suppose that CSR explanations specify all
of the causal pathways through which people come to hold theistic
beliefs.

5. A Probabilistic Version of the Argument

We have now considered some reasons to think that premises 1 and 2 in
the previous deductive argument are false. But perhaps if we reformulate
the atheistic argument in a probabilistic way, we could circumvent at
least some of the objections.
Consider the following version of the argument.

4. According to the CSR, the causes of our belief in the existence of God
are mostly natural instead of supernatural.
5. If God exists, it is likely that the causes of our belief in his existence
are mostly supernatural.
6. Therefore, God does not exist.
 How Relevant Is the Cognitive Science of Religion to Philosophy of Religion? 171

The deductive version of the argument examined in the previous section

attempted to show that God does not exist. The probabilistic version
only attempts to show that CSR theories give us a reason to prefer
atheism over theism.
The theist could respond in various ways here. First, she could grant
that the argument successfully increases the probability of atheism over
theism. The theist could, nevertheless, say that she has good indepen-
dent reasons to believe in the existence of God and claim 6 is not
enough to lower the probability of theism to such an extent that she
should cease believing it. The theist could concede that the causal path-
ways through which people come to believe in the existence of God
are largely natural and this would be something that seems unlikely on
the theistic hypothesis. However, she could still invoke other arguments
or considerations for the existence of God and maintain that theism
does significant explanatory work in other domains (by explaining the
existence, rationality and life-permitting qualities of the universe, for
instance).
The argument could also be challenged by using the IGO points
against premise 4. The opponent of the argument can claim that the
CSR in not complete enough to catch the notions of, for example, reve-
lation or personal religious experience. Nor is there anything in the CSR
that would give us a reason to rule these possibilities out as causes of a
particular individual's belief in God or as a basis of institutional religious
practice. The wielder of IGO would therefore conclude that the CSR does
not, by itself, give us reasons to conclude that belief in God is 'mostly'
(whatever that means) naturally caused.
Premise S is a premise that is seldom stated; instead it is usually situ-
ply assumed. What is presupposed here is that if God exists, he would
probably make his existence known to us and enter into relationship
with us in a way which was mostly supernatural, or at least supernatu-
ral in some relevant respect that can be observed either scientifically or
in some other commonsensical way. No one, not even the most hard-
ened supernaturalist, claims that there are no natural causes at work in
causing belief in God. What the defender of S would then mean is that
there are more natural causes at work than would be expected given the-
ism. That is to say, given theism we would expect God to interact with
humans and cause belief in him in a supernatural way. But what reasons
do we have to suppose this?
The supposition could be challenged by arguing that God could, as
far as we know, use natural causes (e.g. our ordinary cognitive sys-
tems, sociological and cultural mechanisms) to supplement supernatural
172 David Leech and Aku Visa/a

influence (e.g. action of the Holy Spirit, direct revelation etc.) to a large
extent. Since God is, by definition, omnipotent, He would have the
power to do this. The question is whether acting this way would some-
how be against God's purposes or be in contradiction with his goodness
(or some other relevant attribute). There might be a problem for the
theist here and we will be returning to it later.

6. Arguments from Unreliability

In addition to the aforementioned arguments, there is another kind of
argument that instead of attacking the truth of theism, is directed against
the rationality of belief in theism. Earlier we dubbed these unreliability
arguments, since they invoke the unreliability of the mechanisms that
produce belief in God. Take the following claim by psychologist Paul
Bloom as an example:

No finding from the cognitive science of religion can refute either

the existence of God or a theistic account of the origins of religious
belief. But, even so, psychological inquiry can still tell us something
about the rationality, or lack thereof, of religious believers, in the
same sense that it can tell us about the mental status of those who
believe in life on other planets. 12

Unreliability arguments seem at least prima facie plausible: if we can

demonstrate that belief in God is completely naturally caused, that is,
not in any direct way connected to God, then it seems that our belief
that God exists is not formed in a properly truth-tracking way.
Structurally speaking, unreliability arguments follow a common pat-
tern. In the literature, arguments of this kind have been dubbed evolu-
tionary debunking arguments. 13 Most arguments of this sort have to do
with morality, but the same strategy is now also being employed against
belief in God. According to philosopher Guy Kahane, evolutionary
debunking arguments are of the following form: 14

S's belief that p is explained by X

X is a process that does not track the truth
Therefore, S's belief that p is not justified

From the point of view of debunking arguments, it does not really mat-
ter whether God actually exists or not, because the truth of theism is not
a necessary condition for the rationality of belief in God.
 How Relevant Is the Cognitive Science of Religion to Philosophy of Religion? 173

We can now present a somewhat simplified unreliability argument

from the CSR for the irrationality of theistic belief.

7. S's belief that there is a God is explained by CSR (that is, by S's
intuitive cognitive mechanisms).
8. The cognitive mechanisms that cause belief in the existence of God
are not truth-tracking with respect to the existence or non-existence
of God.
9. Therefore, S's belief that there is a God is not justified.

The defender of such an argument can invoke several considerations in

support of premise 8. The basic claim would be that in some sense the
CSR shows that the cognitive mechanisms responsible for causing our
belief in God are not a sufficiently reliable source of beliefs. Here are just
a few possible ways to defend this claim. 15
First, one could claim that there needs to be a proper kind of causal
connection between a belief and its target for a belief to be justified.
But such a connection between our belief in God and gods seems to be
missing, since given CSR there is no need to think that God directly
caused humans to believe in him. Second, one could claim that our
hypersensitive agency detecting device (HADD) produces false beliefs
to such an extent that it should not be regarded as a reliable belief-
forming mechanism. If HADD is unreliable and it is indeed central in
the CSR account of the emergence of belief in God, then belief in God is
not justified. Third, it could be claimed that the mechanisms that CSR
identifies produce all sorts of wildly diverse beliefs about God and gods.
From this actual diversity of beliefs we can then infer that we should
not take such systems as reliable sources of beliefs about God.
Basically, theists have adopted two different strategies in answering
such claims. The first consists of demonstrating that there are no reasons
to suppose that the mechanisms described by the CSR are unreliable in
all religious contexts without begging the question against the theist.
With respect to the claim that there is no sufficient causal connection
between belief in God and its target, some have argued that CSR does
not really show this. For instance, Michael Murray claims that as far as
we know, God might be active further back in the causal nexus and set
up the initial physical conditions of the universe and guided evolution
in such a way that we would develop capacities to form beliefs about
him. According to this 'cognitive fine-tuning' view, there is no need for
God to intervene by special divine action at every step of religious belief
formation and thus the completeness of the CSR account is preserved. 16
17 4 David Leech and Aku Visa/a

The second argument from the unreliability of HADD can also be chal-
lenged. The claim would then be that since HADD is obviously reliable
in some contexts and not others, it cannot be shown to be unreliable
in religious contexts without first assuming that there is no God. Most
of the time our agency detection is correct in attributing agency. Simply
on the basis of the CSR, as far as we know, it could also be reliable in the
case of God. In addition, it could be insisted that since even CSR writers
themselves disagree as to what the explanatory scope and function of
HADD actually is, its role in explaining individual instances of belief in
God might actually be quite smallY
Third, Murray has also argued that the diversity of religious beliefs
does not really count against the reliability of the cognitive mechanisms
that produce them. This is because the diversity is actually produced by
the cultural information that elaborates the basic intuitions produced
by the cognitive mechanisms. The mechanisms themselves only pro-
duce certain intuitions to which different kinds of elaborated cultural
constructions can then attach themselves. Thus, the diversity of reli-
gious beliefs does not give us enough reason to think that the underlying
systems are unreliable sources of belief in God.
Finally, one could also add that our judgements about reliability or
unreliability of belief-forming mechanisms tend to be relatively vague
and depend heavily on background assumptions. All this suggests that
at least premise 8 of the standard unreliability argument could be
challenged.
But there might be another way to respond to arguments based on
unreliability. This strategy utilizes the IGO objection that was intro-
duced earlier. The previous objections were based on refuting premise 8
of the standard unreliability argument. The wielder of IGO will instead
attempt to refute premise 7 and with it the claim that CSR in fact
explains individual belief in God in a sufficient way. Since we have
already discussed IGO earlier, we will keep these remarks brief.
If IGO is correct, then it is the case with respect to any given indi-
vidual's belief in God that CSR underdetermines the causes of that
individual's belief in God. As we pointed out earlier, the rationality of
theism depends at least to some extent on reasons, arguments and expe-
riences that are not explained by CSR-type mechanisms. These causes
cannot be accounted for in terms of non-conscious cognitive systems.
If this is the case, then even if CSR turns out to be true and we could
demonstrate that the cognitive mechanisms that CSR reveals are unreli-
able, the theist might still be justified in believing in God on the basis
of these other reasons. 18
 How Relevant Is the Cognitive Science of Religion to Philosophy ofReligion? 175

We thus conclude that current unreliability arguments along the lines

of premises 7-9 have been unable to show that the cognitive systems
which more or less cause our belief in God (according to CSR) are so
radically unreliable that theistic belief is irrational.

7. The Rationality of Theism and Sensus Divinitatis

Attempts to support theism with findings of the CSR have been rare but
not inexistent. These attempts have mainly focused on the rationality
of theism, rather than its truth. For instance, such an argument is more
or less explicit in an article entitled 'What Theology Might Learn (and
Not Learn) from Evolutionary Psychology: A Postliberal Theologian in
Conversation with Pascal Boyer' by theologian Niels Henrik Gregersen.
Gregersen sees no particular theological problem in the claim that reli-
gious beliefs are caused by our ordinary, biologically produced cognitive
mechanisms. However, he insists (correctly, in our view) that the truth
of religious beliefs should not be evaluated in terms of their biological
origin. 19
Gregersen is mainly interested in the origins and function of rational-
ity in different domains of life. If religious rationality has cognitive and
evolutionary roots in the sense that it is not its own distinct domain
of rationality, all the better, he claims, for belief in God. 2° First, the
fact that religious concepts are easy to understand and process coincides
with the judea-Christian idea that humans are created in the image of
God and thus endowed with abilities to be in communication with him.
This ability to understand God is common to all humans irrespective
of their religious views. So the fact that CSR finds natural mechanisms
that explain the pan-human tendency to believe in godlike agents is
something that theism would give us reason to expect.
Simply put, Gregersen's argument is that if CSR is correct and religion
is indeed a by-product of non-religious cognitive systems (such as HADD
and theory of mind), then it is the case that belief in God is generated by
the same cognitive mechanisms as most of our non-religious beliefs. If,
then, to avoid global scepticism, we should hold the deliverances of our
ordinary belief-forming mechanisms prima facie justified, and if belief
in God is a product of these ordinary mechanisms, then it is prima facie
justified in the same way as all other basic beliefs are.
The problem with proposals such as Gregersen's is that they do not
get the theist very far: prima facie justification can be easily removed
by defeaters of different kinds (reasons, arguments, new experiences).
At most, it could show that religious reasoning in general is justified.
176 David Leech and Aku Visa/a

But it would do nothing to increase the likelihood of theism being

the case.
Psychologist justin Barrett has made arguments along similar lines.
If belief in God is largely produced by normal human cognitive systems
and if we generally trust these systems, then we should not suspect them
in the case of belief in God. Barrett compares belief in God and belief in
other minds and argues that they are psychologically speaking on the
same level. 21 Both types of beliefs - belief in gods and belief in other
minds- (1) lack empirical support, (2) are believed both reflectively and
intuitively and (3) are mostly universal. Barrett points out that it is not
scientific evidence that makes people believe in minds but rather the
ordinary outputs of their cognitive systems. Thus, the belief in minds
and belief in God should be held as foundational or basic. Foundational
beliefs- belief produced by our ordinary belief-forming mechanisms-
need no outside justification to be prima facie rational. What is needed
is that there is no immediate defeater to them. Further, as they are both
basic beliefs, if we doubt the one on the basis of lack of propositional
evidence we should also doubt the other.
This line of argument and the .!dea of the prima facie justification
of basic beliefs is indebted to Reformed Epistemology. Cognitive the-
ories, according to Barrett, may have ended up revealing a form of
what traditional Calvinist theology referred to as the sensus divinitatis,
or 'God-sense' (a God-given capacity to know God). 22 According to this
traditional theological view, the sensus divinitatis is the source of our
basic, non-inferential beliefs about the nature and existence of God and
it is providentially guided by God. Sensus divinitatis is something that all
human beings have, although some have refused to listen to it. 23
Notice that those who link sensus divinitatis with the CSR-type mech-
anisms have two different arguments going on at the same time. The
first argument, as we have seen, is for the rationality of theism and is
based on the Reformed Epistemology account of basic beliefs and their
justification. But in addition to this, there is also an argument of a differ-
ent type- an argument that is rarely spelled out. It would go something
like this. The reasons for postulating the existence of sensus divinitatis
are mainly theological and have been around for a considerable time.
If it turns out to be the case that CSR confirms the existence of such a
mechanism, then CSR would provide a reason to think that theism is
true. In such a case, CSR would function as an independent source of
knowledge that would confirm what is being claimed on a theological
basis, thus giving support to theism.
 How Relevant Is tiJe Cognitive Science of Religion to Philosophy of Religion? 177

Although it might seem promising to link CSR results with the sen-
sus divinitatis doctrine, on closer inspection several problems arise. First,
there is the problem of the IGO. That is to say, belief in God might be
too content-specific to be a product of the mechanisms described by the
CSR (recall Murray's defence). The theist would need a lot more than
just sensus divinitatis to get to full-fledged faith in God. Of course, the
theist can claim that sensus divinitatis does not give us a fully fledged
concept of God in the first place, but rather a vague feeling that 'there
might be something out there', to which she can add other sources for
beliefs about God.
Second, the CSR only makes it probable that some kind of supernat-
ural belief arises, not inevitable. Some theists might not like the idea
of God as a risk-taker. It is true that our cognitive architecture seems to
have a survival advantage over some other types of architectures so it is
probable that something of the sort would come about through natural
selection, but this is by no means certain or necessary. All products of
evolution are to some extent contingent. Therefore, if CSR-type meclla-
nisms were identical with sensus divinitatis, there is always the chance
that sensus divinitatis might not be realized, at least not by natural
selection.
Third, if God chooses to be in contact with human beings only
through the mechanisms described by the CSR, then it would seem that
God would not be in communion with humans in any direct way. Most
theists, however, believe that God is somehow more intimately related
to us, indeed, that they enjoy a personal relationship with him. If we
claim that God interacts with us only through the mechanisms that CSR
describes, then God would come out looking like a distant and deceiv-
ing God. 24 Given these worries, the theist should probably be hesitant
in identifying sensus divinitatis with CSR-type mechanisms.

8. Objection: The Deceiving God

The last point about the possibility of a deceiving or distant God can
be expanded into a more general argument against theism from CSR.
We call this the Deus Deceptor (DD) argument. For instance - and this
picks up on the intuition behind the probabilistic argument in the pre-
vious section - the CSR account might be interpreted as entailing that
humans do not relate directly to a supernatural God, but rather to a
more or less adequate simulacrum of him produced by their cognitive
architectures. But the atheist can object at this point that this causal
178 David Leech and Aku Visa/a

account of religious belief is not compatible with God's putative moral

nature. lf God is supposed to be good, non-deceiving and so forth, then
he should not produce belief in us in the way specified by CSR. So if a
theist suggests that he does produce human beliefs about him in this
way, this should strike the theist and atheist alike as improbable given
what we take to know about the nature of God. In that case, the CSR
account does not render theism more but less probable.
Perhaps the theist could reply at this point that God intervened in
human history at a certain point and henceforth the beliefs he put into
certain humans' minds were correct ones, and indeed the result of direct
divine action. But the atheist can insist that even if this were the case,
it would still be incompatible with God's benevolent nature to have
allowed vast amounts of false, merely natural, religious beliefs. First,
he would have deprived millennia of humans from enjoying meaning-
ful direct loving relationship with him. Second, it would presumably
surprise theistic expectations to learn that God at once allowed mil-
lennia of humans - who, presumably, were non-resistant to the Divine
call- to have false merely natural'religious' beliefs, and selectively put
involuntary true religious beliefs into the minds only of some. 25
The theist can perhaps further object that God may have been giving
earlier humans a foretaste of himself through these beliefs which arose
merely through the mechanisms specified by CSR. However, she must
still face the question of why God did this through simulacra, and did
not reveal himself personally even if in a limited way.
Indeed, the non-theist could produce an argument which would con-
firm prior non-theist assumptions, and thus represent a reversal of the
theist's sensus divinitatis argument. This argument would go as follows:
the non-theist does not find the CSR explanation of religious belief
and behaviour improbable in relation to her background assumptions
(unguided evolution, comprehensive natural causation, natural human
cognitive architectures etc.). By contrast, the traditional theist would
(or should) find the CSR explanation of religious belief and behaviour
improbable in relation to their background assumption that there is an
omnibenevolent, non-deceiving God who would choose to enter into
loving, meaningful, direct relationship with humans.
We can see the extent of the problem for the theist more clearly if
we see it in terms of the problem of divine hiddenness. According to,
for instance, Schellenberg, any evidence for the absence of loving per-
sonal relationship between God and humans counts centrally against
theism. If theism is the case, a relationship with God will be the ulti-
mate good for humans. lf we grant this, then it would look like God's
 How Relevant Is the Cognitive Science ofReligion to Philosophy of Religion? 179

moral attributes would count as reasons for thinking that God would
not realize certain states of affairs if he existed. 26
But the CSR account, as we have seen, can plausibly be interpreted
as implying that God, if he exists, allowed very inadequate simulacra
of relationship with him to exist in human minds in the place of direct
personal relationship for most of human history. Two possibilities con-
cerning how God might have done this come to mind, neither of which
help the case for theism.
First, perhaps God created humans incapable of entering into direct
conscious relationship with God by creating them without the requisite
cognitive and affective capacities to do so. But it is hard to see why a God
with the relevant moral properties would have done this, since it renders
(non-resistant) humans unable to enter into genuine conscious relation-
ship with God, which is the opposite of what the theist would expect.
Second, perhaps God created humans with the capacity to enter into
direct conscious relationship with him, but decided for some reason
to replace this with simulated relationships for most of human history
(including our long prehistory), only enabling a true, deeply mean-
ingful, direct relationship with the advent of Judaeo-Christianity. But
then the question suggests itself: if God could have enabled true, deeply
meaningful, direct relationship for all humans, why did he not do so?
Perhaps here the theist would have to develop some defence of the
necessity of humans coming to more and more adequate knowledge
of God gradually. Possibly a version of theistic evolutionism applied to
humans' cognitive and affective capacities might work here. But the
details of this would have to be fleshed out, and in any case one can
hardly talk about gradualism here, since the difference between having
a simulated experience of God and having a real, direct experience of
him is not one of degree.
We might finally mention a further way in which CSR could frus-
trate theistic expectations which was briefly mentioned above. CSR
theory would seem to imply that large-scale human non-theism (or
deeply imperfect theism) over countless generations cannot plausibly
be blamed on human voluntary resistance to God. The problem can
be stated like this: for most of human history and prehistory, humans
involuntarily held religious beliefs which were either not theistic ones
or very imperfectly 'theistic'. The involuntariness is the problem here
for the theist who wants to reconcile her theism with CSR. On the
CSR account, these humans cannot plausibly be interpreted as resist-
ing deep, meaningful relationship with God and the theistic beliefs
which would arise involuntarily from such a relationship. As a matter
180 David Leech and Aku Visa/a

of fact - and through no fault of their own, but as a result of their cog-
nitive architectures - they did not possess the appropriate beliefs about
a super-person distinct from nature who has the omni-attributes. So not
all nonbelief in the theistic God is the result of voluntary resistance to
God (i.e. sin).
The basic point here is perhaps this: even if the God of love might
accept with regret that humans might resist awareness of him, he would
not - if he was perfect - himself make direct, conscious relationship
with himself impossibleP But the theist who accepts CSR seems to be
constrained to say that this appears to be what God has done (at least
for prehistory and prior to the Abrahamic revelations).
In conclusion, then, it could be said that there are prima facie good
reasons for thinking that while the existence of just any naturalistic
causal account of religious belief formation does not make theism more
or less likely, certain features of the CSR account look like they could
make theism more improbable than not, even while they do not lend
active support to atheism.

9. Conclusions
We think the implications of CSR for the philosophy of religion, con-
trary to how they have often been presented, are very inconclusive. CSR
theory can be something of a double-edged sword for atheists and the-
ists hoping to recruit it to support their respective positions. The same
CSR-based arguments which seem to favour one side in the debate can,
appropriately transformed, favour the other side.
For instance, IGO can be used to show that there are present gaps in
naturalistic explanations of religion, but this can also be re-converted
by the atheist into a standard God-of-the-gaps objection to theism: the
gaps might be explained by the fact that researchers have so far given
insufficient attention to this field (this can gain extra plausibility from
the fact that the CSR field is new). Likewise, appeals to CSR as confirma-
tory of sensus divinitatis can also be turned against the theist as actually
offering reasons for thinking that the new naturalistic explanations of
religion might actually disconfirm theological expectations.
So, on the basis of this survey of arguments it seems to us reasonable
to say that the CSR fails to offer unequivocal support either to atheism
or to theism, and indeed its implications for the atheism-theism debate
are somewhat unclear. We therefore suggest that the CSR as it currently
stands provides very little or no reason to prefer atheism over theism,
but also no reason to prefer theism over atheism.
 How Relevant Is the Cognitive Science of Religion to Pllilosophy of Religion? 181

Notes
1. For an overview of explanations along these lines, see Schloss and Murray,
2009.
2. See, e.g., Lawson and McCauley, 2002; Boyer, 1994, 2001; Barrett, 2004;
Whitehouse, 2004; Atran, 2002.
3. By contrast, some prominent authors have recently tried to imply that
because an empirically supported naturalistic explanation of religion now
exists, religious truth-claims are once and for all debunked. See for instance
Dawkins, 2006 and Dennett, 2006.
4. Mithen, 2009, p. 11.
5. Similar suggestions have been made by, e.g., Bering, 2010.
6. Mithen, 2009, p. 27.
7. Schellenberg, 2005, pp. 37-8.
8. Cf. for instance atheist philosopher John Mackie: 'Even an adequate, uni-
fied, natural history which incorporated all these factors would not in itself
amount to disproof of theism. As William James and many others have
insisted, no account of the origin of a belief can settle the question whether
that belief is true or not' (1982, p. 197).
9. See, e.g., Sperber, 1996.
10. Indeed, critics have questioned CSR's explanatory power precisely on
account of this generality, and suggested that it may be the type of large-
scale anthropological programme which has little to give to the study of
particular cultural traditions because the generalizations it produces are not
very interesting. See Laidlaw, 2007 and Day, 2007.
11. This point is developed further in Leech and Visala, 2011. For a more
comprehensive treatment, see Visala, 2011.
12. Bloom, 2009, p. 126.
13. For evolutionary debunking arguments of religion, see Barrett, Leech and
Visala, 2010 and Leech and Visala, 2011. See also essays in Schloss and
Murray, 2009.
14. Kahane, 2011, p. 103.
15. For arguments and responses see Barrett, 2007; Murray, 2007, 2009. For an
overview, see Leech and Visala, 2011.
16. Murray and Goldberg, 2009.
17. On HADD's explanatory scope, see Barrett, 2004, p. 39.
18. This response to unreliability arguments is developed in Leech and Visala,
2011, pp. 180-3.
19. Gregersen, 2006, p. 320.
20. Gregersen, 2006, pp. 312-14.
21. Barrett, 2004, pp. 95-105. Barrett's argument draws from Alvin Plantinga
(1990).
22. See, Barrett, 2009, p. 98. The sensus divinitatis argument is presented in more
detail by Clark and Barrett, 2010.
23. See Plantlnga's treatment of sensus divinitatis in Plantinga, 2000,
ch. 6.
24. These problems are outlined in Leech and Visala, 2011.
25. On the hard problem of non-resistant disbelief for theism, see Schellenberg,
2007, pp. 198-206.
182 David Leech and Aku Visa/a

26. Schellenberg, 2007, p. 198.

27. Schellenberg, 2007, p. 202.

References
Atran, S. (2002) In Gods We Trust: The Evolutionary Landscape ofReligion (New York:
Oxford University Press).
Barrett, J. (2004) Why Would Anyone Believe in God? (Walnut Creek: AltaMira
Press).
Barrett, J. (2007) 'Is the Spell Really Broken? Bio-Psychological Explanations of
Religion and Theistic Belief', Theology and Science 5: 57-72.
Barrett, J. (2009) 'Cognitive Science, Religion and Theology', in M. Murray and
J. Schloss (eds), The Believing Primate: Scientific, Philosophical, and Theologi-
cal Reflections on the Origin of Religion (New York: Oxford University Press),
pp. 76-99.
Barrett, J., D. Leech and A. Visala (2010) 'Can Religious Belief Be Explained Away?
Reasons and Causes of Religious Belief', in Ulrich Frey (ed.), The Nature of God-
Evolution and Religion, vol. 1 (Marburg: Tectum Verlag), pp. 75-92.
Bering, J. (2010) The God Instinct: The Psychology of Souls, Destiny and the Meaning
of Life (London: Nicholas Brealey).
Bloom, P. (2009) 'Religious Belief as an Evolutionary Accident', in M. Murray
and J. Schloss (eds), The Believing Primate: Scientific, Philosophical, and The-
ological Reflections on the Origin of Religion (New York: Oxford University
Press).
Boyer, P. (1994) The Naturalness of Religious Ideas: A Cognitive Theory of Religion
(Berkeley: University of California Press).
Boyer, P. (2001) Religion Explained: The Evolutionary Origins of Religious Thought
(New York: Basic Books).
Clark, K. ]., and J. Barrett (2010) 'Reformed Epistemology and the Cognitive
Science of Religion', Faith and Philosophy 27: 174-89.
Day, M. (2007) 'Let's Be Realistic: Evolutionary Complexity, Epistemic Probabil-
ism, and the Cognitive Science of Religion', Harvard Theological Review 100:
47-64.
Dawkins, R. (2006) The God Delusion (London: Bantam Press).
Dennett, D. (2006) Breaking the Spell: Religion as a Natural Phenomenon (New York:
Viking).
Gregersen, N. H. (2006) 'What Theology Might Learn (and Not Learn) from Evo-
lutionary Psychology: A Postliberal Theologian in Conversation with Pascal
Boyer', in F. LeRon Shults (ed.), Evolution of Rationality: Interdisciplinary Essays
in Honor of/. Wentzel Van Huyssteen (Grand Rapids: William B. Eerdmans),
pp. 306-26.
Kahane, G. (2011) 'Evolutionary Debunking Arguments', Nous 45: 103-25.
Laidlaw, J. (2007) 'Well-Disposed Social Anthropologist's Problem with the "Cog-
nitive Science of Religion"', in H. Whitehouse and ]. Laidlaw (eds), Reli-
gion, Anthropology and Cognitive Science (Durham: Carolina Academic Press),
pp. 211-46.
Lawson, E. T., and R. McCauley (2002) Bringing Ritual to Mind: Psychological
Foundations of Cultural Forms (Cambridge: Cambridge University Press).
 How Relevant Is the Cognitive Science of Religion to Philosophy of Religion? 183

Leech, D., and A. Visala (2011) 'The Cognitive Science of Religion: A Modified
Theist Response', Religious Studies 47: 301-16.
Mackie, J. L. (1982) The Miracle of Theism: Arguments for and against the Existence
of God (Oxford: Clarendon Press).
Mithen, S. (2009) 'The Prehistory of the Religious Mind', in N. Spurway (ed.),
Theology, Evolution and the Mind (Newcastle: CSP), pp. 10-30.
Murray, M. (2007) 'Four Arguments That the Cognitive Psychology of Religion
Undermines the Justification of Religious Belief', in J. Bulbulia eta!. (eds) The
Evolution of Religion: Studies, Theories, and Critiques (Santa Margarita: Collins
Foundation Press), pp. 365-70.
Murray, M. (2009) 'Scientific Explanations of Religion and the Justification of
Religious Belief', in M. Murray and]. Schloss (eds), Tile Believing Primate: Scien-
tific, Philosophical, and Theological Reflections on the Origin ofReligion (New York:
Oxford University Press), pp. 168-78.
Murray, M., and A. Goldberg (2009) 'Evolutionary Accounts of Religion: Explain-
ing and Explaining Away', in M. Murray and ]. Schloss (eds), The Believing
Primate: Scientific, Philosophical, and Theological Reflections on the Origin of
Religion (New York: Oxford University Press), pp. 179-99.
Plantinga, A. (1990) God and Other Minds: A Study of the Rational Justification of
Belief in God (Ithaca: Cornell University Press).
Plantinga, A. (2000) Warranted Christian Belief (New York: Oxford University
Press).
Schellenberg,]. (2005) Prolegomena to a Philosophy ofReligion (Ithaca and London:
Cornell University Press).
Schellenberg, ]. (2007) The Wisdom to Doubt (Ithaca and London: Cornell
University Press).
Schloss, ]., and M. Murray (eds) (2009) The Believing Primate: Scientific, Philo-
sophical, and Theological Reflections on the Origin of Religion (New York: Oxford
University Press).
Sperber, D. (1996) Explaining Culture: A Naturalistic Approach (Oxford: Blackwell).
Visala, A. (2011) Theism, Naturalism ami the Cognitive Study of Religion: Religion
Explained? (Aldershot: Ashgate).
Whitehouse, H. (2004) Modes of Religiosity: A Cognitive Theory of Religious Trans-
mission (Walnut Creek: AltaMira Press).
8
The Rationality of Classical
Theism and Its Demographics 1
T. J. Mawson

Whatever the naturalness of the religious impulse, it is obvious that

nurture directs it and may either enhance or diminish it. Historians
and pre-historians have long argued that varieties of polytheism (and/or
ancestor- and spirit-worship) were extremely widespread, if not univer-
sal, in the earliest recoverable periods of human development. And it
has long been the consensus that monotheism first emerged, fitfully
but severally, in what is sometimes called the Axial Age, before gain-
ing eventual hegemony over other supernaturalist views. Finally, there
would be little dispute that, with the modern age, naturalism- the view
that there is no supernatural order whatsoever - has emerged from the
shadows as the view of a significant minority; in some countries it has
even managed to become the majority view.
The empirical facts of the demographics of theism, both over time -
considering humanity as a whole, as it has developed - and over
space - considering the set of humans extant at any one time - have
been supposed by some to have implications for the rationality of clas-
sical theism. It is these supposed facts and their supposed implications
that I wish to explore in this chapter. Obviously facts of the past are
harder to discern than facts of the present, ceteris paribus. For this rea-
son, in what follows I shall focus on the contemporary demographics
of theism; the same points that may be made about these facts may be
made mutatis mutandis about the facts as we may think we discern them
concerning the historical and pre-historical demographics of theism.
In a moment, then, I shall sketch the contemporary demographics
of theism as I perceive them, marking the divide between those who
subscribe to some variant of theism and those who do not (lumping

184
 The Rationality of Classical Theism and Its Demographics 185

together into the latter category both agnostics, understood as those

who fail to have a belief either way, and atheists, understood as those
who believe that there's not a God). Before I sketch this pattern, it is as
well for me to make mention of some worries that one may have with
the method that leads me to perceive it, namely- in a manner compara-
ble to much contemporary so-called 'experimental philosophy'- that of
aggregating the results of various surveys, polls and responses to ques-
tionnaires, to make mention of these worries and to do something to
defuse them. 2

II

It must be conceded that it is certain that people responding to a ques-

tion asking them whether or not they believe in God understand 'God'
in a variety of ways and thus understand themselves to be answering a
variety of questions as they reply to this 'one'. This variation of under-
standings cannot but add 'noise' to the data collected thereby. It is, in
addition, a fact notable from a number of surveys that a proportion of
those who say they do not believe in God nevertheless do not describe
themselves as atheists in response to other questions. It may be that they
have agnosticism in mind as their preferred self-characterization, but
one hypothesis which has gained support is that, even in the largely sec-
ular world, there is still a certain social stigma supposed by respondents
to be associated with labelling oneself an atheist, a stigma that is not
supposed to attach to one as much if one says of oneself merely that one
doesn't believe in God. If that is so, the supposed stigma may- one now
inclines to hazard- be supposed to attach to one to some extent even if
one merely says of oneself that one doesn't believe in God. And, if the
last thought hazarded is right, belief in God is therefore likely to be over-
reported by surveys. A variant on this worry would be the following:
in societies where theism is state-sponsored (e.g. Iran), it seems likely
that social/cultural pressures would lead to an over-reporting of theistic
belief, even when the anonymity of the respondents was as assured as it
ever could be to those who have lived under covert surveillance for so
long. Conversely, in societies where naturalism has been state-sponsored
(e.g. North Korea), and the same difficulties attend assuring the respon-
dents that they will remain anonymous, theistic belief is likely to be
under-reported.
There are a number of variants of worries of the sort outlined in
the previous paragraph and I think that one thing one must grant in
response to them is that they certainly do point to barriers that lie
186 T. f. Mawson

between us and knowledge of the prevalence of belief in theism ver-

sus agnosticism and atheism. However, another thing one may insist on
is that these barriers are not insuperable in principle and, I would fur-
ther suggest, that they have been overcome in practice by a variety of
means, most obviously in the case of the thought hazarded that it might
be being supposed by respondents that a 'social stigma' would attach
to someone who labelled themselves an 'atheist', one might be able
to assure anonymity to the participants (perhaps through making the
survey online); one might perhaps misrepresent oneself whilst asking
the questions, as for example a member of the National Atheist Asso-
ciation, keen to find out just how many right-thinking people such as
oneself there are in the country; and so forth. (Of course the last ruse
might lead to an under-reporting of theism, but one could compare and
contrast the results of surveys done with the ruse and ones done with-
out.) With regard to the worries one might have about respondents from
Iran and North Korea, they obviously do not generalize to other coun-
tries, or at least they do not apply to other countries to the same extent;
and indeed the worries can be mitigated to some extent even for those
countries for which they are most pressing; for example, one may ask
ex-patriots what they believed when they were in the country in ques-
tion (although of course care must be taken to control for the fact that
they may have been caused to become ex-patriots precisely because they
had a religious/irreligious view that they feared would lead to persecu-
tion if they stayed in their home country). In short, drawing on a variety
of surveys is the best way to iron-out problems of this sort, problems
which may indeed affect any one of them.
My main worry as I started reading the questions posed by surveys on
this issue was that people who said that they believed that there was a
God (and, even more so, people who said that they believed that there
was not a God) would in general have such confused notions of what
sort of God it was they were believing did or didn't exist that there'd
be no way of tying their answers to any conclusions about the preva-
lence of belief in the classical theistic God. I found this worry mitigated,
however, by the fact that Gallup Junior and Lindsay, for example, had
asked people whether they believed in God or a 'Higher Power'; what,
one might think, could be more inclusive than that? A lot of people
who aren't theists would therefore have been swept into theistic com-
pany in answering that question affirmatively, but, because of this, this
survey cannot have over-reported the proportion of respondents who
are agnostics and atheists with respect to classical theism. That being
the case, I was very surprised to learn that the survey revealed that as
many of 39 per cent of British people do not believe in God or a 'Higher
 Tile Rationality of Classical Theism and Its Demographics 187

Power'. 3 At least 39 per cent of British people, then, one may confidently
infer, are agnostics or atheists with respect to classical theism; were the
classical theistic notion of God to be described to them, they would say
that they did not believe that anything answered to it.
Finally, it is often suggested that response rates to questionnaires
are low enough to render the resulting data statistically questionable
and one may reasonably suspect that non-random factors affect who
responds to at least some questionnaires. In addition, there can be
deliberate attempts to skew the data, motivated by all sorts of pur-
poses. There was recently a campaign - not, one suspects, directed by
a desire for truthful reporting - to encourage people, when filling out
the 2001 British census, to list 'Jedi' as their religion. ln the end a - to
me staggering- number of people, 390,127, did state that they were Jedi
on the form, prima fade suggesting that there were more Jedi in the UK
at the turn of the millennium than there were Sikhs, Buddhists or Jews.
According to the same census, a full two per cent of the population of
my home city, Oxford, were Jedi in 2001. 4 However, in twenty years of
walking around the city (even walking on an 'inter-faith' walk down the
multi-cultural Cowley Road) I've never seen any Jedi at all, or at least
seen any that I recognized as such from their appearing in the apparel
that the Star Wars films would lead me to think is de rigueur for them.
What then to say of this batch of worries?
Again, it seems to me that one must concede that there are prob-
lems of the sorts outlined in the previous paragraph. But again it seems
to me that these problems are not entirely insuperable. There are well-
known mathematical techniques for assessing whether or not results
are statistically significant and for controlling for at least some of those
non-random factors which it has occurred to one may affect response
rates. For example, the filling out and returning of the British census
is a legal requirement, something which reduces - even if it does not
entirely eliminate - grounds for worries of this sort. However, it must
be admitted that campaigns aimed precisely at misleading those con-
ducting surveys can remain a step ahead of techniques to avoid being
misled. That is, I cannot but believe, what happened in the case of what
has become known as 'the Jedi census phenomenon' (it was repeated to
varying extents around the Anglophone world at about the same time
as the 2001 British census). l just cannot believe - despite the data -
that on average one in fifty of the people I see as l walk around Oxford's
streets is a Jedi.
In short, then, I would say in response to worries of the sorts I have
briefly listed above that they give us reasons for caution when drawing
conclusions from the results of surveys of contemporary theistic belief
188 T. f. Mawson

versus agnostic/atheistic belief, but these reasons can be overstated and

these reasons, where and when they are most pressing (and admittedly
sometimes they are pressing, e.g. in assessing how many people in North
Korea really are atheists or how many people in Oxford really are jedi),
do not undermine the confidence we may reasonably place in some sur-
veys and in the insight they give us into some locations. All of that
being so, then, I would maintain that we may sketch with a fair degree
of confidence the following picture of the contemporary demographics
of theism.

III
There are hardly any (less than ten per cent) agnostics or atheists in
North and South America (not including Canada), Africa, the Middle
East, India and Southeast Asia. There are more (10-50 per cent) in most
European countries and in Canada and Australia/New Zealand; Rus-
sia, with 30-40 per cent agnostics and atheists, follows the pattern for
Europe. In the Nordic countries, the percentage of agnostics and athe-
ists is highest - at 50-70 per cent. To this general pattern, the most
striking exceptions are Vietnam and japan, where there is much more
agnosticism and atheism than their geographical positioning would lead
one to expect (70 per cent plus in Vietnam, 60 per cent plus in japan).
Also perhaps exceptional is the Republic of Ireland, where there appears
to be much more theism than in the rest of Europe. (Personally, I am
sceptical that Ireland really is an exception to the European rule, remem-
bering the truth behind the oft-told anecdote of the tourist travelling in
Northern Ireland who is cornered in an alley by a gang and asked threat-
eningly, 'Are you protestant or catholic?' Not knowing the allegiance
of the gang but knowing simply that to pick the wrong one of these
would be his downfall, he replies, 'I'm an atheist.' The questioner comes
back instantly with, 'Ah, but are you a protestant atheist or a catholic
atheist?')
On this basis then we may assert with a high degree of confidence
that theism is currently very unevenly spread across the surface of the
globe: speaking always generally and ceteris paribus, we may say that
people born in Nordic countries or in Vietnam or japan are unlikely
to become theists; people born in Europe generally (not including the
Nordic countries) are slightly more likely to become theists than not;
and people born into the Americas (not including Canada), Africa and
the Middle East are very likely to become theists. In what follows I shall
therefore take these to be established empirical facts. I shall not question
 The Rationality of Classical T/Jeism and Its Demographics 189

them further. My question is what implications, if any, do these facts

have for the rational defensibility of classical theism?

IV

I am not aware of anyone who has advanced an argument to the effect

that facts of this sort have positive implications for classical theism; I am
aware of two lines of argument that have been advanced for thinking
that they have negative implications. I shall describe these two lines of
argument, although in this paper I shall only deal in detail with the
second, having dealt with the first elsewhere. 5
First, it is sometimes argued that these facts present a direct problem
for knowledge of theism, in particular they give one reason to suppose
that one cannot know that theism is true even if theism is true, because
they show that the facts which lead to belief in theism are too distant
from and uncorrelated with the facts that make theism true, too dis-
tant and uncorrelated for the belief in theism that results from them to
count as an item of knowledge. I will not repeat my counterarguments
to such suggestions here, just their conclusion. On the assumption that
the empirical facts are as I've sketched them and that theism is true, it
follows that- again, it should be stressed, only speaking 'in general' and
ceteris paribus - people born in Vietnam have a lesser chance of com-
ing to know that it's true than those born in Great Britain, who in turn
have a lesser chance than those born in the United States of America.
However, that does not imply that those who do come to know it don't
know it after all: the 'higher-level' chanciness (if indeed that's the best
way of thinking of it) affecting whether or not one's a knower doesn't
generate a 'lower-order' chanciness in oneself, one that disables one
from being a knower. We can see this most quickly by noticing that all
sorts of beliefs which obviously and uncontroversially constitute knowl-
edge for some of the people who have them have similar demographics.
Access to higher education is very variable across the surface of the globe
and a certain higher education is a necessary condition of one's coming
to know, for example, that proof of Tychonoff's theorem requires the
Axiom of Choice. Suppose one happens to have had such an education
and now knows this. One may say that it's a matter of chance - mere
good luck - that one was born where one was, had access to the higher
education that one did, and thus came to know this. But even so, that
chance or luck, one may insist, does not cast into any doubt the process
by which one has come to know this, making the process which one's
actually employed unreliable (in a way that should trouble an externalist
190 T. f. Mawson

about knowledge) or the reasons one actually has for believing it any
less reasonable (in a way that should trouble an internalist). So much,
then, for arguments which seek to use the demographics of theism to
undermine it 'de jure', as it were, as an object of knowledge, rather than
undermine it 'de facto', as it were, by giving us reason to think it's not
true. I shall now turn, then, to arguments of the second sort.

v
Second, then, it is sometimes argued that the facts which I have sketched
undermine theism directly as they constitute facts from which one may
run a variant of the Argument from Evil against the existence of God:
belief in God is, they reveal, distributed other than one would expect
were God to exist.
Recently, such an argument has been pressed with much vigour by
Stephen Maitzen. 6 It is his argument with which I shall most closely
engage in what follows, though the same points may be made against
the similar arguments of others. Maitzen suggests that, as a variant of
the Argument from Evil, the argument from the demographics of theism
has at least this advantage over those versions of the Argument from Evil
which focus on the evil of suffering:

The key difference, I suggest, between suffering and non-belief in

God is that suffering is far more evenly distributed than non-belief ...
non-belief in God is anything but uniformly distributed worldwide
[and indeed, one could add, over the course of human history), and
consequently any explanation in terms of features, such as human
freedom, that are uniformly distributed will not work. 7

But what the argument from the demographics of theism gives with
one hand, it threatens to take away with the other. For lack of belief in
God, whilst perhaps much more obviously unevenly distributed than
the sort of suffering Maitzen has in mind - phenomenological pain -
is also much less obviously an evil than that sort of suffering; that sort
of suffering is much more obviously an evil, even if it is also perhaps
less obviously unevenly distributed. For the argument to give with one
hand whilst not taking back in this way with the other, lack of belief in
God needs to be linked to something that is obviously an evil (ideally,
to preserve a comparative advantage, more obviously evil and more evil
than phenomenological pain). What could that be?
 The Rationality of Classical Theism and Its Demographics 191

It is instructive I think that Maitzen narrows his focus to considering

a variant of theism that links belief in God with the Highest Good, we
may call it 'salvation', by making belief in God a necessary condition of
salvation; the variant he has in mind he calls 'evangelical Christianity',
though one presumes he would wish to insist that the same points could
be made, mutatis mutandis, for at least certain Islamic views in addition. 8
Maitzen says this:

Recall that the problem ... arises in the first place from the nature
of theism's personal creator God, whose perfections include unsur-
passable lovingness and who, according to evangelical Christianity
anyway, wants everyone to believe the gospel message. Non-belief
becomes puzzling if a being of that description exists. It therefore
does not refute ADH [the Argument from Divine Hiddenness] to
construe 'theism' more broadly, as, say, the generic belief in the
supernatural: it merely changes the subject. 9

This is somewhat misleading, on a number of fronts.

First, it ignores what we might call the 'Calvinist' and 'Molinist' expla-
nations of the demographics of theism. I don't favour either of these two
solutions, for it seems to me that they don't get rid of the problem of
unfairness - they just shift it to an earlier stage in the divine economy;
it's no longer unfair that some are born without any chance of com-
ing to the beliefs that would save them because they weren't elected
to be amongst the saved anyway (Calvinism) or they had such 'stub-
born essences' that God knew that, were he to create them, there'd be
no situation in which he could place them in which they'd come to
be saved anyway (Molinism). Isn't that just to make the later distribu-
tion of belief not unfair by having the relevant unfairness occur earlier,
in not electing those souls or in instantiating such stubborn essences?
And, in the latter case, could there really be an essence so stubborn that
nothing God could do would bring the person with it to the relevant
beliefs in Him? Be all that as it may, these explanations of the demo-
graphics of theism certainly deserve at least a brief mention. 10 Here's the
Calvinist one.
Determinism is true and souls may be divided into two kinds, the
Elect and the Non-Elect. The Elects' souls are born into societal situa-
tions in which they are determined to come to the ante-mortem beliefs
necessary for salvation; the non-Elects' are born into societal situations
in which they're not, indeed they're determined to fail to come to these
beliefs. ('Double pre-destination', as it's sometimes called, seems to me
192 T./. Mawson

the only way of ultimately making this view coherent.) Calvinism of this
sort commits one to what I have elsewhere called 'rather heavy-duty
metaphysics' - determinism and souls - but it is not a metaphysics
which no theist has ever adopted. 11 Maitzen admits that his 'objection
assumes that individuals with one of these characteristics [the previous
sentence has included as a characteristic 'one's predestination as non-
elect] do not cluster by country or culture'. 12 Indeed it does, but on
Calvinism that assumption is obviously false (the demographics show
it to be so!), so the objection to theism from its demographics fails.
As Jason Marsh has recently done better than I could do in sketching the
Molinist solution, I'll simply refer readers to him for that one. 13 I want
to move past this family of 'solutions' though and see what can be done
if we grant that God does want everyone he creates to believe the gospel
message and hasn't created anyone pre-elected not to do so or incapable
of actually doing so. If we can come up with a solution within these
parameters, we will then be able to preserve the truth of the counterfac-
tual, which common sense suggests is true, namely, a goodly proportion
of those who die atheists in Vietnam would have died evangelical Chris-
tians in the USA had they escaped from Vietnam to the USA when young
enough.

VI

Supposing, then, that we grant that God does want everyone to believe
the gospel message and doesn't create anyone who is pre-elected not to
do so or couldn't actually do so, we can still resist the sub-conclusion
that we should then expect belief in Him to be uniformly distributed
over the surface of the globe. And we can resist it in a number of ways.
First, the argument depends on the strength we posit that particular
divine want to have; more precisely, it depends on our positing that it is
strong enough so that it is not over-balanced by other wants for things
that are incompatible with it. If we do not grant this, the argument fails.
To see why, consider the following by way of analogy.
I want everyone on the planet to buy my latest book, 14 but I also
want everyone who does buy it to buy it freely. Freedom being as it is-
how Libertarians construe it (see my book for more details) - I cannot
then, as a matter of metaphysical necessity, determine things so that
the world meets both conditions, so that everyone buys my book and
everyone who buys it buys it freely. Assuming my power is unlimited,
so I could make the world meet either one of the conditions, I'd have to
choose which I cared about most before deciding which way to exercise
 Tile Rationality of Classical Tlleism and Its Demographics 193

my power. Let's suppose that I care more about the world being one
that meets the latter condition - that everyone who does buy my book
buys it freely - than about it meeting the former - that everyone buys
my book. If so, you'd expect purchasers to be unevenly distributed over
the surface of the globe, in that it would be very unlikely that everyone
would freely buy my book and one of the things that leads to people
buying a book is the effect on them of other people who've bought
it and chosen to recommend it on to their friends; the culture of the
society in which they have been brought up; and so forth. Purchasers
and non-purchasers would be expected to 'cluster' then over the surface
of the globe. Similarly, then, God might well want everyone to believe
the gospel message, but he might well want those of them who do so to
come to do so without His having had to interfere with libertarian free
will; and, if his want for the world to satisfy the latter condition is greater
than his want for everyone to believe the gospel message, then, given
that one of the things people can use their free will to do is either spread
or diminish the spread of the gospel, given that people's receptivity to
the gospel is dependent on the culture in which they are brought up
and so forth, we'd expect to find that theism had an uneven distribution
over the surface of the globe. 15 This is not, note, to suppose that belief is
ever under the control of the will, just that spreading or restricting the
spread of beliefs sometimes is. 16
Now this move cannot be made so straightforwardly if one makes
having belief in God a necessary condition of the highest good, let's call
it 'salvation', for then there's nothing God would want more than it on
the 'whoever wills the end wills the means' principle. And that, it must
be confessed, is what evangelical Christianity, indeed perhaps Christian-
ity more broadly, tends to do: no less an authority than the Athanasian
creed, for example, characterizes its statements as those which 'except a
man believe faithfully, he cannot be saved'. So, we are most charitable
to Maitzen if we think of him as supposing a view of this sort in the
background of his argument. It is safe to suppose that God has an over-
whelming want that everyone believe the gospel message because only
by believing it can they be saved - the highest good.
However, with this bit of charity to the argument, we reveal the par-
ticularity of the view to which the argument is cogent. It is not an
argument against theism per se; it's an argument against this variant
of it. Furthermore, we should note that it's an argument only against a
variant of this variant, the variant of the variant that makes ante-mortem
belief in God a necessary condition of salvation. I turn now to consider
this point.
194 T. J. Mawson

VII
Even if I had the power to make everyone buy a copy of my book
and I wanted this with such strength that I had no other greater want
that over-balanced it (for example, the want that anyone who did buy
it would have bought it freely), then one could take the existence of
non-purchasers as puzzling only if one supposed that, in addition to all
this, I had set myself a deadline before which I was to have reached
the end I most wanted (everyone having purchased my book) and one
supposed that the deadline I had thus imposed on myself had passed.
Otherwise, one may simply think of particular non-purchasers or clus-
ters of non-purchasers: 'well, he's not got round to them yet'. There are
two plausible suggestions for 'deadlines' for God: either that he has set
for each individual his or her own deadline and set it as the moment
of his or her death, or that he has set us collectively one and the same
deadline, the Eschaton/Last Judgement. An alternative, 'The Harrowing
of Hell' model, suggests itself. This was a traditional answer to the ques-
tion of what happened to those who never heard the gospel; they get a
chance when Jesus goes and preaches to them in Hell. Of these three
models, only on the ante-mortem deadline model does the argument
become pressing, because only on this model do the demographics of
theism (and- I suppose one must posit additionally- the lack of univer-
sality of the phenomenon of deathbed conversions to the right religion)
make it that God's missing his deadline. So, contra Maitzen, many vari-
ants of theism - ones which don't just collapse into the vague belief in
the 'supernatural' that he suggests as the (apparently only) alternative
to evangelical Christianity - evade his argument. In fact, many vari-
ants of evangelical Christianity evade it; it is only variants of theism
which make ante-mortem belief a necessary condition of salvation that
are troubled by it. Even those believers who start off believing in the
variant of theism that makes them susceptible to this argument should
not then take this argument as giving them overall reason to reject the-
ism unless they think it is more plausible that 'if theism is true, then
the troubling variant of it is true' than they think that theism is true.
But I take it that it will be much more obvious to everyone who believes
that God exists that God exists than that if he exists, he's arranged the
economy of salvation to depend in this way on people's success in ante-
mortem theological enterprise, in which case, the argument from the
demographics of theism is one that just threatens this peripheral fea-
ture of one branch of theological thought; the argument gives nobody
reason even to abandon evangelical ChristianityY
 The Rationality of Classical Theism and Its Demographics 195

So, to sum up so far, the demographics of theism do give those of

us who are theists reason to believe something about our theism. The
demographics do not give us reason to think that we don't or probably
don't know theism to be true; 18 nor do they give us reason to think that
it's false. They do, however, give us reason to think that our believing
theism to be true ante-mortem isn't as good for us as some have main-
tained; it may yet be very good indeed for us to believe it, but it's not
essential to our ultimate salvation that we believe it. To some evangelical
Christians, that may seem a 'concession too far'. They will then natu-
rally find themselves drawn to Calvinism or Molinism, which, it will be
recalled, I do not myself favour. Finally then, I wish to do something to
show that the 'concession too far' worry is misguided; evangelical Chris-
tianity need not be threatened by giving up the claim that ante-mortem
belief in it is essential to salvation; indeed it is helped by giving it up.

VIII

I shall do this in two interlocking ways. First, I'll draw on passages from
the Bible - most specifically passages from the New Testament- to sug-
gest what the Christian biblical view is and that it does not suppose
an ante-mortem belief condition for salvation. I'll argue this through
establishing that the New Testament speaks consistently in favour of
universalism, which - given the demographics of theism (and the lack
of universality of deathbed conversions) -entails the ante-mortem belief
condition on salvation doesn't obtain. In order to avoid needless mis-
understanding at the outset, let me state that Universalism of this sort is
quite consistent with only those who accept Christ getting into Heaven
and with post-mortem punishment for at least some; the only thing it's
not consistent with is post-mortem annihilation or everlasting Hell for
any. Second, I'll look at whether the Christian biblical answer - as I'll
by then have established it to my satisfaction - is something which
rational reflection on the nature of goodness alone, unaided by reve-
lation, would have led us to expect of God. And I'll conclude that it
is exactly what we would have been led to expect, the second line of
my argument thus giving one a reason to affirm that this is indeed
what the Bible should be interpreted as teaching if we wish rationally
to believe in it as a revelation from God. Together, then, these two lines
of thinking should motivate a rejection of Calvinism and Molinism and,
in the light of the demographics of theism (and the absence of universal
deathbed conversions), a rejection of the ante-mortem belief condition
on salvation.
196 T. J. Mawson

Here's a famous passage from the Bible: 'For as in Adam all die, even
so in Christ shall all be made alive.' 19 The clear statement here is that
all shall be made alive in Christ; the passage doesn't end with 'even so
in Christ shall some be made alive'. Of course, one might suggest that
whilst all are to be made alive in Christ, some of those to be made alive
in him are to be made alive merely to be made dead to Him imme-
diately thereafter. But it would be more natural from the surrounding
context and other passages to consider being made alive in Christ as
being saved. Thus it would be more natural to think that our all being
made alive in Christ is our all being raised, being raised to a form of
judgement assuredly (there are certainly other passages which suggest
judgement), but raised to a form of judgement after which an everlasting
life in perfect communion with God awaits. This is re-enforced by other
New Testament passages. 'Then as one man's trespass led to condemna-
tion for all men, so one man's act of righteousness leads to acquittal and
life for all men.' 20 'For God has imprisoned all in disobedience so that he
may be merciful to all.' 21 We shall all be made alive in Christ; we shall all
be acquitted and given life. God will be merciful to all. One word seems
to be popping up quite a bit, doesn't it? 22 Let's look at a couple of other
passages.
Romans 10:9 tells us this: 'If you confess with your mouth, "Jesus is
Lord," and believe in your heart that God raised him from the dead,
you will be saved.' It is certainly true that not all people confess this
with their lips or believe this in their hearts ante-mortem. In Philip-
pians 2:11 we hear of the - presumably post-mortem then - day on
which every tongue shall confess that Jesus is Lord and every knee bow
before him. I suggest that if one puts the last passage from Romans
and the Philippians passage together, universalism becomes pretty much
inescapable.
The Romans passage tells us that if you (1) confess with your mouth
that Jesus is Lord and (2) believe in your heart that God raised him
from the dead, you will be saved. Philippians tells us that one day every
tongue will confess that Jesus is Lord and every knee bow before him.
So, the Philippians passage tells us that condition 1 as made mention of
in the Romans passage will on that day be satisfied by everyone; every
tongue will confess that Jesus is Lord. The Philippians passage also tells
us that on that day every knee shall bow down before him. That's not
quite as clear a statement of the satisfaction of condition 2 - everyone
believing in their heart that God raised jesus from the dead - as would
have made the case that universalism is the only biblical option airtight,
but it's close. It's just possible to render these two passages consistent
 The Rationality of Classical Theism and Its Demographics 197

without subscribing to universalism as one might suggest that some of

these who on that day confess that Jesus is Lord and bow before him will
nevertheless not believe in their hearts that God raised this person - this
very person whom they're confessing as Lord and bowing down before-
from the dead and thereby they'll nevertheless fail to be saved, failing on
condition 2 as stated in the Romans passage. But to suggest that seems,
frankly, desperate. Remember they're bowing before him and confessing
him as Lord at this stage. Is it really plausible to suppose that some
who are doing all this are really still doubting that God raised him from
the dead? No, it is not. Thus universalism is - I suggest - pretty much
inescapable for anyone who takes the Bible seriously.
I shall now move on to my second line of argument and ask whether
universalism is what reason alone would have led us to expect of a
perfectly good God.

IX

Obviously it would be good for everyone to enjoy everlasting life with

God, and this in itself would give God reason to be a universalist. But
there is another thing that is good for people, that others respect their
freedom of choice, and this opens up the possibility for wondering,
might it be overall good for God to respect our freedom if we freely
chose at the Last Judgement to reject His offer of eternal life through
Christ and go instead to annihilation or torment in an everlasting Hell?
Here I must be brief: it seems to me the answer is 'No'; however good it
is to respect free will, it's ex hypothesi not as great or good as salvation
and thus if God had to choose between them, he'd go for the latter.
Some people have done things which are so terrible that they deserve
terrible punishment, punishment which they did not receive this side
of the grave. Adolf Hitler and Joseph Stalin would be two obvious exam-
ples. Shouldn't God punish them? The answer is obviously 'Yes', but we
cannot think that even these monsters committed such terrible crimes
that only a punishment of infinite duration could be appropriate and,
I suggest, the punishment inherent in their simply being fully in God's
presence at the Last Judgement would in itself be all that justice could
demand.
Consider Jesus' parable of the returning prodigal. As with any repen-
tance, the destitution and humiliation of the son at the moment of his
decision to turn back towards his father is in exact proportion to the self-
ishness and vanity in which he has previously indulged and we may be
assured - if not re-assured - that the same fearful equation will operate
198 T. f. Mawson

on us in the searing furnace of self-knowledge that must accompany

any last judgement. When we are exposed directly to God's glorious
presence, the worse we are, the more hellish that refiner's fire will seem
to us as it burns off our misplaced egotism and self-satisfaction. It is
a fearful thing to fall into the hands of the Living God. There will be
plenty of wailing, weeping and gnashing of teeth then - enough to
validate the various biblical passages that make mention of these far-
from-cheering things. Some have already turned towards God and for
them His judgement may seem a momentary delight - the first words
of His that they hear will be 'Well done good and faithful servant.' Oth-
ers will not be turned until they find themselves before Him and to
them God will need to say more. To some, the painful truths He must
speak on that day will make this judgement seem close to a torturous
eternity. But for each of us - from Mother Teresa to Adolf Hitler - the
Judgement will in fact pass, once it has done its irreversible perfecting
work. And everlasting bliss awaits each of us, once perfected, on the
other side.
So, in summary, reflection on the nature of perfect goodness alone
would lead us to conclude that which I've argued the New Testament
asserts: universalism. That then gives the evangelical Christian reason
to reject Calvinism and Molinism as explanations of the demographics
of theism and to reject the ante-mortem belief condition on salvation.

In conclusion, we who are theists must see the demographics of theism

not as reason to think that we don't know that God exists, let alone
reason to think that God doesn't exist. Rather, we must see these facts
as giving us reason to believe what the Bible and natural reason reflect-
ing on the nature of perfect goodness were already giving us reason to
believe: the God who exists is one who will not suffer any of us finally
to be lost and thus has not made our ante-mortem theological success a
necessary condition of our being saved. The demographics of theism are
thus good news for theists in one sense and they're good news for agnos-
tics and atheists in another (in that they reveal [though not of course
to them] that their failure to be ante-mortem theists won't ultimately
exclude them from salvation). That is the gospel message, and it's obvi-
ous from the nature of the gospel message (that it doesn't require itself
to be believed in order to be true and to be the good news that it is) why
it is that God isn't prioritizing making everyone believe it this side of
the grave. 23
 The Rationality of Classical Theism and Its Demographics 199

Notes
1. I am very grateful to John Cottingham, Douglas Hedley, Dave Leal, Steve
Maitzen, jason Marsh, Yujin Nagasawa and Mark Wynn for their comments
on a draft of this chapter.
2. The sources I draw from are mainly those mentioned in the bibliography.
3. Gallup and Lindsay, 1999, p. 121.
4. The relevant data are obtainable from the Office of National Statistics
website, at www.statistics.gov.uk/census2001/profiles/rank/jedi.asp/.
5. Mawson, 2009. There's another argument that might suggest itself, springing
from the epistemology of disagreement literature: the unevenness of belief
in theism suggests that there is persistent disagreement on theism between
epistemic peers and where one has that on a topic, one has reason to sup-
pose that knowledge of the topic is not to be had. I won't engage with that
argument either.
6. Maitzen, 2006.
7. Maitzen, 2006, pp. 188-9. In fact Maitzen is wrong in this; we'll explore
why in more detail in the main text in a moment, but one quick way of
showing it is the following: one of the things people can use their freedom
to do is either spread or hinder the spreading of belief in God. That being so,
free will alone could be used to explain how, over time, we were collectively
Jed from a state in which belief in God was uniformly distributed over the
human population (originally just Adam and Eve) to one in which it had the
demographics I have sketched. Positing a historical fall would be the simplest
way of thus generating a free-will defence to the variant of the problem of
the evil posed by the demographics of contemporary theism.
8. Here and elsewhere I can be seen to be assuming false Schellenberg's position
that God would have quite generally applicable and overwhelming reasons
to remove at least inculpable atheism this side of the grave. The view that
the sort of perfect communion with God enjoyed in the beatific vision (the
enjoying of which is of course Incompatible with one's being an atheist) is
the highest good is, I take it, more or less definitive of theism, but that this
highest good is also the good of highest urgency is, I take it, not definitive
of theism; nor is it plausible. And, in the context of a Schellenberg/Maitzen
argument, it's degree of urgency that counts, not degree of goodness. It seems
to me that the elimination of atheism can only be made an urgent good by
making its achievement ante-mortem the necessary means for this sort of
post-mortem beatific vision.
9. Maitzen, 2006, p. 179; emphasis in original.
10. In a later paper, written in response to one by Marsh outlining the Molinist
reply, Maitzen says as much himself. It is from this interchange that I take
the phrase 'stubborn essences'. See Marsh, 2008, and Maitzen, 2008. One
criticism of Molinism in the main text, that it just pushes the problem back
a stage, is in fact inapplicable to the variant of Molinism that Marsh pro-
pounds- unlike the more 'traditional' one propounded by, say, Craig. Marsh
Is a universalist, who thinks of the demographics of theism as explained by
God's grouping those with ante-mortem stubborn essences together in Viet-
nam, for example, for they've nothing to lose by being so grouped. 'As long
as God's final victory ... is taken to mean that salvation will eventually be
200 T. f. Mawson

achieved by all I cannot see how any of this would be unfair to any of
the individuals [born in Vietnam]' (Marsh, 2008, p. 468). In other words,
unlike the traditional Molinist view, there are none born with such stub-
born essences that they would be ante-mortem and post-mortem stubborn.
Of course, with this version of Molinism there's still the worry that it doesn't
preserve the truth of the counterfactual that many who die atheists having
been born and lived in Vietnam would have died Christians had they been
moved as children to the USA (see main text).
11. See Mawson, 2009, for further discussion of Calvinist-inspired responses to
the demographics of theism.
12. Maitzen, 2006, p. 184.
13. Marsh, 2008.
14. Mawson, 2011.
15. Of course, unless one goes for a historical fall (see earlier note), it will seem
plausible to maintain that cultural differences have enhanced a pre-existing
uneven distribution, not generated it ab initio.
16. Here Maitzen would start to object, maintaining that given that belief isn't
under the control of the will in any case, so God could, in principle, zap
belief in theism into any given individual without thereby immediately
interfering with that individual's free will. But I would stress that if the prior
lack of belief in that individual had resulted from the free actions of other
individuals, not even God could zap belief in theism into the first individ-
ual without thereby interfering with the effective freedom of these other
individuals.
17. This is a bit 'rough and ready'; note though that I do say that it doesn't give
these theists 'overall reason' to reject theism; I would concede that it gives
them reasons in the following manner. If you start out believing '"A" and
"If A, then B"' is true and I give you reason to think that 'A and B' is false,
that is my giving you some reason to think' "A" and "If A, then B"' is false,
which is also then, one might say, my giving you some reason to think 'A' is
false, as A's being false is one of the ways in which it could be false that '"A"
and "If A, then B".' But I've not given you overall reason to think A is false
if it is 'much more obvious', as I say in the main text, to you that 'A' is true
than it is that 'If A, then B' is true. I've given you overall reason to think that
'If A, then B' is false.
18. See Mawson, 2006, for the defence of this.
19. 1 Corinthians 15:22.
20. Romans 5:18.
21. Romans 11:32.
22. I've italicized it just in case it's not obvious which one I have in mind.
23. These considerations lead me to think that a stronger argument for athe-
ism could be constructed if one were to focus on the demographics, not of
ante-mortem belief in God, but of something that is more plausibly the sort of
thing that God would have an overwhelming desire to see evenly distributed
ante-mortem and which, empirical research suggests, has an uneven distribu-
tion ante-mortem. Discussion with john Cottingham suggests to me that the
best candidate for such a thing would be the ability to lead the morally good
life, which ability in turn depends on having epistemic access to the nature
of the morally good life.
 The Rationality of Classical Theism and Its Demographics 201

References
Bruce, S. (2002) God Is Dead: Secularization in the West (Oxford: Blackwell).
Finngeir, H. (1998) Atheism in India (Mumbai: Indian Secular Society).
Finngeir, H. (2003) Atheism in the World (Oslo: Human-Etisk Forbund).
Gallup, G., ]r, and M. Lindsay (1999) Surveying the Religious Landscape (Harrisburg,
PA: Morehouse).
Greeley, A. (2003) Religion in Europe at the End of the Second Millennium: A Sociolog-
ical Profile (New Brunswick, NJ: Transaction Publishers).
Inglehart, R., M. Basanez, J. Diez-Medrano, L. Halman and R. Luijkx (2004)
Human Beliefs and Values: A Cross-Cultural Sourcebook Based on the 1999-2002
Value Surveys (Mexico City: Siglo Veintiuno Editores).
Maitzen, S. (2006) 'Divine Hiddenness and the Demographics of Theism',
Religious Studies 42: 177-91.
Maitzen, S. (2008) 'Does Molinism Explain the Demographics of Theism?',
Religious Studies 44: 473-7.
Marsh, J. (2008) 'Do the Demographics of Theistic Belief Disconfirm Theism?
A Reply to Maitzen', Religious Studies 44: 465-71.
Mawson, T. ]. (2009) 'Mill's Argument against Religious Knowledge', Religious
Studies 45: 417-34.
Mawson, T.]. (2011) Free Will: A Guide for the Perplexed (New York: Continuum).
National Statistics Office, The (2001) The 2001 Census Online; available online at
www.statistics.gov.uk/census2001/demographic_uk.asp.
O'Brien, J., and M. Palmer (1993) The State of Religion Atlas (New York: Simon &
Schuster).
Zuckerman, P. (2003) Invitation to the Sociology of Religion (London: Routledge).
Zuckerman, P. (2007) 'Atheism: Contemporary Rates and Patterns', in M. Martin
(ed.), The Cambridge Companion to Atheism (Cambridge: Cambridge University
Press), pp. 47-68.
Part V
Religious Tolerance and
Disagreement
9
Coercion, Consequence
and Salvation
Steve Clarke

Thus Augustine says to the Count Boniface: 'What do these people

mean by crying out continually: "We may believe or not believe just
as we choose." Whom did Christ compel? They should remember
that Christ at first compelled Paul and afterwards taught him.'
- Aquinas, Summa Theologiae, 2nd part of the 2nd part,
Question 10, Article 8

1. Salvific Exclusivism

Salvific exclusivists believe that there are necessary conditions that

must be met before salvation can be attained. Different salvific exclu-
sivists believe in different necessary conditions. Common necessary
conditions include: belief in the cardinal tenets of a particular religion,
membership of a particular religious organization, conduct of particular
religious practices and the avoidance of other practices. Salvific exclu-
sivism stands in contrast with salvific pluralism. Salvific pluralists such
as Himma (2002) hold that there is no set of conditions necessary for
salvation. On this view, members of many religions are eligible for sal-
vation, and their actual salvation depends on God's consideration of
their individual merits. A middle position between these two extremes
is one that might be referred to as 'salvific preferentialism'. This is the
view that, although God favours those who hold certain religious beliefs,
conduct certain religious practices, or are members of particular religious
organizations; when deciding whom to grant salvation to, God does
not apply hard and fast rules, and will consider the individual merits of
those who lack the beliefs, practices and/or organizational membership
required for preferential consideration.

205
206 Steve Clarke

Practically all of those who believe in salvation agree that salva-

tion is supremely important. Salvation is typically understood as a
precondition to entry into Heaven and the opportunity to enter Heaven
is usually understood as an opportunity that only comes along once.
On standard Christian and Muslim views entry into Heaven enables
the experience of maximal happiness for eternity. And if that were not
enough motivation to do what is necessary for salvation, Christian and
Muslim theologians often add that some or all of those who are not
saved will spend an eternity suffering in Hell. Salvation, if indeed it is
available, is of overwhelming importance when compared to any other
prudential or 'self-regarding good' (as distinct from 'other-regarding
goods'), or indeed all other prudential goods combined. Salvation is a
necessary precondition to receiving eternal happiness. In the absence
of salvation, our enjoyment of all other prudential goods is necessarily
ephemeral.
Salvific exclusivism has long been a prominent position amongst
both Christians and Muslims. The Catholic Church officially advocated
a fairly strict form of salvific exclusivism up until the time of Vati-
can II (1962-65), holding that salvation is only available to practicing
Catholics (Avalos, 2005, pp. 195-6). 1 Nowadays its position is a salvific
preferentialist one, according to which no one is excluded from the bare
possibility of salvation; however, membership of the Catholic Church is
said to make it much easier to attain salvation Oanes, 1967). The South-
ern Baptist Convention holds that only Christians can attain salvation,
a form of salvific exclusivism that is common amongst conservative
protestant groups. 2 Muslim Salvific Exclusivists are often less exclusive
than their Christian counterparts. An influential view in Islam is that
Salvation is available to 'people of the book' - including Christians and
Jews, but not to Hindus, Confucians and others. However, members of
the Salafi branch of Sunni Islam are more exclusive than most Muslims,
often holding that only very devout Muslims can be saved (Adraoui,
2008).
Consequentialists, who view morality as being exhausted by the con-
sideration of consequence, hold that we ought to do whatever can
be done to ensure that the best possible consequences occur.3 So it
seems that consequentialists who are salvific exclusivists should do
everything they can to attempt to persuade all others to join the
appropriate religious organization, believe the required religious propo-
sitions and/or participate in the required religious practices, so that
they may maximize their chances o~ attaining salvation -the one over-
whelmingly significant consequence. The same claim can be made for
those salvific exclusivists who are not consequentialists, but who accept
 Coercion, Consequence and Salvation 207

that the consequences of salvation are important enough to trump

whatever concerns, other than concerns of consequence, that they con-
sider that morality requires. Such concerns might include deontological
constraints that would override many considerations of consequence
but are not considered to override consequences as significant as the
attainment of salvation. 4
What if the salvific exclusivist who is committed to maximizing the
chances that others are able to attain salvation is unable to persuade
others to do what she believes that they need to do in order to attain
salvation? Sometimes it may be within the power of consequential-
ist salvific exclusivists to coerce others to join religious organizations,
to participate in religious practices, and, in so far as this is possible,
to believe particular religious propositions. 5 Should salvific exclusivists
who are able to coerce others to do these things, in order to ensure
that those others are eligible to receive salvation, do so? Many Kantians
would object to the use of coercion, under such circumstances, on the
grounds that it fails to respect individual autonomy. But consequential-
ists should have no such qualms, particularly when the stakes are as
high as they are when the possibility of salvation hangs in the balance.
Surely the consequentialist salvific exclusivist will consider that she has
a moral obligation to employ coercive means to compel acceptance
of the correct religion. Furthermore, it seems that the consequentialist
salvific exclusivist has a moral obligation to ensure that rival religious
doctrines are not promulgated, if there is some chance that these will
be accepted by some people and that this acceptance will result in those
people being denied salvation.
The last point above is unoriginal. David Lewis (1989) argued, along
similar lines, that a devout defender of an orthodox salvific exclusivist
religion who was also a consequentialist would have a compelling rea-
son to suppress heresy, which is that it threatens to cause followers
of the orthodox religion, who might be susceptible to the influence
of heretical teachings, to lose out on all possibility of obtaining salva-
tion. Craig Duncan (2007) extends this argument. According to him,
in cases where there is even a small probability that orthodox believers
may attain salvation and there is no probability that those who are not
orthodox believers will be saved, consequentialism 'absolutely requires'
the persecution of unorthodox religion (2007, p. 4).
In reasoning along these lines both Lewis (1989) and Duncan (2007)
follow in the footsteps of Aquinas who argued that

With regard to heretics ... there is the sin, whereby they deserve not
only to be separated from the Church by excommunication, but also
208 Steve Clarke

to be severed from the world by death. For it is a much graver matter

to corrupt the faith which quickens the soul, than to forge money,
which supports temporal life. Wherefore if forgers of money and
other evil-doers are forthwith condemned to death by the secular
authority, much more reason is there for heretics, as soon as they
are convicted of heresy, to be not only excommunicated but even
put to death. (Summa Theologiae, 2nd part of the 2nd part, Question
11, Article 3)

Aquinas is no consequentialist, however his argument for harsh treat-

ment for heretics can be understood as a piece of salvific exclusivist
reasoning that appeals to considerations of consequence. Heretics have
a tendency to promulgate views that would, if accepted, prevent the
attainment of salvation. Therefore, considerations of consequence jus-
tify the suppression of heretical teachings and justify harsh punitive
measures against heresy so as to deter others from promulgating hereti-
cal views. Aquinas is far from unique in arguing along these lines. Calvin
defended the execution of heretics, as did his successor in Geneva, Beza.6
Although Aquinas's views made appeal to considerations of conse-
quence in justifying the suppression of heresies, he did not extend this
argument as far as it might go. He did not advocate the general sup-
pression of other religions or the use of forced conversion for the sake
of salvation. Indeed, Aquinas upholds the standard Christian line of his
day, in insisting that belief must be voluntary (Summa Theologiae, 2nd
part of the 2nd part, Question 10, Article 8) but arguing that heretics and
apostates constitute a special case, because their previous expressions of
faith amount to a promise which they can be held to, and which cannot
be renounced.
This standard line was by no means universal amongst Christian the-
ologians. Duns Scotus notoriously argued for the forced baptism of Jews.
His argument turns on the Orwellian sounding doctrine of consenting
virtually, according to which assent to baptism due to fear of injury
or death can be sufficient to count as genuine consent in the eyes of
God {Turner, 2006, p. 196). Of course, once a legitimate baptism had
occurred, Jews were no more able to leave the Church than heretics or
apostates. Duns Scotus supplements this argument with an argument for
the compulsory baptism of Jewish children that rests on a conception
of parental rights which broke with the Christian orthodoxy of the day.
Whereas Aquinas and others had declared that one should not forcibly
baptize the children of jews and heathens, on the ground that this
would be a violation of the rights of their parents, Duns Scotus argued
 Coercion, Consequence and Salvation 209

that parental rights should give way to the rights of higher powers.
According to him the rulers of Christian states, who were intermedi-
ate between ordinary folk and God, have a duty to enforce the rights
of God, the highest power, to have jewish children converted to Chris-
tianity (Krop, 1989, pp. 164-5). If their parents were to resist this then
their resistance could trigger an additional argument for their forcible
baptism. According to Duns Scotus:

it is religiously just for those parents themselves to receive baptism

forcibly with threats and fear, because although they will not be real
believers at heart, the evil for them to be stopped from serving their
law with impunity is less than serving that law freely. What is more,
if their children are well educated, they will be real believers in the
third and fourth generation/

None of the above argument turns on any particular details of Judaism,

so it looks like it can be applied equally to all non-Christians.
Lewis (1989) and Duncan (2007) are concerned, inter alia, to demon-
strate that salvific exclusivist consequentialist arguments for the sup-
pression of unorthodox religions are valid. Here it will be taken as given
that such arguments, along with arguments for coerced religious conver-
sion, are valid. Our focus will be on a different problem. Many of us live
in liberal societies and religious tolerance is a core value of contempo-
rary liberal societies. We may not share the religious beliefs of the Hare
Krishna living at the end of our street and we may not like her religious
practices, however, we put up with her presence and try not to express
our distaste for her behaviour. In some instances our decision to be tol-
erant is pragmatic and in some instances it is based on a commitment to
the value of religious tolerance. This commitment is common to almost
all variants of liberalism; and religious tolerance is often upheld as a core
value of contemporary liberal societies.
Not all citizens of liberal societies are convinced that religions other
than their own should be tolerated. Prominent amongst the uncon-
vinced are salvific exclusivists. There are many who hold salvific exclu-
sivists religious beliefs and from the salvific exclusivist point of view,
as we have seen, there are compelling reasons to oppose the tolera-
tion of other religions. Of course not all salvific exclusivists will find
such arguments compelling. Some will consider that the values of reli-
gious tolerance and/or autonomous choice outweigh the consequence
of other people not being eligible for salvation. But many will follow
Aquinas and Calvin in finding such arguments extremely compelling.
210 Steve Clarke

The problem that will concern us here is the problem of when and how
members of a liberal society who value religious tolerance can share
their society with those salvific exclusivists who find arguments for
religious intolerance that appeal to considerations of consequence com-
pelling. These will include consequentialists who are salvific exclusivists
and they will also include those non-consequentialists who decide that
whatever deontological constraints on their behaviour there are, that
these are not sufficient to override the moral importance of ensuring
that others attain salvation, even if this involves using coercive mea-
sures. For convenience we will refer to all of those who accept a moral
argument for using coercive measures to ensure that others are eligible
to attain salvation as 'interventionist salvific exclusivists'.

2. The Interventionist Salvific Exclusivist

in a Liberal Society
Interventionist salvific exclusivists are motivated to try to ensure that
others are eligible for salvation and so they are motivated to try to ensure
that others meet whatever conditions this involves. Most of the time
these conditions will include accepting some or all of the key aspects of
a particular religion, whether this be acceptance of core beliefs, member-
ship of a particular religious organization, or participation in particular
religious practices. I will refer to any or all of these as 'acceptance' of a
religion.
Here our concern is with interventionist salvific exclusivists who
reside in typical liberal societies and in typical liberal societies individu-
als enjoy freedom of religion. They are free to choose which religion to
accept. They are free to switch allegiance to a different religion at any
time, and they are free to proselytize on behalf of their favoured religion.
In one way freedom of religion should be welcomed by interventionist
salvific exclusivists; it ensures that they are able to attempt to convert
everyone within a society to acceptance of their religion. However, there
is another way in which freedom of religion is unwelcome for interven-
tionist salvific exclusivists. Precisely because liberal states uphold the
value of freedom of religion, they restrict the use of some of the means
to make converts that the interventionist salvific exclusivist may wish
to employ and which may be most effective in securing conversions.
In particular, liberal states restrict the use of coercive means to secure
conversions, especially the use of violence and the use of the threat of
violence.8 They do so because the use of coercive means diminishes a
 Coercion, Consequence and Salvation 211

person's freedom of choice, a core value that liberal states traditionally

aim to uphold (Spector, 2008).
The benefits of coercion for the interventionist salvific exclusivist are
not exhausted by opportunities to make additional conversions. The
interventionist salvific exclusivist is in competition with other religions,
many of which have an interest in making apostates of the followers of
her religion. If believers in these other religions are also interventionist
salvific exclusivists, then, all things being equal, they will be as moti-
vated as she is to make conversions. But this is not the only sort of
threat she faces. Because, in the liberal state adherents to her religion
are not compelled to accept any religion, they may become agnostics or
atheists. They may also create new religions. If these bear a partial resem-
blance to the religion that the interventionist salvific exclusivist seeks to
promote then they are heresies from the point of view of the interven-
tionist salvific exclusivist and they constitute a particular danger to the
interventionist salvific exclusivist because current adherents only have
to change some of their beliefs, while being able to retain others, to con-
vert to the heresy, which is generally easier to do than to convert to an
entirely new religion. 9 All of these threats can be countered by the use
of coercion.
Although liberal states oppose the use of coercion by some citizens
against other citizens, most liberal states typically suffer no qualms
about employing coercive means, through laws backed by standing
police forces and armies to prevent individuals and groups residing
within the state, from using coercive means against one another.
Because the legal and institutional apparatus of the liberal state stands
in the way of the interventionist salvific exclusivist's goals, interven-
tionist salvific exclusivists will be motivated to attempt to overthrow
the liberal state and perhaps to collaborate with other opponents of the
liberal state to attempt to overthrow that state. So, the presence within
a liberal state of interventionist salvific exclusivists who have a dispo-
sition to overthrow the state is a standing threat to that state. If some
interventionist salvific exclusivists are considered too much of a threat
to a liberal state then that state might feel warranted in suppressing
some or all salvific exclusivist religions. It is perhaps unclear whether
a society that decided to do this and to not 'tolerate the intolerant'
would count as a genuine liberal society or not. 10 But this definitional
issue will not be taken up here. From the point of view of tolerant
mainstream liberals, a society that tolerated all religions except some
salvific exclusivist religions would be far preferable to the society that
212 Steve Clarke

the interventionist salvific exclusivist would prefer, in which only one

religion was tolerated.
The interventionist salvific exclusivist is motivated to maximize the
number of people who adhere to the correct religion and are therefore
eligible to attain salvation, and she will feel justified in using coercive
means to achieve this goal. However, if she attempts to use coercive
means in a liberal state, she can expect to be prosecuted by the state, and
may become less able to make converts in the future. Repeated attempts
to use coercive means may also result in the practice of her religion being
suppressed, in which case many who previously followed her faith may
abandon that faith and may become ineligible for salvation as a result.
So, the potential gains to be had by the use of coercive means must be
great if they are to justify the use of coercive means.
Because the liberal state can be expected to try to ensure that there
are no lasting benefits to be gained from the use of coercive means, by
some of its citizens against others, the only benefits that can reason-
ably be expected to be reliably available to the interventionist salvific
exclusivist, which might compensate for the risks involved in using
coercive means, are ones that may be achieved by overthrowing a lib-
eral state and replacing it with a religious state. If this could be done,
then it would enable the interventionist salvific exclusivist to use coer-
cive means in an unhindered manner, to ensure that the great majority
of people living in that state meet the necessary conditions of salvation.
Furthermore, it would enable the interventionist salvific exclusivist to
redeploy the legal and institutional apparatus of the state in aid of this
goal. In effect the salvific exclusivist is faced with a basic choice between
attempting to 'save souls' via non-coercive means, while accepting that
many may be lost, and attempting to overthrow the state in order to
create a religious state in which many more souls may be saved. How-
ever, the act of attempting to overthrow the state is fraught with danger
and may result in the loss of many souls if it fails.

3. Modelling the Interventionist Salvific

Exclusivist's Decision
The interventionist salvific exclusivist aims to maximize the number of
people who accept the correct religion, so as to maximize the number
of people who are eligible to attain salvation. From her point of view
all other considerations are trumped by this one consideration. In order
best to achieve her goal the interventionist salvific exclusivist who is
living in a liberal state must make a basic choice. She can attempt to
 Coercion, Consequence and Salvation 213

operate within the rules of the liberal state or she can involve herself
in the organization of an attempt to overthow the state and replace it
with a religious state. Either course of action involves potential risks
and potential benefits. If she attempts to operate within the rules of the
state and proselytizes on behalf of her favoured religion she may make
converts and she may lose adherents to competitors. If she attempts to
overthrow the state she may succeed and be able to compel acceptance
of her favoured religion and she may fail and her cause may be harmed
as a result. How is she to think about this choice?
Contemporary work on decision-making under risk provides us with
two broad frameworks which may be used to guide her decision. These
are cost-benefit analysis (CBA) and the precautionary principle (PP).
We will now consider how each may be used to shape her decision. 11
Using CBA a comparison of the expected utility of the two courses of
action is made and the course of action with the greatest expected util-
ity is recommended. Suppose, to consider a somewhat overly simplistic
example, that SO per cent of a population of I 00,000 who live in a small
liberal state are adherents to a particular religion and are considered
to be eligible for salvation by an interventionist salvific exclusivist. She
calculates that if she and other interventionist salvific exclusivists who
follow the same religion devote their efforts to non-coercive conversion
over the next ten years, they stand a 7S per cent chance of converting
another 10 per cent of the population, or 10,000 souls and a 2S per
cent chance of losing 20 per cent of the SO,OOO faithful, or 10,000 souls.
So she calculates that the expected utility of non-coercive conversion
is (10, 000 x 0. 7S)- (10, 000 x 0. 2S) = SOOO. She then calculates that if
she and other interventionist salvific exclusivists attempt to overthrow
the state they stand a SO per cent chance of succeeding and a SO per
cent chance of failing. She further calculates that if they succeed they
can expect to convert approximately 40 per cent of the current popu-
lation (she assumes that approximately 10 per cent will flee to another
state) or 40,000 souls. However, if they fail they can expect to lose 80 per
cent of the faithful, or 40,000 souls (she assumes that 20 per cent will
continue to practice their religion even if they are persecuted for doing
so after the failed coup). So she calculates that the expected utility of
attempting to overthrow the state is (40,000 x 0. S)- (40,000 x 0. S) =0.
This application of CBA suggests that the true believers should not now
resort to coercive means, but the difference in expected utility between
the two possible courses of action is not overwhelming, so it seems that
she should review her decision at a later date as the factors informing
her calculation may well change. 12
214 Steve Clarke

It might be thought that only leaders of religious groups that can

count on the support of a significant proportion of the population will
stand any serious chance of overthrowing a state, and that therefore we
can dismiss the possibility that interventionist salvific exclusivists who
apply CBA, and who adhere to a religion that holds sway over a small
minority of the population, will attempt to overthrow the state. But
this would be to misunderstand the calculation at hand. Suppose that a
salvific exclusivist religion was followed by 2 per cent of the population
of 100,000 and suppose that their interventionist leader made the cal-
culation that they had a 5 per cent chance of taking over the state and
converting the remaining 98 per cent of the population and a 95 per
cent chance of failing in their attempt, leading to suppression and the
loss of all of the souls that are currently believed to be eligible to be
saved. On these calculations the expected utility of attempting a coup
is (0. OS x 98000) - (0. 95 x 2000) = 3000. So, it will be rational for the
salvific exclusivist leader to attempt a coup under these circumstances,
unless the expected utility of making conversions over an equivalent
period of time (say ten years) is 3000 or over, which would involve a
very high rate of conversion, for a religion that currently only has 2000
followers in a population of 100,000.
A possible objection to the way I have set up these calculations is that
if the experience of maximal happiness for eternity has infinite value
then the value of the salvation of any number of individuals (above
zero) will be infinity; and therefore it is pointless to try to increase the
number of people who will attain salvation beyond one. So, saving an
additional 3000 souls should make no difference to the interventionist
salvific exclusivist, provided that at least one soul is already going to be
saved. Intuitively, many of us will want to be able to say that a greater
number of people experiencing infinite utility is preferable to a smaller
number, despite the problem of comparing infinities. Vallentyne and
Kagan (1997) appeal to nonstandard mathematics to defend the coher-
ence of this intuitively appealing assertion. 13 It is beyond the scope of
this essay to determine whether they are successful. If they are, then
their approach can be used to meet the objection. Another way of meet-
ing the objection is also available, which is to deny that the experience
of maximal happiness over an infinite period of time has infinite utility.
If the utility of happiness is discounted at an appropriate rate over time,
then infinite happiness can be ascribed a very high (but finite) utility
(Garcia and Nelson, 1994, p. 185).
Examples like the above, where the use of CBA can lead to the rec-
ommendation that we take great risks, has led some to be sceptical of
 Coercion, Consequence and Salvation 215

the appropriateness of CBA as a guide to decision-making under risk. 14

Such sceptics are typically advocates of the PP. Despite being referred
to as the PP, there is no one PP. In fact, there are at least twenty differ-
ent formulations of the PP (Sunstein, 2005, p. 18). What they have in
common is that they all involve an attempt to capture the apparently
commonsensical idea that we are 'better safe than sorry' (Sandin, 2007,
p. 105). In other words, we should err on the side of caution even if
that means foregoing significant potential benefits. The recommenda-
tions resulting from applications of CBA and the PP to the same set of
circumstances will be much the same when potential benefits are not
significant enough to outweigh risks. However, when potential benefits
are significant enough to outweigh risks, and risks are not trivial, then
applications of CBA and typical variants of the PP will lead to very differ-
ent recommendations. While an application of CBA will typically lead
to the recommendation of action in such circumstances, applications of
most versions of the PP will lead to the recommendation of inaction.
From the point of view of a salvific exclusivist religious leader who is
contemplating a coup, in order to be able to compel people to accept
the religion that will enable them to attain salvation, there are almost
always significant risks involved in action as a failed attempt to conduct
a coup may well lead to the suppression of the religion she wishes to
promote. It looks like use of the PP would only ever lead to the recom-
mendation of attempting a coup when it was certain that it would be
successful or when it was certain that there would be no adverse con-
sequences that would result from a failed coup. But this is effectively
never.
Despite the above line of reasoning, I am not convinced that applica-
tions of the PP do invariably lead to the recommendation of inaction
when an interventionist salvific exclusivist leader is contemplating
attempting a coup. To see why we need to consider the following influ-
ential line of criticism of core versions of the PP, I refer to Manson
(2002), Sunstein (2005) and others. These critics argue that it is impos-
sible to apply these core versions of the PP in a logically consistent way
and that actual applications of the PP necessarily involve a failure to
recognize that there are 'risks on all sides', and a concomitantly selec-
tive approach to the consideration of risks. For example, the PP is often
applied to justify restriction of the production of Genetically Modified
(GM) crops on the grounds that the consumption of GM foods might
possibly be a health risk to humans. But this line of reasoning fails to
take into consideration the possibility that a failure to grow GM crops
will pose a risk to human health. Non-genetically modified foods are
216 Steve Clarke

generally more expensive than GM foods and in situations of famine

the worst off face the possibility of being unable to afford proper nutri-
tion if they do not have access to GM crops. If we focus on this risk then
the PP seems to lead to the recommendation that we should definitely
grow GM crops. 15
It might seem that an application of the PP would lead the salvific
exclusivist to decide not to risk losing the souls that are currently eligible
for salvation and therefore to avoid attempting to overthow the state.
However, this appearance is only manifested when we lose sight of the
risks involved that are borne by the unconverted in being denied the
possibility of salvation. A salvific exclusivist who focused on the risks to
the unconverted (and did not consider the risks to those that currently
adhere to the correct religion) might equally employ the PP to argue that
we should avoid the risks to the unconverted involved in being denied
salvation, and so attempt to overthrow the state in order to be able to
coerce acceptance of the correct religion, even if the chances of success
are extremely low.
Manson (2002) and Sunstein (2005) are right to hold that we cannot
apply the PP in a logically consistent manner, but this does not mean
that we cannot apply the PP at all. We can apply it when we take a
selective approach to risk. According to Sunstein selective approaches
to risk are widespread and are the result of the operation of a vari-
ety of cognitive biases, including the availability heuristic, probability
neglect, loss aversion, a belief in the benevolence of nature and systems
neglect (2005, p. 35). The most significant of these biases, in Sunstein's
view, are the biases accompanying the use of the 'availability heuris-
tic', a rule of thumb which is used pervasively in the lay assessment
of risk (Sunstein, 2005, pp. 35-9). Risks will seem highly 'cognitively
available' to us when we are familiar with them or when they are made
highly salient to us by recent events, by the testimony of others or by
the media. If a type of dangerous event has been experienced recently,
is discussed widely, or is highlighted in the media then people will intu-
itively raise their assessment of its likelihood of recurrence and this may
'crowd out' awareness of other risks. After the events of 9/11 the risk of
a terrorist attack while flying was highly cognitively available to many
people, particularly in America. This caused many people to use alter-
native forms of transport, such as driving, for inter-city trips that they
would have previously flown on. These alternatives involve risks of their
own. However, because the cognitive availability of terrorist activity on
airplanes crowded out these other risks, many people did not consider
them when making inter-city transport choices. 16 The pervasiveness of
 Coercion, Consequence and Salvation 217

cognitive biases in human cognition, and of biases resulting from the

use of the availability heuristic in particular, helps explain why the PP
can be both incoherent and widely applied (Clarke, 2010).
It is not clear that a committed interventionist salvific exclusivist who
was contemplating a coup in order to maximize the number of peo-
ple who were eligible for salvation would focus on the risks involved
in attempting to overthrow the state. She might instead focus on the
risks to those who are currently ineligible for salvation. This is, after all,
the primary concern of the interventionist salvific exclusivist. So, in her
hands, an application of the PP might recommend attempting to over-
throw the state, even in circumstances where applications of CBA would
lead to the recommendation of inaction.

4. Liberal Responses to the Threat of Salvific Exclusivism

How should a liberal state respond to the possible threat posed by
interventionist salvific exclusivists, given that the reasoning of salvific
exclusivist leaders who are influenced by considerations of consequence
may lead them to attempt to overthow the state? The first thing to say
is that the state should attempt to understand the moral reasoning that
accompanies particular instances of salvific exclusivism. Some salvific
exclusivists will take the view that they should not attempt to coerce
others into accepting any particular religion, even if this is in their
own interests; in which case they will be unlikely to be motivated to
attempt to overthrow the state. However, others may well be swayed by
the argument that we have rehearsed and will consider it appropriate
to coerce people to accept the religion that offers the only possibility of
their salvation.
The second thing to say is that the state should attempt to monitor
the activities of salvific exclusivist leaders who are liable to be swayed by
interventionist arguments for coercion, as these leaders may be highly
motivated to attempt to overthrow the state or to collaborate with oth-
ers in attempting to overthrow the state. 17 Often states have more than
one group of enemies and interventionist salvific exclusivist leaders who
are disposed to overthrow the state may ally themselves with other
opponents of the state in order to achieve their long-term goals. 18
The third thing to say is that a liberal state should take all reason-
able steps to ensure that when interventionist salvific exclusivist leaders
decide whether to attempt to overthrow the state or not, the answer
they are most likely to be led to is that they should not attempt to over-
throw the state. As we have seen, some interventionist salvific exclusivist
218 Steve Clarke

leaders may try to make this decision on the basis of reasoning that is
guided by CBA, while others may try to make it on the basis of reason-
ing that is guided by the PP. We will consider what the liberal state can
do to influence leaders guided by each of these broad frameworks.
The government of a liberal state will typically be able to exert an
influence on at least three variables that inform the calculation that
the interventionist salvific exclusivist leader who applies CBA will try to
make when deciding whether or not to attempt to overthrow the state.
They can affect (1) the opportunity of the salvific exclusivist to pursue
non-coercive means to achieve their goals, (2) the possibility that an
attempt to overthrow the state may succeed and (3) the consequences
of conducting a failed coup. I will now briefly consider all three of these
variables.
Because the interventionist salvific exclusivist will be comparing the
consequences of non-coercive attempts to make conversions with the
consequences of attempting to overthrow the state, it is vital, from
the point of view of the liberal state, that the interventionist salvific
exclusivist is free to make conversions. If the interventionist salvific
exclusivist is completely unable to make conversions by operating
within the rules of the state then any possibility of overthrowing the
state will be sufficient to motivate the interventionist salvific exclusivist
to attempt to overthrow the state. And if the freedom to make con-
versions is restricted to some degree then this will reduce the prospect
of the interventionist salvific exclusivist making conversions and so
make the consequences of an attempt to overthrow the state appear
comparatively more appealing. In f;eneral, the lower the possibility that
the interventionist salvific exclusivist believes the prospects of making
conversions (and retaining the loyalty of those who are currently com-
mitted to their favoured religion), the more appealing the prospect of
attempting to overthrow the state is going to appear.
The liberal state can take a variety of measures to reduce the chance
that a coup initiated by interventionist salvific exclusivists is successful.
These will include taking reasonable steps to ensure that the loyalties
of those individuals who make up the instruments of state power- the
police force, the army, the judiciary and so on- are primarily committed
to the state rather than to religious organizations. It can also include
ensuring that religious organizations are not able to acquire experience
in military operations, and are not able to create ongoing associations
with paramilitary groups.
When conducting a CBA to decide whether to attempt to overthrow
the state or not, the interventionist salvific exclusivist will need to have
 Coercion, Consequence and Salvation 219

in mind the possible consequences of failure. The possible consequences

that the interventionist salvific exclusivist will care about most will
be the loss of faith by current believers and the possibility of being
prevented from converting others to the interventionist salvific exclu-
sivist's religion in the future. Knowing that these possibilities are matters
of grave concern to the interventionist salvific exclusivist places the
defender of the liberal state on the horns of a dilemma. On the one
hand, one of the core values of the liberal state is freedom of religion.
On the other hand, if the interventionist salvific exclusivist believes that
she and others will still be free to proselytize on behalf of her religion
after an attempted coup, then they will be more likely to attempt a coup,
so a willingness to restrict the freedom of the interventionist salvific
exclusivist under certain conditions makes it more likely that freedom
of religion will be available to all. But this is really just one instantiation
of a more general dilemma that the defender of liberalism must face in
many contexts. Liberals are generally in favour of tolerance but some of
those that they may decide to tolerate represent a threat to the future of
liberal states. It is beyond the scope of this essay to solve this dilemma.
I simply note here that one way that liberals can make it less likely that
salvific exclusivists will attempt to overthrow the state is by making it
clear to interventionist salvific exclusivists that their freedom to prose-
lytize of behalf of their religion will be significantly restricted if they do
attempt to overthrow the state and fail. 19
Suppose that an interventionist salvific exclusivist leader's decision
about whether to attempt to overthrow the state or not is based on the
PP rather than CBA. Is there anything that the state can do to reduce
the chance that she will attempt to overthrow the state? I will argue that
there is. Recall that the interventionist salvific exclusivist leader's deci-
sion under an application of the PP depends crucially on which risks are
salient to her when she is attempting to apply the PP. If she attends to
the risks associated with a failed coup then an application of the PP will
lead her not to attempt to overthrow the state. However, if she attends to
the risk that people may be ineligible for salvation because they do not
accept the required religion then an application of the PP will lead her to
attempt to overthrow the state. If she attends to both of these, then, for
the reasons that Manson (2002) and Sunstein (2005) point out, she will
be unable to apply the PP. However, as we saw earlier, she is unlikely
to attend to both sets of risks when either one is highly 'cognitively
available' to her.
Cognitive availability is influenced, inter alia, by familiarity. If people
know about a risk and are reminded of it on a regular basis, then this
220 Steve Clarke

risk will be less likely to be 'crowded out' by other risks, and it will be
more likely to crowd out other risks. It is heavily in the state's interest
to ensure that the risks associated with attempting a coup are familiar to
the interventionist salvific exclusivist leader who applies the PP and to
any other salvific exclusivists who might become interventionist salvific
exclusivist leaders and might apply the PP. If these risks are familiar then
they will be cognitively available and will be hard to crowd out. If they
are not crowded out then the interventionist salvific exclusivist leader's
application of the PP will either lead to the decision not to attempt a
coup or will lead to a failure to apply the PP.
The risks of attempting a coup will tend to be familiar when salvific
exclusivists have been told about such risks sufficiently often. Such
information can be imparted in various ways. For example, the edu-
cation system could be used to teach people about the punitive
response of the state to attempted coups and about past failed coups,
attempted coups that have failed could be celebrated (the celebration
of Guy Fawkes Night - also known as Bonfire Night - in England is
a precedent here) and current plots to overthrow the state that fail
could be discussed widely by state representatives (and others) in the
media.

5. Concluding Remarks
The combination of acceptance of salvific exclusivism and a commit-
ment to the moral goal of maximizing the best consequences for others
can be extremely dangerous from the point of view of liberal soci-
eties. In virtue of their intellectual commitments, interventionist salvific
exclusivists consider that they have a moral obligation to ensure that
everyone accepts their religion. If others will not accept their religion
voluntarily, then they have a compelling reason to attempt to coerce
them to accept that religion, and a compelling reason to use coercive
measures to prevent proselytizing on behalf of other religions. If the
state tries to prevent them from using coercive means to achieve these
ends then they will have a compelling reason to attempt to overthrow
the state.
There are few sorts of states that would be able to tolerate the pres-
ence of interventionist salvific exclusivists other than religious states
that happen to endorse their religion. However, it is sometimes pos-
sible for a liberal state to tolerate the presence of interventionist
salvific exclusivists. To do so the liberal state needs to ensure that the
interventionist salvific exclusivist will judge that she is more likely
 Coercion, Consequence and Salvation 221

to be able to make converts by non-coercive means than she is by

attempting to overthrow the state and thereby enabling the use of
coercion.20

Notes
1. The Catholic Church has generally made exceptions for morally upstanding
people who lived before the birth of Christ.
2. Specifically, Southern Baptists hold that 'There is no salvation apart from
personal faith in jesus Christ as Lord' (see the Current Baptist Faith and Mes-
sage Statement: www.sbc.net/bfm/bfmcomparison.asp; accessed 9 February
2011).
3. Strictly this claim is only true for 'expected value consequentialists' (Duncan,
2007, pp. 6-7). A further complication is that whereas on most views sal-
vation is available to an unlimited number of people, Jehovah's Witnesses
believe that salvation is only available to a limited number of people.
4. Another possibility is that the same argument might be compelling for
strong salvlfic preferentialists, who hold that God is unlikely to grant sal-
vation to very many of those who do not accept the correct set of beliefs,
join the right organization or conduct the right practices. But we will set this
possibility aside for the sake of simplicity.
S. Is it possible to coerce someone to believe that a particular religious proposi-
tion is true? Locke (1991) famously argued that it is not. Waldron presents a
counterargument (1991). I will not attempt to judge the issue here.
6. Duncan (2007, pp. 2-4) presents relevant texts in which all three authors
argue for the violent suppression of heresy on salvific grounds.
7. Cited in Krop, 1989, p. 165.
8. The exact definition of coercion is a complex exercise and is often said that
coercion includes various forms of psychological pressuring, including the
use of 'social pressure' and emotional manipulation. Most liberal states are
not in a position to prevent these from being used, and may not be opposed
to their use in any case. For more on the definition of coercion see Anderson,
2006.
9. This consideration may go much of the way to explaining why Aquinas
and other leading Christian theologians have advocated particularly harsh
treatment of heretics.
10. For a recent discussion of the limits of tolerance in liberal societies see
Ignatieff, 2004.
11. It is sometimes suggested that CBA and the PP should not be directly com-
pared because, while CBA applies to situations of risk (where the probabilities
of potential outcomes may be specified), the PP applies to situations of uncer-
tainty (where the probabilities of potential outcomes may not be specified).
However, most real world situations involve a mix of risk and uncertainty.
In these many cases both CBA and the PP can be applied, so it seems
appropriate to compare the two (Clarke, 2009, pp. 161-2).
12. The interventionist salvific exclusivist should also consider the possibility
that any new state that is established after a successful coup may fail to
continue to deliver the benefits that motivated the coup (in this case the
222 Steve Clarke

ongoing opportunity to coerce acceptance of a particular religion). Thanks

to Russell Powell for this nice point.
13. Strictly, Vallentyne and Kagan (1997) are defending the coherence of a
slightly broader claim than the one with which I am concerned. They
are concerned to defend the coherence of comparisons involving infinite
numbers of 'locations', which include infinite series of temporal locations
(eternity) as well as infinite populations of people.
14. For a defence of CBA against a variety of criticisms see Schmidtz, 2001.
15. There are some versions of the PP that do allow for the consideration of ben-
efits, which are then weighted against risks. But such 'weak' versions of the
PP are not importantly distinct from CBA, so they are not part of an alter-
native approach to decision-making under risk which we need to consider
here. For more on the distinction between 'weak' and 'strong' versions of
the PP and their relation to CBA see Sunstein (2005, pp. 18-26) and Clarke
(2009, pp. 163-5).
16. Gigerenzer (2004) estimates than an extra 350 road fatalities occurred in
America in the final three months of 2001 as a result of people avoiding
air travel following the events of September 11 2001. This is a higher num-
ber of deaths than the 266 deaths involved in the four crashed flights of
September 11 2001 combined.
17. In some liberal states, such monitoring may be problematic due to concerns
about privacy. It is hard to know how to balance concerns about privacy
with concerns about the security of a state and there is a lack of agreement
in contemporary liberal states about how to strike such a balance.
18. Even failed attempts to overthrow liberal democratic states can be damaging
as liberal democratic states may respond to such attempts by becoming more
willing to restrict the liberties of minorities, in order to minimize the pos-
sibility of being overthrown. Indeed, liberal democratic states have a long
history of reacting in exactly this way in response to real and perceived
threats (Ignatieff, 2004, pp. 54-81).
19. The religion in question will, in all likelihood, receive extremely bad pub-
licity in the media in the event of an unsuccessful coup, and this may be a
further barrier to future conversions.
20. Thanks to participants in 'The Concept of God and the Cognitive Science of
Religion: an International Conference' held at the University of Birmingham
in 2009, as well as to audiences at the Centre for Applied Philosophy and
Public Ethics, Canberra Division, Oxford University and the University of
South Carolina, and also to Russell Powell and Yujin Nagasawa for helpful
comments on earlier versions of this essay.

References
Adraoui, M.-A. (2008) 'Purist Salafism in France', ISIM Review 21: 12-13.
Anderson, S. (2006) 'Coercion', Stanford Encyclopedia of Philosophy; avail-
able online at www.seop.Jeeds.ac.uk/entries/coercion/ (accessed 3 February
2011).
Aquinas, T., Summa Theologiae, English trans. (1265-74); available online at www.
newadvent.org/summa/index.html (accessed 3 February 2011).
 Coercion, Consequence and Salvation 223

Avalos, H. (2005) Fighting Words: The Origins of Religious Violence (Amherst:

Prometheus Books).
Clarke, S. (2009) 'New Technologies, Common Sense and the Paradoxical Precau-
tionary Principle', in P. Sollie and M. Duwell (eds), Evaluating New Technologies:
Methodological Problems for the Ethical Assessment of Technology Developments
(Dordrecht: Springer), pp. 159-73.
Clarke, S. (2010) 'Cognitive Bias and the Precautionary Principle: What's Wrong
with the Core Argument in Sunstein's Laws of Fear and a Way to Fix It', Journal
of Risk Research 13: 163-74.
Duncan, C. (2007) 'The Persecutor's Wager', Philosophical Review 116.1: 1-50.
Garcia, j. L.A., and M. T. Nelson (1994) 'The Problem of Endless Joy: Is Infinite
Utility Too Much for Utilitarianism?', Utilitas 6: 183-92.
Gigerenzer, G. (2004) 'Dread Risk, September 11 and Fatal Traffic Accidents',
Psychological Science 15: 286-7.
Himma, K. (2002) 'Finding a High Road: The Moral Case for Salvific Pluralism',
International Journal for the Philosophy of Religion 52: 1-33.
Ignatieff, M. (2004) The Lesser Evil: Political Ethics in an Age of Terror (Edinburgh:
Edinburgh University Press).
Jones, E. M. (1967) The Church-God's Plan for Man: The Teaching of t11e Second
Vatican Council (London: Burn and Oates).
Krop, H. A. (1989) 'Duns Scotus and the Jews: Scholastic Theology and Enforced
Conversion in the Thirteenth Century', Nederlands Archiefvoor Kerkgeschiedenis
69: 161-75.
Lewis, D. (1989) 'Mill and Milquetoast', Australasian Journal of Philosophy 67.2:
152-71.
Locke, J. (1991) 'A Letter concerning Toleration', in J. Horton and S. Mendus
(eds), John Locke: A Letter concerning Toleration in Focus (London: Routledge),
pp. 12-56.
Manson, N. A. (2002) 'Formulating the Precautionary Principle', Environmental
Ethics 24: 263-74.
Mill, J. S. (1859) On Liberty (London: J. W. Parker and Son).
Sandin, P. (2007) 'Common Sense Precaution and Varieties of the Precaution-
ary Principle', in Tim Lewens (ed.), Risk: Philosophical Perspectives (London:
Routledge), pp. 99-112.
Schmidtz, D. (2001) 'A Place for Cost-Benefit Analysis', Philosophical Issues
(A Supplementto Nous) 11: 148-71.
Spector, H. (2008) Autonomy and Rights: The Moral Foundations of Liberalism
(Oxford: Oxford University Press).
Sunstein, C. R. (2005) Laws of Fear: Beyond the Precautionary Principle (Cambridge:
Cambridge University Press).
Turner, N. L. (2006) 'Jewish Witness, Forced Conversion, and Island Living: John
Duns Scotus on jews and Judaism', In Michael Frassetto (ed.), Cl1ristian Attitudes
toward the Jews in the Middle Ages: A Casebook (London: Routledge), pp. 183-209.
Vallentyne, P., and S. Kagan (1997) 'Infinite Value and Finitely Additive Value
Theory', The Journal ofPIJilosophy 94.1: 5-26.
Waldron, ]. (1991) 'Locke: Toleration and the Rationality of Persecution', in
]. Horton and S. Mendus (eds), fohn Locke: A Letter concerning Toleration in Focus
(London: Routledge), pp. 98-124.
10
Polarized Yet Warranted
Christian Belief
David Efird

Does the world and our experience of it constitute evidence for God's
existence, or does it constitute evidence against his existence? This ques-
tion has inspired seemingly endless debate with no rational resolution
in sight. To cite some classic contemporary exponents of either side
of the debate, on one side, Swinburne (1991a) argues that a variety of
aspects of the world and our experience of it constitutes evidence for
God's existence, an argument he summarizes thusly:

Why believe that there is a God at all? My answer is that to sup-

pose that there is a God explains why there is a world at all; why
there are the scientific laws there are; why animals and then human
beings have evolved; why humans have the opportunity to mould
their characters and those of their fellow humans for good or ill and
to change the environment in which we live; why we have the well-
authenticated account of Christ's life, death and resurrection; why
throughout the centuries men have had the apparent experience of
being in touch with and guided by God; and so much else. In fact, the
hypothesis of the existence of God makes sense of the whole of our
experience, and it does so better than any other explanation which
can be put forward, and that is the grounds for believing it to be true.
(1991b)

On the other side of the debate, appealing to other aspects of the world
and our experience of it, Rowe (1979) argues that the evil we experience
constitutes evidence against God's existence. It seems, then, that the
world and our experience of it constitutes mixed evidence, that is, evi-
dence which seems to confirm and evidence which seems to disconfirm
the existence of God. Now, say that a person has either a fairly confident

224
 Polarized Yet Warranted Christian Belief 225

belief that God exists or a fairly confident belief that God does not, and
she is then presented with the above mixed evidence for God's exis-
tence, a situation many students taking philosophy of religion courses
find themselves in, as they read Swinburne's and Rowe's now classic
arguments. How should the consideration of this mixed evidence affect
the confidence level she has in her belief?
It seems that her confidence in her belief should decrease, since, after
all, the evidence is mixed. But this isn't what tends to happen. Religious
belief, as Elizabeth Weiss Ozorak (1989), Paul Klaczynski and Gayathri
Narasimham (1998), and Paul Klaczynski (2000) have shown, is sub-
ject to a phenomenon known as 'belief polarization', like many of our
other beliefs, such as political beliefs (Chang, 2003), moral beliefs (Epley
and Caruso, 2004) and other types of evaluative belief (Lord and Taylor,
2009), where consideration of mixed evidence prompts an increase,
rather than a decrease, in subjective confidence in that belief because
we assimilate evidence for and against our religious belief in a biased
way: evidence which confirms our belief we assimilate in an uncritical
way while evidence which disconfirms our belief we assimilate in a criti-
cal way. Religious belief, then, tends not only to be preserved in the face
of mixed evidence but also strengthened.
Now, it might seem that the polarization of religious belief poses a
problem for its warrantedness, 1 since, it seems, religious belief, if polar-
ized, could not then be the product of cognitive processes aimed at
the truth. Consequently, religious belief, if polarized, cannot be war-
ranted, or so one might argue. In what follows, after summarizing the
phenomenon of belief polarization, I argue that at least one form of reli-
gious belief, namely, Christian belief, can be polarized yet warranted, by
providing a model on which this is the case.

1. Belief Polarization
Say that you and I disagree about whether capital punishment should
be legal. We each have some subjective confidence in our respec-
tive beliefs, but not overwhelming certainty. We are presented with
'mixed' evidence, that is, some evidence for making capital punish-
ment legal and some evidence against. How should our confidence
in our beliefs change in light of having considered this 'mixed' evi-
dence? It seems that our respective confidences in our beliefs should
decrease since the evidence is mixed. However, Charles Lord, Lee Ross
and Mark Lepper (1979) found that what seemingly should happen
to our respective confidences does not in fact happen; rather than
226 David Efird

decreasing, the subjective confidence we have in our beliefs tends to

increase.
Lord, Ross and Lepper discovered this fact in the following way. They
measured the strength with which a group of people held their par-
ticular belief on capital punishment and then divided the group into
smaller groups. They showed the two groups one of two cards, each of
which had a statement about the results of a (fictitious) research project
written on it:

Kroner and Phillips (1977) compared murder rates for the year before
and the year after adoption of capital punishment in 14 states. In 11
of the 14 states, murder rates were lower after adoption of the death
penalty. This research supports the deterrent effect of the death
penalty.
Palmer and Crandall (1977) compared murder rates in 10 pairs of
neighbouring states with different capital punishment laws. In 8
of the 10 pairs, murder rates were higher in the state with capital
punishment. This research opposes the deterrent effect of the death
penalty. (Lord, Ross and Lepper, 1979, p. 2100)

The participants were then asked about the strength of their beliefs
about the deterrence of the death penalty and the effect the research
had on the strength of their belief. Next, the participants were given
information on the study described in the card they received, includ-
ing details of the research, critiques of the research and the researchers'
replies to those critiques. The participants were then asked again about
the strength of their beliefs about the deterrence of the death penalty
and the effect the research had on the strength of their belief. The trial
was then re-run on all the participants using the card which supported
the opposite belief to that which they had seen initially. Lord, Ross
and Lepper found (1) that participants thought that the research which
agreed with their original belief was better conducted than the research
which disagreed with their original belief and (2) that participants' orig-
inal belief was held more strongly both after reading research which
supported their belief and after reading research which conflicted with
their belief. Participants' beliefs concerning capital punishment were
then polarized having examined mixed evidence regarding those beliefs.
What this study, and many others which confirm its results, 2 shows is
that we tend to evaluate evidence for our strongly held views differently
 Polarized Yet Warranted Christian Belief 227

from how we evaluate evidence against those same views. Evidence

for our views we tend to assimilate uncritically, without subjecting it
to careful scrutiny; evidence against our views we tend to assimilate
critically, subjecting it to critical scrutiny, with the result that we take
ourselves to have discredited that evidence. The result of this biased
processing of evidence is that our confidence level in our views increases
when we consider mixed evidence, since evidence for our views, uncrit-
ically assimilated, confirms our views, and evidence against our views,
critically assimilated and thereby seemingly discredited, further con-
firms our views. Consequently, the confidence level we have in our
beliefs increases; our beliefs become polarized.

2. Polarized Religious Belief

A great variety of our beliefs are subject to this phenomenon, including
our religious beliefs (Ozorak, 1989; Klaczynski and Narasimham, 1998;
Klaczynski, 2000). Given that our religious beliefs are polarized in this
way, what follows regarding their warrantedness? It might be thought
that if religious beliefs are polarized, then they cannot be warranted for
the following reason: 3

Because we assimilate evidence for and against religious belief in a

biased way, uncritically assimilating evidence that confirms religious
belief but critically assimilating evidence that disconfirms religious
belief, which then tends not only to preserve religious belief but also
to increase subjective confidence in religious belief, religious belief
is not among the proper deliverances of our rational faculties, since
it is not the product of properly functioning, truth-aimed cognitive
processes whose purpose is to provide us with true beliefs held with
an appropriate level of subjective confidence. Rather, religious belief
is the result of cognitive dysfunction, of unreliable processes which
are not aimed at the truth but rather at the preservation of religious
belief, regardless of whether the belief is true or not. Therefore, the
polarization of religious belief renders it irrational.

From this objection, it seems to follow that religious belief, if polarized,

cannot be warranted since, it seems, the polarization of religious belief
renders it irrational. 4 In what follows, I argue that this is not the case;
I argue that the polarization of religious belief does not entail its lack
of warrant, for at least one kind of religious belief, namely, Christian
228 David Efird

belief, by providing a model on which Christian belief is polarized yet

warranted.

3. A Model of Polarized Yet Warranted Christian Belief

Following Alvin Plantinga, I take it that

to give a model of a proposition or state of affairs S is to show how

it could be that S is true or actual. The model itself will be another
proposition (or state of affairs), one such that it is clear (1) that it
is possible and (2) that if it is true, then so is the target proposition.
From these two, of course, it follows that the model is possible. (2000,
p. 168; emphasis in original)

The model I propose for showing the possibility of polarized war-

ranted Christian belief begins from the model of warranted Christian
belief developed by Alvin Plantinga (2000) from the thought of Thomas
Aquinas (1948) and john Calvin (1960).
On this model, God has endowed us with a sensus divinitatis in order
that we may come to know him. The sensus divinitatis is a cognitive
mechanism which produces belief in God in a wide variety of circum-
stances, such as the contemplation of the glories of nature, which trigger
the disposition to form this belief. On this model, knowledge of God is
not arrived at by inference; rather, it is basic, immediate knowledge, akin
to perceptual knowledge. This cognitive faculty is aimed at the truth,
since God, on this model does in fact exist, and it functions according
to its design, that is, in accord with how God designed human beings,
in an appropriate environment, the world in which God has created us,
which triggers belief in God. As a result, beliefs produced by this fac-
ulty are counted as knowledge because they are produced by cognitive
processes aimed at the truth, processes which are thereby reliable, and,
in fact, it is true that God exists on this model. Because this model is
clearly possible, and because it entails that Christian belief is warranted,
it shows the possibility of warranted Christian belief.
The model can then be extended to show the possibility of polarized
yet warranted Christian belief in the following way. Not only does God
desire us to know of his existence, but also he desires to protect that
knowledge.
Thus, not only has God designed our minds with the faculty of the
sensus divinitatis and placed us in an environment which would trig-
ger the operation of that faculty, but he has also designed our minds
 Polarized Yet Warranted Christian Belie( 229

and placed us in an environment which would increase our confi-

dence in that knowledge. The environment provides mixed evidence for
God's existence, evidence which includes the glories of nature, which
tends to confirm belief in God's existence, as well as moral and natural
evil, which tends to disconfirm belief in God's existence. But God has
designed our minds in such a way that we assimilate this evidence in a
biased way such that, rather than the confidence in our belief that God
exists decreasing, as one might expect, our confidence increases. So, the
environment together with the design of our minds tends to increase
our confidence in belief in God, thereby protecting knowledge of his
existence.
Therefore, belief polarization, rather than being a product of cogni-
tive dysfunction, is a product of the proper functioning of our cognitive
faculties, at least as it functions with respect to belief in the existence of
God. Furthermore, the mixed evidence for God's existence provided by
the world and our experience of it, rather than constituting a problem
for God's existence, as one might think on some versions of the prob-
lem of divine hiddenness (see Schellenberg, 1991), this mixed evidence
confirms belief in God's existence. Indeed, the evil we experience in the
world, because it forms a part of the mixed evidence for God's existence,
helps to confirm, rather than disconfirm, belief in God. 5 That God is
hidden to some extent, that the evidence for God's existence is mixed,
then protects belief in God given the phenomenon of belief polariza-
tion. Thus, divine hiddenness together with belief polarization is a way
of belief protection for those who do believe in God. Hence, belief polar-
ization is, on this model, a feature of properly functioning cognitive
faculties aimed at the truth, functioning according to a design plan, and
functioning in an appropriate environment, namely, one which pro-
vides mixed evidence for God's existence. Therefore, because this model
is clearly possible, and because it entails that polarized Christian belief is
warranted, it shows the possibility of polarized, yet warranted, Christian
belief. Consequently, the polarization of Christian belief does not entail
that it lacks warrant.

4. Objection: The Persistence of Atheism

It might be objected that the model is not possible, since it entails that
once we hold atheistic beliefs those beliefs would be protected by the
design of our minds, namely, their susceptibility to belief polarization,
together with the environment, that is, the mixed evidence for God's
existence provided by the world, just as much as theistic beliefs would
230 David Efird

be, and this should not be possible given God's desire and ability to have
relationship with us. 6 In response, the model can be extended to show
this possibility.
The extended model begins with the claim that when we, the human
race, that is, were first created, the sensus divinitatis was functioning per-
fectly, and so all of the first humans believed in the existence of God,
and that belief was protected by the design of our minds and the envi-
ronment we were in. However, in addition to endowing us with the sen-
sus divinitatis, God also endowed us with free will, free will in the sense
of the unfettered ability to determine our own actions. Now, the first
humans, Adam and Eve, say, exercised their free will in such a way that
they sinned, that is, acted contrary to the will of God. Call this event
'the Fall'. A consequence of the Fall was the corruption of human nature,
corruption which included the impairment of the sensus divinitatis, such
that atheism now became possible. So, given the design of our minds
and our environment, atheism as a persistent belief became possible.
Now, this possibility was in no way God's intention for us. But it was
not something he could have done anything about, given the fact that
he also endowed us with free will, in the sense defined above. So, as a
consequence of the Fall, persistent atheism became possible. But, also
on this model, at the end of time, God will reveal himself fully, and
all will then not only know of his existence but be in union with him,
and so atheism will not persist, nor, a fortiori will anyone go to a realm
of eternal conscious punishment, Hell, as a result of their atheism. So,
though persistent atheism is an evil, it is merely a transitory evil and is
outweighed by the good of free will. So, it is then possible for God to
create us with the sensus divinitatis and free will and in an environment
that provides mixed evidence for his existence, all of which entails the
possibility of persistent atheism. Thus, that the model entails persistent
atheism does not show the model to be impossible.

5. Objection: The Unfalsifiability of Polarized

Christian Belief
It might be objected that by using a model based on Christian belief
in order to defend the possibility of polarized yet warranted Christian
belief I have rendered Christian belief unfalsifiable. The authors of the
original study which uncovered belief polarization write,

Our subjects' main inferential shortcoming ... did not lie in their
inclination to process evidence in a biased manner. Willingness to
 Polarized Yet Warranted Christian Belief 231

interpret new evidence in the light of past knowledge and experi-

ence is essential for any organism to make sense of, and respond
adaptively to, its environment. Rather, their sin lay in their readi-
ness to use evidence already processed in a biased manner to bolster
the very theory or belief that initially 'justified' the processing bias.
In so doing, subjects exposed themselves to the familiar risk of mak-
ing their hypotheses unfalsifiable- a serious risk in a domain where it
is clear that at least one party in a dispute holds a false hypothesis -
and allowing themselves to be encouraged by patterns of data that
they ought to have found troubling. (Lord, Ross and Lepper 1979,
p. 2107)

To this objection, I can only reply that just as it must be open to the
atheist to be able to assume the possibility of a naturalistic account of
the world if the warrantedness of her belief is in question, so must it
be open to the theist to be able to assume the possibility of a super-
naturalistic account of the world if the warrantedness of her belief is in
question. It might then mean that, because of the phenomenon of belief
polarization and that the world presents mixed evidence concerning
the existence of God, it is impossible to rationally resolve the ques-
tion of God's existence, which might explain why natural theology and
atheology seem to be degenerating research programmes (cf. Lakatos,
1970).

6. Mysterianism about the Existence of God

Those who believe, then, do so by faith, along the lines suggested by
John Bishop (2007), who has recently argued that because all publicly
available evidence can be interpreted in contrary ways, ways which
are either consistent with theism or consistent with atheism, religious
believers should be modest fideists, according to which it is sometimes
morally acceptable to hold a belief which (1) shapes how a person lives
their life, (2) is such that one may genuinely choose to make a prac-
tical commitment to its truth, and (3) cannot be evidentially justified
within any wider framework of beliefs. Such modest fideism seems to
be suggested by the nature of our minds, in particular, how we assimi-
late evidence for and against our religious views, and the nature of the
world, which presents mixed evidence regarding the existence of God.
Colin McGinn (1989) has argued that the solution to the mind-body
problem is cognitively closed to us, that is, that it is beyond the capabil-
ity of the human mind to conceive of it. Similarly, though not exactly
232 David Efird

parallel, I suggest that the nature of our minds, the nature of the world,
and the nature of the rational resolution of the question of God's exis-
tence are beyond the capability of the human mind to decide. Along
the lines of McGinn's verdict on whether we can solve the mind-body
problem,

We have been trying for a long time to solve the mind-body prob-
lem. It has stubbornly resisted our best efforts. The mystery persists.
I think the time has come to admit candidly that we cannot resolve
the mystery. But I also think that this very insolubility- or the reason
for it- removes the philosophical problem. (1989, p. 349)

the following seems to be true, given the lack of progress in rationally

resolving the question of God's existence, a lack of progress explained,
at least in part, by the phenomenon of belief polarization and the mixed
evidence concerning the existence of God presented by the world:

We have been trying for a long time to solve the problem of ratio-
nally resolving God's existence. It has stubbornly resisted our best
efforts. The mystery persists. I think the time has come to admit can-
didly that we cannot resolve the mystery. But I also think that this
very insolubility - or the reason for it - removes the philosophical
problem.

If this is correct, then a 'mysterian' (a term coined by Owen Flanagan

in his 1991, p. 313) position on the existence of God seems true: that
there could (epistemic sense of 'could') not be a rationally compelling
argument, one that should convince everyone concerned, for theism or
for atheism, because of the nature of our minds and the nature of the
world. There is, then, no philosophical problem of God's existence/

Notes
1. By 'warrant', I mean whatever it is that distinguishes true belief from
knowledge.
2. For further studies, see Gilovich, 1991, ch. 3 and Lord and Taylor, 2009.
3. This style of objection, that religious belief is the product of cognitive processes
not aimed at the truth, is familiar from Sigmund Freud (1961) and Karl Marx
(1964).
4. In a recent article, Thomas Kelly (2008) considers the epistemological implica-
tions of belief polarization, and he argues that the polarization of belief does
not affect the reasonableness of holding that belief, for those who are unaware
of its being polarized. He writes, 'I think that if one's beliefs are ones that it
 Polarized Yet Warranted Christian Belief 233

would otherwise be reasonable to hold in light of one's total evidence, then

the fact that it is a highly contingent, fragile matter that one has this particu-
lar body of total evidence rather than some other is not enough to undermine
the reasonableness of believing as one does' (2008, p. 629). However, Kelly
goes on to argue that the polarization of belief does affect the reasonable-
ness of holding that belief, for those who are aware of its being polarized. He
writes, 'In general, the fact that distorting or biasing factors played a role in
one's arriving at total evidence E does not make it unreasonable to believe in
accordance with E, provided that one is unaware of the operation of those
factors; what would be unreasonable would be to fail to adjust one's views
upon learning of the role played by those distorting or biasing factors' (2008,
p. 629). A consequence of Kelly's conclusion is that, for beliefs we believe to
have been polarized by the biased assimilation of evidence, we should be less
confident in those beliefs; indeed, it seems we should tend towards agnosti-
cism regarding the subject matter of those beliefs. Now it has been shown that
a great variety of our beliefs, particularly our evaluative beliefs, such as polit-
ical and moral beliefs, are polarized in this way (see Lord and Taylor, 2009).
Therefore, if Kelly is right, we, that is, you and I, should then tend towards
agnosticism on political and moral matters, amongst many other such mat-
ters. This is a striking consequence, and, if correct, would transform political
and moral debate. Furthermore, not only are political and moral beliefs sub-
ject to belief polarization, but also religious beliefs as well (see Ozorak, 1989;
Klaczynski and Narasimham, 1998; Klaczynski, 2000). Therefore, if Kelly is
correct, then, because religious belief is polarized, those readers who have reli-
gious beliefs should now be less confident of those beliefs, even becoming
religious agnostics.
5. This account of evil provides a theodicy similar to that of John Hick's (1966)
soul-making theodicy, a sort of confidence-making theodicy, which justifies
God's placing humans in an environment containing evil for benefiting the
relationship between God and humans.
6. I am grateful to Yujin Nagasawa for raising this objection.
7. I am grateful for Yujin Nagasawa's most helpful comments on a previous draft
of this essay.

References
Aquinas, Thomas (1948) Summa Theologica, vol. 1, trans. Fathers of the English
Dominican Province (New York: Benziger Bros.).
Bishop, John (2007) Believing by Faith (Oxford: Clarendon Press).
Calvin, John (1960) Institutes of the Christian Religion, trans. Ford Lewis Battles
and ed. John T. McNeill (Philadelphia: Westminster Press).
Chang, Chingching (2003) 'Party Bias in Political-Advertising Processing: Results
from an Experiment Involving the 1998 Taipei Mayoral Election', Journal of
Advertising 32: 83-92.
Epley, Eugene, and Nicholas Caruso (2004) 'Egocentric Ethics', Social Justice
Research 17: 171-87.
Flanagan, Owen (1991) The Science of the Mind (Cambridge, MA: MIT Press).
Freud, Sigmund (1961) The Future of an Illusion, trans. and ed. James Strachey
(New York: W. W. Norton).
234 David Efird

Gilovich, Thomas (1991) How We Know What Isn't So (New York: The Free Press).
Hick, John (1966) Evil and the God of Love (New York: Harper and Row).
Kelly, Thomas (2008) 'Disagreement, Dogmatism, and Belief Polarization', Tile
Journal of Philosophy lOS: 611-33.
Klaczynski, Paul A. (2000) 'Motivated Scientific Reasoning Biases, Epistemolog-
ical Beliefs, and Theory Polarization: A Two-Process Approach to Adolescent
Cognition', Child Development 71: 1347-66.
Klaczynski, Paul A., and Gayathri Narasimham (1998) 'Development of Scientific
Reasoning Biases: Cognitive Versus Ego-Protective Explanations', Developmental
Psychology 34: 175-87.
Lakatos, Imre (1970) 'Falsification and the Methodology of Research Pro-
grammes', in Imre Lakatos and Alan Musgrave (eds), Criticism and the Growth of
Knowledge (Cambridge: Cambridge University Press), pp. 91-196.
Lord, Charles G., Lee Ross, and Mark R. Lepper (1979) 'Biased Assimilation and
Attitude Polarization: The Effects of Prior Theories on Subsequently Considered
Evidence', Journal of Personality and Social Psychology 37: 2098-109.
Lord, Charles G., and Cheryl A. Taylor (2009) 'Biased Assimilation: Effects of
Assumptions and Expectations on the Interpretation of New Evidence', Social
and Personality Psychology Compass 5: 827-41.
Marx, Karl (1964) 'Contribution to the Critique of Hegel's Philosophy of Right,
Introduction', in idem and Friedrich Engels, On Religion, trans. Reinhold
Niebuhr (Chico, CA: Scholar's Press).
McGinn, Colin (1989) 'Can We Solve the Mind-Body Problem?', Mind 98:
349-66.
Ozorak, Elizabeth Weiss (1989) 'Social and Cognitive Influences on the Devel-
opment of Religious Beliefs and Commitment in Adolescence', Journal for the
Scientific Study of Religion 28: 448-63.
Plantinga, Alvin (2000) Warranted Christian Belief (New York and Oxford: Oxford
University Press).
Rowe, William L. (1979) 'The Problem of Evil and Some Varieties of Atheism',
American Pllilosopliica/ Quarterly 16: 335-41.
Schellenberg, J. L. (1991) Divine Hiddenness and Human Reason (Ithaca, NY:
Cornell University Press).
Swinburne, Richard (1991a) Tile Existence of God, rev. edn (Oxford: Clarendon
Press).
Swinbrune, Richard (1991b) 'The Justification of Theism', Truth Journal; available
online at www.leaderu.com/truth/3truth09.html (accessed 4 July 2011).
Part VI
The Compatibility of Science
and Religion
11
Freedom, Science and Religion
Katherin A. Rogers

1. Introduction: Christian Freedom

The interrelated questions of free will, moral responsibility,

self-reflectiveness and formation of character are central for the
Christian life. In recent decades experimental psychologists, and
philosophers working on the data they provide (henceforth shortened
to just 'experimental psychologists'), have produced exciting evidence
related to these issues - evidence about how we go about choosing,
thinking about our choices and developing our characters. This evidence
and future work along these lines can prove invaluable in helping those
who are interested in choosing wisely and well and in building good
character. For example, recent experiments show that it is surprisingly
common for us to misremember or to fail to grasp just what it is we
have done or are doing and why. So experimental psychology has much
to offer to the Christian believer. However, some experimental psychol-
ogists evince an attitude bordering on contempt towards their fellow
human beings and an eagerness to go beyond what the evidence war-
rants to make sweeping claims about human nature including that we
do not have free will or robust characters. These claims are antitheti-
cal to Christianity, or indeed to any view which attributes to human
beings a special dignity as free, rational agents. I will not offer a lengthy
rehearsal of the arguments reconciling the experimental evidence with
the existence of free will and character (Alfred Mele [2009] has recently
produced a good book on the topic). Rather, I will sound a cautionary
note concerning experimental psychology. I will defend the claim that
there is a practical danger in denying human freedom by appealing to
evidence which can be found in the way experimental psychologists go
about their business. This is not a condemnation of the experimental

237
238 Katherin A. Rogers

approach. I will suggest ways in which future experiments can be con-

ducted and evaluated which may help us understand and perhaps even
improve upon our situation as intentional actors, but which avoid some
of the worrisome aspects apparent in present work.
The assumption of human freedom is central to the Christian con-
ception of the relationship of the human being to God. Doctrines
concerning sin and salvation, reward and punishment, depend upon the
underlying claim that humans are free and morally responsible. More
fundamentally, it is in being free, rational agents, with some control
over the sort of people that we are, that we are said to be exceptional
among animals as specially made in the image of God. This insistence
on human dignity entails two practical corollaries concerning human
equality. The Christian claim, most counter-intuitive to the classical
Greco-Roman culture within which Christianity first grew, is that all
human beings- even the poor or enslaved! even women and children! -
are at the core metaphysically equal. We are each of tremendous value
having been made in the image of God (Rogers, 2008, pp. 56-60). The
practical consequence is that, on a fundamental level, everyone deserves
equal respect.
The Christian worldview entails a practical human equality through
another important route. If we are indeed such metaphysically elevated
creatures, how is it that we always and everywhere fall short of behav-
ing as well as we could? The story of humanity contains much that
is dismal and vile. The Christian explanation is that, at the dawn of
human history, things went badly wrong. Humans turned against God
and introduced a state of war between humans and God, between one
human being and another, and even within the human being itself. The
war goes on in all of us under the name of Original Sin. Sometimes one
reads that contemporary anthropology or sociology or psychology has
disproved this doctrine. But alternate explanations for the ubiquity of
evil, for example that we behave badly due to our evolutionary history,
do not really capture the phenomenology of moral evil (Rogers, 2002,
p. 79). If we allow, as the doctrine of Original Sin has it, that we are none
of us as good as we ought to be, we should recognize a basic equality
among human beings. We should see that we are flawed and so no one
among us should set himself over the rest as somehow ultimately and
intrinsically superior. 1 If the traditional doctrine of human freedom is
explicitly and widely abandoned, these corollaries concerning equality
might go with it.
Happily, the sweeping claims regarding the lack of freedom and char-
acter go far beyond what the evidence actually warrants. Often, the
 Freedom, Science and Religion 239

experimental psychologist does not even offer a definition or descrip-

tion of free will. On almost any philosophically plausible understanding
of free will the experiments do not really speak to the question of
whether or not such a thing constitutes part of the human being's men-
tal landscape. As to character, there are myriad ways of understanding
the evidence which purports to show that character has little to do
with behaviour, such that a more thorough examination of the range of
possibilities might well preserve our traditional understanding of char-
acter. Thus the evidence does not actually threaten the Christian view
of human nature. As so often happens, the claim that 'science conflicts
with religion' is unfounded. Still, the overly ambitious claims of some
experimental psychologists may pose a danger; a danger which, I will
argue, is evidenced in their work.

2. Wegner and Doris: Evidence against Freedom

and Character
I take as my prime examples of cause for concern Daniel Wegner's
provocative The Illusion of Conscious Will and, to a somewhat lesser
extent, John Doris's Lack of Character. Each proposes, at book length, a
sweeping revision of our understanding of human nature. My discussion
of their views will be sketchy, I do not say that all work in experimental
psychology exhibits the flaws I point to in Wegner and Doris, and by
'Doris' and 'Wegner' here I do not mean the men, but rather the voices
in the books. My criticism is aimed at attitudes and assumptions which
are implied by the text, but it may be that Doris and Wegner, the men,
do not now hold the views which I associate with their names. It is pos-
sible that they never did. I will attribute unseemly attitudes to Doris and
Wegner- the voices in the text- but one of the themes I develop is the
importance of charity and respect for one's object of study, and I do not
want to violate my own principle. I take it, though, that the only respect
we owe the text is a fair interpretation.
The thesis of The Illusion of Conscious Will is that human beings do
not have the sort of free will which we usually take ourselves to have.
Wegner does not spell out what he means by free will and his positive
view of what is going on in the human being is unclear and seems to
shift over the course of the book. But one major theme is that our 'con-
scious will' -again not carefully explained, but including our conscious
motives and intentions and choices - is merely epiphenomenal. There
are body events (Wegner seems to be a mind/body dualist) which pro-
duce conscious events including intentions and so forth and which also
240 Katherin A. Rogers

produce subsequent overt bodily behaviours. Since the conscious events

are usually closely followed by these subsequent behaviours, we suppose
that the behaviours are intentional actions produced by our conscious
will. But in fact the apparent actions and the preceding conscious 1Will-
ings' are merely constantly conjoined, they are not causally connected.
But if all that mental activity is epiphenomenal then we are not free as
we thought we were (Wegner, 2002, p. 68). This is an extremely rad-
ical position. Should you, right now, decide to raise your right arm
(the reader may try this on his own) and then up goes your arm, it
is not that the decision is the, or even a, cause of the raising of the
arm. Some earlier body event of which you are unaware is the cause of
both. Were Wegner's analysis correct, it would rule out not just libertar-
ian theories of freedom, but most, if not all, compatiblist theories. Most
standard compatibilist views allow that agents may be free and responsi-
ble, even if their choices are causally determined, as long as their actions
are voluntary, the consequence of their own deliberations, interests and
desires. Wegner's theory denies that agents' deliberations and so forth
play any causal role in the production of their actions.
What is Wegner's evidence for this extreme theory? Wegner leans
heavily on experiments by Libet which, apparently, show that a special
sort of brain activity precedes the choice (or urge) to flex a finger. Does
this prove that the unconscious brain event, and not the choice, consti-
tutes the sufficient cause of the finger flex? No, the choice might be part
of the chain of causes of which an earlier link was the unconscious brain
event (Mele, 2009, p. 11). Libet himself holds that freedom may be evi-
denced by the fact that apparently between the brain event and the flex
the agent might ,veto' the choice and not flex. And even a robust liber-
tarian brand of freedom, such as that proposed by Robert Kane, would
expect to find special brain activity before a choice. On Kane's analy-
sis, a choice follows upon a ,struggle' to pursue one of two incompatible
desires or intentions (Kane, 1996, pp. 112-15). The brain event in Libet's
experiments may correspond to this struggle.
The bulk of Wegner's evidence consists in experiments and obser-
vations which demonstrate a disconnect between the actual bodily
behaviours of an agent and what the agent believes about what he or she
will do or has done. This is supposed to show that it is not the agent's
conscious will that produces bodily behaviour. Many of the experiments
and observations are extremely interesting and informative, but they do
tend to focus on the odd, the anomalous and the truly bizarre. That
there is a disconnect between intention and action in the unusual case
does not show that there is always a disconnect. Rather the opposite;
 Freedom, Science and Religion 241

what makes a case unusual is that the connection which seems to be

there in the usual case is missing. Wegner concludes his argument with
a review of Daniel Dennett's criticisms of libertarian freedom, which crit-
icisms do not lack for responses in the literature and which, in any case,
do not support Wegner's epiphenomenalism. The case for his radical
position is excessively thin.
The same can be said for Doris's argument in favour of situation-
ism, the thesis that character as commonly understood rarely if ever
exists and that behaviour is best explained by elements in the situation
immediately surrounding an agent's actions. Ordinarily we suppose that
people have character traits, strong tendencies to act in certain ways,
including virtues and vices. We take it that character plays a significant
role in producing and explaining behaviour. In fact, says Doris, minor
factors in the agent's situation play a much larger role than whatever
character traits we might attribute to a person. Character has so little
explanatory value that it is reasonable to conclude that character as tra-
ditionally understood exists rarely, if ever. Doris chooses the virtue of
compassion as his test character trait and reviews a series of experiments
and observations wherein, he holds, the compassionate act, or the fail-
ure to act compassionately, is most often elicited by a small factor in
the situation. But if something minor can make people act more or less
compassionately, then probably the subjects do not have a strong ten-
dency towards behaving in a compassionate way. Few people, then, can
be said to have compassion as a virtue, and it is the same, presumably,
for other character traits.
As with Wegner's evidence, the experiments and observations are
extremely interesting and informative. But, again, the conclusion far
outstrips the evidence. One problem, on which I will elaborate below,
is that one person's minor factor may be another's powerful motiva-
tor. Moreover, the choice of compassion may be a poor one for making
the broader situationist case. Even if very few people have a fixed ten-
dency towards compassionate activity that would not demonstrate that
other character traits are not present robustly in great swathes of the
population. For example, raising children is a long and difficult task,
but most parents do not let small things interfere with their exercising
parental solicitude. Some do, of course, so we cannot say it is a neces-
sity of human nature to be solicitous towards one's children. Might it
not fit the definition of a virtue? What about gluttony? The gluttons of
my acquaintance are rarely deterred by anything, large or small, from
eating too much and devoting too much time and energy to think-
ing about, watching television shows about and pursuing food. lt seems
242 Katherin A. Rogers

wildly counter to ordinary experience to hold that no one or only very

few people have gluttonous characters.

3. A Dangerous View
Wegner and Doris are overly eager to conclude that human agents
do not have freedom and character. And isn't there a danger of dire
consequences in advertising these conclusions? If so, they should be
disseminated only if there is powerful evidence in their favour. Dennett,
who is often quoted on this point by experimental psychologists,
including Wegner, assures us that the danger is the invention of an over-
wrought imagination. Although we do not have the sort of freedom
that would ground 'before-the-eyes-of-God' guilt, there is still reason
to hold people responsible and punish them. We do so in order to
prevent and deter further crime. So even in the absence of a robust free-
dom, such as libertarians insist upon, we have as much freedom as we
should want or need to maintain the fabric of society (Dennett, 1984,
pp. 156-72). Even the radical claims of Doris and Wegner are consistent
with the thesis that punishment may change behaviour (though Weg-
ner will have to tell an odd story absent any effect of conscious will on
body behaviour). So both could hold, along Dennett's lines, that their
claims do not undermine ascriptions of freedom and responsibility in
any harmful way.
But, from the perspective of the Christian, Dennett rather misses
the point. True, much of the free will literature revolves around justi-
fied punishment. But the question of punishment is a by-product of
the more fundamental question of what sort of things we are. In the
Christian tradition the point of having free will is that we should be
able to be good on our own (in some very limited, creaturely way)
and so be the best images of the divine that we can be. If we are
not free enough for before-the-eyes-of-God guilt, then we are not free
enough for before-the-eyes-of-God praiseworthiness. We are not the
exceptional, made-in-the-image-of-God creature that the Christian tra-
dition has taken us to be. We do not have the human dignity and the
elevated value which the religious perspective ascribed to us. Dennett's
assumption that the whole reason for holding people responsible and
punishing them is so that we may use them to benefit the rest of us
illustrates the point nicely.
This denial of human freedom and character carries with it the dan-
ger of rejecting the corollaries about equality - that all human beings
need to be treated with respect and that no one should set himself up as
 Freedom, Science and Religion 243

intrinsically better than the mass of men. This danger becomes sharply
focused when the denial of freedom comes from those who study the
causes of human behaviour. History demonstrates that people are always
looking for ways to control others and that the 'others' are usually the
worse for it. Knowledge is power. A standard impetus for the study of
the natural world has been to exert control over it. Could the study
of human beings through experimental psychology inspire and enable
those who would like to do the controlling? Not necessarily. If the exper-
iments are done respectfully and the conclusions are drawn cautiously,
experimental psychology need not encourage the power seekers. On the
contrary, in helping all of us understand the connections between our
desires, motivations, intentions, actions and characters, experimental
psychology may confer upon the individual agent more self-control and
autonomy than he could have had otherwise. But what if the respect and
caution are lacking? Suppose the extreme claims of Doris and Wegner
were to become the consensus in our society? Suppose we come to
accept that no one is ever free and responsible except in Dennett's
punishment-serves-a-social-purpose way? Then, in attempting to con-
trol others through behavioural engineering, we do not impose some
new restraints on them; we simply replace one set of stimulae thrown
up by a blind nature with a new set produced and monitored by the
expert in human behaviour. And isn't that an improvement? That, at
least, is how the argument of the power-seekers might proceed.
C. S. Lewis in That Hideous Strength, published in 1943, has one of his
villains explain the point of studying human beings: 'If Science is really
given a free hand it can now take over the human race and re-condition
it: make man a really efficient animal' (1997, p. 383). Later in the book
an even worse villain amplifies: 'You know as well as I do that Man's
power over Nature means the power of some men over other men with
Nature as the instrument' (1997, p. 519). B. F. Skinner in Walden Two,
published five years later, puts almost the same words in the mouth of
his hero. Frazier, Skinners's mouthpiece in the book, defends a contem-
porary version of Plato's old, anti-democratic claim that the best society
is the one in which the expert is in charge: 'Are the people skilled gov-
ernors? No. And they become less and less skilled, relatively speaking,
as the science of government advances ... when we've once acquired a
behavioral technology, we can't leave the control of behavior to the
unskilled' (1976, pp. 250-1). Later (and one must give Skinner credit
for honesty) Frazier notes that there is a 'curious similarity' between
him and God, and then - as if he had just been reading That Hideous
Strength- says, 'When we ask what Man can make of Man, we don't
244 Katherin A. Rogers

mean the same thing by "Man" in both instances. We mean to ask what
a few men can make of mankind. And that's the all-absorbing question
of the twentieth century. What kind of world can we build - those of
us who understand the science of behavior?' (p. 279). Skinner's extreme
behaviourism is now passe, and one might suppose that concern about
experimental psychology is paranoia induced by exposure to A Clock-
work Orange at too impressionable an age. I will argue, to the contrary,
that there is evidence in Doris and Wegner that fear is warranted.

4. 'Using' the Subject

My evidence is of three sorts. First, there is a problematic willingness to
'use' the participant in the experiment in various ways. Second, there
is a lack of charity towards the participant in interpreting the experi-
ment. Both of these points speak to an absence of the sort of respect
that the traditional view of the free human person called for. Third,
and most importantly, Doris and Wegner set themselves above the mass
of humankind in that their operating assumption is that they are not
subject to the sort of limitations they have discovered in all the rest of
us. In discussing each sort of evidence, I will point to ways in which
experimental psychology might proceed without the worrisome factor.
Some experimental psychology involves setting up situations and
then observing how passers-by respond. Some involves experiments in
which the participants know that they are part of a study. In neither
case are people coerced to participate. The Institutional Research Boards
(IRBs) at universities are charged with seeing that experiments using
human beings evince a 'respect for persons' which ensure that partici-
pants are not coerced into participating. At my university a requirement
in many psychology courses is participation in studies, but substitution
of a written paper is usually allowed, and, in any case, no one is forced to
take psychology courses. Coercion to participate is not the problem. But
even a voluntary participant may be 'used' in ways that are troubling.
The famous Milgram obedience experiments cited by Doris and Wegner
illustrate two areas of concern. Apparently the contemporary commu-
nity of experimental psychologists disapproves of these experiments as
they were originally done, but they continue to conduct experiments
which demonstrate the objectionable features to which I will point.
In the Milgram experiments paid volunteers were told that they were
participating in an experiment concerned with the effect of punish-
ment on memory. A lab-coated 'experimenter' explains to the volunteer
that he is to administer a shock to a fellow volunteer if he gives
 Freedom, Science and Religion 245

incorrect answers to questions and that, as the number of wrong answers

increases, the intensity of the shocks increases. Actually the other 'vol-
unteer' is a confederate who does not feel any shock. As (supposedly)
more shocks of more intensity are administered the confederate begins
to scream, to claim that his heart is bothering him, and to beg to be
released. If the 'shocks' continue past a certain point the confederate
falls silent. No response is considered a wrong answer and the volunteer
is told to continue the shocks. If, at any point in the proceedings, the
volunteer balks at administering another shock, the 'experimenter', in a
firm but polite tone, asks the volunteer to continue, says the experiment
'requires' that he continue and so forth. At any time the volunteer can
refuse to administer more shocks with no reprisal. The dismaying result
was that the majority of volunteers were willing to administer shocks
past the point at which the confederate had fallen silent (Doris, 2002,
pp. 39-42).
An IRB today would probably not permit a recreation of the whole
Milgram experiment, though a recreation of part of the experiment
was recently conducted by jerry M. Burger. It produced roughly similar
results (Burger, 2009). Burger noticed, in reviewing the Milgram data,
that a large percentage of those who were willing to administer shocks
up to 150 volts were willing to go all the way. He recreated the experi-
ment, with the difference that once a participant agreed to go past the
150-volt threshold, the experiment was stopped so that the participant
did not have to go through the ordeal of deciding whether or not to
administer more severe shocks. Also, the participant was immediately
debriefed. The IRB would likely judge, of the original experiment, that
the harms might outweigh the benefits - the harms being the imme-
diate stress and anxiety suffered by the participants, any subsequent
psychological harm to them, and the possibility of harm to their repu-
tations if their behaviour during the experiments should become public
knowledge. These are well-taken concerns, but they do not exhaust the
worrisome features. A further cause of concern in this experiment is that
the subject is simply being lied to. Many, if not most, of the experiments
recorded by Doris and Wegner involve lying to or otherwise deceiving
the subjects, as does the recent recreation of the Milgram experiment.
It is a curious phenomenon - at least to the philosopher steeped in the
Christian tradition- that experimental psychologists seem to take it for
granted that lying to people is acceptable in the setting of the exper-
iment. This is not 'play-acting' lying, such as one might do on stage,
since the participants need to be genuinely deceived in order for the
experiment to provide the sort of evidence being sought. But deceiving
246 Katherin A. Rogers

someone is a classic instance of 'using' them. It is treating one as a

means to the ends of the deceiver. Doesn't the liar express an unwhole-
some attitude of superiority over the participant? A common principle
upon which IRBs operate today is that deception is permissible (assum-
ing other principles of beneficence and justice are met) if the participant
is immediately debriefed after the experiment. This does seem to miti-
gate the concern, but does not allay it entirely. Presumably none of us
like being lied to by plumbers or waiters or policemen. (Many of the
untruths in the mouths of politicians probably fall into the 'play-acting'
category, since almost no one is really deceived.) And even if the waiter
tells us later that he lied, it is not clear that that undoes the wrong. (I use
this example because a student of mine argued that deception in exper-
imenting is permissible because it is no more harmful than the lying he
did as a waiter, always recommending the most expensive item so he
would get a big tip.) Is experimental psychology not governed by basic
rules of honesty which one would hope to apply in other professions?
One might respond that lying to the participant in the experiment
is permissible as long as it does not do them any harm - especially if
they are immediately debriefed. A first response is that lying to some-
one, even if it does them no harm, is still wrong. Many well-developed
and defensible ethical theories hold as much. Experimental psycholo-
gists tend (at least this is the impression one gets from the literature
and discussion of the IRB principles) to be consequentialists. And this
should raise questions in that most versions of consequentialism, as
they address the human condition, assign the locus of ultimate value
to something other than the intrinsic worth and dignity of each human
being. A second point to note is that, even if a deceiving experiment did
not harm the participant, it might harm the experimenter. What if Aris-
totle is right and Doris is wrong? What if we do have characters and we
formed them at least in part through our actions? Burger's understand-
ing of the Milgram experiments is of interest on this point. He takes the
evidence to show that someone who continued to give shocks up to the
150-volt level was more likely to continue to the bitter end. This might
be because someone who is willing to give a shock of 150 volts is already
the sort of person who would be equally willing to give a much more
damaging shock. But an Aristotelian could take the evidence to indicate
that choosing to go to 150 volts made one the sort of person willing to
go much further. If so, minor choices can have significant consequences
for the agent themselves. Surely it is not a good thing to be a liar. One
who lies often as part of the job, even if the lying is innocuous in the
workplace context, may become a more general liar. Thus one could
 Freedom, Science and Religion 247

argue that the experimenter is put in danger. The IRB, I am told, does
not oversee the well-being of the experimenter.
If the above point is plausible, then the deceived participant in an
experiment may be harmed in ways which IRBs do not anticipate. The
IRB takes note of stress and anguish which might be suffered during the
experiment, and it looks to the issues of risks of psychological harm or
reputational harm. From the Christian perspective there is more at stake,
however. If we are interested in studying freedom and character, the
most obvious way to construct an experiment is to place a participant in
a situation involving significant moral choice. That is what the original
Milgram experiments did, and, given Burger's thesis that the 150-volt
point is the decisive moment for whether or not a participant is likely
to proceed the whole way, the recreated experiment includes an almost
equally significant moment of choice. Many of the other experiments
which Wegner and Doris cite place participants in morally significant
situations. Let us grant that actually shocking someone badly, possibly
to death, in an experiment concerning the effects of punishment on
memory would be evil. (I do not know that 'evil' is a word much used
in the psychology literature, but, at least in the USA, experimentation
using human subjects is to be guided by the report, issued in 1979, of the
Belmont Commission. The Commission was charged with establishing
basic ethical principles. I assume that means that the scientific commu-
nity is to be guided by what is right and wrong, good and evil. Evil,
then, should be allowed as a reality in the universe of experimental psy-
chology.) From the perspective of the moral choices and character of an
individual, if the participant believes the memory experiment to be real,
then, even in the fake version, the subject is participating in evil. The
Christian standardly holds that doing wrong and making bad choices
does great harm to the agent. One ultimate hope of the Christian is to be
praiseworthy-in-the-eyes-of-God. But you become a murderer by mur-
dering. You become a torturer by torturing (Rogers, 2007). The subjects
in the Milgram and Burger experiments were put at risk of harm - they
were placed in moral peril, danger of becoming wicked people - just by
being thrust into a situation in which they were sorely tempted to do
the wrong thing.
True, the ordinary vicissitudes of life put us in positions where we
have to make moral choices. But the fact that some event is sure to
befall us as we go about our business does not entail that it is permis-
sible for our fellows to engineer that event. Think of death. And, yes,
the few who refused to continue with the shocks may have ultimately
benefited by doing the right thing. But the results of the experiment
248 Katherin A. Rogers

indicate that overall the subjects incurred more harm than good, and,
in any case Milgram and Burger could not gauge how much harm would
be done. We might defend the experiments by noting that the harm of
making the bad choice was up to the subject themselves. True, but they
could not have made the choice had they not been put into the situa-
tion. If the vicissitudes of life put you in situations where you are sorely
tempted to do the wrong thing, that is bad luck for you. That does not
give the experimenter carte blanche to construct some bad luck for you.
The Christian must grant that God may, from time to time, for good
reasons of His own, say 'No' to our prayer that He 'lead us not into temp-
tation'. But God is a special case and to argue that the experimenter may
do what God may do would prove my point about the dangerous atti-
tude associated with denying freedom and character. Perhaps the most
worrisome point about the risk of moral harm to the tempted subject is
that it does not seem even to occur to experimental psychologists and
IRBs. Certainly Wegner and Doris do not mention it.
If lying and putting people in morally tempting situations may cause
harm, the experimental psychologist could respond that the harm
is outweighed by the good of the knowledge produced. The non-
consequentialist answers that it is still wrong to use people. But even the
consequentialist must grant that these are the sorts of harms and goods
it is exceedingly difficult to balance, especially if we add the harm done
to the agent who chooses evil. The 150-volters in Burger's experiment
may have done themselves permanent and serious damage without the
IRB even considering the possible risk. Must experimental psychologists
simply stop studying human freedom and character? It will be difficult
to construct fruitful experiments which put people in morally inter-
esting situations, but which do not deceive them. Knowing they are
being tested morally is almost certain to limit the value of the partic-
ipants' responses. But there are other avenues which the experimental
psychologist might pursue. I will mention three, but I presume there are
many more.
Let us take the question which motivated Milgram and Burger. In Nazi
Germany many otherwise apparently ordinary folk engaged in hideous
behaviour, often in obedience to orders they might have disobeyed
without terrible consequences. Could ordinary Americans in the 1960s
or the 2000s do the same? Can we study that question without using
deception and placing participants in moral peril? First, it might be
possible to construct experiments which do not need deception. They
might involve choices which do not have moral significance, but which
are somewhat similar, and so might shed light on how various factors
 Freedom, Science and Religion 249

affect moral decisions. One might ask agents to make aesthetic choices
or to make judgments about what fictional characters ought to do. If it
is important to include the fellow with the lab coat, but without decep-
tion, one group of subjects could meet an expert who will tell them
how he or she thinks they ought to choose, and then the control group
could be left to its own devices. With careful analysis of the relationship
between the aesthetic choice or the fictionalized choice and an actual
moral choice this could provide at least some evidence.
Second, there are first-person reports. We might compile lists of folks,
past and present, who have done hideous things upon being instructed
to do so. We could find out about them from written records - court
transcripts might be an example - and testimony of family and friends.
Do they seem relatively ordinary? Then we could amass first-person
reports of what they had done and why. If they are alive we could
interview them. If not, perhaps they have left testimony and letters.
Third, once the psychologist's question has been carefully framed,
simply observing people as they go about their business or have gone
about it in the past might provide good evidence. This seems obvious in
the Milgram case. A glance at the French Revolution, Russia under Stalin,
or China under Mao makes it abundantly clear that hideous behaviour
on the part of average people in obedience to orders - especially if a
patina of Progressivism and Science is added - is not unique to Nazi
Germany. The evidence of history is overwhelming and one who insists
that it couldn't happen here is naive or blinded by jingoism. Milgram
really didn't need to do his experiments in order to answer his ques-
tion. There may be other questions which, if suitably framed to focus
on relevant evidence, could be answered from observation without the
need to use deceptive experiments.

5. Interpreting Evidence Uncharitably

A second sort of evidence that there may be something worrisome in
the attitudes of experimental psychologists towards their subjects lies in
their frequently assigning an uncharitable interpretation to the results
of an experiment when a charitable interpretation is available. I offer
two examples, but there are many more. Doris, to defend situation-
ism and undermine the position that people have robust characters,
needs to show that behaviour is significantly affected by relatively minor
variations in situation. This is because we would expect extreme situa-
tional factors to override even robust character traits. For example, even
the most compassionate person is likely to behave without compassion
250 Katherin A. Rogers

under threat of death. Doris notes that in the Milgram case there is no
threat of death, punishment, or indeed negative consequences of any
sort, made against the participant should he stop the shocks. Doris then
takes the Milgram experiments to show that a minor factor, the instruc-
tion of the 'experimenter', has a profound effect on the behaviour of the
subjects. So these experiments provide evidence of lack of compassion.
But there is a more charitable interpretation (which is still consistent
with my point above concerning recent history's unhappy tale about
human capabilities): in our society science is taken as the paradigm
of proper knowledge acquisition. The scientist is treated with enor-
mous respect. In some contexts- doubt about evolution, for example-
those who do not accept the pronouncements of science are held up to
ridicule. And given the wonderful advances science has made in the last
five hundred years this attitude of respect and deference towards the sci-
entist and their discipline seems appropriate in many or most cases. Add
that fear of embarrassment is an extremely powerful motivator. We very
often remember our embarrassing moments with much more vivid pain
and regret than we suffer when we remember those moments in which
we failed morally. Perhaps the subjects in the Milgram experiments, hav-
ing deeply internalized our contemporary attitude towards science with
the entailment that one who is sceptical about the scientist and their
project is deserving of contempt, found it extremely difficult to disobey.
It is interesting that Milgram focuses on authority, and not the author-
ity of the scientist specifically. Milgram did conduct his experiments in
various venues to test the effect of the impact of the institution, but the
'scientist' experimenter was always present (though sometimes through
the telephone). What if the experimenter had feigned a stomach disor-
der and had handed his clipboard - with a statement about delegating
authority - to a confederate disguised as a passing janitor? It is at least
possible that the subjects in the Milgram experiments were, by and
large, compassionate people suffering under severe social pressure not
to doubt or disrespect the scientist. If so, their lack of compassion was
not the consequence of a minor situational factor.
Another example of lack of charity comes from Wegner. The subject
is told he is taking a test involving speed typing. He is at a computer
keyboard in a room where others (confederates) are at work as well. The
subject types until he is told to stop. He is then accused of damaging
the computer by pressing the wrong key. In fact he did not, and usually
subjects initially deny doing the deed. Then a confederate claims he saw
the subject hit the wrong key. 'Those whose "crime" was ostensibly wit-
nessed became more likely to sign a confession .... We are not infallible
 Freedom, Science and Religion 251

sources of knowledge about our own actions and can be duped into
thinking we did things when events conspire to make us feel respon-
sible' (Wegner, 2002, pp. 10-11). The claim that this result, or similar
results from a million such experiments, would provide evidence that
consciousness is epiphenomenal is bizarre. But what worries me here
is Wegner's view that the subject is 'duped'. 'Dupe' in the dictionary is
synonymous with 'fool'. True, the subject is mistaken. But given the evi-
dence available to him, the subject has in fact come to what looks to
be the correct conclusion. Is it more probable, epistemically, that you
might hit the wrong key when typing quickly, or that both the person
conducting the typing experiment and a total stranger would conspire
against you to accuse you of doing something you didn't do? A reason-
able person should probably believe that he had damaged the computer
and perhaps even try to remember doing so. Someone who stuck to
the story that he hadn't damaged the computer would likely be lack-
ing in appropriate humility, motivated by fear of responsibility, and/or
not very committed to telling the truth as he sees it. Wegner offers
many similar experiments throughout his book and never distinguishes
between making a mistake due to some sort of cognitive failure (his
example of 'table-turning' where people believed that spirits were mak-
ing a table move when it was the people themselves doing it might fall
into this category) and making a mistake when the available evidence
renders the mistaken belief epistemically justified. Lumping everyone
together as equally 'duped' fails to respect the abilities of those whose
cognitive processes are functioning properly, but have come up against
the weird and anomalous situation. And since this failure is pervasive
in Wegner's work it suggests a worrisome contempt for the subjects.
My prescription, then, is that in interpreting the results of experiments,
ascribing negative characteristics such as gullibility or a lack of compas-
sion is appropriate only after more charitable interpretations have been
tested and shown to be less plausible.

6. Assuming Superiority
My final sort of evidence is the most serious. Doris and Wegner set them-
selves outside of and above the rest of humankind in drawing their
conclusions. This is more obvious and striking in Wegner's case than in
Doris's, so let us look at the latter quickly and then devote more time to
the former. The case against Doris requires an assumption which seems
intuitively plausible; the proper practice of science (and philosophy -
Doris is actually a philosopher assessing the scientific evidence) requires
252 Katllerin A. Rogers

certain virtues such as perseverance, courage, honesty, humility in the

face of the evidence and perhaps many more. Doris's theory, situation-
ism, holds that human beings do not have virtues and vices, but behave
as they do because of minor variations in situation. Were that the case
we might expect the scientist whose research has turned up a conclusion
that poses a problem for him to be honest if he's found a five-dollar bill
(there is an old experiment supposedly showing that people acted more
compassionately when they had found a dime) but not honest if he's
stepped on chewing gum. The successful practice of science would be
rendered impossible. Doris is engaging in professional philosophy and
his trust that the experiments he reports were conducted properly sug-
gests that he takes himself and the scientists he cites to be successful
practitioners of science and philosophy. But then Doris and his confr-
eres must possess the virtues necessary for the job. Thus Doris seems to
exempt himself in concluding that people do not have characters.
The case with Wegner is more obvious and more extreme. Regarding
Doris one could argue that the 'scientific' virtues are not truly required
for the successful practice of science or that the virtues in question
are not really robust character traits. So, possibly Doris could consis-
tently apply his conclusions to himself. The same cannot be said for
Wegner. Wegner holds that conscious will - that is intentions, moti-
vations and so forth - is epiphenomenal. We are deluded in thinking
that our intentions even partially cause our overt actions. Our actions,
and possibly our intentions themselves, are caused by preceding body
events of which we are unaware. If this thesis were correct, then no one
could ever conduct experiments. A representative dictionary (Webster's)
defines 'experiment' as 'an operation carried out under controlled con-
ditions in order to discover an unknown effect or law, to test or establish
a hypothesis or to illustrate a known law'. An experiment, by definition,
is a series of actions carried out as a means to achieving a consciously
understood purpose. Experiments are done 'in order to' discover, test or
establish something. If intentions are not part of the causal explanation
of overt action then experimentation is illusory.
If the conscious will is epiphenomenal as Wegner describes it, then
the scientist might design an experiment. He may consider a thesis and
think up ways to test it. But on Wegner's view these conscious events
have no causal or explanatory impact on overt actions. The scientist
cannot conduct any experiments. Can we even say correctly that the sci-
entist can engage in certain motions such as arranging blocks, talking
to the subject and writing on a note pad? Without invoking inten-
tions in an explanatory role, it is difficult to analyse the behaviours of
 Freedom, Science and Religion 253

'arranging' versus random putting, 'talking' versus just making sounds,

and 'writing' versus making scratchings. 2 But, however we describe
these motions, on Wegner's analysis, it is not the case that the sci-
entist engages in these movements 'in order to' achieve a conscious
purpose. These actions are the direct result of, and fully explained by,
unconscious causes, which causes may or may not have also caused
the epiphenomenal thoughts. Thus the scientist's motions cannot, by
definition, constitute experiments. If Wegner's conclusion is correct, no
one has, or could, demonstrate experimentally that conscious will is an
illusion.
Yet, when Wegner describes the actions he, his colleagues and his
sources engage in he seems to assume that they perform experiments
under this definition. So, for example, he writes (my italics for empha-
sis), 'Wegner and Wheatley (1999) conducted an experiment to learn
whether people will feel they willfully performed an action' (2002,
p. 74); 'In trying to understand what kind of mental system would make a
suppressed thought come back to mind again and again, my colleagues
and I did a series of experiments' (p. 141); 'Lots of clever studies have
been aimed at this distinction' (p. 179); 'in our experiments a variety of
conditions were tested to see whether this tendency to answer correctly
could be changed' (pp. 203-4); 'To ascertain whether belief was indeed
causal, on one study we directly manipulated belief' (p. 206). The impli-
cation is that, while the rest of us poor dupes are living an illusion when
we suppose ourselves to be capable of intentional actions, Wegner and
his colleagues can do what is impossible for us. They can act for rea-
sons! Wegner allows that 'we' are all often seduced by the illusion, and
so one might think that he includes himself and his colleagues (2002,
pp. 341-2). But he does not go on to grant that no one has ever actu-
ally conducted experiments so there is no experimental evidence for his
views. He seems to take it for granted that qua scientist he is above the
rest of humankind, able to study our delusions but not sharing them.
It is this attitude that is especially disquieting. Unless Wegner is actually
superhuman, he should not exempt himself from the universal claims
he makes. And if he cannot exempt himself, he should rescind the uni-
versal claims. The Golden Rule of experimental psychology regarding
universal claims about human beings ought to be 'Attribute unto others
only those characteristics you would have attributed unto you.'
The Christian tradition coupled freedom with human dignity and
equality. Experimental psychology need not undermine that view. First
and foremost, it should limit its conclusions to what the evidence war-
rants. Further, it can be conducted with respect and charity towards its
254 Katherin A. Rogers

subjects and with the appropriate humility to recognize that conclu-

sions which apply to the subjects of experiments, qua human beings,
apply to the experimenters themselves. My prediction is that, while
experimental psychology may uncover all sorts of interesting limita-
tions regarding our ability to choose, it will not find that we are unfree
in some philosophically interesting way. On the contrary, by apprizing
us of our limitations, it can guide us towards taking more control over
our choices and in that way help us to polish the reflection of the divine
in ourselves - the very activity for which God gave us freedom in the
first place.

Notes
1. Saint Augustine of Hippo is one of the chief architects of Christian philoso-
phy and a consistent defender of the basic metaphysical and practical equality
of human beings. He famously argued that government is a species of 'nec-
essary evil' in that it is necessary due to sin, but it is disordered in that it
gives some authority to use force against others. This is disordered because all
human beings are fundamentally equal. And his antipathy towards some of
the heresies he combated, including Pelagianism, stems at least in part from
his disapproval of the claims of their.'ldherents to be a moral elite superior to
'ordinary' human beings (Rogers, 2008, p. 129).
2. Perhaps a philosophical epiphenomenalist could manage it, but Wegner's view
is both extreme and undeveloped. Note a related problem: his experiments
often involve first-person reporting, but drawing conclusions about beliefs
based on first-person reports seems justifiable only if there is some causal or
explanatory connection between the beliefs and the overt action of making
the sounds that constitute the reporting.

References
Burger, J. (2009) 'Replicating Milgram', American Psychologist 64: 1-11.
Dennett, D. (1984) Elbow Room (Cambridge, MA: MIT Press).
Doris,]. (2002) Lack of Character (Cambridge: Cambridge University Press).
Kane, R. (1996) Tile Significance of Free Will (Oxford: Oxford University Press).
Lewis, C. S. (1997) Out of the Silent Planet, Perelandra, That Hideous Strength
(New York: The Quality Paperback Book Club).
Mele, A. (2009) Effective Intentions (Oxford: Oxford University Press).
Rogers, K. (2002) 'The Abolition of Sin', Faith and Philosophy 19: 69-84.
Rogers, K. (2007) 'Retribution, Forgiveness, and the Character Creation Theory of
Punishment', Social Theory and Practice 33: 75-103.
Rogers, K. (2008) Anselm on Freedom (Oxford: Oxford University Press).
Skinner, B. F. (1976) Walden Two (New York: MacMillan Publishing Company).
Wegner, D. (2002) Tile Illusion of Conscious Will (Cambridge, MA: MIT Press).
12
The Compatibility of Science and
Religion: Why the Warfare Thesis
Is False
Michael Ruse

In the second half of the nineteenth century, a good number of

scientists - prominent among whom was Darwin's great supporter
Thomas Henry Huxley - subscribed to what is known as the 'warfare'
thesis about the relationship between science and religion (Desmond,
1994, 1997). They argued that science and religion are in conflict, and
that, if one holds to the one, one cannot hold to the other (Draper,
1875; White, 1896). The name of Galileo came up frequently in these
discussions, and it was pointed out that it was only to be expected that
the Catholic Church would have shown such opposition to so distin-
guished and fertile-thinking a scientist. Naturally enough, the recently
published theory of evolution of the English naturalist Charles Dar-
win was another topic of conversation by these conflict theorists. Again
it was suggested that one is faced with a stark dichotomy: either one
accepts that organisms had natural origins or one accepts that we all
arrived supernaturally. There is no other option.
Looking back, we can see now that this debate about the relationship
between science and religion was as much social as intellectual. People
like Huxley in Britain and his supporters both in that country and else-
where (especially including the United States of America) were pushing
strongly to reform society (Ruse, 200Sb). They wanted a modern civi-
lization, forward looking, and soundly based on science and technology.
Rightly or wrongly, they saw the in-place churches as the major source of
opposition, supporting as they did the establishment which wanted lit-
tle or no change. The warfare thesis therefore was an important weapon
of attack, whether or not it was fully sustainable either historically or
theologically or philosophically.

255
256 Michael Ruse

At the beginning of the twenty-first century, we find that history is

repeating itself. There is now a substantial fraction, drawn mainly from
the sciences but also including supporters from philosophy and else-
where, that is actively again promoting the warfare thesis. They argue
that there is inevitable conflict and opposition between science and reli-
gion. Notable proponents of this position include the British science
writer Richard Dawkins, author of the smash-hit bestseller The God Delu-
sion (2007), as well as philosopher Daniel Dennett (2006), journalist
Christopher Hitchens (2007), and sometimes student and neuroscien-
tist Sam Harris (2004). As with the nineteenth century, one suspects that
today the opposition is as much fuelled by social causes as by intellec-
tual reasons. There is great tension about the rise and place of religion in
society today. This applies both to the Christian religion, particularly the
evangelical fundamentalist variety which is pushing not only to have its
ideas included in science classrooms, but also a very conservative social
agenda with respect to such things as abortion and homosexuality, and
to other religions, particularly Islam. The horrific events in New York on
September 11 2001 are never far from anyone's mind.
In this essay I shall say little or nothing about the social aspects of the
warfare thesis. I have elsewhere discussed the nineteenth-century situa-
tion and no doubt others will discuss the twenty-first century better than
I am able. I shall rather focus on the intellectual questions. In particular,
I am interested in the question of whether or not science and religion
are necessarily in conflict. The alternative position, one which inciden-
tally has my support, is often known as the 'independence' position
(Barbour, 1990). More recently, in a rather sneering manner, it has been
referred to (particularly by the quartet mentioned in the last paragraph)
as the 'accommodationist' position. As I set out, let me stress that I am
not now trying to argue for the truth of any particular religion. I accept
science but religion must find its own advocates. What I question here
is simply whether how science is understood today poses an insupera-
ble barrier to reasonable religious belief. Mainly because modern science
has developed during the time of Christianity, I shall take that particular
religion as my touchstone, although I trust that what I have to say could
be extended to other religions.
My discussion will fall into two main parts. First I shall offer a general
argument suggesting that the grasp of science is necessarily limited, and
that questions reaching beyond that grasp are not only legitimate but
fair game for the religious person to try to answer. The second part of the
discussion will focus on a number of arguments that have been invoked
to suggest that science and religion are necessarily incompatible.
 Why the Warfare Thesis Is False 257

1. Science and the Machine Metaphor

I start with the general case. This depends crucially on the fact that
scientific understanding is inherently metaphorical (Kuhn, 1993). Scien-
tific theories are far from just faithful descriptions of raw reality. From
the beginning, they are attempts at understanding, at interpretation,
and they always employ metaphors as they try to make sense of what
we are experiencing, as they try to find ways to extend our grasp and
look for new discoveries and connections. Take, for example, Darwin's
theory of evolution through natural selection (Ruse, 2008). From the
beginning, Darwin was thinking metaphorically. As he became an evo-
lutionist, his thinking was done in terms of a 'tree of life' - a metaphor
taken from the earliest chapters of Genesis. Darwin never thought of
organic origins simply in their own terms, but instead as filtered through
his cultural conception of life climbing ever-higher up the branches of
a great diversifying tree. Then, as he started to move towards his key
mechanism of natural selection, he immersed himself in the work of
the animal and plant breeders of his day, thinking metaphorically as he
pushed from his understanding in the human realm to what he thought
might possibly be occurring in the natural order. The key insight for him
came at the end of September 1838, when he was reading a work on pop-
ulation by the Revd Thomas Robert Malthus. Malthus was talking about
population pressures and how there can never be food and space ade-
quate to speak to them. Malthus therefore talked about the 'struggle for
existence', something Darwin in a flash generalized to the whole world
of animals and plants (Ruse, 1975).
From the beginning, Darwin recognized that this was metaphorical
thinking. For a start, although sometimes there is a literal struggle, as
often as not it is meant metaphorically as organisms compete (itself a
metaphor), and moreover it is not really existence that we are talking
about but reproduction. Then came the crucial move to natural selec-
tion. Darwin took the struggle and added in the ubiquitous variation
we find in nature, arguing that there is an ongoing equivalent to the
selective practices employed by animal and plant breeders. This 'natural
selection'- as metaphorical as one can get- is the force of ongoing evo-
lutionary change. But, as Darwin pointed out, it is not random change.
It is rather change in the direction of adaptive advantage. Metaphori-
cally, Darwin saw the organic world as if it had been designed- the eye
is for seeing and the hand is for grasping - and selection speaks to this.
In short, Darwin's theory is as metaphorical as a sermon or a piece of
poetry (Ruse, 200Sb).
258 Michael Ruse

What I want to suggest - and here, I emphasize, I am not being at

all original but rather drawing on the consensus of those who write
on the history of science - is that science today is governed by one
overall, dominant metaphor, what is itself known metaphorically as a
'root' metaphor. This metaphor is that of a machine. Under the so-called
'mechanistic' viewpoint that dominates modern science, the physical
world (and this today extends to the biological and even human world)
is seen as if it is a giant machine (Hall, 1983). This metaphor goes back
to the Scientific Revolution that occurred in the sixteenth and seven-
teenth centuries. The root metaphor for the Greeks, and for subsequent
thinkers in the West down to the time of the Scientific Revolution, was
that of an organism (Merchant, 1980). Plato, in particular, set the scene
when, in his great dialogue the Timaeus, he likened the world itself to an
organism, arguing that it has a soul of its own. (This, in medieval times,
was known as the anima mundi.) It was for this reason that it was appro-
priate to think in terms of ends, of purposes, of what Plato's great pupil
Aristotle was to call'final causes'. With the Scientific Revolution all of
this was swept away. The French philosopher and mathematician Rene
Descartes argued that physical matter has no mind of its own (Garber,
1992). Its only distinguishing feature is that of extension. What we have,
therefore, after the Revolution, is simply matter in motion going on
invariably and inevitably, without end, according to blind laws. The
British scientist-philosopher Robert Boyle (1996) was fond of likening
the world to a clock. Of course you might object that machines, clocks
in particular, have ends or purposes. But as historians of the Scientific
Revolution point out, by the time the Revolution was over most scien-
tists were ignoring the possibility of ends and thinking only of the ways
in which the machinery works. In the words of one of the most distin-
guished commentators on the Revolution, God had become a 'retired
engineer' (Dijksterhuis, 1961).
There is much more that one could say about the development of the
machine metaphor in the four centuries since the Scientific Revolution.
Most particularly, there was the coming of the machine metaphor to the
biological world. It was here particularly that Charles Darwin's genius
was important, as he showed that - as we have sketched above - the
distinctive features of organisms could be explained in terms of blind
law, no less adequately than the distinctive features of inorganic objects
are explained in terms of blind law in physics and chemistry (Ruse,
1979). In the words of Richard Dawkins (1976, p. 22), organisms became
'survival machines'. Most recently, in the second half of the twentieth
century, thanks to the coming of computers, students of humankind
 Wily tile Warfare Tllesis Is False 259

have been making major advances in understanding how our thoughts

and actions can be understood as the results of sophisticated electronic
machines - machines otherwise known as computers. This of course is
the basis of the new area of enquiry known as 'cognitive science'.

2. Unasked Questions, Unanswered Questions

I shall not dwell on details of the coming of the machine metaphor
to science, but shall now move to explore its implications for our
inquiry. The important point I want to make is that metaphors have
both strengths and weaknesses, or if you do not want to talk in terms
of 'weaknesses' then let us say 'limitations'. Thanks to metaphor, you
can not only organize and explain what you have, but you can use
it heuristically to explore new areas and problems. Think just in the
area of biology, and see how regarding parts of organisms as function-
ing contrivances or machines has paid huge dividends. To take one
example, consider the strange plates that run down the back of the
dinosaur Stegosaurus. What are they for? One suggestion was that they
might be for attack or defence, but it turns out that they are not very
strongly made, so that suggestion almost certainly falls. Another sugges-
tion is that they might be for sexual attraction, but since both males
and females have the plates it is thought most unlikely that this is the
case. The most recent suggestion is the plates serve the function of heat
regulation, helping the cold blooded brute warm up in the sun in the
morning and then to cool off in the heat of the day, thanks to the wind
blowing across the plates (Farlow et al., 1976). This suggestion came
because palaeontologists were aware of similar sorts of plates playing
heat regulating roles in electrical generating stations. Because of a literal
machine, we were able to understand the metaphorical machines that
we find in dinosaurs.
The weaknesses or limitations of metaphors, as the late Thomas Kuhn
(1962) pointed out, is that they succeed only by putting blinkers on us.
They direct us to crucial areas, but at the same time rule certain ques-
tions out of court. In the terms of the metaphor, the questions simply
do not make sense and should not be asked. Suppose, for instance, I say
that my beloved is a rose. At once you know a number of things about
her. You know that she is beautiful, at least in my eyes, that there is
something fresh and young about her and that I am attracted to her.
I suppose, if you know that I'm a little bit of a joker or a cynic, you
might wonder perhaps if she is a little bit prickly. What you don't know
is how good she is at mathematics. You don't know if she is strong
260 Michael Ruse

on the subject or weak. You don't know if she is bilingual. You don't
know if she is English or French. It is not that these are not meaningful
questions, they certainly are, but rather that my metaphor just does not
address them at all. The same holds very much in science. On using a
metaphor one focuses on some things, but not on others. Suppose one
says that the DNA molecule incorporates a genetic code. This is a very
useful metaphor, inasmuch as one is now led to ask questions about how
to crack the code. But, if one asks other questions about the code, then
they may well be inappropriate. One is not for instance asking about the
religious affiliation of the molecules involved in the code. A question
like this simply does not make sense.
Now to the root metaphor of the machine. I argue that there are cer-
tain questions that the metaphor simply rules out of court (Ruse, 2010).
Under the metaphor, it is inappropriate to ask questions of this kind.
I will list four such questions, although I do not want to claim that
these are the only four questions that are ruled out. Moreover, I fully
recognize that, in the future with a new understanding of machines and
their power, one might think that some of the questions I focus on are
indeed questions that can be answered in terms of machines. {In fact, as
we shall see, already some would dispute my choice.)
The first question is about origins. One can ask how a machine was
made, but ultimately where the materials came from is simply not part
of the discourse. I can ask about where the copper used in the manufac-
ture of the clock was dug up. But really where the copper came from is
irrelevant to the making of the machine and its function. Translated
to the metaphor of the world as a machine, I suggest therefore that
science simply does not ask what is known as the 'fundamental ques-
tion': Where did everything come from, or more generally why is there
something rather than nothing? This question is simply not a scientific
question in the sense that modern science just does not try to answer it
at all.
The second question is about morality. Machines can be used for
good, machines can be used for ill ..But machines themselves are neither
good nor bad. Take for instance an electric chair. Many people, myself
included, would want to say there they can think of little good use for
an electric chair. We disapprove of capital punishment under any cir-
cumstances. Many other people, people who would be indignant if it
were suggested that they were not moral, would want to say that some-
times the electric chair can be used for good ends. If Hitler had survived
the war, for example, then it would not only have been appropriate but
 Why the Warfare Thesis Is False 261

a good thing to execute him in the chair. Generalizing again about the
world, the point is that the world-as-a-machine says nothing directly
about morality. This is a point that philosophers have made many times
in the past. As David Hume (1739) remarked, you cannot go from state-
ments about matters of fact to statements about matters of obligation.
It is fallacious. What I'm trying to do here is show precisely why such a
move is fallacious.
The next question is probably more contentious. Machines are not
conscious. This is the point made by the great German philosopher
Leibniz (1714). Machines can do many things, but ultimately they are
not conscious. So, science does not speak to the question, What is
consciousness? Perhaps in this day and age one needs to modify this
claim just a little bit. Possibly you might want to argue that, ultimately,
really sophisticated computers will be conscious. Then, I would say that
talking about the machine at the material level tells us nothing about
why exactly the machines are consciousness, if by being conscious you
mean sentience. There is a gap - what the philosopher-psychologist
David Chalmers (1996) has called the 'hard problem' - between the
way the machine works and the consciousness that emerges. So more
generally what I'm saying is that, no matter what the triumphs of cog-
nitive science, ultimately sentience goes unexplained. (As I said, this is
more contentious. There are philosophers today who think that already
cognitive science can explain sentience [Churchland, 1995; Dennett,
1992]. However, when you look at the arguments, there is inevitably
a gap between the premises about the functioning of the brain or the
computer and the conclusion about sentience. Along with many philo-
sophical critics, I argue that there is a gap between the functioning of
the material machine and consciousness.)
The final question is about ends. We have seen that, under the
machine metaphor as used in today's science, questions about ends are
ruled out. Science focuses only on the working of the machine and not
upon the reasons why the machine exists in the first place. It will be
remembered that God is a retired engineer. Hence, what I suggest is that
modern science simply does not set out to answer questions about pur-
poses. The Nobel prize-winning physicist Steven Weinberg (1992) has
remarked that the more he does science the less he sees any purpose to
the world. Not only am I not surprised, I would be very surprised if he
did find purpose. The way we think about the world today using the
machine metaphor does not allow for answers of this kind. Hence it is
pointless to ask questions that demand such answers.
262 Michael Ruse

2.1. Answers?
We have four questions that science not only does not answer but that
science does not even set out to answer. Why is there something rather
than nothing? What is the ultimate basis of morality? What is sentience
and its cause? What is the ultimate purpose of everything? I think it is
possible to be a sceptic or agnostic on all of these questions. It's per-
fectly reasonable simply to say that one has no answers and cannot see
any way of getting answers. Overall, this is my position. It is perfectly
meaningful to ask why there is something rather than nothing. It is
just that I don't have an answer! As far as morality is concerned, you
can explain it away as an adaptation. I myself am inclined to say that
this is enough and there is no need to go looking for foundations. But
if you insist on foundations, then science is silent and so am I. Again,
I am with those who not only do not have an answer to sentience but
who are starting to think that perhaps there never will be an adequate
answer (McGinn, 2000). Perhaps it is part of the way that our brains
have evolved that this is a question that can never be resolved. This may
seem like a bit of a cop out, but already in physics we are told that there
are questions which we cannot answer. I am thinking about whether
an electron is really a particle or really a wave. Heisenberg's Uncertainty
Principle simply says that questions like these are off limits. There are
no answers. Perhaps in cognitive science there is the equivalent to this
kind of inexplicability. We know all about sentience except the most
important questions, namely what it is and how it comes about. And
certainly I have no answers to the ultimate purpose of reality. I don't
see any purpose, so at this level I am with Steven Weinberg. Whether
there is some ultimate purpose, however, seems to me simply beyond
the grasp of science and once again beyond my grasp entirely also.
However, what I would argue is that my inability to offer answers
does not preclude others trying to offer answers, so long as they do not
offer scientific or quasi-scientific answers. You have to offer answers of
a different kind. And here it seems to me is a place for the religious
thinker. If the believer wants to offer answers, then I think it is perfectly
legitimate for the believer to do so. I am not saying that the believer is
beyond criticism, but I do not think that the believer can be criticized
on the grounds of science. I am certainly saying that, if a believer insists
that the Earth is 6000 years old, because this is what he or she learns in
the Bible, then it is perfectly legitimate for the scientist to criticize such a
claim. The believer is making a claim in the realm of science and as such
is open to critique. I'm also saying that if the believer makes a claim that
one wants to criticize on grounds other than science, it is legitimate to
 Why the Warfare Thesis Is False 263

make such a criticism. If, for instance, the believer makes a claim about
a deity which seems internally contradictory or whatever it is legitimate
to criticize. If, for instance, one argues that God is a necessary being,
I think it legitimate for a philosopher to go after this claim, arguing that
existence can never be necessary. (I am not now saying that the philoso-
pher will be successful.) The point is that one is not now criticizing on
the basis of science. And this is my position. Within these parameters -
no trying to slip in science and recognize that you can (potentially) be
faulted on other grounds - I would argue that the believer has every
right to offer his or her solutions to the questions listed above.
Certainly it is the case that believers do offer answers to these ques-
tions. Staying- as restricted above- with Christianity, we find that there
is a clear answer to the first question about why anything at all exists.
The world is the creation of a god who did so as an act of love. God
created out of nothing and He did so simply because He wanted to do
so, and He wanted to do so because He saw that it would be better if He
did so than if He did nothing. This is obviously not a scientific answer.
It is a religious answer. I suggest - putting aside whether or not it is an
adequate answer - in the light of the discussion above, it is certainly a
legitimate answer.
What is the foundation of morality? Again the answer comes through
loud and clear. It is that which God wants us to do. Morality is fitting
in with the will of God. Notice that Christians realize that one has got
to go a little beyond this (Quinn, 1978). One does not want to say sim-
ply that morality is God's will and that God might have done anything.
Clearly, we do not want to say that God might have made killing a good
thing. This sophisticated Christian response to this kind of worry is to
point out that morality is bound up with the creation. God has made
the world in a certain way and we are expected to behave in accordance
with that way. In other words, we are expected to do that which is nat-
ural. So, for instance, wanton killing could never be right because it is
unnatural. It is destroying God's creation without good reason. Moral-
ity therefore finds its foundation in God's will, but understanding God's
will demands that we take due note of God's creation.
What is sentience? Again, note that we have got to be careful not
to offer a scientific answer or an answer that is pretending to be reli-
gious but really is scientific. The Christian answer is that sentience is
that which makes us 'in the image of God'. In some sense sentience or
consciousness is that which puts us in tune with our creator. It is that
which opens up the possibility of knowledge as well as of moral choice.
Without these abilities we could not perform the role in the drama that
264 Michael Ruse

is taking place here on Earth. Sentience does not make us divine, but it
opens up the future that we can share uniquely with the divine.
What is the purpose of it all? Ultimately, obviously, it is to share eter-
nity with our creator. Exactly what this means is a matter of considerable
debate among believers. But it is certainly intended to be a factual claim
in some sense and not just a pious hope. The Earth was created to be the
home of humans (and perhaps other organisms too), but in a sense this
is only a temporary resting place. The Earth is a place of trial, in some
sense, and depending on our performance here our future is determined
accordingly. So the Earth does not have value just in itself but as a place
of preparation for what is to come.
I want to stress again that I am not suggesting here that anybody has
to accept any of these answers- or parallel answers that might be offered
by members of other faiths. What I am arguing is that, given its inherent
nature, modern science does not even attempt to answer the questions
being asked. Therefore, it is legitimate for others to make the attempt.

3. Problems?
I have also stressed that I do not think anything said thus far makes reli-
gious answers immune to criticism. I wanted therefore now to change
track somewhat and to look at three criticisms that have been made in
the name of science, suggesting that the religious answers are not well
taken. In other words, I want to look at criticisms that suggest even
though it may be possible for the religious person (the Christian specifi-
cally) to offer a legitimate response in the light of science, the particular
responses that the religious person wants to make are inadequate.
First let us raise the venerable problem of evil. Many Darwinians
think that Darwin's theory of evolution poses an altogether-much-more-
severe threat on this front than has been encountered hitherto. The
traditional Christian approach to the problem divides the problem into
two (Ruse, 2001). On the one hand, we have moral evil- human-caused,
like Auschwitz- and on the other hand we have natural or physical evil-
earthquakes, like the one that destroyed Lisbon. In both cases, Darwin-
ism poses a threat. Moral evil is supposedly something introduced by
the sin of Adam - God created us perfect - and Darwinism negates an
original pair of humans. Physical evil is simply the result of laws and it is
argued that the laws are a package deal, sometimes causing good things
and sometimes bad things, but overall good. (Pain is uncomfortable but
better pain than accidents where we are unaware that our bodies are in
danger.) But natural selection, as we have seen, depends on the struggle
 Why the Warfare Thesis Is False 265

for existence. The struggle for existence is frequently very painful and
so it seems almost capricious of God to have created using this mech-
anism. Would an all-loving, all-powerful God have chosen selection as
his means of creation? Darwin himself worried about this issue. And
Darwinians down to the present, especially including Richard Dawkins
(1995), have made much of this point.
My aim is not to defend Christianity against the problem of evil.
I am far from convinced myself that this is possible. Here, I am inter-
ested rather in whether or not the Darwinian critics are correct and
whether science - Darwinian evolutionary theory in particular - poses
an altogether-new dimension to the problem. I'm not sure that it does.
Start with moral evil. Following St Augustine, it is argued that thanks
to Adam we are tainted in some way and hence have an inclination
to do bad things. Obviously, the evolutionist cannot accept Adam and
Eve as literal, historical figures. However, the evolutionist does have an
explanation of moral evil, namely that Darwinian selection is going to
produce humans that are somewhat ambivalent mixes of good and ill.
On the one hand, we are social beings and as such therefore have incli-
nations to work harmoniously with others. On the other hand we have
to look after ourselves - in Dawkins's colourful metaphor, we have to be
'selfish genes'- otherwise we simply would not succeed in the struggle
for existence. So what is argued by the Darwinian is that, rather than
evil being the result of one person's bad act at some point in history, it
is the natural outcome of a long, slow process of development.
Is this the end of the Christian story? It is certainly the end of the
Augustinian story. But there are today Christian theologians who sug-
gest that this Darwinian account is a more adequate theological answer
to the problem of evil (Schneider, 2010). Rather than making a whole
Christian story a result of one weak man's act of disobedience, and
then the whole drama of Christ coming and dying on the cross for
our sins being, as it were, a matter of playing catch up - what is often
referred to semi-humorously as 'Plan B' - we see that, in a sense, God
intended everything from the start. This is a tradition which goes back
before Augustine to lrenaeus of Lyons, and many think it a more com-
fortable and reasonable interpretation of what it is to say that God
had become to save us from our sinful nature. Because of our his-
tory we are but partial figures and the Incarnation and the Atonement
were part of finishing the creation, not cleaning up a mess after the
creation.
As far as physical evil is concerned, interestingly it may well be
Richard Dawkins (1983) himself who comes to the Christian's rescue at
266 Michael Ruse

this point. He argues that the only way in which one can naturally get
design or final cause- organic adaptation- is through natural selection.
All other mechanisms are either false (for instance so-called Lamarckism,
the inheritance of acquired characteristics) or inadequate (for instance
saltationism, the idea of evolution through jumps). But if natural selec-
tion is the only way in which orgar:isms could have been produced, and
this includes human beings, then inasmuch as God decided to create
through law natural evil had to follow. Remember it is never the claim
that God can do the logically impossible. Only that he will always do
that which is best. Of course, there is also the matter of whether or not
God had to create through law. However, given the kind of beings that
we are, had God not created through law then he would surely have
been deceiving us - every part of our physical nature attests to our hav-
ing evolved rather than having been created miraculously. So, it seems
that there are strong reasons for thinking that natural evil or pain is not
necessarily something that can be laid down as God's responsibility.
I turn next to a worry that many have about freedom. If one accepts
the scientific perspective then one is committed to the belief that the
world is governed according to universal law. The metaphor is that
the world is like a gigantic clock, simply going through the motions.
The clock obviously has no freedom of its own. If the hands point to
2 o'clock right now, and then in one hour's time the hands will point to
3 o'clock. The clock cannot of its own intent decide to speed things up
and make the hands point to 4 o'clock or slow things down and make
the hands point to 2.30. It is, however, an essential part of the Chris-
tian religion that humans are free agents. We have the power to choose
between good and ill, even though regrettably all too often we choose
the ill. God could presumably have made us all robots, doing nothing
but good. However, in His wisdom, He saw that it is better that we be
free even if sinful, than predetermined and unable to decide for our-
selves. There seems therefore to be a major clash here between science
and religion.
Moreover, there are those who think that particular sciences make
matters even worse. In recent years, it has been claimed by crit-
ics that the attempt to apply evolutionary theory to human nature
leads to a particularly pernicious form of determinism, often labelled
'genetic determinism' (Lewontin, 1991). It is argued that the human
evolutionist - sometimes known as the 'human sociobiologist',
although these days more likely as an 'evolutionary psychologist' - is
committed to the view that everything we do and think is predeter-
mined by the genes, as chosen by natural selection. Hence, for instance,
 Why the Warfare Thesis Is False 267

it is argued that the human evolutionist thinks that males are aggressors
without any choice in the matter, simply because of their genes; whereas
women are sexually coy, again simply without any choice.
That there is an issue here that needs resolving is beyond argument.
However, there are fairly standard ways of answering the general com-
plaint. In particular, many (quite independently of science) would argue
that it is misleading to juxtapose the rule of law with free will. In fact
there are those, notably the philosopher David Hume, who argue that
only if we have the rule of law is freedom possible. Without law, then
everything is purely random and this is not freedom. If for instance
I commit a murder, but it is a purely random event, then I am not
morally culpable. It is only when I'm seen against the background of my
training and human nature and much more that one can properly pre-
scribe praise and blame. This position, often known as compatibilism,
argues that the real distinction must be drawn between being deter-
mined and being confined or in some way oppressed. A person in chains
or hypnotized is not free. The person not so constrained, even though
bound by law, is nevertheless free. So the general charge that science
makes free will impossible is seen to fall to the ground.
What about the specific charge that human evolutionary biology
introduces a new dimension of determinism, one which does indeed
make free will impossible? Here, it is worth looking in a little more detail
at the claims of the human evolutionist. It is indeed agreed that humans,
like all other organisms, are the products of random variation brought
on by natural selection. Hence, the genes that we have do play a vital
causal role in our physical being and in our nature - both our thinking
and our behaviours. But do note that human evolutionary biology today
is a little more subtle than critics often appreciate. Most importantly, it
is recognized that higher organisms like humans have rather different
behavioural strategies than lower organisms like social insects. Ants, for
instance, do everything by instinct. They do not require any teaching or
prior experience. Hence, ants will follow a pheromone trail quite blindly
to find food and then return to the nest fully laden. Given the success
of ants in the world, this is obviously a very good adaptation because it
leads to high reproductive success. However, working purely by instinct
has its dangers and snares. In particular, if something goes wrong then
it is very difficult to regroup and rethink and try a new strategy. Sup-
pose an ant is out foraging and there is a tropical rainfall wiping away
all of the chemical trails. It may well be that as a consequence literally
hundreds of ants get lost and die, failing to return to the nest. The rea-
son why the ant family can afford this kind of cost is that the queen
268 Michael Ruse

produces literally thousands of offspring, and the loss of even a large

number is far from irreparable.
Humans have taken a somewhat different reproductive strategy.
We only raise a few offspring and invest a great deal of parental care in
them. This means that we cannot afford to lose our offspring as soon as
some environmental event threatens our wellbeing. If every time there
was a heavy rainfall we lost one or more of our offspring, we would very
rapidly go extinct. Hence, humans have to have the ability to regroup
and rethink when obstacles or adverse events come into our part. This
we do, showing that at some level we have a degree of freedom not pos-
sessed by lower organisms like ants. We are still completely law-bound
but we have something over and above just brute determinism. The
philosopher Daniel Dennett (1984), for all that he is violently opposed
to religious belief, has offered a good illustration of how humans work
and a sense in which we can be said to have free will. He likens us to the
Mars Rover, which when faced with rocks and other obstacles on the red
planet does not give up but reassesses and moves around the obstacles.
Everything is entirely law-bound but there is a dimension of freedom
that the more simple machine would not have. Hence we can see that
far from modern science standing !n the way of Christian claims about
free will, if anything it helps us to understand how such free will comes
about.

4. Are Humans Necessary?

I want now to turn to a third problem, one which I believe is more
challenging than the two just discussed. This is about the necessity of
human existence. I do not think it is necessarily the case th.tt the Chris-
tian must believe that we humans are uniquely the focus of God. If one
takes seriously the comment in the Gospels about God knowing when
every sparrow falls, it would seem that God has an intensive concern
for all of His creation. But it is surely the case that humans do have a
special place in God's heart, and that it was absolutely necessary that
we humans appear. It might be impossible that we would not have been
exactly as we are today. Perhaps we might have had blue skin, or maybe
had twelve fingers. Possibly, although I'm not sure about this, it might
have been possible for us not to have sexuality as we know it today.
However, it is surely the case that the beings that were intelligent and
with a moral dimension had to evolve. If this had not happened, then
Christianity would be untrue.
Unfortunately, Darwinian evolutionary theory implies strongly that
there is no necessity in the overall process. Stephen jay Gould (1989)
 Why the Warfare Thesis Is False 269

used to say, if you play the tape of life several times, you will get different
end results each time. Because natural selection is a relativistic process -
this does not mean that it is tautological, but it does mean that what
survives in one situation might not be what survives another situation -
there is no absolute inevitability about what will come out at the end.
And if this were not enough, it is always stressed by Darwinians that
the basic building blocks of evolution - the mutations - are random,
not in the sense of being uncaused but in the sense of not appearing
according to the needs of their possessors. All of this seems to speak
strongly against any claim is that human beings had to appear on the
scene. Hence, it does seem finally that we have come to a point where
modern science strikes strongly at the central claims of the Christian
religion.
Those concerned to reconcile science and religion have not been
unaware of this problem. One solution is simply to take the whole
problem out of the realm of science. This is the position taken by the
physicist-theologian Robert j. Russell (2008), who argues that perhaps
God puts in direction down at the quantum level. Evolution is guided
and humans will inevitably appear, even though we can never expect
to see exactly how this works. However, my suspicion is that most feel
rather uncomfortable with a solution like this. It is too close to a 'God-
of-the-gaps' type solution, where when you don't know how to solve
something you bring it in a miracle. It is a modern-day version of the
kind of position endorsed by Darwin's great American supporter Asa
Gray (1876), a position that Darwin totally repudiated. If God wants to
create through unbroken law, then He should let the law get on with
the job without constantly interfering in the process.
There are a number of attempts to solve the problem using just mod-
ern biology. Richard Dawkins (1986) argues that evolution frequently
involves what are known as 'arms races'. Lines of organisms compete
against each other and eventually better adaptations emerge. If one
keeps this up long enough one can expect that large brains will emerge
and so humans will have to appear. Interestingly, Darwin himself rather
wrestled with this problem - although I do not think he was doing it in
order to solve theological issues any more than that Dawkins - and he
came up with a similar solution.

If we look at the differentiation and specialisation of the several

organs of each being when adult (and this will include the advance-
ment of the brain for intellectual purposes) as the best standard of
highness of organisation, natural selection clearly leads towards high-
ness; for all physiologists admit that the specialisation of organs,
270 Michael Ruse

inasmuch as they perform in this state their functions better, is

an advantage to each being; and hence the accumulation of varia-
tions tending towards specialisation is within the scope of natural
selection. (Darwin, 1959, p. 134)

While this is an ingenious solution, it surely does not guarantee that

humans must appear. Suppose that the dinosaurs had not gone extinct
65 million years ago. Then it might well be that the continued exis-
tence of the dinosaurs would bar the rise of the mammals, and humans
would still not have emerged- nor would there be any prospect of them
ever emerging. So while it is the case that arms races may make intelli-
gence somewhat more probable, it certainly does not offer the iron-clad
guarantee that the Christian demands.
Recently Simon Conway Morris, a distinguished palaeontologist who
is also a practising Christian, has been suggesting that the key item in
solving this problem is the existence of ecological niches. He argues that
certain liveable niches exist, independently of organisms. What happens
is that organisms, powered by natural selection, push and strain until
they find and enter into these niches. Thus, for instance, we find that
there is a sabretooth tiger niche, which was entered independently by
the placental mammals and by the marsupial mammals. Conway Morris
suggests that there is obviously a cultural niche, one that we humans
have occupied, and that therefore even had we not occupied it we might
expect at some point that some organisms would so do.

If brains can get big independently and provide a neural machine

capable of handling a highly complex environment, then perhaps
there are other parallels, other convergences that drive some groups
towards complexity. Could the story of sensory perception be one
clue that, given time, evolution will inevitably lead not only to the
emergence of such properties as intelligence, but also to other com-
plexities, such as, say, agriculture and culture, that we tend to regard
as the prerogative of the human? We may be unique, but paradoxi-
cally those properties that define our uniqueness can still be inherent
in the evolutionary process. In other words, if we humans had not
evolved then something more-or-less identical would have emerged
sooner or later. (Conway Morris, 2003, p. 196)

Again, one can see problems. First, many evolutionists are not
convinced that niches do exist independently of organisms. These
evolutionists argue that, to a great extent, organisms make niches
 Why the Warfare Thesis Is False 271

themselves. Hence, although we humans may live in a cultural niche,

it is wrong to think that the cultural niche exists independently of our
particular course of evolution. And in any case, the dinosaur problem
still stands in the way. Had the dinosaurs not gone extinct, even if a
cultural niche did exist, this surely is no hard guarantee that humans or
any other mammals or organisms would have found their way into it.
Again, therefore, we do not have enough for the Christian.
A third solution is to drop Darwinism almost entirely. This is the tack
taken in a recent book by palaeontologist Daniel McShea and philoso-
pher Robert Brandon. They introduce what they call the 'zero-force
evolutionary law' or ZFEL: 'In any evolutionary system in which there is
variation and heredity, in the absence of natural selection, other forces,
and constraints acting on diversity or complexity, diversity and com-
plexity will increase on average' (McShea and Brandon, 2010, p. 3).
Obviously, if their law is well taken, then if there is enough time one
would expect very complex organisms to evolve. The trouble is that
there seems no guarantee that the complexity would be in any sense
adaptive. More particularly, that it would be adaptive in the sense of
producing large, functioning brains. McShea and Brandon quite openly
acknowledge that they are following not so much in the footsteps
of Charles Darwin, but in the footsteps of his contemporary Herbert
Spencer. And this has always been the problem with Spencerian-type
thinking. You may indeed get an end result, but whether it is the end
result you want is another matter. Certainly, without more added, and
the more almost inevitably points one in the direction of natural selec-
tion, one is going to end up with something which is not necessarily
human-like. Again the Christian will remain dissatisfied.
My own feeling is that all of these solutions go about the problem in
the wrong way. The demand about the inevitable appearance of humans
is not so much a scientific demand but a religious demand. It is not that
the Christian is saying what must happen on scientific grounds, but
rather what is needed on theological grounds. For Christianity to work,
humans must have appeared. I suggest therefore that in looking for a
solution we should start with Christianity itself. Given a universe, there
was no guarantee that on any particular playing of the Gouldian tape
that humankind would occur. We could have many, many universes
where either life did not get going at all or if it did get going did not lead
to humans. However, because humans did appear by Darwinian pro-
cesses, we know that they could have appeared by Darwinian processes.
They needed a world in which this could happen. Our world, obviously.
At some point in an infinite string of universes, either in time or space,
272 Michael Ruse

it would work. Now, it is a standard part of Christian theology that God

is omnipotent. So, presumably God could have created all of those uni-
verses and at some point He would get what he wanted. Would not this
be an awful drag on God's patience? Not at all! It is also part of Christian
theology that God stands outside time and space. God is not waiting
endlessly for things to happen. As Saint Augustine argued, for God the
thought of creation, the act of creation, and the product of creation are
as one. So there is no big problem there. Humans appear, and no special
forces or explanations are needed above and beyond pure Darwinism.
Although today there are many physicists inclined this way, I am not
arguing for this 'multiverse' position on scientific grounds, but purely
on theological grounds. You might think that this is awfully wasteful.
Why did God create only the universe that he knew would be the one
where humans evolved? Here, I think we need to go back to quantum
mechanics, and to the claim that there is no predictability about how
individual events will occur. If we have God choosing only one universe
out of many, then we are back to Robert J. Russell's position - God is
guiding events. I want to argue that God simply did not get involved at
this level, and hence he did not know or determine which way things
would occur at the quantum level. This does rather imply that God at
some level is limited, not knowing what will emerge. But this is no more
a limitation on God's powers than that he cannot make 2 + 2 = S. But
what about the waste? God is creating literally millions of universes
which come to nothing. But, God does this already, because our uni-
verse is absolutely vast with what seem to be countless lifeless galaxies.
Why did God not just confine his creative abilities to our solar system?
In any case, who are we to presume that we know God's intentions and
desires? Who is to say that God did not delight in the very act of cre-
ation and that the universe without human beings would be unpleasing
or irrelevant to God's purposes? The Christian claim is that humans have
a special lien on God's affection, not that we are exclusively the objects
of God's concern and interest (Whewell, 2001).

5. Conclusion
My general position is that, because science is metaphorical, it is lim-
ited in the sense that there are certain questions that it does not even
ask let alone answer. I argue also that it is therefore legitimate for oth-
ers to attempt to offer answers, so long as they are not scientific. Here
is a place where religion might function successfully. I am not argu-
ing that one must be religious, and indeed suggest that one could be
 Why the Warfare Thesis Is False 273

entirely sceptical or agnostic. It is simply that if Christianity wants to

try to answer the unanswered questions of science, then it is free to do
so. I suggest also that some of the most obvious objections to this enter-
prise are really not well taken. Hence, overall I want to argue for the
compatibility of science and religion, and suggest that one can legiti-
mately be an accommodationist. More strongly, I would argue that the
reasonable person should be an accommodationist.

References
Barbour, I. (1990) Religion in an Age of Science (New York: Harper and Row).
Boyle, R. (1996) A Free Enquiry into the Vulgarly Received Notion of Nature, ed. E. B.
Davis and M. Hunter (Cambridge: Cambridge University Press).
Chalmers, D. J. (1996) The Conscious Mind (New York: Oxford University Press).
Churchland, P. M. (1995) The Engine of Reason, The Seat of the Soul (Cambridge,
MA: MIT Press).
Conway Morris, S. (2003) Life's Solution: Inevitable Humans in a Lonely Universe
(Cambridge: Cambridge University Press).
Darwin, C. (1959) The Origin of Species by Charles Darwin: A Variorum Text, ed.
M. Peckham (Philadelphia, PA: University of Pennsylvania Press).
Dawkins, R. (1976) The Selfish Gene (Oxford: Oxford University Press).
Dawkins, R. (1983) 'Universal Darwinism', in D. S. Bendall (ed.), Evolution from
Molecules to Men (Cambridge: Cambridge University Press), pp. 403-25.
Dawkins, R. (1986) The Blind Watchmaker (New York: Norton).
Dawkins, R. (1995) A River Out of Eden (New York: Basic Books).
Dawkins, R. (2007) The God Delusion (New York: Houghton, Mifflin, Harcourt).
Dennett, D. C. (1984) Elbow Room: The Varieties of Free Will Worth Wanting
(Cambridge, MA: MIT Press).
Dennett, D. C. (1992) Consciousness Explained (New York: Pantheon).
Dennett, D. C. (2006) Breaking the Spell: Religion as a Natural Phenomenon
(New York: Viking).
Desmond, A. (1994) Huxley, the Devil's Disciple (London: Michael Joseph).
Desmond, A. (1997) Huxley, Evolution's Hig/1 Priest (London: Michael Joseph).
Dijksterhuis, E. J. (1961) The Mechanization of the World Picture (Oxford: Oxford
University Press).
Draper, J. W. (1875) History of the Conflict between Religion and Science (New York:
Appleton).
Farlow, J. 0., C. V. Thompson and D. E. Rosner (1976) 'Plates of the Dinosaur
Stegosaurus: Forced Convection Heat Loss Fins?', Science 192: 1123-25.
Garber, D. (1992) Descartes' Metaphysical Physics (Chicago: University of Chicago
Press).
Gould, S. J. (1989) Wonderful Life: The Burgess Shale and the Nature of History
(New York: W.W. Norton Co.).
Gray, A. (1876) Darwiniana (New York: D. Appleton).
Hall, A. R. (1983) The Revolution in Science, 1500-1750 (London: Longman).
Harris, S. (2004) The End of Faith: Religion, Terror, and the Future of Reason
(New York: Free Press).
274 Michael Ruse

Hitchens, C. (2007) God Is Not Great: How Religion Poisons Everythit1g (New York:
Hachette).
Hume, D. (1739) A Treatise of Human Nature (Oxford: Oxford University Press,
1940).
Kuhn, T. (1962) The Structure of Scientific Revolutions (Chicago: University of
Chicago Press).
Kuhn, T. (1993) 'Metaphor in Science', in Andrew Ortony (ed.), Metaphor and
Thought, 2nd edn (Cambridge: Cambridge University Press), pp. 533-42.
Leibniz, G. F. W. (1714) Monadology and Other Philosophical Essays (New York:
Bobbs-Merrill).
Lewontin, R. C. (1991) Biology as Ideology: The Doctrine of DNA (Toronto: Anansi).
McGinn, C. (2000) The Mysterious Flame: Conscious Minds In A Material World
(New York: Basic Books).
McShea, D., and R. Brandon (2010) Biology's First Law: The Tendency for Diver-
sity and Complexity to Increase in Evolutionary Systems (Chicago: University of
Chicago Press).
Merchant, C. (1980) The Death of Nature: Women, Ecology, and the Scientific
Revolution: A Feminist Reappraisal of the Scientific Revolution (Scranton, PA:
HarperCollins).
Quinn, P. L. (1978) Divine Commands and Moral Requirements (Oxford: Clarendon
Press).
Ruse, M. (1975) 'Charles Darwin and Artificial Selection', Journal of the History of
Ideas 36: 339-50.
Ruse, M. (1979) The Darwinian Revolution: Science Red in Tooth and Claw, 2nd edn
(Chicago: University of Chicago Press, 1999).
Ruse, M. (2001) Can a Darwinian Be a Christian? The Relationship between Science
and Religion (Cambridge: Cambridge University Press).
Ruse, M. (2005a) 'The Darwinian Revolution as Seen in 1979 and as Seen Twenty-
Five Years Later in 2004', Journal oftlte History of Biology 38: 3-17.
Ruse, M. (2005b) The Evolution-Creation Struggle (Cambridge, MA: Harvard Uni-
versity Press).
Ruse, M. (2008) Charles Darwin (Oxford: Blackwell).
Ruse, M. (2010) Science and Spirituality: Making Room for Faith in the Age of Science
(Cambridge: Cambridge University Press).
Russell, R. ]. (2008) Cosmology: From Alpha to Omega, the Creative Mutual Interaction
of Theology and Science (Minneapolis: Fortress Press).
Schneider,]. (2010) 'Recent Genetic Science and Christian Theology on Human
Origins: An "Aesthetic Supralapsarianism" ',Perspectives on Science and Christian
Faith 62: 196-212.
Weinberg, S. (1992) Dreams of a Final Theory: The Search for the Fundamental Laws
of Nature (New York: Pantheon).
Whewell, W. (2001) Of the Plurality of Worlds. A Facsimile of the First Edition of
1853: Plus Previously Unpublished Material Excised by the Author Just before the
Book Went to Press; and Whewell's Dialogue Rebutting His Critics, Reprinted from
the Second Edition (Chicago: University of Chicago Press).
White, A. D. (1896) History of the Warfare of Science with Theology in Christendom
(New York: Appleton).
Index

agency Bible, 195-8, 262

attributions of, 71-94, 173-4 big bang, 4, 96-7, 100-32
agnosticism, 185, 186, 188, 233 biological complexity, 3, 51, 64, 270
Agurrie, Anthony Bishop, John
arrow of time, 118 modest fideism, 231
eternal steady state inflation model, black crunch, 117, 136
112-24, 127 black hole, 100, 105, 106, 111, 116,
Almeida, Mike 120-1, 136, 137, 143
multiverse, 149, 154-5 Blackburn, Simon
altruism, 62 truth condition approach to
Ambjorn, Jan modality, 14-18
loopholes in Euclidean approach, Boethius, 36-42
130-1 Bojowald, Martin
Anselm, 11, 27 loop quantum gravity, 113-18, 123,
Aquinas 125, 127
foreknowledge, 33, 36, 41 Borde, Arvind
heretics, 207-9, 221 singularity theorem, 108-9, 122,
Plantinga, Alvin, 228 123
argument from evil; see problem of Boyer, Pascal, 30-1, 34, 175
evil Boyle, Robert, 258
Aristotle Brandon, Robert
methodology of, 33-5 zero-force evolutionary law, 271
astrophysical cosmogony, 96 brute facts, 39
atemporalist position, 104, 107 Bulbulia, Joseph
atheism value of theology, 30-1, 42
probability of, 171 Burger, Jerry
persistence of, 229-30 burger experiments, 247-8
atheodicy, 50, 58 Butler, Joseph, 61
atomic theory, 54
Augustine Calvin, John, 208-9, 228
Christian theology, 36, 254, 265 causation, 16, 55, 144, 178
creation, 137, 272 backwards, 17-18
autonomy, 207, 243 Hume, David, 16
Chalmers, David
Bach-Weyl theory of gravity, 101, 109, hard problem of consciousness, 261
110 character, 241-2, 247-8, 249, 252
Banks, Thomas cognitive
contracting space, 116-17, 121, 136 process, 3, 71
Barrett, Justin scientists, 2, 5, 28, 29, 31, 42, 43
prevalence of religion, 28, 32, 42, science of religion, 5, 28, 31, 32, 34,
176 43, 86, 89, 165-80
belief polarization, 6, 225-32 compatibilism, 267

275
276 Index

compatibility of science and religion, doctrine of creation, 99, 102, 133, 263
6-7,237-75 Donaldson, Margaret
conscious modes of consciousness, 37-41
mind, 75-6 Doris, John
will, 239-42, 252-3 lack of character, 239-53
consciousness, 14, 41-3, 75, 76, 78, Draper, Paul
79,105,251,261,263 duplicate universes, 144-5
modes of, 36-41 God as benefactor, 148
consequentialism, 207-9, 246, 248 restrictions on universe, 152, 157
Contigentism, 147-8, 158 science and religion, conflict
cosmological argument, 11, 53 of, 255
Kalam, 4, 95-6, 128, 129-33 dual process theory of mind, 75, 89
cosmology, 2, 4, 96-128, 136, 145, 158
quantum, 125-6, 129, 134 Ellis, George
cost-benefit analysis, 213-15, 217-19, contracting universe, 111-14
221 entrap~ 115-21, 125-7, 128, 136, 137
creation epiphenomenalism, 241
creation ex nihilo, 125, 126, 128, epistemological, 12, 15, 232
129, 135, 137, 144 naturalism, lOS
doctrine of, 99, 102, 133, 263 Everett, Hugh, 126, 143
creationism, 3, 68 Quantum mechanics, many worlds
cyclic ekpyrotic model, 96, 135 interpretation of, 143
evolution
dark energy, 118-19 natural selection, 3, 49, 51-68, 177,
Darwin, Charles, 3, 49-68, 69, 103, 257,264,266,267,269-71
255,257,265,269-70,271 theory of, 49-71, 79, 103, 71-5, 78,
Darwinian argument from evil, 85-8,255,257,265,266,269,
49-68 270,271
Darwinian, 63-7 of space-time, 130
Dawkins, Richard, 256, 258, 265, 269 of consciousness, 79
demographics, 184 evolutionary
explanations of, 191-2, 198 analysis, 74
of theism, 5, 184-98 biology, 3, 49-50, 267
Dennett, Daniel, 4, 31, 34, 42, 71-3, psychology, 3, 71, 74, 88, 175
242,256,261,268 exclusivism
intentional stance, 4, 71-3, 81-6, 89 salvific, 6, 205-10, 217-20
Descartes, Rene, 67, 258 interventionist salvific, 210-20, 221
determinism, 191-2, 266-9 experience-enhancing theology, 3,
deus deceptor argument, 177-80 36-43
divine attributes experimental philosophy, 185
omnlbenevolence, 2, 53, 56, 58,
153, 178 Fall, the, 230
omnipotence, 2, 49, 53, 55, 56, 57, Feynman, Richard
58, 64, 65, 172, 272 cosmology, 109, 134
omniscience, 53, 55, 56, 57, 58, 64, fine-tuning, 60, 111-15, 122, 143,
82 151, 173
timelessness, 2, 22, 28-43, 95, 144 Fischler, Willy
necessary existence, 2, 11-28 contracting space, 116-17
divine hiddenness, 178, 191, 229 Fodor, Jerry, 75-7
 Index 277

folk Hume, David, 15-16, 26, 261, 267

physics, 77-82, 90 causation, 16
psychology, 77, 80, 81 Huxley, Thomas Henry, 255
foreknowledge, 33, 35, 40
formation of character, 237-9 incompatibility of science and
free will, 68, 193, 197, 199, 200, 230, religion, 256
237,239,242,267-8 independence position, 7, 256
infinite contraction, 111
Galileo, 255 instrumentalism, 72-3, 85
general relativity, 96, 100, 104-9, intelligent designer, 3, 143
112, 113, 115, 119, 121, 129-31, inverted god-of-the gaps objection,
134, 136 168, 169-70
genetic fallacy, 167-8 Irenaeus, 265
geodesics, 108-9, 123, 128
Geroch, Robert James, William
singularity, 100 pragmatism, 72, 73
God origin of belief, 181
attributes; see divine attributes Jedi, 187-8
being, necessary, 11-28, 263 Johnson, Samuel, 39
concept of, 1, 12, 21, 177 Jurkiewicz, Jerzy
goodness, source of, 21, 23 loopholes in Euclidean approach,
of-the-gaps, 97, 168, 169-70 130-1
Gould, Stephen Jay, 268
Gratton, Steve Kalam cosmological argument, 4,
arrow of time, 118 95-6, 128, 129-33
eternal steady state inflation model, Kane, Robert, 240
112-24, 127 Kenny, Anthony
Gray, Asa foreknowledge, 33-5
Darwin, defender of, 53, Kuhn, Thomas
55,269 science and metaphors, 257,
Gregersen, Neils Henrik, 175 259
Guth, Alan
singularity theorem, 108-9, 122, Laplace's demon, 82, 84, 89
123 Leibniz, Gottfried
best possible world, 146
Hawking, Stephen, 109, 127, 129, 130, conscious machines, 261
131, 132 Lewis, C. S., 243
Hawking-Bekenstein entropy law, Lewis, David, 145, 158, 207, 209
120 libertarians, 192, 242
Hawking-Penrose singularity Libet, Benjamin, 240
theorem, 107 Linde, A. D.
Heisenberg uncertainty principle, 132, eternal inflation view, 143
262 Loll, Renate
Helm, Paul, 32 loopholes in Euclidean approach,
heretics, 207-8 130-1
heuristic concept, 73, 77, 79, 80, 81,
83 machine metaphor, 257-61, 265-6
Hudson, H., 144, 145, 148, 154, 159 Maitzen, Stephen
plentitudinous hyperspace, 153, 160 argument from evil, 190-4
278 Index

Manson, Neil natural theology, 52-3, 97, 99, 231

precautionary principle, criticism of, naturalism, 54, 184
216,219 naturalistic explanations of religion,
Marenbon, John 165-80
Augustine, 36 necessitarianism, 146-8
McGinn, Colin necessity
mysterianism, 231-2 de dicta/de re, 12-14
McHarry, ]. D. projectivist account of, 12,
identity of indiscernibles, 145 15-19
unsurpassable universe, 147 metaphysical, 15, 17
dissolving the problem of evil, 152 divine, 11-15, 21-6
McShea, Daniel logical, 18, 26
zero-force evolutionary law, 271 of humans, 268-72
Mele, Alfred no logical space arguments, 166
free will, 237, 240 no-design hypothesis, 57-9, 63-6,
Milgram experiments, 244-8 68, 69
Mithen, Steven
religiosity as feature of society, O'Connor, Timothy
166-7 objective threshold, 145, 147-9
monotheism, 71, 87, 184 argument from evil, 154-7
Manton, Bradley Ockham's razor, ISO
big bang, 96 omnibenevolence, 2, 53, 56, 58, 153,
duplicate universes, 145 178
multiverse, 148-50, 155, 156 omnipotence, 2, 49, 53, 55, 56, 57, 58,
problem of omnibenevolent being, 64, 65, 172, 272
153-4 omniscience, 53, 55, 56, 57, 58, 64, 82
moral ontological argument, 13-15
agents, 67 Oppy, Graham
character, 61, 67 universe, beginning of, 96
evil,238,264,265 argument from evil, 156-7
standing, 87-9
morality, 172, 206-7, 260-3 Paley, William, 3, 52, 58
Morris, Simon Conway Parikh, Maulik
ecological niches, 270 antipodal observers, 122, 123
Muller-Lyer illusion, 75-6, 78, 80 Parmenides, 100
multiverse, 4, 106-8, 124, 126, 137, Penrose, Roger, 111, 119-21, 127-8,
143-58,160,272 130
theistic, 143-59 cyclic conformal cosmogony,
scientific, 143-51 109-10
timeless, 106-7, 113 Hawking-Penrose singularity
contigentism, 147-8 theorem, 107
level IV, 143 Perkins, R. K., 154-6, 160
mysterianism persistence of, 229-30
McGinn, Colin, 231-2 personhood, 87-9
mystical experiences, 1, 42 philosophy
of mind, 3, 71, 88
natural selection, 3, 49, 51-68, 177, of religion, 4-6, 32, 143, 151, 158,
257,264,266,267,269-71 165,166,180,225
natural systems, 85, 87, 88 physical evil, 264-5
 Index 279

physical reality, 4, 104, 105 Quine, W. V., 12-13, 26, 79, lOS
physics, 4, 83, 96, 97, 99, 103, 105, epistemology, naturalised, 105
109, 110, 112, 120, 145, 158, modal involvement, three grades of,
258, 262 12-15
folk, 77-82, 90
laws of, 84, 89, 90 radiation, 116, 119, 120, 136
quantum, 56, 121 rationality of theism, 174, 175-7
Pike, Nelson Rea, Michael
timeless God, 32 epistemological naturalism, 105
Pitts, Brian J., 96-103, 108-10, 115, hyperspace criticism, 153
128, 133 reformed epistemology, 176
cosmic destroyer argument, religious
100, 133 agency, 71, 85-6, 88
Plantinga, Alvin, 181, 228 belief, 3, 6, 28, 32, 42, 71, 73, 82,
85, 86, 165-81
possible worlds, 14, 25, 102, 110, 133,
146, 147, 150-3, 158 tolerance, 2, 6, 209-10
reproductive success, 60, 73-4, 85-9,
pragmatism, 72-3, 85
267-8
precautionary principle, 213, 215-16,
218-20
Rowe, William
evidential problem of evil, 224
problem of evil, 3, 4, 49-58, 143,
151-8, 160, 190, 264, 265
Russell, Robert J.
direction at quantum level, 269
Darwinian argument from evil,
49-68
salvation, 6, 191, 193-5, 197-8, 199,
projectivist; see necessity
205-21,238
counterfactual thinking, 12-28
Schellenberg, John, 168, 178, 199, 229
psychologists, 5, 67, 74, 75, 237, divine hiddenness, 178, 199,229
239,242,244,2 45,246,
supernaturalism, 168
248,249 Schrodinger, Erwin
psychology, 6, 12, 26, 37, 75, 238 de Sitter space, 122
developmental, 2, 37 scientific revolution, 258
evolutionary, 3, 71, 74, 88, 175 Scotus, Duns
experimental, 7, 237-54 forced baptism, 208-9
of counterfactual thinking, self-knowledge, 43, 198
12-28 Sellars, Wilfred, 81, 90
folk, 77, 80, 81 manifest/scientific imagine, 81, 90
Pyysiainen, Ikka sensus divinitatis, 175-7, 178, 180, 228,
model of creation of theism, 34, 230
36,42 sin, 180,207,231, 238,254,264
singularity, 111, 112, 115, 117-20,
quantum 124, 127, 128, 129, 133, 135
cosmology, 125, 126, 134 Skinner, B. F., 243
fluctuations, 113, ll5, 116, Smith, Quentin
125, 131 criticisms of Hawking, 127-8, 130,
geometrodynamics, 126, 130 133
gravity, 103-6,108,10 9,113,114, Smolin, Lee, 106, 131, 143
117, 125, 127, 128, 130-2, fecund universe theory, 143
134, 136 space-time, 4, 96-101, 104, 106, 110,
physics, 56, 121 119-20, 122-3, 127-33
280 Index

Sperber, Dan theistic metaphysics, 67

representation, epidemiology of, theodicy, 67, 68, 233
170 theology, 1, 9, 26, 28-43, 144,
spiritual stance, 4, 72, 85-9 175, 176
Steinhardt, P. and Turok, N. technical, 3, 35-6, 43
cyclic model, 143 theory of everything, 112
Stewart, M. time's arrow, 121
distinguishing universes, 145 two system hypothesis, 75
suffering, 24, 53-61, 66-8, 153-4,
156-7,190,206,250 universalism, 195-8
supernatural, 1, 28, 29, 30, 170, 171,
191, 194, 231, 255 Vaas, Rudiger, 113, 115
agents, 167-70 van Inwagen, Peter, 149
belief, 177, 184 Vilenkin, Alexander, 108-9, 112,
entity, 96 116, 122, 124, 129, 131-2,
god, 177 135, 137
realm, 167 tunnelling from nothing approach,
supernaturalism, 168 109
Swinburne, Richard, 65
God's existence, argument for, Ward, Keith
224-S Donaldson, Margaret, 38, 40
warfare thesis, 7, 255-73
Talbott, Tom Wegner, Daniel
theistic multiverse, 149 Tile Illusion of Conscious Will, 239-53
technical theology, 3, 35-6, 43 Weinberg, Steven
Tegmark, Mark purpose in the world, 262
level IV multiverse, 143 Weyl Curvature Hypothesis, 119-21
theism Wheeler-DeWitt
classical, 186-8 approach to quantum gravity, 104,
Darwinian, 63-7 106, 125, 126,
demographics of, 5, 184-98 Whitehouse, Harvey
singularity argument for, 96 idea of religion, 30, 31, 34, 42
knowledge of, 189
multiverse-, 157-8 Zeh, H. D.
open,35 double big bang, 125-6
orthodox, SO, 53, 55-6, 58-9 zero-force evolutionary law, 271
rationality of, 174, 175-80 McShea, Daniel, and Brandon,
theistic evolutionists, 63-4 Robert, 271