Tim Mulgan - Purpose in The Universe - The Moral and Metaphysical Case For Ananthropocentric Purposivism-Oxford University Press, USA (2015)

OUP CORRECTED PROOF – FINAL, 3/8/2015, SPi
Purpose in the Universe

Purpose in the
Universe
The moral and metaphysical
case for Ananthropocentric
Purposivism
Tim Mulgan
1
3
Great Clarendon Street, Oxford, OX2 6DP,
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
# Tim Mulgan 2015
The moral rights of the author have been asserted
First Edition published in 2015
Impression: 1
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2015938208
ISBN 978–0–19–964614–2
Printed and bound by
CPI Group (UK) Ltd, Croydon, CR0 4YY
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
Contents
Acknowledgements vii
1. Introduction 1
2. Meta-ethics 33
Part I. The Case against Atheism

3. Cosmological Arguments 65
4. Teleological Arguments 99
5. Mysticism 130
6. Ontological Arguments 161
Part II. The Case against Benevolent Theism

7. Arguments from Scale 193
8. Arguments from Evil 220
9. Religious Diversity 261
10. Immortality 280
Part III. Ananthropocentric Purposivist Morality

11. A Dialogue 325
12. Human Well-being 343
13. Ananthropocentric Purposivist Moral Theory 373
Bibliography 417
Index 429
Acknowledgements
During the ten years I have been working on this book, I have accumulated more
debts than I can hope to acknowledge here. Most of that time was spent in the
Department of Philosophy at St Andrews, and I could not have asked for a more
collegial academic environment. I am grateful to all my St Andrews colleagues, and
especially to two successive heads of school—Peter Clark and Katherine Hawley. I am
also grateful to my colleagues at the University of Auckland, where I have been based
since 2012, and to the Princeton University Center for Human Values for their
generous hospitality in 2010/11.
I have presented earlier versions of the ideas in this book in seminars, conferences,
and public lectures at the universities of St Andrews, Auckland, Dundee, Edinburgh,
Glasgow, Leeds, London, Otago, Oxford, Reading, Rennes, Rome, and Victoria. I am
very grateful to audiences at all these venues for their comments and suggestions. For
discussions of specific chapters or themes of the book, I am grateful to Nigel Biggar,
Timothy Chappell, Peter Clark, Katherine Hawley, Melissa Lane, Janet McLean, John
Perry, and Alan Torrance. I have also learnt much over many years from conversa-
tions with Sarah Broadie.
I am very grateful to the four anonymous readers for Oxford University Press for
their detailed, forthright, and insightful comments; to Peter Momtchiloff for his
advice and encouragement over several iterations of this project; and to Kim
Richardson for copy-editing so efficiently.
For her constant love and companionship, for travelling to the ends of the earth,
and for bringing purpose to my small corner of the universe, I am forever grateful to
Janet McLean.
1
Introduction
Our universe is religiously ambiguous. It can be read in strikingly different ways. Two
familiar readings are materialist atheism and the benevolent God of the Abrahamic
faiths. In this book, I defend a less familiar reading, which I dub Ananthropocentric
Purposivism (AP for short): there is a cosmic purpose, but human beings are
irrelevant to it. AP borrows traditional theist arguments to defend a cosmic purpose,
and then it borrows traditional atheist arguments to reject a human-centred purpose.
My academic background is not in theology or metaphysics or philosophy of
religion, but in contemporary analytic moral philosophy. In this book, I explore the
moral case for, and the moral implications of, AP. Contemporary philosophy typic-
ally begins with some privileged world view (theist or atheist), and then asks where (if
at all) morality fits in. I begin with substantive normative commitments, and then ask
what metaphysical picture best fits those commitments.
In this introductory chapter, I first outline AP in contrast to its two main rivals,
atheism and benevolent theism. I then explore connections between morality and
metaphysics, beginning with the key notion of religious ambiguity. After outlining
my own moral commitments, and their role in the project, I offer some remarks on
the moral impact of AP. The chapter closes with a summary of the rest of the book.
1.1 What is AP?

In 2003, I taught the existence of God in a large first-year metaphysics course,
covering very familiar ground: cosmological, teleological, ontological, evil. I was
struck by the following thought. Theist arguments, if they succeed, establish that
there is a God of some sort. Atheist arguments, if they succeed, show that there is not
a God of one specific sort. So why not cut and paste the positive and negative
arguments, and thus end up with a God of a different sort?
This book is the result. I aim to introduce, defend, and explore the following
underdeveloped alternative to both theism and atheism:
Ananthropocentric Purposivism (AP). The universe has a non-human-centred purpose—one to
which human beings are completely irrelevant or incidental.
AP is a general idea that can be fleshed out in many different ways. For ease of
exposition, I often focus on theist AP, where a personal creator gives the universe
 INTRODUCTION
its purpose. But AP could take other forms—perhaps modelled on John Leslie’s
axiarchism, or on traditional Idealism or Platonism.1 Indeed, a personal God may
seem uncomfortably anthropocentric. Other things equal, AP is more comfortable
with an impersonal cosmic purpose. I try to remain as agnostic as possible
regarding the exact ontological nature of the cosmic purpose and its source. My
primary interest is not in the metaphysical details of AP, but rather in its moral
implications.
I contrast AP with two familiar positions. The first is traditional Western mono-
theism. The universe was created by a God who cares for individual human beings.
Philosophers have offered many competing accounts of God down the centuries.
Richard Swinburne’s definition captures the features that have become standard
among analytic philosophers: ‘God is a necessarily existing person without a body
who necessarily is eternal, perfectly free, omnipotent, omniscient, perfectly good, and
the creator of all things.’2 Unless I explicitly depart from it, I will be operating with
this definition of God.
I dub this opposing view ‘benevolent theism’ (BT for short). The boundary
between BT and AP is vague. BT says that God loves each individual human being;
that human beings are an essential part of God’s plan for the cosmos; and that God
created this cosmos (in part) because it would contain human beings. AP says that
God does not love individual human beings; that God has no interest in the fate of
humanity; and that the presence of human beings is a cosmic accident. Many
intermediate positions are possible. Perhaps human beings are a subsidiary part of
God’s plan for creation—less important than something else, but still not completely
irrelevant; or perhaps God cares about human beings once they emerge, even though
their existence played no role in God’s decision to create this particular cosmos. Some
arguments for AP over BT are consistent with these intermediate positions, while
others are not. Because my primary aim is to introduce AP into the philosophical
landscape, I contrast two ‘pure’ positions: a benevolent God who cares for each
human being and whose purpose for cosmic creation involves human beings; and an
AP that denies both that God cares for human beings at all and that we played any
part in God’s creative decision.
We are familiar with situations where my individual contribution or welfare is
irrelevant. My vote makes no difference to the election, for instance, because my
preferred candidate wins by a large margin. But here aggregate human impact is still
relevant. Our votes together determine the election. AP goes further. Human beings
are completely irrelevant to the cosmic purpose, to objective value, and to God’s
plans, concerns, or reasons for creation. Our cosmic significance mirrors the electoral
significance of mice, whose interests and opinions have no influence on human-
centred elections.
1 2
See, e.g., Leslie, Universes, and chapter 3 in this book. Swinburne, The Existence of God, p. 7.
INTRODUCTION 
However, AP does not insist on a complete separation between human beings and
cosmic purpose. AP can admit that human beings resemble some aspect of the
cosmic purpose; that we can understand that purpose to some degree; or that we
can bring it into our lives in a meaningful way. The cosmic purpose can matter to us,
even if we do not matter to it. These possibilities come to the fore in part III. Without
them, AP would be much less morally interesting.
BT religions often contain both anthropocentric and theocentric strands. Much
Christian theology, for instance, emphasizes the unknowability, the alienness, the
non-human-ness of God.3 While this theology can come close to AP, actual theist
religions typically contain some key doctrine that rules out AP. For Christianity, this
is the Incarnation. The Christian God becomes a human being for the sake of other
human beings. This is not something any AP God would do. AP cannot plausibly be
offered as a Christian apologetic.4
I contrast AP with ‘benevolent theism’ rather than ‘classical theism’, both to
include non-classical theisms such as process theology and also to emphasize the
human-centred element. However, this terminology is idiosyncratic, and may look
like a sleight of hand to conceal AP’s shortcomings. One reader raises the following
objection:
A cosmic purpose necessarily requires a divine person whose purpose it is. (Purposes require
persons by definition, and only a divine person could give purpose to a cosmos.) The simplest
imaginable divine person is the Omni-God of classical theism. Ceteris paribus, the simplest
hypothesis is the most likely. Therefore, the classical theist Omni-God is the most a priori likely
divine person. But this God, by definition, is benevolent to human beings. Therefore, AP is at a
distinct disadvantage relative to classical theism, a disadvantage which is obscured by substi-
tuting ‘benevolent theism’ for ‘classical theism’.5
As we shall see in chapter 3, the step from simplicity to probability is not

straightforward. (How do we measure comparative simplicity? Is theism the simplest
hypothesis or the most complex? And why expect reality to be simple in the first
place?) But AP has two other replies. The first is that the claim that purposes
necessarily require persons is a substantive thesis that must be defended, not a merely
definitional truth. As a moral philosopher, my central interest in cosmic purpose lies
in its connection to objective values. And it is certainly not obvious that the latter
require persons. (Impersonal Platonism may be misguided, but it is not incompre-
hensible.) Perhaps purposivism will ultimately collapse into theism, but this remains
to be seen.
3
One striking recent example is James Gustafson’s two-volume Ethics from a Theocentric
Perspective.
4
Theocentric Christians thus tend to downplay the second person of the Trinity. For instance, in 664
pages, Gustafson’s Ethics from a Theocentric Perspective devotes only five pages to Christology and only
one to Jesus. (I am grateful to Nigel Biggar and Mark Wynne for discussion of this issue.)
5
This paragraph paraphrases a much fuller discussion from an anonymous referee.
 INTRODUCTION
AP’s second reply is more direct and much more central to our project. AP denies
the final step from divine simplicity to divine benevolence. AP initially seems to be a
rival for classical theist metaphysics or perfect being theology. I contend, instead, that
AP can borrow from these traditions, by breaking the traditional link between
perfection (even moral perfection) and concern for human beings. AP need not
deny a divine creator. It can even admit an omnipotent, omniscient, or morally
perfect God. Indeed, AP could borrow Swinburne’s own definition in its entirety,
by denying that a perfectly good being must care for human beings. Swinburne’s
definition is thus neutral between AP and BT. Swinburne himself reads BT into his
definition, because he assumes that perfect goodness implies benevolence towards
humans. But whether human beings are morally considerable from the perspective
of a morally perfect divine person is the very question at issue between AP and
BT. BT is one interpretation of classical theism: the result of combining perfect
being theology with the substantive moral claim that human beings matter. By
denying the latter, AP offers a rival interpretation of classical theism, not a rival to
classical theism.
Similarly, BT and AP can offer competing interpretations of non-classical theist
traditions such as finite theism or process theism. I return to these alternative
traditions briefly in chapters 7 and 8. However, apart from a brief exploration of
Leslie’s axiarchism in chapters 3 and 4, I mostly consider BT and AP as competing
interpretations of the classical theist Omni-God, who is omniscient, omnipotent,
necessary, and morally perfect. Of course, controversy surrounds the coherence and
compossibility of these divine attributes. (Can a morally perfect God be truly free?
Can God create a stone that God cannot lift? And so on.) In so far as it is not
committed to any Omni-God, these controversies are grist to AP’s mill. But I largely
set them aside in this book. I concentrate instead on the central question that
separates AP and BT: does God care for us?
I contrast AP and BT with atheism. If we take atheism to be the denial of BT, then
AP is atheist. If atheism is the denial that there is any God, then AP overlaps with
atheism, because cosmic purpose can take non-theist forms. However, I shall take
atheism to be the denial of all supernatural entities, divine beings, and cosmic
purposes. Indeed, I go further. My atheist is also a naturalist who regards science
as the model for all human epistemic enterprises and the final arbiter of what exists.
(I discuss naturalism at greater length in chapters 2 to 5.) While this rules out some
possible positions, it does cover most contemporary philosophical atheism, and
leaves us with a simple threefold division. BT says the universe has a human-
centred purpose; AP says it has a non-human-centred purpose; while atheism says
it has no purpose.6
6
One problem case for my simple threefold distinction between AP, BT, and atheism is atheist
religions, such as Buddhism, which combine supernatural commitments with the rejection of God.
I briefly discuss Buddhist atheism, and its relationship to AP, in chapter 9.
INTRODUCTION 
The traditional debate between BT and atheism often tracks other philosophical
disagreements in metaphysics, morality, and methodology. BT is often combined
with belief in personal immortality, incompatibilist free will, dualism about mind and
body, and robust moral realism; while atheism is typically combined with material-
ism, determinism, the denial of personal immortality, and a naturalist or anti-realist
account of ethics. AP seeks a middle ground by separating the question of cosmic
purpose from these other debates. In particular, AP separates cosmic purpose from
the idea that human beings are metaphysically special.
AP is not unknown in the history of philosophy. (Similar themes are explored in
Hinduism and Taoism, and in some less orthodox Western forms of Deism and
Idealism.) But AP is definitely not a familiar position in contemporary Western
philosophy. AP is worth exploring in part because it is a comparatively unexplored
option. And there is also a surprisingly strong philosophical case for AP. Parts I and II
develop that case, arguing that it is at least as strong as the case for either BT or
atheism. Any philosopher who takes BT and atheism seriously should also take AP
seriously.
AP is the conjunction of two claims:
(1) The universe has a purpose; and
(2) The universe does not have a human-centred purpose.
AP borrows the best anti-theist arguments from atheists and the best anti-atheist
arguments from BT. BT arguments establish only (1), while atheist counter-
arguments establish only (2). When atheism and BT are the only options on the
table, these arguments succeed. But faced with a three-way choice between BT,
atheism, and AP, they combine to establish AP.
Theism, Atheism, and AP
THEIST (1) not (2)
Not (1) (2) ATHEIST
AP borrows arguments, but not arguments.
Parts I and II explore many different arguments. No single argument is decisive

and none is indispensible. Rather, each contributes to a cumulative case for
AP. Part I develops the case for (1), arguing that AP can borrow the best traditional
arguments for BT. Part II develops the case for (2), arguing that AP can borrow the
best traditional arguments for atheism. The conclusion of parts I and II is that
arguments leading to (1) are much stronger than those leading beyond to the
negation of (2)—and vice versa on the atheist side. Theists and atheists should both
 INTRODUCTION
regard AP as their main rival. If BT and atheism are both worth exploring, then
so is AP.
This will suffice to interest philosophers of religion in AP. But many moral
philosophers have no interest in religion. The more ambitious task of this book is
to convince those moral philosophers that, in addition to its metaphysical merits, AP
is relevant to them. This is the task of part III. It is prefigured in section 1.5, and in the
next chapter.
I claim that there is a philosophical case for AP that rivals BT and atheism. Ideally,
I would defend this claim by presenting an actual case for AP that was as powerful as
the best extant arguments for theism and atheism. Sadly, this task is probably beyond
any single philosopher. It is certainly beyond me. The case for BT contains many
arguments honed over many centuries. For most of its history, the relationship
between God and morality was one of the central questions in Western philosophy
and the philosophical elaboration of BT attracted the most brilliant thinkers of the
age, while the past few decades have seen the best medieval arguments sharpened
using the tools of contemporary analytic philosophy.
The extant philosophical case for atheism is less impressive. Within Western
philosophy, atheism has moved too quickly from a blasphemy that dare not speak
its name to a dominant world view that need offer no defence. Only within the past
generation have atheist analytic philosophers confronted the need to justify them-
selves to their theist colleagues. Atheist apologetics is still in its infancy. Nonetheless,
the philosophical resources available to contemporary atheism, while less formidable
than those available to the theist, still far exceed anything any one individual
proponent of AP could hope to construct in a single lifetime, let alone a single
book (even one as unreasonably long as this).
Aside from sheer quality, the arguments for theism and atheism also display a
diversity that AP cannot (yet) hope to match. The ‘case’ for BT is really a myriad of
divergent but complementary approaches, rooted in distinct philosophical traditions,
methods, and assumptions. Even confining themselves to the analytic mainstream,
theists seeking justification can choose between the diverse approaches of Robert
Adams, William Alston, William Lane Craig, Alvin Plantinga, Alexander Pruss,
Richard Swinburne, Peter van Inwagen, or Linda Zagzebski, to name but a few.
I cannot offer a case for AP that is as rich, deep, or brilliant as those of its rivals. My
goal is much more modest. I seek to illustrate AP’s potential by sketching the outlines
of one particular cumulative case for AP, built on one initial set of pre-commitments
and preoccupations, and emerging from one particular philosophical tradition. My
hope is that those who prefer an alternative starting point may see how it too, in
sympathetic hands, could ground an argument for AP.
I have described my case for AP as ‘cumulative’. Familiar cumulative arguments
for BT start with a clearly defined thesis, and proceed by steadily increasing its
probability. The classic exemplar is Richard Swinburne, who begins with a standard
Christian definition of God, assigns a prior probability to the hypothesis that God
INTRODUCTION 
exists (using a priori appeals to relative simplicity), and then presents a series of
interlocking inductive arguments designed to raise the probability of that hypothesis.
The goal is to demonstrate that God probably exists.7
Things are different for AP. There is no agreed definition and thus no pre-existing
hypothesis to be tested. AP begins as the vague claim that there is some non-human-
centred cosmic purpose. Several key terms are under-defined (What counts as a
purpose? What makes it cosmic? How exactly are humans excluded?), as are some
key collateral concepts (Does a purpose require a person? Does AP need a God? Is
God a perfect being? How does cosmic purpose relate to value, normativity, or
morality?). Rather than settling these controversial questions by fiat at the outset,
I progressively flesh out my own particular conception of AP through the course of
the book. Each successive argument not only contributes to the case for AP; it also
offers new insights into the content of AP. The view that emerges is one precisifica-
tion: what AP might look like if one were persuaded by all the arguments in this
book. Proponents of AP who are persuaded by other arguments, or by only a subset
of those presented here, will endorse different interpretations.
One contrast is especially salient. Simple AP says that the universe was created by
someone who does not care for human beings. Simple AP is easy to imagine and
understand. Nor is it entirely implausible. For instance, if one combines a fine-tuning
argument for theism with an argument from evil against benevolent theism, one will
conclude that there is a creator God who is not benevolent. Many cultures have
creation myths that instantiate Simple AP, where some distant creator sets the world
in motion and then plays no further part in human affairs. (It is only lesser non-
creator deities who concern themselves with the mortal realm.)
If I sought only to defend Simple AP, this book could be much shorter. Instead,
I aim at Normative AP, where the non-human-centred cosmic purpose is a ground
for objective values and external reasons that have normative significance for human
beings. Normative AP clearly encounters difficulties that Simple AP avoids. Norma-
tive AP is much harder to grasp, understand, or imagine. Even if we grant Simple AP,
we may still find Normative AP incredible. This is especially true if our Normative
AP is exclusive: holding that non-human-centred cosmic purpose is the sole source of
normativity for human beings. (Among other unpalatable conclusions, Exclusive
Normative AP implies that human suffering has no objective significance at all.)
I will try to defend Exclusive Normative AP in this book. (Or, at least, to render it
less implausible.) However, I am not committed to this extreme view. In part III,
I explore more moderate variants of Normative AP, where cosmic purpose is only
one source of values and reasons among others. On the other hand, despite its
apparent difficulties, I am committed to Normative AP rather than Simple AP. The
7
For instance, Swinburne himself estimates the probability that God exists at somewhere between 0.2
and 0.8. (Swinburne, ‘The Argument to God from Fine-Tuning Reassessed’, p. 113.)
 INTRODUCTION
former is much more philosophically interesting, and more compelling, than the
latter. Normative AP has resources that Simple AP lacks.
One might think the case for Simple AP must be stronger than the case for
Normative AP, because the latter entails the former. But the real question is whether
one can endorse Simple AP without also endorsing Normative AP. Can the propon-
ent of Simple AP leave open the normative significance of cosmic purpose? I shall
argue that she cannot. The best philosophical case for Simple AP already commits
one to Normative AP. We cannot establish the existence of a cosmic purpose without
also establishing its normative significance. We do not first prove AP using morally
neutral arguments, and then assess its connection to values and reasons. Rather, if it
arrives at all, AP comes already saturated with objective values.
The essential normativity of AP is the central theme of this book. This claim often
strikes secular moral philosophers as absurd. But consider an analogous contrast
between two interpretations of Benevolent Theism:
• Simple (Morally Agnostic) BT: God exists, but the normative significance of this
fact for human beings is left open.
• Normative BT: God exists, and God’s plans and purposes are a source of
objective values and reasons with normative significance for human beings.
The dialectic is the same as for AP. On the one hand, Normative BT faces challenges,
difficulties, and imaginative barriers that Simple BT avoids. Most notorious, of
course, is the Euthyphro dilemma. Drawing on that dilemma, some atheists argue
that, while Simple BT is imaginable (though under-supported by the evidence),
Normative BT makes no sense. God could not have the right kind of normative
significance. God’s cosmic purpose might influence my response to moral facts about
values and reasons, but it cannot ground those facts. (We return to Euthyphro in
chapter 2.) On the other hand, Normative BT has resources that Simple BT lacks,
because morality itself can now provide evidence that God exists.
Despite the extra burdens of Normative BT, no serious theist philosopher defends
Simple BT. No doubt this is partly for doctrinal reasons. But the philosophical case
for Normative BT is also much stronger and more interesting than the case for any
morally agnostic theism. Arguments for BT work much better if they deliver a God
who has an intimate connection to human morality.
1.2 Ambiguity: Religious and Philosophical

I have argued that, if BT and atheism are worth exploring, then so is AP. Of course,
not everyone thinks that BT and atheism are both worth exploring. Some atheists
find traditional theist arguments worthless, arguments from evil unanswerable, and
the case for naturalism compelling. They conclude that there is nothing to be said for
theism. Some theists find some argument for BT compelling and regard arguments
from evil as worthless. They conclude that there is nothing to be said for atheism.
INTRODUCTION 
None of these people is likely to find this book very interesting. But many philo-
sophers agree that knockdown arguments are rare anywhere in philosophy, and
especially here. An emerging theme of the literature in philosophy of religion is
that our universe is religiously ambiguous.8 BT and atheism are both reasonable
interpretations of the available evidence. AP offers an alternative interpretation—a
new way to read our cosmos.
In this book, I assume religious ambiguity. Given our limited understanding,
radically different interpretations of the basic nature of the cosmos are equally
reasonable: idealism versus realism, theism versus materialism, naturalism versus
non-naturalism. Reason alone—scientific, philosophical, mathematical, or moral—
cannot resolve metaphysical debate. Religious ambiguity is a non-eliminable feature
of our current epistemic situation. It may even be intrinsic to the human condition
per se.9 By highlighting the credentials of AP, I hope to strengthen the case for
religious ambiguity. But my project is aimed squarely at those who already find this
idea compelling.
Ambiguity is not peculiar to religion. It is a pervasive feature of all interesting
philosophical questions. Philosophical debate is never settled by rational argument.
Philosophy never answers its own questions. Consider the following philosophical
questions, many of which will occupy us later in this book. Is genuine free will (the
kind necessary for moral responsibility) compatible with determinism? Do human
beings possess genuine free will? Is there a plausible reductionist naturalist account of
knowledge or consciousness or mystical experience or modality or value or morality?
If not, should we become sceptics about these domains or should we reject natural-
ism? Will the best moral theory take a deontological, consequentialist, or virtue
ethical form? Is there any best moral theory? Are moral statements truth-apt? If so,
are any of them true? If not, what on earth do they mean?
Philosophical ambiguity goes hand in hand with methodological pluralism. We
should not presuppose that all philosophical arguments fit a common coin. In
particular, arguments for the existence of God come in a bewildering array of styles.
Many atheists try to reduce this rich variety to a single pattern: any argument for the
existence of x must be an inference to the best explanation and all adequate
explanations are scientific. Of course, some BTs do offer inferences to the best
explanation. And some even present scientific explanations. But these are often the
least successful BT arguments, and the least philosophically interesting. Not all good
philosophical arguments are inferences to the best explanation. Of the classic argu-
ments, only cosmological, teleological, and (some) moral arguments seek to explain
8
For references and further discussion, see chapters 5 and 9.
9
One intriguing question is whether religious ambiguity is also a necessary feature of the epistemic
situation of other (non-human) non-divine agents, such as aliens or superintelligent machines. (The
answer may turn on whether God has some particular reason to remain hidden from all created
intelligences.) I briefly explore the significance of possible non-human agents in chapter 7, and hope to
address it at greater length elsewhere.
 INTRODUCTION
anything. And—even more importantly—not all explanations are scientific. Indeed,

both cosmological and fine-tuning arguments start from the claim that science
cannot explain some striking fact. It would be odd if they then offered their own
scientific explanations!
Methodological pluralism is especially prominent in contemporary moral philoso-
phy. (I personally find it one of our discipline’s most appealing features.) In its search
for new methods, moral philosophy often engages more intimately with its own history
than some other branches of contemporary philosophy.10 Contemporary moral philo-
sophers constantly reinvigorate old ideas, approaches, and theories using modern
analytic tools. Methodological pluralism leads us to take seriously questions and
answers that were prominent in earlier philosophy, even if they are difficult to translate
into contemporary analytic or scientific modes of thought. In particular, anyone
exploring connections between God and morality is naturally drawn away from a
present where both topics are philosophically marginal, and toward a past where their
connection was perhaps the central philosophical question.11
Several other kinds of pluralism feature in this book. Our discussion of mysticism
in chapter 5 appeals to doxastic pluralism—the claim that there are several distinct,
incommensurable, rational ways to form beliefs. Our argument from evil in chapter 8
appeals to value pluralism, where human values do not fit into a common scale, and
are thus not fully commensurable. I devote much of chapter 9 to religious pluralism:
one philosophical interpretation of the fact of religious diversity. Part III explores
pluralism about responses to value: the thought that humans should respond to
cosmic values in many different ways—including contemplation, study, and worship,
as well as promotion.
A final pluralism concerns individual intellectual commitments. Many philosoph-
ical disagreements turn, not on purely factual or logical claims, but on divergent
intuitions, considered judgements, basic orientations, philosophical sensibilities, or
prejudices. Substantive philosophical disagreement often tracks some deeper dis-
agreement about what is possible, valuable, plausible, simple, elegant, undeniable, or
obvious.
Philosophers seek decisive intuitions that any acceptable theory must accommo-
date. But, in fact, most intuitions serve, not to establish or refute theories, but to
distinguish them.12 No intuition is uncontroversially decisive, if only because there is
always a niche in the philosophical marketplace for the first person to reject it. An
intuition that seems decisive to a theory’s proponents often strikes its opponents as
merely peculiar.
10
Consider the continuing influence, within normative ethics, of figures such as Aristotle, Hume, Mill,
and Kant; and of historical works such as Schneewind, The Invention of Autonomy, or Irwin, The
Development of Ethics.
11
On the centrality of God and morality throughout medieval and early modern philosophy, see
Schneewind, The Invention of Autonomy.
12
Mulgan, Future People, pp. 2–4.
INTRODUCTION 
Some people are happy to accept that whatever science cannot explain is a brute
fact. Science will eventually explain most things, and anything left over is simply
inexplicable. Others feel, on the contrary, that the very existence of any universe (let
alone one as remarkable as this) cries out for explanation, and that the search for
such explanations is a central task of philosophy. If science cannot answer the basic
questions of metaphysics, then we must look beyond science.
People similarly divide over the perennial philosophical programme that seeks to
naturalize everything. Some take it for granted that human beings are merely physical
animals who evolved via natural selection, and that all our features must be explicable
in purely physical scientific terms. Our consciousness, rationality, freedom, and
morality are nothing more than by-products of evolution. By definition, the best
naturalist explanation must be adequate. Others think that reductionist projects
inevitably leave out something that is vital to our nature; that a naturalized ethic is
no ethic at all; and that there is something to the realm of normativity and value that
cannot possibly be cashed out in naturalistic terms.
These are not disputes within science. They concern instead the limits and
pretensions of science. Can science explain why there is a physical universe where
life could evolve; why human beings can think; why torture is wrong; or why beauty
is valuable? If not, must these things simply be rejected as illusions, or can they be
accepted as brute facts, or do they instead cry out for explanation? The arguments
that AP borrows from BT all appeal to the thought that there are some vital questions
that science, operating in a purely atheist framework, simply cannot answer—and
that these questions are worth asking.
AP thus responds to a common source of dissatisfaction with contemporary
atheism. It also captures popular dissatisfaction with BT. Some people can accept
that all the evil we see around us is part of the plan of a perfect loving God. They can
believe and trust in such a God—and often see something impious or arrogant in
human attempts to explain evil to our own satisfaction. But others find it obvious that
the amount and distribution of evil in this world are far beyond what any benevolent
God could permit. Indeed, many people find the very business of attempting to
explain away or excuse the evils of the world tasteless or offensive. Not only don’t
they believe in God—they don’t want there to be a God who would create a world
such as ours. The arguments that AP borrows from atheism appeal to those who
cannot imagine how any benevolent God could reconcile the evils of this world—
those who find all extant theodicies unsatisfactory.
Atheism and BT each correspond to a natural package of pre-philosophical
attitudes. Those who find science satisfying, and happily accept as brute fact whatever
science leaves unexplained, often also find it impossible to believe in an evil-
permitting God—partly because their naturalistic ethic leaves no room for anything
that could outweigh the evils we see around us. This combination hangs together
well. But so does the opposite combination. Those who can believe in a perfectly
loving God who reconciles the evils of the world often also feel that both a naturalistic
 INTRODUCTION
ethic and a focus on the animal nature of human beings leave out something vital and
that scientific explanations cannot be the whole story about the universe.
These pre-philosophical pictures explain why, here as elsewhere, philosophical
argument so seldom results in anyone changing their mind. Philosophy is primarily
apologetic rather than evangelical. Arguments serve to clarify the competing
positions—bringing out the logical connections within each package.
The literature on moral faith offers two striking metaphors. One is the leap of faith:
committing oneself to P despite one’s realization that evidence and argument are
insufficient to establish that P. This metaphor suggests one is initially agnostic, and
then leaps to BT (or atheism). The recent literature often talks instead of holding one’s
ground: retaining one’s commitment to P despite one’s inability to establish that P.13
Here, one already believes in BT (or atheism) and merely seeks to rebut challenges to
that view.
The distinction between these two metaphors matters, because it can be rational to
stand firm where it would not be wise to leap. If I have built my life around a shared
belief that P, this could be a compelling reason to hold fast to P, even if I could have
formed an equally strong commitment to not-P. (Perhaps, at a pivotal time in my life,
I happened to fall in with Protestants rather than Catholics, or with analytic logicians
rather than postmodernists.)
The contemporary philosophical literature often takes the internal perspective of
the religious believer, and then asks whether she has reason to abandon her faith. By
contrast, I begin with an external agnostic perspective, and ask what reason there is to
believe in BT or atheism or AP. This difference becomes especially significant in
chapter 8, where it shifts our focus from defences of BT (accounts of God’s possible
reasons to permit evil) to theodicies (probable stories about God’s actual reasons).
While the literature gestures at intermediate positions in logical space, atheist
materialism and benevolent theism are the only lived philosophical positions in our
time. AP is not a lived position. Belief in AP always requires a leap. But AP could still
be the best way to stay put overall. If AP best explains the presence of both genuine
value and genuine evil in the world, then those whose lives are built around the
recognition of good and evil may need to posit AP if they are to resist the threat (or
the lure) of nihilism.
The very fact that BT and atheism are so well suited to popular philosophical
packages should make us suspicious. Perhaps they only dominate because of their
attitudinal fit, and not from any philosophical superiority. As we shall soon see,
proponents of AP diagnose a single moral failing in both traditional packages. They
both rest on immodest premises, or on attitudes that overestimate the capacities and
nature of human beings and give us an unwarranted role in the universe. Perhaps the
traditional packages dominate precisely because, in their different ways, they serve
13
This is a prominent theme in the work of William Alston and Alvin Plantinga (see chapter 5 in this
book).
INTRODUCTION 
our vanity—presenting humanity either as the culmination of the divine plan for
creation or as the sole source of value in an otherwise meaningless cosmos.
On a more positive note, perhaps there is some other appealing combination of
pre-philosophical attitudes that is represented by neither BT nor atheism. Surely one
could feel that the very existence of the cosmos cries out for explanation; that the
limits of science are not the limits of understanding; and that any naturalized ethic is
unsatisfactory—while also finding all extant theodicies unconvincing. AP allows us to
separate these different commitments and explore new combinations.
Unsurprisingly, the pre-philosophical attitudes I have just described are my own.
While I can easily believe that the universe was designed—that it has a purpose—I
cannot easily believe that it was designed by someone who cares for us. My aim in
this book is simply to see where this plausible combination of pre-philosophical
attitudes, which sits uneasily between BT and atheism, might lead. I do not claim that
AP is the only way to accommodate some decisive intuition. But it does capture a
reasonable and coherent picture of the cosmos and our place in it.
1.3 Morality and Metaphysics

AP is a metaphysical thesis: a claim about how the world is. In a book about
metaphysics, many will feel that morality belongs only at the end. Perhaps a small
epilogue will explore the moral implications of the metaphysical truth. But the
metaphysical argument itself should be value-free.
This is perhaps the most common picture of relations between moral philosophy
and metaphysics. Metaphysics is prior to morality. What we ought to do in the world
depends on how the world is. This sounds fine in theory. In practice, however, it is
less compelling. Once we recognize pervasive religious and philosophical ambiguity,
we must abandon the idea that there is a single neutral objective account of ‘how
things are’.
A second popular view reverses the priority between metaphysics and morality.
A long BT tradition, associated most famously with Kant, holds that we have decisive
practical reason to respond to religious ambiguity with a leap of faith. The basic
questions of metaphysics cannot be answered by theoretical reason, and so we turn to
practical reason. We can help ourselves to metaphysical claims that are necessary for
morality—even if theoretical reason sheds no light on them whatever. (We consider
Kant’s own leaps—freedom, God, and immortality—in chapter 10.)
Another response to religious ambiguity is to treat morality and metaphysics as
completely independent domains. Moral philosophy seeks normative views that
could be the subject of agreement among people with very different comprehensive
metaphysical beliefs. (The most famous example, of course, is John Rawls’s political
liberalism.) This strong independence claim is problematic even when confined to
political justice. The search for a metaphysically neutral foundation for the whole of
morality is even more ambitious, and hence even more problematic.
 INTRODUCTION
My own view is that the truth lies somewhere in-between these three positions.
We cannot deduce our morality from our metaphysics or vice versa. But neither
are the two completely independent. Sometimes a thesis in moral philosophy and
a thesis in metaphysics are mutually supporting. While each has some independent
plausibility, they also reinforce one another. Such mutual support is commonly
sought within moral philosophy—and within metaphysics—so it is natural to seek
it between the two. This book highlights one example of mutual support. It thus
aims to bring contemporary moral theory and contemporary philosophy of reli-
gion into a closer dialogue. Both areas of philosophy have blossomed over the past
thirty years. But they have largely done so in isolation. In particular, connections
between consequentialist moral philosophy and contemporary philosophy of
religion are under-explored. With some notable historical exceptions (such as
William Paley), consequentialists have tended to be atheists (or at least agnostics),
while BTs are invariably non-consequentialists. (Indeed, some of the most virulent
recent opponents of consequentialism have been theists.) This book attempts
to correct that imbalance. I therefore make no claims to provide a neutral or
unbiased discussion.
Morality plays four roles in this book.
1. We might need AP to ground or explain moral facts. In the next chapter,
I explore the BT argument that, without God, nothing could be good or evil,
right or wrong; and I ask whether AP can borrow that argument.
2. Moral or evaluative claims often feature as premises in arguments for AP, even
in arguments that seem purely metaphysical. Some of these premises are
familiar, but others are surprising. The ubiquity of evaluative premises within
metaphysical arguments is a central theme of this book and it recurs through-
out parts I and II.
3. Religious ambiguity requires an ethic of belief. We need moral principles to
guide us when non-normative considerations run out.
4. AP influences the content of morality. While AP supports some familiar moral
ideals, it also pushes those ideals in a very unfamiliar direction. Adopting AP
will not leave morality unchanged. Indeed, AP is so unfamiliar and so austere
that one may worry that it will obliterate human morality altogether. This
worry is addressed in part III.
All four roles depend on what our moral commitments actually are. My case for
AP rests on my own moral views, some of which I outline in the next section. This
may seem to drastically limit the book’s relevance. Why should anyone else care?
Why address only one person’s moral commitments?
One reply is that this is my book. Where else could I start but from what I believe?
As a moral philosopher, my strongest commitments are normative rather than
logical or metaphysical. So that is where I begin. A full treatment of AP relative to
every credible ethical view would simply be unmanageable.
INTRODUCTION 
If my moral commitments were idiosyncratic, the book would be of little interest.

But, on the whole, they are not. While I sometimes appeal to very specific evaluative
claims (about possible futures or components of human well-being, for instance), the
central normative commitments invoked in this book are very general: the compara-
tive value of suffering and freedom, the objectivity of value, moral realism, an austere
picture of morality, and a broadly utilitarian ethic of belief. These views are not
universal, but they are widespread.
This book develops only one possible case for AP driven by one specific starting
point. But my broader aim is to demonstrate that anyone who takes morality
seriously, and finds the existing alternatives unsatisfying, has good reason to consider
AP. There are many possible routes to AP, just as there are many possible sources of
dissatisfaction with both BT and atheism.
1.4 My Moral Commitments

One divide in contemporary philosophical discussion of morality is between those
who study morality as an application of some other branch of philosophy, and those
who explore our moral life from the inside. My approach is the latter. I begin with
substantive moral questions, not with epistemology, philosophy of language, or
metaphysics. In particular, I begin with substantive ethical commitments, rather
than a commitment to some metaphysical world view. I take moral talk and moral
life at face value and ask what metaphysical or epistemological picture does them
justice, rather than requiring moral talk to fit some predetermined paradigm of
legitimate philosophical discourse. This commitment to taking moral talk at face
value may lead in surprising directions. At least, some have surprised me, and they
have often dismayed my consequentialist colleagues.14
1.4.1 Utilitarianism and suffering

I operate within the utilitarian tradition, broadly construed. Any account of a
tradition as rich and varied as utilitarianism is bound to be controversial. My aim
here is not to offer a detailed historical exegesis, but rather to draw out some
utilitarian themes. The defining feature of utilitarianism is that it measures every-
thing (acts, rules, moral codes, social institutions, beliefs) in terms of the promotion
of well-being. (Contemporary utilitarians often defend a broader consequentialism
where morality is based on the promotion of values that may go beyond human well-
being. This development is supported by AP, as we shall see in part III.)
14
This section draws freely on my own previous work, especially Mulgan, Understanding Utilitarian-
ism; Mulgan, ‘Mill for a Broken World’; Mulgan, ‘Utilitarianism for a Broken World’; Mulgan, ‘What is
Good for the Distant Future?’; Mulgan, Future People; Mulgan, Ethics for a Broken World; and Mulgan,
‘Ethics for Possible Futures’.
 INTRODUCTION
I begin with a very straightforward utilitarian commitment. Utilitarians emphasize

the central ethical significance of suffering. As a result, they must take the problem of
evil very seriously and cannot accept any traditional solution. In particular, utilit-
arians insist on the comparative importance of suffering. They reject the (as they see
it) exaggerated notions of human freedom central to theodicy. Free will and self-
development cannot justify the sufferings of this world. Faced with the bleak facts of
human life, the utilitarian offers a bleak explanation. If there is a God, then God
permits suffering because God does not care for individual human beings. This
commitment plays a central role in chapter 8, the cornerstone of part II.15
The utilitarian emphasis on the significance of suffering is a double-edged sword.
By linking objective value to a non-human-centred cosmic purpose, AP seems to
imply that human suffering has no moral significance at all. Even the most intense
agony is not bad at all, from the point of view of the universe. Utilitarians, along with
everyone else, will wonder how anyone could endorse such a counter-intuitive claim.
I concede that AP’s position on human suffering is very counter-intuitive. Perhaps
it will ultimately prove too extreme for most utilitarians to accept. However, over the
course of this book, I try to render this bullet easier to bite. One defence, of course, is
comparative. All three competitors have counter-intuitive implications; it is only
familiarity that makes BT and atheism seem more plausible. I also seek to reduce AP’s
moral counter-intuitiveness. In part III, I argue that AP can consistently agree that
human suffering has great significance for human beings, even if it does not matter to
God, and that this is sufficient to ground a recognizable (if very austere) human
morality. The result is far from what we might initially expect, but it could represent a
new reflective equilibrium: what we would think once we had adjusted our moral
vision to a world governed by a non-human-centred God. Finally, as I’ll now argue,
utilitarianism has other commitments that pave the way to the transcendence of its
own human-centred elements.
1.4.2 A utilitarian ethic of belief

Before outlining my other substantive moral commitments, I turn first to the ethics
of belief. Moral philosophers should take the ethics of belief especially seriously. We
naturally see every dispute as a question of what to do. So it is natural for us to see
metaphysical disputes in this light. (And this should be especially natural for
utilitarians, who bring all aspects of life before the tribunal of the utility principle.)
My utilitarian ethic of belief is built on two foundations: Mill’s liberalism and
Bentham’s aversion to caprice. Straightforward application of Mill’s famous utility
principle to individual acts of deciding what to believe would yield a very simple
principle: believe P if and only if believing P is what best maximizes human well-
being. This is obviously too crude. I follow instead Mill’s own liberal utilitarianism,
15
As we shall see in chapter 8, dialectical context is crucial here. It is as theodicy, not as defence, that
utilitarians reject the standard BT story about free will.
INTRODUCTION 
which emphasizes the collective value of divisions of labour, and of each individual’s
following her own desires and values. Each individual life is, in Mill’s words, part of
an experiment in living.
I am especially interested in moral experiments. Some are practical. (Consider a
landowner freeing his own slaves before emancipation, an early advocate of vegetar-
ianism and animal rights, or a contemporary climate change activist.) But a moral
experiment can also be speculative. In a work of fiction or philosophy, one might
imagine a wider collective shift, and then offer that vision to one’s fellow citizens as
an inspiration for further experiments.16
This liberal utilitarian ethic of belief favours a variety of different moral experi-
ments, a diversity of responses to religious and philosophical ambiguity. One
response is epistemicism, where one only believes what is dictated by the evidence.
The epistemicist response to ambiguity is agnosticism. This is one respectable
response, but not the only one. There is room, in a pluralist Millian ethic of belief,
for other responses.
Utilitarians will especially favour moral experiments that both promote well-being
and also correct deficiencies in current commonsense morality. My second utilitarian
foundation is a diagnosis of those deficiencies, based on Bentham’s aversion to
caprice. Utilitarians are suspicious of our natural human tendency to adopt inter-
pretations of the world that favour ourselves or overestimate our own importance.
They worry that current social structures and moral norms exist because they
disproportionately serve the interests of the powerful. (After all, if you are not
counting all interests equally, then you must be giving disproportionate weight to
someone.)
This aversion to caprice inevitably takes a self-directed turn. The utilitarian thinker
seeks out options than downplay her own interests—and also the broader interests of
her own group, caste, class, nation, or even her own species. The history of utilitar-
ianism is a constant tension between Mill’s desire to create a private sphere safe from
the incessant demands of morality, and the nagging Benthamite suspicion that this
affluent safe haven is just another instance of indefensible caprice.
In the theoretical realm, caprice can arise both in our epistemic attitude to a
proposition and in the proposition’s substantive content. Epistemic caprice consists
in overvaluing one’s epistemic capacities or in placing undue faith in one’s own
cognitive resources. It is epistemic caprice to dismiss reports of mystical experience
out of hand when one has never made any serious attempt to cultivate such
experiences oneself; to assume that every question will eventually be answered by
science, by philosophical argument, or by divine revelation (if one is, respectively, a
scientist, a philosopher, or a theologian); or to take the success of empirical science to
show that every significant question is amenable to the precision and tractability of
16
I attempt such a moral reimagining myself in Ethics for a Broken World and ‘Ethics for Possible
Futures’.
 INTRODUCTION
scientific explanation. Perhaps some very significant aspects of the cosmos are such
that human beings can only glimpse them through a glass darkly.
Epistemic caprice is often combined with substantive caprice, where the favoured
proposition has self-serving content. Consider the belief that one’s own interests,
perspective, or values carry more weight than those of others; or that there is
something special about me, about my group, or about human beings in general.
Of course, a belief in one’s own superiority is not always capricious. You may have
good reason to believe that you are better at philosophy than me or that human
beings are more valuable than rocks. But suspicion is appropriate unless and until we
have especially good reason to believe that we are special. Some cases are clear-cut:
racism is capricious, while God’s self-regard is not. Considerable room for reasonable
disagreement lies between these extremes. One fertile source of contested cases is
environmental philosophy. Do we have sufficient justification to place human inter-
ests above those of primates, or ecosystems, or species? Do human freedom, reason,
or morality suffice to mark us out from the (merely) physical world around us?
Faced with religious ambiguity, or the ubiquitous fact that philosophical argument
is never conclusive, we often have to take a leap of faith in one direction or another.
My utilitarian ethic of belief cautions against self-aggrandizing leaps. We are natur-
ally inclined to overestimate the case for our own significance. It does not follow that
we are not significant, nor that it is irrational for us to believe so. An overestimated
case can still be a sound one; and comforting self-appraisals can still be accurate.
However, given our self-aggrandizing nature, we should always be open to discover-
ing new and unexpected kinds of caprice and new sites of illicit self-aggrandizement.
At the very least, any view that questions even our most cherished moral commit-
ments on the grounds of caprice ought to receive a hearing. AP is one such view, and
this book is an attempt to give it its day in court.
My aim, therefore, is to exaggerate the utilitarian suspicion of caprice, push it in
unexpected directions, and follow where it leads. Many readers will feel that this
stretches a legitimate ethical concern too far, especially when the collateral costs
include many other fundamental moral commitments. I can only ask that such
readers withhold judgement until the end of part III, and only reject AP once they
have seen what positive moral vision (if any) it can build over the shattered ruins of
our self-aggrandizing self-image.
If our total package of arguments leaves all three options on the table, then the
choice between BT, atheism, and AP involves its own leap of faith. And here the least
self-aggrandizing option, for human beings, is AP. Consider, first, the choice between
AP and BT. Supposed we are convinced there is a God, but we are undecided between
human-centred and non-human-centred interpretations of God’s values, concerns,
or purposes. Ex hypothesi, we have insufficient evidence or argument to support
BT. (In particular, we lack compelling reason to believe that human beings possess a
moral significance that would matter to God.) In this epistemic situation, it is simply
caprice to leap to BT rather than AP. If we must make some leap of faith, then AP is
INTRODUCTION 
the only non-capricious option.17 (Of course, many theists will deny that this is our
epistemic situation. But that is another matter!)
A leap to BT can thus seem capricious. What about atheism? Of course, most
atheists don’t accept that they have leaped at all. They indignantly insist that atheism
is not a faith position, like Christianity or homeopathy, but merely a dispassionate
inference from objective facts. But this is simply the denial of religious ambiguity.
And the refusal to admit that one has leaped—the conflation of one’s personal
commitments with objective reality—is itself evidence of self-aggrandizing caprice.
If we acknowledge religious ambiguity, then atheism is a leap of faith. But is it a
capricious leap? In stark contrast to BT’s human-friendly cosmos, the atheist’s world
seems bleak, hostile, meaningless. What could be self-aggrandizing about the scien-
tific picture of humans adrift in a vast uncaring universe? One answer lies in the
dialectical fact that most arguments for atheism are actually only arguments against
BT. Before one leaps to atheism, one must first leap to the Human-Centred Condi-
tional: If there is a God, then God cares for human beings. (Otherwise, traditional
‘atheist’ arguments such as the argument from evil get us nowhere.) But this
conditional is, if anything, more self-aggrandizing than BT itself. BT asserts that
God does care for human beings, while the conditional insists that any God must!
To avoid caprice, the atheist must leap directly to a Godless universe, without
recourse to the Human-Centred Conditional and in full recognition of the inde-
pendent case for AP. But a leap to atheism instead of AP is a deliberate rejection of
any non-human-centred source of objective values. The atheist insists that, if there is
to be any value at all, then human beings must be its only source. This insistence itself
is self-aggrandizing.
A Benthamite aversion to caprice thus cautions against any direct leap away from
AP. It also operates within specific arguments, because many premises require their
own leaps. For instance, BT responses to evil often need both personal immortality
and incompatibilist freedom. For a defence, these need only be bare possibilities. But
theodicy requires probability. At this point, evidence and argument run out. And AP
objects that to insist that humans are metaphysically special in these ways is a kind of
caprice.
Like any robust hermeneutic of suspicion, the utilitarian aversion to caprice
eventually turns utilitarianism against itself. Does the utility principle itself give
unjustified significance to human well-being? Utilitarians have long questioned the
17
This argument may look like double-counting. Doesn’t caprice enter the equation twice? If our
aversion to caprice has already been factored into our evaluation of the competing arguments for AP and
BT, then why allow additional suspicion when we are choosing our leap of faith? AP has several possible
answers here. One reply is that, if we want to ensure the avoidance of caprice, then we should be wary of it
at every turn. A second reply is that, even if we prefer agnosticism, we still need an understanding of the
alternatives between which we are undecided. This would suffice to motivate the full exploration of the
resources of AP that this book seeks to inaugurate. Finally, while caprice may feature prominently in our
case for religious ambiguity, it need not do so. In particular, perhaps caprice features in the case for AP only
as a negative factor that rules out leaps to BT within the arguments for BT that AP borrows in part I.
 INTRODUCTION
anthropocentrism of privileging human beings over other sentient animals. But AP

goes further. It denies that human beings matter at all from the point of view of the
universe. Utilitarianism thus points to its own transcendence. The shift from utili-
tarianism to a broader consequentialism is a theme of recent work in the utilitarian
tradition. Perhaps AP is just another, rather unexpected, step along that road. One
obvious worry is that this is an abandonment of utilitarianism rather than a reinter-
pretation. This worry, which arose at the end of section 1.4.1 in relation to suffering,
is addressed in part III.
Aversion to caprice plays a significant role throughout this book. While it arises
within the utilitarian tradition, this Benthamite theme has broader resonance, as
most ethical traditions caution against undue partiality or self-aggrandizement.
Many ethics of belief will thus include some element that could play a similar role.
1.4.3 Two pictures of morality

Our ethic of belief complements a picture of morality that emerges from one strand
within the utilitarian tradition. Much of my earlier work dealt with the demanding-
ness objection to consequentialism: the complaint that consequentialism makes
demands that are unreasonable, alienating, and psychologically impossible. Two
contrasting reactions run through the voluminous literature on this topic.
Academic philosophers are well-paid, well-fed, affluent, upper middle-class pro-
fessionals inhabiting a world of great inequality, need, and injustice. Any impartial
moral theory will find our comfortable lives morally impermissible. Some philo-
sophers find this situation deeply troubling; others do not. Some are distinctly uneasy
living a life of luxury while innocent strangers starve; others feel no particular qualms
about this. As a result, some regard demandingness as the central moral issue, while
others dismiss it as a ‘non-problem’.18
These two reactions to the demandingness debate highlight a broader divide
between two pictures of morality, moral philosophy, and human nature, which
I dub the austere and the complacent. On the complacent picture, our everyday
commonsense morality is taken at face value. Our self-image is in order and there is
certainly no need to give it up simply on the say-so of some philosopher. No plausible
moral theory can demand more than we ordinarily expect. The task of moral
philosophy is to explore our self-image, not to upset it.
On the austere picture, by contrast, morality is a source of demanding external
reasons that compete with our various self-interested motivations. Human beings
have a very strong tendency to confuse our self-interest with morality and to
interpret moral norms and situations in ways that make life easy for ourselves. Our
central moral task is thus to strive to be more impartial, to replace our parochial
18
For the former view, see Parfit, On What Matters, vol. 1, p. 501. The latter is harder to find in print,
but has been the reaction of several prominent moral philosophers when I tell them I work on the demands
of consequentialism.
INTRODUCTION 
viewpoints with the point of view of the universe, and to purge our moral thinking of
self-serving delusions. The truth about our moral obligations and our place in the
cosmos may turn out to be very alienating, demanding, or uncomfortable. Moral
philosophy may be radically revisionary.
While the austere picture is especially associated with consequentialism, austere
and complacent interpretations arise within most moral traditions. Consequentialists
can be either extreme or moderate. Kantians can emphasize either Kant’s pessimism
about human nature and radical evil or his optimistic talk of human freedom and
rationality. Rawlsian liberals can extend their egalitarianism across the globe, with
radically unsettling results, or simply refuse to even speak of ‘justice’ outside the
confines of their own affluent nation state. Christians, and other religious ethicists,
can feel called to give all their wealth to the poor; but they can equally well regard
their own affluence and others’ destitution as morally unproblematic signs of God’s
favour.
Our utilitarian ethic of belief gives us solid reason to reject the complacent picture
on the grounds of caprice. It is not hard to see why this picture appeals to affluent
philosophers! In this project, I follow the austere picture. In a Millian spirit, I do not
present the resulting austere morality as the only possibility, but as one credible
moral experiment.
1.4.4 Lessons from the future
My substantive ethical commitments come from my own recent work on our
obligations to future people. Utilitarians have long embraced temporal impartiality.
Human welfare matters equally, no matter when it occurs. Given our potentially
enormous impact on the welfare of future people, it is no surprise that, for the
utilitarian, obligations to distant future people are of central moral concern. (By
contrast, for non-utilitarians the future is usually at most an afterthought.)
In my recent book Ethics for a Broken World, I present another reason why
utilitarians must focus on future people. Traditional ethical thought sets future
people aside, in part because it assumes that they will be better off and that their
interests do not conflict with our own. Faced with threats such as climate change, this
optimism is no longer viable. We must confront the possibility of a broken future,
where resources are insufficient to meet everyone’s basic needs, where a chaotic
climate makes life precarious, where each generation is worse off than the last, and
where our affluent way of life is no longer an option. Given their commitment to
promoting human welfare impartially, utilitarians must take this possibility espe-
cially seriously.
My most recent work imagines different possible futures and asks how future
philosophers might reinterpret our own moral and political philosophy. I argue that
reflection on possible futures transforms our moral thinking in many surprising
ways. I thus draw on another utilitarian inheritance: Mill’s belief in moral progress.
This is not a naïve optimism about social progress, but rather a fallibilist admission
 INTRODUCTION
that future people will know more than ourselves about what is valuable and that
future ethical inquiry might move in very surprising directions. We should be wary of
projecting either optimistic empirical assumptions or controversial philosophical
theories too far into the future. (This utilitarian openness to ethical revision comes
to the fore in our discussions of moral naturalism, extraterrestrial life, and immor-
tality in chapters 2, 7, and 10.)
For the present book, I draw one central lesson from the future: an adequate
intergenerational morality must be founded on objective values. I have come to
believe that, to make sense of our obligations to distant future people, and especially
to give them adequate normative force, we need an objective list theory of well-being,
together with objective values, and a metaphysically robust non-naturalist moral
realism.
A full defence of this controversial claim would require a book of its own.19
I return to meta-ethics briefly in section 1.4.5 and at length in the next chapter. In
this section, I briefly defend objectivism about well-being, value, and reasons. In each
case, the best argument for objectivity is the inability of subjective accounts to do
justice to our obligations to future people. (I return to these topics at greater length in
part III, especially in chapter 12.)
I begin with well-being.20 Contemporary debate contrasts three positions: hedon-
ism (well-being is pleasure and the absence of pain); preference theory (well-being is
getting what you want); and the objective list theory (OLT), which offers a list of
things that are good in themselves irrespective of the agent’s attitude to them, such as
knowledge, achievement, friendship, individuality, self-development, and so on.21
Objectivists argue that neither hedonism nor preference theory is satisfactory.
Some pleasures are good, some bad, others are neutral. Some preferences improve
your life, while others do not. Consider a child who wants to play in the sand rather
than go to school. Clearly, we make his life go better if we send him to school. The
challenge is to explain why. Education doesn’t simply help people to satisfy their
existing preferences. It also teaches them what to desire and which pleasures to seek.
It is important to satisfy people’s desires only because what they value is independ-
ently worthwhile. The objects are not valuable because they are desired—they are
desired because they are valuable.
While debate between these three positions is ongoing, thinking about the future
tips the balance in favour of OLT. Neither hedonism nor preference theory can
capture our intergenerational obligations. Two familiar thought experiments help us
to see why these subjective accounts fail.
19
I briefly defend the link between moral objectivism and intergenerational ethics in ‘Ethics for Possible
Futures’ and ‘What is Good for the Distant Future?’; and I plan to develop it at greater length elsewhere.
20
For overviews of well-being, especially in the utilitarian tradition, see Griffin, Well-being; Parfit,
Reasons and Persons, Appendix I; Mulgan, Understanding Utilitarianism, ch. 3; and the works cited in
Mulgan, ‘Consequentialism’.
21
Parfit, Reasons and Persons, Appendix I.
INTRODUCTION 
Nozick’s Experience Machine. You are offered the chance to plug into an experience machine
that perfectly simulates any possible human experience. Is it in your interests to do so?22
The Virtual Future. In a broken future, people have abandoned the real world altogether and
spend their entire lives plugged into an experience machine. The natural environment is so
polluted and so resource-poor that people have little choice but to dream away their lives with
no direct contact to any reality outside the machine. But this is all anyone has ever known and
they find it perfectly satisfactory. If our present choices lead to this virtual future, have we done
anything wrong?23
Nozick’s experience machine is often read as a decisive refutation of hedonism.

Life in the machine is phenomenologically indistinguishable from the ‘real thing’. If,
as Nozick persuasively argues, it is a mistake to enter the machine, then there must be
more to human flourishing than the quality of one’s experiences. Our negative
reaction to the virtual future supports this critique, as the hedonist must also find
that world unobjectionable.
If Nozick refutes hedonism, our second tale rules out objectivism’s other rival. The
preference theorist cannot capture our unease about the imposition of any virtual
future whose inhabitants are content with their lot. If we only look at individual
preferences, then we cannot see what is wrong with avoiding our obligations to future
people simply by manipulating their psychology or their environment so that they
never want the good things we destroy.
By contrast, the objective list theory easily captures both Nozick’s reaction to the
experience machine and our reaction to the virtual future. If a connection to the
natural world is intrinsically valuable, then human lives go better (and perhaps can
only go well) when they instantiate that value. Some things matter, and it matters that
people are connected to real values, not virtual ones. Even Peter Singer, the most
prominent contemporary defender of preference utilitarianism, has recently acknow-
ledged, on the basis of very similar examples, that we need a more objective account
of well-being to make sense of our obligations to distant future people.24
Unlike Nozick’s original, and countless other fantastical tales, my virtual world is a
credible future. This realization greatly strengthens the objectivist critique of hedon-
ism and preference theory. In debates over well-being, as in many other philosophical
topics, every theory stumbles over some ingenious imaginary case. Defenders of
hedonism or preference theory can thus set aside Nozick’s experience machine. No
theory is perfect, after all.
We cannot demand that a theory of well-being must perfectly fit all our intuitions
about imaginary cases. But we can reasonably insist that moral philosophy provide
useful guidance about important actual decisions. An acceptable theory of well-being
22
Nozick, Anarchy, State, and Utopia, pp. 42–5.
23
I discuss this example in Mulgan, ‘Ethics for Possible Futures’.
24
Singer, Practical Ethics, p. 244. See also de Lazari-Radek and Singer, The Point of View of the Universe,
chs 8 and 9.
 INTRODUCTION
must help us to think clearly about our obligations regarding credible futures,
especially when our present choices might harm future people.
Singer’s own conversion is instructive here. As a practical ethicist, Singer focuses
on first-order moral issues, such as abortion, our treatment of animals, or our
obligations to the distant poor. His shift away from preference utilitarianism is
driven by the failure of his own attempts to apply it to the newly urgent practical
questions posed by climate change. The practical ethicist can sidestep the experience
machine, but not the virtual future.
The further we look into the future, and the more that future might differ from our
affluent present, the harder it is to believe that predictions about what people will (or
might or could) desire have any real moral significance—or to believe that such
predictions provide a solid foundation for morality.
The inadequacy of subjectivism goes deeper. Subjective stories about morality itself
also cannot accommodate the future. Consider contractualism—the main contem-
porary rival to utilitarian intergenerational ethics. Reciprocity, sentiment, and
mutual cooperation may provide good foundations for intra-generational ethics.
But intergenerational contracts face two barriers: Parfit’s non-identity problem and
the impossibility of reciprocal interaction between present people and distant future
people. How can we begin to imagine contracts, bargains, or cooperative schemes
involving future people whose existence and identity depend upon what we decide
and whose fate is entirely in our hands? By contrast, while utilitarians endlessly
debate the precise details of our intergenerational obligations, they have no difficulty
making sense of them. Obligations to future people are theoretically on a par with
obligations to present people: both derive from the fact that our actions impact on the
well-being of sentient beings. This doesn’t prove that utilitarianism is superior all
things considered, but it does significantly enhance its comparative appeal.
The need for objectivity has three further implications. First, once we acknowledge
an objective account of human well-being, it is natural to posit objective values that
are independent of human well-being. If knowledge makes my life go well, irrespect-
ive of my attitude to it, then knowledge must be good in itself. If it is good for me to
achieve X, then X must be something that is independently worth achieving. Second,
to avoid Parfit’s non-identity problem, utilitarians themselves need impersonal
judgements about the comparative value of possible futures. And these comparisons
must also involve independent objective values. Finally, objectivity about well-being
and impersonal values supports externalism about reasons. Internalists (such as
Bernard Williams) insist that I only have a reason to X if X connects to my
current motivations.25 Externalists (such as Parfit) recognize reasons whose force
is independent of my motivations.26 OLT implies external prudential reasons, and
independent impersonal objective values imply external moral reasons.
25
Williams, ‘Internal and External Reasons’. 26
Parfit, On What Matters.
INTRODUCTION 
These brief remarks do not prove that there are objective values or external
reasons. But they do explain why familiar utilitarian priorities commit one to such
values. We urgently need a foundation for our obligations to distant future people,
and objective values are, at present, the only game in town.
Objective list theory, independent impersonal values, and external reasons recur
throughout this book. I defend these commitments at greater length in subsequent
chapters, when they play specific roles in our detailed arguments. I now explore their
general implications for our project.
One theme of this book is that objective value and cosmic purpose are mutually
supporting. Part I explores the many crucial and unexpected roles that objective
values can play in traditional arguments for BT. Here is one example. Objective value
and cosmic purpose allow us to explain a number of puzzling general features of the
cosmos, from the fact that there is something rather than nothing, to the fact that the
universe is governed by precise elegant mathematical laws, to the fact that it is a place
where conscious rational beings can emerge via processes of biological evolution. The
atheist cannot explain any of these facts. He must regard them as brute facts, cosmic
coincidences, just the way things happen to be. Without objective values, this brute
fact response has some plausibility. If there is nothing special about the way things
are, then why not admit that they just happen to be this way? (After all, things had to
be some way.) But if things are an objectively special way, if the possibility that is
realized is an unusually valuable one, then this does cry out for explanation.
Objective values thus support cosmic purpose. For its part, cosmic purpose
supports objective values. For those suspicious of free-standing moral facts, cosmic
purpose offers something to ground mind-independent values. Perhaps moral facts
are facts about cosmic purpose. This brings us to meta-ethics.
1.4.5 Utilitarian meta-ethics

Utilitarians are split on meta-ethics. Utilitarianism is often associated with attempts
to reconcile morality with naturalism—either by identifying moral properties with
natural ones or by offering a non-cognitivist interpretation of moral talk.27 This is a
good fit for hedonists and preference utilitarians, whose accounts of well-being
already look natural. But a second strand, going back to Henry Sidgwick and made
especially prominent recently by Derek Parfit’s On What Matters, combines conse-
quentialist objective value with a commitment to non-natural sui generis moral facts.
I myself favour this consequentialist non-naturalism, as it alone can ground the
objective values that we need to underpin robust obligations to future people. Our
utilitarian ethic of belief lends further support to a non-naturalist meta-ethic. Faced
with meta-ethical uncertainty, utilitarians should favour options that enable them to
27
The connection between utilitarianism and naturalism goes back to Bentham and Mill. Recent
proponents include Jackson, From Metaphysics to Ethics, Hare, Moral Thinking, and Railton, ‘Moral
Realism’.
 INTRODUCTION
make sense of our obligations to future people. Our aversion to caprice will also play
a role, as it prompts a suspicion of the cosy naturalist consensus that pervades
contemporary meta-ethics. (I explore the connections between meta-ethics and AP
at greater length in the next chapter.)
1.4.6 An objective austere morality
The moral commitments outlined in this section form a united whole, whose
foundation is an austere picture of the demandingness and urgency of our obligations
to future people. I claim that morality is objective in several distinct ways. Objective
List Theory says that each person’s well-being is largely a function of her connections
to values that exist independently of her beliefs, wishes, projects, or pleasures.
Objectivism about comparative evaluations says that these independent values
include a ranking of different possible worlds. Externalism about reasons says that
independent objective values provide the moral agent with reasons for action that are
themselves independent of her desires, projects, or beliefs. Non-naturalist moral
realism, explored at greater length in chapter 2, insists that there are facts about
objective values and external reasons. It also claims that those facts are independent
of the agent’s own attitudes, distinct from any human moral practices, and not
reducible to any natural facts.
My central moral claims are these: that my objectivist and externalist substantive
claims are necessary components of any adequate intergenerational morality; that
such a morality will strike human beings as very austere and demanding; and that
only metaphysically robust moral realism can give us the motivation to follow such a
demanding morality. This last claim is primarily psychological rather than logical or
semantic or metaphysical. In philosophical debates about objectivism, its subjectivist
opponents claim to borrow or mirror all the advantages of objectivism without its
metaphysical extravagances. As it happens, I do not find subjectivism semantically
credible. I regard objectivism as a much more natural interpretation of our everyday
talk about well-being, value, and morality. (I defended this claim briefly in section
1.4.4, and I return to it in chapter 2.) But my central objection is motivational.
Pleasures, preferences, internal reasons, and expressions of one’s attitudes may prove
sufficient to motivate commonsense morality’s relatively undemanding obligations
among contemporaries. But they cannot provide the impetus to sustain the sacrifices
that utilitarianism demands in the face of a possibly broken future. For the hedonist,
preference theorist, internalist, non-cognitivist, or naturalist, the sacrifices demanded
by utilitarian morality must always seem imprudent, irrational, and extreme.
It is fair to criticize a rival metaphysical picture for failing to motivate people to
save the world? Here, once again, we encounter reasonable disagreement. Secular
liberals, naturalist philosophers, and moderate non-consequentialists invariably find
this sort of criticism absurd. But many religious people—and many utilitarians—will
find it both pertinent and decisive. My own position lies somewhere in-between.
Perhaps no world view can motivate the moral demands of utilitarianism. Perhaps
INTRODUCTION 
this is too much to ask. But if one competing world view can ground, explain, and
motivate the morality our future seems to demand, then this should count very
strongly in favour of both that world view and the morality it supports. Nor is such
support out of the question. After all, BT often does motivate extreme self-sacrifice in
the service of its transcendent good, and AP can reasonably hope to borrow this
motivational strength. At the very least, it is worth asking whether or not it can.
For the remainder of this book, I shall assume that morality is very objective,
austere, and demanding, and ask what the world must be like if it is.
1.5 The Moral Implications of AP

The shift to non-human-centred cosmic purpose opens up new theoretical possibil-
ities. On the one hand, AP differs significantly from BT. Theodicy and religious
orthodoxy bring many commitments, both metaphysical and moral, which constrain
the development of BT ethics. Freed from these constraints, AP can use the reality of
cosmic purpose to ground normative theories hitherto neglected in BT ethics. On the
other hand, the addition of cosmic purpose to an otherwise atheist world view gives
moral philosophy new resources—new sources of information about what is valu-
able. If the universe exists because it is good, then learning how the universe is may
tell us what is good. Another possible new source of moral insight is mystical
experience, explored in chapter 5. In the final pages of his 1984 classic Reasons and
Persons, Derek Parfit famously argued that non-religious ethics is in its infancy and it
is not irrational to have high hopes for its future.28 If so, shouldn’t non-religious
ethics be open to all possible sources of ethical insight—including those that arise
within religion itself?
Any attempt to ground human morality in non-human-centred cosmic purpose
faces obvious difficulties. The central challenge is to avoid human moral nihilism. If
objective value is linked to a cosmic purpose that is indifferent to us, how does
anything in our lives get any value at all? I address this challenge in part III. A second
challenge is to defend human knowledge of morality. I explore this question at length
in chapter 5. Existing moral resources (both BT and atheist) must be reimagined so
that they offer genuine insight into the cosmic purpose, but seen through a human-
tinted lens.
A final challenge is more existential. How do we bring a non-human-centred
cosmic purpose into our lives? Why does the metaphysical dispute between atheism,
BT, and AP matter? Suppose you shift from materialist atheism to cosmic purpose.
Would that make any practical difference? (Or suppose you convert from BT to
AP. Is this any different from converting to atheism? Can the new cosmic purpose
play any of the moral roles that God once played? Can there be a non-human-centred
28
Parfit, Reasons and Persons, p. 454.
 INTRODUCTION
religion?) These large questions will occupy us in part III, where I draw on a variety of
models from both secular and religious ethics. Some roles for cosmic purpose are
straightforward. If an accurate knowledge of one’s place in the world is a component
of a flourishing human life—as many atheists and theists alike agree—then we have
good reason to explore any cosmic purpose. More ambitiously, cosmic purpose
might offer a model of a valuable, creative life. Perhaps human lives can resemble
value, even if they do not possess it. We can respond, perhaps in more or less valuable
ways, to cosmic values that transcend us.
Any AP human morality will be strange, unfamiliar, austere, and demanding—
more so than its atheist or theist competitors. But BT ethics also seems alien and
extreme to secular eyes. And any moral realism is too demanding for the nihilist!
This is only an objection to AP if we assume that morality should be comforting and
familiar. But utilitarians, of all people, have no right to assume that.
1.6 Plan of the Book

My commitment to Normative AP structures this book. Our first task is to explore
the moral motivation for AP, which sets the scene for all subsequent arguments.
Accordingly, chapter 2 sets out and defends non-naturalist moral objectivism. It also
develops a moral argument that we need cosmic purpose to ground objective values
and external reasons. The chapter’s key lesson is that AP and BT both agree that God
(or cosmic purpose) has an ineliminable normative dimension. AP is therefore
normative rather than morally agnostic, and it is closer to BT than to atheism.
Chapters 3 and 4 evaluate two very familiar arguments for BT: cosmological and
teleological. I argue that AP can borrow the most plausible contemporary formula-
tions of those arguments. In chapter 3, we learn more about both AP’s normativity
and its content. AP either presupposes or delivers comparative objective evaluations
of possible states of affairs, and these include the judgement that the actual world is
better than the empty world. We also encounter Leibniz’s idea that God must create
the best and therefore that this is the best of all possible worlds. AP is not wedded to
Leibniz’s maximalism. But it does breathe new life into it, for reasons that emerge
more fully in chapters 7 and 8.
Chapter 4 surveys a range of teleological arguments, and concentrates on three
prime candidates for AP to borrow. The most important of these is the fine-tuning
argument, which starts from the observation that a range of cosmological constants
appear to be remarkably fine-tuned to facilitate the emergence of life within the
physical universe. In this chapter we learn that cosmic purpose involves general
features of the cosmos, such as its friendliness to life, its governance by mathematical
laws, and the fact that it can be understand by some of its inhabitants. These content-
based lessons further reinforce Normative AP. Unless we agree that these features are
objectively significant, we cannot hope to ground any successful fine-tuning argu-
ment for cosmic purpose.
INTRODUCTION 
Chapters 3 and 4 also teach us that some empirical questions are more significant
to cosmic purpose (and, therefore, to human morality) than they might initially
appear. Moral philosophers can no longer ignore cosmology! We also see that AP
favours some empirical hypotheses over others, a lesson that is reinforced in
chapter 7.
Cosmological and teleological arguments are explanatory. God is posited as an
essential component of the best explanation of empirical facts about the existence
of the physical universe or its friendliness to the natural emergence of life.
Chapter 5 introduces a different style of argument, based on William Alston’s
defence of the internal rationality of Christian mystical doxastic practices.
I contend that many philosophers make out-of-date assumptions about the
nature of mystical experience, and this leads them to underestimate its moral
significance. I then argue that AP has much to learn from BT mysticism.
Chapter 5 also expands our knowledge of the content of cosmic purpose and
the role of human experience in both morality and mysticism. We learn that the
cosmic purpose is characterized by unity, transcendence, and non-self-
centredness. Mystical experience seems to be non-moral. However, I argue that
either mystical experience is moral experience (because moral facts are super-
natural facts), or it only constitutes reliable evidence for cosmic purpose because
it is linked to human moral improvement or insight. Any successful argument
from mysticism must yield Normative AP rather than Simple AP. (Chapter 5
leaves one objection to mysticism unanswered: the threat of religious diversity.
We pick up that challenge in chapter 9.)
Chapter 6 deals with the third classic BT argument: Anselm’s notorious onto-
logical argument. I argue that, read in its original religious and philosophical context,
Anselm’s Proslogion has many resources that the a-contextual readings of contem-
porary analytic philosophy tend to miss. This chapter explores the source of cosmic
purpose, arguing that AP should posit a perfect being who gives the universe its
cosmic purpose. This perfect being could, in principle, be either personal or imper-
sonal. For convenience, I follow BT and speak of a perfect God. We learn that AP can
legitimately borrow perfect being theology and the philosophers’ Omni-God, and
that all talk about a perfect being is ineliminably normative. Chapter 6 concludes that
any successful ontological argument must link perfection to objective values via
human mystical experience. (Chapters 5 and 6 thus support each other.)
In part II, we turn to arguments against a human-centred cosmic purpose. We
suppose that the arguments of part I deliver a cosmic purpose, and we now seek
further information about that purpose. Chapter 7 investigates a comparatively
unfamiliar objection to BT, albeit one based on a very familiar intuition. The
argument from scale objects that the vast universe discovered by science, where
human beings play such a marginal role, is out of kilter with what we would expect
from a God who was interested in us. After rejecting several unsatisfactory formu-
lations, I conclude that the most plausible argument from scale concerns, not the size
 INTRODUCTION
of the cosmos per se, but rather the number, diversity, and sophistication of its
inhabitants. Simply put: humans are too puny to be the centrepiece of so vast a
cosmic canvas.
Chapter 7 also contains general lessons about God’s creative reasons. I argue that
the arguments of part I suggest a God who has impersonal consequentialist motiv-
ations. God responds to the comparative values of possible worlds and creates the
best world. This divine consequentialism constrains BT’s interpretation of God’s
benevolence. If God cares for finite creatures at all, God must create the best
creatures. Finally, chapter 7 teaches us some surprising things about the cosmic
significance of extraterrestrial life. AP expects God to create beings who are much
better than humans and rejects any presumption that we are alone.
Chapter 7 concludes that, while God would care about superior beings, we have no
good reason to believe that God cares for us. Chapters 8 to 10 complete the case for
AP by arguing that we have reason to believe that God does not care for us. The main
focus in these chapters is on controversial metaphysical claims that BT needs and AP
avoids (such as libertarian freedom and personal immortality), and on BT’s inability
to adequately explain facts about evil and religious diversity. My arguments are
often disjunctive. AP offers several related criticisms of BT and need only argue
that at least one succeeds.
Chapter 8 examines the most popular atheist objection to BT: the argument from
evil. Unsurprisingly, this is where our utilitarian commitments come to the fore.
I present two arguments from evil that AP can borrow—based on animal suffering
and the horrendous evils that humans inflict on one another. Much of the chapter
is a sustained rebuttal of theodicies built around human freedom. I argue that BT
must make very ambitious claims about actual human freedom—claims that AP
can reasonably reject. The main lesson of chapter 8 is negative: God does not care
about individual human beings at all. However, we also learn that AP seeks to
remain agnostic regarding two related claims: that some things are good for
humans and that human well-being has some objective value. We return to those
topics in chapter 12.
Chapter 9 picks up where chapter 5 finished, asking whether the fact of religious
diversity refutes BT. A separate chapter on religious diversity may seem redundant.
Isn’t this just another minor example of evil? (If BT can explain why God permits
horrendous evil or vast animal suffering, surely it won’t be troubled by a few religious
disagreements?) I argue instead that religious diversity is a distinct challenge for BT,
because it represents an unequal distribution of what BT itself regards as the central
human good: right relationship to God. Chapter 9 concludes that AP can borrow
atheist objections based on religious diversity, although we must then modify
Chapter 5’s claims about the significance of mysticism. AP can only borrow insights
from competing mystical traditions when they overlap. Because mystics agree about
abstract moral ideals rather than specific metaphysical or religious doctrines, this
reinforces Chapter 5’s emphasis on the moral dimension of mysticism.
INTRODUCTION 
Arguments from scale, evil, and diversity are explanatory. The objection is that BT
cannot explain the size of the universe, the presence of evil, or the diversity of
religions. In each case, we compare three explanations: BT, AP, and atheist. AP
must demonstrate its superiority over both its rivals. AP’s case against atheism unites
explanatory arguments from parts I and II. Although atheism seems to explain
specific facts taken in isolation, it cannot explain the broader context. Evolution
explains why animals suffer, but not why the universe is a place where sentient
animals evolve in the first place; naturalist sociology may explain why religions
diverge, but not why there is any religion in the first place; and so on. When we
take all the evidence together, only AP offers a satisfactory explanation.
Chapter 10 takes up the question of life after death. Most BT religions believe in
some afterlife, and post-mortem justice is often a key component of BT theodicy. It
may seem obvious that AP can simply treat the afterlife as an unnecessary and
implausible metaphysical commitment peculiar to BT. Unfortunately, things are
not so simple. One perennial BT argument is that immortality is necessary for
morality. Unless we survive death, the demands of morality are irrational, pointless,
incoherent. Atheists who defend undemanding moral theories can easily dismiss this
argument. AP cannot. Its austere demanding morality may well need something akin
to immortality. My first task in chapter 10 is to tease out the metaphysical require-
ments of both BT and morality. I must then demonstrate both that the former
outstrip the latter and that AP can reject BT’s metaphysical extravagances while
still satisfying the demands of morality. A key move here is to increase the meta-
physical demands of BT. I argue that BT must posit both an afterlife for all sentient
animals and a cycle of rebirth. Chapter 10 draws freely on neglected traditions within
BT philosophy, especially British Idealism. It thus illustrates the importance of
philosophical pluralism in our case for AP.
Part III is shorter and more speculative. Suppose we are convinced that AP is true.
How might this impact on our lives? Chapter 11 presents an imaginary dialogue
featuring seven converts to AP, each persuaded by a different argument in parts
I and II. This serves to informally illustrate the possibilities that AP opens up.
Chapter 12 addresses the central worry for any AP ethic: can AP recognize the value
of human well-being? I argue that it can, in a surprising variety of ways. Chapter 13
explores the impact of AP on moral theory, especially within the consequentialist
tradition. I have two complementary aims in part III. First, one overarching theme
throughout the book is that AP mutually supports the package of moral commitments
outlined in section 1.4. Parts I and II show how these commitments support AP. Part
III must then show how AP supports them. But my particular route to AP is not the
only possible route. Non-consequentialists could endorse AP. Part III thus also asks
how moral theorists who don’t share my particular consequentialist orientation might
respond to the discovery of a non-human-centred cosmic purpose.
I cannot provide a full AP moral theory in this book. AP faces serious objections,
but so do both its rivals. (BT cannot explain evil and materialism cannot explain
 INTRODUCTION
morality at all.) As Parfit said, these are early days for non-religious ethics, and we
should not dismiss new possibilities out of hand. And I do hope to show that it would
make some difference to our moral practice, our view of the moral universe, and our
moral theory, if we were to shift from the familiar choice between a universe without
meaning and a universe where humans matter to the less self-aggrandizing thought
that, while it is about something, the universe is not about us.
2
Meta-ethics
Good is indefinable not for the reasons offered by Moore’s successors, but
because of the infinite difficulty of the task of apprehending a magnetic but
inexhaustible reality.
[Iris Murdoch, The Sovereignty of Good, p. 42]
If all realities of intrinsic value were absent from the world, would it then be
intrinsically worthwhile to believe that they were absent? Self-evidently not—
from which it follows that there’d be no real point in believing it, in any sense of
‘real point’ that was truly worth anything. . . . Many philosophers of today seem to
think this would be a satisfactory situation. It strikes me as a profoundly
depressing one. I recognize that, were it our actual situation, then nothing
intrinsically worthwhile could be obtained through being depressed by it. But
belief in realities of objective value is at the heart of my own eagerness to continue
living.
[John Leslie, Infinite Minds, p. 47]
A comprehensive survey of contemporary meta-ethics is beyond the scope of this

book, and the competence of its author. I do not aspire to completeness or neutrality.
In particular, I dismiss some very popular views quite peremptorily. This is because
I believe they owe their popularity, not to their prima facie semantic plausibility, but
to their compatibility with a naturalist world view.
This chapter plays two roles. First, it introduces my own meta-ethical views. In
particular, I defend non-naturalist moral semantics against non-cognitivism and
moral naturalism, and moral realism against moral nihilism; and I challenge the
implicit presumption of global naturalism that underpins contemporary meta-ethics.
Dissatisfaction with the state of contemporary meta-ethics is one motivation for
this book. While I do not expect all readers to share all my meta-ethical prejudices,
I do hope to establish the need to explore alternatives to the current consensus—and
thus to motivate the chapters that follow.
The chapter also sketches a moral argument for AP. Or, rather, an argument for
BT that AP can borrow. There is a very old argument from morality to God. Atheists
can talk the moral talk. They can have sincere moral beliefs and be morally good
people. But, metaphysically, moral truth and moral knowledge are only possible
because of God. I think there is something to this argument. If objective value
 META - ETHICS
makes no sense in a purely materialist universe, then perhaps we must move beyond
materialism. Under BT or AP, value is no longer an isolated anomaly. It is built into
the very fabric of the cosmos. This need not commit us to God, let alone to any
specific religion. But it does get us surprisingly close to cosmic purpose.
This moral argument is not sufficient on its own. Nor is it essential to AP’s case
against atheism. One could find the arguments of part I persuasive without agreeing
that we need cosmic purpose to explain objective values. (Indeed, as we’ll see, the
arguments of chapters 3 and 4 might be most appealing to those who can counten-
ance independent non-natural values.) Nonetheless, a moral argument can form part
of the cumulative case for AP.
2.1 Two Approaches to Meta-ethics

As I explained in chapter 1, my starting point is normative ethics, not metaphysics.
I begin with the moral commitments set out in section 1.4, especially regarding
religious ambiguity and objective values, and then ask what meta-ethical position we
need to make sense of those commitments. This is not the standard approach among
contemporary meta-ethicists, who typically start with some already privileged meta-
physical picture of the world, and then ask where (if at all) our moral talk might fit
into that picture.
Most meta-ethicists begin with a presumption of global naturalism, where ‘the
only kinds of things whose existence we ought to countenance are things that fit into
a unified scientific framework’.1 Contemporary meta-ethics is thus driven by a
dilemma that Frank Jackson dubs the ‘location problem’.2 When we say that things
are good or bad, right or wrong, forbidden or permissible, we make claims that go
beyond a mere description of human practices. Our moral talk seems to demand
moral facts. But where, in the naturalistic world, are moral facts to be found? And
where do they get their authority over the agent and her desires?
The global naturalist has only three broad meta-ethical options:
1. Moral statements do not express propositions. (Non-cognitivism)
2. Moral statements express propositions that are reducible to propositions about
natural properties. (Moral naturalism)
3. Moral statements are all false. (Moral nihilism)
For those who wish to avoid moral nihilism, the only alternatives are non-
cognitivism and moral naturalism. But both are semantically implausible. It is
prima facie obvious that, when we use moral language, we do mean to make claims
that can be true or false, and we do not take ourselves to be talking about purely
1
Joyce, The Evolution of Morality, p. 145. Other definitions of naturalism are explored in chapters 3, 4,
and 5.
2
Jackson, From Metaphysics to Ethics, pp. 113–38.
META - ETHICS 
natural properties. The argument from global naturalism to moral nihilism looks
compelling.
Most of the energy in contemporary analytic meta-ethics goes into squaring this
circle—rehabilitating non-cognitivism or moral naturalism. Much ingenious work
has been done, and continues to be done, but the underlying semantic inadequacy
remains.
I shall argue that all three global naturalist options are untenable, given the plausible
normative commitments set out in the introduction. While these options can seem
adequate for relations between contemporaries (where non-moral motivations bolster
moral ones), none captures the urgency of our ethical talk about the distant future. To
make sense of our moral talk, we must embrace moral non-naturalism. Moral talk
refers to moral facts that are not natural facts. In contemporary philosophy, non-
natural moral realism is a minority position, but not an idiosyncratic one.3 I depart
from my fellow normative ethicists—especially my fellow consequentialists—over the
ontological implications of moral realism. In particular, I believe that atheist non-
naturalism has clear deficiencies that the introduction of God or cosmic purpose can
significantly reduce.
2.2 Against Non-cognitivism

Contemporary meta-ethics is dominated by the dispute between cognitivism and
non-cognitivism.4 Cognitivism holds that moral statements are truth-apt. They
express propositions about states of affairs. (Cognitivism then divides between
moral realists, who hold that some moral statements are true; and moral nihilists,
who deny this.) Non-cognitivism holds that moral statements are not truth-apt, and
do not express propositions.
Every philosophical endeavour has some presuppositions. One presupposition of
my project is the prima facie superiority of cognitivism. Our first task in this chapter
is to justify this assumption.
The primary motivation for non-cognitivism is not semantic, but ontological. By
denying that moral talk is truth-apt, non-cognitivism neatly sidesteps all the epi-
stemic and ontological worries that plague cognitivism. There are no moral facts, and
therefore Jackson’s location problem is dissolved. Semantics aside, non-cognitivism
is impeccable.
If one begins with global naturalism, then non-cognitivism is a serious contender.
However, I begin with substantive moral commitments, and a recognition of religious
3
My defence of non-naturalism in this chapter is especially indebted to Huemer, Ethical Intuitionism;
Parfit, On What Matters, vol. 2, pp. 263–620; Scanlon, What We Owe to Each Other; and Nagel, Mind and
Cosmos.
4
For an overview of the contemporary debate, see van Roojen, ‘Moral Cognitivism vs Non-
Cognitivism’.
 META - ETHICS
ambiguity. I seek a plausible account of the meaning of moral talk. From this perspec-
tive, semantic considerations have priority over ontological ones. The semantic debate
between cognitivists and non-cognitivists is empirical. Their competing claims con-
cern the current function of moral talk in actual human languages. And the evidence
points against non-cognitivism. Huemer puts the point well.
The non-cognitivist makes a claim about actual meanings, but all of the reasonably direct and
objective evidence, from the way we use ethical language, points towards cognitivism. . . .
Evaluative statements act in every way like factual claims.5
For a taste of the difficulties here, consider how the non-cognitivist might analyse the
following utterances:6
• If pleasure is good, then chocolate is good.
• It is better to achieve something worthwhile than to fulfil your desires.
• I wonder whether I did the right thing.
• If murder is wrong, then I did the right thing.
• Did you try to do the right thing?
• It is true that pain is bad.
• Is euthanasia wrong?
• Do the right thing!
• Some things are valuable.
• Morality demands great sacrifices on behalf of future generations.
• If God commands X, then X is required.
• If God does not exist, then nothing is good.
No doubt some non-cognitivists can accommodate some of these linguistic phenom-
ena. But it is hard to see how any non-cognitivist could account for them all as easily
as the cognitivist can. (As we’ll see, non-cognitivists are especially troubled by moral
objectivity, a difficulty they share with moral naturalists.)
The semantic evidence places the non-cognitivist on the back foot. The non-
cognitivist saves moral talk, but only by changing the subject. Such reinterpretation
is necessarily a second-best option. But without a prior commitment to global
naturalism, we have no need to settle for second best. Given its semantic inadequa-
cies, non-cognitivism is best seen as revisionary rather than descriptive. It is not an
account of what moral talk has meant, but rather instead a suggested reinterpretation
whereby moral talk remains useful despite the non-existence of moral facts. (Just as
religious talk might be radically reinterpreted by someone who no longer believes in
the existence of God.) Revisionary non-cognitivism is a form of moral nihilism, and
not an alternative to it. (We return to moral nihilism in section 2.4.)
The unspoken atheism of contemporary analytic philosophy is nowhere more
evident than here. Many non-cognitivists reject moral facts as logically or
5 6
Huemer, Ethical Intuitionism, p. 25. Adapted from Huemer, Ethical Intuitionism, p. 23.
META - ETHICS 
metaphysically incoherent. Indeed, the obvious impossibility of moral facts is often

seen as semantic evidence for non-cognitivism. Any informed person can see that
there could be no moral facts. So no sensible moral talker could ever mean to refer to
such facts. As an argument about actual human moral talk, this is unconvincing—for
the simple reason that most human moral talkers are not global naturalists. In
particular, many are theists. If there is a God, then there might well be moral facts.
The incoherence of moral facts is not obvious, unless the non-existence of God is also
obvious. In a religiously ambiguous universe, moral facts are not especially anomal-
ous. Non-cognitivism has a place in atheist apologetics. But as an ontologically
neutral account of moral semantics, it is a non-starter.
Two features of our project count against non-cognitivism. The first is our
rejection of its presumption of global naturalism. The second is our commitment
to moral objectivism and an austere demanding morality. Moral talk is puzzling
because it combines factual and motivational elements. Moral statements both
express our moral beliefs and motivate us to act. Non-cognitivism claims to capture
the motivational aspect of our moral talk better than cognitivism, because it connects
a person’s moral utterances directly to her motivations. This makes sense if we
presuppose subjectivism about morality and internalism about reasons. But what if
the most urgent moral talk explicitly aims to lead us beyond any actual human
motivation?
In chapter 1, I introduced a package of moral commitments with several distinct
objective elements: an objective list theory of well-being, mind-independent com-
parative evaluations of possible states of affairs, external reasons, and a very
demanding consequentialist moral theory. By contrast, non-cognitivism seems
more at home with a subjective complacent morality, where the link between my
present attitudes and my moral obligations is straightforward and there is no need
for moral motivations beyond my actual current desires. In that undemanding
context, perhaps non-cognitivists can talk the moral talk (albeit somewhat artifi-
cially). But when it comes to expressing the demands of the future, non-cognitivism
comes up short. Non-cognitivist paraphrases may suffice if we only want to use
moral propositions as premises in abstract reasoning. But they lack the external
normative push that this project’s substantive moral commitments demand. Or so
it seems to me.
My primary complaint is not that non-cognitivism cannot construct intersubjective
standards to evaluate the correctness of moral utterances, but that it cannot do justice
to the existence of mind-independent intrinsically motivating objective values and the
external reasons they generate. This motivational deficiency goes hand in hand with
non-cognitivism’s ontological narrowness. Contemporary non-cognitivism claims to
capture (or mimic) the objectivity of realist moral talk. As I suggested earlier in this
section, this seems implausible if we include the everyday moral talk of theist moral
realists. But even if non-cognitivism could mimic the logical and semantic features of
realist moral talk, it would surely lack the urgent normative push provided by real
 META - ETHICS
external moral facts—especially the moral facts available to theists. (We return to the
motivational advantages of theist ethics in sections 2.6 and 2.7, in chapter 5, and in
part III where I ask whether AP can borrow those advantages.)
Once its opponents include the theist positions outlined in sections 2.6 and 2.7,
non-cognitivism’s inadequacies are clearly exposed. They are much less apparent in
non-cognitivism’s traditional rivalry with moral naturalism—precisely because the
latter encounters very similar problems. Semantic objectivity is a challenge peculiar to
non-cognitivism. But substantive objectivity is equally troubling for moral naturalism.
While it can provide objective criteria of moral correctness, moral naturalism struggles
to accommodate objective values. We return to that struggle in section 2.3.2.
2.3 Against Moral Naturalism

The moral naturalist says that moral facts are natural facts. As with non-cognitivism,
the primary motivation is ontological. Moral naturalism appeals to those who wish to
combine moral realism with global naturalism.7 A successful reduction to natural
property talk would be epistemically and metaphysically unproblematic. (Or, at least,
it would inherit only those epistemic and metaphysical difficulties that affect natur-
alism in general.) Unfortunately, moral naturalism fails to capture the real meaning
of moral talk. As with non-cognitivism, the problem is semantic.
The modern semantic case against moral naturalism goes back to Moore’s infamous
Open Question Argument (OQA). There is much controversy surrounding the details
of Moore’s argument, and whether it succeeds. Despite its shortcomings, I believe there
is a compelling objection behind Moore’s OQA. What follows is an attempt to develop
that underlying thought, not an exegesis of Moore or his many successors.8
Contemporary moral naturalism comes in two varieties. The distinction goes by
several names: semantic, analytic, or definitional naturalism is contrasted with
metaphysical, non-analytic, synthetic, or ontological naturalism. To avoid confusion,
I shall call the two positions definitional and non-definitional moral naturalism.
2.3.1 Definitional moral naturalism

The definitional moral naturalist proposes a natural analysis of the meaning of moral
talk. Moore’s OQA offers the following objection. Let ‘N’ be the natural analysis
offered for some moral term such as ‘good’. If the analysis were correct, then the
following question would not arise: ‘I know this particular thing is N, but is it good?’
7
Despite their affinities, moral naturalism and global naturalism are distinct. Non-cognitivists and
moral nihilists are also typically global naturalists: all facts are natural, but there are no moral facts. And the
moral naturalist can agree that some non-moral facts are not natural (such as mathematical, logical, or even
supernatural facts).
8
For a variety of recent verdicts on the OQA, see, e.g., Huemer, Ethical Intuitionism, pp. 67–72; Pigden,
‘Identifying Goodness’; Joyce, The Evolution of Morality, pp. 151–4; Fisher, ‘God, Good and the Open
Question Argument’.
META - ETHICS 
If ‘good’ means ‘N’, then no native speaker would ask this question. (Just as no one
who understands ‘unmarried man’ and ‘bachelor’ can wonder whether this particular
unmarried man is a bachelor.) But this question is open. It always makes perfect
sense, given any natural description of an action or event, to ask whether it is good.
This shows, says Moore, that no natural definition captures the meaning of ‘good’.
Moore’s OQA is semantic, and therefore empirical. The definitional moral natur-
alist makes claims about the current role of moral talk, which Moore seeks to rebut.
Moore appeals to his own linguistic intuitions and those of his readers. While this
does make the OQA ‘intuitionist’ in one sense, it is not intuitionist in a more familiar
(and much more controversial) sense. The OQA appeals to linguistic intuitions, not
moral ones. Its conclusion is not that there are moral facts, or that human beings do
have intuitive access to those facts. While Moore himself did hold these further
positions, the OQA itself merely concludes that moral terms do not designate natural
properties. (We return to ethical intuitionism in section 2.5.)
The OQA is on firm ground against any specific reductive analysis. Consider a
utilitarian claim that ‘right’ means ‘maximizes human welfare’.9 Native speakers can
see that ‘right’ and ‘maximizes human welfare’ are not synonymous. It is absurd to
accuse non-utilitarians of misusing moral language, or failing to speak English
properly. And it is surely significant that the OQA has succeeded against every
other extant reductive analysis.
But is Moore entitled to the more ambitious claim that no natural definition is
possible? Can’t the naturalist simply reply that a successful analysis will be forthcom-
ing in the future? Why not treat the OQA as a spur to further refinement rather than
a knockdown objection to an entire research programme?
A more ambitious OQA argument could simply be inductive. Every past natural
analysis has failed the open question test. Therefore, every future analysis will fail.
Given the number of analyses philosophers have produced, this induction would
carry some weight. (It is certainly more compelling now than it was in 1903.) But, as
with any induction from past failure, it is not decisive. Fortunately, there is a stronger
Moorean argument available, based on a direct appeal to linguistic intuition. When
native speakers reflect on the failures of particular analyses, as revealed by the
original OQA, they can see that no such analysis could ever succeed. The OQA
helps us to realize that moral talk is not a shorthand for some very complex natural
talk, but instead refers to distinct moral properties that cannot be analysed in
natural terms. As we shall soon see, this doesn’t show that moral properties are
not natural properties. But it does strongly suggest that moral terms are not syn-
onymous with natural terms. Like non-cognitivism, definitional moral naturalism is
a second-best option. And it is too soon in our exploration of meta-ethics to settle for
9
Utilitarians need not adopt this definitional claim. There are non-naturalist utilitarians. (Indeed, this
book is proof of that!) There is always a distinction between the meta-ethical claim that rightness is X, and
the substantive normative ethical claim that whatever is X is right.
 META - ETHICS
second best. (In addition, the arguments deployed against non-definitional natural-
ism in the next section would also defeat definitional naturalism.)
2.3.2 Non-definitional moral naturalism

Many moral naturalists agree that the OQA defeats definitional naturalism. Moral
talk is not analytically reducible to natural talk. Moral terms and natural terms are
not synonymous. But this does not mean that moral properties are not, in fact,
identical to natural properties.
According to non-definitional moral naturalism, the identity between a moral fact
and a natural fact is discovered a posteriori. The model is the analysis of natural kind
terms offered by Kripke and Putnam.10 Water is H2O. This is not a definitional truth.
‘Water’ does not mean the same as ‘H2O’. Similarly, ‘good’ does not mean the same as
any natural property term. Moore’s open questions are not semantic questions, but
substantive questions in chemistry or moral philosophy. To discover that water is
H2O, we examine the substance water—not the word ‘water’. Similarly, Richard Boyd,
a prominent recent proponent of non-definitional moral naturalism, advocates the
following procedure for moral terms.11 We first list the features associated with our
idea of goodness, rightness, or any other basic moral property. ‘Good’ refers to
whatever property actually has these features. Assuming that all properties are
natural, we then examine the natural world, and (hopefully) find the property that
has those features. Call it N. We conclude that ‘good’ refers to N. Like many moral
naturalists, Boyd is a utilitarian. He conjectures that, if we were to carry this project to
fruition, we would find that goodness is maximum human well-being.
The most obvious weakness of the analogy between water and goodness is that the
identity with H2O has already been established beyond reasonable chemical doubt, as
have many other scientific accounts of natural kinds. There is no moral equivalent.
Non-definitional moral naturalists gesture at such accounts, and some claim to have
produced them. But no one has actually produced an identity that has anything like
the authority of ‘water is H2O’.
Early chemists wondered whether water was H2O. But this question is no longer
scientifically open. No informed person could now wonder whether this sample of
normal water is H2O. Anyone who does wonder merely reveals his own ignorance.
No moral question has ever been closed like this. There is no putative moral identity
whose questioning would cast doubt on the speaker’s basic competence—whether
linguistic or empirical. ‘How can you possibly doubt that this action which maxi-
mizes human well-being is right? Haven’t you read Mill/Hare/Boyd/Railton/Jackson?
Don't you know anything?’ Unlike the equivalent response in chemistry, this is just
intellectual bullying.
10
Kripke, Naming and Necessity; Putnam, ‘The Meaning of “Meaning” ’.
11
Boyd, ‘How to be a Moral Realist’; Boyd, ‘Finite Beings, Finite Goods’. (Boyd calls his position
‘ontological’ moral naturalism.)
META - ETHICS 
Robert Adams uses this disanalogy to develop a semantic objection to Boyd.

Adams appeals to the following ethical idea:
The critical stance: ‘The stance amounts to at least this. For any natural, empirically identifiable
property or type of action that we or others may regard as good or bad, right or wrong, we are
committed to leave it always open in principle to raise evaluative or normative questions by
asking whether that property or action-type is really good or right, or to issue an evaluative or
normative challenge by denying that it is really good or right.’12
For Adams, this stance is an essential feature of commonsense moral thought, and it
‘captures the real truth behind Moore’s open question argument’.13 To treat any
human consensus as beyond criticism is a form of idolatry. Naturalistic ethics cannot
accommodate the critical stance, while BT ethics can. This gives BT a decisive
advantage.
Suppose Boyd’s empirical enquiry (or some other non-definitional naturalism)
succeeds. We have identified N—the natural property that is goodness. Doubting that
N is good is now on a par with doubting that water is H2O. Our currently limited
moral knowledge means that we do not yet have to abandon the critical stance. But
should empirical ethical enquiry succeed, we would have to abandon it.
Adams replies that we would never abandon the critical stance. Even at the limits
of empirical enquiry, normative questions still remain open in a way that chemical
questions do not. Even if we knew everything there is to know about the natural
world, we could still ask whether or not something that was N was also good.
Goodness must always outrun our knowledge of the natural world. This is part of
the very idea of goodness.
Adams’s argument assumes that the moral naturalist must claim that, while we
don’t currently know which natural property is identical to which moral property,
rational inquiry will (or at least could) discover those identities in future. Moral
naturalists may be tempted to simply deny this. Unfortunately, they cannot. And this
commitment to future convergence leads to deeper problems for moral naturalism.
Adams’s specific objection only scratches the surface.
As moral realists, moral naturalists must recognize a gap between moral facts and
our current moral beliefs. We could be wrong about what is right. But moral naturalism
cannot recognize a gap between moral facts and all human moral beliefs. If ethical
inquiry proceeds as it should, it will eventually converge on a set of moral beliefs—what
Jackson dubs ‘mature folk morality’.14 And these beliefs give us moral facts.
12
Adams, Finite and Infinite Goods, p. 78. The debate continues in Boyd, ‘Finite Beings, Finite Goods’;
and Adams, ‘Anti-Consequentialism and the Transcendence of the Good’.
13
Adams, Finite and Infinite Goods, p. 78.
14
Jackson, From Metaphysics to Ethics, p. 140. (Strictly speaking, Jackson’s own ‘analytical naturalism’
is a form or definitional moral naturalism. But his notion of mature folk morality is one that the non-
definitional moral naturalist can borrow. And, as I argue in the text, every moral naturalist needs
something like this.)
 META - ETHICS
This convergence may never occur. Human beings might become extinct tomor-
row, or slip into a state of barbarism from which they never recover. We cannot
identify moral facts with actual future moral beliefs. Moral naturalism has an
ineliminable counterfactual element. But counterfactuals are notoriously problem-
atic. Can we be confident that moral inquiry would yield convergence? Can we hope
to predict where it might converge?
Moral naturalism optimistically assumes that future inquiry would converge—and
then identifies moral facts with the results of that convergence. Unfortunately, when
they defend this optimism, moral naturalists help themselves to several unwarranted
simplifying assumptions. The most obvious is a presumption of global naturalism.15
It is taken for granted that future inquiry will confirm the truth of naturalism, rather
than BT or Platonism or some as yet unimagined alternative. Most striking, for our
purposes, is the presumption of atheism. Mature folk morality captures the moral
beliefs of a fully informed society of confirmed atheists. The possibility that future
inquiry might converge on belief in God, and then converge on some BT meta-ethic,
is not considered.
In our present project, of course, this presumption is unfounded. Future enquiry
might not converge on global naturalism. It might converge on BT or AP. Perhaps
rational inquirers will eventually see that divine creation and cosmic purpose are the
only way to explain our cosmos (chapters 3 and 4). Or perhaps, at a certain stage of
rational inquiry or human history, God will (necessarily) unambiguously reveal the
cosmic purpose. Or perhaps future moral philosophers will take mystical experience
more seriously as a source of moral knowledge, in ways that affect both their
metaphysical views and their substantive moral philosophy (chapter 5). If rational
metaphysical inquiry converges on BT or AP, and future enquirers then identify
moral facts with facts about God, then Boyd’s own method undermines moral
naturalism.
A deeper worry is that, if our universe is genuinely religiously ambiguous, then
ethical enquiry might fail to converge at all. Metaphysical debate shows no sign of
convergence, and metaphysics influences ethics. Of course, the moral naturalist is
primarily interested in moral inquiry, not metaphysics. So she could insist that future
inquiry will still reach a consensus on ethics, even if it remains forever agnostic about
metaphysics. The rest of this book seeks to undermine this agnosticism, arguing that
moral inquiry cannot remain untouched by cosmic metaphysics.
A second unacceptable simplification is that moral naturalists often make their
task easier by assuming naturalist-friendly substantive moral views, especially in
relation to human well-being, prudential reasons, and the foundations of morality.
The identity between moral facts and natural facts is much closer if the former are
limited to agents’ desires or mental states. It is no coincidence that moral naturalists
15
This presumption is often implicit, but it is clearly present in the work of most contemporary moral
naturalists.
META - ETHICS 
tend to be hedonists or preference utilitarians. By contrast, a commitment to

objective list theory, or to independent and impersonal objective values, sits uneasily
with moral naturalism. Can the objective value of knowledge or achievement ever be
given a purely natural explanation? (Naturalists have great difficulty explaining why
an agent who does not desire achievement nonetheless has a reason to pursue it.)
The naturalist may seek to remain agnostic, and leave the details of human well-
being to be determined by future moral inquiry. (One reason for distinguishing
mature folk morality from current folk morality is to avoid awkward questions like
this!) But even admitting the possibility of an objective list theory is problematic,
because it greatly reduces the likelihood of convergence. Could the dispute between
hedonism, preference theory, and objective list theory ever be eliminated by any
conceivable empirical inquiry? Could any experiment settle whether knowledge is
intrinsically good even for those who don’t desire it, or tell us how to balance sensual
pleasure against accurate perception of the cosmic purpose?
If we are already committed to global naturalism, then it makes sense to limit our
attention to moral theories based on desire or pleasure. And this is one reason why
subjectivism is so popular. But once we acknowledge religious ambiguity, this narrow
focus is untenable.
A third oversimplification of moral naturalism is that, in common with most
contemporary moral and political philosophy, it operates with a very naïve picture of
the possible future of moral inquiry. In Ethics for a Broken World, I argue that many
of our most cherished moral ideas will need to be radically rethought in a broken
future. In other recent work, I argue that other possible futures will be similarly
disruptive. By relying on convergence across different possible futures, the moral
naturalist implicitly denies this. But, to my knowledge, moral naturalists never
adequately defend this conservatism.
These difficulties surrounding convergence support a semantic argument against
non-definitional moral naturalism. The non-definitional moral naturalist must iden-
tify moral facts with the end points of some process of empirical inquiry. Goodness is
whatever future inquirers would identify. By bringing out the complexities buried in
this counterfactual, we see that the identification fails to capture our central ideas
about goodness itself. Whether or not Moorean questions remain open, this is not a
plausible way to close them.
Moral naturalism is often defended by appeal to supervenience.16 Moral properties
are said to supervene on natural ones. (Very roughly: there can be no change in moral
properties without some accompanying change in natural properties; any two
16
See, e.g., Jackson, From Metaphysics to Ethics, pp. 118–25. Jackson uses supervenience to defend
definitional moral naturalism. Supervenience is a platitude, and therefore must be built into the meaning of
moral terms. This ambitious argument is especially prone to the objections raised in the text. Even if moral
properties turn out to be identical to natural properties, it is hard to see how anyone who took religious
ambiguity seriously would accept that this identity is built into the meaning of moral terms.
 META - ETHICS
possible worlds that are naturally identical must also be morally identical.) Indeed,
this claim is presented as an a priori truth, an essential feature of commonsense
ethical thought. Supervenience is one of the ‘moral platitudes’ that any meta-ethic
must respect. But if the moral supervenes on the natural, then moral properties must
be natural properties. And, if we know this, then we must intend our moral talk to
refer to those natural properties.
Rather than embarking on an extended discussion of supervenience, I limit myself
to two remarks. First, and most important, the supervenience of the moral on the
natural is not a platitude for the theist. Suppose I believe I have made a promise to
God. My moral obligations then depend, at least in part, on whether or not God
exists. More broadly, no one could know a priori that the wrongness of X is entirely
independent of God’s attitude to X. BT and AP both insist that, if there are platitudes
about moral supervenience, these cover only the supervenience of the moral on the
descriptive—where the latter includes both natural and supernatural properties.17 But
this supervenience does nothing to support moral naturalism over the moral super-
naturalism discussed in section 2.7.
Second, many non-naturalists accept supervenience, but deny that it shows that
moral properties are natural properties. As with the (alleged) supervenience of the
mental on the physical, one respectable philosophical option is to insist that there are
still two distinct sets of properties here, even if they necessarily co-vary. (Supernat-
ural properties provide also illustrate this possibility. Consider the following plausible
tale. If God’s existence and nature are necessary, and if God’s commands flow from
the essential divine nature, then God would only issue different commands if there
were some natural difference. Supernatural facts about God’s commands would then
supervene on natural properties. But it does not follow that the property ‘com-
manded by God’ is some natural property.18) If non-naturalism commits us to the
possibility of necessarily coexistent distinct properties, then we may be happy to pay
this metaphysical price—especially if there is no other way to salvage moral realism.
I conclude that non-definitional moral naturalism is another second-best option.
It has great difficulty capturing the pluralism and objectivity inherent in our moral
talk. To the diehard global naturalist, second best is as good as it gets. But less
dogmatic thinkers will want to continue their search.
2.4 Against Moral Nihilism

Dissatisfaction with both non-cognitivism and moral naturalism is not unusual.
A popular alternative is semantic non-naturalism. Moral talk is about moral facts,
but it is not about natural facts. Instead, moral talk aims at sui generis non-natural
properties. Once we accept semantic non-naturalism, we must choose between moral
17
Jackson himself concedes this. He defends analytical descriptivism rather than analytical naturalism.
18
Plantinga, ‘Naturalism, Theism, Obligation and Supervenience’, p. 269.
META - ETHICS 
non-naturalism and moral nihilism. Do such properties actually exist, or is moral talk
simply false?
One key question is where the burden of proof lies. Philosophers divide sharply
here, and the choice of default is driven by background philosophical commitments.
For the partisan of global naturalism, the very idea of non-natural moral properties is
so absurd that non-naturalism can be dismissed out of hand. It is this perceived
absurdity that drives so many global naturalists into non-cognitivism and moral
naturalism, despite their glaring semantic inadequacies. For many partisans of
everyday moral talk, on the other hand, moral nihilism is so obviously repellent
that even to entertain it would be, to borrow Anscombe’s colourful phrase, an
indication of a corrupt mind. If moral realism contradicts global naturalism, then
global naturalism must go.
As we began with everyday moral talk, not with global naturalism, our default
option is non-naturalism. In this section, I briefly set out the case against moral
nihilism. Unlike non-cognitivism and moral naturalism, moral nihilism is not
semantically objectionable. It is a second-best option for quite different reasons.
Nihilism requires us to abandon too much of our current view of the world. Imagine
what one’s life would be like if one truly embraced nihilism. One worry is that moral
nihilism threatens to collapse into global normative nihilism, where there are no
normative reasons of any kind—whether moral, prudential, or epistemic. If nothing
is good or bad, right or wrong, better or worse, appropriate or inappropriate, then
nothing can be good or bad for me, and I can have no reason to believe anything.
Some nihilists downplay the potential radicalness of their view, because they
believe that adequate naturalist alternatives are available. In a nihilist world, we
still have desires, hopes, plans, emotions, social practices, and the like. Nihilism
removes one special class of reasons (‘moral reasons’), rather than all reasons. It
eliminates reasons that might constrain one’s desires, but leaves intact one’s reason to
pursue those desires. But this separation is hard to defend. Isn’t a reason to do what
I want just as ontologically suspect as a reason to obey the moral law? One theme of
recent normative ethics is that it is a substantive moral claim that one has any reason
to follow one’s desires.19 A crazy desire may provide no reason at all.
Like both moral naturalists and non-cognitivists, moral nihilists often help them-
selves to controversial substantive moral views that make their positions seem less
extreme. It is much easier to separate moral and prudential reasons, and to retain the
latter while rejecting the former, if one’s account of well-being is purely subjective.
An objective story about well-being makes this separation much trickier. If my life
only goes well if I pursue independently valuable projects, then the removal of facts
about what is valuable also undermines my prudential reasons. A prudent agent must
recognize independent objective values. The committed global normative nihilist
19
See, especially, Scanlon, What We Owe to Each Other, ch. 3.
 META - ETHICS
cannot be prudent. If a good human life is one that balances prudential, moral, and
epistemic reasons, then any attempt to reduce all reasons to desires is doomed to fail.
Joyce ends his exploration of the practical implications of nihilism with the
following rhetorical flourish:
If uncomfortable truths are out there, we should seek them and face them like intellectual
adults, rather than eschewing open-minded inquiry or fabricating philosophical theories
whose only virtue is the promise of providing the soothing news that all our heartfelt beliefs
are true.20
I find this thought congenial. (After all, the theme of this chapter is that contempor-
ary meta-ethics is driven by a reluctance to consider possibilities that global natur-
alists might find uncomfortable.) However, if global normative nihilism is true, then
Joyce’s statement is false. (Or, at least, his use of ‘should’ is merely a shorthand for
some normatively inert personal preference.) If there are no reasons, then there is
nothing we ought to do or believe. Nihilists sometimes present their position as
admirable. Perhaps it is. But only if it is false. If nihilism is true, then nothing is
admirable, and no one has any reason to believe anything. Nihilists cannot consist-
ently criticize their opponents on moral or other reason-based grounds. If nihilism is
true, then I violate no legitimate epistemic or moral norms by simply refusing to
believe it.
Of course, none of this shows that nihilism is false. But it does highlight the extent
to which it is a second-best option. This motivates, yet again, the search for a viable
form of moral realism. As the primary motivation for moral nihilism is dissatisfac-
tion with the existing options, we must explore new alternatives.
On the other hand, I also believe that normative ethicists should take the nihilist
threat more seriously than they often do. Although their position is very counter-
intuitive, nihilists do present a forceful objection to non-naturalism (the only seman-
tically credible form of moral realism available to global naturalists). I address that
objection in section 2.5.
AP has another reason to take nihilism seriously. As well as offering new resources
to avoid nihilism, AP may also support a new kind of nihilism. My main aim in this
book is to construct a human morality based on AP. But this conciliatory strategy
may seem to underestimate the radicalness of AP. Shouldn’t the shift to a non-
human-centred cosmic purpose lead us to abandon human morality altogether?
Shouldn’t AP embrace nihilism, rather than rejecting it?
AP opens up a new nihilist position, where cosmic values (linked to cosmic
purpose) fail to ground any recognizable human morality. There are facts about
objective value, but the only moral fact about anything relating to human beings is
that it has no value. I call this position cosmic realist nihilism. I argue in part III that
cosmic value does ground a credible (if unfamiliar) human morality. I therefore reject
20
Joyce, The Evolution of Morality, p. 230.
META - ETHICS 
cosmic realist nihilism. However, this position is worthy of further exploration.

While its practical implications for us are nihilistic, its moral ontology is closer to
an ambitious moral realism. Cosmic realist nihilism could thus borrow many of part
I’s claims about cosmic purpose and cosmic value. Here is one example. In chapter 3,
I suggest that we may need to introduce cosmic values to explain why there is
something rather than nothing, or why there is a well-ordered or life-friendly
physical universe rather than a chaotic lifeless one. One could accept this argument
while denying that human morality plays any explanatory role.
Cosmic realist nihilism might also avoid the collapse into global normative
nihilism. Nothing about us matters, and therefore there are no facts about what
is good for us, or about how we ought to treat one another. But some things do
matter—namely the cosmic values. So perhaps there are things that ought to be done,
believed, or be—such that any creature able to respond to cosmic values ought to do
so. This would not yield any recognizable human morality, but it might save us from
the abyss of global normative nihilism. (As I argue in part III, AP also provides the
materials for a new kind of moral fictionalism, where we construct a fictional human
morality on a foundation of real cosmic values.)
Once AP enters the scene, the moral nihilist no longer needs to be an atheist, or to
reject cosmic values or cosmic purpose. Of course, most actual moral nihilists are
confirmed atheists. Indeed, most are global naturalists. Within contemporary ana-
lytic philosophy, where the alternative to atheism is BT, this is hardly surprising. But
nihilists need not be naturalists. AP opens up a broader, less metaphysically conten-
tious, disjunctive argument for moral nihilism—one that precludes BT, but is
otherwise metaphysically open.
I return to cosmic realist nihilism again in part III. Until then, however, I set it
aside, and retain my presumption in favour of moral realism.
2.5 Non-naturalism
Recent moral philosophy has seen a striking revival of non-naturalism: moral
statements express propositions about sui generis non-natural moral properties;
some of these propositions are true; and we can (and do) know them. Non-
naturalism is often combined with ethical intuitionism: we know non-natural
moral facts, because we see that they are true. There is a clear link between the
revival of non-naturalism in meta-ethics and the reinvigoration of substantive moral
philosophy. Non-cognitivism and moral naturalism flourished at a time when
philosophers virtually ignored actual moral talk. (And they still flourish primarily
among philosophers who do not do normative ethics.) When actual moral talkers
returned in force to philosophy, and looked for a meta-ethic to make sense of their
substantive discussions, they found the extant options (non-cognitivism, moral
naturalism, moral nihilism) clearly inadequate. This is just not what we have been
talking about.
 META - ETHICS
Non-naturalism captures how things seem to many of us—at least before our
moral phenomenology has been corrupted by philosophical scepticism or global
naturalism. Ethical intuitionism is the natural default position. Moral talk seems to
be truth-apt, to refer to sui generis normative facts, and to deliver moral knowledge.
Why not take things as they seem?
My project is sympathetic to non-naturalism. However, atheist non-naturalism is
open to (at least) four significant objections. In this section, I briefly outline those
objections, together with the most obvious non-naturalist replies. In many cases,
these replies are very plausible, especially given my commitments in this book. (In
particular, non-naturalism complements religious ambiguity, methodological plur-
alism, and objectivism about well-being.) However, I also bring out some limitations
of the replies available to the atheist. These are places where BT or AP might offer
non-naturalism a more robust defence.
The first objection is that moral properties are too weird to admit into any
respectable ontology.21 This is a very common objection, but it is quite hard to pin
down. For the global naturalist, of course, any non-natural property is weird. But
‘weird’ cannot simply be a synonym for ‘non-natural’. And it is surprisingly hard to
find a neutral perspective from which to judge whether non-natural moral properties
are weird. Consider the following list of some of the things that exist: time, space,
numbers, propositions, substances, properties, relationships, logical facts, evidential
connections, mental states, physical states, aesthetic properties, gravitational fields,
the past, dispositional properties, moral properties.22 Each item is weird relative to
the others. But so what? Why not just accept that the world contains many distinct
kinds of things?23
It is often objected that moral properties are weird because they are intrinsically
motivating. But other items on our list are also intrinsically motivating. Logical facts,
evidential connections, and aesthetic properties all equally call forth certain
responses from (normally functioning) rational beings. Intrinsically motivating
facts appear weird to the internalist, for whom all reasons must connect to actual
motives. But non-naturalists can reply that moral facts generate external reasons that
agents should follow. The argument from weirdness thus rejects all external reasons.
Yet, without external reasons, we must abandon the objective list theory, along with
independent values. But the intuitive plausibility (and theoretical utility) of the latter
make it hard to insist that the former are too weird to exist.
I personally find this robust non-naturalist reply compelling. Once we question
global naturalism, the argument from weirdness has little independent force. We
21
This objection is famously associated with J. L. Mackie, who gave it the memorable but now
distracting title: ‘the argument from queerness’.
22
Adapted from Huemer, Ethical Intuitionism, p. 200.
23
In a similar vein, Colin McGinn suggests that, as philosophically interesting phenomena go, meaning
and consciousness are prima facie more odd than morality. (McGinn, Ethics, Evil, and Fiction, p. 19 and
p. 58, respectively.) But few moral nihilists are nihilists about either of these phenomena.
META - ETHICS 
need more than an incredulous stare to undo moral properties. But non-naturalism
does seem precarious if morality is our only departure from global naturalism; and
opponents of non-naturalism argue that all these other alleged non-natural items can
be cashed out naturalistically. This is one place where, as we shall see, BT can bolster
non-naturalism. (Such bolstering is especially important if our substantive morality
is very demanding, because the discovery of non-natural moral facts must then
provide sufficient motivation to overcome the agent’s non-moral inclinations.)
The argument from weirdness is often paired with an epistemic objection to non-
natural moral facts. Even if we grant their ontological coherence, such facts must lie
beyond the ken of finite physical creatures such as ourselves. How can our cognitive
faculties be influenced by non-natural facts? Without such influence, how can we
have moral knowledge? (And, if we have no access to moral facts, how can the non-
naturalist assert their existence?)
The opponent of non-naturalism must argue that moral knowledge fails some
general epistemic test that other human knowledge passes. (Otherwise, the epistemic
objection collapses into global scepticism.) Causal accounts of knowledge are espe-
cially popular here. A knows that F only if A is somehow causally affected by the fact
that F. If all knowledge is causal, then we cannot know non-causal facts. But non-
natural moral facts must be non-causal. (If they could enter into causal interaction
with physical creatures, they would be natural facts.)
The non-naturalist could argue that moral knowledge does satisfy her opponent’s
epistemic criterion: moral facts do causally affect human beings. While it finds little
favour among atheist non-naturalists, this response is very congenial to BT, as we
shall see in section 2.6.24 However, most atheist non-naturalists instead reject their
opponent’s epistemic criterion. Moral knowledge is not causal, but it is none the
worse for that. Here, non-naturalists often seek partners in crime. Moral knowledge
would be anomalous if it were our only non-causal knowledge. But much other
familiar knowledge is also non-causal. Especially popular here is knowledge dear to
the hearts of global naturalists—such as logic or mathematics.25 Some non-
naturalists do take the metaphor of sight literally, and posit a perceptual (or quasi-
conceptual) moral sense. But others interpret ‘see’ only in the same loose sense in
which we ‘see’ that 2 plus 2 equals 4, or P and If P then Q together entail Q. Moral
intuition is a form of rational intuition.
24
Causal access to non-natural facts is much less problematic if one is a dualist about human beings. (If
we ourselves are non-physical souls, then it is no harder for us to interact with non-natural moral facts than
with physical entities.) This is another place where BT assists non-naturalism, because there is a natural
(though not inevitable) affinity between dualism and BT. This connection between dualism and BT recurs
several times in later chapters. It is considered most fully in chapter 10, where I argue that, while AP can
accept some forms of dualism, it rejects any dualism that is sufficiently metaphysically ambitious to
support BT’s claims about the afterlife.
25
See, e.g., Parfit, On What Matters, vol. 2, pp. 488–98. Some extreme empiricists do reject mathem-
atical knowledge on causal grounds. But most non-naturalists would be happy to show that morality is
epistemically no worse than mathematics!
 META - ETHICS
If morality is on a par with logic and maths, it is on safe ground. But is there
parity here? Opponents of non-naturalism cite disanalogies. One is moral dis-
agreement. Non-naturalists reply that reasonable disagreement is found in all
respectable areas of knowledge, and should be no more troubling in ethics than
elsewhere. This reply is strengthened by our commitment to religious ambiguity,
and philosophical ambiguity more broadly. A second alleged disanalogy is the
essential role that maths and logic play in scientific reason. Mathematical objects
must be admitted to our ontology, because they are indispensible. Moral facts,
however, are dispensable. Non-naturalists reply that non-natural facts are equally
indispensable for science. Science deals in reasons to believe, which are just as
non-natural as moral reasons to act. The scientist must admit reasons, and then
moral reasons are merely more of the same. (This echoes our earlier argument
that moral nihilism leads to global normative nihilism.) Our general commit-
ments to methodological pluralism and religious ambiguity do support the non-
naturalist here. But, once again, BT can provide additional support, as we shall
see in section 2.6.
A final disanalogy is the existence of established expert practices in mathematics
and logic—with clear criteria for membership and esoteric bodies of knowledge.
Ethical intuitionists reply that, for moral facts, the relevant experts are ordinary
people. In chapter 5, we explore a more radical BT reply: that the true moral experts
are mystics!
Our third objection to non-naturalism is closely related. Opponents object that
non-natural moral facts play no explanatory role. Ockham’s razor thus tells us to
reject them. This objection is often deployed within a causal epistemology. Non-
causal beliefs such as mathematics may earn their keep by featuring in our best
explanations, while moral facts do not.
Here, BT offers non-naturalism its most ambitious reply: that we need moral facts
for other explanations. This reply is a central theme of this book. Without it, the
atheist non-naturalist has only two weaker options. First, she can deny that respect-
able beliefs need play any explanatory role. Perhaps moral facts play an essential role
in deliberation, without explaining anything. (We explore this Kantian theme, in
relation to immortality, in chapter 10.) Second, the atheist can claim that moral facts
are essential to explain our moral beliefs and practices. However, this second reply
presumes that we lack a naturalistic explanation of those beliefs and practices.
This brings us to our final objection to non-naturalism. Unlike our other objec-
tions, this one does not rely on either the presumption of global naturalism, or the
rejection of methodological pluralism. I believe it is the most troubling objection to
non-naturalism, and therefore I address it at greater length.
Recent popular culture abounds with debunking explanations of human morality,
drawn from psychoanalysis, evolution, Nietzchean genealogy, anthropological rela-
tivism, or economic models of motivation. Each offers some natural history of our
moral beliefs that is designed to undermine their epistemic pretensions.
META - ETHICS 
Many deflationary accounts of ethics are crude and unpersuasive. But not all are so
easily dismissed. I will focus on one recent evolutionary explanation sketched by
Richard Joyce.26 Joyce offers a possible evolutionary history, where the tendency to
make moral judgements conferred an evolutionary advantage on our pre-human
ancestors. Morality is socially advantageous. Groups of moralizers outcompete
otherwise similar groups. Moralizers have this advantage, not only relative to simple
egoists, but also relative to creatures who are disposed to cooperate but who lack the
capacity to make moral judgements. Morality gets its distinctive adaptive edge
because it sometimes produces extreme self-sacrifice. Moralizers will lay down
their lives for the group, or punish offenders even when it is not in their own interests
to do so. Mere inclination alone could not motivate these sacrifices. If human beings
are to reliably make moral sacrifices, they must believe that the social norms involved
are not merely social norms—that they have the ‘practical clout’ that only morality
can possess.27 I will make the moral sacrifice only if I truly believe that I am obliged to
do so. In an evolutionary contest between groups with different attitudes to morality,
moral realists will out-compete moral anti-realists.
This collective advantage explains why evolution might favour moralizing crea-
tures. The next question is how evolution might give rise to such creatures in the first
place. (If they don’t arise, then such creatures cannot be selected for.) How might
moral beliefs arise? Joyce cites empirical evidence that, he argues, supports a projec-
tivist account.28 Moral beliefs arise when human beings project their emotions onto
the world. As Joyce notes, a projectivist account of any phenomenon is self-
consciously counter-intuitive.29 Its central claim is that things are not as they appear
to be—that our beliefs do not correspond to the underlying reality that causes them.
Colour is not out there in the world, and neither is morality. For our purposes, the
crucial point about this story is that, although Joyce’s moral beliefs share the charac-
teristics emphasized by non-naturalists in their semantic battles with non-cognitivists
and moral naturalists, they arise without any input from moral facts. Joyce thus offers a
deflationary explanation, one that undermines belief in moral facts.
Joyce also argues that moral nihilism can avoid global normative nihilism, because
there is a crucial disanalogy between moral beliefs and other beliefs. We cannot hope to
explain the adaptive advantage of perceptual, mathematical, or scientific beliefs with-
out citing their truth. False perceptual beliefs confer no evolutionary benefits. Unlike
morality, these other beliefs only confer evolutionary advantage when they are true.30
26 27
Joyce, The Evolution of Morality. Joyce, The Evolution of Morality, p. 57.
28
Joyce, The Evolution of Morality, p. 12. The classic projectivist, in this sense, is David Hume.
29
30
Joyce, The Evolution of Morality, p. 183. One prominent BT reply is Plantinga’s ‘Evolutionary
Argument Against Naturalism’ (EAAN), which claims that evolution cannot explain why we have true
empirical beliefs after all. (See, e.g., Plantinga, Warrant and Proper Function, pp. 216–37.) A full treatment
of Plantinga’s EAAN would take us too far afield; and, as I note briefly in chapter 4, it is not clear how much
(if at all) AP can borrow from this much-discussed argument.
 META - ETHICS
This is an intriguing tale. But is it true? Joyce claims that, while not yet fully
proven, his evolutionary tale is no mere ‘just so story’. It has significant empirical
support, and therefore deserves to be taken seriously.31 In particular, Joyce claims
that his tale enjoys far better empirical support than any alternative that might
support non-naturalism.
The non-naturalist could simply reject Joyce’s tale. The theoretical possibility
of a deflationary explanation is insufficient. We need an actual deflationary
explanation—a true (or at least plausible) debunking natural history. Until this is
produced, non-naturalists can simply hold their ground. And many non-naturalists
do precisely this.32
Perhaps, at this stage of the enquiry, this is not an implausible move. As Joyce
himself admits (and as any sensible evolutionary ethicist must concede), we do not
actually possess an evolutionary explanation that is complete even in its own terms.
However, this response leaves worrying hostages to fortune. Non-naturalism is on
stronger ground if it can offer some principled reason why we should not expect a
satisfactory evolutionary explanation to be forthcoming in the future. And it is here,
in particular, that the BT non-naturalist is much better equipped than her atheist
colleague.
Atheist non-naturalism is semantically plausible, and preserves moral talk against
the nihilist challenge. But it faces serious difficulties. These motivate our exploration
of non-atheist alternatives.
2.6 Benevolent Theist Non-naturalism

The rehabilitation of non-naturalism is only one aspect of the reaction against global
naturalism. A second recent trend, further from the analytic mainstream but no
longer beyond the pale, is the revival of explicitly theist meta-ethical positions. No
longer content to be silenced by the mere incantation of the name of Euthyphro, BT
meta-ethicists regard God as central to a proper understanding of both moral facts
and human knowledge of them.
BT supports moral realism in two ways. First, it offers a new meta-ethical position
(moral supernaturalism), addressed in section 2.7. Second, and less ambitiously, it
bolsters the existing case for moral non-naturalism. To see why, we reassess our four
objections to non-naturalism, asking what BT can add to the non-naturalist’s
strongest replies.
We begin with the argument from weirdness. While the atheist can reject this
objection, BT offers a much more robust reply. The main ontological objection to
moral properties is their non-naturalness. But, if God exists, there are infinitely many
31
32
Huemer, Ethical Intuitionism, pp. 214–15; Parfit, On What Matters, vol. 2, pp. 511–42; McGinn,
Ethics, Evil, and Fiction, pp. 55–7; Nagel, Mind and Cosmos.
META - ETHICS 
non-natural facts. And the idea of intrinsically motivating facts is familiar to BT, as
divine facts themselves look like excellent candidates. ‘Every normal human being
who believes that God loves X is motivated to do X’ seems at least as compelling as
‘Every normal human being who believes that X is right is motivated to do X’. In a BT
universe—in a world created by a perfect being—non-natural moral facts are not that
surprising. And we should also not be surprised that their discovery can motivate
otherwise unpalatable moral demands.
BT also answers other puzzling questions. The non-naturalist says that many
things in the universe have sui generis moral properties. We might wonder why
this is so. BT can reply that God created this possible world because of its moral
properties, so it is no wonder that it has those properties. (This is the core of the
cosmological and fine-tuning arguments, discussed in chapters 3 and 4.) BT could
even explain the origin of moral facts. Perhaps God makes the moral facts.
BT thus provides ontological support for non-naturalism. And, as we shall see in
part I, that support is mutual. Once we admit non-natural moral facts, it is much
harder to accept either the existence of the universe, or its nature, as brute facts.
BT is perhaps most helpful in moral epistemology. If there are objective values,
and if it is good to know such values, then a benevolent God will ensure that rational
creatures have knowledge of moral facts. A priori, we have no idea how God might do
this. But we do know that, if God is both omnipotent and benevolent, then God can
and will find a way. We also know that we are inclined to treat our moral beliefs,
intuitions, and practices as prima facie reliable. Perhaps this is how God gives us
moral knowledge. BT thus transforms our knowledge of non-natural moral facts
from something utterly mysterious to something completely expected. (The main
question is whether AP can borrow this explanation. We return to that question in
section 2.9.) BT also provides a distinctive model for moral knowledge. As I argue in
chapter 5, mysticism offers both an exemplar of non-natural knowledge, and a rich
source of moral knowledge.
Once again, this epistemic support between BT and non-naturalism is mutual. The
link between moral and epistemic normativity is a double-edged sword for the atheist
non-naturalist. The more we emphasize the normativity of science, and its connec-
tion to knowledge, the more pressing is the challenge to provide a purely naturalistic
account of that normativity. Are we really sure that we have a respectably naturalistic
explanation of how (and why) science itself works? Or must we invoke God to
explain the normativity of science? (We discuss these questions in chapters 3, 4,
and 5.)
This BT epistemic story also offers a very striking reply to the charge of explana-
tory redundancy. The central theme of part I is that moral facts are essential to
explain the existence and nature of our universe. You can’t get much less redundant
than that!
Finally, BT supports the key premise in the non-naturalist reply to Joyce: the claim
that moral beliefs and practices cannot be explained naturalistically. Instead of
 META - ETHICS
merely criticizing naturalist explanations, BT non-naturalists can offer their own

superior explanations. This reply to Joyce has several strands. First, drawing on the
arguments offered in part I, BT will cite facts that evolution alone cannot explain; that
cannot be given any naturalist explanation; and that can be adequately explained by
appeal to God. This opens the door to non-evolutionary explanations. Second, while
some of these non-evolutionary BT explanations involve very general features of the
universe, others are human phenomena (distinct from morality itself) that have no
purely natural explanation—such as our ability to gain insight into the nature of the
cosmos, or our mystical experiences. The latter are especially significant. In chapter 5,
I argue there that it may be impossible to construct a deflationary naturalistic
explanation of mystical experience. If that argument goes through, then it provides
a model for a non-evolutionary explanation of human morality, as well as under-
mining the presumption that we will eventually find some (deflationary) explanation
of morality. BT can also link morality and mysticism more directly, arguing that any
satisfactory evolutionary account of morality must incorporate an explanation of
mystical experience—and that the latter is likely to forever remain unavailable.
The BT non-naturalist need not reject Joyce’s tale. She can accept it, but then deny
that it undermines the epistemic credentials of our moral beliefs. Even if it were filled
out perfectly in its own terms, Joyce’s tale would never offer a complete explanation of
human moral beliefs. Evolution is only part of the story. Joyce’s evolutionary story
ceases to be deflationary in a BT world. BT can interpret evolution as God’s way to
give human beings moral beliefs. God’s benevolence then rules out the possibility
that those beliefs are systematically mistaken.
Non-naturalism is thus much more plausible within a BT world view. Moreover, if
BT is true, then non-naturalism is no longer our only meta-ethical option, as we shall
now see.
2.7 Moral Supernaturalism

The implicit presumption of global naturalism even infects the terminology of its
opponents. ‘Non-naturalism’ ought to cover any view where moral facts are not
natural facts. But it is reserved for the specific claim that moral facts are sui generis.
This ignores the possibility that moral facts might be identical to some other set of
facts. In particular, perhaps moral facts are facts about God. Rather than redefining
well-established terms, I shall retain the standard use of ‘non-natural’, and dub this
alternative anti-naturalist position ‘moral supernaturalism’.
Moral supernaturalism has clear ontological and epistemic advantages over non-
naturalism. It offers simple explanations of both moral facts and our knowledge of
them. Moral facts simply are facts about God. If there is a God, then moral
properties are not metaphysically weird. And, being benevolent, God obviously
wants us to know these important facts. This is the most straightforward BT story
about morality.
META - ETHICS 
Moral supernaturalism only works if there is a God. For the atheist, moral
supernaturalism implies moral nihilism. And atheism is the main reason why most
contemporary meta-ethicists ignore moral supernaturalism. (Indeed, moral super-
naturalism is most often invoked by moral nihilists, who defend a supernaturalist
moral semantics, and then use God’s non-existence to establish nihilism.) But our
project takes the possibility of God seriously. If the best form of moral realism
requires a benevolent God, then this is an argument for BT, not for moral nihilism.
(How strong that argument is, and whether AP can borrow it, are questions to which
we return.)
To succeed where moral naturalism failed, moral supernaturalism must dissolve
the OQA. Philosophical discussion of God and morality has long begun with Plato’s
Euthyphro dilemma. If God exists, then God loves (all and only) good things. But
what is the relationship between the fact that God loves x and the fact that x is good?
Is the good good because God loves it, or does God love the good because it is good?
This is a dilemma because both alternatives are problematic. Voluntarists say the
good is good because God loves it. Morality depends upon God’s will. But that makes
morality arbitrary. If torture is only wrong because God forbids it, then torture would
not be wrong if God had not forbidden it. Indeed, if God had commanded torture,
then not torturing would be wrong. Furthermore, God’s moral perfection ceases to be
a separate, distinct attribute of God, and becomes instead a trivial corollary of God’s
creative power. ‘God is morally perfect’ simply means that God follows God’s will.
On the other side, intellectualists say that God loves the good because it is good.
Moral standards are independent of God, and God recognizes those standards using
God’s intellect. Intellectualism avoids arbitrariness. If God is necessarily morally
perfect, and if the independent standards condemn torture, then God could not
possibly have commanded torture. But intellectualism threatens to render God
morally redundant: at best a moral guide to be imitated, rather than a moral legislator
to be obeyed.
Many atheists regard the Euthyphro dilemma as fatal for any theist ethic.33 In
reply, BT has two broad alternatives. The first is to argue that God still plays a vital
moral role, even under intellectualism. (This is BT non-naturalism, discussed in
section 2.6.) The second broad option is to identify moral facts with some divine
facts. Some defend pure voluntarism. (Recent years have seen a resurgence of divine
command theory within analytic philosophy.) Others seek a middle road between
voluntarism and intellectualism, where morality is neither independent of God, nor
dependent on God’s arbitrary will, but depends instead on something else about God.
I return to these varieties of BT ethics in part III. Here, I illustrate them using Robert
33
In conversation, when asked why they don’t even mention the possibility that there might be a God,
contemporary meta-ethicists often reply that ‘the Euthyphro dilemma shows that God is irrelevant to
ethics’.
 META - ETHICS
Adams’s meta-ethic, which has the merit (for comparative purposes) of being very
similar to Boyd’s moral naturalism.
Adams links morality to God’s being or nature.34 He develops a broadly Platonic
account of goodness, where God replaces the Platonic Forms. Things are good
because, and in so far as, they resemble God. Moral standards are not independent
of God. Without God to resemble, nothing could be good. So God is necessary to
make moral facts true, because moral facts are facts about God.
Euthyphro raises a Moorean open question challenge for moral supernaturalists
like Adams. Even if we know that X resembles God, or that God has commanded Y, it
always makes sense to ask whether X is good or whether Y is the right thing to do.
And surely ‘God is good’ does not mean ‘God resembles God’. In section 2.3, I argued
that native speakers (at least, those not in the grip of a metaphysical theory) can, on
reflection, see that moral properties are not natural properties. Can’t we also just see
that moral talk is not about God?
The OQA does refute definitional moral supernaturalism, just as it refutes defin-
itional moral naturalism. The existence of atheist (or even agnostic) moral talk is
sufficient to show that ‘good’ does not mean ‘loved by God’, and ‘right’ does not
mean ‘commanded by God’. Any identity between divine and moral properties must
be non-definitional. But here there is a salient disanalogy. The moral supernaturalist
can reject the OQA while pressing the same argument against moral naturalism. Both
positions offer replies to Moore, and both replies face problems. However, those
problems are of very different sorts. Moral naturalism has difficulty accommodating
religious ambiguity and objectivism about well-being. Moral supernaturalism, for its
part, appeals to religious ambiguity and moral pluralism in controversial ways.
Depending on one’s precommitments, therefore, one can reasonably find one reply
to Moore more compelling than the other.
Adams’s moral supernaturalism is modelled on Boyd’s non-definitional moral
naturalism, discussed in section 2.3.2. Adams borrows Boyd’s semantics and meth-
odology, but argues that the property that best fits goodness is resemblance to God. As
we saw earlier, Adams rejects moral naturalism because it cannot respect the critical
stance. By contrast, moral supernaturalism can respect that stance. ‘The good (God)
vastly surpasses all other good things, and all our conceptions of the good.’35 Because
God is transcendent, our knowledge of God must always remain very imperfect, and
so the critical stance is always available. We can always ask whether one thing more
closely resembles God than another. No matter how much empirical knowledge we
acquire, debate about goodness is always open.
34
Adams, Finite and Infinite Goods, especially ch. 2. Adams himself combines his theist account of
goodness with a divine command account of wrongness, where it is only the commands of a good God that
constitute obligations. (Adams, Finite and Infinite Goods, chs 11 and 12.)
35
Adams, Finite and Infinite Goods, p. 50. To illustrate his position here, Adams quotes with approval
our epigraph from Murdoch’s Sovereignty of Good.
META - ETHICS 
Unlike moral naturalists, moral supernaturalists defend actual identity claims. No

BT ethicist would say: ‘This moral property is identical to some divine property that
may be unearthed at some future date, but I have no idea what it is.’ Rather they
identify this moral property with that divine property. Goodness is resemblance to
God. Rightness is conformity to God’s will, motivations, or commands. Moral
supernaturalism thus avoids any speculation about the future course of ethical or
metaphysical inquiry. It need not presuppose that inquirers will ever converge on BT,
or on any specific ethical view. Given the difficulties surrounding such speculation,
this is a very significant advantage.
However, by offering actual identities, the moral supernaturalist exacerbates the
open question challenge. Instead of predicting that, even at the end point of empirical
inquiry, she will still doubt some as yet unspecified identity, the opponent of moral
supernaturalism can simply doubt this actual identity. While I can always doubt
whether this finite thing is better than that, Adams must insist that no one could
doubt that X is good without simultaneously doubting that X resembles God. The
real Moorean question is closed. But then doesn’t the persistent testimony of his
atheist opponents simply render this claim absurd?
I believe the Moorean semantic argument is actually less persuasive against moral
supernaturalism. This is because the supernatural realm is much more uncertain, and
much less explored, than the natural realm. We have a pretty good idea what the
natural world is like, which natural properties might be available to underwrite moral
facts, how human beliefs regarding the relations between natural and moral facts
might converge, and so on. Those who now believe that the gap between moral facts
and natural facts is unbridgeable can confidently assert that better acquaintance with
the actual future course of empirical moral enquiry would not lead them to change
their minds. It is reasonable to believe that natural properties are just not the right
sort of properties to play the moral role.
With moral supernaturalism, such confidence is misplaced. Although familiar
from philosophical history, ideas such as resemblance to a morally perfect God, or
conformity to a cosmic purpose, are largely unfamiliar in contemporary analytic
moral philosophy. We cannot be confident that they cannot ground genuine moral
normativity. Could we really know that X resembled God, and still fail to find
X good? Could we really know that Y cohered with the genuine purpose of the
universe, and still fail to see any reason in favour of Y?
It is easy to say that you can imagine knowing that X resembled God and still
wondering about X’s moral properties. Many atheists say they can imagine God
telling them to do Y, and then deciding for themselves whether to do Y. But
imagination is not a reliable guide to possibility—especially when one is speculating
about what one would do in circumstances one considers to be absurd. Global
naturalists regard both God and Platonic Forms as impossible (and possibly unin-
telligible) entities. For that very reason, any global naturalist utterance that begins ‘If
I were to encounter God . . . ’ must be taken with a large dose of salt.
 META - ETHICS
To silence the Moorean OQA, both moral supernaturalists and moral naturalists
must argue that the discovery of identities between moral and non-moral properties
is a life-changing (or at least moral-perspective-changing) experience. Unlike moral
naturalism, moral supernaturalism can cite actual evidence here: namely, the moral
transformation associated with genuine mystical experience. In Christianity, for
example, a genuine encounter with God makes one more benevolent, more charit-
able, and less self-centred. The discovery of supernatural moral facts thus motivates
an extremely austere and demanding morality.
This mystical evidence suggests that the Moorean question is closed. No native
speaker with mystical experience would wonder whether something that resembled
God was good. If one wonders whether what one is perceiving is good, then one is not
truly perceiving God. And no native speaker who lacks mystical experience can be
confident that this question would remain open for her if she were to have such
experience. (Of course, this argument only works if we give some credence to the
testimony of mystics. But, as I argue in chapter 5, we should.)
Moral supernaturalism also accommodates reasonable moral pluralism. Different
views of well-being, obligation, and value can all be reinterpreted as competing
accounts of resemblance to God. Unlike moral naturalism, moral supernaturalism
does not beg the question against any substantive moral view. This is dialectically
very important. Many moral philosophers are drawn to non-naturalism, not by its
positive attractions, but rather because they reject the specific substantive views
associated with moral naturalism, notably hedonism and preference utilitarianism.
And, of course, the views that moral naturalism cannot accommodate include the
commitments I defended in chapter 1.
I conclude that the reduction of moral properties to divine ones is a serious option,
even though their reduction to natural ones is not. The more serious barrier is to
establish that there are divine properties. But that is the central question addressed in
this book.
While moral supernaturalism may soften the OQA, it cannot entirely silence it.
Non-naturalists will remain unconvinced. They can agree that moral and divine
properties necessarily coincide. Necessarily, X is good if and only if X is loved by God.
But the two properties may be nonetheless distinct. Perhaps goodness is neither a
natural property common to all good things nor any divine property. Goodness is a
separate sui generis moral property that all things loved by God have in addition to
the property of being loved by God. (As with moral naturalism, the non-naturalist
denies that supervenience implies property identity.)
No open question test can capture this objection. The non-naturalist can grant
that no one who understands the nature of God and recognizes that X resembles
God can still wonder whether X is good. But she still insists on distinct moral
properties, and on an epistemic distinction between the knowledge that X resembles
God and the knowledge that X is good. Only ethical intuition can tell us that God
is good.
META - ETHICS 
Our project does not require us to decide between moral supernaturalism and
non-naturalism. For most of this book, I remain agnostic between the two forms of
anti-naturalist moral realism. Both provide support for AP. My task in this chapter is
to defend their credentials, and to explore that support. The argument of this section
has simply been that, for those who cannot accept sui generis moral facts, moral
supernaturalism emerges as the most credible alternative to moral nihilism.
2.8 A Moral Argument for Benevolent Theism

We now have the building blocks of a moral argument for BT. Given the semantic
inadequacy of moral naturalism and non-cognitivism, the moral realist must
embrace either moral supernaturalism or non-naturalism. The former presupposes
BT or some other supernatural reality. And BT also makes non-naturalism much
more plausible. BT non-naturalism can draw on non-natural precedents and explan-
ations, and draw strength from the explanatory power of cosmic values.
As an argument for AP, this is incomplete in several ways. Most obviously, I have
presented an argument from moral realism to benevolent theism. Our most urgent
task is to extend it to AP. I address that issue in the next section. Even as an argument
for BT, however, what I have sketched thus far has several limitations.
First, of course, I have omitted many details along the way, including most of the
replies available to defenders of non-cognitivism, moral naturalism, atheist non-
naturalism, and moral nihilism. All these positions have very able defenders, who
would find fault with the argument as it stands.
Second, my argument unashamedly has very controversial premises—especially its
commitments to religious ambiguity and objective value. This is deliberate, as my
aim is to highlight the ways that these commitments support BT. The flip side,
however, is that those who do not share those commitments can easily sidestep my
arguments.
Third, while I have argued that BT strengthens non-naturalism, I do not claim that
the latter cannot stand on its own. Our other commitments (especially methodo-
logical pluralism and objective value) may suffice on their own to deflect the
objections to non-naturalism without any need for God or cosmic purpose. This
would weaken the moral argument for BT, but it would bolster the arguments
presented in part I, where I argue from non-naturalism about objective values to
cosmic purpose. Those arguments are obviously much more interesting if the former
does not already presuppose the latter!
Fourth, as I conceded in section 2.4, I haven’t really offered an argument against
nihilism. The nihilist may see no reason to accept anything I say. (This is not
surprising, as the consistent nihilist sees no reasons at all.) Alternatively, the nihilist
might want to explore the cosmic realist nihilism sketched at the end of section 2.4.
Fifth, my argument from moral realism to BT rests, at a number of crucial points,
on claims I have yet to defend, and which may strike many readers as bizarre. To cite
 META - ETHICS
only two examples, I have appealed to the epistemic credentials of mysticism, and the
use of objective values to explain basic facts about the cosmos. These claims obviously
stand in need of defence, especially in contemporary philosophical circles. That
defence is offered in chapters 3, 4, and 5.
Sixth, any non-deductive argument for the existence of God must be balanced
against arguments on the other side. Those who find the very idea of God, cosmic
purpose, or non-natural reality unacceptable will no doubt prefer the less glaring
inadequacies of non-cognitivism, naturalism, or nihilism. (And, in part II, I myself
argue that there are very good arguments against benevolent theism.)
There is no knock-down argument from morality to God. But there is a relation of
mutual support between BT and moral realism. If BT is true, then moral realism is
almost certainly true; while atheist moral realism is problematic. Conversely, moral
realism also supports BT, because non-naturalism (the only plausible metaphysically
neutral moral realism) supports the arguments of part I.
The inference from moral realism to BT is not watertight. Those who find the
latter totally implausible on independent grounds may find some other way to defend
the former. But their mutual support is sufficient to strengthen the case for both.
2.9 Can AP Borrow the Argument for Theism?

How does this help AP? Benevolent theism has many moral resources—some of
which AP cannot easily borrow. Doesn’t that give BT a decisive advantage over AP?
Our final task in this chapter is to rebut this objection. I concede that BT offers an
explanation of moral facts that is, in some respects, better than anything AP can offer.
But AP has its own advantages, especially regarding the explanation of evil, and other
phenomena discussed in part II. And, as I now argue, AP still offers an account of
moral facts that is distinctly better than anything the atheist can offer.
From a BT perspective, there are three moral objections to AP.
1. AP’s cosmic purpose is morally deficient. Therefore, it cannot directly ground
moral facts. While moral facts can be identical to facts about a benevolent God,
they cannot be identical to facts about an inhuman cosmic purpose. Nor can AP
ever be consistent with plausible non-natural moral facts.
2. AP cannot borrow any BT explanation of sui generis moral facts, because such
explanations essentially involve God’s benevolence towards human beings.
3. AP cannot borrow BT explanations of human knowledge of moral facts, because
these also need God’s benevolence.
My replies to the first two objections come in part III, where I argue that AP can
ground plausible cosmic values. (The success of this task is also presupposed in part I,
as the arguments presented there rely on cosmic values.) The BT God is a morally
perfect, transcendent, cosmic creator. AP can borrow all of these features, by
separating them from concern for human beings. (One central challenge here is to
META - ETHICS 
develop a non-human-centred account of moral perfection. I address that challenge

in chapters 5 and 6.) But BT also includes divine plans for individual human beings,
divine commands, and divine incarnations, and a God who cares for individual
human beings. AP cannot borrow these elements. If there is a God, then God has
no plan for us, issues no commands, does not become human, and does not care for
human beings. So any human morality based on AP must do without these elements.
In part III, I argue that a morally perfect, transcendent, cosmic creator is sufficient
to ground a plausible (if austere) human morality. I attempt to build a full moral
theory that links non-human-centred cosmic values to human moral obligations and
the human good.
This still leaves the epistemic worry. One of the strongest moral arguments for BT
is its ability to explain our knowledge of moral facts. How can AP borrow that
explanation? Surely only a benevolent God would give us epistemic access to moral
facts. And surely AP must reject many traditional religious accounts of moral
knowledge, such as divine commands, scriptures, divine incarnations, personal
revelations, and the like.
I concede two things up front. (1) A benevolent God provides many more sources
of moral knowledge than AP; and (2) God’s benevolence would provide the easiest
explanation for epistemically reliable moral practices. However, it does not follow
either that BT offers the best overall explanation of actual human moral knowledge,
or that AP is inferior to atheism.
The BT explanation has its limitations. If BT were true, we would expect human
beings to have much more reliable moral beliefs, more enlightened moral practices,
and so on. What we actually observe is very fallible and incomplete human moral
knowledge. While BT does offer subsidiary explanations of the lacks and limitations
in human moral knowledge, as part of its theodicy and its response to religious
diversity, its opponents can reasonably reject those explanations. And this is what
I do in part II, where I argue that AP offers the best explanation of our actual
(limited) moral knowledge.
On the other hand, AP’s explanation of human knowledge of non-natural moral
facts is still superior to any atheist explanation. One of the most striking features of
the cosmos is that it is intelligible. In a universe designed to be understood—really
understood by someone somewhere in a way that passes our understanding—our
accidental partial knowledge is less anomalous than it would be in an atheist
materialist universe. Human moral knowledge is a thing of an expectable kind, rather
than an unaccountable anomaly. (We explore cosmic intelligibility in chapter 4.)
Analogously, if the cosmos is designed so that moral facts can be understood within
it, and so that creatures capable of moral knowledge can emerge naturally somewhere
within the cosmos, then it is not remarkable that other creatures emerge elsewhere
who can only gain fleeting glimpses of moral facts. The best explanation of our
human situation is that we are, not the former creatures, but merely the latter. For
instance, AP will argue that, while they do perceive a realm of cosmic purpose and
 META - ETHICS
value, Christian (and other BT) mystics fall into self-aggrandizing error when they
diagnose that experience as an encounter with a divine person who loves them as
individuals. They misread God’s love for the cosmos as love for them. (I develop this
explanation in chapter 5.)
The emergence of creatures with our level of moral awareness, and our imperfect
insight into objective values, is thus at least as likely under AP as it is under either BT
(where we would expect humans to have better moral knowledge) or global natur-
alism (where we would not expect any knowledge of moral facts at all). This pattern
of competing explanations will recur several times in this book, in relation to mystical
experience, evil, and religious diversity. In each case, I argue that AP alone provides a
satisfactory explanation for both our knowledge and its limits.
The AP explanation of moral knowledge requires a tight connection between
cosmic purpose and moral fact. (Otherwise, there is no reason why knowledge of
one would help provide knowledge of the other.) Our reply to the third BT objection
thus depends on our replies to the other two. AP’s moral epistemology rests on its
moral ontology. And that tale must wait until part III.
If AP can make good its promises, then it is a serious rival to both BT ethics and
atheist non-naturalism. Given the problems facing both these positions, it is at least
worth exploring this new alternative.
PART I
The Case against Atheism
3
Cosmological Arguments
‘Why is there anything at all, rather than just nothing?’ [is] the most fundamental
question of metaphysics.
[Martin Heidegger, Einfuhrung in die Metaphysik, p. 11]
It is not how the world is, but that it is that is mystical.

[Ludwig Wittgenstein, Tractatus Logico-Philosophicus, 6.44]
No question is more sublime than why there is a Universe; why there is anything
rather than nothing.
[Derek Parfit, ‘Why Anything? Why This?’, p. 24]
3.1 The Ultimate Question

‘Why is there something rather than nothing?’ Is this the ultimate philosophical
question, or is it a silly question, or even a meaningless one? While it captures a
thought that many non-philosophers (especially religious people) find compelling,
this is also a very strange question, and it is not clear what would count as an answer.
The cosmological argument offers God as the answer to our ultimate question. The
basic argument is simple. There is something, when there might have been nothing.
(Call this fact ‘S’.) Therefore, we can ask why there is something rather than nothing.
And God best explains S.
Cosmological arguments are a posteriori, because S is an empirical premise.
Different arguments posit different relations between God and S. Some argue that
S is impossible without God; others that God provides the only possible explanation
of S; and others that God is the best explanation for S. All agree that the fact that God
would explain S provides some reason to believe that God exists.
We can distinguish three types of cosmological argument. A deductive argument
seeks to prove that, given the fact of S, God must exist. We then borrow a distinction
from Richard Swinburne to separate two kinds of inductive argument: P-inductive
arguments show that a proposition is probably true; while C-inductive arguments
confirm a proposition by raising its probability.2 Swinburne seeks a P-inductive
1
Translated by Grunbaum, ‘A New Critique’, p. 4. 2
Swinburne, The Existence of God, p. 6.
 THE CASE AGAINST ATHEISM
argument. He wants to show that God probably exists. But he claims only that the
cosmological argument is a good C-inductive argument. Because a C-inductive
argument can ‘confirm’ a proposition with low initial probability, we could have
many such arguments and still lack a good P-inductive argument. However, Swin-
burne’s hope is that a series of independent C-inductive arguments will cumulatively
constitute a P-inductive argument. My hope for AP is similar.
There are many distinct cosmological arguments, and the literature surrounding
them is vast.3 My aim is neither to comprehensively survey that variety, nor to
summarize the literature. Instead, I aim to demonstrate the appeal of cosmological
arguments, to highlight their moral dimension, and to vindicate AP’s attempt to
borrow them. While not decisive, the cosmological argument does support cosmic
purpose. Furthermore, the argument leaves entirely open whether that purpose is
human-centred. Therefore, there is a good C-inductive cosmological argument that
AP can borrow.
Strictly speaking, S would be true if only God alone existed. In the cosmological
argument, however, S concerns the existence of contingent things, or of a physical
universe. It focuses our attention on things that exist even though they might not have
existed. I thus take S to be interchangeable with ‘contingent things exist’ or ‘there is a
physical universe’. Our ultimate question asks: Why are there any contingent things?
Why is there any physical universe?
To deny S is to say that there is nothing. Some philosophical traditions do claim
that nothing exists. Buddhists, among others, often deny that anything has perman-
ence, or inherent existence, or continuity. There are no enduring substances, only a
flux of momentary experiences. But these are not denials of S. Here, ‘nothing exists’
denies some particular kind of thing. To deny S is much more radical. Try imaging
nothing. You might begin with a busy universe, and successively remove all physical
objects, all air molecules, and finally all subatomic particles. This is not nothing, as
you still have space-time. Nothing is not an empty or dull physical universe; it is the
absence of any physical universe at all.
S is compatible with our being very misinformed about the nature of the physical
universe. Some Idealists believe that the physical universe is merely ideas in the mind of
God. But no Idealist denies S. Even a Berkelean world of divine ideas is still something.
The difficulty of imagining nothing leads many philosophers to conclude that our
ultimate question is meaningless, that it is easily answered, or that it can legitimately
be ignored. The next three sections address these attempts to undermine the cosmo-
logical argument.
3
My own thinking about this argument has been especially influenced by Nozick, Philosophical
Explanations, pp. 115–64; Swinburne, The Existence of God, pp. 133–52; and the works of John Leslie
cited in section 3.6.
COSMOLOGICAL ARGUMENTS 
3.2 Does the Ultimate Question Make Sense?

Some dismiss our ‘ultimate question’ as utterly pointless. This dismissal takes two
main forms: Wittgensteinian and naturalistic. The Wittgensteinian response is that
the statement ‘nothing exists’ is incoherent. As Grunbaum puts it: ‘Wittgenstein
acknowledged the powerful psychological reality of wondering at the very existence
of the world. Yet logically, he rejected the question altogether as “nonsense”, because
he “cannot imagine its [the world’s] not existing”.’4 S could not have failed to be true,
and thus needs no explanation. When philosophers go beyond ordinary usage, they
bend language too far. We can understand ‘nothing’ if it means ‘no objects of type x’,
but we cannot imagine a possible world where absolutely nothing exists. Is it a world
without mathematical or logical truths? Or is it a place with some positive truths
(such as ‘5 + 7 = 12’) but no things? If so, what counts as a thing? Whatever we
imagine, we must be imagining something.
Our commitment to methodological pluralism cautions against any attempt to
resolve a controversial philosophical question by fiat. While we should be suspicious
of philosophers’ assertions that X is possible, we should also be equally suspicious of
Wittgensteinian accusations of incoherence. As David Lewis noted in a similar
context, an incredulous stare is no substitute for an argument.5 S is false in any
possible world where every possible contingent existential statement is false. In such a
world, mathematics and logic remain true, but there are no physical entities or
physical laws. Many people feel that they can make intuitive sense of such a world.
Even if we cannot picture it in our minds, we can still describe it without contradic-
tion. It is, in this sense, imaginable. We know by observation that the actual world is
not such a world. So we can reasonably ask why. The ultimate question is certainly
odd. But perhaps philosophy just is the search for answers to questions that no non-
philosopher would even bother to ask.
The Wittgensteinian response may also be accused of caprice. We must be careful
not to overestimate our own perspective, or to equate our own imaginative limita-
tions with the limits of thought. I cannot silence my opponents simply by refusing to
understand them. Perhaps our question will turn out to be meaningless or unanswer-
able. But until this is demonstrated, we should heed the words of Robert Nozick:
‘someone who proposes a non-strange answer shows that he did not understand this
question’.6
I conclude that we should not dismiss our ultimate question before we have even
tried to answer it. If there are intelligible answers to a question, then it makes sense to
4
Grunbaum, ‘A New Critique’, p. 3. An extended defence of the meaninglessness charge is Rundle, Why
There Is Something Rather Than Nothing. For a different perspective, see Nozick, Philosophical Explan-
ations, pp. 115–64.
5
Lewis, On the Plurality of Worlds. Lewis makes this observation in relation to his modal realism,
discussed in section 3.3.4 below.
6
Nozick, Philosophical Explanations, p. 116.
 THE CASE AGAINST ATHEISM
ask it. So we may proceed with our investigation—bearing in mind Nozick’s dictum
that a strange question may require a bizarre answer. (In doing so, we do not really
beg the question against informed opponents of the cosmological argument, who
reject our question as meaningless only after they have considered the proposed
answers and found each as unintelligible as the question itself.)
The second dismissal of our question comes from naturalists. Explanation is what
science does. To seek an explanation is to seek a scientific explanation—to do science.
Whatever science cannot explain must be treated as a brute fact. Science only offers
explanations within this physical universe. It thus makes no sense to ask for a
scientific explanation of S. But a ‘non-scientific explanation’ is a contradiction in
terms. So our ‘ultimate’ question makes no sense.
Not all naturalists dismiss our ultimate question. Some instead offer scientific
answers, or accept that, while the question makes sense, S is simply inexplicable.
(We return to these more modest naturalisms below.) But every naturalist accepts
this disjunction: if there is a scientific explanation for S, then we need no supernatural
explanation. If not, then S is a brute fact. Either way, we need neither God nor cosmic
purpose.
BT turns this naturalist dichotomy on its head. Science may tell us how things
work within the physical universe, but it can never tell us why there is such a
universe. As Swinburne puts it, S is a fact ‘too big’ for science to explain.7 But
S must have some explanation. Therefore, we must seek a non-scientific explanation.
‘Naturalism’ is a slippery word in contemporary philosophy. One crucial distinc-
tion is between ontological naturalists, who insist that only ‘natural’ entities exist, and
methodological naturalists, who insist that legitimate theoretical investigation should
only ever posit natural entities, processes, or explanations. (Supernatural entities may
exist, but it is bad form to appeal to them.) We encountered ontological naturalism in
chapter 2, under the name ‘global naturalism’. For consistency, we retain the latter
name.
To dismiss our ultimate question on ontological grounds is obviously premature.
We cannot simply presume global naturalism. A methodological dismissal is also
problematic, on two counts. First, as we saw in chapter 2, our commitment to
methodological pluralism explicitly involves a willingness to countenance both
non-natural entities and non-scientific modes of explanation. If science cannot
answer our ultimate question, then perhaps we must go beyond science. At the
very least, we cannot dismiss potential answers simply because they are ‘unscientific’.
Second, those who dismiss as ‘unscientific’ both our ultimate question and any
proposed purposive answer thereby illegitimately tie science to atheism. ‘Natural’,
‘scientific’, and ‘atheist’ are not synonyms. We saw in chapter 2 that Boyd’s ‘scientific’
ethical methodology could, in principle, lead to moral non-naturalism or even moral
7
COSMOLOGICAL ARGUMENTS 
supernaturalism. A genuinely scientific response to our ultimate question could

easily suffer the same fate. Philosophical attempts to constrain the methods and
ontological commitments of future science are foolhardy. We cannot assume that
science will never answer the ultimate question. But, by the same token, nor can we
take it for granted that science can succeed without positing cosmic purpose. In a
religiously ambiguous universe, we cannot assume that future scientists will be
atheists. (After all, some current scientists endorse BT.) And, even when atheist
scientists try to explain S, they often drift perilously close to a cosmic purpose.8 As we
shall soon see, there is good reason to doubt that atheist science can answer our
ultimate question. But that needn’t mean that science per se is similarly limited. (We
return to the possibility of an alliance between science and cosmic purpose several
times in later sections, and also in chapter 4.)
At the very least, it is premature to dismiss our ultimate question simply because it
moves beyond one parochial contemporary account of how ‘science’ must proceed.
3.3 Three Non-purposive Answers

Not all naturalists dismiss our ultimate question. Some offer scientific explanations
that are clearly non-purposive. If these succeed, then the possibility that future
science might invoke cosmic purpose is moot, and the cosmological argument is
dissolved. This section explores three non-purposive science-based explanations of S:
the instability of nothing, the theory of everything, and the Hume–Edwards prin-
ciple. I argue that all three fail, and our ultimate question still stands.
3.3.1 The instability of nothing

Two quite distinct intellectual traditions argue that, while there could be a time when
nothing exists, there is no possible world where nothing ever exists. Nothing inevit-
ably gives rise to something, sooner or later. Nothing is inherently unstable. Consider
this remark often attributed to Heidegger: ‘nothing nothings itself ’.9 This somewhat
gnomic phrase suggests the following picture. Nothingness is a force of negation. It
acts on whatever happens to be. Faced with other things, nothingness causes them to
cease to exist. But left alone, nothingness acts on itself. It negates itself, and the result
is that something exists. In a more scientific idiom, some physicists trace the
emergence of matter to the inherent instability of the quantum vacuum, where
random fluctuations eventually generate matter. Nothing cannot continue forever.
In the words of one physicist: ‘Our universe is just one of those things that happen
from time to time.’10
8
Consider, for instance, the attempts of Paul Davies and Roger Penrose, among others, to offer
‘scientific explanations’ of fine-tuning, discussed in chapter 4.
9
The original text is Heidegger, Einfuhrung in die Metaphysik.
10
Tyron, ‘Is the Universe a Vacuum Fluctuation?’, p. 244.
 THE CASE AGAINST ATHEISM
Either story might be true. But in neither does nothing give way to something.
Instead, both nicely illustrate the different senses of ‘nothing’. Whatever else it is, a
Heideggerian force of negation is not nothing. And a quantum vacuum is not
nothing either. It is a highly complex physical system. The emergence of matter
from a quantum vacuum is just another case of one thing leading to another. Perhaps
nature abhors a vacuum. But a vacuum-abhorring nature is still something. And we
can then ask why such a nature exists.
3.3.2 Is S necessary?
Imaginability is not an infallible guide to possibility. Perhaps we think we can
imagine what is nonetheless impossible. If you ask why ‘p or not-p’ must be true,
or why 2 + 2 = 4, then a reasonable reply is: ‘Because it is necessarily true.’ Perhaps
our ultimate question has a similar answer. Like the incompleteness of arithmetic, S is
a surprising necessary truth. Our ultimate question is not meaningless. But it is easily
answered.
A proof of the impossibility of nothingness would dissolve the cosmological
argument. But no one has derived a contradiction from the claim that nothing exists,
or otherwise proved that S is necessary. The most popular route here is to eliminate
the gap between mathematics and physics.11 Cosmological arguments need this gap.
We imagine a possible world where there is no physical universe, and then we ask
why that possible world is not actual. If the actual laws of physics are mathematically
necessary, and hold true in every possible world, then there is nothing left to explain.
The gap between maths and physics can seem obvious. Mathematicians explore a
vast array of possible laws, and then physicists perform experiments to discover
which laws apply in the actual world. Maths is a priori, while physics is a posteriori.
(Why else do we spend billions of dollars on particle accelerators?) But some working
physicists, and especially some popular accounts of physics, present an alternative
picture where physics and mathematics combine to yield a theory of everything
(TOE).12 This TOE would explain, not only how the laws of physics actually work,
but why they must work that way. This would explain S, along with much else.
11
Two additional routes to the necessity of S are to prove that there is a necessary being, or that logic
itself demands a non-empty domain of discourse. However, neither route is plausible. BT has long argued
that God is the (only) necessary being who could explain S. Until we prove that this is not so, the appeal to a
necessary being is thus not an alternative to purposive explanation. As for the domain of discourse:
logicians can quantify over abstract entities (numbers, sets); some logical systems countenance empty
domains; and it is surely presumptuous to equate the limits of our logic with the limits of possibility for the
cosmos.
12
The expression ‘theory of everything’ (TOE) is multiply ambiguous. For our purposes, the crucial
distinction is between a modest TOE that merely attempts to provide a complete account of how the
physical universe works (typically by unifying General Relativity and Quantum Field Theory) and a more
ambitious TOE that also explains why the universe exists and works as it does. This is sometimes
characterized as a distinction between physical TOEs and philosophical TOEs respectively. (Neither should
be confused with the even more modest search for a Grand Unified Theory (GUT) to unify the
electromagnetic, weak interactive, and strong interactive forces of the Standard Model of particle physics.
COSMOLOGICAL ARGUMENTS 
We discuss TOE in more detail in the next chapter. At this point, we note two key
points. First, no TOE is yet available, nor is it obvious that physics will ever discover
one. Perhaps no physical theory will ever explain either why the laws of physics are as
they are, or why there must be a physical universe. Until some TOE is available, it is
legitimate to explore alternative explanations. Second, and perhaps most importantly,
it is not clear that TOE is an alternative to purposivism. Why assume that physics will
explain S without recourse to cosmic purpose? This simply begs the question against
the cosmological argument. (In particular, as we’ll see in chapter 4, the most popular
candidate TOEs need objective values to fill other explanatory roles. And it is far from
obvious that one can have these values without some cosmic purpose.)
3.3.3 The Hume–Edwards principle

Another popular naturalist strategy constructs an explanation of S using scientific
explanations of other things. We thus explain S without any new ontological com-
mitments. One famous attempt is the following:13
The Hume–Edwards principle. If we have an explanation of the existence of each member of a
set of items, then we have an explanation of the existence of the whole set.
Consider the set of states of the physical universe at different times (S1, . . . , Sn).
Suppose science can explain any state of the universe in terms of the previous state.
We then have an explanation of each member of the set. By the Hume–Edwards
principle, we can now explain the state of the universe at all points in time. Nothing is
left unexplained. Of course, we do not actually have all these scientific explanations.
Nor is it obvious that we ever could. But the atheist requires no such claim. She does
not say that we have an explanation for S, but rather that, if it can be found at all, such
an explanation lies entirely within a familiar scientific framework. There is no need
for cosmic purpose.
Our ultimate question presupposes that wondering why there is a physical uni-
verse is distinct from wondering why this or that particular thing exists. The Hume–
Edwards principle challenges this. If we have an explanation for each state of the
universe, then we have nothing left to wonder about. S is not some mysterious further
fact. So S needs no separate explanation. (Instead of dismissing our question, the
naturalist now denies that there is any single ultimate question.)
To evaluate the Hume–Edwards principle, we must consider two cases: either the
physical universe had a beginning or it did not. I shall argue that, either way, some
significant fact remains unexplained.
Physical TOE is similar to GUT, except that it also seeks to include gravity.) Only the most ambitious
philosophical TOE could provide an answer to our ultimate question. I am grateful to an anonymous
referee for pressing me to clarify the relationship between GUT, TOE, and AP.
See, e.g., Pruss, ‘The Hume–Edwards Principle’. The principle was developed by Paul Edwards from
13
Hume’s response to Leibniz.

 THE CASE AGAINST ATHEISM
... Did the universe begin? Suppose the universe began at t1. If science explains
each state of the universe in terms of previous states, then we lack a scientific
explanation for its state at t1. We thus lack an explanation for each member of our
initial set, and the Hume–Edwards principle does not apply. We can still ask why the
universe began. And it is not clear how science could answer that question. (As we saw
in section 3.3.1, explanation in terms of a quantum vacuum, or any other physical
system, is no good. We can still ask why that physical system exists.)
There is a broader point here. Our ultimate question seems arcane. It becomes
more compelling when linked to other, more tangible questions. Can we link S to
some other fact that clearly is worth explaining? One perennially popular candidate is
the ‘fact’ that the universe began (B). If B is a fact, then it would be good to explain it.
Ambitious BT philosophers have argued, not only that B is true, but that S implies B.
Every physical universe must have a beginning. Something outside the universe must
have caused it to exist. And this all men call God. (This is the famous medieval
Islamic Kalam cosmological argument—championed in recent debate by William
Lane Craig.14)
The steps from B to God are controversial, to say the least. Why must everything
that begins to exist have a cause outside itself? Is it really logically impossible for
something to cause itself to exist, or to come into existence for no reason whatsoever?
Many atheists would rather countenance these possibilities than posit God. However,
for our purposes the main point is that a connection between S and B would make
our ultimate question harder to avoid. So we must ask whether B is true.
Philosophers have long argued that B must be true, because every universe has a
beginning. Aristotelians famously deny the possibility of an actual infinity of past
times. But these arguments are notoriously controversial, and often misunderstand
infinity.15 Opponents also cite the emergence of infinite models in cosmology as
evidence that an infinite past is, at least, imaginable.
Fortunately, a more modest approach is available. If the Hume–Edwards principle
only applies within a universe that is actually infinite, then it is sufficient to show that,
even if some possible universes are eternal, this one did begin. Some cite modern
cosmology to establish a beginning.
14
See, e.g., Craig, ‘Theism and the Origin of the Universe’. The view that a finite universe strengthens
the cosmological argument is not confined to philosophers. For instance, Pope Pius XII welcomed the Big
Bang theory as evidence for the existence of God; while the atheist astronomer Fred Hoyle resisted the same
theory in part because he thought it would open the door for theism. (Grover, ‘Cosmological Fecundity’,
p. 284.)
15
Consider one Aristotelian argument. If past time is infinite, then some past moment (t1) was
infinitely long ago. To get from t1 to the present requires the completion of an infinite series. (Imagine
someone who starts counting at t1.) But this is impossible, as an infinite series can never be completed. If
past time were infinite, the universe would never have reached the present. This particular Aristotelian
argument misunderstands eternity. An eternal universe does not begin an infinite time ago. It never begins
at all. Past time is infinite if, for any finite number n, there are times that are longer ago than n.
COSMOLOGICAL ARGUMENTS 
Everyone agrees that the universe began with a Big Bang. However, this is not
sufficient to establish B.16 Some cosmological modals posit a series of Big Bangs, each
generating its own universe which then leads eventually to a new Big Bang. In theory,
that series could be infinite. And some cosmologists insist that it is infinite. Each Big
Bang is then merely another event within an infinite cosmos. (It is often suggested
that the theory of everything, discussed in section 3.3.2, will yield an infinite cosmos.)
The existence of infinite cosmological models suggests that the Big Bang does not
entail B. However, infinite models are a minority view within cosmology. While some
cosmologists explore such models, many claim that the current scientific consensus is
against them. Although understandably popular with atheist apologists, infinite
models seem to swim against the tide. We return to this issue in chapter 4, where
I also argue that one popular argument for these models, based on the so called
‘cosmic coincidences’, is incompatible with atheism.
There is an overwhelming scientific consensus in favour of the Big Bang. And a less
universal consensus links the Big Bang to a cosmic beginning. This consensus should
carry significant weight—especially among philosophers who claim to defer to
science. However, I do not want to place too much weight on any claim to scientific
consensus in this rapidly developing field. Therefore, we must now explore the
possibility that our cosmos is infinite.
... Explanation in an eternal universe The cosmological argument is obviously

on stronger ground if we can demonstrate that, even in an infinite universe where
every state is explained by some earlier one, we still lack an explanation for S. We
begin with an example made famous by Leibniz.17
The immortal book. G is a geometry book that has existed eternally. (Perhaps it is continually
copied by an infallible process.) For any n, G’s existence at time t(n) is fully explained by its
existence at t(n-1).
We can explain G’s existence at each moment. Yet two things remain unexplained:
why does an immortal book exist; and why this book? Why not a non-eternal book,
an eternal cookbook, or an eternal refrigerator? Explanations may not be forthcom-
ing. But, until we have them, there are things we would like to know but do not. Any
theory that can explain the immortal geometry book has an explanatory advantage—
as does any theory that explains why this infinite universe exists. There are infinitely
many possible eternal universes. Why are we in this one?
16
For a detailed and balanced account of the scientific issues here, see Pitts, ‘Why the Big Bang
Singularity Does Not Help the Kalam Cosmological Argument for Theism’. One issue is whether the Big
Bang singularity delivers a first moment in time—as required by most formulations of the Kalam argument
(Pitts, ‘Why the Big Bang Singularity’, pp. 679–80).
17
See, e.g., Pruss, ‘The Hume-Edwards Principle’; Swinburne, The Existence of God, p. 143. (The
original example is in Leibniz’s On the Ultimate Origination of Things, pp. 31–2.)
 THE CASE AGAINST ATHEISM
In an eternal universe, the Hume–Edwards principle applies. We can get an

explanation of each member of our initial set of statements. But we then see that
the principle fails as a response to our ultimate question. Even though we now have an
explanation of each member, there is still something left unexplained. We cannot
explain S simply by piling up explanations within the physical universe.
We can see the inadequacy of the Hume–Edwards principle even more clearly in a
finite cyclical universe, where time moves in a loop.18 Suppose there are three times:
t1, t2, and t3. The state of the universe at t1 explains its state at t2, which explains its
state at t3, which in turn explains its state at t1. We can explain the state at any time,
and we can thus say, if we like, that we have ‘an explanation of the set of states’. But it
is very clear that we lack a full explanation of this universe. There are infinitely many
possible cyclical universes: one contains only billiard balls; another only chocolate-
eating dragons; and so on. Why is this cyclical universe actual?
The failure of the Hume–Edwards principle does not show that science cannot
explain S. But it does further undermine the atheist insistence that familiar everyday
scientific explanation will one day explain S without reference to cosmic purpose.
The cosmological argument cannot be so easily dissolved.
3.3.4 Modal realism

Our final ‘familiar’ explanation of S draws on David Lewis’s modal realism, where all
possible worlds are as real as the actual world.19 To my knowledge, no one explicitly
introduces modal realism to explain S.20 But it does seem to provide an explanation.
(Why is there something rather than nothing? Because there is everything.)
Modal realism is a minority view. Opponents object to its metaphysical extrava-
gance. But recall Nozick’s dictum. Perhaps our ultimate question requires something
so extravagant. And modal realism is not obviously more extravagant than the
purposive explanations discussed below.
I shall argue that AP can sidestep this debate, because modal realism does nothing
to undermine the cosmological argument. Indeed, it may strengthen it!21 Our real
interest is cosmic purpose. When we ask our ultimate question, we want to know
whether S makes it more likely that our universe has a cosmic purpose.
Consider all the possible worlds where S is true. In each world, either S is a brute
fact, or S is true for some reason. There are infinitely many possible explanations, but
18
Pruss, ‘The Hume–Edwards Principle’, pp. 152–3. This is not merely a philosopher’s example. Some
physicists offer explanations of fine-tuning—and other phenomena—that seem to involve causal loops of
precisely this kind. Consider, for instance, the causal explanations discussed in Davies, The Goldilocks
Enigma, pp. 288 and 314.
19
Lewis, On the Plurality of Worlds.
20
Leslie briefly compares modal realism to axiarchism in this context, but argues that the latter is
superior (Infinite Minds, pp. 24–30).
21
The argument in the text is modelled on (but does presuppose the validity of) the familiar objection
that if modal realism is true, then induction cannot be a reliable route to knowledge. (See, e.g., Leslie,
Infinite Minds, pp. 24–30.)
COSMOLOGICAL ARGUMENTS 
only one way to be a brute fact. In most possible worlds, therefore, S has some
explanation. Therefore, in the actual world, S probably has an explanation. Rather
than explaining S, modal realism merely rules out accepting S (or, indeed, anything
else) as a brute fact.22
Opponents of AP will reply that, while some explanations of S yield a cosmic
purpose, most do not. In most possible worlds where S has an explanation, there is no
cosmic purpose. Therefore, our world probably lacks a purpose.
These two arguments share the same logical structure. If possible worlds where
S coexists with a cosmic purpose are in a minority, then there is probably no cosmic
purpose. By the same token, if there is a cosmic purpose in the majority of worlds
where S is true, then there probably is a cosmic purpose. Everything turns on the
proportion of those worlds where S is true that have a cosmic purpose. As we have no
independent epistemic access to possible worlds, we can only ask whether the most
likely explanations for S yield a cosmic purpose. But that simply returns us to our
investigation of the merits of competing explanations of S. Modal realism, on its own,
does nothing to threaten the cosmological argument.
3.4 Is S a Brute Fact?

Suppose that we currently lack an explanation of S, and that we cannot be confident
that a purposeless explanation will ever emerge. The final naturalist response is to
accept S as a brute fact. There were two possibilities: S or not S. One of them had to
obtain. Not every fact has an explanation. Some things just happen to be true.
Brute fact is perhaps the most popular atheist response to the cosmological
argument. In this section, we ask whether proponents of the argument can insist
instead that we cannot accept S as a brute fact.
3.4.1 The Principle of Sufficient Reason

The most ambitious BT response is that we should never accept any brute facts. This
reply invokes the following:
Principle of Sufficient Reason (PSR). Every true proposition has an explanation: a sufficient
reason why it is true.23
22
More controversially, we could construct a parallel argument for AP over BT. Explanations of S that
make essential reference to human beings are a vanishingly small subset of possible explanations.
Therefore, the actual explanation of S almost certainly doesn’t mention human beings. Alternatively, if
possible explanations mentioning human beings are in the majority, then (presumably) so too are possible
explanations mentioning any arbitrary X. Instead of BT, we would get a cosmic purpose that included every
possible thing!
23
The most famous advocate of PSR is Leibniz, who argues that even the truths of logic need God to
make them true.
 THE CASE AGAINST ATHEISM
PSR insists that every ‘Why?’ question has an answer. As S is a true proposition,
S must have an explanation. There is a sufficient reason why the universe exists.
PSR does not guarantee a successful cosmological argument. Even if S has an
explanation, we must still find that explanation, and prove that it requires God or
cosmic purpose. However, PSR strengthens the argument considerably. So long as
brute fact remains an option, the atheist can reject any purposive explanation because
it is ontologically extravagant or otherwise incredible. Even if God alone can explain
S, the price may be too high. PSR undermines this response. If S must have an
explanation, and if God alone could explain S, then we must posit God.
PSR is controversial. Swinburne, himself a defender of the cosmological argument,
develops one common objection. Swinburne first argues that PSR implies that every
fact has an absolute explanation: one where ‘the existence and operation of each of
the factors cited are either self-explanatory or logically necessary’.24 (We cannot
allow any brute facts, because PSR rules them out.) By definition, there cannot be an
absolute explanation for any contingent fact, because an absolute explanation of
E must demonstrate that E could not have been false. PSR implies that contingency is
always an illusion based on ignorance. S must be a necessary truth. But, as Swinburne
notes, some facts seem contingent. And S seems to be one of them. (Indeed, the
modal intuition that S is contingent is what drives the cosmological argument.) Until
someone proves that S is necessary, we should seek an explanation of S that acknow-
ledges its contingency.
PSR has its defenders, however. One much-discussed recent defence is Pruss and
Gale’s ‘new cosmological argument’.25 They derive PSR from the weaker, less con-
troversial claim that, for each contingent true proposition, it is possible that it has an
explanation.26 This inference is controversial.27 Even if we accept the inference,
however, opponents can still reject PSR. Davey and Clifton, for instance, argue that
Pruss and Gale’s weaker claim is less intuitive than the atheist claim that some
contingent proposition lacks an explanation.28 Therefore, if Pruss and Gale’s weaker
claim implies PSR, we should deny the former rather than embrace the latter. We
thus have a clash of modal intuitions. For those persuaded by Pruss and Gale, we
have a compelling argument that our ultimate question must be answered. But this
argument will not persuade, much less compel, many atheists.
Our commitment to avoid caprice counts against the cosmological argument here.
If it is presumptuous to assume that science sweeps the board, then it is even more
24
25
Pruss and Gale, ‘A New Cosmological Argument’.
26
Pruss and Gale dub this weaker claim ‘the weak PSR’, as opposed to the strong PSR, which claims that
every contingent true proposition has an explanation. Pruss and Gale, ‘Cosmological and Design Argu-
ments’, p. 125.
27
See, e.g., Almeida and Judisch, ‘A New Cosmological Argument Undone’; and the reply in Pruss and
Gale, ‘A Response to Almeida and Judisch’.
28
Davey and Clifton, ‘Insufficient Reason’. See also, Pruss and Gale, ‘A Response to Oppy and to Davey
and Clifton’.
COSMOLOGICAL ARGUMENTS 
presumptuous to suppose that every fact has some explanation that we can discover
or comprehend. Modesty cautions against inferring that there is no explanation
simply because we (or, more accurately, some of us) cannot yet imagine one. But it
also cautions against assuming that there is an explanation. Modesty rejects PSR, and
cautions against reliance on the new cosmological argument.29
My general strategy is to remain neutral regarding clashes of modal intuitions.
Therefore, while acknowledging that PSR would strengthen the cosmological argu-
ment, we now explore less ambitious alternatives.
3.4.2 Is nothing the default?

Not all BT philosophers accept PSR. Some admit brute facts. But unlike atheists, they
do not accept S as a brute fact. Most people find some brute facts easier to swallow
than others—and different people find different brute facts unswallowable.
For instance, Swinburne’s cosmological argument shifts from absolute to ultimate
explanation, where all the premises are themselves brute facts with no further
explanation. Swinburne then argues that, while science and God both provide
ultimate explanations, God is superior. The cosmological argument can do without
PSR. The question is: which brute facts should we accept? Is God’s creation of a
physical universe easier to swallow than S?
We return to the comparative plausibility of God below. Our present question is
how easy it is to accept S as a brute fact. Some regard brute S as especially unpalatable,
because contingent true positive existential propositions (such as ‘This table exists’ or
‘The physical universe exists’) are especially in need of explanation. The non-
existence of an object needs no explanation, but its existence does. Nothingness
would have been unremarkable, but S is remarkable. As Swinburne puts it: ‘It is
extraordinary that there should exist anything at all. Surely the most natural state of
affairs is simply nothing: no universe, no God, nothing. But there is something.’30
The idea of a default (or ‘natural’) state of affairs is familiar. Objects continue in
motion until something interferes with them, and they continue in existence unless
something destroys them. But these ‘defaults’ concern events within the physical
universe, and we revise our views as we learn more about that universe. (Newtonian
mechanics famously overturned the Aristotelian view that rest is the natural state for
a moving object.) Is there a default state for the universe itself? Can we think
ourselves into some God’s eye point of view, and ask what to expect? Take a trivial
example. Suppose someone claimed that the universe was surprisingly big. What
would we make of this claim? How big would you expect a universe to be?
To avoid the subjectivity of intuitions, some cosmological arguments use objective
evaluations of the ‘likelihood’ (or ‘naturalness’ or ‘intrinsic plausibility’) of different
29
On the hubris of the Principle of Sufficient Reason, see Goldschmidt, ‘The New Cosmological
Argument’, pp. 271–2.
30
Swinburne, Is There a God?, p. 48.
 THE CASE AGAINST ATHEISM
states of affairs. Swinburne notoriously appeals to comparative simplicity.31 Noth-

ingness is the simplest possible state of affairs. Accordingly, S stands in need of
explanation in a way that nothingness would not. The intuitive idea is clear enough.
Whatever our criteria of simplicity, the empty world looks to be among the simplest.
But is nothing simpler than something? And does that make S somehow less likely?
Or does talk of intrinsic or objective degrees of prior probability create a misleading
impression of precision?
Perhaps we reach intuitive bedrock here. Some people just do find the existence of
a physical universe remarkable. To such people, S cries out for explanation. If you
share this reaction, then you at least have reason to seek an explanation for S. Others,
by contrast, find nothing remarkable in the fact that something exists.
This divergence tracks other disagreements, including our central dispute between
BT and atheism. BT sees the physical universe as a creation: fragile, incomplete,
dependent. Unless God continually kept it in existence, the world would cease to be.
Against the eternal and ultimate reality of God, the physical universe appears fleeting
and insubstantial. The existence of such a contingent universe cries out for explan-
ation. Many atheists, by contrast, see the world as solid, independent, ultimately real.
(It is, after all, the most real thing the atheist can imagine.) BT cannot imagine a
world without God, while atheists cannot imagine a world in need of divine support.
BT regards nothingness as the natural thing to expect in the (unimaginable) absence
of God, while atheists equally naturally treat the existence of a physical world as
unproblematic.32
Both sides can accuse their opponents of a lack of imagination here. The commit-
ted atheist cannot see the world as fragile and dependent, because she cannot imagine
the possibility of God; while BT cannot imagine God’s absence. Our agnostic
commitment to religious ambiguity rejects both sides. We cannot presume in
advance that S either is, or is not, the natural state of affairs. In this agnostic situation,
an explanation of S is desirable, but not essential.
Nozick describes this agnostic middle road as ‘egalitarian’, because it regards all
possibilities as equally likely. Egalitarians treat neither S nor nothingness as the
default. S does not demand an explanation, but neither is it so unremarkable that
explanation is altogether redundant. Other things equal, it is better to explain
S than not.
Also, recall that the cosmological argument is only one part of our case for AP. In
the next chapter I argue that, if we admit that any possible world is as likely as any
other, then this greatly strengthens some versions of the fine-tuning argument for
31
Swinburne, The Existence of God, pp. 96–109. For a forceful critique, see Grunbaum, ‘Is Simplicity
Evidence of Truth?’
32
This disagreement also has a historical dimension. For the ancient Greeks, existence was the default:
‘That even a divine creator would, like any craftsman, have to use pre-existing materials is an assumption
that the ancient Greeks apparently never questioned.’ (Sedley, Creationism and its Critics in Antiquity,
p. xvii.)
COSMOLOGICAL ARGUMENTS 
AP. The cosmological and fine-tuning arguments could then be two prongs of a
disjunctive argument for AP. If some possibilities are more natural than others, then
we must explain S. If not, then we must explain why, of all the equally possible
physical universes, it is this particular one that is actual.
3.4.3 Objective values

Our project explores the mutual support between AP and my utilitarian moral
commitments. One of those commitments is to objective values. Such values play
two key roles in our cosmological argument. First, they help motivate the ultimate
question. If S is objectively valuable, then that is one reason to believe that S does
stand in need of explanation. Brute fact is difficult to accept when an especially
valuable possibility obtains. If there are objective values, and if the actual state of
affairs is especially objectively valuable, then this does require explanation. Second,
we shall see that purposive explanations of S are much more credible if they can
either help themselves to objective values or posit entities who deliver those values.
Divine creation only explains why the physical universe exists if God either responds
to objective values or creates them. In other words, to invoke the distinction
introduced in chapter 1: the cosmological argument for AP delivers Normative AP,
and not merely Simple AP.
Objective values are thus central to any successful cosmological argument. But, as
we saw in chapter 2, BT is more comfortable with objective values than atheism is.
Does that render the cosmological argument pointless? Are we merely arguing in a
circle? Not necessarily. Perhaps we are exploring mutually supporting sets of inter-
connected views, rather than simple linear arguments from less controversial prem-
ises to more controversial conclusions. In the terminology introduced in chapter 1,
the cosmological argument marshals one set of distinguishing intuitions. It is not an
argument from decisive intuitions. Also, not everyone who embraces objective values
already endorses cosmic purpose. So the cosmological argument still has a dialectical
impact. Finally, while objective values are very useful, they are not necessary to
motivate the ultimate question. Even if S is no more remarkable than its negation,
it is still better to explain S than to leave S as a brute fact. We can ask our ultimate
question without presupposing the value of S. (Though, as we shall see, we may find it
difficult to answer it without somehow invoking—or deriving—that value.)
Chapters 1 and 2 defended objective values in relation to human well-being and
consequentialist morality. It may seem a huge leap to use those values to motivate our
ultimate question. One central task of the rest of this book is to (gradually) remove
the gap between the two notions of objective value: human and cosmic. I aim to
complete that task in chapter 12. As a first step, my exploration of axiarchism in
section 3.6 shows how the objective significance of S could emerge naturally from
reflection on common consequentialist evaluative commitments. (Chapter 4 extends
that argument to cover other striking features of the cosmos.)
 THE CASE AGAINST ATHEISM
3.5 Explaining S
If S is not a brute fact, then S has an explanation. But what could possibly explain why
there is anything at all? I borrow a useful notion from Parfit.33 A selector is a feature
that explains why x exists. Its utility for nail-hitting is a selector for my hammer,
because its having this feature explains why my hammer exists. A selector for the
universe is a feature of the universe that explains why it exists. To posit a selector just
is to deny that S is a brute fact.
Parfit’s notion of a selector is perhaps best illustrated using God. Anyone teaching
the cosmological argument encounters the question: ‘Yes, but who made God?’ Any
answer posits a selector for God. If that selector had to be a person or process who
made God, then BT would be doomed to regress. Some BTs, like Swinburne, deny
that God’s existence has any further explanation. God is the ultimate brute fact. But
other BTs do offer explanations for God. They say things like this: ‘God is a perfect
being, and God exists because God is a perfect being.’ Here, being perfect is a selector
for God.
Any theory that explains S is superior to one that does not other things being equal.
But other things are seldom equal. Any explanation for S (especially one that invokes
or implies a cosmic purpose) has metaphysical costs. Even if we agree that God would
explain S (and perhaps that nothing else could), we might still baulk at such
metaphysical extravagance.
Opponents of BT often appeal to parsimony. We should not posit entities beyond
necessity or seek explanations where none is available. Parsimony is a good principle.
We should seek the least extravagant hypothesis. But parsimony is not the only
explanatory virtue. It competes with explanatory power. In our present context, any
hypothesis has explanatory power if it offers a better explanation of S than the null
hypothesis that S is a brute fact.
The balance between parsimony and explanatory power depends on the urgency of
the ultimate question. If we endorse PSR, then we may be willing to pay a very
extravagant metaphysical price to explain S. At the other extreme, if our interest in
S is mere idle curiosity, then we may privilege parsimony. As we have seen, our
commitments in this project place us somewhere in-between. The ultimate question
is significant but not inescapable. So we must balance parsimony and explanatory
power. The most parsimonious explanation of S would invoke only familiar entities
and modes of explanation. If a purposeless explanation for S were available—or even
likely to be forthcoming—then the cosmological argument would collapse. But we have
already seen that familiar explanations fail. S is so different from any fact that science
has hitherto explained that there is no reason to expect any scientific explanation—and
especially not to expect a purposeless one. Given our commitments, it is not immodest
33
Parfit, ‘Why Does the Universe Exist?’; Parfit, ‘Why Anything? Why This?’
COSMOLOGICAL ARGUMENTS 
to ask our question, and not unreasonable to explore answers that are both non-
scientific and purposive.
Science and everyday life provide our background of familiar entities and familiar
modes of explanation. If we cannot explain S using only familiar entities and modes
of explanation, then every explanation carries a metaphysical cost. We can evaluate
the metaphysical costs of competing explanations of S along two dimensions: new
ontological commitments and new modes of explanation. I will explore two purpos-
ive explanations of S: BT and axiarchism.
The BT explanation of S is familiar. God created the universe. Recall Swinburne’s
definition: ‘God is a necessarily existing person without a body who necessarily is
eternal, perfectly free, omnipotent, omniscient, perfectly good, and the creator of all
things.’34 God stands outside the physical universe, and God clearly has the power to
bring it into existence. The main question for BT is why God would choose to create.
Here, BT explanations fall into two broad camps. The intellectualist says that it is
better for a physical universe to exist than not. This fact is God’s reason to create.
Being omniscient, morally perfect, and omnipotent, God will know, be motivated by,
and act on this reason. Therefore, God will create. We know God has this reason,
because we can discover the objective values that provide God’s moral reasons. On
the voluntarist view, by contrast, God’s creative act is itself the source of values and
reasons in our cosmos. God simply chooses to create a physical universe. There is no
further explanation.
As I said in chapter 2, while I am committed to objective values, I aim to remain
neutral between voluntarism and intellectualism. When it comes to constructing a
human morality, objective values that flow from God’s decision to create are just as
good as ontologically independent values.
BT introduces one new entity (God) and one new mode of explanation (divine
creation). Whether BT introduces a new kind of entity or explanation depends on our
background assumptions about human persons and objective values. Is God another
person like us, or something entirely new? Are we already committed to objective
values, or are these also new?
Axiarchism is much less familiar than BT. The axiarchic explanation of S is very
simple.
1. The physical universe is better than nothing.
2. The physical universe exists because of (1).
For BT, (1) might capture God’s reason to create. In axiarchism, comparative value is
itself the direct reason why the universe exists. The physical universe exists because it
is good period. Axiarchism can be regarded as either introducing a new mode of
explanation using familiar entities (objective values), or as introducing a radically
34
 THE CASE AGAINST ATHEISM
new kind of entity: directly efficacious values. Like the intellectualist, the axiarchist
must defend independent objective moral values. (By contrast, the voluntarist offers
to explain the emergence of value itself. Voluntarism thus provides BT with distinct-
ive resources that both intellectualism and axiarchism lack.)
We thus have three explanations for S: axiarchism, intellectualist BT, and volun-
tarist BT. Are there alternative explanations?
On one level, there are innumerable alternatives. As Hume famously pointed out,
an explanation of the cosmos based on a person or persons of less than infinite power
would seem more familiar.35 Moving beyond persons, any number of impersonal
creators might explain S. And each possible creator offers both intellectualist and
voluntarist explanations, as they may either respond to external objective values, or
create on the basis of their own whims or internal reasons. Finally, Parfit notes that
goodness competes with a myriad of other direct selectors. Just as the cosmos might
exist because it is good, it might also exist because it is governed by beautiful
mathematical laws, or because it is ‘ontologically maximal’.36
Fortunately, we can reasonably treat our three purposive explanations as repre-
sentative of this much larger set of possible explanations. Any explanation of S is
either direct or indirect. Axiarchism is a direct explanation: the cosmos has property
P, and this is why it exists. There is no intermediary linking goodness and existence.
There is nothing more to be said about why the best possible world exists. An indirect
explanation posits some mechanism, intermediary force, or creative agent to bridge
the gap between goodness (or some other property) and existence. We might say that
a direct explanation involves a selector (in Parfit’s sense), while indirect explanation
involves a creator.
As our primary interest is in cosmic purpose, we focus on two questions: (1) Does
the explanation yield a purpose? (2) Can AP borrow that purpose? It may seem that
AP cannot remain too agnostic. We can imagine selectors without purpose, and
creators without value. Surely these possible explanations threaten AP. But things are
not so simple. Our vocabulary of ‘purposes’ might itself be an illegitimately
anthropocentric leap. Selection needs no purpose in the sense that we have purposes,
just as it needs no personal creator who is like us. ‘Purpose’ is a metaphor, as is
‘selector’. The real question is whether our explanation of S warrants talk of values
and reasons. I argue in section 3.6.4 that we can legitimately assimilate all direct
explanations to axiarchism, because any selector will ground objective values; while
section 3.7.2 argues that AP can borrow any plausible creator. In part III, I argue that
any explanation involving objective values will yield something practically equivalent
to a cosmic purpose, as will any creator who is both (a) powerful enough to create a
physical universe and (b) not obviously morally deficient. Therefore, any explanation
for S yields cosmic purpose.
35
Hume, Dialogues Concerning Natural Religion. 36
Parfit, ‘Why Anything? Why This?’
COSMOLOGICAL ARGUMENTS 
AP is compatible with a broad disjunction of metaphysical positions. Limiting

ourselves to obvious alternatives, any of the following supports AP: axiarchic values,
an ultimate impersonal creative force, a lesser God or lesser gods, or a perfect being.
AP need not choose between axiarchism, BT, and impersonalism; nor between
intellectualism and voluntarism.
Until an alternative explanation for S is forthcoming, we can legitimately focus on
axiarchism and the two forms of BT. For our purposes, the crucial questions are
whether a creator must care for human beings, and whether objective values must be
human-centred. BT says ‘Yes’, AP says ‘No’. We must therefore ask whether human-
centred elements enhance the explanatory power of axiarchism or BT. I shall argue
that they do not.
3.6 Evaluating Axiarchism

To evaluate axiarchism, we must address the following questions. (1) Does axiar-
chism offer a good parsimonious explanation for S? (2) Do our moral commitments
support that explanation? (3) Does axiarchism yield a cosmic purpose? (4) Can AP
endorse that purpose—and thus borrow the axiarchic explanation of S?
3.6.1 Is axiarchism a good explanation?

Our first question runs together parsimony and explanatory power, and also runs
together ontological commitments and new explanatory modes. These are all very
difficult to separate. Axiarchism says that objective values explain S. Opponents
accuse it of metaphysical extravagance, as neither science nor everyday life requires
us to posit objective values. We could focus on explanatory power, and ask how
values directly give rise to facts; or we could focus on modes of explanation, and ask
whether this explanation is familiar or intelligible; or we could focus on ontological
commitments, and ask if we have independent reason to posit objective values. These
three questions are obviously related, and they are all best answered by rendering the
axiarchic mode of explanation, and its values, more familiar. As we shall see, our
moral commitments play a significant role here.
What most starkly distinguishes axiarchism from BT is that its explanations cite
goodness itself, rather than some good thing or person. The physical universe exists
because it is good, not because it resembles the Good or serves the purposes of a good
God. Many people find axiarchism bizarre—much more bizarre than BT. (One of my
senior undergraduate philosophy of religion students said it was the weirdest thing he
had ever heard in philosophy. And he is not alone.) But is this only because BT is more
familiar? In a series of works since 1979, John Leslie has built up a surprisingly
compelling case for axiarchism.37 And the view certainly has a venerable philosophical
37
See, especially, Leslie, Value and Existence; Leslie, Universes; Leslie, Infinite Minds, pp. 155–88. Leslie
himself acknowledges a debt to A. C. Ewing’s Value and Reality. Other recent defenders of axiarchism
 THE CASE AGAINST ATHEISM
pedigree. While this is no argument in its favour, it should at least make us hesitate
before rejecting axiarchism.
Parfit offers a succinct summary of axiarchism. ‘We are now assuming that, of all
the countless ways that reality might be, one is both the very best, and is the way that
reality is. On the axiarchic view, that is no coincidence. That claim makes, I believe,
some kind of sense.’38 The alternatives to axiarchism, in Parfit’s view, are either to
accept some astronomically improbable coincidence as a brute fact, or to offer some
more complex explanation (such as BT) which simply raises more questions than it
answers. In this context, given that explanation must run out at some point, axiar-
chism is not an absurd place for it to do so.
On Parfit’s account, axiarchism appeals to a second-order brute fact. The first-
order question is: ‘Why does this universe exist?’ The axiarchist answers: ‘Because it
is the best’. If we then ask why goodness functions as a selector for possible worlds,
there might be a second-order explanation, but there might not. Parfit’s suggestion is
that, if attempts to produce a second-order explanation fail, then we must accept
axiarchism as a brute fact. But we may still have made progress—so long as this
second-order brute fact is less implausible (easier to swallow as a ‘coincidence’) than
a first-order brute fact.39 Axiarchism thus puts great weight on the objective remark-
ableness of some feature of the actual world.
Our commitment to objective values supports axiarchism’s most striking onto-
logical commitment. In my teaching experience, the most common objection to
axiarchism, once students grasp what it actually involves, is that moral properties
cannot possibly play such a cosmic role—because morality is merely a human
artefact. One very common reaction is: ‘How can our subjective preferences make
the universe exist?’40 Axiarchism only gets off the ground if we accept objective moral
facts. For those who believe in such facts, axiarchism may seem unlikely or extrava-
gant, but it is not unintelligible. And in the context of our overall project, we can set
aside general scepticism about objective values.
Recent developments in meta-ethics also support axiarchism by separating it
from the metaphysical extravagance of Platonism. In contemporary meta-ethics,
naturalists and anti-realists often identify the position they jointly oppose—that
there are non-natural moral facts—as ‘Platonism’. This is typically a term of abuse.41
And John Leslie, the leading contemporary axiarchist, does identify axiarchism with
Platonism. But, as Parfit himself has recently emphasized, non-naturalists need not
include Parfit (Parfit, ‘Why Does the Universe Exist?’; Parfit, ‘Why Anything? Why This?’); and Rice, God
and Goodness. (Rice uses axiarchism as a step to BT: God exists because it is good that God exists.)
38
Parfit, ‘Why Does the Universe Exist?’, p. 3.
39
Parfit, ‘Why Does the Universe Exist?’, p. 3.
40
Nor is this reaction confined to undergraduates. John Leslie reports an equally dismissive response
from Richard Dawkins. (Leslie, Infinite Minds, p. 170.)
41
See, for instance, Jackson, From Metaphysics to Ethics.
COSMOLOGICAL ARGUMENTS 
be Platonists.42 Objective moral facts need no metaphysical Form of the Good to

make them true. Perhaps there simply are irreducibly normative brute facts. A moral
non-naturalist can thus embrace axiarchism without Platonism. Perhaps the same
non-natural properties both underlie human morality and explain the existence of
the cosmos.
Axiarchic non-naturalism is more ambitious than moral non-naturalism, but they
are on the same page. The two forms of non-naturalism are mutually supporting. Not
only does moral non-naturalism help axiarchism appear less crazy, but axiarchism in
turn supports non-naturalism. Axiarchism says the universe exists because it is good. If
moral facts can make a universe exist, then it is no great leap to think they can motivate
human beings. (It is some leap, to be sure—and we evaluate that leap in part III.) The
moral argument sketched in chapter 2 provides further mutual support, because we
may need to cite God or cosmic purpose to explain moral non-naturalism.
Even the staunchest moral realist may baulk at the idea that values somehow bring
a universe into existence. The axiarchist replies that, if we reject brute fact, then,
strange as it seems, axiarchism is the most parsimonious possible explanation. Any
other explanation must posit some mechanism whereby a good universe comes to
exist. Axiarchism cuts out the middle man. There is no multiplying of entities or new
properties. We already have the universe and its goodness. Axiarchism merely adds
‘because’. If goodness and existence are linked, then axiarchism is the simplest link.
As Parfit puts it: ‘[The axiarchist’s] use of “why” may seem utterly mysterious. But we
should remember that even ordinary causation is mysterious. At the most funda-
mental level, we have no idea how some events cause others.’43 Axiarchism cannot
remove mystery, but it does minimize it.
Opponents reply that axiarchism is very unparsimonious, even if we accept moral
realism and objective values. This is because it posits a whole new mode of explan-
ation. One common BT argument is that the only familiar alternative to scientific
explanation is personal explanation via the intentional actions of a free rational being.
We are familiar with both scientific and personal explanation. And these are the only
familiar explanations. Axiarchism is unintelligible, because we have no experience of
moral requirements directly bringing something about. If scientific explanation fails,
then personal explanation is the only intelligible alternative.44 (We return to the
positive case for personal explanation in section 3.7.)
42
Parfit, On What Matters, vol. 2, pp. 263–620.
43
Parfit, ‘Why Does the Universe Exist?’
44
A prominent contemporary defender of the dichotomy between scientific and personal explanation is
Swinburne. (Swinburne, The Existence of God, especially pp. 201–19.) Swinburne explicitly rejects axiar-
chism (Swinburne, The Existence of God, p. 47, footnote 16). However, this rejection is at variance with his
official methodology. Swinburne rejects axiarchism because it is unfamiliar. In his own cosmological
argument (the first step in his cumulative case), this should be a claim about prior probability (Swinburne,
The Existence of God, p. 146). As such, it is very odd. If we have no empirical evidence other than S, then
how do we decide that ‘God exists’ is simpler than ‘Ethical requirements are metaphysically efficacious’ or
‘The impersonal absolute gives rise to a physical cosmos’? If anything, axiarchism looks simpler than BT.
 THE CASE AGAINST ATHEISM
Axiarchism could retreat to reiterating Nozick’s dictum. We are in unfamiliar

territory, where unfamiliar explanations may be required. But, on its own, this dictum
can sound like an irrational mantra—an excuse for invoking any old nonsense. A much
stronger reply is to show that the axiarchist explanatory mode is neither unfamiliar, nor
lacking in independent motivation. To dispute the charge of unfamiliarity, axiarchists
might accuse their opponents of parochialism—historically, culturally, and even
within contemporary analytic philosophy. The demise of axiarchism, after a long
and distinguished philosophical heritage, is not based on anything a priori, but on
contingent matters of philosophical sociology. To someone immersed in Platonic
philosophy, in nineteenth-century Idealism, or in non-Western religious traditions
where the ultimate reality gives rise to a physical world via its own (ethical) nature,
axiarchism may seem as familiar (as intrinsically plausible) as either BT or global
naturalism. It is philosophical cultural imperialism to prefer these tales simply because
they are more familiar to us.
Nor is the search for alternative modes of explanation unknown in contemporary
philosophy. Logic, mathematics, morality, and other minds offer distinct modes of
explanation—and contemporary philosophers increasingly take those explanations
at face value. For our purposes, three points are especially salient: in both logic and
mathematics, non-contingent premises explain the existence of objects; in morality,
moral facts play an explanatory role; and personal explanation cites non-physical
properties of persons. In this company, the appeal to non-natural moral facts to
explain S is not so radical.45 While the thoroughgoing naturalist will find axiarchism
bizarre, the committed non-naturalist should not.
Consider the example I used to introduce the notion of a selector in section 3.5.
Some BTs accept God as a brute fact, but many do not. They must posit a selector
for God. Typically some divine property explains God’s existence. God exists
because God is a perfect being, a necessary being, and so on. This is very close to
axiarchism: God exists because it is good that God exists.46 Whatever we think of its
success, this axiarchic explanation for the existence of God is philosophically
familiar. Axiarchism applies that explanation to the physical world as well as to
God: both exist because it is good that they exist.
Some will no doubt insist that axiarchism is metaphysically extravagant if invoked
merely to explain S. But the cosmological argument is not our only metaphysical
Axiarchism only posits the existence of what is strictly necessary to explain S, whereas BT posits a person
with many other attributes. We should be guided by the conceptual coherence of axiarchism, not by its
familiarity to us.
45
While he never draws any explicit connection between the two, it is presumably no coincidence that
Parfit is a leading proponent of both axiarchism and moral non-naturalism.
46
See, e.g., Rice, God and Goodness. Leslie himself argues that, while BT needs axiarchism to explain
God, axiarchism does not need BT. An axiarchic atheist could consistently argue that God is impossible,
and that the physical universe that exists because it is good is a Godless one. (Leslie, Infinite Minds,
pp. 155–88.)
COSMOLOGICAL ARGUMENTS 
argument. Perhaps, once we see the many other things that axiarchism can explain,
its ontological commitments are easier to swallow. (Indeed, these other axiarchic
explanations may even persuade those who reject the axiarchic explanation of
S altogether, perhaps for the reasons explored in section 3.6.3.) Other arguments in
this book may also support axiarchism, not by offering new facts for it to explain, but
by reinforcing its picture of metaphysically real moral facts (chapters 5 and 6).
3.6.2 Is this the best possible world?

One obvious reply to any axiarchic explanation is that the alleged selector is not a
property of the actual world. Opponents allege that axiarchism must follow Leibniz,
and claim that this is the best of all possible worlds. If goodness leads to existence,
then presumably the world with most goodness will be the one that is actual. But we
can imagine better worlds lacking the bad aspects of this one. This is not the best
world. If axiarchism were true, this world would not exist. But it does. So axiarchism
is false.
BT, of course, faces an analogous objection. Some BTs reply that God simply
chooses, for no further reason, to create a world that, while not the best, is nonethe-
less ‘good enough’. Axiarchism cannot use this reply, because it has no substitute for
God’s arbitrary free choice. If value is directly efficacious, how could the best world
fail to exist? (The claim that divine freedom allows God to choose not to create the
best is itself very controversial, as we shall see in chapter 7.)
Leslie offers a different defence of axiarchism.47 He concedes that this is not the
best possible world. However, if the general axiarchic principle is that a possible
world exists because it is good, then all good possible worlds will exist. Axiarchism
implies a limited plenitude, where every good possible world exists.48 We need not
explain why this world exists alone—because it does not.
Opponents of axiarchism might still ask whether this world is good enough.
Perhaps its evils and imperfections render our world bad overall. Even if it need
not be the best, our world must still be better than nothing. In particular, given
plenitude, all the good things in our world already exist in many other better worlds.
Is it still good to add this particular combination of good and bad?
Parfit regards the argument from evil as the most serious objection to axiar-
chism.49 His response is to consider alternative selectors. If this world is not the
best, then it may still have some other unique and remarkable property. And that
property may be the selector. For instance, Parfit himself suggests that, if this world
turns out to be governed by the most beautiful possible mathematical laws of physics,
then, rather than accepting this as a remarkable coincidence, we might conclude that
our world exists because it has this feature.
47
Leslie, Infinite Minds, pp. 15–19.
48
On Leslie’s own Idealist metaphysics, these worlds all exist as ideas in the divine mind.
49
Parfit, ‘Why Anything? Why This?’
 THE CASE AGAINST ATHEISM
AP is untroubled by the argument from evil. Indeed, that argument is a central

plank of our case against BT in part II. We cannot explain S, or anything else, by
appealing to human-centred values. AP rejects those values, not the underlying
axiarchic explanation. If we adopt a non-human-centred standard of excellence,
then there is no reason to doubt that this world is the best—and certainly no reason
to deny that this is a good world.
Parfit’s alternative selectors seem to be competitors for AP, as they explain
S without mentioning values or purpose. On the contrary, I believe that these
alternatives support AP. Alternative selectors cannot be separated from objective
values. As Parfit himself explains, not every unique feature of a possible world is a
plausible selector. After all, every possible world has some unique features. ‘The
universe exists because it contains fifty-seven inhabited planets’ is a bad explanation,
even if the universe does have this feature—because there is nothing remarkable
about it. A selector must be an objectively remarkable feature, as well as an unlikely
one. I argue in chapter 4 and part III that it is very hard to separate objective
remarkableness from objective value. Indeed, the former presupposes the latter.
Any credible Parfitian selector will thus give AP everything it needs.
We have thus far assumed that the axiarchist comes to the cosmological argument
with independent value commitments, and then uses these to explain S. We might
instead reverse the order of epistemic priority. Perhaps objective values are among the
conclusions of the cosmological argument, rather than its premises. Suppose we
observe that this universe has the uniquely remarkable property P, and conclude
that P is the selector for this universe. We then derive the claim that, at least within
this universe, P is valuable. (This is not voluntarism, because the ontological priority
is unchanged. The independent value of P is still what explains S. But our knowledge
of that value is a conclusion of our cosmological argument, not a premise.) We
explore this possibility of epistemic reversal at greater length in chapter 4.
3.6.3 Is something better than nothing?
AP can avoid—or at least weaken—general objections to axiarchism. We now
consider a more particular objection: even if we have some knowledge of objective
values, we don’t know that something is better than nothing. Even if we can compare
the values of different possible physical universes, it is meaningless to compare the
values of something and nothing. Therefore, whatever else it can explain, axiarchism
cannot explain S.
If subsequent chapters demonstrate that the axiarchic explanation of other facts is
sufficiently compelling, then we could, of course, simply dispense with the cosmo-
logical argument altogether. Axiarchism tells us nothing about our ultimate question,
but it still earns its metaphysical keep in other ways. In the present section, however,
we explore the resources available to the axiarchist who does wish to explain S.
We first note that axiarchism need not claim that something is better than nothing
period. The axiarchist says that this world exists because it is the best possible world
COSMOLOGICAL ARGUMENTS 
(or, at least, one good world among many). But it does not follow that the mere
existence of a physical universe has any value at all. S may be true in many other
possible worlds that are not better than nothing—and, indeed, in many that are much
worse than nothing. The axiarchist need only claim that this possible world contains
more value than the empty world. There is a physical universe because this physical
universe is good. Our selector can be more fine-grained, more specific, or indeed quite
different from, what we seek to explain. (This is another possibility that comes to the
fore in chapter 4.)
Axiarchism need only claim that this world is better than one where nothing exists.
But many people still regard this comparison as meaningless. I now sketch several
possible axiarchic defences of this specific comparison between existence and non-
existence. First, axiarchism could simply appeal to moral intuition. While many
people find the bald claim that something is better than nothing unappealing—or
unacceptably theory-laden—many others feel the force of the more restricted com-
parison between this world—with all its rich variety—and an empty void.
As ever, appeal to brute intuition is not the strongest form of argument. Fortunately,
contemporary moral theory provides axiarchism with more theoretical defences. Two
of these draw on consequentialist responses to loosely analogous debates that arise in
connection to our obligations to future people.50 The parallel objection there is that,
while it makes perfect sense to compare the values of different possible lives, it makes
no sense to compare existence with non-existence from the point of view of an
individual person. We can ask whether I would have been better off if I had been
born in Scotland, but not whether I would have been better off if I had never been born.
In the context of the future people debate, defenders of objective values offer two
main replies. The first is substantive.51 The objection to comparing an actual human
life to non-existence assumes that human well-being has only conditional value.
Certain things (such as pleasure or preference-satisfaction) are good for a person if
she exists, but they cannot make it good—in any independent sense—that she exists.
But many contemporary moral philosophers deny that the value of well-being is
purely conditional. In particular, under the objective list theory, a good human life is
one that contains certain objectively valuable components (such as the successful
pursuit of independently valuable goals) whose value is not reducible to the pleasures,
desires, or experiences of the agent. These objective values are independent, not only
of the agent’s mental states, but also of her existence. Independent values can explain,
not only why it is better for me to live one life rather than another, but also why it is
better that this life is lived. Similarly, independent values might explain why it is good
that this physical universe exists.
Our moral commitments support this substantive argument. However, the argu-
ment is ambitious, and will not convince everyone. Some will still object to the
50
For references to these debates, see Mulgan, ‘Consequentialism’.
51
Here I draw on McMahan, ‘Wrongful Life’.
 THE CASE AGAINST ATHEISM
assumption that values are independent, not only of humanity, but of the very
existence of a physical universe. Fortunately, there is another argument the axiarchist
can borrow. While it also draws on the future people literature, this second argument
is formal rather than substantive.52 We begin with a future people analogy. We have
many very robust intuitions regarding the comparative values of possible lives, and
many other well-supported moral intuitions. The best way to account for all of these,
within a logically coherent moral theory, is to first posit, for each individual, a
complete scale of objective values, allowing us to compare all possible states of affairs
from her point of view—including possible states where she herself does not exist.
We then construct a complete aggregate scale, evaluating all possible states of affairs
from the perspective of humanity as a whole—including worlds with no humans. We
thus extrapolate from reliable intuitions (comparing different possible lives) to
marginal cases where our intuitions are less secure (comparing existence to non-
existence).
If a moral theory built on objective values best respects our secure intuitions, then we
can reasonably extend that theory to cases of intuitive uncertainty. Suppose chapter 4
convinces us that axiarchism is plausible when it appeals to the comparative value of
different physical universes. We can then argue as follows. We observe that, of all the
non-empty possible worlds (those containing physical universes) this is the best. We
conclude that this is no coincidence: this world exists because it is the best physical
universe. On the scale of objective value, this physical universe ranks above all others. If
we then ask why there is any physical universe at all, we can offer the same answer. The
empty world must fit somewhere into that scale of objective value, and this world exists
instead of the empty world because it is better. The only alternatives are either to
conclude that the empty world is better than all physical universes, or to say that it
somehow stands outside the comparative ordering of possible worlds.
So long as some features of our cosmos are objectively valuable, we can justifiably
invoke objective values to explain S. For instance, if we need axiarchism to explain
fine-tuning (chapter 4), we can then use it explain S. Indeed, the axiarchist can even
accept that there is no prior reason to believe that this physical universe is better than
nothing. As we saw earlier, objective values may be among the conclusions of the
cosmological argument, rather than its premises. S is only explicable if something is
better than nothing. Axiarchists conclude that something is better than nothing,
because they observe S. In either case, its ability to explain S still counts in favour
of axiarchism.
3.6.4 Axiarchism and cosmic purpose

We have just argued that any plausible axiarchic (or selector-based) explanation
yields objective values; and part III argues that any objective values sufficient to
52
Here I draw on Broome, Weighing Lives.
COSMOLOGICAL ARGUMENTS 
explain S also yield a cosmic purpose. Axiarchism thus supports a cosmic purpose.
But is it a cosmic purpose that AP can borrow?
Axiarchism says only that this universe exists because it is good. This appears to be
neutral between AP and human-centred purpose. Therefore, while the axiarchic
explanation of S won’t help AP against BT, AP can borrow that explanation in its
case against atheism—unless human-centred elements play some hidden role.
Human-centred axiarchism is superior to AP axiarchism if it either explains more
things or offers better explanations. As to the first, chapter 4 argues that, while its
rival seems to explain extra facts (notably the existence of human beings) this
advantage is illusory. Furthermore, part II argues that even if human-centred axiar-
chism does have some explanatory advantages, these are outweighed by AP’s super-
ior explanation of other facts—notably those relating to scale, evil, and religious
diversity. AP offers a better overall explanation of the striking features of this
universe.
Human-centred axiarchism offers a better explanation of S if its values are more
plausible than AP’s. And human-centred values do seem more plausible—especially
to human beings! To borrow axiarchism, AP must defend its own values. Part III
addresses this challenge.
Axiarchism offers one possible explanation of S. Our moral commitments support
that explanation. If it can defend its own value claims, AP can borrow that explan-
ation. Its ability to explain S counts in favour of axiarchism. While the cosmological
argument for axiarchism is not decisive on its own, it can form part of a compelling
cumulative case. But AP is not committed to axiarchism. It can borrow other
explanations instead. In particular, AP can borrow from BT.
3.7 Evaluating Benevolent Theism

BT’s very familiarity can prevent us from seeing it clearly. One aim of this book is to
see BT afresh—to separate its commitment to cosmic purpose from its human-
centred elements. This is why I began with axiarchism, and will now explicate BT
by comparison with that less familiar alternative.
As with axiarchism, we have four questions to ask. (1) Does BT offer a good
parsimonious explanation for S? (2) What do our moral and methodological com-
mitments add to the plausibility of that explanation? (3) Does BT yield a cosmic
purpose? (4) Can AP endorse that purpose—and thus borrow the BT explanation of
S? Section 3.7.1 addresses (1) and (2); while section 3.7.2 covers (3) and (4).
3.7.1 The BT explanation
Of our two forms of BT, intellectualism is closer to axiarchism. Intellectualism and
axiarchism both involve objective values, and a non-scientific mode of explanation.
Here, our commitments to objective values and methodological pluralism support
intellectualism as well as axiarchism. However, intellectualist BT differs from
 THE CASE AGAINST ATHEISM
axiarchism: it makes different claims about the content, strength, and nature of
objective values; it posits an additional metaphysical entity (God); and it offers a
different mode of explanation. Do these differences give intellectualism an advantage
over axiarchism?
The intellectualist explains S by showing that God has reason to create a physical
universe. Like the axiarchist, she need not claim that any possible universe is better
than nothing—only that this universe is better than nothing. Just as a direct selector
may be more discriminating and more fine-grained than what we seek to explain, so
God’s reason for creation might be the goodness of some specific feature of this
universe. (We can then explain other facts about our universe, beyond its mere
existence.)
Intellectualist BT must address a new challenge. The axiarchist need only argue
that this world is better than nothing. The intellectualist must defend the same
claim—and for this she can borrow the arguments of section 3.6.3. But she must
also argue that this world contains good things that would not exist if God existed
alone. A good BT explanation must rebut the sufficientist objection that, being
perfectly self-sufficient, God needs no (imperfect) world, and thus would not create.
Things are perfect when only God exists. A physical universe could only make things
worse.
The sufficientist objection is not decisive. Intellectualist BT offers many possible
reasons for divine creation.53 For an omnipotent God, creation is effortless. So any
positive value is sufficient. God might create a deterministic universe simply to
admire it; or create some particular indeterminate universe just to see how it turns
out. Or perhaps Idealism is correct, and our world is God’s imagining of it. Wouldn’t
we expect a perfectly imagined world to feel real to its imaginary inhabitants?54 Or if
BT countenances gods of less than infinite power and imagination, it can explain
S more directly, as a physical universe is then more likely to contain good things not
found in the imperfect imagination of a lesser God. (We return to God’s reason to
create in section 3.7.2, when we ask whether human beings provide the best such
reason—one that AP cannot borrow.)
Intellectualist BT says that this physical universe exists because God, recognizing
its objective value, elected to create it. If we are already committed to both God and
objective values, then this looks like a very good explanation. But is the ability to
explain S itself a reason to believe that God exists? Or are the ontological demands of
BT too great? Intellectualism initially seems less parsimonious than axiarchism.
53
Swinburne, The Existence of God, pp. 99–109.
54
Axiarchism and BT both fit well with Idealism, a philosophical tradition that has fallen out of favour
in recent analytic philosophy. A willingness to consider unfashionable traditions thus assists both of the
solutions that AP seeks to borrow. We return to Idealist themes several times in subsequent chapters, and
explore Idealist arguments for immortality at length in chapter 10.
COSMOLOGICAL ARGUMENTS 
Indeed, it seems very unparsimonious. Isn’t the introduction of a perfect being the
ultimate metaphysical extravagance?
Many atheists reject God as unintelligible, meaningless, internally contradictory,
or absurd. Any BT explanation is too metaphysically costly. At the other extreme, of
course, BT replies that God’s existence is necessary. BT thus explains S at no
metaphysical cost. Our methodological commitments reject both extremes. We
grant that God is a coherent, and not metaphysically absurd, entity. But, when
evaluating the cosmological argument, we proceed on the supposition that we cannot
prove that God must exist. (We return to the necessity of God in chapter 6.)
Suppose we agree that God is the least metaphysically extravagant possible being
who can create a physical universe in response to objective values. (Whether this God
must be, in all respects, the traditional BT God is, of course, another matter.) Such a
God seems coherent. But is it too metaphysically extravagant? Does BT’s ability to
explain S justify its metaphysical cost? In particular, is BT superior to axiarchism?
BT claims two related advantages over axiarchism: its objective values are less
‘weird’, and its mode of explanation is more familiar. The intellectualist claims that
her objective values are less metaphysically loaded than the axiarchist’s, as they do
not directly bring a cosmos into existence. We are familiar with values that ‘work’ in
the world only because agents respond to them; while axiarchism’s directly effica-
cious values are unfamiliar (to say the least). BT thus offers the only familiar or
intelligible alternative to scientific explanation. I argued earlier against rejecting
axiarchism altogether on grounds of unfamiliarity. But we might nonetheless feel
that, if two explanations are equally effective, the more familiar one has a clear
explanatory advantage. We must then ask whether a personal explanation really is
familiar when it involves God.
Atheists will reply, of course, that divine personal explanation is not familiar—
precisely because it involves such an unfamiliar person. Even if divine creation is not
a new mode of explanation, it certainly involves a new kind of entity. Some BTs deny
that God is an unfamiliar entity. Swinburne, for instance, argues that God is just like
one of us, only much less complex.55 We are persons with bodies, limitations, and
imperfections. God is a person with none of these. This argument will especially
appeal to substance dualists, for whom human persons are spiritual entities. Any
argument requiring the truth of substance dualism has limited dialectical value,
simply because atheists are typically not dualists. However, I believe the argument
can succeed using only conceptual dualism, where mental and physical predicates
capture different kinds of facts about persons. In particular, if our concept of a person
is moral, then God is a familiar being: what really matters about God is familiar. Like
us, God has purposes, reasons, and values. And if we also embrace moral non-
naturalism, then God’s resemblance to humans includes recognition of non-natural
55
Swinburne, The Existence of God, pp. 96–109.
 THE CASE AGAINST ATHEISM
facts. This is one place where dissatisfaction with naturalist approaches to ethics
supports both BT and AP. (This argument has affinities with Nagel’s argument that
atheism cannot explain the existence of rational creatures who respond to moral
reasons, to which we return in section 4.2. We also return to the general affinity
between BT and dualism in chapter 10.)
If we deny that God’s existence is logically necessary, then we must accept some
brute fact. The atheist accepts S as brute: this physical universe simply exists
uncaused. The axiarchist accepts as a brute fact that objective values are directly
efficacious. This universe exists because it is good—and for this fact there is no
further explanation. For BT, God’s existence is a brute fact. The intellectualist also
accepts brute facts about objective value; while the voluntarist’s extra brute fact is
God’s decision to create.
BT is superior to both axiarchism and atheism, as an explanation for S, if and only
if a brute God is easier to swallow than either brute axiarchism or an uncaused
physical universe. Ambitious BTs, such as Swinburne, seek to attach objective
probabilities to these different brute facts: axiarchism has lower intrinsic plausibility
than BT because it is less familiar; while an uncaused universe is less likely than an
uncaused God because it is so much more complex.56 I am suspicious of these
pretensions to numerical precision. Perhaps the most we can say with confidence is
this. If you find that the existence of this physical universe cries out for explanation,
then you are likely to find a brute God more appealing. If we already believe in
objective values, and are willing to countenance non-scientific explanations, then a
person with simple, interrelated properties who recognizes and responds to those
values may offer a more parsimonious explanation than the claim that the existence
of such an objectively remarkable world is a coincidence. Our reasons for finding BT
a plausible answer thus dovetail with our reasons for asking the ultimate question in
the first place. If you find the ultimate question compelling, and you reject axiar-
chism, then you may find BT the least implausible alternative.
Thus far, we have focused on intellectualist BT. Voluntarist BT dispenses with
independent objective values. God’s creative decision has no external motivation.
Voluntarism thus has an ontological advantage and an explanatory disadvantage. We
begin with the latter. Without objective values, how can we know whether a volun-
tarist God would create? This difficulty is especially acute if we seek to attach precise
probabilities to God’s creation of particular possible worlds. (It is thus no surprise
that Swinburne favours intellectualism.57) A more modest voluntarism argues that,
even though we cannot assign definite probabilities, divine creation is still not
surprising. We are familiar with a person choosing for no independent reason;
56
Swinburne, The Existence of God, p. 47, footnote 16.
57
Swinburne, The Existence of God, pp. 99–109. In his ‘Second Reply to Grunbaum’, p. 922, Swinburne
explicitly acknowledges that, without independent objective values, his arguments would not get off the
ground.
COSMOLOGICAL ARGUMENTS 
whereas we are not familiar with objects that come into existence either for no reason
at all, or simply because this would be good. Unexplained divine choice is better than
either the atheist’s uncaused universe or the axiarchist’s directly effective values.
Voluntarism’s advantage is that it is more ontologically parsimonious than either
axiarchism or intellectualism. God explains moral facts, rather than simply recog-
nizing them. For those unwilling to accept objective moral facts without further
explanation or metaphysical grounding, BT voluntarism offers the best explanation
of S. (Just as voluntarism is also the best meta-ethical position for those moral realists
who are unwilling to accept unsupported moral facts, as we saw in chapter 2.)
Unfortunately, ontological parsimony in one dimension can lead to metaphysical
extravagance elsewhere. If God explains moral facts, then God is less like an ordinary
person.
Even if God is familiar in some ways, God remains a new entity—one apparently
not needed by science or everyday life. Unless the ultimate question is especially
urgent, it seems extravagant to posit a divine person simply to explain S. Here, we
might invoke BT’s other explanatory (and non-explanatory) resources, as outlined in
subsequent chapters. As with axiarchism, if BT does explain other things or play
other roles, then we can cite its ability to explain S as an extra mark in its favour.
The argument from S to BT is far from conclusive. Some will prefer to accept S as a
brute fact. But BT is a serious contender—especially for those with our substantive
and methodological commitments. (And not only for them. Unlike axiarchism, BT
has many actual adherents.) Our project does not require us to choose between these
alternatives explanations for S—either between axiarchism and BT, or between
intellectualism and voluntarism. My aim in this section has merely been to show
that BT offers one plausible explanation of S, and that our commitments support that
explanation. Our next question, as with axiarchism, is whether BT offers a cosmic
purpose that AP can borrow.
3.7.2 Can AP borrow BT’s cosmological argument?

Swinburne ends his own discussion of the cosmological argument with the following
summary:
There is quite a chance that, if there is God, he will make something of the finitude and
complexity of a universe. It is very unlikely that a universe would exist uncaused, but rather
more likely that God would exist uncaused. Hence the argument from the existence of the
universe to the existence of God is a good C-inductive argument.58
Suppose we agree. Does this traditional argument for the existence of the BT God
offer a cosmic purpose that AP can borrow?
In either its intellectualist or voluntarist form, BT almost trivially yields a cosmic
purpose. If God created the universe for a reason, then that reason gives the universe
58
Swinburne, Is There a God?, pp. 52–4.
 THE CASE AGAINST ATHEISM
a purpose. Even if God created for no (prior) reason, divine creation is typically
thought to bring a cosmic purpose into existence. While there are some intriguing
issues here, to which we return in part III, we set them aside for now, and assume that
BT does imply cosmic purpose. The real question is whether AP can borrow that
purpose.
Nothing in the idea of a person who creates the cosmos is inconsistent with
AP. (As we’ll see in chapter 6, AP can even accept a morally perfect God.) So it
seems that, if AP can show that the true objective values are not human-centred, then
it can borrow either intellectualism or voluntarism. (If AP cannot show this, then it
fails anyway.)
Unfortunately, this argument is too swift. Connections between God and human
beings play two distinct roles in many cosmological arguments: human beings
provide one common reason for God to create; and our personhood provides the
model for God. Both roles support human-centred theism over AP.
To defeat the sufficientist, BT needs a reason for divine creation. One popular
answer is the presence of free, finite, rational beings who can choose between good
and evil—notably human beings. Unlike God, we can overcome obstacles, avoid
temptation, and do good even when we could have done evil.59 We bring a new kind
of value into the universe. And, because God cannot predict or control our behaviour,
even divine imagination is no substitute for creation.
AP cannot accept this human-centred reason. It therefore needs alternative
reasons for divine creation. But as we saw earlier, many such reasons are available:
even a perfect God might create because a universe is beautiful, unpredictable,
elegant, complex, or otherwise valuable. Or AP might borrow the Idealist thought
that God’s imagining is God’s creation. AP could also develop a non-human-centred
analogue of BT’s own reason. Perhaps God creates so that, in some far-flung corner
of the universe, there will be creatures far superior to ourselves, who possess real
freedom or really understand the purpose of the universe. (We return to these
possibilities in chapters 4 and 7.)
AP can also reject the presumptions behind the sufficientist challenge. BT is
vulnerable to this challenge because it combines two claims: God is perfect, and the
world is very imperfect. But AP is committed to neither claim. An AP God might be
far from perfect, and thus need to create in order to realize some purpose. On the
other hand, once we transcend human-centred values, there is no reason to deny that
this is the best possible world, and therefore it might be more perfect than its creator.
Finally, even if the removal of one possible divine reason does weaken AP’s case
against BT, this loss is compensated by AP’s other advantages. In particular,
chapter 8 argues that the evils of this world show that concern for human beings
cannot be God’s reason to create it.
59
See, e.g., Swinburne, Is There a God?
COSMOLOGICAL ARGUMENTS 
I conclude that the removal of human-centred reasons does not prevent AP from
borrowing the BT cosmological argument.
BT’s second explanatory advantage over AP is that it provides a model for God. BT
is superior to axiarchism because it is more familiar. This, in turn, depends on God’s
resemblance to familiar persons—namely human beings. But if we, of all the known
universe, most resemble its creator, isn’t it natural to believe that that creator cares
for us?
AP has several replies. First, a theme of this chapter is the need to take unfamiliar
explanations seriously. BT could thus claim to satisfactorily explain S even if its
creator is a very unfamiliar entity. And other arguments for BT (notably the
ontological arguments explored in chapter 6) might yield a God who is very unlike
human persons. Second, AP might seek to reduce the gap in familiarity between its
God and BT’s. Some arguments make BT’s God less familiar, while others may
render AP’s God more familiar. In particular, I argue in chapters 5 and 9 that AP
offers the most natural interpretations of mystical experience and religious diversity;
while chapter 7 suggests that the discovery of superior extraterrestrial rational beings
might point towards a non-human-centred God. Third, AP can accept that our very
faint resemblance to God is prima facie evidence that God cares for us, but then insist
that this evidence is simply outweighed by counter-evidence such as scale, evil, or
religious diversity. Despite initial appearances, the best explanation for all the evidence
is that God does not care for individual human beings. As ever, our cosmological
argument must be evaluated as part of a complex cumulative case for AP.
A plausible BT explanation of S need not either yield a human-centred reason for
creation, or rest on controversial metaphysical claims about resemblances between
human beings and God. So AP can borrow it.
AP also strengthens the cosmological argument. AP is more parsimonious than
BT, because it can avoid BT’s most controversial metaphysical claims about God. An
AP creator need not be omnipotent, perfect, or even a person. (On the other hand, as
I argue in chapters 5 and 6, AP need not deny these claims either.) To explain S, AP
only posits something beyond the physical universe that either responds to or
generates objective values.
Our discussion of the cosmological argument has been comparatively brief, and
has omitted many details of the current debate. The aim has been to establish that the
argument is worthy of serious attention, and in a form that AP can borrow.
3.8 Conclusion
There are good cosmological arguments for both BT and axiarchism. Each can
explain S. But there is a better cosmological argument for AP—precisely because
AP can borrow both BT and axiarchism. Those who reject axiarchism as unfamiliar
may find BT more attractive; while those who reject the metaphysical extravagance of
BT may prefer axiarchism’s elegant simplicity. S raises the probability of BT,
 THE CASE AGAINST ATHEISM
axiarchism, and AP. But it raises the probability of AP more. S raises the probability
that there is some cosmic purpose; and nothing about S tells us whether that purpose
is human-centred or not. And the most plausible cosmological argument delivers
Normative AP, where the cosmic purpose is intimately linked to objective values.
The cosmological argument has two real limitations. The first is that it tells us so
little about the cosmic purpose. How exactly would one respond positively to the fact
that the actual world exists because something is better than nothing? By valuing
contingent things (all of them?) simply because they are contingent? For more
information about the cosmic purpose, we need other arguments.
A second limitation is that the cosmological argument is obviously not decisive.
Although our various substantive and methodological commitments strengthen it at
crucial points, this will not suffice to dissuade those who prefer to accept the physical
universe as a brute fact—or to faithfully await the emergence of some as yet
unimagined purposeless explanation. We have three radically different possibilities:
axiarchism, BT, and an uncaused physical universe. Some hard-core atheists find the
first two possibilities unintelligible, while some hard-core BTs think that only the
second is possible. AP says that all three are intelligible, but it is a matter of
judgement or philosophical temperament how we decide between them. It is not
unreasonable to accept S as a brute fact. AP is on stronger ground if it explains things
that simply cannot be brute facts. This is the task of the fine-tuning argument, to
which we now turn.
4
Teleological Arguments
On a purely materialist understanding of biology, consciousness would have to be

regarded as a tremendous and inexplicable brute fact about the world.
[Thomas Nagel, Mind and Cosmos, p. 44]
To me, the contrived nature of physical existence is just too fantastic for me to
take on board as simply ‘given’. It points forcefully to a deeper underlying
meaning to existence. Some call it purpose, some design. These loaded words,
which derive from human categories, capture only imperfectly what it is that the
universe is about. But that it is about something, I have absolutely no doubt.
[Paul Davies, ‘The Appearance of Design in Physics and Cosmology’, p. 152]
Teleological arguments are based on purpose or design. Teleological and cosmo-

logical arguments both posit God to explain something. But now we seek to explain
some fact about (or within) the universe, rather than the mere fact that a universe
exists. In this chapter, I construct a composite teleological argument that combines
features of several distinct arguments. I conclude that teleological arguments
establish a strong case both that the universe has some purpose and that this
purpose is connected to cosmic values; but they do not support a human-centred
purpose.
In the popular culture, teleological arguments are the most common, popular, and
influential arguments for the existence of God. Recent commentators have also
developed teleological arguments aimed, not at BT, but at more general conclusions
about purpose, design, or meaning. Especially striking is the emergence of explicitly
purposive accounts within science itself. This is fertile ground for AP.
4.1 Varieties of Teleological Argument

Teleological arguments fall into two broad categories. Specific design arguments
explain the existence of some particular kind of thing, or some specific feature of
particular things. Specific design arguments have been offered for all of the
following: stable physical objects, complex molecules, stars, life, biochemical sys-
tems, complex organisms, ecosystems, conscious life, intelligent life, human beings,
rational beings, moral beings, creatures able to comprehend the point or purpose of
 THE CASE AGAINST ATHEISM
the universe (such as physicists or theologians), the human eye, the fact that our
eyes and our feet point in the same direction, or even the fact that the moon looks
like a face.1
Arguments from order, by contrast, explain some general feature of our universe,
such as the fact that it appears to be precisely fine-turned so that life can emerge
‘naturally’, or the fact that it is governed by simple mathematical laws that are
intelligible to some of its inhabitants.
Each teleological argument begins with a fact (F) to be explained. As with the
cosmological argument, we have three general responses: (1) F is a brute fact,
requiring no explanation; (2) F can be explained without mentioning cosmic pur-
pose; and (3) F can only be explained via cosmic purpose. Explanations appealing to
cosmic purpose could be axiarchic, BT, or impersonal. For simplicity, I concentrate
on BT’s explanation, where F is true because God chose to create a world where
F is true.
Opponents of cosmic purpose must choose between brute fact and non-purposive
explanation. While brute fact is a popular response to the cosmological argument, it
is much less popular here. This is the most significant difference between the two
arguments. Almost no one defends this response to any specific design argument. (As
we’ll see in section 4.2, one exception is that some atheists do offer brute fact
responses to specific BT claims about human beings.) The existence of complex
molecules or human beings is not a brute fact. And while brute fact is more popular
for arguments from order, it is still a minority option there too.
Most atheists instead offer scientific explanations. The facts invoked by teleological
arguments are legitimate objects of scientific investigation in a way that S is not.
Science progresses at a pace, and with a degree of definiteness, that philosophy lacks.
The teleological argument has been radically transformed over the last two centuries
by two sciences in particular: evolutionary biology and cosmology. I shall argue that
both scientific developments support AP. Evolutionary biology undermines specific
design arguments (especially those for a human-centred purpose), while cosmology
invigorates arguments from order.
Prior to Darwin, the classic example of design was the human being. Darwin
showed that the existence of human beings is not remarkable given the existence of
earlier species. Attention then shifts to those earlier species, gradually moving back to
the ‘simplest’ organic forms. Two general themes emerge: (1) the existence of human
beings is not remarkable given the existence of the most ‘primitive’ forms of life; and
(2) even the most primitive forms of life are themselves much more remarkable (in
the everyday sense) that we might have expected. Their existence, in turn, cries out
for explanation.
1
For the last two examples, attributed to Paley and Kepler respectively, see Sober, ‘The Design
Argument’, pp. 31 and 27.
TELEOLOGICAL ARGUMENTS 
Philosophers’ claims that some specific physical thing has been designed have a
habit of coming unstuck, as each hitherto inexplicable fact about how the universe
operates is explained in terms of statistical regularities, natural laws, and pre-existing
conditions. This history of philosophical failure and scientific success should make us
very wary of any claim of the form ‘X will not be explained by scientists in terms of
some more basic substance(s), together with the immense age of the universe, and
the laws of nature and/or statistical probability.’ The specific design argument is
looking shaky.
On the other hand, as scientific success shows us the remarkable complexity of our
universe, the intuitive force of the fine-tuning argument strengthens. The more facts
we can explain given the way things are, the more remarkable it seems that things are
that way. In a universe like this, it is not remarkable that life emerged. But it is
remarkable that there exists a universe where that emergence is not remarkable.
And this fact seems much less amenable to a naturalistic explanation. As evidence of
fine-tuning builds up—and cries out for explanation—the claim that the universe has
some purpose grows stronger.
Since Darwin, our epistemic situation has thus been moving in a direction
uniquely congenial to AP.
4.2 The Human Features Argument

Specific design arguments continue to receive attention. Two that are prominent in
popular debate are the notorious ‘Intelligent Design’ (ID) arguments, and the argu-
ment that evolution cannot explain the origin of life itself. AP could borrow either of
these arguments, because their conclusions only point to some intelligent designer,
not to BT. (Proponents of ID, in particular, often endorse BT. But this is said to be
independent of their ‘scientific’ work.) Unfortunately, ID is not credible2; and, given
the relentless march of science, the current lack of a scientific explanation of the
emergence of life is not sufficient to ground a successful specific design argument.
I explore instead a more philosophical argument. According to the human features
argument (HFA), some special features of human beings both (1) cannot be given a
non-purposive explanation; and (2) cannot be accepted as brute facts. (Alleged
special features include: rationality, language, consciousness, freedom, the ability to
form mental representations of the external world, the possession of a soul separate
from our physical body, and personal immortality.) I shall argue that our moral
commitments lend surprising support to HFA, and that AP can borrow this
argument.
2
Against ID, I myself am especially persuaded by Kenneth Miller’s critique. (See, e.g., Miller, Finding
Darwin’s God; Miller, ‘Answering the Biochemical Argument from Design’.) The main philosophical
problem is that ID rests on a defective inference from present uselessness (part P has no independent
utility) to past uselessness (forebears of P had no independent utility).
 THE CASE AGAINST ATHEISM
While most HFAs defend BT, Thomas Nagel has recently defended a distinct non-
theist version.3 I focus on Nagel’s argument, which complements my own project.
Following Nagel, I call the opponent of HFA the ‘Darwinian’.4 The Darwinian insists
that every feature of human beings either has an evolutionary explanation, or can
easily be admitted as a brute fact. Nagel argues instead that human beings are natural
phenomena that do not fit our current paradigm. Darwinian evolution, whose sole
criteria are survival and reproduction, cannot explain the emergence of conscious,
reason-following, value-discerning beings such as ourselves.
One popular HFA appeals to dualism.5 Each human being consists of a physical
body and a non-physical soul. Our ability to represent, reason, and act comes from
our souls, not our bodies. These abilities therefore cannot be explained physically, but
only by citing some other spiritual substance, such as God.
A good specific design argument should explain some fact that is common ground
between BT and atheism. After all, BT makes many controversial metaphysical
claims about human beings that atheists simply deny. If BT invokes God to explain
why we have immortal souls, the atheist is entirely unmoved. Why explain what does
not exist? Darwinians rejects dualist HFAs on similar grounds. Human beings are
physical, not spiritual, and therefore there is nothing ‘non-physical’ to be explained.
Our commitments point in opposite directions here. Methodological pluralism
urges us to take dualism seriously, especially once we reject a presumption of global
naturalism. On the other hand, our aversion to caprice cautions against the idea that
human beings are metaphysically special—that we somehow stand out from the rest
of creation. We return to dualism in more detail in chapter 10. In the meantime,
however, we seek a less self-aggrandizing HFA.
Nagel’s HFA begins from the inside: asking what it is like to be a conscious reason-
following moral being. His phenomenological explorations yield accounts of human
phenomena which, he argues, make them unsuitable for Darwinian explanation. One
example is consciousness. Drawing on his own earlier work in philosophy of mind,
Nagel argues that ‘[o]n a purely materialist understanding of biology, consciousness
would have to be regarded as a tremendous and inexplicable brute fact about the
world’.6
By contrast, the Darwinian studies human beings from the outside, as she would
any other species. She thus focuses on observable behavioural manifestations of
consciousness, and presumes that some reductionist materialist analysis of con-
sciousness is available.
3
Nagel, Mind and Cosmos.
4
Nagel is an atheist, and he explicitly calls not for a rejection of scientific naturalism, but for a new
scientific paradigm. So neither ‘atheist’ nor ‘naturalist’ is an appropriate label for his opponent.
5
See, e.g., Swinburne, The Existence of God, pp. 192–212.
6
Nagel, Mind and Cosmos, p. 44.
TELEOLOGICAL ARGUMENTS 
Like Nagel, our project takes the internal human perspective seriously. And we
resist the philosophical hegemony of reductionist materialism. Nagel’s consciousness-
based HFA is thus worthy of further attention. However, as we aim to remain neutral
regarding controversies elsewhere in philosophy, we want to avoid controversial
claims about consciousness. Fortunately, as we shall see, Nagel’s other HFA is even
closer to our project.
Before leaving consciousness, however, we first explore the structure of Nagel’s
argument. The failure of Darwinian explanation is not sufficient for a successful
HFA. If Nagel insists on a non-reductionist analysis of consciousness, the Darwinian
will reply with brute fact. Evolution has produced creatures who just happen, in
addition to their biologically explicable properties, to possess consciousness. But so
what? What is so special about consciousness? Evolution throws up many anomalies.
Not every biological fact has an explanation in terms of natural selection. (To think
otherwise is to commit the familiar ‘adaptationist fallacy’.) Perhaps consciousness is
just another anomaly. Like the existence of the physical universe itself, the emergence
of conscious beings is a brute fact.
A successful HFA must rebut this brute fact response. Nagel’s reply to brute fact
explicitly cites objective values. Our non-Darwinian features cannot be accepted as
brute facts because they are objectively significant. Nagel’s objective values are
essential to his HFA. This is where our commitments most clearly support his
argument.
Crucially, to defeat the Darwinian, Nagel needs a non-naturalist account of
significance. Moral naturalism explains ‘objective’ significance via evolution. But
this allows the Darwinian to pose a dilemma for Nagel. If consciousness has natural
significance, then it must play an evolutionary role. But then it will have an evolu-
tionary explanation. On the other hand, if consciousness plays no evolutionary role,
it must lack significance. But then why not accept it as a brute fact? Either way,
Nagel’s HFA collapses.
Nagel rejects moral naturalism. He agrees with the evolutionary moral nihilist that
moral facts have no place in evolutionary explanation: ‘an evolutionary self-
understanding would almost certainly require us to give up moral realism’.7 But
Nagel turns the nihilist argument on its head. Moral realism is more compelling than
Darwinism. If the two are incompatible, then Darwinism must go.
Moral non-naturalism is key to Nagel’s consciousness-based HFA. It undermines
the brute fact response, forcing the Darwinian to offer an evolutionary explanation of
consciousness. Non-naturalism is even more central to Nagel’s other HFA, because
here the human feature to be explained is our ability to respond to reasons and
values. Nagel’s moral HFA insists that Darwinism cannot explain these abilities, and
that they are significant. Both steps rely on Nagel’s non-naturalist moral realism.
7
Nagel, Mind and Cosmos, p. 27.
 THE CASE AGAINST ATHEISM
The explanatory failure of Darwinism goes hand in hand with the meta-ethical
failure of moral naturalism. If ‘value’ and ‘reason’ are not biological categories, then
they confer no evolutionary advantage, and thus Darwinism cannot hope to explain
them. This connection is borne out in the critical reaction to Nagel’s book. Nagel’s
Darwinian critics fail to engage with his non-naturalism.8 Like the global naturalist in
chapter 2, the Darwinian presumes the disjunction of moral naturalism or moral
nihilism. If we have access to reasons or values, that access must be a purely biological
phenomenon, and therefore open to evolutionary explanation. Darwinians cannot
take non-naturalist moral realism seriously. Nagel’s HFA highlights the stark choice
faced by the semantic non-naturalist. If we find non-cognitivism and moral natur-
alism implausible on semantic grounds, then we must either embrace moral nihilism
or take the first step down the road to cosmic purpose. Of course, non-cognitivists,
moral naturalists, and moral nihilists will all see this as further reason to reject non-
naturalism. But Nagel seeks to follow the moral appearances wherever they lead. And
this is where our project most clearly aligns with his.
Our next question is whether AP can borrow Nagel’s moral HFA. There are two
obvious problems. Both arise because HFA focuses on human beings. First, how do
we get a cosmic purpose at all? Won’t HFA deliver, at best, a human purpose—a
reason why each human exists? BT solves this first problem. Only God could be
responsible for the emergence of creatures whose features have no Darwinian
explanation, and God has a plan for humanity. But that brings us to our second
problem. Even if HFA does deliver a cosmic purpose, won’t it be an ineliminably
human-centred one? God explains our special features only because God has a plan
for each of us.
AP must reject this aspect of any BT HFA. Indeed, given its anthropocentrism,
shouldn’t our aversion to caprice lead us to reject all HFAs out of hand? How can AP
justify treating human phenomena as objectively significant?
Nagel’s conclusion to his moral HFA transforms human-centred local values into
a cosmic purpose that AP can borrow. We begin with the issue of cosmic purpose
itself. Nagel explicitly rejects a cosmic designer, and describes his alternative as
‘non-purposive’. However, terminology can mislead here, because Nagel associates
‘purposive’ explanation with divine intention. Nagel own tentative explanation is
teleological and Aristotelian. He distinguishes between value-free and axiarchic
teleological explanations.9 And his remarks suggest that Nagel regards the value-
free alternative as unsatisfactory—precisely because it cannot explain why this
universe exists with these significant features.10 Indeed, it is hard to see how any
explanation of non-material facts that acknowledged their significance could work
8
See, e.g., Godfrey-Smith, ‘Not Sufficiently Reassuring’; Leiter and Weisberg, ‘Do You Only Have a
Brain?’; Matthen, ‘Thomas Nagel’s Untutored Reaction of Incredulity’; Orr, ‘Awaiting a New Darwin’; and
Sober, ‘Ending Science As We Know It’.
9 10
Nagel, Mind and Cosmos, p. 64. Nagel, Mind and Cosmos, pp. 88 and 92.
TELEOLOGICAL ARGUMENTS 
without either presupposing or delivering objective values. This reinforces one of the
main lessons of chapter 3. Like the cosmological argument, HFA delivers Normative
AP rather than Simple AP. There can be no cosmic purpose without objective
values.
Nagel’s moral HFA thus delivers an axiarchic explanation of human moral
abilities, one that cites objective values. Furthermore, Nagel agrees that it would be
presumptuous (and very theoretically messy) to treat ourselves as anomalies in an
otherwise mindless irrational universe. Nagel therefore suggests a radical revision of
our scientific understanding. The capacity for consciousness, reason, and value is not
unique to human beings. Instead, it is somehow built into the very fabric of the
universe. ‘It is trivially true that if there are organisms capable of reason, the
possibility of such organisms must have been there from the beginning.’11 Nagel
admits that ‘[a]t this point such a theory is a complete fantasy’.12 But his HFA
suggests that, without such a revolution, science cannot capture the central reality of
our human lives.
Chapter 3 introduced a distinction between our evidence for cosmic purpose and
the content of that purpose. This distinction is central to AP’s ability to borrow any
HFA. Even if human beings are our best evidence of design, this is compatible with a
non-human-centred purpose. Perhaps our ‘special’ features are a step towards some-
thing higher, or a by-product of some larger process, or an accidental result of general
laws designed for some other end. (Chapters 5 and 7 explore these possibilities in
relation to our human capacity to perceive cosmic values.)
Because our case for AP is cumulative, we need not endorse any HFA. However,
our commitments lend particular support to Nagel’s moral HFA. AP takes Nagel’s
call for a scientific revolution in a new direction, asking us to transcend human
specialness altogether, and seek instead significant general features of the cosmos.
Any HFA that AP can borrow thus shades into the cosmic features argument, to
which we now turn.13
4.3 Cosmic Features Arguments

Arguments from order posit cosmic purpose as the best explanation for some general
feature of our universe. We explore two variants: cosmic features arguments and fine-
tuning arguments. Our argument for AP will combine the two.
11 12
Nagel, Mind and Cosmos, p. 82. Nagel, Mind and Cosmos, p. 48.
13
Another source for an HFA is Plantinga’s Evolutionary Argument Against Naturalism (EAAN).
Plantinga argues that evolution cannot explain why we have true empirical beliefs. (See, e.g., Plantinga,
Warrant and Proper Function, pp. 216–37.) Like Nagel, Plantinga challenges AP to provide an alternative
explanation of the emergence of rational moral beings. However, Plantinga’s argument is quite distinct
from any standard HFA. It would therefore take us too far afield to first develop a Plantinga-inspired HFA
and then ask whether AP can borrow it.
 THE CASE AGAINST ATHEISM
Suppose we agree with Nagel that the potential for consciousness is built into the
universe from the start. If so, this is a striking feature of our universe. Why does the
universe have this potentiality? BT has an obvious explanation. God designed the
world to have this feature, perhaps because it is objectively valuable. Axiarchism
offers a similar story. Our universe has this feature because it is good. These are
cosmic features arguments (CFAs).
Other striking features of our universe that have been cited here include the
following:
M: Our universe is mathematical. It is governed by beautiful, elegant, simple
mathematical laws.
I: Our universe is intelligible. At least some of its inhabitants are capable of
comprehending the laws by which it is governed.
U: Our universe is (or will be) understood. Eventually, some of its inhabitants will
understand the laws governing our universe.
BT and axiarchism explain these features by citing cosmic purpose. How else could
we explain them? One simple answer is that they are necessary. If this means logically
necessary, it is implausible. We can easily imagine very different universes.14 On the
other hand, physical necessity seems irrelevant. Of course it follows from the laws of
physics that our universe has laws of physics. But why do those laws hold? Similarly,
perhaps it is necessary that this universe is mathematical and intelligible—in the
sense that any universe lacking these features would be a different universe. But that
merely highlights our initial question: why does this universe exist?
Can science help? Science has certainly taught us that the universe is mathematical,
and perhaps provides our clearest evidence that it is intelligible. If the universe is ever
understood by human beings, science will play a key role. But can science explain
why the universe is mathematical, intelligible, or understood? Can science explain the
preconditions of its own success?
As with the existence of the universe, a scientific (or other non-purposive) explan-
ation of cosmic features is hard to imagine. Opponents of CFA must instead treat
cosmic features as brute facts. The universe just happens to exist, and to be math-
ematical and intelligible. These strike us as significant features. But perhaps this
merely reflects our own preoccupations, rather than tracking any objective moral
fact. It is no surprise that scientists regard its suitability for scientific investigation as
a remarkable feature of our universe. But surely this is anthropocentric hubris that
AP should reject.
14
Nor is this merely a philosopher’s prejudice. ‘There is not a shred of evidence that the Universe is
logically necessary. Indeed, as a theoretical physicist I find it rather easy to imagine alternative universes
that are logically consistent, and therefore equal contenders for reality.’ (Davies, ‘The Appearance of
Design in Physics and Cosmology’, p. 148.)
TELEOLOGICAL ARGUMENTS 
We could reply that any feature is worth explaining, even if we cannot establish its
objective significance. (Just as, in chapter 3, one modest cosmological argument
argued that the existence of a physical universe is worth explaining whether or not
it is surprising.) But a stronger reply would defend the objective value of M, I, and
U. A comparison with the human features argument is instructive. Cosmic features
are certainly much less parochial than human ones. Indeed, as we saw in the previous
section, the significance of cosmic facts emerges naturally from non-human-centred
reflection on striking human features. Following Nagel, we might move from wonder
at human consciousness, reasoning, or scientific achievement, to wonder at a uni-
verse where such things emerge ‘naturally’.
In chapter 3, we distinguished two possible connections between objective value
and cosmic purpose. Values can be premises in an argument for cosmic purpose. We
are driven to explain X because X is objectively valuable, and cosmic purpose emerges
as the best explanation on the assumption that there are objective values. Alterna-
tively, objective values can be among the conclusions of a CFA. Even though we do
not already believe in objective values, X still cries out for explanation. We then posit
objective values in order to explain X: either directly as in axiarchism, or indirectly
via a divine choice that either responds to values or creates them.
CFAs have two strengths that cosmological arguments lack. First, our present
comparative value claims are more secure, because they compare possible physical
universes to one another, rather than comparing existence to non-existence. It is
easy to deny the significance of S. It is harder to deny the value of M, I, or
U. Second, the features to be explained are more obviously contingent. While
some deny that there might have been nothing, it is much harder to deny that
there might have been a physical universe without mathematical laws, or one
whose laws were not understood.
Our commitment to objective values helps CFA only if M, I, and U are valuable.
But this is not implausible, especially given our other utilitarian commitments. In
chapter 1, I argued that, to accommodate obligations to future people, we need an
objective list theory of human well-being. This offers a list of independently valuable
things, typically including knowledge (both cosmic and mathematical knowledge
often feature explicitly), as well as elegance, order, and other general features of a
valuable life. I noted in chapter 1 that a satisfactory utilitarian intergenerational ethic
also needs impersonal comparative evaluations of states of affairs. These take us from
a list of the valuable components of individual lives to the aggregate value of the
whole universe. Utilitarians thus need objective values of exactly the kind that
underpin CFA. A final relevant utilitarian commitment is our aversion to caprice.
If we stand back from our objective lists and impersonal comparisons, and try to
remove any human-centred errors, then we will end up endorsing the independent
value of M, I, and U. If there are any objective values, then M, I, and U are very
plausible candidates. (We address the plausibility of these particular values at greater
length in chapter 12, when we ask whether they are really consistent with AP.)
 THE CASE AGAINST ATHEISM
Can AP borrow CFAs? Some arguments are clearly unproblematic. The value of a
mathematical life-friendly universe says nothing about the value of human beings.
But other CFAs begin from human-centred evidence. We know the universe is
intelligible or understood only because it is (partially) intelligible to us. Furthermore,
these CFAs naturally suggest human-centred explanations. If God made the universe
so that it would be understood, and we understand it, then how can we be irrelevant to
God’s cosmic purpose?
We must tread a fine line here. While AP must avoid human-centred purpose,
U and I are useful precisely because they offer a cosmic purpose that human beings
can meaningfully incorporate into their lives. These cosmic features play a key role in
the development of AP’s human morality in part III. Furthermore, as we’ll see in
section 4.7, we may also need M, I, and U to support the best fine-tuning argument.
Therefore, AP has very good reason to value these cosmic features.
How can AP recognize the objective value of I and U, and still deny a human-
centred cosmic purpose? Drawing once again on chapter 3, AP can separate evidence
of purpose from content of purpose. Understanding the universe comes in degrees.
Our evidence for I and U is that some human beings understand the universe to some
degree (d1). Our conclusion is that some degree (d2) of understanding is sufficiently
significant to explain the existence of the universe, and therefore that the cosmic
purpose involves that degree (d2) of U. But these two degrees (d1 and d2) need not
coincide. Perhaps the highest degree of understanding available to human beings falls
short of the lowest degree needed for objective significance or cosmic purpose.
Perhaps the universe is designed to give rise to some other (better) form of intelligent
life. If you set out to design a universe containing intelligent rational free beings who
are capable of comprehending the purpose of the universe and entering into genuine
communion with God, then it is not especially surprising if, in addition to beings like
that, you also get creatures like us. The universe is designed to be intelligible, but not
to humans. (We return to the possibility of non-human extraterrestrial rational
beings in chapter 7, where I argue that AP cautions against dismissing this unsettling
possibility.)
Understanding differs in kind as well as degree. Perhaps truly cosmically significant
understanding would relate to the purpose of the universe, or the mind of God, rather
than the laws of physics. Our feeble grasp of the laws of physics is but a pale shadow
of real understanding. The vaguer, and more obviously inadequate, insights of
moralists and mystics might be closer to real understanding than the precise know-
ledge of scientists. Our commitments to pluralism and the avoidance of caprice
caution against blithely supposing that our currently most revered mode of under-
standing is the best.
Human understanding could thus be evidence of cosmic purpose, or of objective
value, without itself being either part of the cosmic purpose or an instance of
objective value. (We develop this possibility in more detail in chapters 5 and 7, and
in part III.)
TELEOLOGICAL ARGUMENTS 
Like HFA, any CFA based on I or U does suggest a human-centred purpose—or at

least an explanation where human beings have some cosmic significance. If CFA
were the whole picture, then BT would be a reasonable conclusion. But it is not. Part
II shows that we have independent reason to reject any human-centred purpose. We
must (and can) reinterpret our evidence for I and U in a non-human-centred way.
There are credible explanations that do not yield a human-centred purpose. While
they seem to support BT, CFAs can also support AP against atheism.
I conclude that AP can borrow the best CFA. (And, like HFA, CFA delivers
Normative AP.) For the consequentialist, CFA is more persuasive than the cosmo-
logical argument, mainly because it avoids the anomalous comparison between
something and nothing that puzzled us in chapter 3. Consequentialists who found
the latter persuasive, should definitely consider the former. And the conclusion of
chapter 3 was that we should take the cosmological argument seriously. So we should
also take CFA seriously. On the other hand, CFA does have its weaknesses. In
particular, CFA cannot provide any comparative measure of unlikelihood. We cannot
yet claim that M (or I or U) is unlikely or improbable. But the atheist opponent of
CFA then objects that it makes no sense to say that M is surprising. She can simply
shrug and insist that M is a brute fact. Cosmic feature arguments share this weakness
with both cosmological arguments and HFA.15 It is this defect that the fine-tuning
argument seeks to avoid.
4.4 The Fine-tuning Argument for BT

In contemporary philosophical debate, the fine-tuning argument (FTA) often begins
with the failure of specific design arguments. The basic idea is simple. While the
emergence of X is not remarkable, this very fact is itself remarkable. We need cosmic
purpose to explain the remarkable fact that our universe is a place where the
emergence of X is not remarkable. ‘X’ can take many values. I will focus on life.
Most other plausible candidates for X are specific kinds of life: conscious life, self-
conscious life, rational life, moral life, or beings able to comprehend the universe. We
return to these options in section 4.7.3, as they serve to link FTA to both HFA and
CFA, thereby strengthening all three.
However, we initially develop FTA independently of either human features or
cosmic features arguments. We thus assume there is a non-purposive explanation of
the emergence of all life, including human beings. (If there is no such explanation,
then we can construct a specific design argument instead of a fine-tuning one.) But
15
Swinburne does try to inject probabilities into the cosmic features argument: ‘if God brings about the
world there is a significant probability (say between 0.2 and 0.8) that He will bring about an orderly,
spatially extended world in which humans have a location’. (Swinburne, ‘The Argument to God from Fine-
Tuning Reassessed’, p. 113.) But, as with the cosmological argument, I think this introduces an artificial air
of precision. We explore alternative attempts to measure improbability in section 4.7.3.
 THE CASE AGAINST ATHEISM
what makes the emergence of life possible? Life as we know it can only emerge
naturally if a number of specific conditions are met. These include the following:
stable physical objects that endure for more than a few nanoseconds; stars that
produce carbon or some other complex elements to form the basis of life; a stable
planetary environment that endures for billions of years; and regular laws of nature.
Life could not emerge in a chaotic universe, or in one lacking stable molecules,
complex elements, stars, or stable planetary ecosystems.
Let us say that a universe is ‘friendly-to-life’ if and only if evolution (or some other
non-purposive process) can account for the emergence of life within it. We thus have
a new fact to explain:
(FL) Our universe is friendly-to-life.
On this definition, a universe would not count as friendly-to-life if life could survive
only if created and then constantly protected by an omnipotent God. To take a
mundane analogy, a planet is friendly-to-life in our present sense only if life could
emerge there on its own. A planet where super-advanced aliens could survive by
insulating themselves from the local environment is not friendly-to-life. So far as we
know, the Earth is friendly-to-life in this sense, while the Moon is not.
To deny FL is, in effect, to accept that life requires specific design. So opponents of
cosmic purpose must accept FL. But FL turns out to be very difficult to explain.
Almost no one would argue that FL is logically necessary. We can easily imagine
universes that are not friendly-to-life. Of course, FL may be ‘necessary’ if that means
‘necessary given the laws of physics and the initial conditions of the universe’. But
this is unhelpful, as the fact that those laws and conditions are friendly-to-life is what
we seek to explain. (Some opponents of FTA attempt to prove the necessity of FL via
a ‘theory of everything’. We explore that option in section 4.6.)
The core of our FTA is the denial that FL could be a brute fact. An astonishing
array of cosmic coincidences are required for life to evolve. ‘Being friendly-to-life’ is
not a basic property of our universe. Instead, it depends on the precise values of
several constants and ratios that figure in the basic laws of physics. Modern physics
suggests that our universe would not be FL if those values were ever so slightly
different. Examples discussed in the literature include: the value of the strong
electromagnetic force; the ratio of the masses of the proton, the electron, and the
neutron; the value of the cosmological constant; the amount of dark energy in the
universe; the amount of entropy per baryon; and the ratio between the electromag-
netic force and of the gravitational force.16 This last example is due to Brandon
Carter, who pioneered the modern FTA. In the actual universe, these two forces are
delicately balanced to produce two distinct kinds of stars. Radiative stars lead to
16
See, e.g., Collins, ‘Evidence for Fine-Tuning’; Leslie, Infinite Minds, pp. 201–5. A very readable
introduction is Davies, The Goldilocks Enigma. For a more sceptical view, see Weinberg, ‘A Designer
Universe?’.
TELEOLOGICAL ARGUMENTS 
supernova explosions, while convective stars are like our sun. The former are neces-
sary to spread the basic elements through the universe, while the latter are needed for
the formation of planets where life can evolve. The emergence of life thus needs stars
of both kinds. Accordingly, it requires a very precise balance between the two forces.
The same ratio needs to be ‘fine-tuned’ in two different directions to produce the two
different kinds of stars. If all possible ratios are equally likely, the chance of getting
the correct one is 1 in 1040. (Other numbers in the literature are even more
astronomical. One estimate puts the chances of a suitable value for the entropy per
123
baryon at 1 in 1010 .)
Paul Davies estimates there are thirty constants in basic physics and modern
cosmology that must be fine-tuned for the emergence of life. Not all require coinci-
dences as remarkable as those listed here, but many do. A universe with the same
general physical laws as ours is astronomically unlikely to be friendly-to-life. If you
selected a universe like ours at random, it is almost certain that you would get a
lifeless one. Surely a cosmic coincidence on this scale cannot be a brute fact.
The cosmic coincidence involved here is nicely illustrated by the following popular
analogy:17
Leslie’s Dart. Imagine a very large target with one red dot and many blue dots. A dart is thrown
at the board, and it hits the red dot. To capture the scale of the cosmic coincidence required for
life, suppose the dart is thrown across the entire visible universe and hits a target the size of an
atom. Could something that unlikely just happen?
If FL is not a brute fact, what other explanations are available? Two rivals dominate
the current debate: BT (or axiarchism or some other purposivism); and the multi-
verse hypothesis. Both alternatives deny the following conjunction: (1) this is the only
universe; and (2) it just happens to be FL. BT denies that this universe just happens to
be FL, while the multi-verse hypothesis denies that this is the only universe.
The basic BT FTA is simple.18 If God exists, then FL is quite likely; while FL is
astronomically unlikely if God does not exist. Therefore, we must posit God to
explain FL. Unlike the cosmological argument, HFA, or CFA, this argument prom-
ises objective comparisons of probability. If we accept Carter’s coincidence claim
alone, then the probability of FL occurring by chance is of the order of 1 in 1040. BT
offers a better explanation of FL than brute fact, unless either (a) it is astronomically
implausible that a creator God would favour life, or (b) the intrinsic probability of a
creator God is astronomically low. In practical terms, given the astronomical
improbability of the cosmic coincidences, God will be the better explanation unless
we can prove either that God is logically impossible or that God would definitely not
17
Leslie, Universes, p. 10.
18
My FTA owes most to John Leslie’s formulation: Leslie, Immortality Defended, pp. 77–86; Leslie,
Infinite Minds, pp. 205–16. Leslie defends axiarchism rather than BT, but the basic form of the argument is
the same.
 THE CASE AGAINST ATHEISM
create. Neither response seems plausible. God is not impossible. And a lifeless
universe would be unlikely to suit all of God’s purposes. If God creates, then FL is
very likely.
Some opponents object that, for all we know, God is equally likely to prefer a
universe where life appears miraculously. However, we must recall the present
dialectical context. We are assuming the failure of specific design arguments. Perhaps
the best argument for BT is disjunctive. God has overwhelming reason to prefer life.
Therefore, God will either create a universe where life emerges miraculously or one
that is friendly-to-life. (Another option, explored in section 4.7, is that the best FTA
relies on the fact that another feature of the universe is equally remarkable—namely
the fact that it is governed by regular mathematical laws. The value of this feature
gives God a reason to create a universe that is FL instead of one where life appears by
magic.)
As ever, we must distinguish between evidence for purpose and content of purpose.
FL is put forward as evidence that the universe was fine-tuned by God. But it does not
follow that the universe was fine-tuned for life. Life might be a by-product or
precondition of some other purpose. For instance, if God desires a universe contain-
ing free beings who emerge naturally and discover the wonders of the cosmos, then
God will create an FL universe. Life itself is a means to this higher end. Or God might
want a universe with many different kinds of stars. Life would then be an accidental
by-product. (We return to these issues in section 4.9, when we ask whether AP can
borrow the BT FTA.)
BT’s God is one possible purposive explanation of FL. But there are other possi-
bilities, notably axiarchism and impersonal analogues of BT. The axiarchist says that
FL is true because FL universes are good. Instead of separating BT from axiarchism—
or intellectualism from voluntarism, or a personal God from impersonal creation—I
consider them together. All offer explanations of FL that yield a cosmic purpose. In
the recent philosophical literature, the focus of debate is not on the details of
competing purposive explanations, but on the viability of non-purposive alternatives.
Also, we have already covered general objections to purposive explanations in
chapter 3. So I will treat BT as a proxy for all purposive explanations. (One exception
is in section 4.6, where we ask whether ‘scientific’ explanations of the cosmic
coincidences imply cosmic purpose.)
4.5 Evaluating the Fine-tuning Argument

HFA, CFA, and FTA seem quite distinct, and are often considered separately in the
literature. I shall argue, however, that the most plausible teleological argument,
especially in the context of our overall project, is one that combines elements of all
three. We can construct a composite explanation of the striking features of our
universe (FL, M, I, U) that is superior to any available alternative.
TELEOLOGICAL ARGUMENTS 
HFA and CFA are both much stronger if they can make intelligible claims about
the improbability involved in treating the relevant facts as brute. Embedding these
arguments within an FTA delivers this. Conversely, as we’ll see, FTA needs objective
values to shore up its own probability claims. Crucially, the objective value of life is
not sufficient. A good FTA also needs the objective significance of some other cosmic
feature. In particular, these objective cosmic values give the FTA for purposivism a
decisive advantage over its main non-purposive rival—the FTA for a multi-verse.
This further supports our emphasis on Normative AP. If there is a good FTA, then it
will automatically give us a close connection between cosmic purpose and objective
values.
We begin our evaluation by focusing on the original FTA, introducing other
elements along the way. We must ask four questions.
1. Can science explain FL? (Section 4.6)
2. Can we treat FL as a brute fact? (Section 4.7)
3. Can a multi-verse explain FL without positing cosmic purpose? (Section 4.8)
4. Can AP borrow FTA? (Section 4.9)
4.6 A Theory of Everything

Many physicists regard arbitrary constants as a theoretical defect. A good theory
should explain the value of each constant. One option is the multi-verse hypothesis—
explored below. But some regard that hypothesis itself as even more inelegant than
the facts it seeks to explain. They seek a theory of everything (TOE) to explain why
this universe is FL.
We first encountered TOE in section 3.3.2. We saw there that only the most
ambitious TOE will explain why the universe exists. Similar remarks apply here. We
seek a TOE that removes cosmic coincidence and proves that any universe must be
FL. As with scientific explanations of the fact that there is something rather than
nothing, AP has several distinct replies. The first is that we have no TOE, nor any
clear idea what a TOE would look like. We can explore the hypothetical implications
of a TOE, but it is a leap of faith to assume one will ever be found. This is especially
true for the most ambitious TOE—one that aims to explain FL. At this stage of
scientific inquiry, it is perfectly reasonable for AP to deny that science will ever
explain this.
In reply, TOE proponents often offer an FTA of their own. We should believe there
is a TOE because the alternative is to accept our cosmic coincidences as brute fact.
This is a reasonable move. Rejecting brute fact significantly raises the epistemic
credentials of any alternative explanation. We now have three alternatives: purposi-
vism, multi-verse, and TOE. How should we choose between them?
AP’s second reply is that we need not choose—at least not yet. For all anyone now
knows, and certainly for all anyone has ever shown, it may be that a TOE can only
 THE CASE AGAINST ATHEISM
explain FL via cosmic purpose. Perhaps a complete explanation must invoke object-
ive values, or derive the values of constants from the purposes of God.
Cosmic purpose might especially be needed to explain why the TOE is true, even if
it does not feature in the statement of the TOE itself. After all, being governed by an
FL-TOE is a remarkable feature of a universe. (Indeed, it may seem even more
remarkable than FL itself.) Even if we discover the TOE, we can still ask why that
theory governs our universe. A TOE can relocate the point where science runs out,
but it cannot remove it. It cannot prove that FL is logically necessary.
The explanatory inadequacy of TOE is even more striking if we recall our cosmic
features arguments. Some hope that TOE will demonstrate the mathematical neces-
sity of elegant intelligible laws. The goal of TOE, after all, is to show why every
universe must have these laws. But TOE could only ever yield local necessity. It might
explain why all universes of a certain type have certain laws; but it cannot explain
why there is a universe of that type—rather than some other sort of universe, or none
at all.
Until the TOE is actually produced, we cannot be confident that it explains
anything without appeal to objective values or cosmic purpose or something equiva-
lent. (As we’ll see in section 4.8.2, a purposeless multi-verse also cannot explain M, I,
or U—for the same reason.)
I argue in section 4.7 that FTA only works if it assumes that being governed by
intelligible mathematical laws is itself a remarkable feature of this universe—and one
that cries out for explanation. Yet how could a TOE, working within the realm of
mathematical physics, explain the condition of its own applicability? And, even if it
could, how can we be confident that it would do so without any reference to cosmic
purpose?
In The Goldilocks Enigma, Paul Davies explores a variety of speculative explan-
ations for the emergence of conscious life within a universe governed by elegant
mathematical laws—on the assumption that neither brute fact nor a multi-verse can
provide the full answer. While different from BT (and especially from any popular
Western religion), these speculations have a purposive flavour, and are thus fruitful
avenues for AP to explore. For instance, here is one of Davies’s own accounts of what
needs to be explained:
Our universe possesses laws and states which not only permit self-simulation, they also permit
self-comprehension. The cosmic rule book, in being fit for life and in facilitating the eventual
emergence of consciousness, has not only ensured that the universe has constructed its own
awareness. The cosmic scheme has also constructed an understanding of the cosmic scheme.19
We cannot assume that any successful TOE must give us the BT God. Atheist
physicists are right to see TOE and God as potential rivals. But AP is much more
19
Davies, The Goldilocks Enigma, p. 288. Elsewhere, Davies explicitly endorses something akin to
cosmic purpose. (See the epigraph to this chapter.)
TELEOLOGICAL ARGUMENTS 
modest. It can borrow any explanation that either assumes that certain features of our
cosmos are valuable or allows us to infer that their role in explaining its existence
makes those features valuable. It is not unreasonable to suggest that, for all anyone
knows to the contrary, any TOE must meet one of these conditions. If so, AP can
borrow any successful TOE.
4.7 Rehabilitating Brute Fact

4.7.1 Is FL significant?
FTA promises an advance on our earlier arguments because of the extreme improb-
ability of FL. If the chance of FL really was only 1 in 1040, then surely this is too much
of a coincidence to swallow. Some opponents deny the fine-tuning probability claim
(section 4.7.2). Others agree that FL is true in only 1 in 1040 of possible worlds, but
still deny that FL needs explanation. We accept some statistically unlikely events as
brute facts. Every possible world has some extremely unlikely features. Why not just
accept that FL is one of them?
Consider some analogies. I enter a lottery with one billion tickets and win.
Obviously, this is remarkable for me. But if I conclude that the lottery must have
been designed so that I win, you might reasonably reply: ‘Someone had to win. Why
not you?’ Or suppose I throw a dart at a dart board with 360 million regions, and hit
region 224,689,423. Is this remarkable? No. I had to hit some region—why not that
one? Or suppose the sole survivor of a plane crash concludes that God wanted him to
survive. This is just the sort of egoism that AP rejects.
The literature on personal identity provides an intriguing parallel argument.20
1. If my identity depends on my genetic makeup, then it was astronomically
unlikely that I would ever exist.
2. My identity depends on my genetic make-up.
3. Therefore, my existence was astronomically unlikely.
4. My existence is objectively significant.
5. If X is both astronomically unlikely and objectively significant, then the best
explanation is that the universe was designed so that X.
6. Therefore, the universe was designed so that I will exist.
The conclusion is an expression of megalomania. Something has gone wrong. Zuboff
presents this argument as a reductio ad absurdum of the genetic theory of personal
identity (premise 2)! But this is an extreme reaction. The obvious alternative is to
reject premise 4. Like my lottery win, my existence is significant for me. But it is not
objectively significant that I exist rather than someone else. The improbability of my
20
Carlson and Olsson, ‘Is Our Existence in Need of Further Explanation?’; Zuboff, ‘One Self: The Logic
of Experience’.
 THE CASE AGAINST ATHEISM
existence suggests instead that God didn’t design the universe to produce me.
Someone who wanted me to exist would have created a universe where I was likely
to exist. The genetic theory of personal identity may have implications for ethics,
especially those associated with Parfit’s famous non-identity problem. But the claim
that I am the purpose of the universe is not one of them!
Let us introduce some arbitrary terminology: an event is remarkable if and only if
it is both astronomically unlikely and objectively significant. FTA says that remarkable
events must be explained by cosmic purpose. The present reply is that there is
nothing objectively significant about life. (Life only seems significant to us because
we are alive.) FL is thus not remarkable—no matter how unlikely it is. Despite the
cosmic coincidences, FTA fails.
The strongest FTA reply will cite independent objective values. Our universe is
objectively significant because it is one of the tiny minority of (especially) valuable
possible worlds. Objective values also reinforce purposivist explanations. If FL
universes are objectively valuable, then God will choose one and axiarchism’s directly
efficacious values will produce one. Our commitments support FTA here. For those
already committed to objective values, FTA is hard to avoid.
FL is the most familiar basis for FTA, but it is not the only one. If life is not
objectively valuable per se, perhaps some special kinds of life are. Perhaps the
universe is fine-tuned to produce conscious, or rational, or moral beings. (So far as
we know, unless there is divine intervention, such beings can only emerge in our
universe after the evolution of life itself.) Value claims relating to M, I, and U are less
anthropocentric than those relating to life. While an interest in life can be dismissed
as parochial, it is much harder to similarly dismiss an interest in M, I, or U.
One kind of life is especially relevant to our project. If there are objective values,
then the existence of beings capable of responding to such values is (plausibly) itself
significant. And the emergence of such beings requires FL. The opponent of FTA
must reject, not only the objective value of life in particular, but the very idea of
objective value itself.
Given our moral commitments, this appeal to objective values is legitimate. But
can we construct a less contentious reply? Could FTA be the foundation for belief in
objective values, and not just a consequence of such belief?
Even those who reject objective value may share the intuition that life cries out for
explanation—in a way that the bare existence of a lifeless cosmos would not. This
intuition does seem very widespread. But can it be justified? It is hard to know how to
answer this question, partly because it can be hard to believe that anyone really asks
it. Peter Van Inwagen goes so far as to describe the view that there is nothing to
explain about the existence of life as ‘one of the most annoyingly obtuse in the history
of philosophy’.21
21
Van Inwagen, Metaphysics, p. 151.
TELEOLOGICAL ARGUMENTS 
One explanation is that, in everyday life, terms such as ‘surprising’ or ‘remarkable’

often signify the availability of an alternative explanation. (They are synonyms for
‘suspicious’.) Suppose the lottery winner is the person who designed the lottery
machine. We naturally suspect her of cheating. That suspicion is not removed by
the observation that ‘someone had to win’. But chance did satisfy us when a stranger
won. So why not now? The answer is that an obvious alternative explanation now
presents itself: namely, that the winner cheated. More generally, we reject chance
when both (a) someone could have interfered with an otherwise chance process, and
(b) the result is what that person would have chosen. Analogously, BT claims that life
is what someone would choose if they were selecting a universe. So the fact of FL
naturally suggests that someone did choose our universe. As van Inwagen puts it: ‘the
setting of the cosmic dials does suggest an explanation: the dials were set by a rational
being who wanted the cosmos to be a suitable abode for other rational beings’.22
Completely gerrymandered explanations will not do. The ‘salient’ alternative must
be somehow familiar, or independently credible. Van Inwagen’s alternative is BT,
with its God and divine purpose. As we saw in chapter 3, the familiarity of BT is open
to question. But we also saw that, if either BT or AP is supported by other arguments
(such as those presented in chapters 3, 5, or 6), then it may become sufficiently
familiar to offer a salient alternative to brute fact.
Finally, FTA might accept that we cannot form antecedent judgements of com-
parative significance, but deny that we need to. Any improbable occurrence is worth
explaining—especially if the chances are 1 in 1040. We should thus listen very
carefully to any proposed explanation. If a hypothesis can really explain something
that improbable, then this is a clear point in its favour. A convincing non-chance
explanation might then tell us why the event is significant. Perhaps we should
conclude that life is objectively valuable because, without this value claim, we cannot
hope to explain FL. (As with our earlier arguments, objective values are then among
the conclusions of FTA, rather than its premises.)
Given our commitments, it is very implausible to treat FL as insignificant. FL is
then remarkable if and only if it is astronomically unlikely. But is it?
4.7.2 Is FL unlikely?
This section draws on an objection to the FTA for the multi-verse hypothesis,
associated especially with Hugh Mellor.23 While many commentators think it applies
equally well to the BT FTA, I argue below that Mellor’s objection is actually much
22
Van Inwagen, Metaphysics, p. 153.
23
Mellor, ‘Too Many Universes’. See also McGrew and McGrew, ‘On the Rational Reconstruction of
the Fine-Tuning Argument’; McGrew, McGrew, and Vestrup, ‘Probabilities and the Fine-Tuning Argu-
ment’; Pruss, ‘Fine- and Coarse-Tuning’. (For discussion of the related ‘measure problem’ in cosmology,
see Kragh, ‘Contemporary History of Cosmology’, p. 543.) Monton, ‘God, Fine-Tuning, and the Problem of
Old Evidence’ defends a subjective interpretation of the fine-tuning probabilities.
 THE CASE AGAINST ATHEISM
stronger against the multi-verse. For now, however, we apply the objection to the BT
FTA.
Many opponents of FTA reject its use of probabilities. To convince an impartial
observer, the argument needs objective probabilities rather than subjective ones. The
probabilities must track real facts, rather than merely measuring epistemic uncer-
tainty, or individual commitment, or whatever. But objective probabilities only make
sense within a given physical universe, where they are determined by (or grounded in)
the laws of nature. (Consider what makes it true that the chance of rolling a six on a
fair six-sided die is one in six, and not one in seven.) Yet FTA asks us to step outside
our universe and evaluate ‘the probability of getting a universe like this’. This takes us
beyond the realm where talk of objective probabilities makes any sense.
BT has two basic options: defend objective probabilities or use subjective ones.
I will argue that, armed with objective values, BT can defend objective probabilities.
But first we explore the latter option.
BT could reject objective probabilities. After all, FTA turns on a comparison
between two probabilities: the probability of FL occurring by chance and the prob-
ability of FL occurring if God exists. But could anyone really claim that the latter is
‘the objective probability that God will produce an FL universe’? Isn’t it much more
plausible to treat this as an epistemic claim—a measure of our uncertainty, rather
than any objective fact? If this probability is subjective, then presumably the other
probability is too.
Some opponents regard this concession as fatal to FTA.24 But it need not be. (By
contrast, in section 4.8.2 I argue that this concession is fatal for the multi-verse, which
seeks a non-purposive scientific explanation of FL.) While subjective probabilities
introduce an element of relativism, FTA may still have force for many people. Faced
with the seemingly astonishing cosmic coincidences outlined above, many people feel
convinced that some explanation is called for. Puzzles around objective probability
do not dissolve this sense of wonder—at least not for many philosophers and
scientists, atheists as well as believers. FTA would then become, in effect, an add-
itional cosmic features argument. FL would rank alongside M, U, and I as a striking
feature of the cosmos that must be explained.
However, FTA is clearly much stronger if it can claim that the objective probability
of FL is 1 in 1040. The vocabulary of ‘objective’ and ‘subjective’ suggests a misleading
dichotomy, where objective probabilities are ‘out there in the world’ and subjective
probabilities are ‘entirely in the head’. But the probabilities in FTA are best seen as
objective estimates, given the publicly available evidence, of epistemic probability.
There may be some unknown factor that determines whether or not FL is true. But
given our current knowledge of the nature of the laws of physics governing this
24
‘To retreat to the point where the argument rests on unargued intuitions is to deprive it of anything
more than devotional significance.’ (McGrew, McGrew, and Vestrup, ‘Probabilities and the Fine-Tuning
Argument’, p. 206.)
TELEOLOGICAL ARGUMENTS 
universe, the probability that FL will be true in a universe governed by such laws is (at
best) 1 in 1040. Similarly, it may be that God’s existence is logically necessary or
logically impossible; or that a God must (or cannot) create. But given our current
knowledge, the probability that God exists is neither 0 nor 1 (nor, for that matter, is it
close to either); and the probability that God would create a universe where FL is true
is significantly greater than 1 in 1040. These are not subjective claims about our
mental states, but objective claims about the link between evidence and hypothesis.
This more modest claim to objectivity is sufficient for FTA. And it requires, not a
proof that a certain order exists ‘out in the world’, but merely a reasonable belief
about the available evidence.
Unfortunately, BT cannot rest here. To ground objective probabilities about FL, we
need an objective order. Something must provide the space of probabilities.25 One
option is to partition the set of possible universes into FL and non-FL subsets. The
probability of FL is then a function of the relative size of the two subsets. One
immediate obstacle is that the number of possible universes is probably infinite, as is
the number that are FL. Infinity is not an insuperable barrier. Probabilities can still
make sense when comparing infinite sets, but only if their members are naturally
ordered. Compare the real numbers between 0 and 1 with those between 0 and 10.
Both sets are the same infinite size. But if we choose a real number between 0 and 10
at random, there is a 10 per cent chance it is between 0 and 1. We can assign
probabilities because the real numbers have a natural order.
Or consider a spatial analogy. Your dart hits its atom-sized target. I argue that this
was not unlikely. Space is continuous. So the set of spatial points in the target area is
infinite, and thus no smaller than the total set of spatial points in the universe. You
would object that sets of spatial points are irrelevant. What matters is that the region
of space in the target area is clearly smaller than the region of space outside the area.
Hitting the target is very unlikely.26
Probabilities apply here because, like the natural numbers, physical space has a
natural order. But are possible universes naturally ordered? Without a natural
ordering, we can always reorder them so that FL universes are no longer a small
subset. If the real numbers were randomly scattered around in ‘number space’, we
could not tell whether those between 0 and 10 outnumbered those between 0 and 1.
We can establish a one-to-one correlation between the two sets, and without a
natural order we cannot say which is bigger.
Objective probability requires an order that is itself objectively significant. One
option is to arrange possible universes using similarity relations. But different
25
McGrew, McGrew, and Vestrup, ‘Probabilities and the Fine-Tuning Argument’, p. 203.
26
Indeed, probabilities even make sense for infinite regions of space. Imagine an infinitely large dart
board, with a centre but no edges, divided into 360 million separate areas radiating out from the centre. To
hit the only red area from a great distance is unlikely. The red region (although infinite) is much smaller
than the black region.
 THE CASE AGAINST ATHEISM
similarity relations produce different orderings, and comparative similarity is notori-

ously subjective and interest-dependent. (Which is more like the actual world: the
world of Narnia or the world of Star Trek?) For objective probabilities, we need to
know which similarity relations matter.
Mellor’s original target is a purely scientific explanation of FL, built on the claim
that a multi-verse renders FL more probable. Against this target, Mellor’s objection
succeeds. The fine-tuning probabilities cannot be given a respectable scientific
interpretation, as they lack a plausible grounding in either physical or mathematical
reality (section 4.8.2). But this is grist to our mill. Our project favours the serious
consideration of non-scientific explanations, where objective values may provide
what science cannot.
4.7.3 Combining special features and fine-tuning

Before we ask how objective values might ground an objectively significant order—
and thus help us to ground objective probabilities—we must first examine another
objection to FTA.
So far, we have kept the basic laws of physics fixed when claiming that FL universes
are uncommon. But is this approach too parochial? Many possible worlds are very
different from the actual world. Our cosmic coincidences tell us nothing about
universes with wildly different laws of nature or unimaginable forms of life. Even
within our own universe, many alternative forms of life have been suggested: silicon-
based life, life within stars based on plasma or radiation, life at the quantum level, and
so on. Philosophers have long imagined purely spiritual worlds, or those governed by
Aristotelian biological laws. (Indeed, if Idealism is even credible, such worlds may be
epistemically indistinguishable from this one.) If we admit these possibilities, how
can we hope to estimate the proportion of possible worlds that are FL?
We can distinguish two kinds of fine-tuning: local fine-tuning occurs if FL uni-
verses are very rare among possible universes relevantly like this one; while global
fine-tuning needs them to be rare relative to all possible universes. The present
objection is that FTA must deal in global fine-tuning, while our evidence supports
only local fine-tuning.27
This is a powerful objection. AP itself emphasizes the limits of our intuitions and
cognitive powers, and the dangers of trusting them too far from home. We cannot
establish the global improbability of FL. We must focus on local improbability. What
is remarkable is not FL simpliciter, but rather the fact that we live in a universe
governed by physical laws of a certain kind, which is nonetheless friendly to life.
Consider a revision of Leslie’s analogy. Suppose your dart hits a red dot within a
vast area of blue. But beyond this local area, red dots predominate overall. This feat
27
For discussion, see Leslie, Immortality Defended, pp. 78–9; Collins, ‘Fine-Tuning Arguments and the
Problem of the Comparison Range’; Pruss, ‘Fine- and Coarse-Tuning, Normalizability, and Probabilistic
Reasoning’.
TELEOLOGICAL ARGUMENTS 
still cries out for explanation. Even if our evidence only supports local fine-tuning,
FTA is still compelling.
If possible universes have no objective order, then there is no real distinction
between local and global fine-tuning. (We cannot say that the red dot is in a blue
region unless the dots are regionally arranged.) As well as something that makes FL
objectively significant, FTA thus also needs an objective order that renders FL
(locally) unlikely. If objective values supply the latter, perhaps they can also provide
the former.
Here is one possibility. Our universe is governed by precise and elegant mathem-
atical laws. Call this fact ‘M’. If M is objectively significant, then it generates an
objective order. We can arrange possible universes according to the similarity of their
basic mathematical laws. Universes with identical laws (but different constants) are
more relevantly similar to our own than worlds with different laws of physics. Given
our current knowledge of the laws of physics, the objective epistemic probability that
FL will be true in a world like this is less than 1 in 1040.
This new FTA appeals to values in two places: once to establish the objective
significance of FL itself, and then again to justify an ordering of possible worlds by
similarity of physical laws.
FTA needs the objective value of cosmic features, because the laws of physics must
have objective significance. It is arbitrary and parochial to simply assert that the
specific laws of physics governing our cosmos are special. Better, surely, to say that
what really matters is whether a cosmos is governed by mathematical laws—and
then, within the class of M-worlds, a universe’s type of mathematical law provides a
natural (objectively significant) ordering. We focus on universes with similar laws
only because M is objectively significant.
The objective value of M also strengthens FTA in another way, especially if I and U are
also valuable. As it stands, FTA makes two unrelated value claims concerning FL and
physical laws. Both claims can also seem parochial and immodest. What is so special
about our kind of life or our physical laws? M, I, and U offer a way to connect these
claims: what is valuable is that a physical universe be governed by mathematical laws
that its inhabitants can discover and understand. M, I, and U are more general features
of a cosmos, and their objective value seems more intuitive—or at least less parochial.
A successful FTA must connect FL to objective value. But the value of FL itself
could be either intrinsic or instrumental. FL might be valuable in its own right;
valuable because it gives rise to life; or valuable because it gives rise to something else,
such as intelligent life or the understanding of the intelligible mathematical laws
governing the cosmos.
This connection between CFA and FTA is crucial in our dialectical context. FTAs
are offered for both main alternatives to brute fact: BT and multi-verse. The objective
value of cosmic features sits easily within BT. By contrast, it sits very uneasily within
the scientific (and non-purposive) framework of the multi-verse. The hidden role of
M thus supports BT against ‘scientific’ explanations such as the multi-verse.
 THE CASE AGAINST ATHEISM
However, our FTA for BT now faces a dilemma. Either cosmic features are
significant or they are not. If not, then FTA fails. But if they are significant, why
not dispense with fine-tuning altogether and appeal directly to the (simpler) CFA?
We then posit God, not to explain FL, but to explain M.
My reply is that the two arguments are mutually supporting. Just as cosmic
features support an FTA, so fine-tuning can also support a CFA in two ways. First,
we can appeal to fine-tuning to explain the improbability of cosmic features such as
I or U. CFA does better than FTA in terms of objective significance. But our original
CFA was much weaker regarding probability. Is it unlikely that a universe will be
governed by mathematical laws that are intelligible to its inhabitants? If so, how
unlikely? Without answers to these questions, we cannot rule out brute fact as
decisively for I or U as for FL.
I argued earlier that CFA can proceed without probabilities. But it is much
stronger if it can cite probabilities. Fine-tuning offers the best hope here.28 We
might construct a local improbability argument regarding U in relation to M. U is
true in very few similar possible worlds where M is true. Out of all possible universes
governed by mathematical laws like ours, very few have inhabitants to whom those
laws are intelligible—let alone inhabitants who do understand them.
This comparative probability claim relies on the cosmic coincidences. Under-
standing requires intelligent life, which in turn requires life. It is very difficult to
estimate the probability that intelligent life will emerge once life has emerged, or the
probability that intelligent life will eventually reach the point of real understanding. It
is certainly impossible to credibly place these degrees of improbability as high as 1 in
1040. Therefore, to establish that U is rare given M, we must first show that FL is rare
given M, and then cite the fact that U presupposes FL.
Fine-tuning also supports CFA in a second way. The opponent of FTA argues that,
without an objective ordering of possible worlds, FL is not unlikely. Therefore, we can
admit FL as a brute fact. We have seen that, to reply that FL cries out for explanation,
one must cite the objective significance of M. Some will no doubt respond by
accepting FL as brute fact. But others will still feel the full force of both the objective
significance of FL and its improbability. For those people, reflection on the implica-
tions of FTA may thus reveal their own implicit commitment to the objective
significance of M. (If you don’t think M is objectively significant, then why do you
28
Another option is to construct a local improbability argument for M. Suppose we agree that M is
valuable. It follows that mathematics provides an appropriate objective ordering for possible worlds. We
can then use mathematical criteria of complexity or elegance to arrange possible worlds. Relative to such
an ordering, worlds with mathematical laws as complex or elegant as ours may turn out to be very rare.
This argument is not circular, as it uses the value of M to ground the improbability of M. However, this
argument will be controversial, as competing measures of complexity or elegance might ground different
probability claims. Unlike FTA, this argument offers little hope of any measure of comparative
improbability.
TELEOLOGICAL ARGUMENTS 
find FL surprising?) But once we conclude that M is objectively significant, then a

CFA based on M is much harder to resist.
Neither M nor FL nor U is a brute fact. Opponents of cosmic purpose must look
elsewhere. Unsurprisingly, they turn to science (or, as their detractors would say, to
science fiction).
4.8 The Multi-verse

I really do believe that the case for design stands or falls upon whether we can
find another explanation in terms of multiple universes. (Paul Davies)29
In the recent FTA literature, the most significant rival to BT is the multi-verse.30
I shall focus on a single central objection: that the multi-verse hypothesis cannot
explain FL without positing cosmic purpose.
4.8.1 The multi-verse explanation of FL

Imagine a reality TV game, where a player and a computer each select a number
between 1 and 1,000,000. If both select the same number, the player wins £1. If not,
she is electrocuted on live television. Desperate to appear on television, you play this
game. You win. Intuitively, this cries out for explanation.
FTA claims that living in an FL universe is like winning this original game. But
now suppose I have just watched the game on television. I am suspicious. My friend
at the television network then tells me this was merely the first televised game. The
broadcaster agreed to transmit only if the player wins. While the threat of live
electrocution boosts ratings, an actual execution would deter advertisers. The game
has been played many times before. You were just the first winner. I might find this
disturbing, but I am no longer surprised that you won.
The multi-verse explanation for FL is analogous to this repeated game. It is based
on the following hypothesis.
The Multi-verse Hypothesis (MVH). Our universe is one of many universes that actually exist.
Among these many universes, FL is true in some and not in others.
This hypothesis enables observer selection (also known as the anthropic principle)
to explain FL. Of course we find ourselves in a universe where FL is true—no
observer could find herself anywhere else. On its own, observer selection is inad-
equate. It is like saying, in the one-off game: ‘Of course you won—no player could
observe any other result.’ By itself, this explains nothing. But if there are many
29
Remarks from an interview with Paul Davies in 2002, quoted in Manson, God and Design, p. 17.
30
For a range of views, see Carr and Ellis, ‘Universe or Multi-Verse?’; Kragh, ‘Contemporary History of
Cosmology’; Walker and Ćirković, ‘Astrophysical Fine Tuning, Naturalism, and the Contemporary Design
Argument’; Holder, ‘Fine-Tuning, Multiple Universes, and Theism’.
 THE CASE AGAINST ATHEISM
universes, then it is unsurprising that FL is true in some; and it is then unsurprising

that observers find themselves in such universes.
Despite superficial similarities, MVH is not modal realism. It is not a philosophical
theory about possible worlds, but a physical theory about the actual world.31 To avoid
confusion, I borrow some terminology from Hugh Mellor: our universe is whatever is
spatiotemporally continuous with us; while the universe includes all spatiotemporal
realms that actually exist.32 MVH then says the actual world is much bigger than our
universe. Our universe is only one of many in the universe. Within the universe, FL is
true in some universes (such as ours) and not in others.
The strength of this explanation depends on the details of our multi-verse. Recall
how astronomically fine-tuned our universe appears to be. Without an enormous
number of universes, MVH does not offer a competitive explanation for FL. If MVH
and brute fact are the only options on the table, then obviously any multi-verse
renders FL more likely than it would otherwise be. But if the leading alternative to
MVH is BT rather than brute fact, then we cannot so easily conclude that MVH
would explain FL. FL might support MVH over brute fact, but still support BT even
more strongly.
To get an FTA for MVH over BT, we cannot merely assert that there are many
universes. We also need (a) an estimate of the proportion of possible universes where
FL will be true; and (b) an estimate of the actual number of universes. This requires a
detailed account of the ‘universe-generating mechanism’. MVH must be embedded
in a broader theory. And to provide an alternative to AP, that broader theory must be
a purely naturalistic one, whose universe-generating mechanism comes from physics,
rather than metaphysics, ethics, or theology. (Leslie’s axiarchic plenitude or a multi-
tude of ideas in the mind of God will not do!)
Proponents of MVH offer many such stories. One currently popular explanation is
that universes are generated by a matrix of ‘inflating’ space.33 ‘Bubbles’ within that
matrix cease inflating and become ‘pocket universes’. When this happens, these
pocket universes are said to ‘nucleate’ out of the eternally inflating space. Variations
occur because nucleation is governed by quantum mechanical probabilistic laws.
Given infinite time, all possible low-energy universes eventually arise. As there are
10500 of these universes, this theory predicts a multi-verse containing 10500 universes.
This is only one possible interpretation. MVH is not committed to its details. But
nor can MVH leave all the details open, as it perhaps could if brute fact were its only
rival. We need a multi-verse with a sufficient number of universes that FL is more
likely than it would be if BT were true.
31
As we saw in chapter 3, modal realism alone cannot explain anything.
32
Mellor, ‘Too Many Universes’, p. 221.
33
Inflation is an idea introduced into recent cosmology to explain both the very rapid expansion of the
universe immediately after the Big Bang, and the more or less uniform distribution of matter throughout
the universe. For an accessible overview, see Guth, The Inflationary Universe.
TELEOLOGICAL ARGUMENTS 
4.8.2 Against the atheist multi-verse

MVH is not inconsistent with AP, with axiarchism, or even with BT. Just as any
sophisticated BT can see evolution as God’s way to produce human beings, so any
universe-generating mechanism can serve as God’s way to ensure that FL is true.
Indeed, given the mathematical elegance and explanatory power of MVH, a multi-
verse may be exactly what a perfect God would create. Why settle for one isolated
universe when you could create a plenitude? If we can imagine the glories of a multi-
verse, then so can God.34
Our present concern is with an atheist MVH with no cosmic purpose. Given the
slippery slope from objective value to cosmic purpose, the atheist will hope to also
dispense with the former. However, without objective values, MVH cannot actually
explain FL. Indeed, an FTA without objective values may seem a non-starter. In
section 4.7.3, I argued that FTA must show that FL is remarkable: that it is both
objectively significant and improbable. Without objective values, we have no object-
ive significance, and hence FL is not remarkable.
This is too swift. The atheist can reply that, if FL is sufficiently improbable, then it
stands in need of explanation even if we cannot otherwise demonstrate its objective
significance. The real problem is that, without objective values, the atheist cannot
plausibly claim that FL is improbable. In other words, the atheist has no justification
for denying that FL is a brute fact.
This brings us back to Mellor’s objection. Mellor argues that the proponent of
MVH is not entitled to her initial (im)probability claim that the likelihood of FL is
only 1 in 1040. Without an objective ordering for possible worlds, we can attach no
significance to the claim that FL is more probable if MVH is true, and the FTA for
MVH collapses.
It may seem that AP cannot consistently use Mellor’s objection against MVH, and
then also borrow the FTA for BT. If BT has an adequate response, surely the
proponent of MVH can borrow it? However, this is not so. To offer an alternative
to AP, MVH must be a scientifically respectable explanation with no hint of objective
values or cosmic purpose. The BT response to Mellor explicitly relies on the objective
value of cosmic features such as M, I, or U. The proponent of MVH cannot directly
borrow this response in our dialectical context. Yet it is hard to see where else she
might look. As Mellor himself argues, the only scientifically respectable response is to
accept FL and cosmic coincidence as brute facts.
As we saw earlier, a second BT response to Mellor’s objection is to reject prob-
ability claims altogether. FL is not improbable, but it is still better to explain it than
not. Unfortunately, the atheist cannot borrow this reply either. BT claims that FL
cries out for explanation because, although not objectively improbable, it is
34
Cf. Holder, ‘Fine-Tuning, Multiple Universes, and Theism’. We return to the connection between BT
and MVH in chapter 7.
 THE CASE AGAINST ATHEISM
objectively significant. But, ex hypothesi, our atheist cannot make the latter claim. The
ability to explain insignificant facts that are not improbable does not support a
hypothesis.
Mellor’s objection is thus a decisive comparative blow to any value-free FTA for
MVH.
4.8.3 From value-based multi-verse to cosmic purpose

Once we introduce objective values, MVH can help itself to many features of the FTA
for BT. Suppose FL and M are both objectively significant. (Either because both are
intrinsically significant, or because FL owes its derivative significance to some higher
value such as U.) The significance of M then allows us to claim that FL is improbable.
FL is thus remarkable, and cries out for explanation. MVH can then explain FL by
rendering it much less improbable.
The main question is whether the atheist can borrow this value-based argument
for MVH. As it stands, the argument sits uneasily between atheism and BT. Once we
admit the value of M (or some other cosmic feature needed to generate an ordered
space of possible universes), we are on a slippery slope to cosmic purpose.
Any atheist argument from FL to MVH faces a further hurdle. MVH cannot
explain FL unless MVH is true. But if MVH is true, we can ask why it is true. Can
we explain MVH? To explain MVH, we have the same options as with FL itself:
necessity, brute fact, and cosmic purpose. Necessity will not do. Physical necessity
only operates within the actual world, while MVH cannot be logically necessary.
Most people have no difficulty imagining possible worlds where MVH is false.
Some hope that a TOE will demonstrate the necessity of MVH.35 But this only
prompts the further question: why is this TOE true? And, more importantly, how can
we be confident it is a purposeless TOE? Once we have embraced objective value, can
we really expect a TOE that explains why MVH is true without yielding anything
corresponding to a cosmic purpose?
We are thus left with two familiar alternatives to explain MVH: brute fact or
cosmic purpose. Brute fact may seem hopeless. A multi-verse is much more remark-
able than an ordinary universe. If we didn’t accept FL as a brute fact, how can we
accept MVH?
But perhaps brute fact is more plausible for MVH than for FL. Recall Leslie’s dart
thrown across the universe to hit a red dot. With MVH, the surrounding area is not
the set of universes like this universe, but rather the set of multi-verses like this multi-
verse. As we know very little about other possible multi-verses, we cannot say
whether the surrounding region is mostly blue or mostly red or whatever. We cannot
35
Other physicists argue that any real TOE will explain away the need for anything as inelegant or
metaphysically extravagant as a multi-verse. (Kragh, ‘Contemporary History of Cosmology’; Rees, ‘Other
Universes’; Ellis, ‘Cosmology and Verifiability’; Carr and Ellis, ‘Universe or Multi-Verse?’. An accessible
overview is Woit, Not Even Wrong.)
TELEOLOGICAL ARGUMENTS 
make any estimate of probability. Therefore, we cannot say whether it is remarkable

that a multi-verse like this one contains FL universes. MVH moves us away from the
region of modal space where our judgements of probability might be reliable.36
Even if this brute fact response could dissolve an FTA based on MVH, we would
then have a very compelling CFA. Recall our present dialectical context. The
proponent of MVH must claim that, in the absence of MVH, FL is very improbable.
To capture this claim, she needs an objective ordering of possible worlds. The most
likely candidate to provide that objective ordering is M. Suppose our proponent of
MVH agrees that it is objectively significant that this universe is governed by
mathematical laws. However, if we arrange possible worlds in terms of the com-
plexity of their mathematical laws, then we see that a multi-verse is an especially
complicated world. Even if we cannot assign probabilities, can we reasonably accept
that M is significant, that MVH is only true in a world governed by an incredibly
impressive set of laws, and then simply accept MVH as a brute fact? If possible
worlds are ordered by complexity of mathematical laws, multi-verses will only
appear very far up the scale.37
If the original CFA has any force, then surely the burden of proof is now firmly on
the proponent of MVH. She must explain how her objective ordering of possible
worlds both (a) gives her a high probability for FL, but still (b) allows her to treat
MVH as a brute fact. And she must show this despite the fact that her order is based
on a feature (M) that multi-verses exemplify to a much higher degree than single
universes.
As we saw in chapter 3, we must accept some brute facts. Our current choice is
between FL, (atheist) MVH, and the disjunction of axiarchism and BT. The crucial
question, it turns out, is whether or not we accept objective values. If we do, then it is
very hard to accept either FL or MVH as brute facts. The probability of FL arising by
chance is astronomically low. While we cannot estimate the improbability of MVH, it
is such a remarkable hypothesis that it seems extraordinary to treat it as a coinci-
dence. Given objective values, either axiarchism or BT seems much more palatable
than either atheist option. Alternatively, if we reject objective values, then we have no
reason to prefer MVH to FL as a brute fact. Neither is either objectively significant or
objectively improbable.
In other words, either there is no problem of cosmic coincidence, or (atheist)
MVH is not the solution!
36
Of course, before we can accept this explanation, we need much more detail about MVH and its
TOE. Otherwise, we cannot be confident that it does not involve cosmic coincidences every bit as striking
as those required for a single universe to be FL.
37
Analogous difficulties arise if we recognize the significance of life. A multi-verse contains a lot more
life than a uni-verse. If life has objective significance, then presumably multi-verses are much more
significant than uni-verses.
 THE CASE AGAINST ATHEISM
4.8.4 Does MVH help purposivism?

So far, we have considered an argument for MVH based on the same cosmic
coincidences that ground the FTA for BT. MVH is introduced to explain FL. But
suppose we have independent evidence for MVH. Would this affect our FTA for BT?
Could MVH eliminate the need to introduce God to explain FL?
MVH has been defended empirically. This may seem impossible. By definition,
everything we experience belongs to our universe. We cannot directly observe any
other universe. However, some popular physical theories, such as string theory or
M-theory, imply MVH. If any such theory were independently empirically con-
firmed, then this would indirectly confirm MVH.38 Unfortunately, an empirical
test of string theory—or any rival—is not yet available. There is no certainly no
current scientific agreement that empirical evidence supports a multi-verse. MVH is
speculation, rather than established fact or scientific orthodoxy.39
For the sake of argument, suppose we did come to believe in MVH without relying
on cosmic coincidences. It is usually assumed that this would count against BT. We
can now explain FL without bringing in God. However, as we saw in section 4.8.3,
MVH actually either supports cosmic purpose or fails to explain FL. (And the
consistent atheist nihilist probably finds MVH even harder to accept as a brute fact
than FL!) The possible future development of science does not threaten purposivism.
Indeed, it supports it, because MVH only explains FL if it is embedded in a
purposivist framework.
MVH naturally complements AP. MVH is a non-human-centred view. It pictures
a larger cosmos, with human beings even further from the centre. MVH supports the
argument from scale developed in chapter 7, and is more in keeping with the
grandeur of either a perfect creator God or directly efficacious values. We should
expect either of these to produce a cosmos that is at least as impressive as anything
any human being has ever imagined.
4.9 Can AP Borrow the Fine-tuning Argument?

AP can borrow any teleological argument that does not have essential human-centred
elements. Our FTA cites as evidence the existence of life, the fact that life has emerged
naturally, the cosmic coincidences, and the fact that our universe is governed by
mathematical rules. None of this evidence is human-centred, parochial, or anthropo-
centric. And the explanation it offers for these phenomena appeals to the objective
value of life and of M. (For BT, our universe is FL and M because God chooses it. For
38
Bradley, ‘Multiple Universes and Observation Selection Effects’. Davies explores other possible
evidence for MVH in The Goldilocks Enigma, pp. 200–1.
39
For an accessible overview of doubts about MVH, see Woit, Not Even Wrong.
TELEOLOGICAL ARGUMENTS 
the axiarchist, our universe is FL and M because such universes are the most
objectively valuable.) Again, there is nothing human-centred here.
An austere version of AP could stop here, and deny the significance of I or
U. What is both unlikely and objectively significant is the conjunction of FL and
M. The universe is remarkably friendly to the emergence of life. But given its
friendliness to life, it is not remarkably friendly to the emergence of human life. If
you set out to design a universe that is both M and FL, then it is not surprising that
you end up with some creatures who understand the laws behind that universe.
However, our case for AP is stronger if we can recognize the significance of U. This
strengthens FTA, and grounds an independent CFA. Fortunately, as I argued in
section 4.3, AP can recognize the significance of I and U. While these new commit-
ments do initially suggest a human-centred purpose, they are consistent with non-
human-centred explanations. FTA won’t support AP against BT. But it can support
AP against atheism.
4.10 What Have We Learnt about Cosmic Purpose?

Different teleological arguments teach us different things about cosmic purpose. Let
us be ambitious, and suppose we accept all three arguments defended in this chapter:
human features, cosmic features, and fine-tuning. We have therefore learnt that the
potential for the emergence of conscious, reason-following, value-perceiving agents is
built into the fabric of the cosmos; that the cosmos is designed to be governed by
intelligible mathematical laws; that it is arranged so that, against all odds, creatures
will emerge within it who can comprehend it; and that all these features of the
cosmos are objectively significant. This is more useful than what we learnt in the
previous chapter.
One particular kind of intelligibility is insight either into objective value or into the
cosmic purpose itself. In the next chapter, we explore one argument for BT based on
reports of such insight. This will also enable us to explore the connection between
human understanding and the kind of understanding that has genuine cosmic
significance.
5
Mysticism
There is little difference between the man who drinks too little and sees God, and
the man who drinks too much and sees snakes.
[Bertrand Russell, Religion and Science, p. 188]
My main thesis . . . in the whole book is that Christian Mystical Practice is

rationally engaged in since it is a socially established doxastic practice that is
not demonstrably unreliable or otherwise disqualified for rational acceptance.
[William Alston, Perceiving God, p. 194]
I find unacceptable an a priori rejection of supernatural explanations because of

contemporary modes of inquiry . . . [or] in the name of a dominant current
fashion in explanation.
[Jerome Gellman, Mystical Experience of God, p. 71]
All religious traditions abound with descriptions of religious, mystical, divine, or

miraculous events and experiences. In this chapter, I argue that mystical traditions
may offer a new (and largely unexplored) source of insight into both cosmic purpose
and objective value. In doing so, I borrow from one recent philosophical response to
mysticism: William Alston’s defence of the rationality of a Christian mystical dox-
astic practice.
Secular moral philosophers ignore mysticism. They assume that religious experi-
ence is a delusion, that moral philosophy has nothing to learn from mystical
traditions, and that a deflationary purely materialist explanation of these experiences
either is or soon will be available. (This materialist dismissal of mysticism is perfectly
captured in our epigraph from Bertrand Russell.) In stark contrast, within religious
traditions themselves mysticism is regarded as a respectable source of both know-
ledge about the divine and moral transformation.
I shall argue that even secular moral philosophers should side with believers rather
than atheists. Our prior commitments to methodological pluralism, and to the
inadequacy of global naturalism, predispose us to take mysticism seriously. And
our earlier arguments uncover two important roles for mysticism. One question for
any moral realist is how human beings gain knowledge of objective moral values.
When do we come into contact with this mysterious non-natural realm? Mysticism
offers a solution. Perhaps mystics are moral expects. Perhaps this is how human
MYSTICISM 
beings attain moral knowledge. And, in chapter 4, I suggested that the universe exists
in order that both objective value and cosmic purpose be understood. This left us
with a puzzle: how (if at all) can human beings gain knowledge of these cosmic
values? Mysticism offers a solution here too.
These earlier arguments both mutually support mysticism. On the one hand, they
enhance mysticism’s independent credibility. (If the universe is designed so that its
purpose can be understood, then mysticism is what we should expect.) On the other
hand, mysticism provides new evidence to support our earlier arguments. (Why think
the universe is designed to be understood? Because we have independently credible
reports of cosmic understanding.) The key to these two roles is a close connection
between mysticism and morality, to which we return in section 5.5. This connection
ensures that, like all our earlier arguments, our discussion of mysticism will support
Normative AP rather than Simple AP.
For the analytic philosopher, there is something familiar about the situation of
mystics in the modern world. Mystics remove themselves from the world and its
concerns, speak an arcane language barely intelligible to outsiders, claim esoteric
knowledge of non-physical reality, and struggle to differentiate themselves (in the
public mind or the categories of the airport bookstore) from nutters on the internet.
It is surprising, then, that academic philosophers are not more sympathetic. (Is
mystical theology any more odd than modal metaphysics?) And moral philosophers
should be especially sympathetic. If moral philosophers claim reliable but imperfect
epistemic access to non-natural moral facts, then why shouldn’t some other profes-
sion have a better, deeper—or at least different—access to those same facts? If we
know something about value, then perhaps there are others who know more.
This chapter differs from the previous two in several ways. While cosmological and
fine-tuning are well-established arguments for BT that AP can borrow in their
familiar forms, our argument from mystical experience is constructed from materials
originally designed for a quite different purpose. Alston defends the first-personal
rationality of actual Christian believers. He never claims to provide agnostic third
parties with any reason to accept BT. My focus on Alston may thus seem odd, as
many BT philosophers have offered arguments from religious experience that AP
could borrow more directly. However, those arguments are typically very similar to
other explanatory arguments and thus any discussion of them would largely repeat
earlier chapters. By contrast, Alston’s approach is representative of a distinct
approach commonly known as ‘reformed epistemology’, and therefore the present
chapter illustrates AP’s ability to borrow from a variety of BT philosophical idioms.
Furthermore, exploring mysticism from the inside also offers AP new moral
resources by opening up new connections between human morality and cosmic
purpose.
The philosophical argument of the chapter begins in section 5.2, with an outline of
Alston’s original defence of Christian mystical practice. Section 5.3 develops our
third-personal Alston-based argument for AP and defends my decision to focus on
 THE CASE AGAINST ATHEISM
Alston. Section 5.4 demonstrates how our prior commitments and earlier conclu-
sions support Christian mysticism against familiar objections. Section 5.5 addresses
the moral dimension of mysticism, and shows how AP mutually supports the
traditional moral claims of Christian mystics.
A final distinctive feature of this chapter concerns the evidence. The basic facts
behind our cosmological and fine-tuning arguments are familiar common ground.
Everyone agrees that the physical universe exists and is friendly to life. By contrast, in
this chapter the nature of the underlying phenomena is itself a site of scholarly
controversy and philosophers often operate with obsolete empirical assumptions.
Therefore, in section 5.1 we first ask what mystical experience is.
5.1 The Phenomena: From Isolated Experience

to Contemplative Tradition
‘Religious experience’ is a nebulous term. I concentrate on the narrower phenom-
enon of mystical experience.1 For over a century, academic discussion of mysticism
has been dominated by William James’s magisterial study The Varieties of Religious
Experience. James describes his object of study as follows: ‘In mystic states we both
become one with the Absolute and we become aware of our oneness. This is the
everlasting and triumphant mystical tradition, hardly altered by differences of clime
or creed.’2
James’s approach is psychological, individual, a-contextual, and perennialist. He
treats mysticism as a psychological phenomenon—a subjective experience. The mys-
tical experience is the ‘root and centre’ of all human religion.3 Whatever else it may
contain—scripture, tradition, ritual, theology, social institutions, or moral codes—
religion is everywhere and always founded on the same experience. This presumption
is the key to James’s analysis.
Having decided to look for an experience, James then focuses exclusively on the
experience of ‘individual men . . . in their solitude’.4 James’s Varieties owes its popu-
larity and influence, in large part, to the fascinating array of reports of individual
experiences woven throughout his narrative. While James reports the phenomen-
ology of these experiences in great detail, he ignores their wider context. Many
informants are anonymous. Adherents of different religions, practitioners of wildly
divergent contemplative traditions, and people from different nations or millennia
are lumped together. A medieval monk immersed in a highly structured daily routine
of scripture-based prayer, worship, and contemplation sits alongside some late
1
My interpretation of the scholarship on mysticism is especially indebted to Jones, Mysticism Exam-
ined; Harmless, The Mystics; McGinn, The Foundations of Mysticism; Pike, Mystic Union; and Gellman,
Mystical Experience of God.
2 3
James, Varieties, p. 332. Horne, The Moral Mystic, p. 23.
4
Harmless, The Mystics, p. 16.
MYSTICISM 
nineteenth-century American farmer who experiences an epiphany while walking

through a field.
James’s a-contextualism does not indicate laziness or haste. He is scrupulous to
record those details he considers relevant. A-contextualism reflects a philosophical
presupposition: there is a single mystical experience, and its distinctive features are
exclusively phenomenological. Throughout human history, whatever their superficial
differences, exceptional individuals all share the same experience. This presuppos-
ition is now known as perennialism.5
James’s perennialism is perhaps his most enduring legacy. By the 1970s, perennial-
ism had become the scholarly orthodoxy.6 Unsurprisingly, perennialism especially
appeals to philosophers, who always prefer universal phenomenology to detailed
context-dependent empirical fact. However, perennialism came under sustained attack
in the second half of the twentieth century, along with James’s preoccupation with
individual psychology, and his refusal to consider context.7 Recent empirical studies of
mysticism highlight the role of factors ignored by James—notably scripture, theology,
and institutional setting. While perennialism still has its defenders, it is fair to say
that—at least among historical theologians and scholars of religion—a new orthodoxy
is emerging. Stephen Katz, one prominent exponent, labels this alternative contextual-
ism.8 Kukla and Walmsley summarize the contrast as follows: ‘Whereas [contextual-
ists] claim that the mystical experiences of each culture and era are influenced by that
culture’s or that era’s conception of the world, perennialists maintain that the mystical
experiences in different cultures and eras are essentially the same.’9
Contextualists complain that, by ignoring context to seek a perennial mystical
experience, James does not respect the phenomena, namely the first-personal reports
of actual mystical practitioners. As Bernard McGinn observes: ‘no mystic (at least
before the [twentieth] century) believed in or practised “mysticism”. They believed in
and practised Christianity (or Judaism, or Islam, or Hinduism), that is, religions that
contained mystical elements, as part of a wider historical whole.’10
5
The name is from Huxley’s influential The Perennial Philosophy.
6
Supporters of perennialism include Underhill, Mysticism; Stace, Mysticism and Philosophy; and
Danto, Mysticism and Morality. As we shall see in chapter 9, perennialism has also been very popular
with religious pluralists. For critiques, see Kukla and Walmsley, ‘Mysticism and Social Epistemology’,
p. 144; Katz, ‘Language, Epistemology, and Mysticism’; Zaehner, Mysticism, Sacred and Profane; Harmless,
The Mystics, pp. 254–9; Pike, Mystic Union; and Gellman, Mystical Experience of God.
7
The debate between perennialism and contextualism intersects with another topic beloved of
philosophers: the identification of mysticism with non-conceptual experience. (See, e.g., Stace, Mysticism
and Philosophy.) This identification owes more to perennialist dogma, and a preoccupation with fashion-
able interpretations of Eastern religion, than to any detailed empirical study of actual mystical practices. In
particular, Christian mystics rarely report non-conceptual experiences. (Pike, Mystic Union, p. 115;
Harmless, The Mystics, p. 255.) For our purposes, therefore, we can set this issue aside.
8
Katz, Mysticism and Sacred Scripture.
9
Kukla and Walmsley, ‘Mysticism and Social Epistemology’, p. 144. (Kukla and Walmsley label the
alternative position ‘constructivism’. But I prefer Katz’s ‘contextualism’.)
10
McGinn, Foundations of Mysticism, p. xvi.
 THE CASE AGAINST ATHEISM
The classical mystics—such as Teresa of Avila, or Meister Eckhart, or practitioners

of Zen Buddhism—were, in William Harmless’s apt phrase, members of ‘professional
elites’.11 Far from being isolated individuals blessed with mystical insight, ‘mystics
are almost never prodigies and almost never autodidacts. Most spend long years of
apprenticeship within the confines of their religious communities. There they receive
training at many levels—intellectual, moral, ritual, contemplative.’12 And this training
is always located within some specific religious tradition with its own sacred texts,
conventions, theological presuppositions, and institutions.
James’s a-contextual perennialism enables philosophers to easily dismiss mysti-
cism. Many people report that they have experienced God. But people report all sorts
of things. If phenomenological content is what matters, and if all mystical experi-
ences are phenomenologically indistinguishable, then they must have equal eviden-
tial weight. As some reports clearly carry little weight, so do all the others.
An emphasis on de-contextualized individual experience also supports the
materialist reductionism that James himself dismissed as ‘medical materialism’:
‘the too-simple-minded system of thought which . . . finishes up Saint Paul by
calling his vision on the Road to Damascus a discharging lesion of the occipital
cortex, he being an epileptic. It snuffs out Saint Teresa as a hysteric, Saint Francis
of Assisi as a hereditary degenerate.’13 If we are examining a common psycho-
logical phenomenon, then it is natural to seek a universal psychological or neuro-
physiological cause. If some mystics are mentally ill, then presumably they all are.
(James’s ‘medical materialism’ is the default position of most global naturalists, as
we shall see in section 5.4.6.)
Perennialists first read into mysticism a ‘peculiarly modern bias that locates truth
in the subjective’,14 and then use the unreliability of some putative mystical experi-
ence to undermine them all. But mystics have never shared James’s faith in untutored
experience. For instance, from its origins in the deserts of Egypt in the third century,
Christian mysticism has been obsessed with the possibility of madness, delusion, or
demonic deception.15 Christian mystics developed elaborate techniques to distin-
guish true mystical experiences from false ones. Nor is this preoccupation with
authenticity peculiar to Christianity. Within Sufism (the most prominent Islamic
mystical tradition), the madness of tackling the spiritual path without a teacher was
proverbial; while the Ch’an Buddhists of China (precursors of today’s Zen) went so
11
12
Harmless, The Mystics, p. 243. Indeed, as Harmless notes, the Greek word ascesis—from which we
derive our description of mystics as ‘ascetic’—‘was a sports term before it was a spiritual one’. Its original
meaning was ‘training’. (Harmless, The Mystics, p. 145.)
13
James, Varieties, footnote 15. For a recent example, see Oliver Sachs’s popularization of Charles
Singer’s infamous ‘diagnosis’ of Hildegaard of Bingen as suffering from ‘migraines whose symptoms
included visual auras’. (Sachs, Migraine, pp. 106–8.)
14
15
See, e.g., Brakke, Demons and the Making of the Monk.
MYSTICISM 
far as to issue certificates attesting to the transmission of Dharma from an enlight-

ened master to his disciples.16
No mystic ever regarded experience alone as sufficient. For Christians, experience
is tested against scripture, not the other way around.17 Prior to the seventeenth
century, practitioners did not speak of mysticism, but of mystical theology.18 Recent
scholarship thus rejects the separation, implicit in James’s perennialism, between
mysticism and theology. Many Christian mystics were highly expert theologians or
philosophers.19 Mystical theology was a special route to knowledge of God, but not
an alternative to scholastic theology or philosophy.
The shift from perennialism to contextualism makes it difficult to offer a phe-
nomenal definition of mystical experience to rival James’s famous formulation. For
the contextualist, mystical experience is whatever successful practitioners of estab-
lished mystical traditions actually experience. However, we can make some general-
izations. Mystical experiences typically bring the individual into direct contact with
some divine or supernatural reality. Harmless thus offers this alternative ‘working
definition’: ‘A mystical text is a religious text that describes a profound experiential
knowledge of God or ultimate reality.’20 (Or, at least, this is what mysticism seems to
offer. Whether or not it does is another question, which will occupy us for the rest of
this chapter.)
5.2 Alston on Christian Mystical Doxastic Practice

One contemporary philosopher whose exploration of mysticism reflects the shift
from perennialism to contextualism is William Alston.21 Alston explicitly adopts the
internal perspective of a participant in a religious tradition, rather than the external
perspective of an imaginary impartial observer. Alston argues that, under the right
conditions, the religious person is rationally warranted in retaining her beliefs. He
focuses on one particular religious tradition (Christianity), and one central religious
experience (the perception of God):
16
Harmless, The Mystics, p. 165 and p. 193, respectively.
17
Even the term ‘experience’ can mislead. Introduced into the Western tradition by Bernard of
Clairvaux, it originally lacked the subjective element so central to James. (Verdeyen, ‘Un theologien de
l’experience’; cited in Harmless, The Mystics, p. 280.)
18
19
Consider Augustine, Bonaventure, Eckhart, or even Thomas Aquinas, whose own mystical experi-
ence was so powerful it led him to conclude that all his writings were as straw.
20
21
Alston, Perceiving God. My reading of Alston also draws on Alston, Beyond ‘Justification’; Alston
et al., ‘Symposium on Perceiving God’; and the essays in Battaly and Lynch, Perspectives on the Philosophy of
William P. Alston. More broadly, my philosophical approach to mysticism draws on Kukla and Walmsley,
‘Mysticism and Social Epistemology’; Pike, Mystic Union; Gellman, Mystical Experience of God; and
Plantinga, Warranted Christian Belief.
 THE CASE AGAINST ATHEISM
The central thesis of this book is that experiential awareness of God, or as I shall be saying, the
perception of God, makes an important contribution to the grounds of religious belief. More
specifically, a person can become justified in holding certain kinds of beliefs about God by
virtue of perceiving God as being or doing so-and-so.22
Alston’s key notion is the doxastic practice (DP): a social practice designed to separate
reliable beliefs from unreliable ones, and to enable practitioners to recognize, culti-
vate, and expand their reliable perceptions.
Alston contrasts two DPs. The first is Christian mystical practice (CMP). In CMP,
mystical practitioners acquire knowledge of God. This knowledge is then transferred
to non-practitioners within the broader Christian community, who regard practi-
tioners as reliable experts. (I argue in section 5.5 that CMP also delivers moral
experts, whose moral beliefs outsiders can borrow.) CMP need not involve inference
or argument. Alston observes that, in real life, the religious person typically takes her
experiences at face value. She makes no inference to the existence of God; nor does
she posit God to explain her experiences (or anything else). This is simply how she
sees the world. Just as I do not infer that I now see a tree. (In both cases, argument
comes later, if my initial belief is questioned and I must defend it.) This last example
leads us to Alston’s second DP. Alston contrasts CMP with a sensory doxastic
practice (SP). SP is ‘a constellation of habits of forming beliefs in certain ways on
the basis of input that consists of sense experiences.’23 Alston’s central claim here is
that SP is a sensory doxastic practice. Reliable sensory beliefs are not unproblemat-
ically given to us. We learn to cultivate them using a complex collective practice:
SP. The individual practitioner of SP faces the same questions as the religious person.
Can I trust my perceptual beliefs? And can I rely on the beliefs of other practitioners,
especially experts like scientists or wine tasters?
Unless we are radical sceptics, we accept that some DPs are reliable. That is, they
‘would yield mostly true beliefs in a sufficiently large and varied run of employments
in situations of the sorts we typically encounter’.24 And SP is our paradigm of a
reliable DP. Alston then argues that our grounds for believing that SP is reliable also
carry over to CMP. ‘CMP is rationally engaged in since it is a socially established
doxastic practice that is not demonstrably unreliable or otherwise disqualified for
rational acceptance.’25
Alston begins with a presumption, borrowed from Thomas Reid, in favour of all
socially established DPs. DPs are ‘innocent until proven guilty. They all deserve to be
regarded as prima facie rationally engaged in, pending a consideration of possible
reasons for disqualification.’26 Alston is explicit that contextualism underpins this
reversal of the burden of proof.27
22 23
Alston, Perceiving God, p. 1. Alston, Perceiving God, p. 153.
24 25
Alston, Perceiving God, pp. 104–5. Alston, Perceiving God, p. 194.
26 27
MYSTICISM 
Alston argues that we need this Reidian presumption to avoid global scepticism.
Unless our basic DPs are prima facie reliable, we cannot answer the sceptic. No DP
can be given any external justification. Consider SP. The radical sceptic who simply
refuses to credit the senses at all cannot be refuted on his own terms. We can never
step outside our basic doxastic practices and justify them to a consistent interlocu-
tor who denies their reliability. Once external justification is demanded, none is
forthcoming. This is no idle objection. The history of philosophy abounds with
demands that SP be externally justified. But that history is also a history of failure.
(And, of course, most attempts to provide external support have been very con-
ducive to cosmic purpose: consider Platonic Forms, or Descartes’s or Leibniz’s or
Plantinga’s appeals to God. SP’s need for external support has long been the first
step on the road to God. So atheists had better agree that SP needs no external
support!)
SP is not alone here. Other DPs fare no better. What could externally justify our
knowledge that two plus two equals four? Or consider the enormous range of social
and doxastic practices built on our ability to reliably recognize other people as people.
Can these be externally justified?
We do not doubt that our everyday DPs are reliable. We do not put them on hold
pending a reply to the sceptic. Alston concludes that the lack of external support is a
philosophical problem, and not a pressing practical issue. We should thus abandon
the search for external justification, and deal with each DP on its own terms.
DPs can support one another. For instance, our practices of memory, introspec-
tion, sense perception, and deduction are all mutually supporting. What Alston
denies is that we can ever step outside our package of doxastic practices taken as a
whole. If we carefully work through the ways that different doxastic practices support
one another, we see that any ‘proof ’ of the reliability of a given practice itself invokes
that very practice.28 We cannot break out of the epistemic circle of our doxastic
practices: ‘There is no appeal beyond the doxastic practices to which we find
ourselves firmly committed.’29
Alston’s Reidian presumption is rebuttable. And many socially entrenched DPs
have been rebutted: ‘doxastic practices have fallen by the wayside in the course of
history and prehistory through being undermined by conflict with more firmly
established rivals. This is what has happened with a great variety of magical practices
and practices of divination.’30 Practices can fail because they are internally incoher-
ent, or because they are incompatible with other, more reliable, DPs (notably SP).
A presumption in favour of CMP is not the end of the story, but only the beginning.
28
Here is one example. The defence of any doxastic practice must appeal to claims about memory and
other minds. But how could we justify those practices without trusting some sense perceptions? (Alston,
‘Reply to Bonjour’.)
29 30
 THE CASE AGAINST ATHEISM
Unsurprisingly, Alston’s defence of CMP is not uncontroversial. Some critics

question his general epistemological framework, and especially his Reidianism.31
My aim in this book is to remain neutral wherever I can. For our purposes, the
crucial issue is whether CMP enjoys epistemic credentials comparable to those of
SP—whatever those credentials may be. Defenders of CMP need not buy into all the
details of Alston’s epistemology.32
5.3 From First-person to Third-person

Two features of Alston’s approach seem ill-suited to our project: his emphasis on
Christianity and his first-person focus. Despite these differences, however, I think AP
can borrow from Alston, and also that AP supports CMP. In this section, I defend my
decision to focus on Alston.
Why not replace CMP with a mystical DP based on a non-human-centred cosmic
purpose? One reason is that my aim in part I is to ask whether AP can borrow extant
philosophical arguments. And within analytic philosophy, discussion of mysticism
almost exclusively draws on BT, and specifically on Christianity. Second, although
AP-friendly elements can be found in both Western and non-Western religious
traditions, it is not easy to find an actual mystical tradition that endorses AP. And
the whole point of contextualism is that we need an actually socially entrenched
DP. If AP is to borrow any mystical tradition, it must tread carefully. It thus seems
best to begin with an established non-AP mystical tradition rather than one chosen
for its (potentially misleading) similarity to AP. (We return to the contrast between
AP and CMP in section 5.6.)
Alston’s agenda is modest and first-personal. He does not try to prove the existence
of God to third parties. He merely concludes that the Christian who already trusts in
CMP can rationally ‘sit tight’. Our project adopts the third-personal perspective of
the agnostic observer. Alston thus seems a strange choice. After all, the literature
contains many third-personal arguments based on religious experience.33 Why not
borrow one of them?
As I said at the start of this chapter, one reason I use Alston is to show how AP can
borrow different strands in contemporary philosophy of religion. (By contrast, a
third-party argument from religious experience would be too similar to the argu-
ments canvassed in chapters 3 and 4.) And Alston is not idiosyncratic. First-personal
31
See, e.g., Bonjour, ‘Sense Perception, Epistemic Practices, and Skepticism’; Pasnau, ‘Justified until
Proven Guilty’.
32
Consider the very different epistemological positions of Pike, Gellman, Swinburne, or Plantinga—
who all defend something like CMP. (See the references in footnotes 21 and 33 of this chapter.) Our
broader focus also allows us to sidestep debates about the whether finite physical beings such as ourselves
can truly perceive or experience God. What matters, for us, is whether CMP reliably delivers knowledge of
God (and morality), however we choose to categorize that knowledge.
33
For instance, Swinburne argues that religious experience raises the probability that God exists.
(Swinburne, The Existence of God, pp. 293–307.)
MYSTICISM 
accounts are very popular, especially within the ‘reformed epistemology’ tradition
championed by Plantinga.
There is also a deeper reason to prefer Alston. A third-personal argument would
allow AP to borrow the conclusions of CMP: its beliefs about God. But Alston focuses
instead on the epistemic credentials of CMP practitioners. My hope is that AP can
borrow those credentials, and legitimately treat CMP practitioners as religious (and
moral) experts.
However, before asking whether AP should borrow from Alston, we must first
show that it can. Therefore, the rest of this section shows how to generate a third-
personal response to Alston’s first-personal mystical epistemology.
Philosophers often contrast a purely first-personal approach (such as Alston’s) and
a very ambitious third-personal argument designed to convert the committed atheist.
This gap is very difficult to bridge. My third-personal approach is more modest. I will
argue that, if Alston’s first-personal argument succeeds, then similar considerations
should also persuade moral philosophers sympathetic to our project to take CMP (and
other mystical DPs) more seriously.
We begin with a general epistemological principle adapted from Swinburne:34
Principle of testimony. If x reports that things seem p to x, then (unless there is good reason to
doubt x’s credibility or reliability) this report is reason for others to believe that p.
Swinburne focuses on individual experiences, not doxastic practices.35 But this

difference is easily removed. We need only claim that it is rational to rely on x’s
testimony when x’s belief is itself formed within a prima facie reliable doxastic
practice. This gives us two stronger epistemic principles:
Principle of doxastic testimony. If x reports that things seem p to x, and x’s belief arises within a
reliable DP, then I have reason to believe that p.
Principle of deference. If D is a reliable DP, and x is an expert practitioner of D, then I have good
reason to treat x as a reliable expert within the domain of D.
Most of our knowledge comes via testimony and deference. The crucial question is
whether we have good reason to suspend these principles for reports of mystical
experiences. Do we have good reason to doubt this testimony or these experts?
Kukla and Walmsley draw a useful analogy between the epistemic position of the
philosopher standing outside a mystical doxastic practice, and that of a novice consid-
ering embarking on that practice.36 Imagine a new monk in a medieval monastery
about to embark on an arduous training in the techniques of CMP. Standing outside
CMP, he has no direct access to Alston’s justification. He can only defer to the
34
Swinburne, The Existence of God, pp. 322–4. This principle combines Swinburne’s own principles of
credulity and testimony.
35
Alston, Perceiving God, p. 195.
36
Kukla and Walmsley, ‘Mysticism and Social Epistemology’, p. 141.
 THE CASE AGAINST ATHEISM
judgements of those who are already expert in CMP. But the same is true for any DP. If
it is never rational to defer to experts, then no one will ever master any DP.
If CMP is a reliable DP, then we should treat its practitioners as mystical experts.
(And, I shall argue, as moral experts.) But is CMP reliable? Or is Alston’s Reidian
presumption rebutted in this case? By exploring these questions, we also illustrate the
many ways that AP supports CMP.
5.4 Common Objections to CMP

Opponents argue that, even if there is a presumption in favour of CMP, that
presumption is rebutted. We now examine some common objections to CMP, to
illustrate how our project supports Alston, and what AP might learn from
CMP. I begin with a general observation. Many objections to CMP implicitly deny
doxastic pluralism. This is the most obvious place where our commitments support
CMP. We share Alston’s commitment to doxastic pluralism. In particular, we agree
that SP is not the only reliable DP.
Some partisans of SP claim it is comprehensive. SP is the only reliable DP. Other
apparently distinct practices are merely extensions of SP. (For instance, extreme
empiricists argue that, despite appearances, even mathematics is purely empirical.)
The doxastic pluralist denies that any DP is comprehensive. SP is certainly not
comprehensive, because many uncontroversial areas of human knowledge involve
domains not accessible to SP. This is why empirical reductions of those domains fail.
In earlier chapters, we encountered several areas of knowledge that are not reducible
to SP. The doxastic pluralist’s favourite examples are mathematics, logic, and modal-
ity. In chapters 2 and 3, we implicitly agreed that these DPs are not merely extensions
of SP. Another popular analogy is our knowledge of other minds.37 Chapter 4
defended the autonomy of this DP. If we accept Nagel’s anti-materialist analysis of
consciousness, then introspection and the recognition of other minds are not redu-
cible to SP as it is understood by most atheists.
Doxastic pluralism allows CMP to draw analogies with these other reliable DPs. If
our practices of mathematics and recognizing other people are not reducible to SP,
then such reduction cannot be a necessary feature of any reliable DP. Given our
commitments, the strongest analogy is morality. CMP claims knowledge of a distinct
realm of non-natural facts. If this claim were unique, then CMP’s credentials would
be suspect. But we have already recognized one other non-natural DP: morality itself.
If we accept some human knowledge of non-natural moral facts, then we should at
least accept the conceptual possibility of human knowledge of the divine. (Indeed, as
we’ll see in section 5.5, the connection between morality and mysticism may be even
more intimate.)
37
This example owes its prominence in recent debate to Plantinga, God and Other Minds,
pp. 187–211.
MYSTICISM 
Our commitment to avoid caprice also supports CMP, because it makes us wary
of the standard atheist third-personal dismissal of mystical experience. I am not a
practitioner of any mystical doxastic practice, nor do I have any mystical experi-
ence. It is not that I have found some mystical tradition wanting, but rather that
I have never seriously tried. And among secular philosophers, I imagine my
situation is the norm. In such circumstances, it is presumptuous to reject mystical
DPs out of hand.38
Of course, aversion to caprice also cautions practitioners of CMP to be wary of
overstating their case. But this worry is not peculiar to CMP. Every minority doxastic
practice must guard against the temptation to replace rigour and discipline with the
smug assertion of esoteric knowledge. Yet many minority practices do offer genuine
experiential knowledge—think of music appreciation, wine tasting, moral sensitivity,
philosophical insight, or science itself. We should be wary of minority DPs, but they
cannot be dismissed altogether.
Having seen how our project supports CMP in general, we now proceed to
examine specific objections to CMP.
5.4.1 Lack of empirical verification
One common objection is that, if mystics were reliable, they would make verifiable
predictions. But they fail to do so. Therefore, CMP is not reliable.
This objection nicely illustrates a general point about Alston’s defence of
CMP. Alston diagnoses two unfair attacks on any doxastic practice. The imperialist
unfairly judges one doxastic practice against the standards of another. The dis-
tinctive features of one reliable practice are presented as necessary conditions of
reliability. This is clearly unreasonable. If there are even two distinct reliable
practices, then each will lack some features of the other. A second unfair attack
is to apply a double standard, where we expect one doxastic practice to meet a
standard not required of others. The unreasonableness of this demand is even more
obvious.
Alston argues that the demand for ‘verification’ always involves either imperial-
ism or a double standard. It is imperialist to demand empirical verification, because
this treats SP as normative for all doxastic practices. But to demand external
justification introduces a double standard, because SP itself cannot be externally
justified.
If we take doxastic pluralism seriously, then this particular objection fails. How-
ever, it is closely related to other, more plausible challenges, to which we now turn.
38
As Wendy Donner notes, this is one place where even Bentham’s greatest disciple displays caprice:
‘Although as a skeptic [Mill] has not taken part in religious institutions and groups or undergone any of the
techniques of prayer or meditation which have been used in many spiritual traditions to make students
receptive to the sacred, he feels entitled to judge and dismiss the claims of those who have undergone these
processes of development of their spiritual capacities.’ (Donner and Fumerton, Mill, p. 142.)
 THE CASE AGAINST ATHEISM
5.4.2 The Great Pumpkin Objection

In his criticism of Swinburne’s argument from religious experience, William Rowe
complains that mysticism lacks an ‘effective over-rider system’.39 Not all experiences
are reliable. People report seeing all sorts of crazy things. Recall Russell’s provocative
remark: what distinguishes the man who sees God from the man who sees snakes?
Plantinga christens this the Great Pumpkin Objection, after the Peanuts comic strip.
How do we separate the Christian’s belief in God from Linus’s belief in the Great
Pumpkin?40
Alston’s reply to Rowe appeals to contextualism. Each mystical tradition is
embedded in a religious community with sources of authority (beyond the individ-
ual’s own experiences) to distinguish reliable perceptions from unreliable ones, and
techniques to develop one’s mystical aptitude and make one’s perceptions more
reliable. Every mystical tradition thus has a very developed over-rider system. This
separates mysticism from madness and delusion, which operate outside any such
communal tradition.
Alston offers the following criteria drawn from the teachings of ‘spiritual directors
and mystical theologians’: genuine mystical experience is characterized by interior
peace, trust in God, patience in pain, simplicity, sincerity, and charity.41 (As this list
suggests, a moral test is central to CMP’s over-rider system. We return to this moral
test in section 5.5.)
We now see the benefits of Alston’s doxastic practice approach, and his context-
ualism. The perennialist who emphasizes isolated individual experiences has no reply
to the Great Pumpkin Objection. But the contextualist has a very powerful reply. Our
project supports CMP here, because it supports contextualism. One implication of
methodological pluralism is that we should defer to experts in other fields. And the
consensus of relevant expert scholarly opinion favours contextualism. The failure of
naturalist philosophers to defer to (or even investigate) that consensus is a clear case
of caprice!
The lesson we took from the scholarly literature is that mystical theology is closer
to the collective intellectual endeavours of a scholarly community than to the isolated
ravings of a disturbed individual. Pike draws an interesting analogy between mystical
theology and history, two disciplines ‘where the facts are observed, but the investi-
gator has no direct access to them’.42 In both cases, this indirectness can lead to
scepticism. If we give historians the benefit of the doubt, then why not extend it to
mystical theologians as well? (Imagine how impoverished contemporary moral
philosophy would be if we took historical scepticism seriously.)
39
Quoted in Alston, Perceiving God, pp. 209–10.
40
Plantinga, Warranted Christian Belief, pp. 342–5.
41
Alston, Perceiving God, pp. 202–3. (cf. Pike, Mystic Union, p. 35.)
42
Pike, Mystic Union, pp. 167–8.
MYSTICISM 
CMP’s over-rider system involves input from other DPs. (As we saw earlier, Alston
acknowledges that our DPs form a package, and cannot be evaluated in isolation.)
Consider our imaginary medieval monk, wondering whether to embark on a life
immersed in CMP. Suppose he worries whether this arduous exercise is simply a
waste of time. How can he be sure to avoid delusion?
Our novice’s superiors would have much to say in favour of the reliability of CMP
(or ‘mystical theology’, as they would call it), building on other medieval monastic
DPs.
1. We know that God exists—not solely from individual experience, but also via
reason and revelation.
2. Reason and revelation also independently confirm the picture of God revealed
by CMP.
3. Reason and revelation lead us to expect human beings to be capable of
acquiring the knowledge that mystical theology seeks, because this is part of
God’s plan.
4. We expect any successful mystical theology to be morally transformative, and
this is exactly what we do find.
These claims could also be offered to aspiring practitioners of CMP today. And, to
a surprising degree, our project supports similar claims (once the anthropocentric
bias of BT is removed).
5.4.3 CMP is not as reliable as SP
Alston himself lists three desirable features of SP that CMP appears to lack.43
1. Cross-checking. SP is built on five different senses. Information from one sense
checks the reliability of another, and scientific methods offer additional cross-
checking. CMP has only one source of information—mystical experience.
2. Complexity. SP provides an enormously complex and ever-expanding picture of
its world; whereas CMP has not taught us anything new about God for some
time.
3. Universality. Almost everyone employs SP. This facilitates cross-checking. It
also means that the sceptic’s demand for external justification is purely hypo-
thetical. Practitioners are never really confronted by an outsider who asks them
to justify SP. By contrast, mystical experiences are not shared by everyone, and
sceptics are real.
An extreme doxastic pluralist would insist that different doxastic practices are
incommensurable, and therefore any comparison between SP and CMP is irrelevant.
However, this would leave CMP vulnerable to the Great Pumpkin Objection. If we
43
Alston, Perceiving God, ch. 6.
 THE CASE AGAINST ATHEISM
cannot compare doxastic practices, how can we say that any are unreliable? How can
CMP distinguish itself from astrology?
If we admit the comparison with SP, three defences are available. (1) CMP has
other desirable features that SP lacks; or (2) CMP does possess the three listed
features, perhaps under new interpretations; or (3) while not as reliable as SP,
CMP is reliable enough.
CMP draws support here from other reliable DPs. Our mathematical DP lacks the
range of distinct cross-checking devices available in SP. Yet mathematics has coun-
terbalancing advantages that seem to make it more reliable than SP. Our mathem-
atical knowledge is a priori, and attains a degree of certainty unimagined in SP. We
don’t feel the need to confirm a mathematical proof using our senses! Indeed, many
philosophers regard mathematics as the epistemic gold standard that SP itself fails to
reach.
Similarly, BT rationalists regard our knowledge of God as more reliable than
SP. (Descartes’s response to scepticism only works if the reader agrees that his
knowledge of God is more certain than the deliverances of his senses.) This is credible
if BT is based on a priori arguments (chapter 6), or on a cosmological argument
backed up by the principle of sufficient reason (chapter 3). Or perhaps this superior
knowledge of God comes from CMP itself. Mysticism offers direct access to its object,
unmediated by the unreliable and distorting prism of the human senses.44
CMP might therefore be superior to SP even if it lacks the three features Alston
cites. But perhaps CMP does possess those features in its own way. Other established
DPs have their own cross-checking. Mathematicians check intuition against rigorous
proof, or proof in one branch of mathematics against proof in another; our moral DP
balances intuitions about specific cases, general principles derived by rational argu-
ment, meta-ethical constraints, and the opinions of recognized moral experts; we re-
identify persons with many mutually reinforcing criteria; and so on. CMP’s own
over-rider system seems to offer cross-checking of comparable sophistication, as
witnessed by the roles of revelation, reason, and tradition.
Or consider complexity. Is there any neutral sense in which our practices of
mathematics and logic, our moral systems, and our interpersonal arrangements are
less epistemically complex than SP? Under perennialism, of course, mystical experi-
ence is perfectly simple. But the contextualist explicitly denies that CMP lacks
complexity. And all mystics agree that CMP delivers the degree of complexity
appropriate to its subject matter. (If a DP seeks knowledge of a realm of absolute
unchanging simplicity, then lack of complexity would be a mark of reliability.)
Atheists often argue that, while SP continually offers new insights, CMP never
teaches us anything new or surprising. Perennialists reject this demand for
44
This line of thought will especially appeal to dualists. This is yet another place where the affinity
between BT and dualism might assist our case against atheism. We explore AP’s complex attitude to
dualism in more detail in chapter 10.
MYSTICISM 
innovation as imperialist. Mystical insight is essential for supernatural knowledge.

Without mystics, we would know nothing about the supernatural realm. But later
generations of mystics cannot acquire new supernatural information, because mys-
tical experience by definition is always and everywhere the same. Contextualists can
borrow this reply, but they need not rely on it. Contextualists agree that it is
imperialist to expect mysticism to mirror the constant innovation of science. But
they also deny that CMP is epistemically inert. CMP is open to change. Mystical
experience need not be unchanging, and the other DPs in CMP’s over-rider system
certainly can broaden our knowledge of God or God’s purposes, and therefore teach
us new things about the objects of mystical experience. As I argue in section 5.5, these
new lessons of CMP may include new moral lessons.
5.4.4 Must a reliable DP be universal?
Our third distinctive feature of SP is universality. CMP is clearly not universal. But is
this a defect? Some reputable DPs are extremely rare. Most people cannot do
mathematical logic. Does this undermine its epistemic credentials? Isn’t it sufficient
that practitioners have an over-rider system that they can operate? ‘Why suppose that
the outputs of a practice are unworthy of acceptance because it is engaged in by only a
part of the population?’45 Indeed, few DPs are truly universal. Not everyone grasps
even the rudiments of morality, and some individuals cannot recognize other
persons. But we don’t really think this undermines our moral or interpersonal DPs.46
Minority DPs are often continuous with more common ones. Modal philosophy
and higher mathematics are simply more rigorous developments of intellectual
faculties that we all possess to some rudimentary degree. (The same is true within
SP itself, where wine tasters or experimental scientists deploy more developed senses
rather than entirely new ones.) Perhaps the heights attained by mystics are continu-
ous with more everyday religious experiences. And rudimentary religious experience
is not that uncommon.47 A minority DP requiring years of training and discipline
may also be available to many more people than actually take advantage of it.
Perhaps most serious practitioners do (eventually) see God.
Lack of universality is not, per se, any objection to a doxastic practice. But
partisans of SP will now argue that they can explain why some unfortunate individ-
uals lack one sense or another, and perhaps why some might even lack all senses.
Science can tell us why this person is blind or that one is deaf. CMP cannot do the
same. It is simply a mystery why religious experience is not universal.
This new objection threatens every doxastic practice distinct from SP. After all,
mathematics does not itself explain why some people are no good at maths. It relies
on SP to explain this. (And it is not easy to explain!) But if the Christian philosopher
45 46
Alston, Perceiving God, p. 169. Cf. Plantinga, God and Other Minds, pp. 187–211.
47
Alston cites a US survey suggesting that more than 50 per cent of people report having experienced
God in their lives. (Alston, ‘Religious Language’, p. 231.)
 THE CASE AGAINST ATHEISM
is allowed to appeal to her own broader package of doxastic practices, then she too
will claim to explain CMP’s relative rareness. Perhaps God’s plan requires that God
remain at least partly hidden from human beings.48
We return to religious diversity in section 5.4.8, and at length in chapter 9, where
I argue that this particular BT explanation fails, and that BT cannot explain why a
benevolent God would hide from human creatures or permit religious diversity. At
this point, AP must substitute a different explanation, based on the gap between
God’s purposes and human capacities. My present claim is merely that the need to
explain the rarity of CMP does not, in itself, render CMP unreliable.
I conclude that the differences between CMP and SP do not automatically discredit
CMP. But perhaps they do suggest that SP is more reliable. Mystical perception is
controversial, its claims go beyond the physical world, and it is regarded (even by its
defenders) as a rare and difficult capacity.
Alston himself once argued that CMP is as reliable as SP.49 Their relative advan-
tages and disadvantages cancel out. His later work is more modest. Alston accepts
that CMP is less reliable than SP, but denies that this shows that CMP is unreliable.
After all, no one seriously believes that SP only just meets the threshold for reliability.
Suppose we discovered tomorrow that sense perception is slightly less reliable than
everyone had hitherto believed. (Perhaps a psychological experiment reveals that our
senses systematically misrepresent the size of objects, or some epistemologist
uncovers a new argument for scepticism.) We would not suddenly become sceptics.
Nor would we abandon SP. Or imagine a species with a rudimentary SP based on a
single sense, delivering only a vague picture of the world. Their sensory doxastic
practice could still be reliable. Gellman puts it well: ‘Our ordinary physical-object
beliefs are way overjustified by confirming evidence.’50 SP is extremely reliable. CMP
could be much less reliable and still be reliable enough. And our discussion thus far
suggests that (apart from difficulties relating to religious diversity that we defer until
part II), CMP is reliable enough.
However, this modest strategy has a price, as Alston himself notes. If SP is more
reliable than CMP, then it may undermine it.
5.4.5 Does SP undermine CMP?
In response to Rowe’s objection to Swinburne, Alston embeds CMP in a historical
religious tradition. This makes CMP vulnerable to empirical refutation. If SP falsifies
CMP’s background claims, then it indirectly undermines CMP. Christianity makes
both metaphysical and historical claims. Atheists argue that SP falsifies these claims:
science proves the non-existence of God, while history falsifies Christianity’s histor-
ical claims.
48
49
On the evolution of Alston’s view, see Alston, Perceiving God, ch. 6.
50
Gellman, Mystical Experience of God, p. 27.
MYSTICISM 
Alston’s reply separates science and history from philosophical theories such as
physicalism or materialism. He argues that no neutral scientific or historical inves-
tigation can demonstrate that there is no God, or that Christ did not rise from the
dead. A conflict with philosophical materialism is not a conflict with SP itself.51
We can sidestep this debate here. If there is a decisive SP-based argument against
those of CMP’s background claims that are relevant to AP, then it belongs elsewhere
in this book. (And, in part II, I argue that there are such arguments.) In this chapter,
therefore, we assume that other arguments (whether based on SP or not) do not
decisively refute CMP. (After all, many atheists reject mysticism because they find the
very idea of God incoherent. But this is not an appropriate response at this point in
our inquiry!)
5.4.6 A deflationary explanation
SP might undermine CMP in a more subtle way, by providing an alternative
explanation for mystical experiences. This is the goal of James’s medical materialist,
and of global naturalists down the ages. We must tread carefully here. BT claims that
mystical experience is caused by God. The mere availability of a ‘natural’ explanation
does not necessarily undermine BT’s claim. Indeed, an explanation of the experi-
ence’s origins could even support BT. After all, BT accepts that God sometimes uses
natural mechanisms to achieve some purposes. Why shouldn’t God use this particu-
lar mechanism to communicate with human beings?
A naturalist explanation only threatens CMP if it is deflationary. A deflationary
explanation undermines the epistemic credentials of a belief. Classic examples
include Nietzsche’s ‘genealogical’ explanation of our current moral beliefs; psycho-
analytic explanations of religious beliefs in terms of sexual repression or various
‘complexes’; or false memory syndrome explanations of alleged memories of alien
abduction. Deflationary explanations are also often evolutionary, as when moral
nihilists use the natural history of our moral beliefs to undermine those beliefs.
By definition, a deflationary explanation would undermine CMP’s epistemic
credentials. CMP must insist that no deflationary explanation has yet appeared.
We have no natural explanation of mystical experience, because the broader search
for a naturalistic account of experience and consciousness in general is still in its
infancy.
Many atheists reply that, whether or not we have found it yet, there must be a
materialist explanation for mystical experiences—because there must be a materialist
explanation for any observable phenomenon. But this presumption is a dogma that
CMP simply rejects. Gellman dubs it the disappearance theory, and dismisses it as
follows: ‘I find unacceptable an a priori rejection of supernatural explanations
51
Alston, Perceiving God, p. 245. See also Plantinga, Warranted Christian Belief, ch. 12.
 THE CASE AGAINST ATHEISM
because of contemporary modes of inquiry . . . [or] in the name of a dominant current

fashion in explanation.’52
The disappearance theory is presumptuous for two reasons. First, for all anyone
knows, mystical phenomena may have no materialist explanation at all. Even the best
materialist explanation may fail to capture some vital feature of mystical experience.
Here are four possible gaps in the naturalist picture.
1. If the universe is not fully deterministic, then no complete materialist explan-
ation of mystical experiences is possible.53 Neuroscience cannot distinguish
random quantum events from divinely caused ones. Perhaps God is the imme-
diate cause of quantum events leading to mystical experiences.
2. If substance dualism is correct, and no human experience is purely physical,
then physical processes cannot exhaust the causal story about mental events,
and we must invoke God (or some other non-physical cause).54
3. If Nagel is correct, and ordinary facts about consciousness and reason require a
radically non-materialist paradigm shift, then the eventual ‘scientific’ explan-
ation of mystical experience may well be perfectly consistent with CMP’s claim
that such experiences put us in touch with cosmic purpose.
4. If non-naturalism is the correct account of morality, then there is no naturalist
explanation of morality, and therefore any natural explanation of mysticism
cannot capture its moral dimension. Needless to say, things are even worse for
the global naturalist if moral supernaturalism is correct! (We return to these
possibilities in section 5.5.)
This brings us to the second way that the disappearance theory is presumptuous.
Because not every explanation (whether materialist or not) is a deflationary one, we
cannot assume in advance that the best explanation, when it is found, will undermine
CMP. Perhaps, once science does explain mystical phenomena, it will cite God or cosmic
purpose. This is not idle speculation. One popular neurophysiological account is that
developed by d’Aquili and Newberg, who explicitly do not regard it as deflationary.55
Indeed, they develop their own theology based on their neuroscientific explanation.
As with the cosmic coincidences discussed in chapter 4, the insistence that there
must be a deflationary scientific explanation of mysticism simply begs the question.
5.4.7 CMP is dispensable
Partisans of SP argue that, while SP is an inescapable feature of the human condition,
CMP is optional. Human beings cannot survive without relying on their senses, but
52
Gellman, Mystical Experience of God, p. 71.
53
Mawson, ‘How Can I Know I’ve Perceived God?’, p. 116.
54
Mawson, ‘How Can I Know I’ve Perceived God?’, p. 115. Here we see, yet again, the mutual support
between dualism and BT, to which we return in chapter 10.
55
See, e.g., d’Aquili and Newberg, The Mystical Mind. For discussion, see Albright, ‘Neuroscience in
Pursuit of the Holy’; Gellman, Mystical Experience of God, pp. 75–102.
MYSTICISM 
they can survive without mysticism. The sceptic’s challenge to CMP has a real-life
force that his challenge to SP lacks. The lack of external justification is a prima facie
objection to any doxastic practice. We must (reluctantly) accept indispensible prac-
tices such as SP. But we should reject those, like CMP, that are merely optional.
This response is especially salient for third parties. Perhaps Alston’s argument is
sufficient for the Christian who has already built her life around the reliability of
CMP. But, for those who are not already invested in it, CMP is best ignored if we
want to avoid false beliefs.
This sounds like good advice. But is it? We must first ask what it means to say that
one DP is indispensible while another is optional. One interpretation is psycho-
logical. SP is indispensible because people cannot abandon their sensory beliefs. But
this seems too weak. (Psychological indispensability could indicate madness or
fanaticism rather than knowledge.) We need something less subjective. The most
obvious interpretation is survival: individuals who dispense with SP will perish. (No
sceptic has ever really abandoned all his empirical beliefs.) As we’ll see, even this new
criterion is problematic. But, for the moment, we grant (for the sake of argument)
that SP is necessary for individual survival while CMP is not. Does this imply that
(unlike SP) CMP is dispensable?
CMP rejects this suggestion as doxastic imperialism. SP enhances survival because
it provides knowledge about the external world. SP is designed to produce that
knowledge. So survival-enhancement is a good indicator of SP’s reliability. But
other reliable doxastic practices do not seek knowledge of the external world. So
why should they be survival-enhancing? Higher mathematics famously didn’t help
Pythagoras to survive. But so what? Perhaps there are domains where knowledge has
no practical benefit whatever. This might render the corresponding doxastic practices
pointless, but it doesn’t undermine their reliability.
In our present dialectical context, SP has a reply. Our current concern is not
reliability, but indispensability. Perhaps no doxastic practice is reliable, and we would
like to dispense with them all. We are stuck with SP, but not with CMP.
CMP will reply, in its turn, that other DPs offer their own alternatives to indis-
pensability, based on values other than survival. Our mathematical doxastic practice,
while perhaps not indispensible, is irreplaceable. Mathematical reasoning provides
knowledge we could not get any other way, because it covers a realm of fact that SP
cannot reach. We cannot dispense with mathematical reasoning without significant
epistemic loss.
If SP is not a comprehensive practice, then we must accept a plurality of irreplace-
able doxastic practices, each offering the best knowledge available in its own realm.
However, while indispensability requires irreplaceability, the reverse implication
does not hold. No replaceable doxastic practice is indispensable, but an irreplaceable
practice is still dispensable if its unique knowledge is itself dispensable. And sceptics
will then insist we should only enter a dispensable domain if our DP is more reliable
than SP.
 THE CASE AGAINST ATHEISM
The strongest response to this scepticism is to show that CMP is indispensible—not

for mere survival, but for a flourishing human life. Perhaps CMP provides irreplace-
able knowledge of God, without which human life is diminished. Or perhaps what is
essential to a truly flourishing life is the moral knowledge that only CMP offers.
These claims may seem implausible. Can’t people flourish without practising any
mystical DP? Don’t many atheists live well? One ambitious reply, found in many
religious traditions, is that earthly flourishing is as nothing against true flourishing—
where the latter does require a direct relationship with God.56 Practising CMP might
enlarge our notion of what it is to flourish. We would then recognize that CMP was
indispensible.
A less ambitious reply is that CMP is essential to some valuable but non-essential
component of a rich human life. Under the objective list theory introduced in
chapter 1, this kind of indispensability may be quite common. A direct relationship
with God, or a knowledge of one’s place in the cosmos, is found on many objective
lists. Even though these items need not be present in every flourishing life, they do
enrich human lives. If CMP is indispensible for these items, and if other minority
doxastic practices with similar epistemic credentials underpin other list items, then
we might construct a disjunctive indispensability argument. One can only flourish if
one participates in some minority doxastic practice such as CMP. Our commitment
to the objective list theory thus supports CMP. (We pursue these issues further in
chapter 12.)
Whenever we adopt any DP, we must balance epistemic risks. If we form new
beliefs, we run the risk of building our lives on falsehood. If we never form new
beliefs, our lives will be very narrow. To estimate the costs and benefits of trusting
CMP, we must judge both CMP’s reliability and its contribution to human well-
being. As I argue in section 5.5, the latter may be very significant indeed. If so,
perhaps the risk of falsehood is worth taking.
5.4.8 Religious diversity
Alston rejects a perennialist approach to mysticism, partly because any imaginary
universal mystical doxastic practice would lack an over-rider system. (The contra-
dictory claims of different religious traditions would undermine one another.) He
thus focuses on individual religious traditions such as Christianity. And Alston’s
first-personal defence of CMP is modest. While Christians can reasonably trust
CMP, Alston explicitly leaves open the possibility that practitioners of other (non-
Christian) mystical DPs may enjoy the same epistemic rights.
This raises the problem of religious diversity. Why doesn’t the existence of all these
equally reliable (but contradictory) mystical DPs undermine CMP? And religious
56
For some, this supreme human good is only available in heaven. But others think we can at least
glimpse it in this life. We return to that debate in chapter 10.
MYSTICISM 
diversity is obviously much more worrying for third parties. We return to this
objection in chapter 9, where I argue that it is fatal for BT, but not for AP.
Setting diversity aside, however, I provisionally conclude that CMP is not
decisively refuted. It is reasonable for both practitioners and outsiders to treat
CMP as a reliable source of knowledge. We now explore the content of that
knowledge, and especially its connection to morality and cosmic purpose.
5.5 The Moral Test: How Morality Mutually

Supports CMP
If CMP has an over-rider system, then it needs ways to distinguish genuine mystical
experience from error and delusion. As we saw earlier, several items on Alston’s list
of possible criteria are explicitly moral. Nor is he alone. Many commentators regard
moral criteria as the definitive test of true mystical experience.57 The successful
mystical practitioner is morally superior to the non-practitioner in many ways,
displaying exceptional moral insight and exceptional moral virtue. Her compassion,
benevolence, moral purity, and lack of egoism bespeak a supernatural origin. No one
could be that good, or that wise about goodness, who had not been touched by the
divine.
Alston himself places great weight on the moral test. ‘I know of no canonical
systematization of these criteria [of genuine perceptions of God], but it seems to me
highly plausible to take the moral criteria to be the most fundamental.’58
For our project, this link to morality is mysticism’s most striking feature. What
exactly does the moral test show? Suppose CMP does exhibit the desired moral
improvement. Why should this matter?
For CMP practitioners, and their co-religionists, moral improvement is an end in
itself. But we seek an epistemic pay off. Recall that CMP, on Alston’s interpretation, is
itself a doxastic practice, subject to epistemic criteria. If the moral test is a credible
test, moral improvement must be evidence of epistemic reliability in CMP’s chosen
domain. And that domain is knowledge of God (or supernatural reality more
broadly). The moral test is only credible if moral improvement tracks access to
supernatural facts. And this, in turn, requires a link between supernatural facts and
moral facts. The epistemic pay off is now clear. If CMP passes the moral test, and if
that test is credible, then CMP will deliver insight about both supernatural and moral
57
In addition to Alston, both Pike, Mystic Union, and Gellman, Mystical Experience of God, defend a
tight connection between morality and mysticism. For further general discussion, see Horne, The Moral
Mystic; and Jones, Mysticism and Morality. On the moral test in particular, see Harmless, The Mystics, esp.
p. 189; and Heck, ‘Mysticism as Morality’. On the other hand, Danto, Mysticism and Morality, objects that
mysticism undermines morality, because it obliterates the morally central distinction between individuals.
However, Danto bases his argument on a perennialist non-conceptual account, which is not relevant to
CMP as it is actually practised. (For more on non-conceptualism, see footnote 7.)
58
 THE CASE AGAINST ATHEISM
facts, and we will be justified in treating its adherents as both supernatural and moral
experts.
However, this analysis prompts an obvious reply. Many contemporary analytic
philosophers will regard the moral test as absurd. How could improved behaviour be
evidence of the veracity of experience? Why should moral insight or moral improve-
ment be evidence of anything supernatural? Aren’t there perfectly adequate natural-
istic explanations of human beliefs and actions? What is so remarkable about
compassion, benevolence, altruism, or selflessness? Aren’t most ordinary people
capable of behaving perfectly well without any inkling of mystical experience? We
don’t normally expect a reliable DP to improve people. Scientists, wine tasters, or
even moral philosophers, are not necessarily better people. Why is CMP so different?
I shall argue that, if we grant a number of controversial views in meta-ethics,
normative ethics, and moral psychology, then the moral test is very plausible. And
our broader project, in turn, supports those very assumptions. Given our moral
commitments, the moral test has considerable force.
The most straightforward connection is meta-ethical. The moral test presupposes
moral realism. For the non-cognitivist or moral nihilist, even if moral improvement
makes sense, it surely cannot involve insight into any reality. Moral improvement is
not a reliable indicator of knowledge. It is no coincidence that BT mystics are
invariably moral realists. Indeed, some commentators regard moral realism as
essential to mysticism: ‘The heart of the religious perspective . . . is the conviction
that the values one holds are grounded in the inherent structure of reality.’59
The moral test has additional meta-ethical implications. Not all moral realists will
be impressed. Moral naturalists may agree that veridical experience of moral facts
leads to both superior moral knowledge and moral improvement. But moral facts are
themselves natural facts. Therefore, moral improvement does not indicate supernat-
ural insight. Atheist non-naturalists will be similarly unmoved. Moral facts are sui
generis, with no connection to God or cosmic purpose. If God is only incidentally
related to morality, then why would an experience of God have moral effects?
The answer, of course, is that, for the Christian, morality is supernatural. The
meta-ethic that best supports the moral test is moral supernaturalism, where moral
facts are facts about God. But another congenial meta-ethical picture is one where,
although moral facts are ontologically independent of God, they constitute God’s
reason to bring the cosmos into existence. This supernatural meta-ethical disjunction
was the main lesson of chapter 2. Therefore, to experience God is either to experience
moral facts directly, or to encounter someone uniquely placed to reveal moral truth.
(The moral test also operates in non-theist mystical traditions that posit a similarly
tight connection between moral and supernatural facts. We return to non-theist
religions in chapter 9.)
59
Geertz, Islam Observed, p. 97.
MYSTICISM 
However, a congenial meta-ethic is not sufficient. Even if we assume that God

exists, that moral facts are (or are closely related to) facts about God, and that God
wants human beings to learn moral facts and undergo moral improvement, we still
might think that mystical experience is superfluous. God has many other ways to
reveal morality. Indeed, given the difficulty and rarity involved in CMP, we might
expect a benevolent God to choose something simpler and more universal, such as
conscience or everyday moral practice. Morality would have no need for any special
moral insight available only to mystical experts. Moral improvement would then fail
to provide evidence of mystical success.
In chapter 1, I distinguished two general approaches to normative ethics: compla-
cent and austere. The moral test is in tension with complacent moral philosophy. If the
test of an ethical theory is its closeness to our considered moral judgements, then the
basic principles of morality must be common knowledge. And if no acceptable moral
theory can demand more than we ordinarily regard as reasonable, then everyday moral
behaviour must be relatively okay. Complacent moral philosophy has no need for
supernatural insight to make us better people. Morality is a human-sized phenomenon,
and our everyday level of knowledge and attainment is all we need.
The moral test makes sense only if we reject this complacency, and instead
embrace the austere moral picture I defended in chapter 1. We now expect a very
demanding normative ethic that conflicts with conventional human morality. At the
very least, the moral test presumes a very demanding moral ideal. The vast majority
of human lives fall short of what is required—leaving room for moral behaviour that
is exceptional, not only in the statistical sense of being uncommon, but also in the
moral sense of being clearly superior. CMP’s moral claim is not that its practitioners
are different from the rest of us, but that they are better. (And, at the extreme, that
they alone are good.)
One very striking feature of moral reports about mystical practitioners is their
extreme impartiality. They really do seem to adopt a perspective from which their
own interests count no more than those of others. Consequentialists and Kantians
applaud this impartiality. Contemporary normative ethicists who reject the rigours of
both consequentialism and Kantianism will regard this, not as moral insight, but as
moral error; or, at best, as a purely optional and somewhat perverse lifestyle choice.60
One can be too benevolent. If it leads people too far in this direction, and is too impartial,
then mystical experience yields, not moral improvement, but moral corruption.
If we arrange contemporary normative ethical theories along a ‘demandingness
spectrum’, then traditional Christian ethics lies at one extreme alongside (extreme)
utilitarianism.61 Both present a moral ideal of extreme self-sacrifice and impartial
60
See, e.g., Wolf, ‘Moral Saints’.
61
This synergy between Christian ethics and utilitarianism is nicely illustrated by a pair of papers
delivered at a conference in Oxford in May 2011 on Christian responses to the philosophy of Peter Singer
 THE CASE AGAINST ATHEISM
benevolence. This austere element has been a key part of Christian morality from the
beginning. (‘Be ye therefore perfect.’)
Within Christianity itself, the religious communities that nurture mystics have
always been at the more demanding end of the moral spectrum, often insisting that
only the monastic life could be morally acceptable.62 And renowned mystics often
stand out as exceptionally impartial even within those otherworldly enclaves of a
demanding faith. Nor is this impartiality seen as supererogatory. Both mystics and
non-mystics often believe that mystics alone see morality clearly. Mystics are the true
moral norm.63
There is a clear connection here between meta-ethics and normative ethics. If
morality is a human artefact, an evolutionary adaptation, or something to which we
have reliable everyday epistemic access, then it is reasonable to expect its demands to
track what statistically normal individuals find plausible. But moral supernaturalism
justifies no such expectation. If morality’s source is beyond our world, then why
shouldn’t its demands be beyond both our expectations and our capacities?
Finally, the moral test rests on assumptions about human moral limitations. Why
is the moral insight of CMP both rare and indispensible? We can imagine creatures
whose theoretical reason provided perfect knowledge of supernatural moral facts,
and whose motivations then sufficed to produce morally perfect behaviour. (We
might call them ‘angels’.) The question of why we are not such creatures is, of course,
an aspect of the problem of evil discussed in chapter 8. But that we are not such
creatures is obvious. A recurring theme in the Christian tradition is the extent to
which we fall short of the angelic ideal. Both our reason and our will are clouded by
self-interest. We are incapable either of reasoning our way to the truth of morality, or
of responding appropriately to that radical truth if we encounter it only in abstract
propositional form. Without the gift of moral vision, human beings cannot see moral
truth or live fully moral lives. It is not enough to be told to love the whole world as
yourself. Given our cognitive and other limitations, we need to experience that kind
of love directly before we can display it in our lives. Only a direct encounter with a
divine person (or some other supernatural reality) can do this. On this picture, moral
improvement is evidence of supernatural encounter because such improvement
would otherwise be inexplicable.
Given all these background assumptions—ranging across meta-ethics, normative
ethics, and human psychology—the moral test of genuine mystical experiences is
perfectly reasonable. We should expect genuine mystics to display a truly extraor-
dinary level of benevolence, compassion, and impartiality, because they see the world
by Toby Ord and Eric Gregory. (Later published as Ord, ‘Global Poverty and the Demands of Morality’;
and Gregory, ‘Remembering the Poor: Duties, Dilemmas, and Vocation’.)
62
See, e.g., Chadwick, ‘Introduction’, p. 9.
63
See, e.g., Harmless, The Mystics, p. 266; Rahner, ‘Teresa of Avila’, pp. 362–3.
MYSTICISM 
in a morally transformed way. And, given those assumptions, that transformation

does represent improved moral knowledge.
Indeed, mystical experience may be essential to all human moral knowledge.
Mysticism could be necessary, not only to enable us to live up to the moral ideal,
but for the emergence of that ideal in the first place.64 This is not just unfounded
speculation. Historical evidence suggests that the moral ideal of genuine altruism (as
opposed to kin-based or reciprocal ‘altruism’) first arose alongside the great world
religions within which contemplative mystical traditions emerged.65 Perhaps this was
no coincidence, and some of our most basic moral knowledge has its roots in mystical
experience.
I conclude that, on its own terms, the moral test of genuine Christian mystical
experience is legitimate. And the background assumptions that render the test
plausible are either familiar from earlier chapters (especially chapters 1 and 2), or
they are defended in part III. AP mutually supports both a supernatural meta-ethic
(whether moral supernaturalism itself or a non-naturalism with strong supernatural
connections), and a very demanding and austere moral theory. AP itself has no
specific account of human nature. It is not committed to the specific negative picture
of human frailty that lies behind the Christian justification of the moral test.
However, AP does endorse the general idea that morality transcends human social
conventions; and therefore that both correct moral insight and appropriate moral
behaviour are rare and difficult.
Given our commitments to a supernatural meta-ethic, to a very demanding
normative ethic, and to a non-self-aggrandizing account of human beings, we should
not be surprised to find that some essential moral knowledge requires direct contact
with a supernatural realm, that such knowledge triggers a significant moral trans-
formation, and that it may not be available to all human beings. In our project, the
moral test makes sense.
The moral test is a good test. But does CMP pass that test? Do successful
practitioners of CMP display an otherwise inexplicable (or at least very surprising)
moral transformation? Unfortunately, this is a very difficult question to answer.
Clearly, defenders of CMP (and especially practitioners themselves) believe that
CMP passes the moral test. And reports of successful practitioners certainly provide
strong anecdotal evidence. For Alston’s first-personal purposes, this evidence may be
sufficient. For the outside observer, however, it is less compelling.
One pressing problem is religious diversity. To pass the moral test, does each
individual mystical doxastic practice need only moral improvement relative to the
general population, or must it also demonstrate its moral superiority to other
contemplative traditions? Suppose, for instance, that Christianity and Buddhism
both display similar moral improvement. Do both pass the test, or do they both fail?
64
See, e.g., Wainwright, Religion and Morality, pp. 232–8.
65
See, e.g., Armstrong, The Great Transformation.
 THE CASE AGAINST ATHEISM
Another problem is the lack of independent evidence of moral transformation.

While there are sociological studies of connections between (e.g.), religious affiliation
and moral behaviour, the specific kind of moral transformation that we have
identified as central to the moral test is, by its nature, much less amenable to
systematic empirical study.66
This lack of certainty may seem to support AP against BT. While Christianity
predicts that CMP will pass the moral test, and BT predicts that some BT mystical
practice will do so, AP can remain agnostic. A Christian God will provide human
beings with some epistemic access to moral facts. In an AP universe, things are much
less certain. Even if earlier chapters lead us to conclude that the universe is meant to
be understood, it is not designed to be understood by human beings. In an AP
universe, any human understanding of cosmic purpose or objective value is an
accident. Such understanding is not inexplicable (as it would be under global
naturalism). But nor it is expected. Unlike Christians (who must show that CMP
passes the test) or atheists (who must hope that it fails), AP seems untroubled either
way.
Unfortunately, it is very hard for AP to remain agnostic regarding the moral test.
Indeed, AP may need mysticism more than Christianity does. This is because AP is
much more sceptical about ordinary human moral beliefs, and therefore much more
in need of some new source of moral knowledge. And, as we’ll see several times in
later chapters, it is much easier to develop a plausible human morality founded on
AP’s cosmic values if we can rely on a close connection between morality and
mysticism.
We thus have compelling reasons to hope that some mystical doxastic practice
does pass the moral test. Fortunately, because it is not wedded to any one mystical
tradition, AP is much more flexible than specific religions such as Christianity. In
particular, as we shall see in chapter 9, AP is not troubled by religious diversity.
Suppose, as seems most likely, that many different religions are all equally morally
transformative. AP can accommodate this diversity more easily than any of its rivals.
While religious diversity is fatal for specific BT religions, it does not prevent AP from
borrowing the general moral lessons of CMP.
What about the lack of independent evidence? In reply to this objection, we can
draw several lessons from our exploration of Alston’s defence of CMP. Independent
evidence is not essential to the credibility of a doxastic practice—especially if
‘independent’ implies some evidential standard that would convince those who
66
On correlations between religious affiliation and moral behaviour, see Haidt, The Righteous Mind,
pp. 295–300; and Putnam, American Grace. A further difficulty for those who seek to establish a moral
epistemic role for mysticism is that some research suggests that the impact on moral behaviour comes, not
from religious belief (much less from mystical experience), but rather from religious attendance. For
instance, Putnam argues that unbelievers who attend church with a believing spouse show the same
moral improvements as believing churchgoers—while believers who are not regular churchgoers display no
such improvement.
MYSTICISM 
embrace complacent morality, or reject non-natural meta-ethics, or expect all evi-

dence to be empirical (and therefore scientifically verifiable). The moral test is not a
purely natural test. Perhaps only those who themselves believe in supernatural (or at
least non-natural) moral facts can correctly perceive the moral transformation
produced by CMP. Those who expect moral facts to be natural, familiar, and
unthreatening will see neither the need for the moral test nor the evidence that
supports it. But this does not undermine either the test itself or the doxastic practices
it confirms.
To apply the moral test, we must examine the moral reactions of those who already
believe in (or are willing to countenance) an austere and demanding non-natural
morality. Do they see any moral transformation arising from mysticism—either in
the lives of individual practitioners, or within mystical communities, or across
humanity as a whole? (How else can the non-naturalist explain the emergence and
persistence of moral knowledge?) The application of the moral test thus brings our
third-personal sceptical outsider into line with Alston’s first-personal perspective.
After all, the vast majority of actual austere non-naturalists are also religious insiders.
In so far as we have any reliable evidence, it is therefore positive. Among those who
are best placed to judge, a majority do think that mystical doxastic practices pass the
moral test. Moral transformation is, almost invariably, what is witnessed by those
who know what they are looking for.
AP can reasonably conclude that the moral test does establish the epistemic
credentials of CMP. (Setting aside religious diversity, to which we return in
chapter 9.) At the very least, even if they are not convinced, moral philosophers
must take this possibility very seriously. (And therefore much more seriously than
any of us currently do.) While I am not suggesting that secular moral philosophers
should rush to immerse themselves in mysticism, nor can they (continue to) dismiss
the suggestion as absurd. There is no substitute for a careful, open-minded study of
reports of successful mystical DPs, and their specific moral and epistemic benefits.
5.6 What Can AP Borrow from CMP?

Even within a BT religion such as Christianity, there is much that AP can borrow.
Many mystical reports concern facts about cosmic purpose, God, metaphysical
reality, or even morality that are consistent with AP. AP can agree that God loves
God’s creation, that the cosmos is good, and so on. Indeed, one common criticism of
mysticism is that its morality is too austere, impersonal, and alienating. In so far as
mystics push human morality in an austere impersonal direction, their pronounce-
ments are especially congenial to AP.
The only elements of CMP that AP cannot borrow occur when God’s human-
centred purpose, or God’s love for individual human beings, are essential either to the
content of a given mystical report, or to the case for the reliability of CMP. Human-
centred content is quite common. Christian mystics report God’s impartial love for
 THE CASE AGAINST ATHEISM
human beings. AP must reject these human-centred elements, and offer an alterna-
tive explanation. CMP misperceives the cosmic purpose, its source, and the realm of
objective value. Mystics correctly believe they are in contact with a supernatural
realm, but the natural human propensity to inflate our own interests and perspectives
leads them astray. Christian mystics mistake God’s love of creation for a direct
personal concern for them.
What justifies AP’s diagnosis of misperception? It would be nice if empirical
scrutiny of mystical reports revealed some phenomenal difference, where abstract
intimations of a cosmic purpose appear more genuine than the mystic’s personal
relation to God. Perennialists often defend just such a differentiation. But it is not
supported by any serious scholarly examination of the vast mystical literature, and
even a casual acquaintance with the literature (both mystical and secondary) suggests
it to be very unlikely. The AP-friendly and AP-contradictory elements of Christian
mysticism are so interwoven that nothing internal to the tradition can separate them.
We must look elsewhere.
AP’s diagnosis of misperception needs some independent reason why BT mystical
claims must be false. If we are convinced that the cosmic purpose is not anthropo-
centric, and that God does not love individual human beings, then any contrary
report must be mistaken. The motivation to seek an alternative explanation thus
comes in part II.
Naturalists offer a deflationary explanation. Mystical experience is entirely mis-
taken, and its practitioners are not in touch with anything beyond their own
disordered mental states. If its only opponent were BT, AP could borrow this
deflationary explanation. However, mysticism would then offer AP no support
against atheism, and AP would lose a potentially vital source of moral insight.
AP prefers a less revisionist explanation. In the absence of independent arguments
against a non-human-centred cosmic purpose, God, or supernatural realm, the
atheist’s demand that we should expect a fully deflationary explanation of mystical
experience is under-motivated. Unlike her claim to have experienced the personal
love of God, the mystic’s claim to experience a morally transformative cosmic
purpose can be accepted at face value. AP’s less revisionist explanation thus respects
the phenomena better than any more radical atheist revision.
AP does not simply assert that some elements of mystical reports are accurate
while others are erroneous. Instead, it offers a genuine explanation, citing the human
tendency to self-aggrandizing interpretations of experience or evidence. This explan-
ation is not ad hoc, because there is ample independent evidence for this tendency. If
there is a non-human centred cosmic purpose, then the reports of BT mystics are just
what we might expect from comparatively acute (but still all too human) observers.
As with human knowledge of moral facts, BT offers a clear explanation of why our
mystical doxastic practices reliably yield knowledge of God. This is exactly how we
would expect a benevolent God (who presumably wants to be known) to have
designed rational beings. BT argues that, without a benevolent deity, there is no
MYSTICISM 
reason to trust any mystical experience. Therefore, only BT can reasonably believe
that any mystical DP is reliable.
AP agrees that the global naturalist must deny the reliability of mysticism. In a
naturalist universe, mysticism can only be a delusion. But as we saw in chapter 4, AP
argues that, if there is a cosmic purpose, then the cosmos may well be designed or
arranged so that its purpose can be understood. Perhaps evolution was designed to
produce creatures who truly perceive and understand the cosmic purpose. We are
not those creatures. But the faltering efforts of human mystics may be inklings of true
cosmic insight. Human rudimentary mystical awareness is then a happy accident,
rather than an inexplicable anomaly. (We return to this possibility in chapter 7.)
Reliable mystical DPs are more likely under BT than under AP. But, conversely,
BT offers a much worse explanation of why (and when) human mystical practices are
unreliable. Even if it can explain why mystical experience is not universal, BT cannot
explain the prevalence of either evil or religious diversity. I argue in chapters 8 and 9
that a world of competing mystical DPs, most delivering very false information about
God and cosmic purpose, is much harder to reconcile with BT than with AP.
If CMP were the only prima facie reliable mystical DP, then it would provide BT
with strong arguments against both atheism and AP. AP aims to borrow the former
and rebut the latter. To undermine mysticism’s support for BT over AP, we need
something that BT cannot explain. We will find it in part II. BT cannot explain
religious diversity. Religious diversity thus also threatens CMP itself. However, I am
confident that AP can salvage enough from CMP to construct its own argument
against atheism. Mysticism thus supports AP against atheism. A theme of this
chapter has been the tight connection between mysticism and morality. In so far as
it supports AP at all, mysticism supports Normative AP—reinforcing the connection
between cosmic purpose and objective value. However, AP can only borrow an
argument from mysticism if it can provide a plausible human morality based on
non-human-centred cosmic values. As in chapters 3 and 4, the success of this chapter
thus rests on the success of part III.
5.7 How Does CMP Shape AP Morality?

As well as reinforcing the connection between cosmic purpose and objective value,
CMP also offers new insights into both. Even after we filter out the human-centred
elements of CMP, we may be left with a residue of new moral information. If mystical
experience reveals God’s love for creation, and leads to a moral transformation away
from egoism, then this will influence our interpretation of cosmic purpose.
Our focus on actual mystical doxastic practices teaches other lessons. The moral
knowledge that emerges from CMP—and other contemplative traditions—is an
uncertain glimpse of a supernatural reality that is granted only rarely; rather than
something all human beings have access to as a matter of course. This change of
emphasis may lead us to reconsider older elitist themes in Western moral thought
 THE CASE AGAINST ATHEISM
that lie buried under the enlightenment insistence that moral knowledge is easily
available to all. (The truly unsettling thought for secular moral philosophers is that
some other professional elite might now provide the true ‘moral experts’!)
Most important of all, our exploration of mysticism, like our earlier foray into
cosmology, has uncovered a surprising area for further ethical research. Moral
philosophers seeking insights into the realm of cosmic value should take seriously
the often overlooked world of contemplative mystical traditions.
In the next chapter, we use this new perspective to put a very old, and overly
familiar argument, in a new light.
6
Ontological Arguments
That Anselm’s Proof of the Existence of God has repeatedly been called the
‘Ontological’ Proof of God, that commentators have refused to see that it is in a
different book altogether from the well-known teaching of Descartes and Leibniz,
that anyone could seriously think that it is even remotely affected by what Kant
put forward against these doctrines—all that is so much nonsense on which no
more words ought to be wasted.
[Karl Barth, Anselm, p. 171]
I think that ontological arguments for the existence of God are very much mere
philosophers’ arguments and do not codify any of the reasons that ordinary
people have for believing that there is a God.
[Richard Swinburne, The Existence of God, p. 9]
There is no argument in the history of philosophy that means more but says less
than the argument of Proslogion 3.
[Gregory Schufreider, Confessions of a Rational Mystic, p. 148]
Either worship is self-contradictory (rather than merely mistaken), or its object

exists necessarily.
[Charles Hartshorne, Wisdom as Moderation, p. 79]
In 1077, Anselm, prior of the Benedictine monastery of Bec, experienced a revela-

tion.1 As prior, Anselm was responsible for the spiritual development of novice
monks. Anselm took his pastoral duties very seriously. While preparing his charges
for a life of discipline, contemplation, and worship, Anselm often addressed the
nature of God, and the possibility of human understanding of God. At the urging of
his brethren, Anselm recorded his ideas in a philosophical treatise. This Monologion
presents several distinct arguments for the existence of God. Anselm remained
dissatisfied. Monologion was too disjointed, unworthy of its divine subject. Anselm
sought a single argument to tie everything together—the elusive unum argumentum.
He became obsessed with this search, going without food, drink, and sleep. ‘Then
1
For Anselm’s biography, see Southern, Saint Anselm; Visser and Williams, Anselm. The main original
source is Eadmer, The Life of St Anselm.
 THE CASE AGAINST ATHEISM
suddenly one night, during the night vigil, the grace of God illuminated his heart and
the matter became clear to his understanding, and immense joy and jubilation filled
his whole being.’2
To record this new insight, Anselm wrote, not another philosophical treatise, but a
single continuous prayer. This is Proslogion—one of the most enigmatic and puzzling
works in Western philosophy, and the source of the infamous ontological argument.
The ontological argument has been described as ‘the single most discussed
argument in the history of philosophy’.3 It is also one of the most divisive. As
Schufreider puts it: ‘among the strangest features of the ontological proof is the fact
that some of the subtler minds in our history have found it thoroughly convincing,
while other equally profound thinkers have found it altogether unconvincing and
vehemently so’.4 Its supporters include Bonaventure, Descartes, and Leibniz; its
most illustrious opponents were Aquinas, Hobbes, Hume, and Kant.5 In modern
times, the ontological argument has been taken up by analytic philosophers (Godel,
Plantinga, Leftow), continental philosophers (Rorty), process philosophers (Hart-
shorne), systematic theologians (Barth), mystical theologians, deconstructionists
(Mark Taylor), and others.6 In the process, the argument has often drifted far from
Anselm’s original text.7 In his long posthumous career, Anselm has been a scho-
lastic, a mystic, a sceptic, a dogmatist, and, perhaps most oddly of all, a practitioner
of late twentieth-century modal logic.
One brief chapter cannot do justice to this vast array of competing literatures.
Instead, I sketch two ontological arguments that, given our prior commitments and
the conclusions of earlier chapters, lend some support to BT; and that AP can
plausibly borrow. One is based directly on Anselm’s original text, and combines
Proslogion with chapter 5’s defence of Christian mysticism. The other argument is
the more familiar modal ontological argument. I shall argue that, if we are committed
to objective values and we take mystical doxastic practices seriously, then ontological
arguments may have considerable force. As with all our other arguments in part I, the
2 3
Eadmer, The Life of St Anselm, pp. 29–30. Rogers, Perfect Being Theology, p. 1.
4
Schufreider, Confessions of a Rational Mystic, p. 248.
5
On the history of the ontological argument, see Logan, Reading Anselm’s Proslogion, pp. 136–72;
Dombrowski, Rethinking the Ontological Argument, pp. 7–31.
6
For the contrasting range of contemporary readings of Anselm, see: Logan, Reading Anselm’s
Proslogion, pp. 136–72; Dombrowski, Rethinking the Ontological Argument; Schufreider, Confessions of a
Rational Mystic; Oppy, Ontological Arguments; Hick and McGill, The Many-Faced Argument. The
arguments of Godel, Plantinga, Leftow, and Hartshorne are discussed below. For the other examples listed
in the text, see Rorty, ‘Response to Charles Hartshorne’; Taylor, nOts; Barth, Anselm.
7
Anselm’s first commentator, Gaunilo, established the pattern of misreading Anselm (Schufreider,
Confessions of a Rational Mystic, p. 117). Medieval discussion was dominated by Aquinas, who responded
not to Anselm but to Bonaventure; early modern discussion took its lead from Descartes, whose letters
suggest he never read Anselm; and many philosophers follow Kant, who aims his refutation at an argument
he seems to get from Descartes via Wolf, and once described Anselm as ‘a Parisian scholastic’ (Logan,
Reading Anselm’s Proslogion, pp. 140, 152, 157).
ONTOLOGICAL ARGUMENTS 
resulting argument delivers Normative AP. The connection between cosmic purpose
and objective value is crucial.
The ontological argument, in its most common form, is extraordinarily ambitious.
It attempts to prove the existence of God entirely a priori. There are no empirical
premises, partisan commitments, or leaps of faith; and the conclusion is one that any
reasonable atheist must accept.
This is too ambitious. Oppy reasonably objects that ‘there is no a priori argument
from uncontroversial premises that will persuade any reasonable nontheist that God
exists’.8 Atheism may be false, but it is not incoherent. Anyone committed to
religious ambiguity, or suspicious of caprice and hubris in philosophical argument,
will agree. Our commitment to metaphysical pluralism cautions against expecting to
find knock-down a priori proofs.
Oppy argues that a ‘dialectically successful’ ontological argument must be this
ambitious, and therefore concludes that the argument fails. This is too swift. We can
construct less ambitious ontological arguments, by introducing modesty along sev-
eral dimensions. Some arguments only defend the conditional claim that, if God
exists, then God exists necessarily. Others openly rely on controversial metaphysical,
logical, modal, or moral premises, or they appeal to mystical doxastic practices.
Others conclude, not that God does (or must) exist, but that God probably exists.
(Perhaps the ontological argument is merely one chain in a cumulative case for God.)
Our commitments might support these modest ontological arguments. In particular,
our earlier discussion of mysticism, and our initial commitment to objective values,
can both bolster ontological arguments at crucial points.
This chapter is continuous with chapter 5, in so far as it both presupposes and
extends our earlier discussion of the epistemic and moral credentials of mystical
experience. We also continue our first-personal perspective—starting from Anselm’s
contemplative advice to his novice monks. Our ontological argument is thus much
closer to our argument from mysticism than to our earlier cosmological and teleo-
logical arguments. However, this chapter is also distinctive, especially in its focus on a
single historical text. In chapter 5, I argued that contemporary analytic moral
philosophers should engage with the recent scholarly literature on mysticism. In
this chapter, I suggest a deeper engagement with Anselm’s original text and the
(largely non-philosophical) secondary literature devoted to it. My own discussion of
both Anselm and his interpreters barely scratches the surface, but I hope it will
stimulate others to go further.
Section 6.1 introduces Anselm’s Proslogion in its original philosophical and reli-
gious context, sets out the ontological argument, explores its intimate connections
with Anselm’s rational mysticism, and shows how our project supports Anselm’s.
Section 6.2 explores the modern modal ontological argument (MOA), the most
8
Oppy, Ontological Arguments, p. 282.
 THE CASE AGAINST ATHEISM
popular contemporary analytic reading of Anselm. Although I engage with some

recent analytic literature, my goal is not to cover familiar ground, but rather to
illustrate how our earlier conclusions regarding mysticism and objective value might
bolster the most plausible MOA. Finally, section 6.3 shows how AP can borrow the
best ontological arguments, so long as it can ground non-human-centred objective
values. (As we have seen throughout part I, the argument of this chapter thus requires
the success of part III.)
6.1 Anselm
Anselm’s intellectual world is not ours. His life, and that of his audience, was a daily
round of prayer, worship, tradition, and scripture. Without some appreciation of this
historical distance, it is too easy to misread Proslogion, and underestimate Anselm’s
resources.
6.1.1 Anselm’s goals in Proslogion
Proslogion is not a long work. But analytic philosophers read only one very brief
excerpt: Proslogion 2 and 3. This omits not only Anselm’s subsequent intricate
discussion of the attributes of God, but also his opening prayer of invocation
(Proslogion 1). This omission is a mistake. Anselm deliberately wrote Proslogion as
a prayer addressed to God, rather than a philosophical treatise. (Proslogion differs
here from both Monologion and Anselm’s later works.) The original text is one
continuous prayer, with no separation between religious and philosophical themes.
(The familiar chapter divisions are a later addition.)
Proslogion 1 sheds light on Anselm’s notion of God. It also addresses the relation-
ship between human and divine, and on the limits of our knowledge. Anselm
explicitly asks God for illumination: ‘show yourself ’. This is no mere form of
words, but an acknowledgement that the unaided human mind cannot possibly
fathom the mysteries of the divine nature.
We naturally look to Proslogion for an argument for the existence of God. And we
expect that argument to persuade the atheist, or at least the reasonable agnostic.
There are several reasons why we should not expect such an argument from Anselm.
First, it is far from clear that, in Proslogion, Anselm intends to offer any argument for
the existence of God. This is certainly not his primary purpose. Anselm’s audience are
young men entering a life of religious contemplation. His task is to prepare them for
that life. Proslogion is, first and foremost, an aid in that task. Anselm’s prayer shows
the aspiring monk how to approach the contemplation of the divine reality—how to
think about God. There is an epistemic element here. Anselm does aim to remove the
reader’s uncertainties. But these are the doubts of the believer: the perennial doubts of
the contemplative or mystic. How can I be sure that I am seeing, contemplating,
adoring, or thinking about God? How can I be confident that my visions are not vain
illusions, or temptations sent by the devil? How can I be certain that the object of my
ONTOLOGICAL ARGUMENTS 
thoughts is real—that it exists outside my mind as well as within it? (As we saw in
chapter 5, such questions are a constant preoccupation for Christian mystics.) The
answers to these questions may also provide the raw materials for a reply to the
doubts of the unbeliever, infidel, or confirmed atheist. But for Anselm, these non-
Christian doubts are largely irrelevant.
In modern philosophical terms, then, Anselm’s Proslogion is closer to Alston’s
defence of Christian mystical practice, or Plantinga’s argument that BT belief can be
properly basic, than to traditional arguments for the existence of God. Anselm
addresses the internal rationality of his religious community, not the conversion of
third parties.9
Does Anselm intend to persuade the unbeliever? This question has excited much
debate. Some argue that, for Anselm, ‘rational non-theist’ is a contradiction in terms.
There is no question of reasoning with an unbeliever.10 This interpretation is sup-
ported by Anselm’s much-quoted comparison between people who do not begin with
faith, or who contest the authority of the Church Fathers, and ‘bats and owls, who see
the sky only at night [and yet] dispute about the midday rays of the sun with eagles’.11
But others argue that Anselm does attempt to persuade an imaginary unbeliever,12 or
perhaps that he addresses different audiences in Proslogion and Responsio.13
Second, if Anselm were to defend his faith, he would not offer Proslogion on its
own. (And he would certainly not regard the ‘ontological argument’ of Proslogion 2
and 3 as sufficient.) Anselm would not defend BT simpliciter, but rather his broader
Christian faith. And that defence would involve scripture, tradition, religious experi-
ence, moral insight, and so on, as much as any metaphysical argument. Even among
the latter, the ontological argument would not stand alone. Anselm could also have
invoked ancestors of the familiar teleological or cosmological arguments examined in
earlier chapters. Anselm’s overall defence of God would thus resemble Swinburne’s
cumulative argument.14 As we shall see, the ontological argument may be much more
plausible in the context of other Christian commitments.
Third, although he claims to use everyday language, Anselm writes within a
specific philosophical framework inherited (via Augustine) from Boethius and the
Neoplatonists.15 This common ground, together with a shared religious tradition,
means that several of Anselm’s argumentative strategies were much less problematic
for his original audience than they are today. Anselm’s Boethian metaphysics
9
Adams, ‘Anselm on Faith and Reason’, pp. 37–8.
10
Logan, Reading Anselm’s Proslogion, p. 179; Barth, Anselm.
11
Adams, ‘Anselm on Faith and Reason’, p. 43. (The original source is Anselm’s De Inc. Verbi.)
12
See, e.g., Visser and Williams, Anselm, p. 18.
13
Logan, Reading Anselm’s Proslogion, p. 24.
14
Swinburne’s own cumulative argument does not include an ontological argument (see the epigraph to
this chapter). However, as we shall see, other philosophers, notably Hartshorne, do incorporate ontological
arguments into a cumulative case for BT.
15
On the complex relationship between Anselm and Neoplatonism, see Logan, Reading Anselm’s
Proslogion, 14; Matthews, ‘Anselm, Augustine and Platonism’.
 THE CASE AGAINST ATHEISM
supports two common commitments of mystics throughout the ages: that purely
intellectual activity touches a realm that is more real than the everyday physical
world; and that the resulting knowledge is more reliable and more valuable than
everyday empirical knowledge. Given Platonic assumptions, it is natural to expect a
priori philosophical enquiry to yield genuine and vital knowledge of reality.
Anselm does not defend this common framework in Proslogion. Proslogion alone
could not convince an unbeliever who is not already a Platonist.16 Yet few contem-
porary philosophers are Boethian Platonists. This metaphysical divide leads some to
conclude that, whatever its historical significance, Anselm’s Proslogion has no con-
temporary relevance. We want an argument to convince us, and Anselm cannot
provide it. As Oppy concludes: ‘If the argument is taken to rely on Neoplatonic
metaphysics, then it is completely lacking in probative force; there are very few non-
believers who are committed to such a metaphysics.’17
This is too swift. Despite its association with an unfashionable metaphysic, I believe
that we should not dismiss Anselm’s Proslogion out of hand. While he inevitably
expresses it within his own religious and philosophical framework, Anselm’s key insight
has an enduring fascination, one that can be illuminatingly re-expressed in new
frameworks in each succeeding age. This chapter explores two such re-expressions.
6.1.2 Anselm’s basic argument

While the details of Anselm’s argument are controversial, its basic form is, on one
level, clear enough.18 Let G abbreviate the phrase: ‘that than which no greater can be
conceived’.
1. Antecedent premise: G is understood.19
2. Conditional premise: If G is understood, then G exists.
3. Interim conclusion: Therefore, G exists.
4. Identity premise: G is God.
5. Final conclusion: Therefore, God exists.
The argument is valid and has only three premises. The identity premise, and the
distinction between G and God, are crucial.20 Anselm deliberately chooses an
16
While Anselm’s other works often offer more detailed philosophical argument, it is very unlikely that
they would suffice to convert a modern global naturalist to medieval Neoplatonism.
17
Oppy, Ontological Arguments, p. 170.
18
I borrow here from Logan, Reading Anselm’s Proslogion, especially p. 125.
19
My formulation abandons Anselm’s notorious expressions ‘exists in the understanding’ and ‘exists in
reality’. I do not think they help us to understand Anselm’s thought. I am persuaded by Visser and
Williams that Anselm chose these expressions, not to express some obscure metaphysical doctrine about
two modes of existence, but rather because he ‘just thought the phrasing sounded nice and made the
argument memorable’ (Visser and Williams, Anselm, p. 74).
20
By treating the identity premise as a separate step in Anselm’s argument, I assume that he does not
offer G as a definition of ‘God’. In this, I follow the consensus of recent commentators. Indeed, as Logan
notes, within Anselm’s Boethian metaphysics it makes no sense to offer a definition of ‘God’, as God does
not fall under a genus (Logan, Reading Anselm’s Proslogion, p. 18).
ONTOLOGICAL ARGUMENTS 
expression that is not a standard formulation of Christian prayer.21 The believer must
set aside his preconceived notions and approach God anew. (For our project,
Anselm’s gap between God and G opens up the possibility of an Anselm-inspired
argument from G to AP.)
Philosophical debate concentrates on the antecedent and conditional premises.
Both have been interpreted in many different ways. The antecedent premise might
claim that someone is thinking about G; that G exists in someone’s mind; that G is
conceivable; that G is understood; that the concept G is coherent, clear, or otherwise
acceptable; that G exists qua intentional object; or (most commonly in recent
analytic philosophy) that G is possible. The conditional premise is then interpreted
analogously: if someone is thinking about G, then G exists; if G is possible, then
G exists; etc.
The antecedent premise seems unproblematic. We are familiar with claims about
possibility or understanding. The conditional premise is more puzzling. The trick is
to bridge the gap between the concept of G and the existence of G—to go from
possibility to necessity. The link is provided by the special nature of G. G might be a
perfect being; a maximally great being; an eternal uncreated creator of all other
beings; an unsurpassable being; a being worthy of worship; or (most frequently in
analytic philosophy) a necessary being.
Since Gaunilo’s celebrated reply to Anselm, every ontological argument faces the
threat of parody.22 The parodist constructs some alternative concept (G2) and then
alleges that, if both of Anselm’s premises hold true for G, then both are also true of
G2. G2 is also coherent and possible; and G2 mirrors the special feature of G that
justifies the conditional premise.
Parodies have been put forward that ‘prove’ the existence of the following: a greatest
island, a golden mountain, a necessarily existing golden mountain, a perfectly evil
being (‘devil’), a being who knows that there is no necessary being (‘knowno’), or a
being who exists in a possible world where there is no necessary being.23
Parodies threaten the ontological argument for two reasons. First, they simply
make the argument appear ridiculous. Any argument that obliges us to accept the real
existence of Gaunilo’s perfect island or Kant’s golden mountain is obviously
unacceptable. Even more troubling is the fact that some parodies yield entities,
such as devil and knowno, that are logically inconsistent with the existence of
Anselm’s God. If these entities exist, then Anselm’s God does not. Even if he were
21
Some commentators even suggest that Anselm chooses G precisely because it was first employed by
non-Christian classical authors, notably Seneca. G thus offers a form of words that is neutral between
believers and non-believers (Logan, Reading Anselm’s Proslogion, p. 142; see also pp. 92–3).
22
While many philosophers find Gaunilo’s reply decisive, Anselm himself clearly did not. Indeed, he
ordered that Proslogion should always be published together with Gaunilo’s reply and his own Responsio.
23
The greatest island is Gaunilo’s original parody, while the golden mountain is attributed to Kant.
Devil parodies became popular in the late twentieth century, introduced by Haight and Haight, ‘An
Ontological Argument for the Devil’. The knowno appears in van Inwagen, Metaphysics, p. 107.
 THE CASE AGAINST ATHEISM
willing to countenance the ontological extravagance of golden mountains or perfect

islands, Anselm could never accept these contradictory parodies.
A successful ontological argument must be parody-proof. While either of its two
key premises may apply to other concepts, G is the only case where both apply. To
defeat Gaunilo, Anselm must show that his argument applies to God and only to
God. Two strategies present themselves. Some parodies fail at the conditional
premise. If one understands the concepts involved, then one sees that one can easily
imagine the entity not existing, because the possibility of non-existence is built into
the concept. This blocks the move from possibility to existence. Other parodies fail at
the antecedent premise. While they seem to satisfy the conditional premise, they turn
out to be incoherent. Despite appearances, the concept contains some hidden
contradiction. (As we’ll see, subsequent defences of the ontological argument all
deploy some combination of these two strategies.)
6.1.3 Anselm on thinking, understanding, and an incomprehensible God

While Anselm’s Monologion explicitly eschews any reference to scripture and trad-
ition, Proslogion makes free use of biblical allusions and quotations. The most famous
is the Psalmist’s ‘fool who says in his heart there is no God’.24 Anselm wants to help
the believer to avoid being a fool. The contrast between true believer and fool is
central to Anselm’s reply to Gaunilo, and also offers the best reply to all subsequent
parodies.
The fool presents a prima facie challenge to Anselm. The fool thinks about God.
He understands the term ‘God’. Yet he believes that God does not exist. If he admits
the existence of fools, how can Anselm argue that God’s non-existence is
inconceivable?
Anselm does not deny that there are fools. Some people think there is no God. His
solution distinguishes different ways of thinking—different cognitive activities that
connect the human mind to its intentional objects. Anselm follows the apophatic
tradition in Christian theology. There are levels of understanding (especially per-
taining to God) that are unavailable to mortal beings, at least in this life. In one very
important sense, we can never hope to know God, understand God’s nature, or grasp
God’s essence.25
At the other extreme lies the fool. The fool grasps the meaning of individual words,
and is adept at manipulating words, signs, and logical symbols to construct new and
bizarre concepts. (In short, the fool is a gifted scholastic or analytic philosopher.)
After a fashion, fools do ‘think’ about many things: from God to unicorns; from
islands whose wealth exceeds all others to necessarily existent mountains made
24
Psalm 53.
25
Some non-philosophical commentators conclude that Anselm’s view is that we can never know God
at all. (See the epigraph from Karl Barth; and also Dombrowski, Rethinking the Ontological Argument, p. 3;
Logan, Reading Anselm’s Proslogion, p. 172.)
ONTOLOGICAL ARGUMENTS 
entirely of gold; from perfectly evil beings to those who know the falsehood of
necessary truths. This is how the fool can think there is no God.
Anselm’s believer cannot aspire to inaccessible knowledge that passes all mortal
human understanding. But nor need he settle for the mere thinking of the fool.
Anselm argues that the believer can reasonably hope for something deeper. A more
robust, discerning, and reliable cognitive relation to objects is available to human
beings in this life. Let us call this understanding, to distinguish it from the fool’s
thinking. Anselm’s fool thinks about God, but without understanding God. If the fool
did understand, he would see that God cannot fail to exist.
What is the understanding the fool lacks? It is not, in the general case, a mode of
cognition that guarantees the existence of its object. Anselm is very clear on this
point. Indeed, understanding sometimes teaches us that the object cannot exist.
Anselm’s example is the chimera: a beast combining parts of a lion, a goat, and a
serpent.26 Hearing a description of this terrifying beast, an ignorant child (or a
foreigner unacquainted with the local mythology and fauna) may wonder whether
it exists. Are there any chimera around here? Should I be on my guard? This ‘fool’
thinks about the chimera without any idea whether it exists. But the educated person
knows that the chimera is a mythical beast, and as such cannot exist. (That is what it
is to be a mythical beast.) Similarly, the educated reader does not seek Mr Sherlock
Holmes in Baker Street. She understands that, as a fictional character, he cannot be
real.
For mythical and fictional entities, understanding involves realizing the kind of
being one is dealing with. One then knows what modes of existence are available. The
same applies to contingent beings. Perhaps a fool or a very young child can think of a
tree, an island, or a mountain, without any idea what kind of being it is. But to truly
understand these objects is to realize they are contingent physical objects. Any actual
tree might not have existed. In a different history of the physical universe, this
particular tree never appears. When I imagine that alternative history, I think of
this tree as not existing. As a physical being, any actual tree is also surpassable. Its
properties are purely physical, and therefore measurable on some finite scale. It is
always possible, and in most cases very easy, to imagine a better tree: one further up
the scale that measures desirable tree features. By definition, there is no perfect tree.
The fool, who cleverly gerrymanders his concepts by adding properties willy-nilly,
thinks of many things that true understanding reveals to be nonsensical. This is
Anselm’s response to Gaunilo, and the response he would give to all subsequent
defenders of the fool. There cannot be an unsurpassable island: we can always
imagine a greater one. A necessary golden mountain is a contradiction in terms:
for any mountain, we can always imagine a possible history where that mountain
never comes to exist.
26
Logan, Reading Anselm’s Proslogion, pp. 94, 193.
 THE CASE AGAINST ATHEISM
To understand a contingent thing, then, is to know that, in this case, the condi-
tional premise is false. Non-existence is always possible, and always conceivable. On
the other hand, to add ‘necessity’ to the concept of a contingent thing is absurd.
‘Necessary mountain’ may satisfy the conditional premise. But this imaginary con-
cept cannot be understood. (What could ‘necessary contingent thing’ mean?) The
fool is now talking nonsense. This is also how Anselm would deal with the perfect
devil and knowno. Despite appearances, these concepts cannot be understood. Each
hides a contradiction.
The threat is that, if we raise the threshold for the antecedent premise, Anselm’s
own preferred candidate (G) may also fail to be understood. The challenge for any
ontological argument is to distinguish its claims to understanding from parodies.
I explore the most popular analytic route in section 6.2. First, however, I present a
more phenomenological or mystical reading, based on Anselm’s own contemplative
context.
6.1.4 Anselm as rational mystic

As Anselm was a monk writing for other monks, it is natural to interpret Proslogion
as an exercise in mystical theology. Religious interpretations of Proslogion are often
seen as rivals to philosophical or logical readings. However, we saw in chapter 5 that
Christian mysticism and contemporary philosophy can be partners rather than
rivals. In this section, I offer a philosophical reinterpretation of Anselm’s rational
mysticism in Proslogion.27
Anselm’s central goal is to help his novices to think clearly about God, to
understand God. He leads them on a contemplative journey in search of that than
which no greater can be conceived (G). Once they complete that journey, and explore
the true nature of G, they will realize that it is none other than God. They will then
understand God in a way that they did not before.
Anselm’s monk can begin his contemplation where he himself is most comfort-
able. There is no need to try to picture God straight off. Instead, he might begin by
examining everyday objects, such as a chair or a tree, and asking one simple question:
can I conceive of something greater than this? He now attempts to picture this new
greater object in his mind; and then repeats his question: can I now conceive of
something greater than that? By repeating this process, the monk follows his own
contemplative path, gradually approaching the limits of conceivability. He is groping
towards an object than which he can conceive none greater.
Anselm’s formula G leads the monk on a largely negative path of discovery. He
concentrates on the limitations of each object. The chair is small, transient, lifeless.
He then imagines an object that lacks this or that limitation, only to focus on its
27
The section draws on Barth, Anselm; Dombrowski, Rethinking the Ontological Argument; Taliaferro,
‘Review of Oppy’; and especially Schufreider, Confessions of a Rational Mystic.
ONTOLOGICAL ARGUMENTS 
limitations. The mountain is larger and more enduring that the chair, but it is still
small relative to a galaxy. And even the galaxy is still lifeless. The tree is not lifeless,
but nor is it rational. A human being is rational, but still has many limitations of
power, insight, and goodness. And human life is especially contingent and precar-
ious. Any human being’s existence is the result of a highly improbable series of
events. (There is nothing easier than imagining a world where I do not exist.28) Some
things exist longer than others, but anything that comes into (or goes out of )
existence is limited. Some things are partially responsible for their own existence,
nature, or circumstances, or even for the existence and circumstances of other things.
But anything that is not wholly responsible for every feature of both itself and
everything else is still limited. And so on.
At each stage, the monk adds to an expanding list of things that G is not: G is not a
tree, a mountain, a galaxy, a human being; G is not transient; G cannot have any
physical size or temporal extent, because any size or duration can easily be exceeded
in our imagination; G is not inanimate, or non-rational; G is not dependent on
anything else; G’s power and goodness have no limits, etc.
Each monk follows his own contemplative path. But Anselm offers many sign-
posts. The prayer that opens Proslogion addresses God as ‘you who cannot be thought
not to be’. Anselm’s then introduces the fool who doubts the existence of God. This
will prompt Anselm’s monk to ask, of each object he brings before his mind: can
I imagine this thing not existing? Is there a possible history where it never appears?
Behind his ability to imagine a thing’s non-existence, Anselm’s monk sees a limita-
tion of that thing. So he searches for an object whose possible non-existence cannot
even be imagined.
The central question for my mystical interpretation is whether Anselm’s negative
procedure generates a positive result. Anselm’s monk doesn’t just want to stop
thinking about things that are not God. He wants to start thinking about God.
Anselm’s procedure will help him to know when he falls short of his goal, and
perhaps also to know when he has reached it. But will it help him to reach it? How
does a mere list of limitations that God lacks lead one to understand God? Wasn’t the
fool mocked for just this sort of list-making enterprise?
There are several worrying possibilities for Anselm here. One is that his procedure
will generate different positive results. What if different monks reach incompatible
understandings of G? (One experiences Anselm’s God, another a maximally evil
being, another an impersonal creative value, and so on.) This would obviously spell
disaster for Anselm.
28
Today, a philosophically educated monk might dwell on Parfit’s non-identity problem, and the
enormous antecedent improbability of his own existence. (Parfit, Reasons and Persons, ch. 16.) It is crucial
here to recognize, as Anselm does, the gap between imagining my non-existence and doubting my existence
(Schufreider, Confessions of a Rational Mystic, p. 151). Our concern is the former, not the latter. Anselm’s
project in Proslogion is not Descartes’s project in Meditations.
 THE CASE AGAINST ATHEISM
Anselm’s procedure works by identifying limitations: properties that it is better for

a thing to lack than to have. The mountain is greater than the tree, a human being is
greater than either, an angel is greater than all three. An eternal being is greater than a
transient one; the spiritual is greater than the physical; necessary existence is greater
than contingency; creators are greater than creatures; and so. These judgements of
greatness appear subjective. And subjectivity breeds parody. If each Anselm-inspired
monk works with his own idiosyncratic notion of ‘greatness’, then we face a prolif-
eration of (incompatible) beings than which no greater can be conceived.
If G could lead competent practitioners to incompatible understandings, then
Anselm’s theology would collapse. Anselm must dismiss all these dissenting monks
as fools. They think they have followed his procedure, but they have not. They have
not understood G. Anselm is not inviting us to explore our own subjective prefer-
ences or whims, but to explore an objective world of mind-independent values.
Anselm’s commitment to objective values is crucial. Without some independent
standard of comparative greatness, Anselm’s G cannot escape parody. As with the
cosmological argument and fine-tuning arguments (see chapters 3 and 4), I have
often found that it is this commitment to objective values—rather than any of its
more baroque metaphysical or logical features—that today’s philosophy students
(especially those inclined towards naturalism and atheism) find most alien in An-
selm’s Proslogion.
Many critics simply dismiss Anselm’s evaluative objectivism along with his
extravagant Boethian metaphysics.29 But this is surely too swift. Anselm himself
does use Neoplatonic language that is foreign to us. But his comparative evaluative
claims can be rephrased in more familiar ways, and the result need not embarrass
contemporary moral realists, especially consequentialist non-naturalists.
This is where our broader commitments and earlier arguments most clearly
support Anselm’s general strategy. However, because it defends different objective
values, AP will reject many of the particular conclusions about G that Anselm draws
in the later chapters of Proslogion. (We return to this question in section 6.3.)
Suppose, for the sake of argument, that we grant Anselm objective values to which
human beings have epistemic access. (This might be independent access, or moral
insight might come from mystical practice itself.) This is sufficient to dissolve the
possibility of multiple positive results. If monks following Anselm’s procedure can
29
To illustrate the close connection between suspicion of Anselm’s Boethian metaphysics and rejection
of his moral objectivism, consider this objection from Oppy: ‘Conceiving is always conceiving by someone.
What is conceivable is always what is conceivable for someone. It is a big step to suppose that there is a—
greatly idealized—limit in which the conceivings—and judgments about conceivability—of all reasonable
persons converge’ (Oppy, Ontological Arguments, p. 167). This is strikingly similar to the moral naturalist
reliance on convergence that we encountered in chapter 2. If one believes that all human claims about the
conceivability of a ‘greater being’ are merely subjective, then this is a powerful objection. However, if one
agrees with Anselm that humans can access a realm of objective moral facts, then Oppy’s complaint is
much less compelling. If moral knowledge is objective and accessible, then all morally informed agents will
converge on the moral truth.
ONTOLOGICAL ARGUMENTS 
track independent standards of greatness, then they will not reach radically incom-
patible end points. But this raises a second worrying possibility. What if Anselm’s
procedure fails to produce any positive result at all? We can easily imagine a very
conscientious monk following Anselm’s procedure to the letter, and coming up with
nothing, with a vision of pure emptiness. Indeed, this is often how Buddhist medi-
tation does proceed. My description of a monk reflecting on the impermanence of all
things could describe Buddhist meditative techniques as easily as Christian ones.
Anselm has several possible replies. First, he never claims that his Proslogion
procedure guarantees a positive experience or understanding of God. Anselm
would reject the blasphemous suggestion that any human endeavour could ensure
an experience of God. The contemplative monk is not a magician who can compel
God to appear. By clearing his mind of limited objects of thought, Anselm’s monk
creates a space where God can appear. G leads to a place where God might be found.
But whether God does appear is entirely beyond the monk’s control. Any under-
standing or experience of God is a free gift of divine grace. The monk can only wait
and pray for understanding to be given. But, if he faithfully follows Anselm’s
procedure, he will at least recognize this true understanding if it does come.
An individual failure to get a positive result is thus not a threat to Anselm. (By
contrast, even a single incompatible positive result would be fatal.) Indeed, even if no
one achieved a positive result by following Anselm’s procedure, this would not prove
either that God did not exist, or that the procedure was flawed. On the other hand, if
everyone who followed Anselm’s procedure came up empty, we would be left without
any Anselmian argument for the existence of God. Whether or not Anselm himself
sought such an argument, we do. Therefore, we must argue that, while rare and
difficult, mystical insight does occur. The intellectual exploration of G sometimes
leads to an understanding of God.
We must not set our sights too high. We may fail to recognize (or not be open to) a
positive result because we have unrealistic expectations of what it will involve. A full
understanding, or a vision of the divine, is not available in this life. In later chapters of
Proslogion, Anselm emphasizes that God is beyond thought and dwells in inaccess-
ible light. Indeed, this thought is present from the beginning of Proslogion 1. Any
human understanding of God in this life must be very fragmentary and inadequate.
And even this fragmentary positive experience may only be possible for those who
begin by searching for God. Perhaps unbelievers, who have no idea what they are
looking for, never achieve a positive result. The failure of non-Christian mystics
would not worry Anselm! (Another way to reduce our expectations of Anselm’s
mystical procedure is to combine it with the modal ontological argument, so that our
positive result need only provide evidence of the possibility of God. We return to this
combined argument in sections 6.2.4 and 6.2.6.)
On my new reading, Anselm’s Proslogion offers a new test of the reliability of
Christian mystical practice—a new way to ensure that the visions of mystics are
genuine experiences of God. As we saw in chapter 5, a proper evaluation of any
 THE CASE AGAINST ATHEISM
mystical practice requires close attention to actual results, and often involves a moral
transformation. If Anselm’s procedure is a good one, we would expect to find that, at
least occasionally, it did yield a positive result, and one with a powerful impact on the
individual’s moral life and moral vision. This is an empirical question, and perhaps
we simply have too little evidence to judge. We should add an exploration of
Anselm’s procedure to the investigation of mysticism that chapter 5 has placed on
our agenda as moral philosophers.
A third worrying possibility is that Anselm’s procedure, when it is followed
correctly, only ever leads to a positive result that contradicts Anselm’s Catholic
faith. In fact, this is exactly what AP predicts. If I truly understand G, in light of
everything else I learn in other chapters about cosmic values, I will see that a being
who cares for human beings is less great than one who does not. (We return to this
possibility in section 6.3.)
6.1.5 How our project supports Anselm

Any historically informed reading of Anselm will sound strange to us. Our commit-
ment to methodological pluralism, and to avoiding the caprice of privileging cur-
rently fashionable philosophical idioms, urges us to at least take Anselm’s
G seriously. Many ontological arguments are grounded in metaphysical systems
that are no longer in vogue, such as Anselm’s own Boethian Neoplatonism. (Another
example is Leibniz’s metaphysic of simple properties, discussed in section 6.2.3.)
Methodological pluralism offers some support here, via its general edict to take
alternative philosophical modes seriously.
The most common objection to any ontological argument is that it is open to
parody because it rests on purely subjective claims. A successful ontological argument
needs objective, mind-independent foundations. And it needs reliable human access
to those objective values. Anselm needs a reliable moral or mystical doxastic practice.
This is where our substantive commitments most clearly support the argument. Our
combination of objective values and credible mysticism is just what Anselm needs.
Debate over the ontological argument vacillates between two extremes. On the one
hand, the argument itself is presented as purely a priori—compelling the agreement of
all rational observers. When this argument fails, opponents allege that we are left only
with the trading of subjective intuitions. We will see this pattern repeated for modal
claims in section 6.2. And, we have seen it before: for the cosmological argument in
chapter 3, and for the fine-tuning probabilities in chapter 4. Once again, I defend a
middle way. There are objective mind-independent facts to be discovered, and we can
reasonably claim some epistemic access to them. A commitment to non-natural (or
supernatural) moral facts is the foundation of our project, and our most striking
departure from atheist philosophical orthodoxy. The admission of such values
strengthens any ontological argument, because its evaluative claims can no longer
be automatically dismissed as subjective.
ONTOLOGICAL ARGUMENTS 
Anselm’s naturalist opponents often accuse him of ‘descending into mysticism’.

These opponents tend to lose interest at this point—confident that the very word
‘mysticism’ removes all rational credibility. As we saw in chapter 5, this is simply
naturalist prejudice. If mysticism offers insight into some supernatural realm, then
perhaps it can ground an ontological argument.
A theme of chapter 5 was the close connection between morality and mysticism.
Anselm himself sought moral lessons from mystical contemplation. To the modern
analytic reader, this seems bizarre. But if there are objective moral facts, if human
beings can access them epistemically, and if that access sometimes operates via
mystical contemplation, then Anselm’s procedure is no longer absurd.
Anselm’s procedure, focusing relentlessly on unsurpassable greatness, may offer
new insights into the realm of objective value. AP argues that our exploration of that
realm is still in its infancy. Surprising new insights are thus to be expected, from the
most unlikely sources. As with CMP itself, it is too soon to speculate about the details
of this Anselm-led moral transformation.
Finally, our earlier arguments may themselves reinforce some of Anselm’s specific
evaluative claims. In particular, cosmological and fine-tuning arguments suggest that
a being who creates a physical universe is greater than one who does not create at all,
and that a being beyond the physical universe is greater than any physical being.
(This suggestion is especially compelling for moral supernaturalists, for whom God
or cosmic purpose are ultimately identical to objective value.)
I conclude that, while it is not unproblematic, a reading of Anselm as a rational
mystic can make a real contribution to a cumulative argument for BT or AP.
6.2 The Modal Ontological Argument

Charles Hartshorne, the most prominent modern champion of the modal ontological
argument (MOA), describes it as ‘Anselm's great discovery’.30 Hartshorne credits
Anselm with the refutation of atheism. Before evaluating Hartshorne’s claim, we
must first ask what he means. Hartshorne distinguishes three positions: theism says
that God exists; atheism admits that God could have existed, but denies that God
actually exists; and positivism regards the very idea of God as incoherent. Anselm’s
discovery is that atheism is untenable. If we grant that the idea of God makes sense,
then we cannot then deny that God exists. God, by definition, is a being who could
not fail to exist. God is a necessary being. The positivist rejects this modal element as
meaningless. Every imaginable being is a contingent being. All talk of ‘necessary
beings’ is nonsense.
Hartshorne argues that Anselm refutes atheism in Proslogion 3. But he concedes
that Anselm does not refute positivism. Anselm fails to establish that the necessary
30
Hartshorne, Anselm’s Discovery. On Hartshorne’s MOA, see Dombrowski, Rethinking the Ontological
Argument, pp. 84–154.
 THE CASE AGAINST ATHEISM
being is possible. (Hartshorne himself appeals to mystical experience to plug the gap.
We return to that possibility in sections 6.2.4 and 6.2.6.) But after Anselm, the only
intellectually credible options are theism and positivism. Those who wish to deny
God’s existence must deny God’s possibility.
This section proceeds as follows. Section 6.2.1 analyses MOA and introduces the
perennial threat of parody arguments. Section 6.2.2 highlights the importance of
modal facts, and shows how our project supports MOA’s ontological commitments.
Section 6.2.3 critiques Leibniz’s attempt to prove that MOA’s necessary being is
possible. I conclude that, while our project casts doubt on Leibniz’s ambitious MOA,
it may support more modest MOAs. Sections 6.2.4 and 6.2.5 cover recent modest
MOAs based either on Pruss’s presumption that the ontological commitments of
motivationally central religious beliefs are possible or on Leftow’s modal interpret-
ation of worship-worthiness. Section 6.2.6 brings us back to Anselm’s rational
mysticism, arguing that the most promising MOA may be one based on the defence
of CMP offered in section 6.1 and chapter 5.
6.2.1 Analysing the modal ontological argument

In our earlier argument schema, the basic MOA is as follows.
1. Antecedent premise: The necessary being is possible.
2. Conditional premise: If the necessary being is possible, then the necessary being
exists.
3. Interim conclusion: Therefore, the necessary being exists.
4. Identity premise: The necessary being is God.
5. Final conclusion: Therefore, God exists.
All three premises are controversial. The identity premise is crucial. To play its
dialectical role, the argument must establish the existence of God. (As opposed to,
say, a necessary abstract entity.) For the moment, however, we focus on the argument
for the interim conclusion.
We borrow the explanatory device of possible worlds, and assume a modal logic
where each possible world represents a genuine possibility from the perspective of all
the others. (In the language of modal logic, every world is ‘accessible’ from every
other world.31) A proposition is possible in one world if and only if there is some
31
The literature on MOA typically assumes either S5, or some other modal logic containing the Brouer
axiom: for any two possible worlds w1 and w2, w1 is accessible from w2 if and only if w2 is accessible from
w1. On the links between modal logic and MOA, see Plantinga, The Nature of Necessity, ch. 10; Leftow,
‘The Ontological Argument’. Oppy argues that ‘there is a valid ontological argument . . . based on any
propositional modal logic’ (Oppy, Ontological Arguments, p. 72). For reasons of space and personal
competence, I set aside debates between competing systems of modal logic. While a comprehensive
treatment would need to engage those debates, I do not believe they offer either a watertight proof of
MOA or a knock-down objection. On the other hand, as we shall see in section 6.2.2., the ontological status
of possible worlds is of crucial importance.
ONTOLOGICAL ARGUMENTS 
world where it is true. ‘Unicorns exist’ is possible in the actual world if and only if
there is some possible world where unicorns exist.
In MOA, both the antecedent premise and the conditional premise seem clear and
uncontroversial. We begin with the former. The trading of modal intuitions is central
to contemporary analytic metaphysics. (Modal metaphysics thus has a contemporary
respectability that mysticism and Neoplatonism lack.) The concept ‘necessary being’
involves no obvious contradiction, and philosophers say seemingly intelligible things
about necessary beings. So why not grant the antecedent premise?
We can prove the conditional premise by reductio. Suppose the necessary being
(N) is possible but not actual. If N is possible, then there is some possible world (W)
such that N exists in W. If N exists in W, then N must also exist in every world
accessible from W. (This is what it means to be a necessary being.) Given our
assumption that every possible world (including the actual world) is accessible from
W, it follows that any being who exists necessarily in W also exists in the actual
world. Therefore, no necessary being can be possible without also being actual.
Unfortunately, identical reasoning threatens to establish an infinite array of
conflicting necessary beings.32 I concentrate on two parodies: devil (a necessarily
existent perfectly evil being) and knowno (a being who knows that there is no
necessary being). Neither involves any obvious contradiction, and we seem to
speak of them intelligently. If there is a presumption of possibility, then it surely
extends to these parodies. But this leads to contradiction. There cannot be two
distinct omnipotent beings (one good, the other evil). And knowno and necessary
being cannot coexist. If knowno is even possible, then there is some possible world
where there is no necessary being. But the necessary being cannot exist at all unless it
exists in every possible world.
We now see the importance of the identity premise. To avoid parody, MOA must
show both (a) that only one necessary being is possible, and (b) that that being is
God.33 To evaluate MOA, I focus on the antecedent premise. By definition, a
necessary being is one that satisfies the conditional premise. The real question is
whether such a being is possible.
Any solution must meet two desiderata. First, we need some objective standard to
separate God from devil or knowno. Second, our epistemic access to the antecedent
premise should be independent of our knowledge of the argument’s conclusion.
Obviously, if I already know that God exists, then I know that God is possible. But
then I don’t need MOA at all. MOA is helpful only if I can know that God is possible
without (yet) knowing whether God exists.
32
Oppy, Ontological Arguments, pp. 171–9; Tooley, ‘Plantinga’s Defense of the Ontological Argument’,
p. 424.
33
Technically, proponents of MOA can admit some necessary beings in addition to God, so long as the
existence of those beings is not incompatible with God’s. (Necessary abstract beings are one obvious
candidate.) Whether any other necessary being could coexist with God depends on whether its independent
existence would compromise God’s sovereignty. Fortunately, we can safely set that tangled issue aside.
 THE CASE AGAINST ATHEISM
6.2.2 Are there modal facts?

MOA seems to appeal to individuals’ modal intuitions. But this quickly leads to
stalemate. The parodist can simply cite his intuition that devil or knowno is possible.
This is an instance of the general dialectical impotence of modal intuitions. (Indeed,
van Inwagen claims that there are no interesting examples of modal intuition-trading
leading a philosopher to change his mind.34) MOA needs objective modal facts,
which we can know. Modal nihilists deny that there are such facts; while modal
sceptics deny that we have any epistemic access to them. MOA thus rests on
controversial metaphysical and epistemic assumptions.
This is a familiar situation, and our options are also familiar. The simplest solution
is to claim direct epistemic access to modal reality. Our project offers indirect support
here, given its general support for objective non-natural facts and the possibility of
human epistemic access to them. If human beings can know moral or mystical truths,
why not modal ones? After all, modal notions are common currency in analytic
philosophy. (Where would we be without ‘possibly p’ and ‘necessarily p’?)
This ambitious route will appeal to modal realists, for whom other possible worlds
are just as real as the actual world. But most philosophers prefer instead to treat
possible worlds as logical constructs or convenient fictions: useful tools, but not fit to
bear the metaphysical weight required for MOA. A more modest approach is to treat
modality as another site of reasonable disagreement. MOA will only appeal to those
who already endorse both the general idea of objective modal facts and one particular
account of those facts.35 However, this introduces a new controversial metaphysical
commitment, and will thus reduce the argument’s dialectical force. If possible, we
aim to remain neutral regarding any prior commitment to modal facts. I will aim
instead to link modal facts to other non-natural facts that we have already accepted.
One venerable BT tradition rejects independent modal facts altogether, because
such facts are an insult to divine sovereignty. If everything outside God owes its
existence to God, then either modal facts are facts about God (perhaps possibilities
are ideas in the divine mind) or God is the creator of modal facts. We need God to
explain modal facts.36
This connection between God and modal facts would raise two difficulties for
MOA. First, if God is invoked to explain modal facts, then we cannot cite those facts
to prove God’s existence, on pain of circularity. Second, if modal facts are so closely
tied to God, then the atheist will simply reject modal facts, thus depriving MOA of
any dialectical force.
The first worry is easily addressed. We simply supplement MOA with a new
disjunctive argument. Either modal facts need further explanation, or they do not.
34
Van Inwagen, ‘Modal Epistemology’.
35
In particular, MOA assumes that the modal facts obey the Brouer axiom (see footnote 31).
36
This argument is especially associated with Leibniz (Adams, Leibniz, pp. 177–91).
ONTOLOGICAL ARGUMENTS 
If they do, we must posit God to explain them. If not, we can use modal facts to
ground MOA. Either way, we get the conclusion that God exists.
To address the second worry, we need an independent reason to endorse objective
modal facts. One promising route, given our project, is to connect modal facts with
moral and mystical facts. Here, our response to modal nihilism and scepticism
draws on Anselm. It is highly controversial whether Anselm himself put forward a
modal argument.37 What is clear is that, even if his argument has modal elements,
Anselm does not begin with divine necessity. Rather, he starts from another much
richer concept, of which necessary existence is only one corollary. Anselm seeks that
than which no greater can be conceived (G). He then discovers that any such being
exists necessarily, and is identical to the Christian God. This rules out any necessary
beings that are incompatible with God, but it does not commit Anselm to the claim
that every necessary being is identical to God.
As we saw in section 6.1.4, Anselm’s use of formula G explicitly relies on human
knowledge of objective moral facts, obtained within a reliable mystical doxastic
practice. If Anselm’s procedure yields a positive result, then this also vindicates
other claims he makes about G. Some of those claims are modal. So Anselm’s
procedure could vindicate modal facts. (We discuss a more direct connection
between Anselm’s mysticism and MOA in sections 6.2.4 and 6.2.6.)
Plantinga’s much-discussed contemporary MOA also follows Anselm’s lead. Plan-
tinga begins with the idea of a being of maximal greatness.38 A maximally great being
is one that is maximally excellent in every possible world. Therefore, a maximally
great being exists in all possible worlds. (You can’t be maximally excellent where you
don’t exist!) Plantinga’s MOA has only one premise: ‘maximal greatness is possibly
exemplified’.39 In a similar vein, other MOAs talk of ‘perfect beings’ or ‘beings
worthy of worship’. These notions also have their roots in Anselm, and all share
Anselm’s original melding of modal, moral, and religious notions.
Without objective modal facts, MOA cannot get off the ground. Once we recog-
nize the evaluative dimension of MOA’s modal facts, and their connection to
mystical practice, the arguments of chapter 5 vindicate this necessary first step.
MOA no longer seeks unprecedented knowledge of an otherwise inaccessible and
unnecessary realm. But the recognition of modal facts is only the beginning. We still
need to avoid parody. Can we defend the antecedent premise without admitting the
possibility of devil or knowno? The next three sections explore three ways to ground
the antecedent premise: proof, presumption, and experience.
37
Visser and Williams note that Anselm does not explicitly mention God’s necessary existence in
Proslogion, introducing it only in his response to Gaunilo (Visser and Williams, Anselm, pp. 130, 160;
Logan, Reading Anselm’s Proslogion, p. 125). On the other side, both Hartshorne and Dombrowski argue
that the modal ontological argument is contained in chapter 3 of Proslogion. (As do Visser and Williams,
Anselm, p. 92.)
38
Plantinga, The Nature of Necessity, p. 214.
39
Plantinga, The Nature of Necessity, p. 214.
 THE CASE AGAINST ATHEISM
6.2.3 Proving the antecedent premise

The most ambitious MOA claims to prove the possibility of God. Leibniz is often
credited as the first to explicitly note the need to establish that possibility.40 Leibniz
also offers a proof.41 God is a being who has all perfections. A perfection is the
infinite degree of any positive property (power, knowledge, goodness, and so on).
Every positive property is a simple property that is not made up of other properties.
Leibniz then introduces a rebuttable presumption of possibility. For any complex
concept C, C is possible unless there is a contradiction among the properties that
constitute C. There can be no contradiction between two or more simple properties,
because each simple property has its own distinct domain. (Power cannot contradict
knowledge, because each concerns a different aspect of the being in question.)
Therefore, no concept containing only simple properties can harbour a contradic-
tion. The concept of God involves no internal contradiction. Therefore, God is
possible.
How might Leibniz deal with our parodies? He will first claim that no parody deals
exclusively in simple properties, and then search for hidden contradictions. Leibniz
could borrow from Anselm’s reply to Gaunilo, and diagnose a contradiction in the
very idea of any necessary physical being because ‘physical’ implies ‘contingent’. Or he
could argue that divine unity entails the impossibility of a perfect-but-evil being,
because an omnipotent omniscient being could not fail to know and be motivated by
moral truths. (For Leibniz, evil is parasitic on goodness, and therefore not a simple
property.)
Here is how Leibniz might undermine knowno. I can know there is no necessary
being only if I know that no necessary being is possible. God is one putative necessary
being. So the knowno must know that God is not possible. A being is impossible only
if its concept hides a contradiction. But Leibniz has proved that the concept of God
hides no contradiction. And that proof yields a necessary truth, because it cites only
the concept of God and the logical features of simple properties. There is no possible
world where the concept of God hides a contradiction. And nothing can be known
that is not true. Therefore, there is no possible world where someone knows that God
is not possible. Despite appearances, the knowno is not possible. This concept does
hide a contradiction. Fools think knowno is possible, but understanding reveals that
it is not.
If we grant his metaphysic of simple properties, Leibniz’s argument is sound.
Unfortunately, Leibniz’s metaphysic is no more popular today than Anselm’s Neopla-
tonism. Neither will convince any modern atheist. However, some other metaphysical
40
Logan argues that, in fact, Scotus was the first to spot this famous lacuna in MOA (Logan, Reading
Anselm’s Proslogion, p. 145).
41
The text offers a very loose and anachronistic paraphrase of Leibniz. For more detailed discussion, see
Adams, Leibniz, pp. 135–56.
ONTOLOGICAL ARGUMENTS 
proof of possibility may succeed, and some do have their supporters.42 For the modern
philosophical audience, perhaps the most unsettling aspect of any Leibnizian dissol-
ution of parodies is the ineliminable evaluative dimension. (Why are those the simple
properties? Why is that the correct account of divine unity?) Leibniz needs objective
evaluative claims to distinguish the true necessary being from parodies. Whether we find
any such proof plausible depends on our attitude to objective moral facts. Given our
commitments to the metaphysical significance of evaluative facts, Leibniz’s argument
(or something similar) could be a fruitful avenue for AP metaphysics to explore. I leave
that task for another day. While the literature often focuses on Leibniz’s ambitious
attempt to prove that God is possible, more modest MOAs are increasingly popular and
arguably nearer to Anselm’s own intentions. I now turn to these modest alternatives.
6.2.4 Pruss and the presumption of possibility

If I claim that X exists, and you deny this, then I must give you some reason to believe
that X exists. But if I claim only that X is possible, and you deny even that, then many
people would say you now bear the burden of proof. There is a presumption of
possibility that you must rebut. If there is a presumption of possibility, then MOA
can help itself to the antecedent premise, at least initially. The obvious problem is
parody. A generic presumption of possibility will apply to the devil and knowno as
well as God. But then the presumption collapses into contradiction.
One option is to argue that a general presumption of possibility is rebutted for
parodies and not for God. (This is one place where a modest version of Leibniz’s
‘proof ’ from section 6.2.3 could be redeployed.) By its nature, this approach is
piecemeal. We rebut the presumption one parody at a time. The obvious disadvan-
tage is that new parodies can always be generated. (This is what drives the MOA
literature.) Under this first strategy, a successful MOA is only ever provisional.
An alternative is to restrict the presumption itself. One common thought is that
the presumption of possibility only applies to real concepts, not gerrymandered ones.
We must then ask what makes a concept real. One popular answer is that we should
only presume possibility when a concept plays a central role in some established
social practice. (It must be a central role, because even parodies have some role in the
social practice of philosophy.)
Alexander Pruss has recently defended the following argument based on this
stricter presumption:43
(1) If X is a human individual or community that leads a flourishing and intel-
lectually sophisticated life of significant length while holding a motivationally
central belief that P, then, probably, P is possible.
42
For instance, another recently popular MOA is that of Godel, which draws on a notion of positive
properties that seems close to Leibniz. (See, e.g., Hazen, ‘On Godel’s Ontological Proof ’.)
43
Pruss, ‘The Ontological Argument and the Motivational Centres of Lives’, pp. 233–4.
 THE CASE AGAINST ATHEISM
(2) A number of individuals and communities have led a flourishing and intel-
lectually satisfying life of significant length while holding a motivationally
central belief that there is a maximally great being.
(3) Therefore, probably, it is possible that there is a maximally great being.
Pruss’s argument has obvious affinities with Alston’s defence of CMP (see
chapter 5). Both argue that a central role in the life of a flourishing community,
over a significant period of time, creates some presumption in favour of a belief.
And, like Alston, Pruss’s favoured example is the role of beliefs about God in the
Christian community.
However, there are several key differences between Pruss and Alston. We begin
with two ways that Pruss is less ambitious than Alston. First, Pruss’s presumption
concerns, not truth or rationality, but merely possibility. Second, Pruss’s focus on the
‘flourishing and intellectually satisfying life’ of a community is wider than Alston’s
focus on mystical doxastic practices. Christian communal life does include Christian
mysticism. But a community could have a flourishing and intellectually satisfying life
without any specifically mystical doxastic practice. And Pruss’s argument makes no
specific claims about the reliability of mysticism. Pruss’s MOA thus provides BT with
a new resource, one that could be used to convince those who are suspicious of
mysticism.
On the other hand, Pruss is also more ambitious than Alston in two ways.
While Alston focuses on first-personal rationality, Pruss tries to convince third
parties. The presumption of possibility should be recognized by everyone, not just
by those within community X.44 Second, to support the antecedent premise of
MOA, Pruss must show that modal beliefs are socially central, whereas Alston can
remain agnostic here. Suppose a community has built its life around worship of
God. It is not enough for some obscure philosopher in that community to
speculate that God is a necessary being. God’s necessity must itself be a central
motivating belief for ordinary believers. (Or, at the very least, necessity must
follow very directly from some canonical divine attribute, such as perfection,
omnipotence, or unsurpassability.)
To establish the centrality of modal beliefs, Pruss borrows Plantinga’s notion of a
being of maximal greatness. He argues that this concept is central to the Christian
community’s life of faith. The Christian God is maximally great, unsurpassably
perfect. And the modal claims that MOA needs flow directly from this central
commitment. As an account of the Christian religious tradition, this seems cred-
ible. Furthermore, this is one place where our earlier interpretation of Anselm as
rational mystic could directly support MOA. Anselm’s Proslogion arose within a
44
By contrast, Plantinga, who is Pruss’s inspiration, presents a first-personal MOA, where belief in the
possibility of a maximally great being is rational for Christians (Plantinga, The Nature of Necessity, p. 221).
ONTOLOGICAL ARGUMENTS 
contemplative Christian tradition focused on the unsurpassability of God, and

Anselm presents his meditation as a natural development of that existing tradition.
Perhaps it is professionals, not amateurs, for whom modal notions are central. If he
were willing to privilege specialized mystical doxastic practices over everyday
Christian worship (and perhaps then link the two as Alston does), Pruss could
borrow more directly from Anselm.
Whether based on everyday Christian worship, or on the specialized thinking of
mystics, Pruss’s empirical claim about the central role of modal concepts seems
plausible. I think we can reasonably grant Pruss his second premise.
That leaves Pruss’s first premise: the presumption of possibility itself. Pruss must
deny that any parody satisfies his presumption. This seems straightforward. Parodies
are a paradigm of gerrymandered concepts! No flourishing communal life has ever
centred on belief in golden mountains, omnipotent devils, or knownos. However,
Pruss himself concedes that, while parodies don’t qualify for the presumption of
possibility, non-Christian religious beliefs do.
Religious diversity is an obvious threat to Pruss. Many religions have central beliefs
that contradict BT. Indeed, some endorse atheism. (Isn’t the belief that there are no
enduring substances central to many Buddhist communities?) Surely, by Pruss’s own
lights, all these beliefs are possible. But if God’s non-existence is possible, MOA will
collapse.
Pruss discusses atheism. He argues that, while positive beliefs can play a central
motivating role, the negative belief that there is no God cannot. (It is no coincidence
that atheist groups typically self-identify in more positive terms: rationalist, human-
ist, naturalist, liberal, and so on.) No atheist community has flourished for any length
of time on this negative foundation.
But what about Buddhist atheism? Here, Pruss offers a different objection. He
dismisses ‘illusionist metaphysics’ (e.g. traditional Buddhism) because it is inca-
pable of motivating genuine human love, an essential component of a flourishing
life. Perhaps a flourishing community can profess an illusionist metaphysic. But
its flourishing will then be in spite of that metaphysic, and not thanks to it. Pruss
thus relies on a controversial account of what counts as a flourishing and
intellectually sophisticated life. This will not impress those opponents who
already suspect that MOA smuggles in subjective preferences masquerading as
objective moral values. Like Anselm and Leibniz, Pruss must appeal to objective
moral facts.
Our commitments to objective values, and to taking religious traditions seriously,
support Pruss’s general strategy. It is reasonable to have some presumption in favour
of socially entrenched beliefs. And when two equally well-entrenched traditions
make conflicting claims, it is reasonable to use objective moral criteria to decide
between them. Moral or mystical error can rebut the presumption of possibility.
However, as we’ll see in section 6.3, and at greater length in chapter 9, AP turns
Pruss’s strategy against BT (and therefore against Christianity).
 THE CASE AGAINST ATHEISM
6.2.5 Leftow’s argument

Brian Leftow’s recent ontological argument is similar to Pruss’s MOA, but suffi-
ciently different to merit separate consideration.45 Instead of Plantinga’s and Pruss’s
‘maximal greatness’, Leftow focuses on worthiness to be worshipped. (This is obvi-
ously distinct from actual worship. Many unworthy things have been worshipped.)
Suppose you are actively engaged in religious practices involving worship. Worship
includes giving thanks for the good things in life. You reflect that, if they were created
by God, then these good things provide appropriate reasons to worship God. You
thus imagine God as a possible being who created the good things in this world and is
worthy of worship. To be truly worthy of worship, God must be omnipotent. God is
omnipotent only if there is nothing independent of God, nor could there be. Leftow’s
omnipotence thus requires both (1) that all other actual things owe their existence to
God; and (2) that there is no possible thing that would have been independent of God.
As you have imagined God, you know that God is possible. There is a possible
world (w) where God exists. Now consider the alleged possibility that, despite existing
in w, God does not exist in the actual world (@). To prove that this is not possible,
you proceed by reductio. Suppose God exists in w and not in @, and that there
are good things in @. Imagine someone in w, who imagines the good things in @.
Because God does not exist in @, these things in @ are not dependent on God.
However, from the point of view of w, things that exist in @ are possible things.
Therefore, in w, there are possible things that are not dependent on God. If God does
not exist in @, then God is not omnipotent in w. But a being who is not omnipotent is
not God. Therefore, if you are sure you can imagine God, then the being you are
imagining must exist in @ as well as w. A worship-worthy being cannot exist in w
without also existing in @. QED.
We can strengthen Leftow’s conclusion if we suppose that the same things exist in
both w and @. This supposition is reasonable. When she thinks of God, the ordinary
religious believer does not picture some distant possible world containing bizarre
unfamiliar things. She looks at the good things in this world, and imagines a God who
created those things. We can now say that, in w, there are some actual things that
might have existed independently of God. w’s ‘God’ is now very clearly not
omnipotent.
The crucial question is whether worship-worthiness requires Leftow’s strong
omnipotence. Consider some analogies. A human creative artist (A) creates a
timeless work of art (X). If X—the very same work of art—might have existed in a
nearby possible world without any input from A, then A is less creative than if
X could only possibly have existed through A’s creation. Or consider the scientific
45
In the text, I offer an informal, abbreviated account of Leftow’s argument, drawing especially on
Leftow, ‘The Ontological Argument’. See also Leftow, ‘Anselmian Polytheism’; and Leftow, ‘Individual and
Attribute in the Ontological Argument’.
ONTOLOGICAL ARGUMENTS 
achievements of a Godel or an Einstein. Aren’t those achievements greater if no other

possible scientist could have discovered them?
Now consider the creation of the whole physical universe. In w, the physical
universe owes its existence to God. Suppose, for reductio, that the actual physical
universe does not. Therefore, in w, there could have been a physical universe that was
not dependent on w’s God. Intuitively, such a God is not as perfect as one without
whom no physical universe could have existed. If this universe could have existed
without a divine creator, then no possible creator is truly worthy of worship.
Leftow’s argument is not a modal ontological argument. A worship-worthy being
need not be a necessary being. While Leftow’s argument rules out possible worlds
where there are other things and no God, it does not rule out empty possible worlds
with no God and nothing else.46 However, Leftow still faces analogues of the standard
modal parodies. Knowno becomes a being who knows there is no being who is
worthy of worship; the devil is an evil being worthy of worship; non-Christian
religions posit alternative objects of worship who are incompatible with Leftow’s
(Christian) God; and so on.
Like Pruss, Leftow appeals to actual religious practice both to develop his notion of
worship-worthiness, and to establish its possibility. If, in their daily worship, study,
and contemplation, Christians have been fervently engaged in imagining an omnipo-
tent being for some two millennia, then this is some evidence that such a being is
possible.
Leftow’s argument raises two questions. First, does Leftowian worship-worthiness
have a central place in Christian religious practice? Of course, Christians suppose
that their God is worthy of worship. But what about Leftow’s modal claims? Critics
will object that Leftow has grafted his own particular views about worship-worthiness
onto the Christian tradition.
Leftow does not claim that Christian believers have always (or indeed ever) had
abstruse modal concerns in mind. (Any more than Plantinga or Pruss would claim
that ordinary Christians who think God is maximally great are deliberately endorsing
S5!) The modal implications are implicit in everyday Christian practice. What is
explicit is that, to be worthy of worship, a being must be an omnipotent unsurpass-
able creator. Christians have long distinguished their God from pagan deities on the
grounds that such lesser beings were never worthy of worship. And, once we think
through the logical implications of divine omnipotence, we will reach Leftow’s modal
conclusions. Or so Leftow claims.
Like Pruss’s parallel one, Leftow’s empirical claim is not implausible. And
I propose to grant it for the sake of argument. This brings us to our second question.
Even if Christian worship does presuppose a worship-worthy being, is this a good
reason to believe that such a being is a coherent possibility? Suppose we agree that
46
Leftow himself does believe that God is necessary (Leftow, God and Necessity, ch. 7). But his
ontological argument does not require divine necessity.
 THE CASE AGAINST ATHEISM
God’s worship-worthiness is implicit in the Christian tradition. Many critics will

simply refuse to grant this fact any evidential force. Perhaps two millennia of
Christian thinkers and believers have been talking nonsense, and directing their
worship to an impossible object.
Many opponents will regard Leftow’s exercise as an expression of individual
subjective preferences or idiosyncratic value judgements. Worthiness of worship is
an ineliminably moral (or at least evaluative) concept. Leftow’s argument is only
plausible if he is exploring (however contestably and tentatively) some realm of
objective moral facts.
Our commitments support Leftow here, just as they support Pruss. Leftow’s modal
claims are not eccentric preferences, but claims about the structure of a realm of
objective moral (and perhaps mystical) facts. As with Pruss, however, the greatest
threat comes from religious diversity. Consider, again, a Buddhist who argues that his
religious tradition centres on the search for enlightenment, and that enlightenment
essentially involves the realization that one inhabits a cosmos where no worship-
worthy being exists. (The enlightened person is a knowno!) If the central beliefs of
Buddhism enjoy the presumption of possibility, then there is no (actual) worship-
worthy being. This threatens to reduce Leftow’s argument to merely first-personal
relevance. Buddhists and Christians can both appeal to a presumption of possibility,
but only to defend the rationality of their own beliefs. There is no neutral third-
personal presumption of possibility.
A similar threat comes from religious traditions with alternative concepts of what
makes a being worthy of worship. If we are exploring objective moral reality, then we
can be mistaken about what is worthy of worship. As we’ll see in chapter 9, AP argues
that, on balance, the best explanation of actual religious diversity is one that supports
AP but is incompatible with Christianity.
6.2.6 Experience
A third way to ground the antecedent premise of MOA is to defend direct human
insight into modal truths. The way we naturally talk about modality mirrors our talk
about mathematical or moral truths. (‘Don’t you see that a round square is impos-
sible?’, ‘My considered judgement is that a golden mountain is impossible’, etc.) If the
latter two are credible objects of human knowledge, then why not the former?
Perhaps the proponent of MOA simply sees that the antecedent premise is true. Of
course, this cannot be literal sight, any more than with morality or maths. But
rational insight into other possible worlds seems not entirely hopeless.
Recall our earlier interpretation of Anselm as rational mystic. Suppose we combine
that reading with MOA. In chapter 5, we asked whether CMP offers an experience to
confirm the existence of God. We could look to Anselm to provide that experience.
But perhaps his procedure yields (only) an insight into what is possible. If I follow
Anselm’s procedure, I will not experience God, but only the coherence of the concept
of God. I can then use MOA to infer that God exists.
ONTOLOGICAL ARGUMENTS 
Like Pruss and Leftow, Anselm could thus look to established religious traditions
only for a presumption of possibility. This presumption is more modest than Alston’s
presumption that established doxastic practices reliably yield true beliefs. MOA
would then magnify the epistemic power of mystical experience—taking us from
possibility to actuality. Perhaps this is the most fruitful reading of Anselm’s Proslo-
gion, both historically and philosophically.
This new strategy has a price. The shift to MOA exacerbates the threat of parody.
As with both Pruss and Leftow, the greatest threat to our new appropriation of
Anselm is religious diversity. Anselm cannot concede that anyone could even
understand the possibility of any other unsurpassable (or necessary) being. Nor
could anyone understand the possibility that there is no God. Like Pruss and Leftow,
Anselm must now deny even the possibility of Buddhism. This would not have
bothered Anselm himself. The idea of a plurality of equally epistemically credible
doxastic practices would be anathema to him. Whether it should bother us is another
question, to which we return in chapter 9 (and briefly in section 6.3).
6.2.7 Conclusions re MOA

MOA is not decisive. We cannot establish the conditional premise without invoking
controversial claims about objective moral standards, the comparative superiority of
one religious tradition over others, or the reliability of mystical experience. However,
at this stage in our project, these limitations are no surprise. Aside from religious
diversity, to which we return shortly, the strongest opposition to MOA concerns the
existence of objective values and the plausibility of a presumption of epistemic
credibility for religious communities and especially their mystical practices. For the
atheist moral-nihilist global-naturalist, MOA looks hopeless. Given our commit-
ments, and the conclusions of earlier chapters, MOA is promising. I conclude that
MOA may play a role in a cumulative case for BT. Our final question is whether AP
can borrow it.
6.3 Can AP Borrow the Ontological Argument?

MOA looks easy for AP to borrow. The bare notion of a necessary being does not
favour BT over AP. However, as we have seen, to establish the antecedent premise
MOA may need other concepts with specific BT implications. ‘Perfect being’, ‘max-
imally great being’, ‘worship-worthy being’ all deliberately carry implications incom-
patible with AP. Nor is this surprising, given the origin of the ontological argument.
Anselm’s God is not compatible with AP. Proslogion is filled with direct appeals to
God, and reflections on God’s concern for human beings. What can AP borrow from
this distinctively Christian work? And how could AP borrow Pruss’s and Leftow’s
defences of the antecedent premise, which explicitly cite social facts about Christian
religious practice?
 THE CASE AGAINST ATHEISM
AP could simply reject the ontological argument. It could even embrace parody,
and argue that the central concepts of Christianity are incoherent. BT does not need
the ontological argument.47 But it does need an intelligible concept of God. AP could
thus use parody to refute BT, and then defend a less metaphysically loaded cosmic
purpose. While this negative option has some appeal, I believe instead that we can
incorporate an ontological argument into our cumulative case for AP.
AP needs to reinterpret the human-centred elements of the best ontological
arguments. For illustration, I focus on the notion of a maximally great being. (Similar
remarks apply to perfect, unsurpassable, or worship-worthy beings.) AP is impover-
ished if it concedes this concept to BT. (‘Yes, a maximally great being would care for
individual human beings. Sadly, there is no maximally great being.’) Such a conces-
sion threatens to severs the link between AP’s non-human-centred cosmic purpose
and objective values. Instead, AP should insist that a maximally great being would
not care for human beings. We could offer this as an analysis of the concept of
maximal excellence. Perhaps more credibly, we could instead say that philosophical
analysis leaves open the choice between BT and AP interpretations of excellence. We
would then need other evidence to choose between the two interpretations. For
instance, given the amount of evil the world contains, it cannot have been created
by an omnipotent human-centred being. If a maximally great being must be omnipo-
tent, then any maximally great creator of this world must be non-human-centred.
Our ontological argument thus mirrors our explanation of the combined evidence
of mystical tradition and religious diversity (chapters 5 and 9). Atheism dismisses all
mystical traditions as mistaken; while BT arbitrarily singles out one. AP argues that
all religious traditions offer partial insight into the divine realm. However, because
these insights are filtered through a self-aggrandizing lens, they must be purged of
human-centred accretions. In the same way, different religious all posit some max-
imally great being or ultimate reality, but they differ radically as to its specific
features. AP borrows the general notion of maximal greatness, and then offers a
non-human-centred interpretation. In isolation, the evidence offered by Anselm,
Pruss, and Leftow supports a specifically Christian God. But our total evidence
instead supports AP. (Pruss himself acknowledges this possibility, when he observes
that an argument from evil could rebut his presumption that the Christian’s perfectly
benevolent creator of this world is possible.48)
Suppose the other arguments in parts I and II go through. AP does offer the best
explanation of all religious and other phenomena, and it can construct a credible
non-human-centred picture of objective moral facts where a maximally great or
perfect being is one that is indifferent to human beings. AP can then borrow the most
plausible ontological arguments. And, as we have seen throughout this chapter, those
47
Indeed, many BT philosophers explicitly reject the ontological argument. (Two notable contempor-
ary examples are Swinburne and van Inwagen.)
48
Pruss, ‘The Ontological Argument and the Motivational Centres of Lives’, p. 242.
ONTOLOGICAL ARGUMENTS 
arguments are considerably strengthened by earlier chapters—especially chapter 5’s

links between mysticism and morality, and the defence of objective values in chapters
2, 3, and 4. As with cosmology, teleology, and mysticism, the best ontological
argument will give us Normative AP, not merely Simple AP.
What would a successful ontological argument teach us about cosmic purpose or
objective value? Here, the ontological argument is closer to the cosmological than to
the fine-tuning argument. Careful attention to Anselm’s formula G might reveal the
hitherto underappreciated modal dimension of objective values. Beyond that, how-
ever, substantive moral lessons are likely to be limited. Even if the ontological
argument tells us that there is a maximally great, unsurpassable, or perfect being,
we must look elsewhere for any detailed knowledge of what maximal greatness is.
This concludes our survey of the arguments that AP borrows from BT. We have
seen that, taken together, they constitute a plausible case for Normative AP against
atheism. However, all our arguments also support BT against atheism. And some
favour BT over AP, at least in their most familiar forms. Therefore, AP must turn to
atheist arguments against BT, to see what it can borrow from them.
PART II
The Case against Benevolent
Theism
7
Arguments from Scale
Is the universe as it is revealed to us by modern science roughly the sort of

universe which we would antecedently expect a God of traditional theism to
create? Is this an apt universe, given the admittedly sketchy conception we have
of his nature and his intentions?
The short answer to this is ‘No’. In almost every respect, the universe as it is
revealed to us by modern science is hugely unlike the sort of universe which the
traditional thesis would lead us to expect.
[Nicholas Everitt, The Non-Existence of God, p. 216]
The sheer size of the universe is no reason to depreciate the significance of the evolution
of mind in one tiny corner of one solar system in one galaxy among myriads. The
quantitative dimensions of the physical base of the evolution of life, consciousness, and
intelligence are totally irrelevant to the question of the significance of these qualitative
developments . . . the appearance of the noosphere, to use Teilhard de Chardin's term, is
a qualitative leap of vast significance, given its nature and its results, quite irrespective of
the sheer size of the mindless universe out of which it has appeared.
[Brian Hebblethwaite, In Defence of Christianity, pp. 11–12]
In part I, AP borrowed from BT against atheism. Part I established a case for cosmic
purpose, with AP and BT emerging as competing interpretations of that purpose. In
part II, AP borrows from atheism against BT. AP also contends that the best
‘arguments for atheism’ merely rebut the human-centred component of BT. They
do not support atheism against AP.
The cornerstone of part II is the argument from evil. But we begin with something
much less familiar. Unlike the other arguments in this book, the argument from scale
has received little philosophical attention.1 I include it because it captures a common
intuitive reaction to BT that AP finds especially congenial, and because it highlights a
number of general themes of AP.
1
A search on the Philosophers Index for ‘argument from scale’ yields no hits, and all the items found in
a routine Google search relate to Everitt’s brief discussion. Perhaps the argument sometimes goes by
another name. If so, I have not discovered it. (For an exploration of similar themes from a theological
perspective, see Gustafson, Ethics from a Theocentric Perspective, especially vol. 1, pp. 88–99.)
 THE CASE AGAINST BENEVOLENT THEISM
This chapter is especially speculative, partly because the argument from scale is
underdeveloped and partly because I am unsure exactly how considerations of scale
impact on the debate between AP and BT. After setting out the logical structure of
the argument (section 7.1), we ask what BT and AP predict about God’s creation
(section 7.2). A preliminary conclusion is that the relevant issue is not the size of the
universe, but the nature and distribution of its inhabitants. Opponents of BT argue
that the best possible world would contain more numerous, more various, or more
accomplished inhabitants. Picking up a debate prefigured in chapters 1 and 3, section
7.3 then argues, against prominent arguments from Robert Adams and Richard
Swinburne, that God must create the best possible world. (Or, at the very least,
that BT cannot defeat the argument from scale simply by pointing to gaps between
divine choice-worthiness and objective value.) Sections 7.4 and 7.5 then examine two
empirical possibilities: that human beings are alone in the cosmos, and that we are
not. Section 7.4 argues that, while BT does provide a superior explanation of human
uniqueness, its superiority here is not decisive; while section 7.5 argues that the
existence of superior beings elsewhere in the cosmos would strongly support
AP. I conclude that we should therefore add the search for extraterrestrial life to
our list of surprising questions of vital importance to moral philosophy! Section 7.6
concludes by asking how we should respond to our empirical uncertainty about
extraterrestrial life, arguing that the same considerations that support AP also
support a reluctance to presume we are alone.
7.1 What Does the Argument Show?

The arguments from scale and evil share a common structure. Both allege that some
feature of our universe is incompatible with divine creation. The argument from evil
claims that BT is inconsistent with the evils of this world. As we’ll see, the objec-
tionable feature that motivates the argument from scale is harder to pin down.
Everitt suggests that God would create a universe ‘on a human scale’,2 and this is
not the world we inhabit. He contrasts two possible worlds. Genesis World, suggested
by a literal reading of the Bible, is human-sized. Human beings occupy a significant
place at the centre of the physical universe. We emerge near the beginning of the
universe, and are there until the end. Science World is very different. The universe has
existed for billions of years and will continue for billions more. Human beings
emerged only very recently and are unlikely to survive very long. There are billions
of galaxies and many billions of stars. Our Sun is one small star in one nondescript
galaxy. Life on Earth has existed for several billion years, and humans are a tiny
fraction of all living things. And so on.
2
Everitt, The Non-Existence of God, p. 215.
ARGUMENTS FROM SCALE 
Genesis World is human-sized, while Science World is not. Some BTs notoriously
defend the Genesis narrative, or argue that modern science could be mistaken. I set this
response aside. Our BT does not reject science, but attempts to accommodate it. As
Everitt notes, while modern science might be mistaken in details, it is hard to imagine
that future science will ever restore the human-sized universe of traditional BT.3
We can now state the argument from scale:
1. God would create a human-sized universe.
2. Our universe is not human-sized. (Call this fact ‘not-H’.)
3. Therefore, God does not exist.
This schematic argument raises three questions. What should we expect God to
create? Are those expectations realized? What does a gap between reality and
expectation tell us about God’s existence? We will concentrate on the first two
questions. BT’s strongest reply is that we have no reason to expect God to create a
universe unlike this one. But we begin with the third question.
Suppose we would expect God to create a different universe. The most ambitious
atheist would conclude that God’s non-existence has been demonstrated, or at least
rendered highly probable. To my knowledge, no one has defended such ambitious
arguments from scale. There is no contradiction involved in God’s creating a human-
sized universe. Nor is this single unexpected fact sufficient, on its own, to outweigh all
the positive evidence for BT. The argument from scale is C-inductive: ‘The claim is only
that the findings of modern science significantly reduce the probability that theism is
true.’4 Just as S (there is something, when there might have been nothing), FL (our
universe is friendly-to-life), or M (our universe is governed by precise and elegant
mathematical laws) might raise the probability of BT, so not-H reduces that probability.
C-inductive arguments are comparative. Some fact F supports BT only if BT
explains F better than atheism does. In the cosmological argument, we saw that
atheism treats S as a brute fact, while BT offers a superior explanation. The argument
from scale has a different structure. Atheism does not explain not-H. Like S or FL in
chapters 3 and 4, the atheist treats not-H as a brute fact. This is just how things are.
Atheists have no prior expectations about the scale of universe. Not-H supports
atheism over BT because BT does create expectations about the universe God will
create, and those expectations are not met. BT expects H. Atheism cannot explain
not-H, but BT cannot explain it away.
If neither atheism nor BT can explain not-H, then perhaps some other position is
superior to both, because it explains why the universe is not-H. One obvious
candidate is AP. We should expect a non-human-centred God to create a universe
like Science World. AP therefore aims to use the argument from scale against both
BT and atheism.
3 4
Everitt, The Non-existence of God, p. 219. Everitt, The Non-Existence of God, p. 218.
 THE CASE AGAINST BENEVOLENT THEISM
No C-inductive argument is decisive. BT could simply accept the atheist’s com-

parative conclusion. Not-H is evidence against BT, but it is outweighed by our
reasons to accept BT. As we saw in chapter 5, some BTs, following Alston or
Plantinga, believe it is rational to retain one’s belief in God unless it is defeated.
They could then argue that not-H is insufficient to overturn the presumption
favouring BT. However, unlike Alston’s or Plantinga’s, our dialectical situation is
third-personal, not first-personal. Our BT wants to persuade third parties. She must
offer arguments for BT. (This dialectical context becomes especially important in
chapters 8 and 10.) Our BT must therefore argue that the balance of arguments
favours BT over atheism. One powerful option is to counter scale with fine-tuning
(chapter 4). If the atheist must accept FL as a brute fact, then BT’s inability to
adequately explain not-H is less harmful. Given the astronomical improbability of
FL, BT has the advantage here even if God was extremely unlikely to create a universe
like this.
AP reduces the power of this counterargument. As we saw in part I, the traditional
arguments for theism strongly support the disjunction of BT and AP, but only (at
best) weakly support BT over AP. BT is no longer alone in explaining FL. It is now
much harder to prove that BT offers the best explanation of all the facts. If the
argument from scale has some force, then, to outweigh its inability to account for
not-H, BT must do better elsewhere. BT’s explanation of other facts must be
significantly better than AP’s.
Other features of our dialectical context are also crucial throughout part II. My
case for AP is cumulative. This means both that our arguments fit together, but also
that later arguments can legitimately presuppose the conclusions of earlier ones,
along with any interim conclusions we have established along the way. And, because
part I’s case established Normative AP, those earlier conclusions include a tight
connection between cosmic purpose and objective value. In particular, the value
commitments defended in chapters 3 and 4 can now appear as premises in AP’s
arguments from scale and evil, as can our conclusions about divine perfection in
chapter 6. These commitments reinforce my consequentialist pre-commitments. As
we shall see, while they may seem metaphysical, arguments from scale and evil
actually rest primarily on moral premises.
In our dialectical context, BT is unwise to simply concede that the argument from
scale is strong. Instead, BT will try to weaken the argument. We first briefly address
two attempts to sidestep the argument: finite BT and sceptical BT.
Arguments from scale or evil exploit a tension between two traditional features of
God: omnipotence and benevolence. BT traditionally defends both. Finite BT aban-
dons omnipotence.5 God has created a deficient world. But this was the best cosmos
5
On finite BT, see Dilley, ‘A Finite God Reconsidered’; Bishop, ‘Can There Be Alternative Concepts of
God?’ One prominent finite theist position is Hartshorne’s ‘process theology’. (See, e.g., Hartshorne,
Omnipotence and Other Theological Mistakes.) In Chapter 1, to justify my terminological preference for
ARGUMENTS FROM SCALE 
God could create. Finite BT is a genuine rival to AP. Both agree that we must depart
from traditional BT. However, an adequate treatment of finite BT would make this
book too long. Therefore, I limit myself to three brief remarks. First, finite BT has
difficulty borrowing the arguments in part I, especially if it seeks to defend BT over
AP. If God’s choices are so constrained, then how can we be confident that God has
any interest in human beings, or rational beings, or any other actual feature of the
cosmos? Perhaps our existence also reflects God’s limited creative options. Second,
positing a benevolent finite God instead of an indifferent unconstrained God puts
undue weight on our evaluation of our own cosmic significance. If we must depart
from classical BT, why privilege benevolence towards humans over omnipotence?
(Especially when, as we saw in chapter 6, moral perfection need not involve concern
for us.) Finite BT is thus, somewhat surprisingly, more capriciously anthropocentric
than traditional BT. Finally, any adequate case for benevolent finite theism requires a
careful consideration of the best case for AP. This book focuses on building that case.
The argument from scale turns on expectations about God. Sceptical BT replies
that God is inscrutable.6 We cannot grasp God’s reasons for creation. We cannot
form expectations about what God will do. No empirical discovery ever counts
against BT. Our dialectical context rules out sceptical BT, because divine inscrutabil-
ity undermines the case for BT. The arguments of part I suppose that humans can
understand divine purposes, albeit indirectly or imperfectly. (This is especially true if
we base human moral knowledge on insight into cosmic purpose.) We can reason-
ably ask whether God would prefer H or not-H.
Our moral commitments also come into play here. One obvious source of insight
into God’s motives is our own knowledge of objective values. And, as we’ll see, the
best arguments from scale explicitly rely on such values. Our BT cannot simply reject
such values, or human knowledge of them, out of hand! Our BT will speak, not of an
inscrutable God, but of a God whose purposes are more complex than atheism
allows. Our BT must therefore move beyond both finite theism and sceptical theism,
and seek a different reply to the argument from scale. We begin by asking why
Science World is thought to be a problem for BT.
7.2 What Exactly is Wrong with This Universe?

The atheist intuition is easy to grasp. (And it is one that AP shares.) Looking at the
size and scale of the universe, it is easy to conclude that human beings are insignifi-
cant. But what exactly is the argument? Popular debate suggests several different
‘benevolent theism’ over ‘classical theism’, I listed process theism as the main alternative to classical theism.
It may seem inconsistent that I now fail to pursue process theism further. My defence is simply lack of
space. A full defence of AP would need to engage this alternative theist tradition, especially in the context of
theodicy, where it offers distinct new resources to BT. However, I leave that task for another day. (I am
grateful to an anonymous reader for prompting me to clarify my attitude to process theology.)
6
For references, and further discussion, see chapter 8.
 THE CASE AGAINST BENEVOLENT THEISM
objections. One is that reality is fundamentally physical. The laws of physics do not
mention humans, and our existence is accidental. The presence of human beings is a
very uninteresting, incidental feature of the cosmos. Therefore, we cannot be part of
the cosmic purpose.
This atheist objection is not really about scale. BT has familiar replies, which our
project supports. If we accept a fine-tuning argument, then physics alone cannot
explain everything and the existence of rational beings is not an accident. Indeed, all
the arguments in part I reject the hegemony of global naturalism. Any objection to
BT that presupposes that reality is merely physical simply begs the question. This is
not the way to build a persuasive argument from scale.
Imagine waking up tomorrow to find a new scientific consensus. Scientists now
agree that we do inhabit Genesis World. The cosmos has a physical centre: our Earth.
It is only 6000 years old, and only a few thousand miles across. Who would celebrate
this conceptual revolution: atheism or BT?
Perhaps the thought behind the argument from scale is that this imaginary reversal
would support BT. Therefore, the actual shift from Genesis World to Science World
favours atheism. However, we need to tread carefully here. Intuitively, it may appear
that our imaginary revolution back to Genesis World would support BT. But this
does not necessarily support the argument from scale. Other arguments for (and
against) BT will also be affected. For one thing, intelligent design suddenly looks a
whole lot more plausible! Without very long periods of time, it is almost impossible
to explain the natural emergence of life. And, as we’ll see in chapter 8, the argument
from evil would also be weaker if we removed the eons of animal suffering before
human beings evolved. Perhaps these other impacts of the shift to Science World are
much more important, and drown out mere considerations of size. BT will hope to
address those other difficulties directly: replacing intelligent design with more robust
teleological arguments and constructing a theodicy that justifies pre-human animal
suffering. Whether or not these attempts succeed, scale itself is not an additional
threat to BT. After all, many BTs accept Science World. And many scientists accept
BT. These sociological facts at least suggest that scale alone is not an insuperable
difficulty for BT.
BT can object that the shift from Genesis World to Science World is irrelevant.
(See the epigraph from Hebblethwaite.) What matters is that the cosmos contains
rational creatures. God will create a suitable abode for rational beings like ourselves.
Beyond that, we have no reasonable expectations. In particular, we have no reason to
expect a universe of any size, either temporal or spatial. God has no preferences
regarding merely physical properties. Our commitment to non-natural moral facts
supports this BT reply. If the objective values that explain the existence of the cosmos
are non-physical, then why should a God (whether human-centred or not) have any
physical preferences?
BT also denies that God needs any reason to choose Science World over Genesis
World. BT accuses atheism of illicit anthropomorphism here. We expect an agent to
ARGUMENTS FROM SCALE 
choose the least costly route to her chosen goal. If you choose a large expensive house
over a small cheap one, we assume you need all that extra space. But efficiency has no
meaning for God, whose every possible creation is costless. Why create a vast
universe rather than a human-sized one? Well . . . why not? Out of the infinitely
many possible universes that include rational beings, God chose this one. No further
explanation is needed. Perhaps God’s decision to opt for Science World over Genesis
World is itself a brute fact.
Alternatively, BT can suggest possible divine reasons to prefer Science World.
Perhaps God wants a universe that is both governed by the most perfect physical laws
and peopled by rational beings. Or perhaps God wants rational beings to emerge
naturally. This combination requires not-H. (If God had no interest in humans, or if
miracles would do, then a much smaller universe would be fine.)
Indeed, the scale of the universe might even support BT.7 Atheist scientists often
claim that Science World is much more awe-inspiring than Genesis World. Science
reveals a more intricate, vast, and impressive world than anyone previously
imagined. But surely we should expect God to create a world more majestic than
any human imagining. BT predicts that science will continually amaze us. (For
instance, some philosophers have argued that BT favours the multi-verse hypothesis,
discussed in chapter 4, precisely because this would make God’s creation more
impressive.8) If science did restore the Genesis story, that would be evidence against
God. Scientific BTs might lose their faith when confronted with the smallness of
reality! (AP supports this divine preference for grandeur. I argue in section 7.5,
however, that it leads us away from BT.)
I conclude that size alone need not trouble BT. Our argument from scale must look
elsewhere. Another atheist thought is that, if a God who cared for human beings did
create such a vast universe, then God would give humanity a prominent role within
that universe.9 If human beings are so special, why are there so few of us? Why are we
all in one insignificant corner of the cosmos? Why is the universe so sparsely
inhabited? If human beings are so central to God’s plan, why aren’t we everywhere?
The problem is not that the universe is too big, but that we are too few. (Or perhaps,
as I argue below, that we are too small.) A benevolent God would create a universe
teeming with life.
The rest of this chapter explores this interpretation of the argument from scale.
This argument complements chapter 4. We need God to explain why the universe is
friendly to the emergence of life. But benevolent BT also generates more specific
7
While this argument for BT is seldom defended explicitly, it is implicit in the musings of BT scientists
on the grandeur of the universe revealed by science. Of course, grandeur is not merely a feature of size, but
nor are the two entirely unrelated.
8
Holder, ‘Fine-Tuning, Multiple Universes, and Theism’.
9
Here I draw loosely on Everitt, The Non-Existence of God, p. 221.
 THE CASE AGAINST BENEVOLENT THEISM
expectations about the nature and distribution of its inhabitants, and these do not
hold true in the actual universe.
This is where our moral commitments come to the fore, especially our views about
value. This new argument from scale is clearly evaluative. It rests, not on direct
insight into God’s motivations or reasons, but on the assumption that God will
respond to independent evaluative reasons. The argument has a familiar structure:
1. God will create the best possible world.
2. This is not the best possible world.
3. Therefore, God does not exist.
If this is not the best world, what exactly is wrong with it? I distinguish three
objections:
1. The Quantity Objection. There are too few rational beings.
2. The Variety Objection. This world lacks variety among rational beings.
3. The Quality Objection. This world lacks superior rational beings.
A better world would contain more beings like us, or many other species of rational
beings, or better rational beings. I shall argue that, of the three, the quality objection is
the most pressing. Human beings are simply too puny to be the main inhabitants of
such a vast cosmos.
The scenarios painted in these objections are all familiar from science fiction. They
pose a dilemma for BT. On the one hand, the atheist argues that, given BT’s value
commitments, we should expect to find a more populated universe, and our failure to
do so counts against BT. On the other hand, within popular science fiction, the
discovery of other rational civilizations (especially more advanced ones) often creates
a crisis of faith both for specific religions and for our general belief that human beings
matter. Both the absence of ETs and their discovery are problematic for BT.
As we saw in chapters 1 and 3, our utilitarian commitment to objective values
includes comparative evaluations of possible worlds. So this style of argument is
congenial to our project. But even the atheist moral nihilist can run the argument
using conditional evaluative claims: if human beings are valuable, then this is not the
best possible world. BT accepts the antecedent. So this argument has force against
BT, even if the atheist herself rejects all values. (As we would expect, AP will also
distance itself from evaluative claims about human beings in a similar way, because it
denies that human welfare matters to God.)
The real significance of the historical shift from Genesis World to Science World is
that it opens our eyes to the possibility of grander civilizations, larger populations, or
superior beings. Reflecting on the size of the universe, we can imagine greater rational
beings who would be at home in such a vast and complex universe, and who would
truly comprehend it. AP argues that, once we recognize how good the universe could
be, we see how capricious it is to expect to find ourselves as its most impressive
inhabitants.
ARGUMENTS FROM SCALE 
7.3 Would God Create the Best Possible World?

Our new schematic argument is valid. If God would choose the best possible world,
and this is not that world, then there is no God. BT must deny one of the two
premises. Either God will not create the best possible world, or this is the best possible
world. We examine each option in turn.
Our question is whether God must create the best. We briefly touched on this
question in chapters 1 and 3, where we saw that AP’s consequentialist commit-
ments and explanatory arguments work best if God does maximize value. We now
examine the issue more closely. This section explores two arguments that God need
not create the best possible world. Both are originally deployed against the argu-
ment from evil, and we will revisit them in chapter 8. Our present concerns are
whether they enable BT to avoid our new argument from scale, and also whether
they pose any threat to AP.
7.3.1 God is not a consequentialist

To claim that God will create the best possible world is to attribute consequentialist
morality to God. Non-consequentialists may simply reject this attribution. God is not
a consequentialist. God has no general reason to create the best possible world. God’s
goodness is perfect virtue. When creating, God will be perfectly kind, perfectly
generous, perfectly benevolent, perfectly just, and so on. But none of these divine
virtues requires the maximization of value.
My project explores the metaphysical implications of consequentialism. So I could
simply insist on divine consequentialism. And, at this stage of this book, we have
amassed several arguments that God is a consequentialist. Our explanatory argu-
ments for cosmic purpose in chapters 3 and 4 work best if we imagine that God
responds to cosmic value directly. God is motivated by facts about the comparative
value of possible worlds. And, in chapter 5, I defended AP as an interpretation of
perfect being theology, where God responds perfectly to reasons grounded in object-
ive values. Surely the perfect divine response to any value is to maximize it? For God
as for human agents, gratuitous satisficing is irrational.10 Finally, in part III I explore
the suggestion that any credible human-centred elements in human morality are best
understood as departures from a pure consequentialist morality, motivated by
human fallibility. The appropriate morality for a perfect creator is therefore act
consequentialism. Unless there is a positive reason not to create the best possible
world, God will do so. If BT relies on arguments from cosmic value to God, then it
should agree that God will create the best possible world. God will not create less than
the best.
10
On the irrationality of gratuitous (or ‘blatant’) human satisficing, see Mulgan, The Demands of
Consequentialism, ch. 3.
 THE CASE AGAINST BENEVOLENT THEISM
Even if they are not decisive, these consequentialist arguments do at least suggest
that if there is a best possible world that God can create, then God will create that
world unless God has some positive reason not to do so. If BT denies that God will
create the best, then it must provide that reason. The most prominent suggestion here
is from Robert Adams’s 1970 article ‘Must God Create the Best?’, which inaugurated
recent philosophical discussion of whether God must create the best.11
Granting for the sake of argument that there is a best possible world, Adams argues
that, if one accepts ‘ethical views typical of the Judeo-Christian religious tradition’,
then one can deny that a perfectly good creator must create ‘the very best world that
he could create’.12 Suppose God creates a world that is less than the best he could
have created. In particular, although God creates creatures with flourishing lives, God
does not create the best possible lives. Adams argues that, if we are to find fault with
God’s sup-optimal creation, it must either be because it wrongs some particular
individual creature, or because it manifests some defect of character. Adams then
argues against both disjuncts. First, prefiguring Parfit’s non-identity problem, Adams
argues that, if God gives each creature the best life it could have enjoyed, then God
does not wrong any individual creature by failing to create other possible creatures
who would have enjoyed better lives. Different possible creatures exist in different
possible worlds. God can be perfectly virtuous to each creature without creating the
best possible world, so long as each creature has the best life it could have had.13 We
return to non-identity in chapter 8, where I argue that, contrary to Adams’s hopes,
person-affecting BT finds it harder to defeat the argument from evil, because God has
not given each creature the best life it could have enjoyed. We set that worry aside for
now. Our present question is whether a person-affecting solution could at least avoid
our quantity, variety, or quality objections.
Adam then argues that God’s suboptimal creation does not manifest any defect of
character, because it results from God’s grace, which is ‘[o]ne important element in
the Judeo-Christian moral ideal’. Adams offers several definitions of grace:
• ‘grace may be defined as a disposition to love which is not dependent on the
merit of the person loved’.14
• ‘the gracious person sees what is valuable in the person he loves, and does not
worry about whether it is more or less valuable than what could be found in
someone else he might have loved’.15
• ‘grace is love that is not completely explained by the excellence of its object’.16
11
Adams, ‘Must God Create the Best?’ Adams’s paper has generated much discussion. My critique
draws particularly on Wielenberg, ‘A Morally Unsurpassable God Must Create the Best’; and Rowe, Can
God be Free?
12
Adams, ‘Must God Create the Best?’, p. 317.
13
Following Melinda Roberts, we could describe Adams’s God as a person-affecting consequentialist
(Roberts, ‘A New Way of Doing the Best We Can’).
14
Adams, ‘Must God Create the Best?’, p. 324. 15
16
Adams, Finite and Infinite Goods, p. 151.
ARGUMENTS FROM SCALE 
A gracious God will love less excellent creatures as much as more excellent ones.
Therefore, a gracious God might create (and then love) the former rather than the
latter. Adams’s grace-based argument is worth quoting at length:
A God who is gracious with respect to creatures might well choose to create and love less
excellent creatures than He could have chosen. This is not to suggest that grace in creation
consists in a preference for imperfection as such. God could have chosen to create the best of all
possible creatures, and still have been gracious in choosing them. God’s graciousness in
creation does not imply that the creatures He has chosen to create must be less excellent
than the best possible. It implies, rather, that even if they are the best possible creatures, that is
not the ground for His choosing them. And it implies that there is nothing in God’s nature or
character which would require Him to act on the principle of choosing the best possible
creatures to be the object of His creative powers.17
A gracious God, who creates without regard to excellence, can thus choose the less
excellent over the more excellent. Of course, God is not indifferent to the welfare of
creatures. Gracious love would not allow the creation of lives that are deficient or not
worth living. (We return to these conditions in chapter 8.) But God can freely create
flourishing lives that are less valuable than the best possible lives. God respects the
unique value of human beings by creating and loving them, not by seeking to
promote or maximize that value.
Adams’s virtue of grace is person-affecting. It involves obligations to, and responses
to the value of, specific individuals. Consequentialists typically reject person-affecting
morality. And Adams himself concedes that grace ‘is not part of everyone’s moral
ideal’.18 AP could stick to its consequentialist commitments, and insist that a morally
perfect being will not love in a gracious way—or indeed that God would have no
person-affecting virtues at all. However, the case for AP is obviously stronger if we
can accommodate this popular and appealing moral ideal, and argue instead that a
gracious person-affecting God would create the best.
Three terminological clarifications are in order before we proceed. First, the
literature often talks of God’s ‘defects’, or of blaming or criticizing God. I do not
find this language helpful here. (Who are we to criticize God?) So I will speak instead
of God’s reasons. Or, more straightforwardly, of what we might expect a morally
perfect being to do. Second, all our discussion of divine benevolence (and other
divine attitudes) towards humans is hypothetical. AP denies that God has person-
affecting concern for human beings, because God has no interest in humans at all.
This leaves entirely open the question whether God cares directly for other superior
beings who do matter. Perhaps God’s concern for those creatures who do matter is
person-affecting. Finally, to allow for this possibility and to avoid anthropocentric
bias, I shall speak of creature-affecting divine motivations rather than person-
affecting ones.
17
Adams, ‘Must God Create the Best?’, p. 324. 18
 THE CASE AGAINST BENEVOLENT THEISM
We must separate two claims about God’s gracious reasons:

1. If God is gracious, then God’s gracious love provides no reason for God to create
the most excellent possible creatures rather than less excellent creatures (who
are nonetheless happy and lovable).
2. If God is gracious, then no other aspect of God’s character provides any reason
for God to create the most excellent possible creatures rather than less excellent
creatures (who are nonetheless happy and lovable).
The first claim seems to follow directly from Adams’s definition of grace. (This is
just what it is to be a gracious God.) But Adams also needs the second claim, which
does not follow either from the definition of grace or from first claim. Nothing about
grace implies that no other divine disposition provides a reason to create the best
possible creatures. Erik Wielenberg puts the point very well:
[Adams’s argument from grace] goes wrong when it moves from the claim that (1) God
exhibits the virtue of grace and acts in a supererogatory fashion in actualizing the less valuable
world, to the claim that (2) God does not exhibit a lack of moral virtue in actualizing the less
valuable world. The latter claim does not follow from the former.19
Consequentialists will argue that, in addition to love for individual creatures

(whether gracious or not), God also has motivations related to the excellence of
states of affairs. Other things equal, a morally perfect being will aim to create the best
possible world. And grace does leave other things equal. Adams himself concedes
that a gracious God could create the most excellent creatures, and that grace gives
God no reason to prefer less excellent creatures. Therefore, if God has any preference
for better possible worlds, then God will create the best possible world. And that
world presumably contains the best possible creatures.20
Pure consequentialists, of course, argue that God’s only motivation is to produce
the best possible world, without any direct regard for the interests of creatures per se.
This is clearly incompatible with grace. But moderate consequentialists can accept
that gracious love is one divine motivation, while still insisting that God also has
some ‘dispositions to pursue intrinsically good states of affairs’.21 The latter disposi-
tions do not compete with grace, and therefore the consequentialist need not claim
that they are stronger. It is sufficient that God has some non-creature-affecting
disposition. Adams’s argument from grace only goes through if God is entirely
indifferent to the intrinsic values of states of affairs. Given the broadly consequen-
tialist commitments of our project, AP can reasonably set this extreme anti-
consequentialism aside, and conclude that Adams’s argument fails. Even a gracious
19
Wielenberg, ‘A Morally Unsurpassable God Must Create the Best’, p. 52.
20
In chapter 8, we’ll see that BT theodicy often denies that the best world available to God necessarily
contains the best possible creatures. But we can set this complication aside in our present context, as the
possibilities exploited by theodicy do not affect the arguments addressed in this chapter.
21
Wielenberg, ‘A Morally Unsurpassable God Must Create the Best’, p. 46.
ARGUMENTS FROM SCALE 
God would still create the best possible creatures. In our dialectical context, Adams’s
reply is not sufficient, on its own, to rebut the argument from scale.
Against Adams, William Rowe argues that God must display perfect benevolence,
and that perfect benevolence implies that God will create the best possible creatures
enjoying the best possible lives.22 Because we borrow perfect being theology, AP can
borrow Rowe’s argument. At the very least, AP can offer the following conditional
analogue of Rowe: if human beings have objective moral significance (or if human
lives have objective value or if human beings matter), then God will be perfectly
benevolent to each individual human being, and God will be perfectly just and
impartial between competing human claims, and God will perfectly promote the
value instantiated by flourishing human lives.23 AP then turns Rowe’s original
argument against its own antecedent. We do not find ourselves in the kind of
world a God with these human-friendly motivations would create. Therefore, con-
trary to our most deep-seated moral intuitions, human beings have no cosmic moral
significance.
AP’s critique of BT combines creature-affecting and impersonal consequentialist
themes. The arguments of part I, in combination with our utilitarian pre-commitments,
support impersonal consequentialism where God promotes objective value. This
element is present in BT itself, where God is pictured as recognizing the goodness of
creation. But BT also has strong creature-affecting commitments. God is said to love
each individual human being, and to want what is best for each of us. We can put these
two elements together. A perfectly benevolent agent will seek outcomes that are both
best overall and best for each individual. We return to these themes several times over
the next four chapters.
While AP can borrow from Rowe, it need not share all his commitments. As our
previous discussion shows, Adams’s argument fails if God has any non-creature-
affecting motivations. Perfect benevolence (or any other moral perfection) is not
required. As I argued earlier (especially in chapter 6), AP can acknowledge God’s
moral perfection. But this is not a necessary commitment of AP.
Recall our three scale-based objections: quantity, variety, and quality. Creature-
affecting moral theory can sidestep the quantity objection: God has no creature-affecting
reason to create more creatures. Here, AP must cite impersonal consequentialist
motives. But creature-affecting morality cannot enable BT to avoid the variety and
quality objections. Consider a world whose only inhabitants were happy slugs. The
atheist says: ‘God would not create such a world, because it is not the best possible.’
Creature-affecting BT replies that God has done what is best for each slug. This is
clearly inadequate. A God who valued slug lives would see greater value in human
22
Rowe, Can God be Free?, pp. 74–87. (For critical discussion of Rowe, see e.g., O’Connor, ‘Review of
Rowe Can God Be Free?’; and Hasker, ‘Can God Be Free? Rowe’s Dilemma for Theology’.) I am grateful to
an anonymous reader for alerting me to similarities between my argument and Rowe’s.
23
A further question, to which we return shortly, is whether perfect promotion implies maximization.
 THE CASE AGAINST BENEVOLENT THEISM
lives, and thus create human lives as well. (Or at least create a wider variety of insects!)
In the same way, AP argues that something of greater value is missing from any world
containing only human lives. Once we imagine superior (or very different) rational
beings, we see that a God who cared about the value we embody would create them as
well. God will not create lesser rational beings instead of greater ones. The perfect
efficiency of divine creation now counts against BT. If God could effortlessly create
better creatures who live happier and deeper lives, then why not create them?
Parfit introduces the useful notion of ‘wide’ creature-affecting benevolence, in
contrast to the ‘narrow’ interpretation we have been discussing hitherto.24 A wide
creature-affecting benevolent agent aims, not just to give each creature the best life
it can enjoy, but also to create creatures who can enjoy the best possible lives.
Unlike impersonal consequentialism, wide creature-affecting benevolence does not
permit trade-offs. One creature is never harmed to benefit another. Wide benevo-
lence is also more demanding than narrow benevolence, because it forbids the
creation of clearly suboptimal lives when much better lives were available (albeit to
different possible creatures). The wide interpretation thus appeals to those who are
uneasy about using non-identity to justify suboptimal creation, but who also reject
impersonal trade-offs. If we expect God to be perfectly benevolent, then a wide
creature-affecting interpretation seems most appropriate. And on that interpret-
ation Adams’s argument clearly fails.
7.3.2 There is no best possible world

Adams argues that God need not create the best possible world, even if there is such a
world. Richard Swinburne argues that God cannot create the best possible world,
because there is no such world.25 Every possible world is improvable. Swinburne’s
favourite example is that the addition of another flourishing human life would always
make things better. God can do whatever is possible. But creating the best possible
world is not possible. God cannot be faulted for failing to do what is not possible.
Atheists will object that even if God cannot create the best possible world, God will
not create a deficient world. And the quality objection claims that this world is
deficient. Consider again the world of slugs. When the atheist objects that this is
not the best possible world, Swinburne replies that any world is improvable, because
God could always add more slugs. Therefore, God cannot be faulted for creating
an improvable world. The atheist will remain unconvinced. The slug world is not
merely improvable, it is deficient—because it lacks both variety and superior lives.
Adding more slug lives could never compensate for the absence of humans. Similarly,
any world inhabited only by humans is deficient, because it lacks superior rational
beings. I take this first reply to be decisive, as was the analogous reply to Adams.
24
Parfit, Reasons and Persons, pp. 396–401.
25
Swinburne, Providence and the Problem of Evil. See also Hasker, ‘Can God Be Free?’
ARGUMENTS FROM SCALE 
(In chapter 8, I argue that this world is deficient in a second way, because a
benevolent God would also prevent great human suffering.)
While it cannot save BT, Swinburne’s argument is still worth exploring—not least
because it also raises difficulties for AP. To see why, we begin with another argument
against BT presented by William Rowe. Rowe turns Swinburne’s argument on its
head, arguing that BT collapses if there is no best possible world. Rowe begins with
the following moral principle:
Rowe’s principle B. ‘If an omniscient being creates a world when it could have created a better
world, then it is possible that there is a being morally better than it.’26
God is traditionally said to be morally unsurpassable: there is no possible being that is

morally better than God. Yet principle B implies that a morally unsurpassable being
cannot create any world if it could have created some better world. One moral virtue
is the disposition to respond to the intrinsic values of possible worlds. (Call this the
consequentialist virtue.) A morally unsurpassable creator will, perhaps alongside
other motivations, be disposed to optimally respond to those values. Suppose A is
an omnipotent being who creates a world (w1) when there is some other better world
(w2) that A could have created instead. Let B be some other omnipotent being who
creates w2 instead of w1. If A could have created w2, then B is possible. And if w2 is
better than w1, then B possesses the consequentialist virtue to a greater degree than
A. As there are no other morally relevant differences between A and B, it follows that
B is morally better than A. (We saw in our discussion of Adams that A is not, for
instance, any more gracious or loving than B.) Therefore, A is not morally unsur-
passable. If God is morally unsurpassable by definition, then A cannot be God. Our
only assumption about A is that A creates a world when A could have created a better
world. We can therefore conclude that every suboptimal creator is surpassable, and
no suboptimal creator can be God. If there is no best world, then a morally
unsurpassable God will not create any world. Yet the actual world exists. Therefore,
we have an inconsistent triad, consisting of Rowe’s principle B, the existence of a
morally unsurpassable God, and the claim that there is no best possible world. Rowe
argues that his principle is necessarily true. Therefore, BT must follow Leibniz and
agree that this is the best of all possible worlds.27
26
Rowe, Can God Be Free?, p. 4.
27
Hasker objects that Rowe criticizes God for failing to do what is impossible (Hasker, ‘Can God Be
Free?’). If there is no best world, then God cannot create the best—any more than God can do anything else
that is impossible. God’s omnipotence is limited by what is logically possible. Therefore, it is no defect in
God that God fails to create the best. Rowe replies that, while A cannot be faulted for failing to (impossibly)
create the best possible world, A does possess the defect of having failed to create w2—a particular world
that would have been better than the actual world (Rowe, Can God Be Free?, pp. 104–13). Perhaps it would
be unfair to ‘criticize’ A for this failure. But it is a defect—and therefore sufficient to show that A is not
morally unsurpassable. A fails to respond optimally to reasons based on objective values, and therefore B is
superior to A.
 THE CASE AGAINST BENEVOLENT THEISM
Our project supports Rowe’s principle B. In particular, the most controversial

element in Rowe’s use of his principle is the inclusion of the consequentialist virtue in
the specification of moral unsurpassability. As we saw in our discussion of Adams,
our pre-commitments and previous conclusions strongly support this consequen-
tialist element. God will respond to the intrinsic value of possible worlds.
The claim that God is morally unsurpassable is a traditional component of BT. But
it is also very helpful to AP, because it allows us to apply consequentialist morality to
God. Unless God responds optimally to the independent objective values of possible
worlds, we have no reason to believe that God’s existence helps to explain the
remarkable features of the actual world. God’s moral unsurpassability was also
central to our ontological argument in chapter 6, and our exploration of mysticism
in chapter 5. Rowe’s argument thus suggests that, if AP denies that God creates the
best, then it deprives itself of virtually all the arguments for cosmic purpose can-
vassed in part I.
Rowe’s argument thus poses a threat to AP. Whether or not BT can survive
without part I’s arguments, AP cannot. One possible solution is to distinguish my
particular cumulative case for AP from AP per se. AP can reject divine unsurpassa-
bility altogether, and instead offer itself as a rival interpretation of non-classical
theism. If Rowe demonstrates that an unsurpassable God could not be the creator
of this world, then theism must be radically reimagined. AP could then argue that its
reimagining is superior to BT’s. This concession to Rowe would mark a radical
departure from our project, however, and thus we seek alternative replies.
AP’s next option is to insist that, if God must create the best possible world, then
this is the best possible world. AP must then explain away Swinburne’s seemingly
compelling example. Doesn’t the addition of a flourishing human life always make
things better? And doesn’t it follow that any possible world, however wonderful, is
improvable?
If we want to insist that this is the best possible world, we have two possible
strategies. The simpler option is to deny the value of individual lives altogether. The
addition of flourishing human lives does not improve things because human lives do
not matter. For those who believe in cosmic purpose, this is the most theoretically
parsimonious option. If we reject human-centred values, then Swinburne’s familiar
examples provide no reason to conclude that there is no best possible world. (This
counter-intuitive move may seem undermotivated on its own. But in conjunction
with all the other arguments in parts I and II, it may well be sufficient to establish
non-human-centred cosmic values as the most plausible option.)
Unfortunately, this simple strategy has a significant cost for AP. Aside from its
counter-intuitiveness, it prevents AP from admitting the value of other rational
beings. (If there are beings who matter, then any possible world could be improved
by adding more of those beings.) Yet several of our arguments in chapters 4 and 7 rely
on the cosmic significance of superior beings. Ideally, AP seeks a response to Rowe
and Swinburne that does not rule out these arguments.
ARGUMENTS FROM SCALE 
Another way to insist that this is the best possible world is to deny that familiar
principles of aggregation apply to populations of superior beings. Human lives do
not matter at all, and superior rational beings matter in a non-aggregative way.
Here are some possibilities: perhaps finite populations of superior beings are
evaluated holistically according to their collective ability to respond to cosmic
purpose in ways that truly matter and additional numbers make no difference to
that ability; or perhaps the correct aggregative principle for superior beings is
some analogue of average utilitarianism rather than total utilitarianism; or per-
haps the actual population of superior beings is infinite, and therefore cannot be
improved by the additional of extra superior lives. This last possibility is especially
worthy of exploration, as an infinite population seems more fitting for an unsur-
passable God. While any finite population can be improved simply by adding
extra lives, an infinite population cannot.28 If this world contains a finite popu-
lation of rational beings, then it is improvable. But if an infinite population is
possible, then there are possible worlds that are not improvable.29 God will choose
one of those worlds. God will create infinitely many rational beings. After all,
surely an infinite collection of flourishing human lives is more worthy of a cosmic
creator? We would then be left with three options: either the actual universe
contains infinitely many rational beings, or individual rational beings have no
value, or there is no God.30
Unlike BT, AP can credibly claim that this is the best possible world. However, this
strategy involves additional controversial commitments. (In particular, it commits us
to specific aggregative claims about finite or infinite populations of superior beings.)
A more robust AP reply accepts that there is no best possible world, and then asks
how an unsurpassable God might respond. Rowe argues that, if there is no best
world, then God will not create at all. But this is too swift. The most that his argument
establishes, if it succeeds, is that God will not create any one possible world on its
own. But perhaps God instead creates all good worlds: all possible worlds that meet
some minimum threshold of divine choice-worthiness. If every good world is
28
On infinite utility, see Vallentyne and Kagan, ‘Infinite Value and Finitely Additive Value Theory’;
Mulgan, ‘Transcending the Infinite Utility Debate’.
29
I argue in the text that an infinite population cannot be improved simply by adding extra good lives.
However, it does not follow that no infinite population can ever be improved at all. On the contrary,
standard solutions to the infinite utility puzzles imply that worlds with infinite populations can be better or
worse even if they contain the same total utility. (To take one common example: if you double the
happiness of every member of an infinite population, then you have made things better. For other
examples, and further discussion, see Vallentyne and Kagan, ‘Infinite Value and Finitely Additive Value
Theory’.) Therefore, BT cannot appeal to infinite utility to salvage Swinburne’s argument. We return to
infinite utility in chapters 8, 10, and 13.
30
We saw on page 208 that, as with Rowe’s argument against Adams in section 7.3.1, AP does have
another option here. AP can deny that God is morally unsurpassable, just as it can deny the related claim
that God is morally perfect. But this would lead us back to finite theism, where BT and AP offer competing
imperfect creators. And, for this book, we have set finite theism aside.
 THE CASE AGAINST BENEVOLENT THEISM
improvable, then there are infinitely many good worlds. But that is no problem for an
omnipotent God, who could easily create an infinite array of worlds. Rowe’s principle
B is now irrelevant, because there is no better world that God fails to create.
The result is a theist analogue of Leslie’s axiarchic plenitude, which we discussed in
chapter 3. In this way, God can respond optimally to objective values even if there is
no best possible world. The arguments of part I thus survive intact. This reply is
available to either BT or AP, with the proviso that both must show that the actual
world is not deficient. (As I argued at the start of this section, and as we’ll see more
clearly in chapter 8, BT cannot do this.)
I conclude that Swinburne’s argument cannot save BT, but also that Rowe’s
argument poses no serious threat to AP.
7.4 BT vs AP If We Are Alone

I conclude that, given our commitments and the arguments of part I, Adams and
Swinburne’s arguments fail, at least in relation to the argument from scale. We can
reasonably expect God to create the best possible world. BT must now claim that the
actual world is the best one possible. At the very least, BT must show that this world
is not deficient. I consider two cases. First, I assume we are alone in the universe, and
ask whether BT or AP could explain this situation (section 7.4). I then ask whether
BT or AP could cope with scenarios where we are not alone (section 7.5). Section 7.6
asks which scenario is true.
We are now comparing BT and AP. But we must also remember our third
competitor: atheism. At this stage of our cumulative argument, we assume that
part I’s arguments have established a strong case against atheism. In particular,
atheism cannot satisfactorily explain the emergence of rational beings at all. Atheism
cannot overturn the conclusion of part I simply by providing a slightly better
explanation of specific facts about the cosmos and its inhabitants. However, if either
BT or AP proves unable to explain human uniqueness (or its negation) at all, then
this will be a significant victory for atheism. AP’s explanations don’t need to be
perfect, but they cannot be hopeless.
7.4.1 Can BT explain why we are alone?

Suppose we are alone in the universe. The cosmic population of rational beings is
finite, consisting solely of human beings. BT expects this result only if a finite
population is best. Some theories of aggregation imply that a very small finite
human population could be the best one possible. The most familiar of these theories
is average utilitarianism, where aggregate value is a function of global average well-
being. The addition of extra lives does not add value unless they are above the global
average.
ARGUMENTS FROM SCALE 
In the contemporary philosophical literature, average utilitarianism is a minority

option, subject to very serious general objections.31 Most consequentialists agree that,
if human flourishing has any value, additional flourishing lives do always add value.
This reply would thus tie BT to an eccentric position within consequentialist ethics.
Recall that, for God, all things are effortlessly possible. Instead of a finite population,
God could have created an infinite population of equally flourishing people. Does BT
really want to insist that all those additional lives would have added no aggregate
value at all?
Average utilitarianism also falls foul of our variety objection. Fans of variety will
argue that, even if more human lives would not improve things, the addition of
entirely new rational species would. BT must deny that this additional variety has any
value. A more serious difficulty is that BT still has no answer to our quality objection.
The quality objection can accept average utilitarianism, and then deny that this world
contains the highest possible average quality of life. We discuss the possibility of
happier human populations in chapter 8. Our focus here is on happier non-human
ones. Recall our example of slugs and humans. We can imagine beings as far above us
as we are above slugs. (Call them angels.) Angels’ lives would be much better than
ours. If we are alone, then God has created only humans and slugs. But average
utilitarianism calls for angels instead of either humans or slugs.
BT could deny that our replacement by angels would increase average happiness,
by insisting that each rational life is infinitely valuable. Every world containing
rational beings is infinitely valuable, and no inhabited world is better than any
other. But AP will simply reject this as the height of caprice. Why think that, from
the point of view of the universe, each human life is not only valuable, but infinitely
valuable? Can we really picture all the possible superior beings and still insist that
none is more valuable than us in any way? Presumably BT posits a gap between
God’s value and ours, so there must at least be some hierarchy of infinite values.32
BT cannot plausibly cite our independent knowledge of objective values to estab-
lish that a finite human population is best. BT could instead infer this evaluative
claim from the fact that the actual population is finite. If we are sufficiently confident
both that God would create the best possible world, and that we are alone, then our
independent evaluative judgements must give way. Strange as it may seem, the best
cosmic population is a finite human one. For instance, in the brief response to an
argument from scale that is our epigraph for this chapter, Hebblethwaite suggests
31
The classic discussion remains Parfit, Reasons and Persons, ch. 19. For further references, see Mulgan,
‘Consequentialism’. My own view is that the intuitions that seem to favour average utilitarianism over total
utilitarianism are better captured by wide creature-affecting benevolence. For instance, average utilitarians
worry that total utilitarianism sacrifices individual well-being for the sake of aggregate well-being. Wide
creature-affecting benevolence addresses this worry.
32
The infinite value view is more credible if each human life is infinitely long; or if we borrow the
Eastern Orthodox Christian idea of deification, where human beings participate post mortem in the glory
of God. We return to these possibilities in chapter 10.
 THE CASE AGAINST BENEVOLENT THEISM
that, if we are alone, then this fact supports BT.33 If the actual world contains only a
few billion rational beings, then the natural conclusion is that there is something very
cosmically special about those particular beings.
One interpretation of Hebblethwaite’s remarks is that the fact that we are alone
grounds a teleological argument for BT, by undermining the atheist’s brute fact
response. If life has only emerged on earth, then it is no longer plausible to treat
the emergence of life as statistically likely rather than miraculous. Even if our
uniqueness creates a prima facie problem for BT (because we would expect a more
populous cosmos), it still favours BT overall.
If we are very confident that we are alone, that God would create the best possible
world, and that God would care for humans, then our best option is to revise our
theory of aggregation. Although it agrees that God would choose the best, AP is
suspicious of our confidence in the other two judgements. We return to the question
of human uniqueness in section 7.6, where I argue that we should not be so sure that
we are alone—in part because belief in our uniqueness mutually supports belief in our
specialness. We concentrate here on our judgements about values. AP argues that,
faced with a choice, we should reject our human-centred values before our aggrega-
tive ones—because the former are much more likely to be contaminated by self-
aggrandizing caprice.
This AP position may seem particularly implausible. How can we retain any
utilitarian moral commitments once we deny that human beings have moral signifi-
cance? We here confront AP’s most radical departure from moral common sense: its
rejection of human-sized values. While a full reply must wait until part III, I offer
three preliminary remarks. First, AP will argue that part I provides independent
reason to believe that our intuitions about the structure of value are more solidly
grounded than our (human-centred) intuitions about the substance of value. The
former play an ineliminable role in the best explanation for fundamental facts such as
S, FL, and M; while the latter do not. Second, AP need not deny that some things are
good or bad for human beings. Our intuition that (e.g.) pain is bad is correct as a
claim about human well-being. It is only (!) false when extended to objective value or
divine motivation. Third, as I argue in part III, AP can even concede that human
well-being has a degree of objective value, so long as that value falls short of the
threshold required for cosmic significance. At this stage, these are no more than
promissory notes. But if AP can deliver on them, then it can reasonably defend its use
of structural intuitions about value.
BT cannot rely on independent knowledge of human values, nor can it reliably
infer human significance from our observation of the actual world. AP can reason-
ably deny that BT offers a compelling explanation of human uniqueness. BT’s final
defence is ad hominem. Even if BT’s explanation is not perfect, it is superior to
33
Hebblethwaite, In Defence of Christianity, pp. 11–12.
ARGUMENTS FROM SCALE 
anything that either atheism or AP can offer. To evaluate this reply, we must ask what
AP predicts.
7.4.2 Can AP explain why we are alone?
AP must deny Hebblethwaite’s conclusion. Human beings are not special, even if we
are alone in the universe. But AP must then account for the fact that we are alone.
This is less straightforward than it may seem.
AP per se has no direct implications regarding the size of the cosmos, or the
number of its inhabitants. But any interesting AP must tell us what is objectively
significant. And part I contained several suggestions: FL, M, I, and U. M is inde-
pendent of rational beings. But FL suggests that some kind of life has cosmic
significance, while I and U suggest what kind of life that is. Cosmic significance
resides in beings able to comprehend the cosmic purpose itself.
This raises two problems. First, AP is now vulnerable to the quantity, variety, and
quality objections. If life is so significant, why is there so little of it? If it is good that
the universe be understood, why is it understood so sporadically? Atheists can object
that, like BT, AP should expect the world to be teeming with life. A second problem is
that AP cannot borrow BT’s best reply to this atheist challenge. AP must deny that
human beings matter. But if rational life matters, and we are the only rational beings,
how can we not matter?
AP must argue that every feature of human beings that seems valuable is either (a)
a pale reflection of something greater that is genuinely valuable; or (b) not relevant to
cosmic value at all. Many of the arguments in part I begin from facts about human
beings that seem valuable. And the argument of part III will be that, in some ways,
human lives can resemble cosmic values. If AP always denies any connection between
humanity and cosmic values, then it will be very austere, unable to avail itself of many
positive arguments, and not a promising basis for human morality.
AP wants to say that some features of humanity are pale reflections of genuine
value. Obvious candidates are our rationality, freedom, creativity, mystical aware-
ness, and moral insight. The most natural interpretation is that this ‘something
greater’ is the analogous feature of some actual superior rational being. (For instance,
our mystical insight is a pale reflection of the mystical insight of angels or aliens who
do see God clearly.) AP works best if the cosmos includes superior rational beings.
But the most natural interpretation is not the only interpretation. AP is not com-
mitted to the existence of extraterrestrials. Human creativity, for instance, could
instead reflect either some inanimate feature of the cosmos or some feature of its
creator. (It is probably easier to model human creativity on a personal creator, rather
than on some inanimate cosmic purpose. If there are no other rational beings in the
cosmos, then AP thus becomes more sympathetic to theism rather than impersonal
alternatives such as Platonism or axiarchism.)
Recall the distinction, introduced in chapter 3, between our evidence of creation
and the purpose of creation. If we fail to discover superior beings, then our best
 THE CASE AGAINST BENEVOLENT THEISM
evidence that there is a cosmic purpose may be our own existence. But AP still insists
that it is self-aggrandizing to conclude that we are part of the cosmic purpose. When
Hebblethwaite infers our cosmic value from the ‘fact’ that we are alone, AP diagnoses
an illicit leap of faith. (Actually, as I argue in section 7.6, this is the second of two
illicit leaps here. The first is to believe that we are alone in the first place.)
AP can insist that it offers a credible explanation. This is sufficient if BT fails to
explain some other fact, such as evil or religious diversity. We must distinguish
between the strong BT claim that AP is incompatible with human uniqueness, and
the weaker BT claim that AP offers an inferior explanation. The latter would suffice if
we were evaluating the argument from scale on its own, but our dialectical context
requires the former. If we are alone, then this fact considered in isolation might place
AP at a disadvantage relative to either atheism or BT. But our case for AP is
cumulative, and both its rivals fare much worse elsewhere.
7.5 What If We Are Not Alone?

The significance of Science World is not its actuality, but its possibility—and the
other possibilities it enables us to imagine. Once we discover the physical size of the
universe, we can imagine a more populous cosmos, containing many different
rational beings. As we saw in chapter 4, recent cosmological speculation even raises
the possibility of an infinite cosmos that is home to infinitely many rational beings.
(In a multi-verse, we are almost certainly not alone!) We must therefore consider the
possibility that we are not alone. In this section, I ask what this possibility might
involve, whether it troubles BT, and how it supports AP.
If we are not alone, then the actual universe contains non-human rational beings.
These could be our distant evolutionary descendents, intelligent machines, or extra-
terrestrial life forms. The first two cases pose their own special issues, to which we
return in section 7.5.2. For now, however, I concentrate on the possibility of rational
ETs.
7.5.1 BT and ET
Suppose we share the cosmos with rational ETs. Should this worry BT? Kenneth
Miller argues that it should not. God would love non-human rational beings.34 God
creates a world where rational beings emerge. The result is human beings, and God is
happy. But God would have been equally happy with rational reptiles, silicon-based
life forms, or whatever. Miller’s ecumenical BT naturally lends itself to consequen-
tialist expansion. If non-human rational beings would be equally good, then God will
be even happier to get both human beings and other rational beings. Therefore God
will seek this result. God will prefer a universe that is home to infinitely many
34
Miller, Finding Darwin’s God, pp. 273–4. Miller’s original aim is to show that BT is not threatened by
the fact that the evolutionary emergence of homo sapiens was, ex ante, extremely unlikely.
ARGUMENTS FROM SCALE 
radically different species of rational being. BT religious texts seldom predict extra-
terrestrials.35 But BT need not fear them.36 Indeed, as I argued in section 7.2, perhaps
BT should welcome ETs as evidence that God has created the best possible world.
Unfortunately, things are not so simple. Some ETs are a threat to BT, as we shall now
see. The real challenge for BT is the quality objection. If the universe contains (perhaps
infinitely) many species of rational being, then it probably contains beings who are
much better than us. AP further insists that, whether or not they actually exist, superior
beings are possible. A cosmos with better beings is more valuable than one containing
only human beings (or other equally valuable creatures). If we instantiate value, then
the best possible world is one where we are not its main instantiations. BT leads us to
expect a world that prompts us to move beyond BT to AP.
Can BT accept superior rational beings? In one sense, the answer is obvious. BT
has long posited angels. BT can easily reply that, although we are not the most
valuable beings, we still matter to God. If atheism were its only opponent, BT
could dismiss the quality objection. It might be interesting to discover whether the
universe contains other (or better) rational beings, but that possibility is no threat to
BT. AP changes this. We now have two interpretations of the same (possible)
phenomenon. Suppose the universe contains, alongside several billion humans,
other rational beings who possess, to a much higher degree, the very features that
(according to BT) make humans valuable. BT judges that, in this world, we are (still)
cosmically valuable. We are part of God’s reason to create. AP concludes instead that
we do not matter, that we are no part of God’s reason to create, and that cosmic
significance is found (only) in higher rational beings or in other cosmic features.
Suppose the actual world contains three kinds of being: slugs, humans, and angels.
BT and AP both place a threshold on the scale of comparative value, such that only
creatures over that threshold matter. BT puts the threshold between slugs and
humans, while AP puts it between humans and angels.37 Once we admit superior
beings, BT has no explanatory advantage over AP. If God’s goals include superior
beings, then this is sufficient to explain the observable features of the universe.
Perhaps we are a mere by-product in a world designed to house greater rational
beings. We thus have no reason to posit divine concern for human beings.38
35
Scientology is an obvious counterexample, but it is not clear that it is either a form of BT or a religion.
36
Some specific doctrines within BT religions do have difficulty with ETs. (In so far as Christianity is
committed to the uniqueness of the Incarnation, it has long struggled with the possibility of non-human
souls—a theme that runs through much Roman Catholic science fiction in particular. See, e.g., Roberts, The
History of Science Fiction.) Another theme in religious science fiction is the danger that we will infect
extraterrestrials with our own sinful nature. (See, e.g., C. S. Lewis, ‘Religion and Rocketry’.)
37
An intermediate view would posit, not a threshold of cosmic significance between humans and
angels, but a lexical divide. We would still matter, but no finite number of human lives could outweigh the
value of a single angelic life. It is not clear whether this is a genuine form of BT, rather than a variant of
AP. We consider such intermediate positions in more detail in part III.
38
BT will reply by citing some divine plan that requires human beings as well as superior beings. (For
instance, in the traditional Christian picture, humans instantiate values or virtues that angels lack.) AP
must reject any such story as self-aggrandizing human caprice. (Do angels tell a similar tale about our
 THE CASE AGAINST BENEVOLENT THEISM
The existence of superior beings would thus cancel out any explanatory advantage
BT has gained over AP in part I. Our package of explanatory arguments (cosmo-
logical, teleological, scale) would leave BT and AP exactly on a par. While this may
seem a small victory for AP, it is actually a very significant one, because it leaves BT in
a very vulnerable position. If any other argument supports AP against BT, then AP
will have the upper hand. The rest of part II presents several such arguments. And
even without other arguments, AP can still insist that any remaining human prefer-
ence for BT over AP is solely caprice.
If we are not alone, then this fact supports AP over BT.
7.5.2 Between BT and AP

As we saw in the introduction, the boundary between BT and AP is vague. The
possibility of superior beings nicely illustrates that vagueness. Superior cosmically
significant beings could take many forms, including the following:
a. Aliens who have already emerged (via evolution or perhaps divine intervention)
elsewhere in the universe;
b. Aliens who will emerge elsewhere in the universe in the future;
c. Supernatural beings (angels, lesser deities, etc.);
d. Future inhabitants of Earth who are not directly descended from humans (i.e.
terrestrial evolution has cosmic significance, but we are one of its dead ends);
e. Future inhabitants of Earth who are the distant evolutionary descendents of
humans;
f. Superintelligent (or super-moral or super-mystical) machines or artificial
intelligences created by human beings (or descended from earlier AIs created
by us);39
g. Human beings who have turned themselves into superior beings using tech-
nology in this life (e.g. we are the superintelligences we create);
h. Human beings who have turned themselves into superior beings by supernat-
ural means in this life (e.g. we become super-aware or super-powerful through
mystical contemplation);
i. Human beings who are turned into superior beings by technological or super-
natural means outside their control in this life;
cosmic significance?) We return to this issue in chapter 8, because the most common suggestion here is that
humans possess a unique freedom to choose between good and evil.
Another option for BT is to borrow an alternative formulation of the argument from scale itself, and
attribute our existence to God’s preference for variety. AP would then reply that, given the facts about evil
explored in chapter 8, God’s concern for variety is best interpreted as an impersonal concern for a general
structural feature of states of affairs, rather than any concern for the fate of specific individuals.
39
Artificial superior beings are also possible under a, b, c, or d. They could exist on other worlds, or as
creations of later naturally emerging inhabitants of Earth. AIs might even be built by angels. But it is only
AIs produced by humans that raise special issues for AP.
ARGUMENTS FROM SCALE 
j. Human beings who become superior beings in the next life, whether by their
own efforts or not, whether by technological means or supernatural ones. (We
return to this possibility at length in chapter 10.)
Suppose God cares for superior beings but not for (non-superior) human beings.
Which of these categories are consistent with AP, and which mark a shift to BT? (a)
through (d) are clearly AP. If God cares only for superior beings that are unconnected
to us, then BT is false. Equally clearly, (g) through (j) fall under BT, even if the
resulting theology would be extremely eccentric and heretical within any actual BT
religion. In all these cases, God cares for at least some human beings, even if only for
those who become superior.
The tricky cases are (e) and (f), where humans are steps (either knowingly or
unknowingly) along the road to cosmically significant beings. This is a grey area, but
I am inclined to think that AP captures these cases better than BT does. God cares
about our impact, but not about us. (Just as the animal-indifferent gardener cares
about the impact of cats and birds and bees, but doesn’t consider their welfare.) We
matter (only) in the way that our evolutionary ancestors matter on those familiar
anthropocentric theist stories where they exist merely as a prelude to our arrival.
This intermediate form of AP has distinctive features. In particular, it is easier to
find human meaning here, as we can do things that matter cosmically. (Namely: we
can play our role in facilitating the emergence of superior beings.) Studying the
cosmic purpose could thus play a key role in AI research, as it may tell us what kind
of superior beings we should strive to create. (Should we aim at superintelligence,
moral superiority, superior value-perception, superhuman mystical awareness,
or . . . ?)40
7.6 Are We Alone?

The overall conclusion of sections 7.4 and 7.5 is rather surprising. While BT and AP
both struggle to explain the ‘fact’ that we are alone, AP is clearly more comfortable if
we are not. The possibility of superior beings, in particular, favours AP over BT.
BT thus prefers a universe where we are alone. But are we? Or does the universe
contain (perhaps infinitely many) other rational beings? (And, if so, are some of them
40
Conversely, as I argue elsewhere, once we recognize the possibility of non-human-centred cosmic
purpose and non-human-friendly Gods, the very idea of superintelligence becomes much more worrying
from a human point of view (Mulgan, ‘Moral Philosophy, Superintelligence, and the Singularity’). The
popular literature on superintelligence typically assumes either that superintelligent AIs will naturally care
for human beings, or that humans can engineer a human-friendly superintelligence. These presumptions
underlie our optimism about technological futures. However, they only hold true if we beg several
contested philosophical questions, especially in meta-ethics, moral psychology, and philosophy of religion.
In particular, any presumption of super-friendliness collapses if we take AP seriously! I hope to explore
these connections at much greater length in the future.
 THE CASE AGAINST BENEVOLENT THEISM
significantly more impressive than us?) These are empirical questions that we cannot
yet answer.
As traditionally interpreted, BT says that we should expect to be alone. AP replies
that we should expect to find other (and better) instances of value: either greater
rational beings or cosmic features that dwarf humanity. Because they predict differ-
ent things, there could be evidence that favoured BT over AP, or vice versa. Perhaps
evidence of extraterrestrial life will one day decide between AP and BT.
It therefore does matter whether there are other rational beings. Not just because it
would be nice to know, but in the deeper sense that our considered moral judgements
may be radically altered. Our view of our place in the cosmos is at stake. We must add
the search for extraterrestrial life to our list of hitherto unexpected sources of moral
knowledge (alongside the explorations of cosmology and mysticism we added in
chapters 3, 4, and 5).
Several hypotheses are live options: we are alone; we are among peers; we share the
cosmos with our betters. Which is correct? Our current information does not settle
the question of extraterrestrial rational life. Hebblethwaite argues that our current
evidence suggests that we are alone.41 He dismisses belief in extraterrestrials as a
‘secular prejudice’. My own contrary view is that, given the complexities inherent in
the question, and especially the possibility that we inhabit a multi-verse or other
infinite cosmos, it is simply too soon to say where the evidence points. For us today,
the question of extraterrestrial life remains open.
This is not a standard case of religious ambiguity, because evidence could become
available. We could encounter other rational beings, and become convinced that their
form of life is more valuable. On the other hand, some new scientific theory might
predict that life is vanishingly unlikely to emerge elsewhere in the universe. (A
decisive refutation of the multi-verse would obviously help here.42)
Evidence could be forthcoming one day. But we do not possess that evidence. Our
present situation requires a leap of faith. And here AP argues that the modest option
41
Hebblethwaite, In Defence of Christianity, pp. 11–12.
42
Another possibility is plenitude, introduced in chapter 3. If there are many different worlds, then the
total existing population (across all existing worlds) is very large and varied, even if we are alone in ‘our’
actual world. If (as in Leslie’s version), every worthwhile possible world exists, then that will include both
worlds with infinite populations and worlds with non-human rational populations. Plenitude is thus not an
alternative to infinite or non-human populations, but rather an instance of them. I argued in section 7.3.2
that plenitude may be the most elegant theist response to the discovery that there is no best possible world.
In our present context, plenitude has the additional advantage that, unlike claims about our world, it is very
hard to refute empirically. This is especially relevant if it is capricious to believe we are alone. Even if we
conclude that extraterrestrials are absent from the actual world, the logical coherence of Leslie’s plenitude
means that we can never rule out the possibility that we are not alone. We thus face an empirical question
that can only be decisively settled in one direction. Future evidence could persuade us that we are not alone,
but nothing should persuade us that we are. Taking Leslie’s plenitude seriously would thus greatly
strengthen the argument from scale. Any attempt to invoke plenitude to dissolve other objections to
BT—such as the argument from evil—would thus be a double-edged sword. (We return to connections
between plenitude and evil in chapter 8.)
ARGUMENTS FROM SCALE 
is to proceed on the assumption that, somewhere in the universe, there is a value

greater than anything found in human life. It would be very surprising if we turned
out to be the most splendid beings in the cosmos. Given the size of the cosmos, we
will never be in a position to know that we are unsurpassed. In this uncertain
situation, it is epistemic caprice to proceed on the assumption that we are alone,
and therefore cosmically special.
Extraterrestrials could also provide new evidence relevant to other arguments.
They might support AP’s case against BT in the rest of part II. One obvious
possibility is that extraterrestrials with radically different religious, spiritual, or
mystical practices could reinforce our argument from diversity in chapter 9. On
the other hand, if it turned out that all other rational beings converge on one
particular world view, and that metaphysical disagreement is found only on Earth,
then this would lend support to the cosmically dominant world view—whatever it
might be.
Extraterrestrials might also offer direct insight into cosmic purpose itself, or into
cosmic values or morality. Their attitude to us could be especially revealing, and
potentially unsettling. Superior beings will lack our bias towards ourselves. So their
estimation of our worth or cosmic significance might be more reliable than our own.
If superior beings systematically do not care for us, that would support AP.
What have we learned from our tentative speculations about our cosmos and its
inhabitants? Perhaps the most significant lessons came in section 7.3, when we
concluded that AP can reasonably insist that God will choose the best possible
world. This conclusion is significant throughout part II. Our most surprising lesson
is that discovering whether we are alone should be a central concern of both
metaphysics and moral philosophy.
The central lesson of this chapter, however, is that we should be wary of any
argument for BT based on seemingly neutral empirical assumptions about the
cosmos—such as human uniqueness—that may themselves reflect our human-
centred biases. This is not a new lesson, but it is an important one.
I conclude that, while certainly not decisive, the (not entirely well-named) argu-
ment from scale does lend some support to AP over BT. And, as we shall now see,
other facts about the cosmos and its inhabitants are much more troubling for BT.
8
Arguments from Evil
Science has made only two contributions to the data of natural theology: (1) the
discovery that there were sentient animals long before there were rational
animals; and (2) the discovery that the physical world does not have an infinite
past.
[Peter van Inwagen, The Problem of Evil, p. 112]
. . . the only defence that has any hope of succeeding . . . is the so-called free-will
defence.
[Peter van Inwagen, The Problem of Evil, p. 70]
God did in fact create significantly free creatures; but some of them went wrong
in the exercise of their freedom: this is the source of moral evil. The fact that these
creatures sometimes go wrong, however, counts neither against God’s omnipo-
tence nor against his goodness; for he could have forestalled the occurrence of
moral evil only by excising the possibility of moral good.
[Alvin Plantinga, The Nature of Necessity, pp. 166–7]
. . . there are facts about evil that are logically inconsistent with the perfect
goodness of an omnipotent and omniscient creator of the world, at least
relative to certain value commitments which a reasonable person may have.
[John Bishop and Ken Perszyk, ‘The Normatively
Relativized Logical Argument from Evil’, p. 110]
8.1 AP and the Argument from Evil

The argument from evil is my central argument for AP against BT. This argument is
very familiar and many atheists cite evil as their principal reason for rejecting BT. In
this introductory section, I briefly outline the argument from evil, and explain what is
distinctive about my approach.1
Here are three classic example of ‘evil’:2
1
From the vast literature on evil, I have learnt most from van Inwagen, The Problem of Evil; Adams,
Horrendous Evils and the Goodness of God; Bishop and Perszyk, ‘The Normatively Relativized Argument
from Evil’; Oppy, ‘Arguments from Moral Evil’; and Dougherty, The Problem of Animal Pain.
2
The fawn is due to Rowe. (See, e.g., Rowe, ‘The Problem of Evil’.)
ARGUMENTS FROM EVIL 
1. Animal suffering. A fawn burns to death in a forest fire, perhaps long before any
humans exist.
2. Natural evil. A human child is born with an incurable disease that gives her a
brief life containing nothing but excruciating agony.
3. Moral evil. An innocent person is slowly tortured to death.
Other examples involve similar events on a much larger scale: natural disaster, famine,
war, genocide. At this stage, we need only agree that these are bad things in themselves;
and that it would be better if they did not occur other things being equal.
AP does not predict evil, but it easily accommodates it. The sufferings of humans
(and other animals) do not matter from the point of view of the universe. The
universe is not designed by someone who cares for individual humans or animals.
Therefore, it is no surprise that animals and humans suffer. Of course, AP must then
explain how an inhuman cosmic purpose gives rise to a recognizably human mor-
ality. This is the challenge of part III. Whether or not that challenge can be met, the
existence of evil is not a separate problem for AP.
By contrast, evil is a very serious explanatory problem for BT, whose God is
benevolent and wants to avoid human and animal suffering. The existence of evil
is a prima facie difficulty for BT. This chapter argues that both the arguments of part
I and the moral commitments of our project strengthen the traditional argument
from evil.
My goal in part I was to convince atheist moral philosophers to take seriously a
collection of arguments they tend to ignore. While each argument has generated a
vast philosophical literature, I did not assume that readers were familiar with
contemporary debates, nor that they had given the issue much independent thought.
As a result, my discussion was often quite introductory. My exploration of the
argument from scale in chapter 7 also did not presuppose acquaintance with the
philosophical literature—because there isn’t any!
The present chapter proceeds quite differently. Atheists can ignore cosmology,
fine-tuning, and mysticism. I have argued that they shouldn’t. But they can—both
professionally and existentially. Asked why the universe exists, the professional
secular moral philosopher can reply, without shame or embarrassment: ‘How should
I know? I’ve never really thought about it. No one expects moral philosophers to
think about philosophy of religion.’
It is hard to imagine any thoughtful theist, let alone a professional philosopher,
responding in similar vein about evil. (‘Why does a loving God allow so much
suffering? You know, I’ve never thought about that.’) Even those who don’t think
evil presents any philosophical problem for theism, even those who deny there is
even a prima facie defeater here, will still have something to say about why God does
or might or could permit widespread human and animal agony. And no one would
seriously think you could pursue theist ethics without having some views here. The
objections raised in this chapter will not, on the whole, be news to theists.
 THE CASE AGAINST BENEVOLENT THEISM
Atheists are often much less familiar with contemporary philosophical (and
theological) debates about evil than theists. But that doesn’t stop them having
views about those debates. Many ‘secular’ moral philosophers find it obvious that
evil rules out a benevolent God. The broad outlines of the position I defend in this
chapter will be familiar, even where the opposing arguments are not. Indeed, in the
utilitarian tradition since J. S. Mill, it has become a commonplace that utilitarians
must find the argument from evil compelling.
Many atheist readers will thus be tempted to skip this chapter, taking its conclu-
sions as read. On the other hand, many theists will feel they have heard it all before,
and find it no more convincing this time around. Where, they will naturally object, is
a careful rebuttal of the theodicies offered by Marilyn Adams, Alvin Plantinga, Peter
van Inwagen, Richard Swinburne, or John Hick, to pick only a small sample from the
more analytic end of the spectrum?
My defence is that any adequate treatment of contemporary theodicy would require
at least another book as long as this one. My aim is not to survey familiar ground, but to
offer a specific perspective on evil, shaped by my prior utilitarian commitments and
also by the conclusions we reached in part I. Here I follow the lead of much Christian
theodicy, which seeks to develop its own story about the cosmic role of evil, rather than
concerning itself with direct rebuttal of specific atheist objections.
My argument from evil is distinctive in several respects. The most obvious is that
I make no attempt at moral neutrality. Arguments from evil typically seek to avoid
moral controversy. The argument is presented as morally neutral. If either side makes
controversial moral claims, this is seen as objectionable. But moral philosophers need
not shy away from moral controversy, and our project is unashamedly morally
controversial. This chapter will not persuade those committed to BT, any more
than part I will persuade the committed atheist. But it does show that, from one
respectable moral perspective, AP offers a better account of this world’s evils than
anything available to BT.
The literature divides arguments from evil into logical and evidential (or prob-
abilistic). The former concludes that evil is logically incompatible with the existence
of God. Evil thus demonstrates God’s non-existence. Evidential arguments conclude
that evil makes it improbable that God exists (P-inductive), or at least that evil
reduces the probability that God exists (C-inductive).
The logical argument is very ambitious, and our methodological pluralism cau-
tions against expecting something so powerful. But my own argument is not strictly
inductive or probabilistic either. In terms of the existing literature, it might better be
characterized as a normatively relativized logical argument.3 Given our moral com-
mitments, the existence of a benevolent God is inconsistent with the amount and
distribution of evil in this world. Because our moral commitments are themselves
3
I borrow this term from Bishop and Perszyk, ‘The Normatively Relativized Logical Argument from
Evil’.
ARGUMENTS FROM EVIL 
controversial and unproven, the resulting argument against BT is not nearly as

decisive as a purely logical argument would be.
Our argument from evil comes late in our cumulative case for AP. We can therefore
presuppose the conclusions of earlier arguments, as well as our prior utilitarian pre-
commitments. As we shall see, these presuppositions rule out many popular BT moves.
Our commitments influence the argument from evil in several ways. Most obviously,
our utilitarian commitments regarding the significance of suffering and the compara-
tive value of suffering and freedom strengthen both arguments based on animal
suffering and atheist replies to BT’s ‘freewill theodicy’. More sophisticated forms of
liberal consequentialism also play a supporting role, allowing us to argue that, even if
freedom is valuable, it lacks the specific kind of value that BT needs. A third distinct-
ively consequentialist comparative evaluative claim also does considerable work in this
chapter. Following Thomas Hurka, I argue that the value of human responses to
suffering cannot outweigh the disvalue of the suffering itself.
Distinctively consequentialist values thus constrain BT’s responses to evil. Conse-
quentialist responses to value also play a role. In part I and chapter 7, I argued that a
morally perfect God will respond optimally to comparative evaluations of possible
worlds. If there is a best possible world, God will create it. (If there is no best possible
world, God will create all good worlds.) This prior conclusion constrains BT’s ability
to explain evil. On the other hand, I do not want my case against BT to rest entirely
on this controversial claim about divine motivation. Therefore, I attempt to construct
an argument from evil that is compatible with person-affecting accounts of God’s
reasons as well as impersonal consequentialist ones.
Another distinctive feature of our dialectical context is that our BT seeks argu-
ments to persuade third parties. This rules out two modest BT moves: defence and
scepticism.4 BT replies to arguments from evil with both theodicy and defence.
A theodicy is an account of God’s actual reasons for permitting evil; while a defence
offers God’s possible reasons. The same story could be either theodicy (‘This is what
happened’) or defence (‘This might have happened’). If BT seeks only to demonstrate
the logical possibility of an evil-permitting God, then a defence will suffice. If it aims
to show that our total evidence (including evil) makes it probable that God exists,
then BT needs theodicy. Many contemporary BT philosophers offer only a defence.
But our BT must provide a theodicy. Our dialectical situation forces BT to become
more ambitious.
Sceptical BT sidesteps the need for theodicy. No argument from evil has any force,
because we cannot fathom God’s reasons for creation.5 As we saw in chapter 7, our
4
As in chapter 7, we also sidestep a third modest option: finite BT. We assume that BT defends the
traditional Omni-God, who is omnipotent, omniscient, and omni-benevolent.
5
Bergmann, ‘Sceptical Theism and Rowe’s New Evidential Argument from Evil’; Bergmann and Rea,
‘In Defence of Sceptical Theism’. For critique, see Almeida and Oppy, ‘Sceptical Theism and Evidential
Arguments from Evil’.
 THE CASE AGAINST BENEVOLENT THEISM
BT cannot be sceptical. If we have no idea about God’s reasons, then we cannot

borrow arguments for God, especially teleological and cosmological arguments based
on probabilities. Sceptical BT rules out any philosophical defence of BT. Our project
compares the philosophical case for AP with the philosophical case for BT, and the
latter cannot avoid claims about God’s purposes.
While it rejects full-blown scepticism, AP agrees that speculation about God’s
possible purposes is fraught with difficulty. Our AP argument from evil thus seeks to
avoid direct claims about God’s purposes. It appeals instead to claims about human
well-being—especially regarding the comparative importance of suffering and
freedom—and claims about possibility. While still controversial, these are less ambi-
tious than any direct appeal to God’s purposes.
AP denies that human well-being has any cosmic value. As in chapter 7, we
therefore proceed by reductio. We assume that God does value human well-being,
and then conclude that a God who valued human life could not have created this
world. AP agrees that we know very little about God’s purposes. But it argues that we
can reasonably conclude that God does not care for human beings, because our
knowledge of human well-being tells us that this world is not well designed to foster
flourishing human lives.
One of my goals in this chapter is to bring out the fact that BT responses to evil
often rely on controversial metaphysical claims. Faced with philosophical contro-
versy about human nature, BT must embrace our metaphysical specialness: in a
universe of contingency, impermanence, and decay, we are immortal; in a material
world otherwise governed by physical laws, we possess a freedom that is incompatible
with determinism; and so on. BT’s metaphysical claims are not impossible or absurd.
But they require a leap of faith that AP avoids.
Our aversion to caprice obviously plays a role here. (It is self-aggrandizing to
assume our metaphysical specialness unless we have very good evidence.) But AP has
other arguments, and need not rely solely (or even primarily) on caprice. Human
metaphysical specialness is an unnecessary complication. AP offers a simpler, less
controversial picture of the universe. (One striking metaphysical commitment of BT
is that all theodicies invoke an afterlife. Chapters 8 and 10 are inextricably linked.
AP’s argument against theodicy is based, in large part, on rejecting specific post-
mortem claims.)
A final introductory topic is modal scepticism. The argument from evil posits better
possible worlds that a perfectly benevolent God would create instead of this one.
Modal sceptics, such as Peter van Inwagen, deny that those better worlds are
possible.6 A world must be described in much greater detail before its possibility
can be established. (Van Inwagen cites the actual laws of physics to illustrate the level
of detail required.)
6
Van Inwagen, The Problem of Evil, p. 117. For a reply, see Trakakis, ‘Is Theism Capable of Accounting
for Any Natural Evil at All?’, pp. 55–6.
ARGUMENTS FROM EVIL 
In previous chapters, our response to modal disagreement has been to presume

possibility. If a possible world can be described without contradiction, the onus is on
the person who denies possibility. Our dialectical context is crucial here. Suppose an
argument from evil rests on a controversial modal claim (e.g. ‘God could have created
a world like this one, but with less suffering’). For a defence, BT only claims that is it
possible that p is not possible. To defend a theodicy, BT must show that it is probable
that p is not possible. A presumption of possibility favours defence, but it counts
against theodicy.7 The relevant notion of possibility here is epistemic. Van Inwagen’s
favoured phrase is: ‘for all anyone knows’. This suffices for defence. For theodicy, it
does not. We must, somehow, try to determine the balance of epistemic probabilities.
And a presumption of possibility seems a reasonable place to start. (The most
significant modal disagreements in this chapter concern human freedom. In this
particular case, as we’ll see in sections 8.3 to 8.9, our moral commitments count
against BT.)
I present two distinct arguments based on two different kinds of evil: the suffering
of animals and the horrendous moral evils inflicted by human beings on one another.
The two arguments are separate but complementary. Each stands or falls on its own,
and either would suffice to refute BT. However, the two arguments explore overlap-
ping themes, and each appeals to a distinct set of utilitarian concerns. In particular,
because my argument from animal suffering is especially designed to persuade
hedonists who privilege suffering over freedom, my argument from moral evil
often adopts a more complicated liberal consequentialism where freedom is accorded
greater value. To complicate matters further, I also present hedonist and liberal
alternatives within my argument from moral evil. The overall argument from evil
is thus multiply disjunctive. I don’t expect anyone to be persuaded by all the
arguments that follow, but I do expect anyone with some utilitarian sympathies to
be persuaded by at least one of them!
8.2 Animal Suffering

Modern biology has taught us the sheer scale of suffering in the animal world, much
of it long before the arrival of human beings.8 A world with so much suffering seems
very suboptimal. BT must either deny that animal suffering is evil; deny that God
must choose the best possible world; or argue that the evil of animal suffering is
outweighed by some greater good.
Descartes’s notorious denial that animals suffer is no longer empirically credible.9
Some animals are sentient. And agony is paradigmatically bad for any sentient
7
Van Inwagen himself deploys modal scepticism in the context of a defence.
8
Van Inwagen, The Problem of Evil, p. 112; Everitt, The Non-Existence of God, pp. 227–8.
9
For a thorough discussion of animal consciousness in relation to evil, see Dougherty, The Problem of
Animal Pain, pp. 77–95.
 THE CASE AGAINST BENEVOLENT THEISM
individual. In principle, BT could simply deny that God cares about animals. Perhaps
God is benevolent only to humans. But few actual BTs take this line today. And two
of our utilitarian commitments oppose it. First, utilitarians emphasize the signifi-
cance of pain in human well-being. Utilitarians debate what to add to hedonism. But
all agree that agony is very intrinsically bad. If our agony is so bad, it is hard to see
how animal agony could count for nothing. Second, utilitarian animal rights advo-
cates highlight similarities between humans and other animals. It is anthropocentric
caprice to restrict one’s benevolence to human sentience. Such speciesism is
unworthy of God. A consistent benevolent God will care about animals. The vast
animal agony in our world is a very great evil.
8.2.1 Creature-affecting benevolence

The argument from evil assumes that God must create the best possible world, and
therefore that atheists need only discover some defect in the actual world. BT often
denies that God must create the best. Chapter 7 introduced both Adams’s argument
that God’s benevolence is person-affecting, and Swinburne’s argument that God
could not choose the best possible world because there is no such world. We rejected
Swinburne’s argument in chapter 7, and evil offers him no new resources.10 We
therefore leave Swinburne behind.11 (Except that section 8.3 introduces a related
objection, based on Plantinga’s claim that human freedom makes it impossible for
God to choose the best possible world even where it exists.)
The present section thus continues our discussion of Adams’s claim that God
might choose not to create the best possible world. As we saw in chapter 7, much
turns on our account of God’s perfect benevolence. I shall consider three alternatives:
narrow creature-affecting, wide creature-affecting, and impersonal. (As in chapter 7,
we replace ‘person-affecting’ with ‘creature-affecting’ to cover non-human animals.)
I agued in chapter 7 that perfect benevolence may require all three. God must do what
is best for each creature, and create the happiest creature, and maximize value overall.
However, my overall argument in this chapter does not presuppose this controversial
tripartite view. Instead, I argue that BT fails under any credible interpretation of
divine benevolence.
Under a narrow creature-affecting interpretation, a benevolent God cares about
each individual, not about maximizing overall well-being. God aims to give each
sentient creature the best life it could possibly enjoy. Can this approach dissolve the
10
In chapter 7, I argued that AP’s best response to Swinburne may involve plenitude: if there is no best
possible world, then God creates all good worlds. However, we can set that complication aside in this
chapter. Plenitude cannot save BT, because God will not create deficient worlds, and any benevolent God
will regard worlds containing undeserved and unnecessary horrendous evil as deficient.
11
Another similar BT move is van Inwagen’s argument that there is no least number of acceptable evils,
and therefore God cannot be faulted for creating a world with some unnecessary evils. We briefly consider
van Inwagen’s argument in chapter 10, where I argue that it fails for much the same reasons as
Swinburne’s.
ARGUMENTS FROM EVIL 
argument from evil? Does this world provide the best possible life for each living
creature?
This is a very demanding standard, especially for an omnipotent God whose ability
to improve creatures’ lives is only limited by logical possibility. Creature-affecting BT
can appeal to Parfit’s non-identity cases.12 A benevolent agent could give a creature a
life containing agony, so long as both (a) that life is worth living; and (b) the creature
would not otherwise have existed at all. Consider a disease that is essentially linked to
a creature’s identity. Or suppose it is essential that this creature evolved via a process
of natural selection involving considerable pain and suffering. In these cases, a
creature-affecting God could create a world with great agony, and still give each
creature its best possible life.
BT would be unwise to rely too much on non-identity. For one thing, non-identity
is much less likely for God’s choices than for human ones. Non-identity arises when,
as a matter of fact, I would not have existed if things had been different. If my parents
had never met, then I would not exist. But it is logically possible that I exist even
though my parents never meet. (Perhaps my genetic material is brought together in a
laboratory or by magic.) We cannot realize these possibilities. But God could.
How much actual animal agony is so closely tied to individual identity that even
God could not separate them? Some genetic disorders may be identity-determining.
This animal could only escape this disease by possessing a genetic make up that
would make it a different individual. But many diseases are not genetic in this way.
Those diseases are avoidable. The connection between individual identity and evo-
lutionary history is even more tenuous. As a matter of fact, actual evolutionary
history was necessary for the existence of this individual animal. If evolution had
gone slightly differently, there would be no mammals, no dogs, and certainly no Fido.
But couldn’t God have created Fido using a slightly different, less agony-filled,
evolutionary process?
Another difficulty is that the conditions for individual animal identity across
possible worlds are controversial. What makes Fido Fido? Do we have the same
animal if we vary her genetic make up, or her evolutionary history, or place her in an
artificially pain-free environment? Non-identity claims are too controversial a foun-
dation for theodicy.13
Non-identity may excuse some animal agony. But it cannot explain all of it. It is
vanishingly unlikely that every animal who has ever lived enjoyed the best life it could
12
Parfit, Reasons and Persons, ch. 16. Prefiguring Parfit, Adams gestures at this defence of BT in ‘Must
God Create the Best?’.
13
BT creature-affecting theorists face an extra problem not encountered by their secular counterparts.
Following Parfit, the non-identity literature presupposes some secular account of personal identity. But
some BTs favour non-secular accounts, where personal identity depends on God’s will. For instance,
Stephen T. Davis suggests that the fact that God wills that a certain future person is my future self is
sufficient (in the right circumstances) to make that person me (Davis, Risen Indeed, p. 119). On this
criterion of identity, God cannot face a non-identity problem. (We return to Davis’s divine will account of
personal identity, and its implications for personal immortality, in chapter 10.)
 THE CASE AGAINST BENEVOLENT THEISM
have had. However animal identity is determined, there is some poor actual Fido who
would have enjoyed a better life in some other possible world.
Creature-affecting benevolence has a second deficiency. It imposes a necessary
condition that is not sufficient. God must give each creature the best life it could
possibly enjoy. But suppose God creates sentient creatures whose lives contain only
unrelieved excruciating agony. Suppose this agony is all identity-linked. No creature
could have existed without its agonies. Each enjoys her best possible life. Has God
been perfectly benevolent?
The obvious answer is: No. God had the option of not creating those creatures at
all. And a benevolent God would take that option. Some possible lives are worse, for
the creature itself, than no life at all. Some possible lives are not worth living. No one
who cared for a creature would give it such a life.14
Any plausible creature-affecting benevolence must therefore combine two require-
ments: give each individual both its best possible life and a life worth living. The latter
‘zero’ constraint sets a minimum condition that each individual’s life must meet. The
literature on non-identity also offers more stringent thresholds, where a benevolent
creator must ensure that each creature enjoys a flourishing life, much better than
merely worth living.
Two additional thresholds emerge from the literature on evil itself. Some BTs
argue that God will not allow unredeemed evils.15 An evil is redeemed when it
somehow becomes good. While animal suffering is bad in itself, it can be part of
some good that could not otherwise exist. The evil part somehow makes the whole
more valuable.
A less stringent constraint would prohibit uncompensated evils. With redemption,
the evil itself becomes valuable. With compensation, the good outweighs the evil. The
evil is still bad. The animal is still worse off than it would have been if it could have
enjoyed the good without the evil. But that desirable combination is not possible.
Good and evil cannot be separated, and the former outweighs the latter.
Human redemption arises because we can understand our own suffering, see the
point of it, and place it within a satisfying narrative of our life. So far as we know,
animals are incapable of these insights. Redemption may be impossible for this
individual animal. God could transform Fido into a creature capable of redemption.
But would the resulting super-dog still be Fido? Can God compensate Fido instead of
offering redemption, or should God instead refrain from creating animals who suffer
and cannot experience redemption?
14
Consequentialists dispute whether it makes sense to say that non-existence is better than existence, or
that existence is worse. (See, e.g., Parfit, Reasons and Persons, App. G; McMahan, ‘Wrongful Life’; Broome,
Weighing Lives.) We can sidestep this debate. What matters is that, if you cared for a creature, you would
not give it such a life.
15
See, e.g., Adams, Horrendous Evils and the Goodness of God.
ARGUMENTS FROM EVIL 
Compensation is easier than redemption. But it is still a very high standard. Every
evil must have its compensating good within the animal’s own life. And, if the ‘best
possible life’ condition is also satisfied, those compensating goods must be logically
inseparable from the evil itself.
There are earthly animal lives that are clearly not worth living, where great
suffering is never redeemed nor compensated. None of this is controversial, on any
plausible account of well-being. If a creature’s earthly life is her whole existence, then
no creature-affectingly benevolent God would create this world. Any adequate
theodicy for animal agony needs an afterlife where evil can be redeemed or compen-
sated, or at least where every individual’s life can become worthwhile. Immortality for
both humans and animals is an essential component of any acceptable theodicy. Even
if the atheist argument in this chapter fails, BT must still defend this extravagant
metaphysical commitment. (We return to immortality in chapter 10.)
Even with a suitable afterlife, BT still faces significant obstacles. We require a
theodicy, not a defence. So we need a credible greater good that probably outweighs
animal agony. The primary conclusion of this chapter is that it is very hard to
imagine such a good in the human case, where BT can appeal to our freedom,
moral responsibility, and ability to respond to evil. It is therefore even harder to
imagine a good that could outweigh animal agony in the animal’s own life. This
suggests that BT may need to abandon creature-affecting benevolence.16
8.2.2 Impersonal consequentialism

Creature-affecting benevolence is very demanding. It is also controversial. Many
moral theories are not creature-affecting, and it is far from obvious that God’s
benevolence will be creature-affecting. Given our interest in consequentialism, we
should consider replacing creature-affecting benevolence with impersonal conse-
quentialism. (Or, as I argued in chapter 7, perhaps moral perfection combines the
two, incorporating both person-affecting and consequentialist virtues.) How would a
shift from the former to the latter affect the argument from evil?
In one sense, creature-affecting benevolence is less demanding. If different crea-
tures inhabit different possible worlds, then a creature-affecting God can opt for less
16
When I originally wrote this section, I thought I was presenting original objections to BT. But Trent
Dougherty’s recent book (The Problem of Animal Pain) has persuaded me otherwise. Dougherty’s project is
very different from mine. (He offers a Christian theodicy, for one thing!) But we share many assumptions:
that the argument from evil cannot be explored without explicit moral commitments, that BT needs a
theodicy, and that BT’s theodicy needs very strong claims about what happens to individual animals after
death. Dougherty thinks those strong claims are plausible, and I don’t. I discuss this disagreement further
in chapter 10. Another difference between my approach and Dougherty’s lies in our divergent moral
theories, and especially our different accounts of virtues such as compassion and courage that are
intrinsically linked to suffering. I return to that disagreement briefly in section 8.2.2. (I am grateful to an
anonymous for referee for alerting me to Dougherty’s book, which appeared only while I was making final
revisions to my manuscript.)
 THE CASE AGAINST BENEVOLENT THEISM
than optimally happy creatures, while an impersonal consequentialist God must

create different creatures who are capable of better lives. However, impersonal
consequentialism is also less demanding in one crucial way, because it allows trade-
offs where one individual’s suffering benefits others. Creature-affecting benevolence
forbids trade-offs. Each individual must have the best life possible for her. An
impersonal consequentialist God might allow one individual to suffer evil, if that
gave other individuals some greater good that could not possibly exist otherwise.
In chapter 7, we also explored another interpretation of perfect benevolence:
Parfit’s ‘wide’ creature-affecting benevolence. Our benevolent agent now aims, not
just to give each creature the best life it can enjoy, but also to create creatures who can
enjoy the best possible lives. If this creature’s life necessarily contains this evil, but
some other possible creature could enjoy the same good things without that evil, then
a wide benevolent agent will create the latter creature—and then give it a life without
that evil.
Unlike impersonal consequentialism, wide creature-affecting benevolence does
not permit trade-offs. One creature is never harmed to benefit another. This
interpretation is also more demanding than narrow creature-affecting benevolence,
because it forbids the creation of clearly suboptimal lives when much better lives
were available for different possible creatures. I argued in chapter 7 that, if we
expect God to be perfectly benevolent, then a wide creature-affecting interpretation
seems most appropriate. But judged by this standard, our world is clearly not
acceptable.
AP’s argument from evil need not decide between creature-affecting benevolence
(wide or narrow) and impersonal consequentialism. All three are too demanding to
permit a benevolent God to create the actual world. On its own, animal suffering is an
evil that God wants to avoid. BT must show that some greater good outweighs that
evil. For creature-affecting benevolence, the greater good must occur within the
creature’s own life. We saw in section 8.2.1 that such goods are unlikely to exist.
Impersonal consequentialism allows for external greater goods outside the creature’s
own life. Unfortunately, these goods are also hard to find.
Given the volume of animal suffering, we need a very significant good. And,
because we demand a theodicy (not merely a defence), we need to know what that
greater good is. Furthermore, good and evil must be logically inseparable. Otherwise,
God can simply arrange for the good without the evil. BT apologetic literature offers
two basic models. Evil is either a by-product of the greater good (e.g. animal suffering
is a by-product of the evolutionary process), or a precondition (e.g. human compas-
sion for animals could not exist unless animals suffered.)
The ultimate cause of much animal suffering is the evolutionary process. So BT
must find some good in evolution. Otherwise, God would create animals without
evolution. One common BT story is that God designed evolution to produce
human beings. We questioned the plausibility of that explanation in chapter 4.
But we grant it now for the sake of argument. Suppose God does value the
ARGUMENTS FROM EVIL 
evolutionary emergence of humans. Could this provide a sufficient justification for

animal suffering?
There are two problems with this explanation. First, all by-product explanations
share a common flaw. It is very difficult to establish that any by-product is essential.
Is there no equally valuable possible world with less animal suffering? Couldn’t
sentient beings have emerged via a gentler process of evolution, where creatures
only experience mild discomfort instead of agony, or where all sentient life is
herbivorous? Or why couldn’t evolution be more efficient, thus shortening the
period of pre-human animal suffering? If God really cared about animal suffering,
surely the evolutionary process could have been reconfigured to reduce it. Modal
sceptics deny that we know these are genuine evolutionary possibilities. The shift
from defence to theodicy, with its accompanying shift in the burden of proof, is
crucial here. BT needs to demonstrate that these alternatives are not possible. (We
return to these issues in section 8.9.)
A second problem is speciesism. The same similarities that rule out a God who
cares for humans but not for animals also rule out a God for whom anything human
outweighs animal suffering. Human pleasures and achievements may be greater than
those of animals, but are they so much greater? Perhaps evolution will one day, on a
distant planet or in the distant future, give rise to creatures whose accomplishments
dwarf the animal suffering that preceded them. But it is simply egoism to believe that
we are such creatures. (In chapter 7, we explored the possibility that God aims to
produce superior beings far above human beings. If animal suffering is necessary for
this divine goal, then that would support AP rather than BT.)
A second BT explanation involving external goods is that suffering provides the
opportunity for morally desirable responses to the suffering of others. This good is
inseparable from evil. Suffering is no longer a by-product, but an essential precon-
dition. Without suffering, there could be no compassion.
One obvious objection is that our world would contain ample opportunities for
valuable responses even with much less suffering. Indeed, animal suffering is entirely
redundant. Humans can feel compassion for one another. Why create a world where
animals suffer as well? (This objection is especially powerful when animals suffer
before humans evolve.)
Our utilitarian commitments provide an additional reply. Animal suffering is
too high a price to pay for morally admirable responses. Our responses could never
justify millennia of pre-human animal suffering. Compare the popular debate over
the respective merits of charity and compulsory redistribution as responses to the
suffering of the poor. Proponents of charity argue that it offers the rich the
opportunity to be virtuous. Utilitarians object that the price of this freedom is
too high. When rich people choose not to be charitable (as many do), the poor
suffer. In terms of aggregate well-being, the sufferings of the poor outweigh the
freedoms of the rich. Similarly, animal suffering outweighs human freedom. Given
the similarities between humans and other animals, the value of a human response
 THE CASE AGAINST BENEVOLENT THEISM
cannot outweigh the disvalue of the original agony. Once again, BT must appeal to
human specialness.17
BT could reply that, even if humans and animals are morally equal, human moral
responses can still outweigh animal suffering—because human responses would also
suffice to outweigh human suffering. Theodicy arises naturally from our urgent
human need to make sense of suffering, misfortune, and injustice. Because we inhabit
a world where suffering (both animal and human) is rife, many of our central moral
concerns relate to suffering, and our most cherished moral virtues are often
responses to suffering (both one’s own and others’). In such a world, compassion,
courage, and self-sacrifice are highly prized. It is natural to conclude that these virtues
are the highest goods, and therefore that suffering can earn its keep because it is their
essential precondition. In possible worlds without suffering, these virtues do not
exist—and therefore something of great value is lost.18
Consequentialists have several related replies to this common BT argument. First,
hedonist utilitarians insist that value resides primarily in goods and bads themselves,
and only derivatively in human responses to them. This intuitive reaction can be
given a formal justification, and extended beyond hedonism. For instance, Thomas
Hurka has recently argued that, as a matter of moral logic, the value of a virtuous
response to any evil cannot exceed the disvalue of the evil itself.19
Theodicy often treats compassion, courage, and other responses to suffering as
virtues. Consequentialists offer their own accounts of virtue.20 A virtue is a character
trait that happens (in the actual world) to promote what is independently valuable. In
a world full of suffering, courage and compassion are virtues. In such a world, it is
appropriate that people acquire these virtues, and therefore feel compassion only in
response to great suffering. However, in a world without suffering, the disposition to
feel compassion only when confronted with great suffering would be pointless. In
that world, it is better to feel compassion when someone misses out on a significant
good, even if they do not suffer much. (In that possible world, this is the worst kind of
thing that ever happens to anyone, and so it deserves the deepest compassion.) The
utilitarian says, not that the highest virtues are missing in that world, but that a
different set of virtues are the highest ones.
Consequentialists can thus reject what they see as theodicy’s over-valuing of
responses to suffering. Our Benthamite aversion to caprice plays no direct role in
this argument. But AP will diagnose something similar: a natural but regrettable
17
To highlight the arbitrariness of this BT appeal, it is instructive to consider beings above humans.
Consider the following analogous justification of human suffering: God permits human agony because
evolution will one day produce greater beings whose appreciation of our suffering will dwarf its disvalue.
What justifies taking animal suffering as a means without treating our suffering the same?
18
The very different theodicies of Adams, Stump, and Dougherty all cite the importance of virtuous
responses to suffering.
19
Hurka, Virtue, Vice, and Value.
20
In addition to Hurka, see also Driver, Uneasy Virtue.
ARGUMENTS FROM EVIL 
human tendency to make sense of our situation by treating contemporary actual

connections as necessary ones. Some kinds of understanding, moral awareness,
virtue, appreciation of goodness, desirable interpersonal relations, and so on, are
actually made possible by great suffering. As a matter of fact, without great suffering,
these human goods would not exist. We infer that goods like these cannot exist
without great suffering.21 But this simply reflects our lack of moral imagination.
AP can concede that these particular goods would not have existed without this
particular suffering. But AP argues that equally valuable (or more valuable) goods
could exist without any comparable suffering. Those evil-free goods are unavailable
to human beings in this life, and perhaps we could never enjoy them. But God could
have created beings who would. And a perfectly benevolent God would so create. We
fit great suffering into a narrative that makes sense of it, because we have to. But that
doesn’t make it right.22 (We return to this debate throughout this chapter, especially
in relation to liberal-w3 in section 8.9.)
Even if they were to succeed, of course, these BT replies are only available to
impersonal consequentialists. Creature-affecting BT rejects them out of hand. A God
who cared for each creature would never use animals as a means to the emergence of
humans, or the exercise of our capacities.23 And if we want to retain BT’s emphasis
on both God’s moral perfection and God’s concern for each individual person, then
our BT probably will include some creature-affecting interpretation of benevolence,
whether wide or narrow.24
I conclude that our project strengthens the already strong utilitarian argument
from evil based on animal suffering. We now turn to another distinct argument,
based on a different kind of evil, and designed to appeal to those who are suspicious
of the hedonist utilitarianism that motivates the argument from animal suffering.
8.3 Free Will Theodicy

The literature on evil divides human suffering into two categories: natural evil arising
from purely natural causes (disaster, disease, death); and moral evil resulting from the
actions of other humans.
21
In Dougherty’s model, the highest good is the making of saints, and saints must suffer. In his own
words: ‘All saints display some of the highest virtues. All opportunities for sainthood involve very
significant trials . . . All the best worlds where God creates have saints, and all saints suffer’ (Dougherty,
The Problem of Animal Pain, pp. 124 and 127).
22
Dougherty claims that the best narratives are tragic (Dougherty, The Problem of Animal Pain,
pp. 131–3). And he is right. But who wants to live in a great work of literature? Who would want their
children to live in interesting times? The most compelling narrative is not the best life.
23
Dougherty replies that the superhuman level of compassion for animals is felt by the animal itself in a
future life. Therefore, the animal’s suffering improves its own existence. We return to this possibility in
chapter 10.
24
BT is not logically committed to God’s moral perfection or God’s concern for each individual person.
But these are very common commitments of many actual BT religions and much actual BT philosophy. If
we abandon them, we reject much of the content of actual BT belief.
 THE CASE AGAINST BENEVOLENT THEISM
Many commentators regard natural evil as the greater threat to BT, largely because
(as we’ll see) BT has a prima facie plausible explanation of moral evil. Utilitarians see
suffering as very bad for humans. Natural evil is very significant. However, I will not
discuss this aspect of theodicy in detail here, as the arguments largely parallel those
regarding either animal suffering or moral evil. (If my arguments based on animal
suffering and moral evil fail, then AP could in principle fall back on arguments based
on natural evil. However, I doubt that the failure of the first two sets of arguments
would leave much room for the successful development of the third.)
In moral evil, one human being’s suffering results from another’s wrongdoing. I will
concentrate on ‘horrendous evils’: ‘evils the participation in which (that is, the doing or
suffering of which) constitutes prima facie reason to doubt whether the participant’s
life could (given their inclusion in it) be a great good to him/her on the whole’.25
The argument from moral evil is simple. An omnipotent benevolent God could
create a world without horrendous moral evil. This world contains such evil. There-
fore, there is no such God. Most BT responses to moral evil appeal to free will. Evil is
an unavoidable side-effect of human moral freedom. God could only avoid evil by
creating automatons. Despite its evils, our world is better than any world without free
agents.
Contemporary debate often begins with Plantinga’s reply to Mackie’s ‘logical’
argument from evil.26 Mackie argues as follows. For any free agent (F) and any
time (t), it is possible that F does no evil at t. So it is possible that F never does evil.
This is true of all free agents. Therefore, for any population of free agents, there is a
possible world where those very agents never do evil. A benevolent God creating free
agents will choose that possible world. God will not create any free being who ever
does evil. But actual free beings sometimes do evil. Therefore, God does not exist.
Plantinga does not deny that there are possible worlds where free agents never do
evil, nor that such worlds are better than those where evil is done. But he denies that
God could choose a possible world where no free being ever does evil. Free beings
choose without outside determination. (That is what freedom is.) If F is free, then
God cannot choose what F will do. God can create free Fs, but God cannot choose
between different possible worlds where those Fs do different things. God can only
create the Fs, and then wait to see what they actually do. If free agency is sufficiently
valuable, then God might create free agents who end up choosing evil. God and evil
are not incompatible.
Many commentators (including many atheists) take Plantinga’s rebuttal of Mackie
to be decisive.27 But Plantinga only attacks Mackie’s ambitious claim that evil and
25
Adams, Horrendous Evils and the Goodness of God, p. 26.
26
Plantinga, The Nature of Necessity, ch. 9. For the logical argument, see Mackie, ‘Evil and Omnipo-
tence’; Mackie, The Miracle of Theism, pp. 150–76.
27
On the philosophical consensus that Plantinga defeats Mackie, see Oppy, ‘Arguments from Moral
Evil’, footnote 2. (As we’ll see, Oppy himself has reservations about Plantinga’s success.)
ARGUMENTS FROM EVIL 
God are logically incompatible. Plantinga offers a defence of BT. Our BT needs a
theodicy, and not merely a defence. We will therefore examine a theodicy based on
Plantinga’s defence. (This is not eccentric, as many BTs offer free will based
theodicies.)
The rest of this chapter evaluates a free will theodicy (FWT). I will conclude that
any such theodicy is incompatible with our central moral commitments. As ever,
I can only consider one version of FWT, and one set of utilitarian replies. But I trust
my selections are not too idiosyncratic. While other theodicies could be constructed,
I am confident that the case for BT stands or falls with free will. If this theodicy
cannot work, then none can. (As the epigraph from van Inwagen attests, many BTs
agree.)
Section 8.4 draws out the metaphysical demands of FWT. Section 8.5 argues that
the actual world is not as FWT claims. Sections 8.6 to 8.9 argue that, even if the actual
world did meet FWT’s requirements, we can still imagine a possible world that a
benevolent God would have preferred. As that world is not the actual world, there is
no benevolent God.
What does FWT need? First, it needs what I will call contra-divine free will (CDF).
A creature enjoys CDF if and only if God cannot choose between different possible
worlds where that creature makes different choices. Now let F2 be the most valuable
freedom that is not contra-divine. FWT claims that a world where F2 creatures
always do the right thing is worse than one where CDF creatures sometimes inflict
horrendous evil. This comparison between CDF and F2 is crucial. F2 creatures need
not lack all freedom. (Indeed, I will argue that, for all anyone knows, we ourselves
might be F2 creatures.)
FWT compares four possible worlds.
w1. Creatures with CDF never inflict horrendous evils.
w2. Creatures with CDF sometimes inflict horrendous evils.
w3. Creatures with F2 never inflict horrendous evils.
w4. Creatures with F2 sometimes inflict horrendous evils.
CDF F2
Never Horrendous Evil w1 w3
Sometimes Horrendous Evil w2 w4
w1 is better than w2, and w3 is better than w4. Mackie argues that God must choose
w1 or w3. But the actual world contains horrendous evil. So it is clearly not w1 or w3.
Therefore, God does not exist.
FWT rests on three claims:
1. Contra-divine freedom. God cannot choose between w1 and w2. (Otherwise,
God would have chosen w1.)
 THE CASE AGAINST BENEVOLENT THEISM
2. Actuality. The actual world is w2. (If the actual world were w4, then God would
have preferred w3. And, by definition, God can choose between w3 and w4.)
3. Superiority. w2 is better than w3. (Otherwise, God would choose w3 instead of
risking horrendous evil in w2.)
For the sake of argument, suppose that both w2 and w4 mirror the behaviour of
human beings in the actual world. Actuality then implies that actual horrendous evil
is performed by humans who enjoy CDF. (As we saw in section 8.2, if God allows
horrendous evil simply to promote the freedom of superior non-humans, that would
support AP rather than BT.28)
Plantinga’s defence requires only the logical compossibility of FWT’s three claims.
Theodicy requires plausibility. Do these three claims, taken together, offer a credible
explanation of the horrendous evil we find in our world?
Contra-divine freedom merely spells out a definition. FWT’s substantive claims are
Actuality and Superiority. AP need only undermine one of these two claims. I shall
argue that it can deny both. We begin by exploring the metaphysical presuppositions
of Actuality. What kind of freedom must w2 contain, if God is unable to choose
between w1 and w2?
8.4 What is CDF?

The defining feature of w2 is that God cannot choose w1 instead of w2. If w1 and w2
were deterministic worlds, then God could easily choose. By varying laws of nature or
initial conditions, God can select any specific deterministic world. In deterministic
worlds, God chooses how everyone acts. Therefore, w2 is not a deterministic world.
Furthermore, w2’s indeterminism must extend to human actions. If w2’s freedom is
compatible with determinism, then God can choose w1 instead of w2 without
compromising that freedom. CDF is not compatibilist freedom.
FWT is committed to incompatibilism. Genuine free will is incompatible with
determinism. And, because w2 must reflect the actual world, FWT also asserts
libertarianism. Human beings have incompatibilist freedom.29
28
As an aside, AP can thus borrow one popular explanation of natural evil. Some BTs treat natural evil
as an instance of moral evil, caused by the free actions of non-human agents. For instance, Alvin Plantinga
follows Augustine in attributing natural evil to fallen angels such as Satan (Plantinga, The Nature of
Necessity, p. 192; Plantinga offers a defence, so he puts Satan forward merely as a logical possibility). This
explanation is congenial to AP, which concludes that God values satanic freedom more than human
flourishing. But AP would also question BT’s suggestion that our suffering is necessary for the exercise of
valuable satanic freedom. Surely Satan could find better things to do? (Doesn’t Milton’s Satan, otherwise
one of fiction’s most engaging heroes, lose the reader’s attention when he abandons his brave explorations
to pursue his petty revenge against the utterly uninteresting Adam and Eve?)
29
I use ‘incompatibilism’ to denote a claim about what genuine freedom is; and ‘libertarianism’ to
denote the claim that we possess that freedom. While terminology is contested here, this is a fairly common
usage.
ARGUMENTS FROM EVIL 
CDF implies incompatibilism. But does incompatibilism imply CDF? Could a

creature with incompatibilist free will still lack CDF? Can God choose between possible
worlds where creatures with incompatibilist freedom make different choices? If in-
compatibilist freedom is logically distinct from CDF, then the F2 creatures in w3 may
enjoy incompatibilist freedom. Even if there is no logical gap here, there is certainly an
epistemic gap. Given our limited understanding of the metaphysics of both incompa-
tibilist free will and divine action, we cannot be confident that God could not (some-
how) choose how creatures with incompatibilist freedom act.
To see how CDF may go beyond incompatibilism, consider counterfactuals of
freedom.30 A counterfactual of freedom is a statement of the form: ‘If F is ever in
situation S, then F will do x.’ Metaphysicians debate whether prior counterfactuals of
freedom are consistent with incompatibilist freedom.31 I aim to sidestep this debate,
by arguing that such counterfactuals do undermine CDF. If incompatibilism allows
counterfactuals, then CDF is more metaphysically restrictive.
Are there true counterfactuals of freedom in w2? If there are true counterfactuals,
they are either fixed prior to God’s decision to create or not. We begin with the
former. (As we’ll see, the latter possibility is even worse for FWT.) Suppose God
contemplates creating a free creature F. Before F is created, there are true counter-
factuals of freedom regarding F’s future choices. Being omniscient, God knows these
counterfactuals. God foresees F’s future wrongdoing. Oppy objects that F lacks the
freedom that FWT needs. If there are true counterfactuals of freedom before God
creates F, then the truth-makers of those counterfactuals must be something other
than F’s own free choices. (Ex hypothesi, if F does not even exist, then F has not yet
chosen.) But then F cannot enjoy CDF, because God could vary those truth-makers.
There is an interesting parallel here with the debate between narrow and wide
creature-affecting benevolence introduced in chapter 7. Suppose counterfactuals of
freedom are fixed prior to God’s creation of F. Even God could not create F without
those counterfactuals. Suppose one of the counterfactuals of freedom for F states that,
in situation S, F will inflict horrendous evil. Suppose also that F cannot avoid S. An
agent who was never in situation S would not be F. (This may seem unlikely. But
FWT itself needs this claim. If F could exist without ever being in situation S, then
God could create F and still reliably avoid horrendous evil.) In this scenario, God
cannot create an F who never inflicts horrendous evil. But there is some other
possible free agent (F*) whose actions are governed by different counterfactuals of
freedom. F* will never inflict horrendous evil. If we are dealing with purely logical
possibilities, surely the default assumption is that F* is possible.32
30
I draw here on Oppy’s critique of Plantinga (Oppy, ‘Arguments from Moral Evil’).
31
This debate connects with medieval debates about divine foreknowledge, so-called ‘middle know-
ledge’, and the like.
32
This is especially plausible if, as FWT supposes, the individuals in question have incompatibilist
freedom. The choices of such agents do not depend on anything outside of themselves. Therefore, if F was
 THE CASE AGAINST BENEVOLENT THEISM
God can create either F or F*. In Parfit’s terminology, God faces a different people
choice. Perhaps, a narrow creature-affectingly benevolent God might create F. But
this highlights the implausibility of narrow creature-affecting benevolence. It is
clearly better, from the standpoint of human well-being impartially considered, to
create F* instead. A wide creature-affecting or impersonal consequentialist God
would create a world of different free creatures with no horrendous evil.
We can distinguish two senses of CDF: narrow and wide, to parallel Parfit’s two
interpretations of creature-affecting beneficence. FWT focuses on narrow CDF:
whether God could create this very individual and avoid horrendous evil. But
I argued in chapter 7 that God’s creature-affecting benevolence is more likely to be
wide. If so, we should shift to wide CDF: can God create an otherwise identical
individual whose future choices are captured by different true counterfactuals of
freedom, and who never does evil?
Indeed, wide CDF is more compelling than wide creature-affecting benevolence.
F’s choice of horrendous evil impacts primarily, not on F herself, but on other people.
God’s primary reason to create F* instead of F is not concern for F. Rather, God acts
either out of concern for other individuals or from a desire to promote some
aggregate good. Therefore, even a narrow creature-affectingly benevolent God or
an impersonal consequentialist God will favour wide CDF over narrow CDF.33
If counterfactuals of freedom are fixed prior to God’s creation, then God cannot
choose a possible world where these very individuals never do evil. But God can
choose a possible world where different free creatures never inflict horrendous evil.
And this is still an instance of w1, because what matters for FWT is the presence of
horrendous evil, and not the identity of individuals. Therefore, God will not
choose w2.
If there are prior true counterfactuals of freedom, then God can select a world
where no one ever inflicts horrendous evil, by selecting possible agents with the right
counterfactuals of freedom. Instead of determining how those agents will act, God
selects them because they will freely choose to act rightly. God is like many human
agents who select individuals for particular purposes due to their predictable free
choices. (The police select the informant who will respond best to a bribe; the mafia
select the assassin who will carry out her mission; and so on.) While these earthly
agents select already existing people, God chooses among possible people.
free to inflict no horrendous evil, then there must surely be some otherwise identical agent F* who does
inflict no horrendous evil.
33
One apparent exception is where the identity of F’s victim (V) is also tied to this horrendous evil, so
that V could not exist without suffering this evil. However, such a tight connection seems implausible,
especially if F’s infliction of horrendous evil is (narrowly) CDF. If F can choose a possible world where
V doesn’t suffer horrendous evil, why can’t God? And, in any case, unless God’s only motivation is narrow
creature-affecting benevolence, God will have strong reasons to create instead some other creature (V*)
who would not be a victim of F. (We saw in chapter 7 that God almost certainly does have some non-
creature-affecting motives.)
ARGUMENTS FROM EVIL 
Wide CDF is very significant. W3’s inhabitants can now enjoy both incompatibi-
list freedom and narrow CDF. While wide CDF is what matters for God’s choice, it is
presumably narrow CDF that is valuable to the agent. Therefore, w3 can include
every freedom we value.
FWT collapses if there are counterfactuals of freedom prior to God’s creative
decision. It collapses even more directly if God’s decision is what fixes those coun-
terfactuals. God can now easily decide what F does by fixing F’s counterfactuals. F’s
freedom is clearly not contra-divine.
FWT must therefore insist that there are no true counterfactuals of freedom. Or, at
least, not until after F has actually chosen. Until situation S occurs, there is no fact of
the matter regarding what F will do in S. (A fortiori, there is no such fact if F never
exists at all, but remains for ever a merely possible person.)
W2 is a world without true counterfactuals of freedom. Because it insists that w2
is actual, FWT must deny that human choices are governed by such counterfac-
tuals. This obviously raises the metaphysical costs of BT. It also drastically alters
God’s choice. Without counterfactuals of freedom, God has no prior knowledge
how F will behave in any situation where F enjoys freedom. God does not know that
F will not always do the worst possible thing. And the same is true for all free
beings. Not only can God not choose between w1 and w2, but God cannot avoid a
much worse world (w2-minus), where all moral agents always inflict horrendous
evil whenever they can.
If there are no counterfactuals of freedom, then, by creating free creatures, God
always risks w2-minus. But could any benevolent God run that risk? Wouldn’t it be
better not to create free beings at all? Or will a prudent benevolent God always
intervene after creation to prevent w2-minus? But then why not intervene more
often, to ensure there is even less horrendous evil?
The world that God risks in creating free agents is much worse than Plantinga’s
discussion suggests.34 This argument thus complements AP’s primary reply to FWT: that
w3 is better than w2. (A fortiori, if w3 is better than w2, then it must be better than w2-
minus.) We return to the comparative value of freedom and suffering in sections 8.6 to 8.9.
Exploration of counterfactuals of freedom thus opens up a gap between incompa-
tibilist freedom and CDF. A more general, but more speculative, argument for that
gap draws on our prior commitments and the conclusions of earlier chapters. One
very popular argument for incompatibilism relies on our intuition that outside
34
In theory, BT could concede that w3 is better than w2-minus (and even better than w2), but then
argue that creating free agents is still worth the risk. The expected value of creating free agents exceeds
the expected value of creating unfree automatons. As I suggest in the text, AP rejects this risk-taking
picture of divine benevolence. There are also further difficulties. If BT concedes that w3 is better than
w2, then expected value calculations require that w1 must be much better than w3. But without true
counterfactuals of freedom, how can we even make sense of ‘expected value’ here? What grounds God’s
‘expectations’? Our main argument against FWT bypasses these complexities by arguing (in effect) that
w1 is not better than w3.
 THE CASE AGAINST BENEVOLENT THEISM
interference is a threat to moral responsibility. For practical purposes, we identify

incompatibilist freedom by the absence of outside factors that would threaten human
moral responsibility. But then we have good reason to doubt that incompatibilist
freedom must be CDF. The familiar threats to moral responsibility are human
interference and causal determinism. But divine creative choice is very different.
Any extrapolation to its impact on human moral responsibility is very unreliable.
A guiding thought behind our cosmological and fine-tuning arguments in chapters 3
and 4 was that divine creation is different from naturalistic causation. God’s creative
decisions transcend the realm of scientific causes. AP also emphasizes the distance
between God’s creative activity and ordinary human agency. Drawing its inspiration
from Nozick’s dictum, AP argues that to explain the existence of our universe we
need something quite unlike any familiar cause. We therefore cannot assimilate
divine choice to either human coercion or causal determinism.
Our discussion of moral facts and their mystical dimension (in chapters 2 and 5)
provides further reasons for doubt. If human recognition of moral facts is (in part)
a connection to supernatural reality, then both our moral knowledge and our
moral behaviour are significantly different from any natural process. But then
God’s decision to create a possible world where all rational beings respond appro-
priately to moral facts is not like choosing between worlds governed by different
scientific laws.
Recent naturalistic incompatibilist theories offer another possible gap between
incompatibilism and CDF. While some libertarians have claimed that human action
is distinct from the ordinary physical world, others attempt to integrate incompatibi-
list freedom into a purely naturalistic world view.
Naturalistically respectable incompatibilism is a double-edged sword for BT. On
the one hand, such accounts raise the plausibility of incompatibilism, especially for
those who are suspicious of any suggestion that humans stand outside the natural
order. On the other hand, opponents will contend that metaphysically modest
libertarianism cannot satisfy the demand of FWT, because it falls short of
CDF. Naturalistically respectable incompatibilist freedom is not contra-divine.
There is a real logical gap between the freedom defended by metaphysicians and
the freedom demanded by theodicy. I will illustrate this claim using Helen Steward’s
recent account of incompatibilism.35
35
Another prominent naturalist libertarian is Robert Kane (see, e.g., Kane, The Significance of Free
Will). I am grateful to an anonymous reader for drawing my attention to this strand in contemporary
libertarian thought. That reader suggested these new theories are problematic for my overall argument,
because they eschew any extravagant metaphysical assumptions about human beings, and are thus not
vulnerable to my later objections based on caprice. I argue in the text, on the contrary, that, rather that
counting against AP, the existence of modest libertarian alternatives strengthens my overall argument.
Instead of relying on caprice—which some readers may feel is an overused device in this book—AP can
point to a real logical gap between the freedom defended by metaphysicians and the freedom demanded by
theodicy.
ARGUMENTS FROM EVIL 
In A Metaphysics for Freedom, Steward aims to ‘put incompatibilism on a more

solidly naturalistic footing than has sometimes been thought available to it’.36
Steward argues that the contemporary debate over incompatibilism operates at too
high a level of sophistication. In particular, when philosophers ask whether freedom
is compatible with determinism, the relevant notion of freedom ‘has been conceived
in ever more lofty and sophisticated terms’37—to the point where it is no longer
obvious that even the most sophisticated human agents actually enjoy the relevant
kind of freedom. By contrast, Steward argues that determinism is incompatible with
much more basic capacities that are enjoyed, not only by normal human adults, but
also by many non-human animals. Instead of focusing on human moral responsi-
bility, incompatibilism should be ‘a more general claim about how we must think
about causality and evolution of reality over time in a world in which animals, and
their distinctive powers of self-direction, are present’.38
Steward operates within a purely naturalistic framework, where the capacities of
both humans and other animals result from evolution. She argues that animals have
evolved as self-movers: entities able to ‘settle’ certain facts about themselves and their
world through their decisions and choices. This is evolution’s solution to the problem
of enabling creatures to negotiate very complex environments. Deterministic pro-
gramming offered an alternative solution, but evolution has gone down a different
non-deterministic route.
Steward’s novel incompatibilist position is controversial.39 As we seek to avoid
unnecessary metaphysical controversy wherever possible, it would be nice if AP
could co-opt her view, rather than having to oppose it. I shall argue that, in fact,
Steward’s naturalistic incompatibilism supports AP against FWT. This is because it
opens a large gap between incompatibilism and the kind of CDF that FWT needs.
Self-moving animals are able to settle facts about their behaviour. These facts
depend upon the animal’s decisions, and are not fixed or determined in advance.
‘Where animal agents exist in a world, the unfolding of that world through time must
wait upon decisions and choices.’40 However, the relevant facts are typically specific
details rather than general goals. A particular animal may be ‘free’ to decide how it
will satisfy its needs for shelter, food, water, or reproduction. But it is typically not
free to decide whether to satisfy them. A lion can decide when to hunt, and which
gazelle to devour. But no lion could decide to become a vegetarian. Self-moving
animals do not have ‘any very grand capacity to transcend such things as the
promptings of instinct’.41 And neither, on Steward’s view, do human beings: ‘I am
36 37
Steward, A Metaphysics for Freedom, p. 4. Steward, A Metaphysics for Freedom, p. 2.
38
Steward, A Metaphysics for Freedom, p. 5.
39
See, e.g., Broadie, ‘Agency and Determinism in A Metaphysics for Freedom’; and Levy, ‘Are We
Agents at All?’.
40 41
Steward, A Metaphysics for Freedom, p. 19. Steward, A Metaphysics for Freedom, p. 20.
 THE CASE AGAINST BENEVOLENT THEISM
sure I could not, under the normal sorts of circumstances, jump off a cliff to my
certain death or leave my children to starve.’42
The freedom of self-movers operates within strict constraints of instinct, inclin-
ation, and the like. Nor is this a coincidence. Evolution can select self-movers only if
they reliably act in fitness-enhancing ways. Otherwise, deterministically programmed
competitors would be superior. Unconstrained self-movers would not win the battle
for survival. Evolution can only ‘select’ agents whose behaviour is, in broad outline,
predictable.
Steward does not address theodicy, and her naturalistic incompatibilist framework
has no role for God. But nothing in Steward’s indeterminist picture rules out the
possibility that God could create a population of self-moving animals possessing the
same capacities and freedoms as actual animals, and yet still ensure that they
instantiate some specific pattern of behaviour quite distinct from that of actual
animals. (In particular, as Steward’s own human example suggests, self-moving
animals could remain incapable of inflicting horrendous evil on one another. At
the very least, God’s creative choices would thus include my liberal-w3, discussed in
section 8.9.) Naturalistic incompatibilism seems very unlikely to deliver what FWT
needs.
We have explored three possible gaps between incompatibilism and CDF, based
on counterfactuals of freedom, the uniqueness of divine creation, and the natural-
ization of incompatibilism. I conclude that we can reasonably posit some such gap.
Agents in w3 can enjoy incompatibilist freedom.
W2 freedom has one final essential component. FWT requires freedom with a very
particular scope: the freedom to inflict horrendous evils. In w2, this freedom must be
contra-divine. Otherwise, God can prevent horrendous evils. Conversely, this is the
only place where the freedom found in w3 cannot be contra-divine. W3 can include
any kind of freedom elsewhere. F2 creatures can enjoy CDF everywhere except when
contemplating horrendous evil. (This possibility comes to the fore in section 8.9.)
To recap: w2 contains creatures with incompatibilist freedom, CDF (both narrow
and wide), and no true counterfactuals of freedom, at least when they decide whether
to inflict horrendous evil. For the sake of argument, I grant that w2 is possible. The
real questions are whether it is actual, and whether any possible w3 is better.
8.5 Is W2 Actual?
FWT must first show that the actual world is an instance of w2, not w4. We therefore
run through the requirements listed in section 8.4, to see whether they match our
actual freedom.
42
Steward, A Metaphysics for Freedom, p. 20.
ARGUMENTS FROM EVIL 
8.5.1 Do we have incompatibilist freedom?

FWT insists that ordinary human freedom is incompatibilist. But is it? Our actual
freedom is one of the most contested philosophical questions.43 Libertarians typically
appeal to introspection or morality.44 We either experience our (libertarian) freedom,
or we infer it from the fact that we are morally responsible. Opponents reject
metaphysically loaded interpretations of human experience, defend the moral cre-
dentials of compatibilism, or embrace moral nihilism.
What does our project add to these overfamiliar controversies? A modest option is
to avoid taking sides in this contested metaphysical debate, grant FWT’s libertarian-
ism, and move on. I am sympathetic to this option, and my overall argument in this
chapter does not presuppose the defeat of libertarianism. However, I also believe AP
could reasonably take a stand here against libertarianism (or, at least, against any
libertarianism that would support FWT).
Our project initially seems to support libertarianism in several ways. First, we take
introspection seriously. Chapters 5 and 6, in particular, suggested that introspection
does offer insights into the nature of reality. And we certainly do not join those
global naturalists who dismiss any alleged insight into non-natural or anomalous
facts. If our freedom seems incompatibilist, then this appearance must be taken at
face value. Second, we endorse non-natural moral realism. So we must reject moral
naturalist or nihilist replies to the moral responsibility argument for libertarianism.
If moral facts require incompatibilist freedom, then AP must endorse libertarianism.
In particular, many prominent arguments for compatibilism presuppose moral or
global naturalism—and then try to fit human freedom into a naturalist framework.
We support FWT’s rejection of those arguments. Finally, in earlier chapters (espe-
cially in relation to mysticism in chapter 5), I suggested that methodological
pluralism and aversion to caprice imply deference to experts. Philosophers who are
not metaphysics experts should defer to those who are. While metaphysicians are (of
course) divided, the consensus of experts appears to favour libertarianism.45
Despite this common ground, our commitments also point away from libertar-
ianism. Deference to ‘experts’ cuts both ways. Libertarian metaphysicians argue that
43
For a taste of the different contemporary approaches, see Fischer et al., Four Views on Free Will.
44
Steward’s naturalistic incompatibilism, discussed in section 8.4, might offer a third way here, as it is
based on observation of animal behaviour, rather than either introspection or morality. However, as we
saw earlier, such naturalistic alternatives are very unlikely to support FWT overall, because they increase
the distance between incompatibilism and CDF. A final option for BT is to relocate our libertarian freedom
to another life. If BT embraces pre-existence, then humans can enjoy full CDF in a previous life even if all
our actions in this earthly life are determined. We return to this possibility in chapter 10.
45
Here I defer to van Inwagen’s summary: ‘It is probably still true that most philosophers are
compatibilists. But it is also true that the majority of philosophers who have a specialist’s knowledge of
the ins and outs of the free-will problem are incompatibilist’ (van Inwagen, ‘How to Think about the
Problem of Free Will’, p. 338). For a summary of the opinions of philosophers in general, see the PhilPapers
Survey, where compatibilism enjoys 59% support and libertarianism only 13.7%. http://philpapers.org/
surveys.
 THE CASE AGAINST BENEVOLENT THEISM
consequentialist moral philosophers are out of touch. The latter return the compli-
ment, replying that metaphysicians rely on naïve common-sense accounts of moral
responsibility that have been superseded. Modern (utilitarian) moral responsibility is
not incompatible with determinism. We thus reach an impasse. (And, of course,
philosophers are not best placed to judge impartially whether anything in philosophy
constitutes deference-warranting expertise!)
One objection to libertarianism is based on caprice. In chapter 5, even when
deferring to mystical experts, our aversion to caprice cautioned against metaphysic-
ally loaded interpretations of experience, especially when these are also self-
aggrandizing. AP is very suspicious of the idea that human beings are metaphysically
special. In a universe otherwise governed by physical laws, why think we are special?
Like any appeal to caprice, this argument presumes that we lack a compelling
argument for libertarianism. Caprice only comes into play when theoretical reason
does not determine a metaphysical question, and we must make a leap of faith. Our
Benthamite aversion to caprice can only justify a presumption against libertarianism.
AP thus still needs an independent rebuttal of the positive arguments for libertar-
ianism. However, our earlier discussions do provide the materials to construct that
rebuttal, at least for some arguments. First, given the variety of competing interpret-
ations of our experience of freedom, AP can reasonably insist that introspection
alone is not sufficient to overturn the presumption against libertarianism. Second, we
saw in section 8.4 that naturalistic arguments for incompatibilism, while they seem to
escape any objection based on caprice, will not deliver the kind of incompatibilism
that FWT needs.
That leaves us with those arguments for libertarianism that cite its connection to
moral responsibility. If AP wishes to reject libertarianism on the basis of caprice, then it
must first undermine these arguments. AP has several options at this point. First, while
some consequentialist defenders of compatibilism are driven by naturalism, others are
not. (After all, Henry Sidgwick—the most famous utilitarian compatibilist—was no
moral naturalist.46) Consequentialists focus on the content of moral decisions, and on
our practice of holding others to account. Libertarian moral responsibility is not
necessary for these practical purposes. Whatever the underlying metaphysics, we can
still strive to make better decisions, to live better lives, and to model those lives on the
cosmic purpose.
Our project supports this compatibilism. Our primary moral facts concern the
comparative objective values of different possible worlds. These moral facts acquire
human significance because we can discover truths about value, and respond to them.
These abilities are not affected by metaphysical debates about our freedom. Similarly,
our moral practices require that we believe that human deliberation can influence
human behaviour. This does not require incompatibilist freedom. Careful deliberation
46
For Sidgwick’s compatibilism, see his Methods of Ethics, bk 1, ch. 5. On Sidgwick’s non-naturalism,
see Phillips, Sidgwickian Ethics, ch. 2; Hurka, British Ethical Theorists from Sidgwick to Ewing, ch. 4.
ARGUMENTS FROM EVIL 
clearly can have a beneficial impact on decision-making, whatever its ultimate meta-
physical nature. (We return to this argument in section 8.7.)
A more radical option is to bite the bullet, and deny that human beings have moral
responsibility. If ‘real’ moral responsibility does have stringent libertarian metaphys-
ical requirements, then human beings do not possess it. This is just another place
where we fall short of what really matters. Unlike God and the angels (and perhaps
the superior extraterrestrials of chapter 7), we lack real moral responsibility. We have
some feeble insight into moral reality, but the moral facts do not apply to us. This
would lead to the combination of cosmic moral realism and human moral nihilism
prefigured in chapter 2 and explored further in part III. Our need to develop a human
morality may justify a fiction of libertarianism, but not the metaphysical conclusion
that we actually enjoy incompatibilist freedom. And FWT insists that libertarianism
is a metaphysical reality, not merely a practical fiction.
I conclude that, given our various consequentialist commitments, we could deny
libertarianism. FWT would then collapse. On balance, however, I think AP’s most
sensible course is to accept that we do have incompatibilist freedom, but then deny
that we have the extra freedom that FWT requires.
8.5.2 Do we have CDF?
We saw in section 8.4 that FWT’s claims about human freedom go well beyond
libertarianism. FWT also needs CDF and the freedom to inflict horrendous evil. But
neither introspection nor morality can establishes these extra claims.
Introspection may teach me that some of my decisions are free from causal
determination or interference by other (earthly) agents. But I cannot hope to
introspect that God cannot choose a possible world where I do this rather than
that, or that there are no true counterfactuals of freedom about my future actions.
These are controversial philosophical questions, not matters of experience. Even
more obviously, I cannot introspect that God could not have created someone else,
otherwise identical to me but subject to different counterfactuals of freedom. This is
not really a question about me at all. I cannot introspect that I enjoy wide CDF.
Arguments from moral responsibility are no more promising. Perhaps, to be
morally responsible, I need to know that my actions are not determined by outside
forces. But do I also need to be confident that I enjoy CDF, or that counterfactuals of
freedom are absent? Once again, wide CDF is especially problematic. My moral
responsibility surely does not rest on whether God could have created other possible
people.
A final option for BT is to insist that incompatibilist freedom implies both CDF
and the absence of counterfactuals. We then infer these extra factors from the fact
that our actual freedom is incompatible with determinism. But this inference signifi-
cantly increases the metaphysical price of FWT. If any freedom governed by any true
counterfactuals of freedom counts as compatibilist, then it will take much more than
contested interpretations of introspection or morality to establish that our freedom is
 THE CASE AGAINST BENEVOLENT THEISM
incompatibilist in the first place. Once the stakes are this high, AP can reasonably
regard libertarianism as a metaphysical step too far.
8.5.3 Are we free to inflict horrendous evil?

FWT makes one additional demand. Our CDF must include inflicting horrendous
evil. If introspection teaches me anything about my own freedom, it is that
I genuinely do not know whether I could inflict horrendous evil. I trust that the
same is true for most readers of this book. We do not know, by introspection, that
anyone is free to inflict horrendous evil.
This response may seem flippant. We know that some people do inflict horrendous
evil. So they must be free to do so, and therefore our actual freedom does extend to
horrendous evil. However, our present question is what we can introspect about our
actual freedom. And it is unclear what (if anything) introspection about paradigmatic
free choices tells us about choices involving horrendous evil.
Paradigm experiences of freedom often involve choices between incommensurable
goods. I borrow Joseph Raz’s canonical definition: ‘A and B are incommensurable if it
is neither true that one is better than the other nor true that they are of equal value.’47
When we make such decisions, we are aware that nothing—not even the balance of
reasons—determines our choice. It is then that we experience our freedom. Unfor-
tunately for FWT, incommensurability applies to competing goods. Good and evil are
not incommensurable. If our experience of freedom is experience of incommensur-
able choices, then we cannot introspect that we are free in the same way to choose
between good and evil.
Incommensurability is controversial, and the overall argument of this section does
not presuppose it. However, for those who do find it plausible, incommensurability
strengthens the case against FWT. We return to this connection in section 8.9, as
incommensurability is especially congenial to liberal consequentialism.
If introspection fails, what about moral responsibility? One problem is that many
people believe that those who inflict horrendous evil are less morally responsible than
others. Would moral responsibility disappear if it turned out that horrendous evil
only ever resulted from compulsion or psychosis, and was never really freely chosen?
Perhaps some people do freely choose to inflict horrendous evil. But the evidence is
not compelling. (More importantly, as we’ll see in section 8.9, it is hard to see how
this freedom makes things better.)
We have no good reason to believe that we enjoy incompatibilist freedom, CDF, or
the freedom to inflict horrendous evil. If there is a presumption against human
metaphysical specialness, then it remains undefeated. I conclude that our actual
freedom falls short of the demands of FWT. The actual world is w4 rather than
w2. Therefore, a benevolent God would have chosen w3 instead.
47
Raz, The Morality of Freedom, p. 322. See also Chang, Incommensurability, Incomparability, and
Practical Reason; Raz, ‘Incommensurability and Agency’.
ARGUMENTS FROM EVIL 
FWT’s Actuality claim is false. This conclusion would suffice to defeat FWT. But
our case for AP will be stronger if it does not rest on notoriously controversial claims
about our actual freedom. Accordingly, we will now accept Actuality for the sake of
argument, and focus our attack on FWT’s Superiority claim that w2 is better than w3.
8.6 Varieties of W3
For the sake of argument, suppose our actual freedom does fit FWT. We enjoy
incompatibilist CDF freedom, governed by no true counterfactuals of freedom,
extending to the infliction of horrendous evil. The actual world is w2. AP now argues
that w3 is a possible world, and that it is better than w2.
Recall our discussion of modal scepticism at the start of this chapter. For a defence, it
is sufficient to cast doubt on the opponent’s possibility claims. A theodicy must show
that those possibility claims are probably not true. If we accept that imaginability is
some guide to possibility, then the burden of proof is now firmly on BT. This reversal of
modal presumptions is perhaps the crucial difference between our dialectical situation
and Plantinga’s, and it does considerable work in the remainder of this chapter.
W3 is as close as possible to w2. The only two differences are (1) w3 contains no
horrendous evil; and (2) while God cannot choose between w1 and w2, God can choose
between w3 and w4. I will discuss three variants of w3, highlighting different kinds of
freedom that w2 lacks. (These three variants are neither exhaustive nor mutually
exclusive.) Compatibilist-w3’s inhabitants enjoy freedom that is compatible with deter-
minism. Incompatibilist-w3’s inhabitants enjoy incompatibilist freedom, but their
choices are either governed by true counterfactuals of freedom, or fall short of CDF in
some other way. Liberal-w3’s inhabitants enjoy CDF, but only when choosing between
competing goods. They do not possess CDF when choosing to inflict horrendous evil.
(They may be free to choose horrendous evil according to some other standard.)
Different strands of the utilitarian tradition will favour different arguments here.
Liberal consequentialists, who concede that freedom can outweigh suffering, will
invoke incommensurability to argue that liberal-w3 contains all the value of actual
human freedom, and therefore that it trumps w2. More conventional hedonist
utilitarians, who reject incommensurable values, will deny that freedom ever out-
weighs suffering, and conclude that compatibilist-w3 is sufficient to trump w2. The
best overall consequentialist argument is therefore disjunctive rather than cumula-
tive. AP argues that some version of w3 is both (a) a possible world that God could
choose, and (b) better than w2. AP only needs to establish this conjunction for one
variant. FWT must deny the conjunction in all three cases.
8.7 Compatibilist-w3
On one level, compatibilist-w3 is definitely possible. God certainly could create
creatures whose behaviour was determined by physical laws. (Indeed, perhaps
 THE CASE AGAINST BENEVOLENT THEISM
many actual animals are such creatures.48) But can they be moral agents? Can they
enjoy all the good things found in actual human lives? The real question is not
whether compatibilist-w3 is a possible world, but whether it is better than w2.
At this point, we have agreed that compatibilist-w2 is possible but not actual. We
are now arguing, not about our actual freedom, nor about logically possible connec-
tions between freedom and morality, but about what makes freedom valuable. This
book is primarily an exercise in moral philosophy. We seek to remain agnostic in
metaphysical disputes. But we claim no such neutrality in moral philosophy. We are
thus more willing to rely on contested claims at this point in the argument than in
our earlier discussion of actuality or possibility. Even if our actual freedom is
incompatibilist, this is not what makes our freedom valuable.
If w3 is possible, then FWT must show that a benevolent God would still prefer w2.
Presumably God will not choose a deficient world. The burden of proof is on AP to
show that w2 is deficient relative to w3. But w2 contains horrendous evil, while w3
does not. And this is a very significant deficiency in w2. W2 is deficient overall unless
it contains some greater good to outweigh its horrendous evil. The burden of proof is
back on BT. I shall argue that w3 contains all the good things found in human lives
in the actual world. Therefore, if w2 is actual, then it is deficient regarding human
well-being.
Isn’t this focus on human well-being unduly human-centred? BT never claims that
God cares only about human beings. Perhaps horrendous evil is necessary for some
other feature of the actual world. My reply is that w3 and w2 only differ regarding
human beings. If w2 contains an extra good, it must be logically inseparable from
human activity. (Otherwise, the same good could also exist in w3.) But what possible
non-human greater good has human freedom as an essential precondition? This is
even more self-aggrandizing than the standard BT claim that God cares directly for
our well-being.
Recall our response to sceptical BT in section 8.1. While our knowledge of God’s
possible reasons is very incomplete, our knowledge of human well-being is not so
lacking. We are not infallible judges of what is good for humans. But we do know
quite a lot. If we can imagine no additional human good in w2, then FWT is definitely
on the back foot.
8.7.1 The value of freedom

The differences between w3 and w2 concern human freedom and horrendous evil.
FWT must show how these differences make available a greater good that is suffi-
ciently valuable—to human beings—to make w2 better than w3 for us.
Perhaps freedom itself is the greater good that makes w2 better. Horrendous evil is
very bad. So the value of freedom must be very great. Many consequentialists will
48
Even Steward, who denies that animals are deterministic creatures, agrees that evolution could have
produced such creatures. And if evolution can, then presumably so can God.
ARGUMENTS FROM EVIL 
deny that any freedom outweighs horrendous evil. We could follow this utilitarian
tradition, and simply insist that the value of freedom cannot outweigh horrendous
evil.
I have considerable sympathy for this hedonist response. However, those utilit-
arians who are most sympathetic to it will already have been convinced by our
discussion of animal suffering in section 8.2. Furthermore, while hedonism and
compatibilism are commonly combined, there is no necessary connection between
them. A compatibilist need not be a hedonist, and our argument from evil should aim
to persuade those compatibilists who place a higher intrinsic value on freedom.
Indeed, our case for AP is much stronger if we can avoid haggling over the
comparative value of freedom and suffering altogether. I argue instead that w3
contains all the elements of freedom that have any value.49 The extra freedom
introduced by FWT has no value at all. A fortiori, the value of that freedom cannot
outweigh the disvalue of suffering. Furthermore, w3 contains more freedom than w2,
and w3 is therefore much better in regard to freedom. We can prefer w3 to w2 without
comparing the values of freedom and suffering.
Unlike traditional hedonist utilitarianism, which privileged suffering over free-
dom, liberal consequentialism is open to the possibility that freedom sometimes
outweighs suffering. However, I shall argue that, if we follow recent trends in liberal
consequentialism itself, we see that the true value of freedom is available in w3 as well
as in w2. (The influence of liberal consequentialism is strongest in section 8.9, when
we turn to liberal-w3.)
A commonplace of modern liberalism (both popular and philosophical) is that
freedom is intrinsically (as well as instrumentally) valuable. As philosophers living in
liberal societies, reflecting on our own experience, we see freedom, not merely as a
means to human flourishing, but as an essential component of it. The freely chosen
life is valuable in part because it is freely chosen. Liberals often endorse very
significant risks, sacrifices, and costs in the name of freedom. However, while it
recognizes the value of freedom, modern liberalism also denies that the freedom in
w2 has any extra value.
Freedom contributes to human well-being in many ways. I focus on three.
1. Instrumental value. People are often the best judges of their own interests, and
so their lives tend to go better if they make their own choices.
2. Valuable achievements. On many objective list theories of well-being, one
central component of any flourishing human life is the successful autonomous
pursuit of valuable goals. Achievements are especially valuable when freely
49
In the text, I argue that w3 contains the very features that make our actual freedom valuable. Another
more complicated possibility is that w3 has other equally valuable features. However, any appeal to possible
valuable features of which we know nothing comes perilously close to sceptical BT, and is therefore best
avoided. AP is on much stronger ground if it confines itself to what makes our actual freedom valuable.
 THE CASE AGAINST BENEVOLENT THEISM
chosen. Creatures who cannot choose their own goals have less valuable
achievements, and hence less valuable lives.
3. Moral responsibility. Being a morally responsible agent is a distinct (and
perhaps essential) component of a flourishing human life. If only free agents
are morally responsible, then only free agents can enjoy this good.
These claims are all controversial. But they are congenial to our project, and they
support FWT. (Without claims like these, freedom cannot outweigh horrendous
evil.) We can therefore grant them for the sake of argument. We can also set aside the
instrumental value of freedom. W3 can contain any instrumental good, because God
can arrange any combination of contingent connections. Therefore, we need consider
only valuable achievements and moral responsibility.
Actual human achievements derive their value from two factors: the value of
what is achieved, and the process of reasoning and deliberation that precedes the
achievement. For the vast majority of achievements, the former is independent of
freedom, which only enters into the latter. Deliberation contributes to the value of
achievement through the agent’s ability to recognize reasons, to assess the com-
parative weight of those reasons, to decide which is weightier, and then to act on
those judgements. These contributions are available in compatibilist-w3. Why
should deliberation require incompatibilist freedom? An agent’s achievements are
no less valuable simply because she is so constituted that she always chooses the
most valuable option.
Moral responsibility is also available in compatibilist-w3. At this point in the
argument, we have accepted that our actual freedom is incompatibilist. But FWT
must now demonstrate that incompatibilist freedom is logically necessary for moral
responsibility. The argument for incompatibilism is therefore much more metaphys-
ically ambitious here. Consequentialist compatibilist interpretations of the actual
world will now be redeployed to illustrate the possibilities available to God. Compa-
tibilists believe that incompatibilist freedom and moral responsibility are logically
distinct. This does not prove that God could separate them. But it does place the
burden of proof on those who deny this. As outsiders to these metaphysical debates,
AP moral philosophers can reasonably refuse to place that much faith in any
contested interpretation of moral responsibility.
We should also note that our real question is not whether w3 captures human
moral responsibility, but whether its inhabitants enjoy lives as valuable as our own.
Even if they lack our ‘real’ moral responsibility, the inhabitants of w3 might none-
theless enjoy a different kind of responsibility that is compatible with determinism,
and that does allow for valuable achievements.
I conclude that FWT has not demonstrated that a compatibilist creature could not
enjoy achievements or responsibilities that are as valuable as our own. Angels whose
freedom is fully compatible with determinism would differ from us. But their
freedom is no less valuable, nor are their lives less worthwhile.
ARGUMENTS FROM EVIL 
8.7.2 Responses to horrendous evil

If freedom doesn’t give w2 an advantage, then that leaves w2’s other distinctive
feature: the presence of horrendous evil. Of course, horrendous evil is bad in itself.
But is it also linked to some greater good?
Many actual human achievements respond to evil done by others. This value is
found in w2, but not in w3. This argument applies equally to any w3. Indeed, it is
largely independent of FWT. If responses to evil are sufficient to outweigh horren-
dous evil, then w2 is better than w1. Even if God could choose a possible world where
actual human beings never inflict horrendous evil, the actual world is better.
As with animal suffering in section 8.2.2, AP has three main responses. First, many
actual victims of horrendous evil enjoy no valuable ‘response’ at all. If our account of
divine benevolence is creature-affecting (either narrow or wide), then the responses
of third parties cannot justify God’s creation of w2. BT replies with the afterlife,
where responses to horrendous evil are available to all. (We return to that topic in
chapter 10.) Second, other equally valuable responses are available even without
horrendous moral evil. Animal suffering and natural evil already call for courageous
sacrifices on behalf of others, and engage our sympathy. None of my variants of w3 is
entirely free from disease, disability, suffering, or death. Morally valuable responses
to agony and injustice are available in w3. Third, even if it were necessary for some
greater good, horrendous moral evil is too high a price to pay. The value of good
human responses to evil cannot (as a matter of moral logic) outweigh the disvalue of
the original evil. Would any benevolent agent really prefer a world where x suffers
horrendous evil and others feel compassion for x to a world without horrendous evil
at all? Would any parent, given the choice, want her own child to suffer horrendous
evil so that others could learn to respond well?50 Perhaps there is a reasonable moral
disagreement here between consequentialists and non-consequentialists. As I argued
in section 8.2.2, given its consequentialist commitments, AP can deny that responses
to suffering outweigh the suffering itself.
8.7.3 Transcendent goods

BT may agree that, if we consider only earthly goods, then w3 beats w2. But w2 can
also contain transcendent goods, such as a free loving relationship with God, union
with God, divine illumination, or contemplation of Platonic Forms.
Some people claim to experience these transcendent goods in this life. In
chapter 5, we took this claim seriously. However, most victims of horrendous
evil do not enjoy transcendent goods here. If transcendent goods are to compensate
50
Dougherty argues that loving parents are willing to expose their children to the risk of great
suffering in order to develop their virtues (Dougherty, The Problem of Animal Pain, pp. 96–117, especially
100–3). But it is not clear how closely his imaginary choice situation resembles our present choice
between w3 and w2.
 THE CASE AGAINST BENEVOLENT THEISM
or redeem horrendous evil, then an afterlife is required. But BT is already com-

mitted to the afterlife, so this is not a new metaphysical cost.
AP cannot dismiss transcendent goods or deny their value. It must argue instead that
these goods are available in w3 as well as w2. Why can’t creatures in w3 also enjoy a
loving relationship with God? BT must explain why transcendent goods are available
only in w2. Here is one familiar story. The highest good for any rational creature is
union with God, which is only valuable if freely chosen. And the relevant notion of
freedom here is CDF: my love of God cannot be something that God chooses.
In compatibilist-w3, no choice is CDF. In this world, union with God is not free
from divine control. Defenders of compatibilist-w3 must instead argue that this
transcendent good is equally valuable even when, in some sense, God has chosen
it. Or, more modestly, they could drop the strong claim that w2’s extra freedom has
no value, and argue that any extra value is outweighed either by the greater avail-
ability of transcendent goods in w3 or by the presence of horrendous evil itself. For
the committed utilitarian, these comparative evaluations are not implausible.
A second common BT argument is that transcendent goods, in the afterlife,
redeem horrendous evil—thereby making the evil itself good.51 AP cannot deny
that redemption would be desirable. But it can question the likelihood that it
produces a unique good. At issue here are not the resources available to God in
w2, but what is possible in w3. Redemption occurs in w2 only if its horrendous evil
becomes part of a whole that is more valuable than any transcendent good that God
could make available in any possible w3. Our knowledge of the relevant possibilities
is too slender to justify BT’s redemption claim.
Post-mortem transcendent goods can compensate for earthly horrendous evil. But
compensation without redemption is insufficient, especially if our model of divine
benevolence is either wide creature-affecting or impersonal. Creatures in w3 can
enjoy the ‘compensating’ good without the original horrendous evil. And that is
clearly better.52
I conclude that utilitarians can reasonably judge that compatibilist-w3 is superior
to w2. However, those sympathetic to the value of freedom may prefer to base their
case against FWT on our other variants of w3.
8.8 Incompatibilist-w3
Compatibilist-w3 is clearly possible. The possibility of incompatibilist-w3 is more
controversial. (Recall that creatures in incompatibilist-w3 enjoy incompatibilist
51
See, e.g., Adams, Horrendous Evils and the Goodness of God.
52
BT might be tempted to argue that, because transcendent goods have infinite value, horrendous evil
becomes irrelevant once they enter the picture. However, drawing a lesson from the infinite utility debate,
AP can reply that the inclusion of horrendous evil within a life always makes it worse, even if evil’s disvalue
is finite and the transcendent good is infinite. We examine this argument in detail in chapter 10, in our
exploration of possible connections between infinity and immortality.
ARGUMENTS FROM EVIL 
freedom, but not CDF.) If FWT rejects counterfactuals of freedom, it will argue that
such counterfactuals are incompatible with genuine freedom. If we accept those
arguments, and if incompatibilist-w3 requires counterfactuals of freedom, then it
seems to follow that incompatibilist-w3 is impossible.
This argument is too swift. While FWT uses the same argument against both the
actual claim that our actual libertarian freedom is governed by counterfactuals of
freedom, and the possible claim that such freedom is possible, the two claims
involve different standards of proof. A controversial metaphysical argument
might undermine the actual claim without threatening the possible claim. And as
I argued in section 8.5, some possible gaps between incompatibilism and CDF are
unrelated to counterfactuals of freedom. FWT needs separate arguments against
each alleged gap. In each case, we may agree that there is no actual gap without
being confident that no gap is possible. The leap from what God has done for actual
humans to what God could possibly have done for any possible rational creatures is
very large. AP can reasonably refuse to take that leap and insist, in the absence of a
compelling argument to the contrary, that incompatibilist-w3 is possible. God
could have chosen this possible world. The crucial question is whether a benevolent
God would choose it.
We now compare incompatibilist-w3 with w2. As before, we are more willing to
rely on contested claims at this point in the argument than in the early debates over
actuality or possibility. We can proceed quickly, because the issues are familiar from
our discussion of compatibilist-w3 in section 8.7. The main exception is that
incompatibilist-w3 is much closer to w2, and therefore it is now correspondingly
harder for w2 to claim an advantage that might outweigh its horrendous evil.
Even if incompatibilist freedom is essential to the value of our achievements, it is
hard to see how that value could depend upon the absence of true counterfactuals of
freedom, or on whatever other feature creates the gap between incompatibilism and
CDF. I am not aware of any discussion of human well-being that links the value of
achievement or moral responsibility to narrow CDF, let alone wide CDF. Even if
compatibilist-w3 doesn’t contain freedom as valuable as ours, incompatibilist-w3
surely does.
What about transcendent goods, especially the value of a freely chosen relation-
ship with God? As with compatibilist-w3, AP must deny that the real value of
transcendent goods requires CDF. AP is now on firmer ground here. In
incompatibilist-w3, my decision to align with God, while not strictly contra-divine,
is incompatibilist. My decision to love God will feel as free in incompatibilist-w3 as
in w2. It is governed by no physical cause or outside human interference, but
merely by God’s mysterious choice between possible worlds. Why think the latter
threatens the value of my (otherwise) free decision to seek God? This thought is
especially compelling if I enjoy narrow CDF and lack only wide CDF. God has
created me knowing that I will freely choose to love God. How does this render my
choice less free?
 THE CASE AGAINST BENEVOLENT THEISM
I conclude that, if we admit its possibility, then incompatibilist-w3 is a real threat

to FWT. And denying that possibility seems very presumptuous, given our limited
knowledge of God’s creative power, and our rudimentary understanding of both the
metaphysics of human freedom and the nature of the physical world.
8.9 Liberal-w3
Our final variant of w3 is more controversial and more complex. The argument based
on freedom is strongest here, but the possibility of this variant is harder to establish.
8.9.1 Is liberal-w3 possible?

BT will deny that we can imagine creatures who exercise genuine freedom only when
choosing between goods. In reply, I sketch a version of liberal-w3.53
Liberal-w3’s moral agents evolved by natural selection, but in a gentler world than
our own. All living creatures are herbivores. (Perhaps all animal flesh is poisonous to
all animals, or perhaps no animal has ever developed the ability to digest flesh, or . . . )
Liberal-w3 lacks our aggressive legacy from our own carnivorous ancestors.
Liberal-w3 is not a paradise. Food is not unlimited, and natural disaster is not
unknown. As finite physical creatures, liberal-w3 agents do not escape disease,
death, or bereavement. Their lives are not perfect, and contain many opportun-
ities for growth, struggle, disappointment, and hard choices. Liberal-w3 agents
respond to reasons as they see them. They respond to values, and to the instan-
tiation of values in their own lives and the lives of their fellows. But they are not
infallible. They are rational, but not perfectly so. Liberal-w3 agents sometimes
mistake their reasons, or fail to act on the reasons they believe they have. They
often experience weakness of will; falsely believe that one good is greater than
another; or pursue more immediate lesser goods rather than striving for greater
future ones. Each liberal-w3 agent attaches more weight to her own interests, and
to those of her nearest and dearest, than to the interests of strangers. Liberal-w3
agents are not perfect altruists, moral saints, or perfectly impartial consequential-
ist agents.
However, no-one in liberal-w3 ever inflicts horrendous evil. This incapacity is not
physical. The barrier is psychological. No liberal-w3 agent ever regards another
agent’s suffering as a reason for action; none has any desire to see another suffer;
53
I claim no originality for my tale. Similar themes are explored by Tooley, Trakakis, Rowe, Everitt, and
countless writers at the optimistic end of speculative fiction. (Trakakis, ‘Is Theism Capable of Accounting
for Any Natural Evil at All?’; Plantinga and Tooley, Knowledge of God, pp. 110–15; Rowe, ‘The Problem of
Evil and Some Varieties of Atheism’; Everitt, N., The Non-Existence of God, ch. 12.) One early philosophical
precursor is the optimistic sentimentalism of David Hume, especially when contrasted with the nastier
Hobbesian vision so beloved of much contemporary philosophy. Hume’s genial portrait of human nature
may be mistaken, but it hardly seems impossible.
ARGUMENTS FROM EVIL 
and each always regards another agent’s suffering as a very strong prima facie reason
not to act in a particular way. Their sympathetic identification with one another’s
agony both drives them to relieve it whenever possible and makes them psychologic-
ally incapable of even considering deliberately inflicting such agony.
The agents in liberal-w3 have moral practices. Scarcity of resources and conflicts of
interest mean that crime and immorality are not unknown to them. They apply
sanctions to those who infringe moral norms, they institute practices to encourage
rehabilitation, and so on.
Beyond the provision of basic needs, the society in liberal-w3 emphasizes freedom.
Each is free to focus primarily on living her own life and making her own choices
between competing goods. As in our own society, different agents prefer different
goods, and many choices seem incommensurable. Indeed, agents’ experience of
choosing between these incommensurable goods is their primary experience of
freedom. As in the actual world, philosophers in liberal-w3 debate whether this
freedom is compatible with determinism, governed by true counterfactuals of free-
dom, and so on.
This is obviously not the actual world. Nor do I claim that we (or any humans)
could live such lives. But is this a possible world? We can describe creatures who have
‘freedom just as valuable as ours, but lacking the CDF to inflict horrendous evil on
one another’. But are such creatures genuinely possible?
I believe they are. Liberal-w3 agents are not radically different from many human
beings, at least those who have had the benefit of a comfortable and secure upbring-
ing. Liberal-w3 is very close to the actual world. The only difference is a very slight
change in freedom. Given our uncertainty regarding the nature of our own freedom,
we must accept the epistemic possibility that, for all anyone knows, liberal-w3 agents
do have the same freedom as many of us. The barrier against horrendous evil in
liberal-w3 need not be mysterious or occult. It could be a psychological reluctance of
an entirely familiar kind. Many real human beings are (so far as we can tell) incapable
of seriously considering horrendous evil. This suggests that liberal-w3 is possible.
A benevolent God could have created such a world. If it is better than our world, then
there is no such God.
8.9.2 The value of freedom in liberal-w3
Liberal-w3 contains everything that makes actual human freedom valuable. In
liberal-w3, where agents enjoy full CDF when choosing between competing goods,
w2’s additional freedom is an unnecessary (and disastrous) distraction. Even if some
valuable achievements require some CDF, CDF to choose evil is redundant. Freedom
to choose between goods is more valuable than freedom to choose between good and
evil. The freedom that enhances the value of achievement is exclusively the former.
The latter freedom adds nothing.
My defence of this position draws on the utilitarian tradition of J. S. Mill, on recent
work on incommensurability, on Joseph Raz’s work on freedom, and my own earlier
 THE CASE AGAINST BENEVOLENT THEISM
work.54 It thus draws unapologetically on controversial moral claims. (Utilitarians

who reject these claims were probably already persuaded in section 8.7 that
compatibilist-w3 is better than w2!)
Our own lives include choices between competing goods, and between good and
evil. We face barriers to freedom, such as sanctions, threats, or imprisonment. If they
only prevent us from choosing evil, then these barriers do not reduce our morally
valuable autonomy. My awareness that inflicting horrendous evil is severely punished
does not compromise my autonomy, or otherwise affect my life. Inflicting evil is not
something I need the freedom to do. By contrast, interference with choices between
valuable goods does reduce my well-being, sometimes quite severely.
The intrinsic value of freedom is found only in choices between competing goods.
(Or, to be more precise, once we have a choice between competing goods, the
addition of a choice between good and evil adds no new intrinsic value.) Of course,
some (political) libertarians insist that additional choice is always good. But this is an
extreme position that our liberal consequentialist can reasonably reject.
Liberal consequentialism sees the shift from a focus on good and evil to a focus on
competing goods as moral progress. This is not naive or optimistic. No one denies the
role of evil in our world. But that is precisely what makes the actual world worse than
liberal-w3.
Another relevant difference is that choices between goods can involve incommen-
surability, while the choice between good and evil cannot. We saw in section 8.5 that
incommensurability may underlie our experience of freedom. It also helps to explain
the value of freedom. If A is clearly the best option, then it seems odd to say that F’s
achievement in doing A is more valuable if she is able to (irrationally) consider B or
C instead. But if A and B represent different kinds of value, and if the balance of
reasons does not settle F’s decision, then F’s free choice between A and B exercises
her powers of discrimination. In the act of choosing, F identifies herself with one or
other kind of value. F’s choice is now genuinely autonomous. This suggests that
freedom is only valuable in cases of incommensurability. The freedom to choose
between good and evil thus has no value at all. (Even if incommensurability is not
essential to the value of achievement, it certainly enhances it, so that choosing
between good and evil adds nothing new.) I noted in section 8.5 that incommensur-
ability is controversial. However, it is a hallmark of contemporary liberal consequen-
tialism. Without incommensurability, there is no reason to value freedom so highly.
Consequentialists thus offer a disjunctive objection to FWT. If there are incom-
mensurable values, then, while human freedom can sometimes outweigh suffering,
liberal-w3 is better than w2. Conversely, if there are no incommensurable values,
54
Mill, Considerations on Representative Government; Mill, On Liberty; Chang, Incommensurability,
Incomparability, and Practical Reason; Raz, The Morality of Freedom; Raz, ‘Incommensurability and
Agency’; Mulgan, Future People.
ARGUMENTS FROM EVIL 
then human freedom cannot outweigh suffering, and even compatibilist-w3 trumps
w2. Either way, there is some w3 that a benevolent God would prefer to w2.
The value of achievement does not require the freedom to inflict horrendous evil.
Neither does moral responsibility. Whether or not actual individuals who inflict
horrendous evil are morally responsible, an agent could be perfectly morally respon-
sible without ever considering horrendous evil. Our present question is whether the
distinctive element in w2-freedom is essential to moral responsibility per se, not
whether actual moral responsibility involves that freedom.
In liberal-w3, moral life centres on the choice between competing goods. W2’s only
distinctive feature is that some lives centre on the choice between good and evil, and
some people opt for evil. The freedom enjoyed in w3 has wider scope; but this is
simply not a way that w2 is superior to liberal-w3 at all.
Liberal-w3 thus contains all the value of actual freedom. It also contains more of
that value. At their best, our own lives are characterized by richly nuanced choices
between various competing distinct kinds of goods. But the vast majority of human
lives were (and are) not like this at all. Prehistoric humans, the poor and destitute
throughout the ages, and many people in broken, developing, or even developed
societies today ‘enjoy’ only the freedom to struggle for survival. Even many who are
comparatively well off have little real choice in their major life choices.
One of our utilitarian commitments is Mill’s belief in epistemic moral progress
(chapter 1). Humans continually learn more about what is good or bad. We now
know that a life with the freedom to choose between valuable goods is much better
than one without. This new moral knowledge highlights a striking deficiency of the
actual world. The best freedom is very sparsely and unequally distributed. And the
recent human development literature has shifted the boundaries between natural evil
and moral evil.55 Famine, poverty, and lack of freedom and democracy, previously
attributed to the natural environment or to cultural factors beyond conscious human
control, are now attributed instead to contingent failings of human organization or
motivation. Liberal-w3’s inhabitants care more about one another’s well-being than
we do. They will not eliminate these problems. But they will reduce them. Freedom
will be much more widespread in liberal-w3 than in the actual world.
The inequality of evil is a rarely noted dimension of the argument from evil.
We expect human institutions to work to reduce the unequal distribution of suffering
among innocent people. So why think any benevolent God could accept an unequal
distribution? We expect human rulers to be impartial between humans. Why expect
less from God? A perfectly benevolent God will be perfectly impartial. Such a God
would not create a world where some fare so much better than others, with respect
to things that really matter for human beings, through no fault or merit of their
own. Evidence that the actual world contains great inequality is thus evidence
55
See, especially, the influential work of Amartya Sen (e.g., Development as Freedom).
 THE CASE AGAINST BENEVOLENT THEISM
against BT. (BT solves the unequal distribution of earthly well-being with an afterlife,
where God matches well-being to desert. In chapter 10, we will see how far-reaching
that afterlife needs to be.)
This argument will especially appeal within any moral theory that has impartial
foundations, where human partiality must be justified as an undesirable departure
from an impartial ideal. If we were morally perfect, then we would be perfectly
impartial. Our agent-centred prerogatives and permissions are regrettable conces-
sions to our finitude. But God has no such limitations. Absent compelling evidence to
the contrary, we should expect God to be impartial. This also bridges the gap between
creature-affecting benevolence and impersonal consequentialism, because an impar-
tial impersonal God who avoids unfair inequality has much in common with a
creature-affecting God who cares for each individual. (As we saw in chapter 5, our
interpretations of both morality and mysticism support this strong impartialism, and
I develop it further in part III.)
Even if we confine ourselves to the value of freedom, liberal-w3 is better than w2.
Liberal-w3 contains the best freedom in w2, and much more of it. We can judge that
liberal-w3 is better than w2 without ever comparing horrendous evil and freedom.
8.9.3 Transcendent goods in liberal-w3
Some people in liberal-w3 believe in a benevolent God, others don’t. Even without
horrendous evil, the imperfections of their world lead some to doubt God’s goodness;
while others simply see no positive reason to believe in any God at all. Belief in God
involves a leap of faith. Some agents seek fulfilment in transcendent goods such as
union with God, experience of God, or contemplation of Platonic Forms. They
regard these goods as valuable in part because they are freely chosen over other
competing values.
Choices between transcendent goods and other goods are thus identical in both
w2 and liberal-w3. Differences only emerge in the choice between God and evil.
Therefore, FWT must argue that this choice is especially significant. The real
choice for human beings is between God and evil; and this choice must be
contra-divinely free. Our lives go best if we chose God over evil without God’s
controlling that choice.
The idea that human life is a battle between God and evil is common in BT
religion. But can it provide w2’s advantage over liberal-w3? AP need not deny that a
connection to God could outweigh horrendous evil, nor that such a relationship is
more valuable if freely chosen and not controlled by God. But AP argues that this
value is equally available in liberal-w3. Liberal-w3 creatures can also freely choose a
relationship with God. But they choose it instead of other valuable goods. This choice
can be contra-divine in liberal-w3. How is this any less valuable than choosing God
over a life spent inflicting horrendous evil? If a relationship with God is the only
transcendent good, then liberal-w3 offers a choice between transcendent goods and
earthly goods. If God is one transcendent good among others (such as contemplation
ARGUMENTS FROM EVIL 
of Platonic Forms), then liberal-w3 offers choices between different transcendent

goods. Either choice is sufficiently valuable. What is the value in adding evil to a
choice between goods? Far from helping FWT, transcendent goods only make its task
harder, by increasing the significance of choices between goods. Earthly goods and
transcendent goods involve such radically different values that the choice between
them must be significant.
Everyday relationships provide our only model for the relationship between the
human and the divine. In human relationships, love is more valuable if freely given,
and if chosen in preference to other goods. And perhaps love is more valuable when
the other party cannot choose whether we love. (That would justify an insistence that
our love of God must be CDF.) But is love more valuable when chosen instead of life
as a mass murderer? How could this make the relationship more valuable for either
party? Would my marriage be more valuable if my alternatives had included a life
devoted to terrorism? If not, why is my worship of God less valuable (to me or to
God) if I lack the freedom to even consider terrorism?
BT could reply that a human being cannot choose between God and other goods.
Our only real alternatives are a life committed to God and a life of evil. On one
reading, this is simply false. Not every atheist commits horrendous evils. And if we
really are so pessimistic about human beings, then AP will reply that creatures who
could choose between God and a life of mundane goods would inhabit a better world
than human beings. Why think God would prefer a battleground between good and
evil to a more civilized liberal competition between distinct goods? Why would God
glorify a war that we ourselves have outgrown? Why shouldn’t God value fine
discrimination between similar goods, or choices involving genuine incommensur-
ability, over the morally uninteresting choice between good and evil?
A more nuanced reading is that a life with other goods, but without God, is
impossible. God is the Good. To be not for God is to be against the Good; which,
in turn, is to be for evil. This story is congenial to the supernatural meta-ethic we
developed in chapter 2. But it is self-defeating in our present dialectical context. If
every alternative to God is ‘evil’, then a creature could choose between God and evil
without ever inflicting horrendous evil on other rational beings. Liberal-w3 creatures
now enjoy CDF for all choices between God and evil, except those involving horren-
dous evil. Liberal-w3 now contains ‘evil’ without horrendous evil. W2’s horrendous
evil is still redundant.
A final possibility is that w2’s advantage over liberal-w3 comes, not from differ-
ences in freedom, but from the presence of horrendous evil, and the human
responses it makes possible. We rejected this option for compatibilist-w3, and it
seems no more plausible here. Liberal-w3 is no paradise. It contains ample oppor-
tunity for sympathy and compassion towards both humans and animals. (And, as we
saw before, this reply is a non-starter if we consider only earthly human responses.
Once again, FWT must fall back on the mysteries of the afterlife, to which we return
in chapter 10.)
 THE CASE AGAINST BENEVOLENT THEISM
8.10 Conclusion
Each variant of w3 is a credible possibility that is arguably superior to w2, especially
given our utilitarian commitments. AP needs to succeed in only one of the three
cases. (This is fortunate, as we have seen that different variants appeal to different
strands of utilitarianism.) BT must dismiss them all. I have argued that, given our
moral commitments, BT fails to discharge this obligation. I conclude that there is
some w3 that is better than w2 with regard to human well-being. A benevolent God
would not create w2 instead of w3. Whether or not it is an instance of w2, the actual
world is clearly not w3. So there is no benevolent God.
FWT claims both that actual human freedom makes it impossible for God to
create a world where creatures like us never inflict horrendous evil, and that any
divinely creatable world without horrendous evil would be inferior to the actual
world. I have argued against both claims. Proponents of AP can reasonable insist that
God could have created a world where rational agents enjoy lives and freedoms as
valuable as ours and yet no one ever inflicts horrendous evil. On any plausible
interpretation of divine morality, a benevolent God would choose some such
world. Yet actual humans frequently inflict horrendous evil. So FWT fails and BT
must be false.
This completes our discussion of arguments from evil. As with all our arguments,
the case for AP is not decisive. Our argument unashamedly appeals to controversial
moral claims, and we have left many alternative BT arguments under-explored. (In
particular, we have left open the possibility that BT might explain evil by appeal to
the afterlife. We close that loophole in chapter 10.) However, we have also seen that
AP’s prior arguments and moral commitments significantly strengthen a number of
already powerful arguments from evil. No impartial, liberal consequentialist can
reasonably believe that the universe was created by a benevolent God. If there is a
cosmic purpose, then it must be a non-human-centred one.
9
Religious Diversity
A contemporary apologetic for belief in the transcendent . . . must start from the
new situation revealed by our modern awareness of religious plurality and
conceptual relativity.
[John Hick, An Interpretation of Religion, p. 9]
It is evident as a matter of logic that since [the great religions of the world]
disagree, not more than one of them can be true.
[Bertrand Russell, quoted in Hick, An Interpretation of Religion, p. 229]
In the absence of any external reason for supposing that one of the competing
practices is more accurate than my own, the only rational course for me is to sit
tight with the practice of which I am a master and which serves me so well in
guiding my activity in the world.
[William Alston, Perceiving God, p. 274]
Alongside evil, one of the oldest objections to BT comes from religious diversity
(RD). Human beings endorse a vast range of seemingly incompatible religious beliefs.
How can we reasonably pick any one religion and regard it as true?
Like scale or evil, RD is presented as a fact that BT cannot explain. RD is also
thought to undermine the positive case for BT, especially by undermining the
epistemic credibility of mysticism and other alleged sources of supernatural insight.
The challenge for AP is to borrow the atheist argument against BT based on RD,
without abandoning chapter 5’s appeal to the moral and epistemic credentials of
Christian mysticism. This challenge structures the chapter. Section 9.1 explains why
RD should be regarded as a distinctive difficulty for BT, and not merely another
instance of evil. Section 9.2 critiques atheist explanations of RD, arguing that these
fail because atheism cannot explain why there is any religion at all. (My strategy here
loosely parallels that deployed in chapter 7, where I argued that atheism cannot
explain why the universe has any rational inhabitants.) Section 9.3 presents AP’s
explanation of RD, which was already sketched in chapter 5. Section 9.4 briefly
discusses one religious tradition that especially troubles AP: namely, Buddhist
atheism. Section 9.5 argues that BT’s attempts to explain RD fail. Section 9.6 inves-
tigates John Hick’s pluralist explanation, which seeks to exonerate the epistemic
credentials of all major religions, including BT. While AP has much in common with
 THE CASE AGAINST BENEVOLENT THEISM
Hick’s pluralism, the two differ in important ways; and where they differ, AP is
superior. Section 9.7 concludes by arguing that AP’s explanation of RD does not
undermine the epistemic or moral credentials of mystical doxastic practices. Our
cumulative case for AP can thus include both BT arguments based on religious
experience and atheist arguments based on religious diversity.
9.1 What RD Adds to the Argument from Evil

The basic argument against BT is familiar. God would not want RD, and could easily
avoid it. RD thus contradicts BT. As ever, our BT needs theodicy, not merely defence.
She seeks a plausible account of God’s reason to permit (or even encourage) RD. As
with evil, my argument against BT relies on controversial moral premises, drawn
from our prior commitments and the conclusions of earlier chapters. I shall conclude
that BT cannot explain RD, and that AP can.
We could simply treat RD as an instance of evil, and subsume the present
discussion under chapter 8. However, RD raises some distinct issues, especially in
light of our discussion of mysticism in chapter 5. Therefore, this chapter sets evil
aside, and asks whether RD is a separate difficulty for BT.
Suppose BT can explain evil. RD is an additional worry only if BT cannot explain
RD, despite having an otherwise satisfactory theodicy. Other instances of evil can be
explained away, but this one cannot. On the face of it, this seems very unlikely. RD
seems innocent compared to the horrendous evils of this world. If God permits the
latter, why not the former?
RD differs from horrendous evil in ways that make it very difficult to explain. First,
RD threatens a very significant BT good. Knowledge of God and of God’s purposes
are great goods for human beings. RD implies that these goods are not given to all. If
BT is true, then all non-BT religions involve false beliefs about one’s place in the
cosmos. And even adherence to the wrong specific BT religion may lead one
dangerously astray. Every religious group is a minority. Wherever the truth lies,
most people do not believe it. The unbeliever’s life would go better if she believed.
The world would be better if everyone had correct beliefs about God, and especially
about God’s love and plans for them. Therefore, RD is something a benevolent God
should seek to avoid, whatever God’s attitude to human suffering, or to the com-
parative value of suffering and freedom.1 The arguments of chapters 2 and 5
reinforce this worry. If cosmic knowledge is necessary for moral knowledge, then
correct cosmic knowledge is essential to human flourishing. How could a benevolent
God choose to make that knowledge so rare?
A second difference between RD and evil concerns modal scepticism. As we saw in
chapter 8, BT responses to evil often cite the limits of our modal imagination. Can we
1
Indeed, in some theodicies cosmic knowledge is the greater good that justifies suffering. RD thus
threatens to undermine theodicy itself.
RELIGIOUS DIVERSITY 
really imagine possible worlds where people enjoy different kinds of freedom, or
where humans flourish without evil? With RD, by contrast, the atheist’s imaginary
worlds are straightforward. Every BT religion agrees that God has revealed the
cosmic truth to some people. The atheist imagines a world where this happens
more often, or where God prevents the rise of false religion. This seems much less
modally controversial. Our new imaginative task does not involve complex counter-
factuals of freedom, or speculation about alternative evolutionary paths. We need
only imagine that God performs miracles more frequently and less ambiguously. If,
as BT often insists, knowledge of God is a freely bestowed gift, then surely God could
bestow it more often. Modal sceptics who reject the argument from evil will thus find
RD harder to explain away.
A third difference is that, while AP and atheism largely agree in their response to
horrendous evil, they disagree sharply over RD. While AP uses RD against BT, AP
must also deny that RD undermines chapter 5’s mysticism-based argument against
atheism. The next two sections explore this divide between atheism and AP.
9.2 Can Atheism Explain RD?

RD is a problem for BT only if we acknowledge the following two theses:
Substantive conflict. Different religions make incompatible substantive claims.
Epistemic parity. Different religions enjoy (approximately) equal epistemic credentials.
If different religions do not disagree, or if one is clearly epistemically superior, then

RD is innocuous. As we’ll see shortly, BT typically denies either substantive conflict
or epistemic parity. Atheism claims to explain RD without denying these prima facie
plausible claims. The atheist endorses both conflict and parity. Religions contradict
one another, and there is no rational basis to choose one over the others. Therefore,
we should simply reject them all. The atheist explanation of RD is simple. Religion is
a purely natural phenomenon, bearing no relation to any ‘supernatural’ truth.
Religions arise within, and are conditioned by, human societies. Given the differences
between societies, it is therefore no surprise that, without any objective facts to track,
religious beliefs diverge. RD is exactly what we should expect.
This atheist position combines two claims: that BT cannot explain RD and that
atheism can. Many atheists stop at the first step. If BT cannot accommodate RD, then
atheists need only gesture at their alternative ‘explanation’. However, BT has offered
explanations of RD and AP offers another. For our comparative purposes, we need
more detail on how atheism actually explains RD.
RD is the fact that there are several different religions. To fully explain this fact, we
must contrast RD with two other possibilities. There might have been only one
religion, or none at all. BT struggles to explain why we find RD rather than only one
true faith. But atheism struggles to explain why we find any religion at all.
 THE CASE AGAINST BENEVOLENT THEISM
We discussed the deficiencies of atheist explanations of religion in chapter 5. The

atheist must provide a credible deflationary explanation of religious phenomena—
especially religious experiences, mystical practice, and moral transformation. Not
every naturalist explanation is deflationary. (BT agrees that God uses some natural-
istic mechanisms for some purposes.) This is why we need to see the naturalist
explanation in detail, before we can be confident that it supports atheism. As yet,
however, atheists have not presented the necessary naturalist explanation of particu-
lar religious phenomena. There is no generally accepted deflationary account of
mysticism. A fortiori, we lack an actual explanation of the diversity of religious
experiences, practices, and beliefs.
We saw in chapter 5 that one barrier to any deflationary explanation of mysticism
is its connection to morality. If mysticism produces a moral transformation that has
no natural explanation, then the atheist cannot hope to fully explain religion.
Atheism accommodates parity by concluding that the epistemic credentials of com-
peting religions are all equally low. But the moral supernaturalist replies that, in this
crucial area, those epistemic credentials are very high indeed.
As in chapter 5, our prior commitments, especially in meta-ethics, count against
atheism here. We cannot assume that the best explanation of RD is both naturalist
and deflationary. Indeed, if we accept the argument of chapter 5, we should expect
the opposite!
As with mysticism itself, the phenomena to be explained in RD are themselves
disputed. Proponents of each individual religion insist that it has clearly superior
epistemic and moral credentials, while atheists dismiss them all. I argued in chapter 5
that our commitments, especially to methodological pluralism, the avoidance of
caprice, and supernatural morality, prompt us to take the claims of Christian mystics
very seriously. But these same commitments also make us wary of the suggestion that
there is something special about Christianity. (As agnostic third parties, we should be
especially suspicious when Christians themselves argue that there is!) We thus seek
an explanation that preserves both the positive thought that competing mystical
traditions have good epistemic and moral credentials, and the negative thought that
they enjoy epistemic and moral parity. This is what AP seeks to provide.
9.3 How AP Explains Diversity

AP’s explanation of RD draws on themes from chapters 5 and 7. Human beings can
(and sometimes do) perceive a non-natural realm of non-human-centred objective
value and cosmic purpose. This is a happy accident, because the universe is not
designed with us in mind. But it is not altogether surprising. Our limited insight into
cosmic reality is the sort of thing that might well happen in a cosmos designed so that
other rational beings will comprehend its purpose.
This explains why we have some cosmic knowledge. However, that knowledge is
partial, in both senses of the word. We have only very limited access to the realm of
RELIGIOUS DIVERSITY 
cosmic values and purpose. Our tendency to favour ourselves then leads us to
misinterpret that realm, giving it a human-centred flavour and anthropomorphizing
its source.
AP thus explains both the successes and failures of religion. And the mechanism it
uses to explain diversity is not ad hoc. We have ample independent evidence of the
human tendency to favour self-aggrandizing interpretations of the world. AP’s
explanation of RD flows very naturally from its explanation of the epistemic and
moral credentials of mysticism. AP expects that, in so far as human beings have access
to supernatural reality, that access will be very partial. Like atheism, AP is not
surprised by RD.
AP’s advantage over atheism is that we need not dismiss all mystical experience as
delusion, or pin our hopes on some as yet unspecified naturalist explanation. AP
better respects the phenomena. Where the atheist diagnoses delusion with no real
object, AP diagnoses misperception. The human-centred elements of each religion
are errors, but established mystical doxastic practices can still offer genuine insight
into cosmic purpose and value. AP can thus account for RD without retracting its
account of mysticism.
Prima facie, one might expect AP to actively deny parity between competing
religions. If AP is the correct view, then some religions are closer to the truth than
others. Non-human-centred religions are more correct than human-centred ones.
This suggests that the former have greater epistemic credentials. And if cosmic
knowledge is linked to moral knowledge, we should also expect non-human-centred
religions to be morally superior—producing more ‘saints’ than their human-centred
rivals. In chapter 5, we placed great weight on a moral test for reliable mystical
experience. The argument from diversity is often motivated by the natural thought
that different religions perform equally well in this test. It is arbitrary to favour any one
religion, because they produce equivalent moral transformations. If one religion had
much better epistemic credentials, we would expect it to have equally extraordinary
moral credentials. Conversely, equal moral transformation suggests moral parity. And
the evidence, such as it is, seems to favour moral parity over moral superiority.
Commitment to moral parity is often associated with pluralism, addressed in
section 9.6. For instance, John Hick, the most prominent recent pluralist, argues
that all religious traditions are on a par in terms of depth of insight, transformative
power as measured in moral terms by the production of saints, volume or depth of
religious experiences, or any other measure of religious ‘success’. Hick concludes his
own survey of the evidence with this tentative conclusion: ‘I . . . venture . . . the
impressionistic judgement that no one tradition stands out as more productive of
sainthood than another.’2
2
Hick, An Interpretation of Religion, p. 307. For a similar defence of moral parity, see Armstrong, The
Great Transformation.
 THE CASE AGAINST BENEVOLENT THEISM
Hick’s moral parity claim is controversial. But it is very congenial to our project.
AP seeks to remain neutral here, and to avoid leaving hostages to fortune. (So far as
I am aware, there is as yet no independent evidence that non-human-centred
religions are morally superior.) When evidence is lacking, moral parity is certainly
the least capricious position to adopt. Accordingly, in the absence of compelling
evidence that one religion is morally superior, I propose to accept moral parity and
explore its implications. Our first task is to demonstrate that, despite its theoretical
preference for some religions over others, AP can still respect moral parity.
AP seems to lead us to expect moral disparity. If non-human-centred religions are
closer to the supernatural truth, shouldn’t they also be closer to the moral truth? And
if so, shouldn’t we expect them to perform better in the moral test? AP’s reply is that
other things are not equal. The epistemic advantages of non-human-centred religions
are outweighed by a motivational disadvantage. Human-centred religions appeal
more directly to existing human motivations, and their moral lessons are easier to
accept. This motivational advantage balances (and could even outweigh) any epi-
stemic deficiency. Those who misperceive supernatural reality, and ‘see’ a God who
loves them, may respond better than those who accurately perceive a non-human-
centred purpose.
AP can thus endorse moral parity even if it denies epistemic parity. On the other
hand, AP need not entirely dismiss epistemic parity either. One lesson from
chapter 5 was the significance of the distinction between first- and third-person
defences of BT or any other world view. Alston’s question is whether someone
immersed in one specific religion can continue to trust its mystical practices. Our
question is how third parties should respond to BT in light of RD. AP can admit
that, for first-personal purposes, competing religions enjoy epistemic parity. Each
believer can reasonably remain where she is. It does not follow that impartial third-
parties cannot recognize some salient difference that favours one over the others. Each
religion offers accounts of evil and diversity that fit its own background beliefs. Each
insider reasonably prefers her own package of beliefs. Any ‘neutral’ third party is likely
to find some of those packages more congenial than others, given her own background
commitments.
Our project recognizes religious ambiguity. Our universe is open to radically
different interpretations. AP has its reasons for favouring one interpretation over
others. But those reasons are primarily moral commitments. We could thus acknow-
ledge epistemic parity between BT, atheism, and AP (and thus between more specific
religious views), and then opt for AP on moral grounds.
I conclude that AP’s explanation of RD can respect the phenomena, even if those
phenomena include both epistemic and moral parity. The atheist’s strongest reply is
to reject AP’s account of the phenomena, and deny the epistemic and moral
credentials of mysticism altogether. But that reply was fully canvassed in part I,
especially in chapter 5. If we accept part I’s case against atheism, then the fact of RD
does not give the atheist any new ammunition.
RELIGIOUS DIVERSITY 
9.4 AP and Buddhist Atheism

I conclude that, given our prior commitments and earlier discussions, AP’s explan-
ation of RD is better than anything the atheist can yet offer. Before asking whether BT
does any better, we first address a dimension of religious diversity that seems
especially problematic for AP.3
To explain RD, AP must account for the emergence of each distinct kind of
mystical doxastic practice. In section 9.3, and earlier in chapter 5, we explored BT
traditions, and the possibility of non-human-centred religions similar to AP itself.
AP can accommodate both. But this does not exhaust human religion or mysticism.
Some religions have no cosmic purpose at all. The classic example is Buddhist
atheism, where the mystic discovers the ultimate impermanence and meaningless-
ness of all existence. Buddhism also comes in BT, impersonal, and pluralist forms.
But Buddhist atheism is both the purest form and the one most troubling for AP.4
AP places Buddhism in the same broad category as BT. Both involve a misper-
ception of the real cosmic purpose, fuelled by our human tendency to prefer self-
aggrandizing or comforting views. Buddhism might involve caprice if the Buddhist
mystic infers that there is no cosmic purpose either because he cannot locate that
purpose or because (having recognized the defects of BT) he mistakenly assumes that
a cosmic purpose must be human-centred. Alternatively, if one’s metaphysical world
view involves perpetual reincarnation in a world of endless suffering, then Buddhism
appeals to our self-interest because it alone offers individual salvation.
AP’s deflationary explanation is less secure for Buddhism than it was for BT. BT
and AP share a common commitment to cosmic purpose. If AP is correct, then
Buddhism is perhaps even more mistaken than BT. Therefore, AP must allege that
Buddhism involves a greater degree of misperception. On the other hand, AP’s
allegation of caprice is less prima facie plausible here. Even if both involve self-
aggrandizing leaps, BT’s caprice is much clearer than Buddhism’s.
Fortunately, AP can bolster its explanation. AP need not insist that all cosmic
misperception flows from caprice. Sometimes we just make mistakes. Perhaps, while
BT is capricious, Buddhist atheists are simply unlucky. Atheism is a one credible
world view, and a chain of unfortunate historical accidents happened to produce an
atheist mystical tradition. (Perhaps, if atheism continues to be popular in the West,
atheist mystical traditions will eventually emerge there too.) Part I offered independ-
ent arguments that atheism is mistaken. If we find those arguments sufficiently
credible, then we can reject Buddhism on those grounds, and then offer human
error as the best explanation of Buddhism’s apparent mystical credentials. (Just as we
3
My argument in this section is particularly tentative. I hope to explore the connections between AP
and Buddhist atheism at greater length elsewhere.
4
In chapter 6, we saw that the rejection of Buddhist atheism is also essential to the ontological argument
that AP hopes to borrow.
 THE CASE AGAINST BENEVOLENT THEISM
conclude that a seemingly clear-sighted person who claims to perceive something

that we know not to exist must be mistaken.)
Even if it cannot explain how Buddhism emerged, AP will insist that only it can
explain Buddhism’s degree of success. In particular, only AP accounts for the equal
moral success of BT and Buddhism (alongside many other religions). Conventional
(Western) atheism cannot explain the success of any mystical tradition. BT cannot
concede moral parity—unless it breaks the link between moral transformation and
cosmic knowledge altogether. Buddhism itself has no problem with RD. (If there is
no God, then why shouldn’t false religions thrive?) But moral parity is a prima facie
difficulty for Buddhism. If Buddhism is true, then why does BT seem to work just as
well? Like other atheists, the Buddhist cannot explain the moral phenomena associ-
ated with RD. I conclude that the need to explain Buddhism does not undermine
AP’s superiority over atheism.
This concludes our exploration of the comparative merits of AP and atheism
regarding RD. AP offers a superior explanation, not of RD per se, but of religious
phenomena taken as a whole. Our exploration supports two further conclusions.
First, RD alone does not force AP to abandon the arguments developed in chapter 5.
(However, as we’ll see in section 9.7, RD does have some impact on AP’s positive use
of mysticism.) Second, if BT cannot explain RD, then RD will favour AP over both
its rivals.
9.5 Can BT Explain RD?

Religious diversity is not new. BT has always confronted other religions.5 And the
initial adherents of every great religion were a tiny minority. Every religion claims to
explain RD, just as each offers an account of evil. Even if BT’s explanation of RD is
less than perfect, BT might still offer the best explanation of all the relevant facts. BT
could cite the atheist’s inability to explain other unrelated facts, such as the existence
of the universe or its friendliness to life. If BT alone can explain these, then it can
afford a few anomalies such as religious diversity. A more ambitious argument is that
BT offers the best explanation of specifically religious phenomena. While BT has
difficulty explaining the diversity of religious beliefs, practices, and experiences,
atheism cannot explain the existence of religion. As ever, AP undermines these
responses by offering its own explanations. AP also explains S (there is something,
when there might have been nothing), FL (our universe is friendly-to-life), and
mystical experience. (And AP offers superior explanations of both scale and evil.)
BT thus needs to offer a better explanation of RD.
Suppose we agree that RD is a fact and that it is prima facie undesirable. Other
things equal, God would prefer a world without RD. BT will argue that other things
5
For instance, for an account of early Christian responses to diversity, see Wilken, ‘Religious Pluralism
and Early Christian Theology’.
RELIGIOUS DIVERSITY 
are not equal. RD is inextricably linked to some greater good. If we already have a
satisfying explanation of evil, then that explanation is the obvious place to start. In
chapter 8, we focused on a theodicy based on human free will (FWT). Can FWT also
explain RD?
There are several possible connections between free will and RD. The greater good
that justifies RD is either our freedom itself, or something essentially linked to
freedom, such as a valuable autonomous relationship with God. As with evil, RD
could be either a by-product of freedom, or a precondition of it.6
FWT’s explanation of horrendous evil is straightforward. If humans possess the
contra-divine freedom to inflict horrendous evil, then some people will choose
horrendous evil. Evil is thus a by-product of our valuable freedom. Similarly, perhaps
some people exercise their contra-divine freedom to choose false religions.
One obvious barrier here is that, unlike horrendous evil, it is not clear that RD
results from free choice at all. Most people do not choose their religion, even if they
do choose whether to follow the religion in which they are raised.7 The liberal fantasy
of a ‘free’ choice from a smorgasbord of religions is not the normal human situation.
Our project is open to the kind of epistemic choice displayed in leaps of faith in
response to religious or philosophical ambiguity. But such leaps of faith are not
sufficiently widespread to explain the actual diversity of religious belief.
We saw in chapter 8 that freedom comes in many different varieties, and that the
metaphysical requirements of FWT are very demanding. FWT requires incompatibi-
list freedom that is contra-divine and includes the freedom to inflict horrendous evil.
The case for human freedom may not meet the requirements of FWT. Similar
worries arise here. It is not sufficient for BT to assert that RD is connected to ‘valuable
human freedom’, because many possible human freedoms are not contra-divine. God
could thus choose a possible world that contains valuable freedom but entirely lacks
RD and its accompanying false beliefs.
The diversity of freedom also casts doubt on BT attempts to extend FWT to cover
RD even if FWT succeeds. FWT makes certain very specific claims about human
freedom, while BT explanations of RD make other specific claims. Can we be sure
that the two sets of claims overlap? Why suppose that the freedom needed to explain
RD is the very same freedom needed to explain horrendous evil? If its explanation of
RD requires a new kind of freedom, then RD raises the metaphysical cost of BT.
BT could reply that, although false religion is not directly chosen, it is an indirect
result of human freedom. For instance, drawing on familiar Christian themes,
Plantinga suggests that RD is a consequence of ‘our fallen nature’, which is itself a
6
Another possibility is that RD arises when frail human beings are seduced by demons, fallen angels, or
other false gods. God permits RD because non-human freedom is especially valuable. We briefly discussed
analogous arguments in chapters 7 and 8. I argued there that this possibility supports AP instead of BT. AP
offers a simpler explanation of RD: God doesn’t really value us at all, and only values these higher beings.
Therefore, we set this possibility aside in the text, and focus on human freedom.
7
See Himma, ‘Finding a High Road’, pp. 11–18, for a discussion of the relevant empirical literature.
 THE CASE AGAINST BENEVOLENT THEISM
consequence of our unwise exercise of our God-given freedom.8 The freedom we

need to form loving relations with God also leads us, in a different context, into
religious error.
Here, our dialectical context is crucial. Within his own reformed epistemology,
Plantinga treats actual Christian belief as prima facie reliable unless it is defeated. He
therefore explicitly offers his model of human cognitive failing as a possibility that
grounds a defence. But our BT must be more ambitious than Plantinga. Our BT needs
a theodicy to convince third parties, not merely a defence to comfort believers. BT
must argue that Plantinga’s story is probably true. At this point, AP can reasonably
side with those atheists who reply that there is no independent empirical evidence to
support Plantinga’s model. Before we accept that our epistemic powers are corrupted
by human choice, we need some independent evidence of a connection between the
two. And the empirical evidence suggests that humans do not, in any sense, choose
their religion. Unlike horrendous evil, RD does not result from human choice.
A second possible connection is that RD is a precondition of the value of human
freedom. One possibility here is that God has good reason to remain ‘hidden’. God
wants human beings to love God, and to live in accordance with God’s will. But these
achievements are only truly valuable if they are both freely chosen and difficult to
acquire. If God’s existence or purposes were too obvious, then neither religious
knowledge nor correct practice would be any great achievement. Valuable choice
must involve a leap of faith to what cannot be proved. But if God does remain
‘hidden’, then some people will (regrettably) form false beliefs about God. RD is the
inevitable accompaniment of this valuable freedom.
A related possibility is for BT to draw on the liberal utilitarianism of J. S. Mill,
deployed by AP in chapter 8. To be truly (valuably) free, my lifestyle must be chosen
from among a range of genuine alternatives—Mill’s famous ‘experiments in living’.
Everyone benefits from this diversity. Even if you would not choose a minority
lifestyle option yourself, the presence of such options increases the value of your
own lifestyle by ensuring that it is freely chosen. BT could similarly argue that God
wants RD so that those who choose the right path will do so freely.
As we saw in chapter 8, some utilitarians will simply deny that freedom is
valuable in these ways, at least to the degree necessary to justify widespread false
belief. However, our project aims to accommodate alternative liberal views and not
8
Plantinga, Warranted Christian Belief, pp. 199–240, 437–57. Here is one of Plantinga’s summaries of
his central thesis: ‘The most serious noetic effects of sin have to do with our knowledge of God. Were it not
for sin and its effects, God’s presence and glory would be as obvious and uncontroversial to us all as the
presence of other minds, physical objects, and the past. Like any cognitive process . . . the sensus divinitatis
[a special sense implanted by God to give us knowledge of the divine] can malfunction; as a result of sin, it
has indeed been damaged. Our original knowledge of God and his glory is muffled and impaired’
(Plantinga, Warranted Christian Belief, p. 214). (In footnote 22 on p. 214, Plantinga makes it explicit
that: ‘It is no part of the model to say that damage to the sensus divinitatis on the part of a person is due to
sin on the part of the same person. Such damage is like other disease and handicaps: due ultimately to the
ravages of sin, but not necessarily sin on the part of the person with the disease.’)
RELIGIOUS DIVERSITY 
to rely too heavily on controversial utilitarian claims about the comparative values
of suffering and freedom. A stronger reply is that the relevant freedom (and its
value) do not require RD. A person could still choose whether to love God even if
she had never doubted God’s existence. And God could remain ‘hidden’—or at least
not obvious—among a people who only ever encountered one religion. Horrendous
evil alone would generate sufficient doubt. Belief in God always requires a leap
of faith.
This reply has support within BT. For instance, in response to Hick’s suggestion
that epistemic distance is necessary for free choice, Alston argues that even if we were
as certain about Christian doctrine as we are about the physical world, ‘I fear that not
all of us would automatically lead the new life of the spirit’.9 Freedom would still
exist—and thus still be valuable—even if we knew of God’s existence and God’s plans
with certainty.
We must note the extent of actual religious diversity. The world contains many
radically different forms of religion—from classical theism to polytheism to imper-
sonal world views such as Taoism. BT must explain the necessity of all this diversity.
Opponents will object that God could have created a world with much less religious
variation, and still left room for choice. Consider a world containing ‘only’ Judaism,
Christianity, and Islam. Its inhabitants only choose between variants of BT. No one
ends up worshipping false gods. But surely the interminable disputes between (and
within) these three traditions still provide plenty of room for religious choice.
We can easily imagine possible worlds where God reveals religious knowledge
more reliably, or distributes miracles so that one religion had a clear monopoly. (This
is why modal scepticism is much less powerful for RD than for evil.) And it seems
hard for BT to deny that these alternative worlds would be better than the actual
world. A possible world where human beings are only exposed to the one true
religion contains much more cosmic knowledge. Atheists can dismiss this good,
but for BT it is necessary for the greatest good of all: relationship to God. What actual
greater good could possibly outweigh the loss of so much cosmic harmony?
As with horrendous evil, one striking aspect of RD is that God’s gifts are distrib-
uted, not on the basis of any human merit, but entirely arbitrarily. Those born into
the wrong religious tradition cannot even know where true belief might be found.
This seems unfair. I argued in chapter 8 that that, given our moral commitments, it is
more plausible to posit a non-human-centred God than one who is capriciously
benevolent. The bestowal of arbitrary favours is not consistent with any modern
understanding of perfect justice. AP denies that God would even allow goods and
evils to be distributed so unequally. RD is even more striking than evil, because here
God actively promotes the unequal distribution. This is not how a morally perfect
benevolent being would behave.
9
 THE CASE AGAINST BENEVOLENT THEISM
I conclude that AP can reasonably deny that RD confers any additional benefits.
Even if God would permit horrendous evil, there is no reason to permit the diversity
of the actual world. BT cannot point to other goods that might give any benevolent
God sufficient reason to permit RD. If RD is bad for humans, then a benevolent God
cannot allow it. BT’s only remaining option is to deny that RD is bad.
9.6 Pluralism
No one denies that horrendous evil is bad in itself. But many people in the modern
world regard religious diversity as a positively good thing. If RD is good, then God
will prefer it. RD would no longer count against BT at all.
Accordingly, another BT option is simply to deny that RD is bad. Despite
appearances, there is no undesirable religious diversity. We see only a snapshot.
The appearance of undesirable diversity would disappear if we took a broader view.
One obvious source of a broader outlook is BT’s commitment to life after death. We
saw in chapter 8 that every theodicy needs an afterlife, where horrendous evils are
compensated or redeemed. Perhaps, in that afterlife, God provides universal cosmic
knowledge.10 An earthly life spent worshipping false gods (or none) fades into
insignificance compared to an eternity of correct religion. Posthumous knowledge
could easily outweigh earthly ignorance. And it might also redeem it. The ignorance
of childhood is not bad at all, because it leads to the wisdom of adulthood. Perhaps
our current ignorance stands in the same relation to our posthumous wisdom. People
born into false religion are no worse off than those who grow up in the correct faith.
Even a wide person-affectingly benevolent God could thus permit RD.
If God can posthumously redeem horrendous evil, then the negative impact of RD
may be relatively easy to remove. However, the need to accommodate RD as well as
horrendous evil places an extra burden on the afterlife. Recall, as ever, that BT needs a
theodicy, nor merely a defence. We need a plausible story about why God allows RD in
this life and how it is redeemed in the next life. If the present chapter fails to find any
good in RD at all, then we may wonder what posthumous benefit could possibly justify
it. We return to that difficulty in chapter 10. For the remainder of this chapter, we ask
whether BT can account for RD without simply shifting that task to the next life.
RD adds variety and richness to our shared cultural life. This is why many secular
liberals like diversity. But can BT agree? For BT, RD means that many people build
their lives on very false beliefs about the fundamental nature of the cosmos and their
place in it. If there is one true God, how can this state of affairs be good for those
people? How could God value false beliefs about God?
RD is prima facie bad only because it involves widespread false belief. Without
substantive conflict, RD is unobjectionable. Religious pluralists deny that diversity
10
Consider the Christian tradition that, between crucifixion and resurrection, Christ descended into
Hell to offer salvation to those born before his Incarnation.
RELIGIOUS DIVERSITY 
involves any false belief. Properly understood, the main world religions do not
disagree. A benevolent God has no reason to avoid RD. Indeed, RD is good, because
it teaches us that there are many different ways to experience the divine.
This is the pluralism of John Hick.11 Hick’s starting-point is a religiously ambigu-
ous universe open to both religious and materialist interpretations. Although materi-
alism is reasonable, it is also epistemically reasonable to see the universe in religious
terms, and to structure one’s life accordingly. However, given epistemic parity, it is
arbitrary to believe one religion rather than another. If we cannot choose one religion
over another, but we wish to retain a religious interpretation of the universe, then we
must deny that different religions really conflict. Supernatural reality is beyond all
human understanding. Different religions represent that reality from different cul-
turally situated perspectives. But their real messages do not conflict. Substantive
religious disagreement is only apparent.
Many religious people agree that religions agree about many important things—
whether a moral code, a commitment to moral realism, a rejection of materialism, or
a belief in life after death. Hick goes further. Different religions are all equally true,
even when their claims apparently conflict. Hick draws a Kantian distinction between
(a) the Real as it is in itself; and (b) the Real as it is experienced by human beings.12
We all perceive, and refer to, the same object—but from very different perspectives.
Christian, Muslim, Hindu, and Buddhist all experience the same divine reality, even
though they express themselves in very different ways. At the noumenal level, all
religions have the same object: the Real as it is in itself. But all our religious concepts
are confined to the phenomenal level of human experience.
Hick’s pluralism has very broad scope. For a Christian pluralist, all Christians
worship the same God. We can also imagine a BT pluralist, who believes that, despite
their differences, all BTs experience the same ultimate reality. In both cases, the
ultimate reality is God. When Christians or other BTs speak of God, their utterances
are true of the ultimate reality.
Hick’s pluralism runs deeper. The Real as it is in itself is beyond both personal and
impersonal. The Real can be experienced as either a personal God or an impersonal
ultimate reality. But the Real itself is neither personal nor impersonal. It cannot be
identified with God rather than Tao, or Platonic Form, or whatever. Hick’s pluralism
includes every ‘major world religion’.
Hick inverts the familiar argument from RD to atheism. All the religious traditions
are epistemically on a par. Therefore, either all are correct or none is. But they cannot
all be incorrect. Therefore, they must all be correct. And the onus is then on the
philosopher to explain how they can all be correct. Kantian pluralism emerges as the
only response to RD that preserves religious realism.
11
Hick’s own output is vast. I rely primarily on Hick, An Interpretation of Religion.
12
Hick, An Interpretation of Religion, pp. 240–6. Hick applies Kant’s account of ordinary perception to
religious experience, thereby diverging markedly from Kant’s own account of religion.
 THE CASE AGAINST BENEVOLENT THEISM
This brings us to another distinctive feature of Hick’s pluralism. Unlike many

people who self-identify as ‘pluralists’, Hick is a realist about religion. Religious talk is
descriptive. Hick rejects anti-realist interpretations, where religious talk is merely the
expression of emotion, or some other non-truth-directed ‘language game’. (Nor is
Hick a religious ‘fictionalist’ who reluctantly admits that religious talk is, strictly
speaking, false.) For Hick, anti-realism is too radically revisionary. It takes us too far
from the ordinary meaning of religious language, and effectively abandons trad-
itional religion. Pluralism, he argues, defends what ordinary believers want to say.
The Christian creeds, for instance, are true at the phenomenal level.
Hick’s pluralism has limits. Not all actual views are included. Like Alston’s (see
chapter 5), Hick’s epistemic presumption in favour of religion only applies to
established historical religions. And it does not extend to all actual religions. Hick
dismisses a number of cults and minority religions, on the grounds that only the
‘major world religions’ have the right moral credentials. Hick stresses the ‘trans-
formational power’ of religious experiences in individual lives. Across all major
religions, religious experience makes people less self-centred and morally better.
‘The basic criterion must be soteriological. Religious traditions and their various
components . . . have greater or lesser value according as they promote or hinder the
salvific transformation.’13 (Namely: ‘the transformation of human existence from
self-centredness to Real-centredness’.14) As well as enabling Hick to dismiss minority
religions, this moral test rules out materialism. If materialism were true, then religion
could not have such a great moral impact.
Hick’s attitude to materialism is ambivalent. By embracing religious ambiguity,
Hick acknowledges the epistemic respectability of materialism. But his pluralism
does not include materialism. Materialism is not itself a reasonable perception of the
Real, but rather a denial that there is anything transcendently Real behind the
material world. Pluralism only covers positive religious views. A Hickian pluralist
believes that, while epistemically reasonable, materialism is false.
In Hick’s vocabulary, it is therefore important to distinguish between ‘materialism’
and ‘atheism’. In the everyday sense of the word, Hick’s pluralism embraces some
atheist views. There are major religions without gods, such as Buddhism and Taoism.
For Hick, these atheist religions still count as (non-materialist) perceptions of the Real.
Aside from materialism and minority religions, Hick’s main opponent is the religion
exclusivist: the person who claims that her religion alone is true. This is where Hick’s
commitment to moral parity comes into play. If only one religion were true, then surely
its adherents would be morally better than others. Conversely, once we accept moral
parity, it is egoistic to prefer one religion over others. Hick is thus more hostile to
exclusivism than to materialism. Materialism, while false, is a reasonable interpretation
of our religiously ambiguous universe. Exclusivism is simply unreasonable.
13 14
Hick, An Interpretation of Religion, p. 300. Hick, An Interpretation of Religion, p. 300.
RELIGIOUS DIVERSITY 
Hick’s objection to exclusivism is worth exploring, because it sheds light on both

the strengths and the weaknesses of his pluralism. One prominent recent defender of
exclusivism is Plantinga.15 Plantinga’s exclusivist denies epistemic parity. As an
externalist in epistemology, Plantinga denies that internal rationality is the whole
epistemic story. Plantinga’s Christian acknowledges that practitioners of other reli-
gions are equally (internally) rational in staying put. She also accepts that, like
Christians, other folks can reasonably claim that, if their religion is true, then their
beliefs probably have warrant. If Confucianism is true, then Confucian beliefs
probably have warrant—and likewise for other established religions. But Plantinga’s
Christian believes that her situation is epistemically unique. Her religion—unlike all
the others—is actually true. As a matter of fact, for whatever reason, God has given
the Christian insights into the true nature of things.16 While the beliefs of other
religions would have had warrant if those religions had been true, only her Christian
beliefs actually have warrant, because only Christianity is true.
One common objection to Plantinga’s exclusivism is that his Christian appears
arrogant or egocentric.17 This is Hick’s response, and it is a very common reaction
among liberal undergraduates! Plantinga’s exclusivist continues in her religious
beliefs, in the face of religious diversity, despite acknowledging that she has no
knock-down argument to convince an impartial third party that hers is the true
religion.18 How can this be epistemically responsible? Isn’t such a person simply
arbitrarily privileging her own intuitions?
Plantinga points out that a similar pattern arises for many other beliefs. ‘Perhaps
there is no way you can find much moral common ground with a member of the Ku
Klux Klan. Perhaps you can’t find any premises you both accept that will serve in a
good argument for your views and against his. Would it really follow that you don’t
have a right to give decisive assent to the proposition that racial bigotry is wrong?
Hardly.’19
Plantinga concludes that it cannot always be arrogant to retain one’s basic beliefs
in the face of disagreement. And, as Plantinga himself notes, if his Christian is
arrogant, then isn’t the religious sceptic, atheist, or pluralist equally arrogant? ‘If
one is arrogant by way of believing a proposition others don’t, isn’t one equally
arrogant by way of withholding a proposition others don’t?’20
15
See, e.g., Plantinga, Warranted Christian Belief, pp. 437–57.
16
Plantinga offers an elaborate explanation here, based on Aquinas and Calvin. He argues that, if
Christianity is true, then something like his Aquinas/Calvin model is probably true as well (Plantinga,
Warranted Christian Belief, pp. 241–89).
17
For references, and Plantinga’s most detailed reply, see Plantinga, Warranted Christian Belief,
pp. 442–77 (with references in footnote 23 at p. 443).
18
This last clause is essential to Plantinga’s definition of exclusivism. Someone who believes her religion
can be demonstrated to be true, and that unbelievers are simply irrational, is not an exclusivist in
Plantinga’s sense.
19 20
Plantinga, Warranted Christian Belief, p. 450. Plantinga, Warranted Christian Belief, p. 445.
 THE CASE AGAINST BENEVOLENT THEISM
I agree that it is not exclusivism per se that is arrogant. The charge of arrogance
must be directed at the substance of beliefs—in this case, at the illicit anthropocen-
trism of Christianity’s background claims. A central theme of this book is that this
particular charge of arrogance does have considerable force.
Our question, as ever, is not whether this exclusivist response succeeds on its own
terms, but whether it offers a convincing explanation to unbelievers or agnostics.
From this perspective, the obvious problem with Plantinga’s strategy is that each
different religion can make the same move. For instance, Tien presents a detailed
defence of neo-Confucian beliefs that is explicitly analogous to Plantinga.21 Like
Plantinga’s Christian, Tien’s neo-Confucian believes that, while basic Christian
beliefs would have had warrant if Christianity had been true, his own basic neo-
Confucian beliefs do have warrant—precisely because it is neo-Confucianism and
not Christianity that is actually true. Many religions have the property that, if they are
true, then their adherents are epistemically favoured. But without first-personal
reason to favour one religion over another, third parties will prefer an explanation
that preserves epistemic parity. This is what AP offers.
Our project has some affinity with Hick’s pluralism. Like Hick, we take seriously
the moral and epistemic credentials of religious (especially mystical) experience, and
we are suspicious of exclusivist claims on behalf of one religion. AP also seeks to
acknowledge epistemic parity without abandoning religion altogether. Finally, both
our defence of mysticism in chapter 5, and the AP explanation of RD in section 9.3,
share Hick’s emphasis on the connection between moral transformation and epi-
stemic reliability.
However, we also depart from Hick in several key ways. One obvious difference is
that we seek to avoid unnecessary metaphysical commitments. If possible, we do not
want our case for AP to rest on anything as controversial as Hick’s Kantian model of
religious perception. Furthermore, the primary motivation for that Kantian model is
one that we must reject. Our commitment to defer to contemporary scholarship
outside philosophy conflicts with Hick’s approach to mysticism. Hick’s claim that
pluralism captures the true essence of all religious traditions relies on an explicit
perennialism.22 In chapter 5, we saw that perennialism is no longer the dominant
interpretation. Modern scholars of religion are more likely to highlight differences
between contemplative traditions. This makes it harder to argue that all religions
ultimately experience the same transcendent Real. Without perennialism, Hick’s
Kantian model is under-motivated.
The shift away from perennialism, and towards more context-sensitive interpret-
ations of mysticism, also threatens Hick’s claim to respect the self-understandings of
21
Tien, ‘Warranted Neo-Confucian Belief ’.
22
See, especially, Hick, An Interpretation of Religion, p. 26. While he sometimes distances himself from
perennialism (e.g., p. 295), Hick explicitly borrows Underhill’s perennialist accounts of religious experience
(p. 46).
RELIGIOUS DIVERSITY 
each religious tradition.23 Exclusivists reply that Hick rejects traditional religion
rather than offering a reinterpretation. Hick’s real view is that most actual religions
are false, and that something else (namely, Hick’s Kantian reinterpretation) is true
instead. Can a Christian really come to believe that her vision of God as a loving
personal creator is no more accurate than polytheism or impersonalism? Do Bud-
dhist atheists really believe they are experiencing a transcendent Reality? If not, how
can Hick claim to respect existing religious discourse? Plantinga argues persuasively
that the facts are against Hick here: very few religious believers ‘accept anything like
Hick’s pluralism’.24
The distinctive feature of Hick’s pluralism is that, despite their differences, actual
religions are all true. If Hick fails to capture the self-understanding of actual believers,
his pluralism collapses. This is whether AP departs from Hick most starkly. AP offers
an alternative account of the ultimate metaphysical and moral reality. For Hick, all
claims about God, ultimate reality, cosmic purpose, or divine relations with humans
relate only to human perceptions of the Real, not to ultimate reality as it is in itself. By
contrast, AP’s central claims are about ultimate reality: there is a cosmic purpose; it is
not human-centred; if there is a God, then God is not benevolent; and so on. The
dispute between AP, BT, and (religious) atheism is not merely a difference of
perception. AP does not claim to preserve the truth of all major religions. Where
they conflict with AP, other views are false. (Epistemically reasonable and morally
transformative, perhaps, but false nonetheless.)
9.7 Does Religious Diversity Undermine Mystical

Doxastic Practices?
In chapter 5, we did not treat mysticism primarily as something to be explained, but
rather as an alternative mode of epistemic access to cosmic purpose. Does RD
undermine that earlier argument? As in chapter 5, we focus on Alston’s epistemology
of doxastic practices.25 Alston argues that it is rational to treat Christian mystical
practice (CMP) as prima facie reliable. He concedes that RD threatens CMP. Recall
that, for Alston, CMP is not threatened either by the existence of other reliable
doxastic practices, or by the fact that CMP is a minority practice. RD only threatens
CMP if two familiar conditions are met: substantive conflict and epistemic parity.
Competing mystical doxastic practices must make incompatible claims about
important matters. This conflict often arises, not from mysticism itself, but from
the background claims that support each contemplative tradition. (Consider the
23
The argument in this paragraph draws on Alston, Perceiving God, pp. 264–5; and Plantinga,
Warranted Christian Belief, pp. 43–63, and 437–57.
24
Plantinga, Warranted Christian Belief, p. 438, footnote 18.
25
Alston, Perceiving God, pp. 255–85.
 THE CASE AGAINST BENEVOLENT THEISM
conflicting historical claims of Christianity and Islam.) And those conflicting prac-
tices must enjoy epistemic credentials on a par with CMP’s own.
The threat is now obvious. If CMP is confronted by several equally respectable
doxastic practices that offer conflicting verdicts, then CMP is undermined by the
weight of opposing verdicts. Its prima facie reliability is rebutted. And each other
mystical doxastic practice suffers the same fate. RD undermines them all.
This is very similar to our earlier debate about explanation. But the two debates are
slightly different. Some explanations will automatically dissolve the threat to
CMP. Most BT explanations deny either substantive conflict (Hick) or epistemic
parity (Plantinga). Atheist explanations, on the other hand, ground a different (and
more serious) threat to CMP. (If there is a deflationary explanation of all religious
experience, then CMP is undermined by SP [our sensory doxastic practice] before
RD enters the picture.) Our present question is whether an explanation of RD that
acknowledges both substantive conflict and epistemic parity, as AP does, will also
undermine CMP.
Alston himself defends CMP in the face of both substantive conflict and epistemic
parity.26 Indeed, it is substantive conflict that leads Alston to focus on a specifically
Christian mystical doxastic practice in the first place. (The conflicting background
claims of different religions would lead any putative unified mystic practice into
incoherence.) And Alston explicitly concedes epistemic parity, if only for the sake of
argument.27 Practitioners of any established mystical tradition can reasonably regard
it as prima facie reliable.
Alston argues that, while RD may reduce the practitioner’s faith in CMP, it is not
irrational for her to stick with CMP. It would be irrational to persevere in the face of
conflict with a superior practice. I should abandon alchemy once I discover chemis-
try. But no other mystical practice is superior to CMP. (As I argued in chapter 5,
following Alston: while SP may be superior to CMP, SP itself does not undermine
CMP.) RD thus gives the Christian no reason to abandon CMP. She can reasonably
sit tight.
Alston adopts the first-personal perspective of the practitioner of CMP and her co-
religionists. He asks whether the Christian must abandon her commitment to
CMP. Our interest is third-personal. Is it rational for the impartial outsider to
defer to Christian mystics? Suppose we agree that all mystics can rationally stay
where they are, and that co-religionists can rationally defer to their own mystics. This
does not help the outsider with no existing connection to any specific mystical
tradition. We cannot, on pain of self-contradiction, defer to all of them. And surely
it is arbitrary to select one doxastic practice (at random?) and defer (only) to it.
26
In Perceiving God, Alston discusses substantive conflict and epistemic parity at pp. 259–66 and
266–70, respectively. For debate, see Quinn, ‘Is Alston’s Response to Religious Diversity an Overstated
Case?’; and Alston, ‘Response to Quinn’.
27
RELIGIOUS DIVERSITY 
Unlike Hick’s pluralism, AP regards all non-AP religions as mistaken. CMP, other
BT mystics, and even Buddhist atheists all profoundly misperceive ultimate reality.
But AP can still borrow from these mistaken practices. Mysticism is misperception,
not delusion. Once we conclude that a doxastic practice essentially involves a deeply
mistaken understanding of its subject matter, then we cannot ourselves become
insiders in that practice. And there may be sociological or psychological reasons
why a mystical tradition cannot arise except within a religion that promises individ-
ual salvation (as both BT and Buddhism do, albeit in very different ways). Perhaps we
could never create a truly AP mystical practice. But it does not follow that AP must
abandon mysticism. Suitably reinterpreted—through a lens that their practitioners
would emphatically reject—the deliverances of CMP and Buddhism may teach us
valuable (and irreplaceable) lessons about cosmic purpose. AP’s partly deflationary
explanation of RD thus does not undermine chapter 5’s argument from mysticism to
cosmic purpose. While RD provides a new argument against BT, it leaves AP’s case
against atheism intact.
In addition to constructing an argument for cosmic purpose, chapter 5 also
suggested that AP could borrow the metaphysical and moral insights of mystics.
Does religious diversity undermine this more specific use of mysticism? When
mystical doxastic practices disagree, we cannot consistently borrow from all of
them. And we have accepted, contra Hick, that different religions do disagree. AP
has two alternatives at this point. First, we can use independent metaphysical
arguments (cosmological, fine-tuning, or evil) to eliminate broadly erroneous mys-
tical traditions (notably those that are either atheist or BT), and then borrow the
insights of those traditions that remain. This alternative has a high price, however, as
it greatly reduces the value of mysticism. For instance, Buddhism and Christianity
could teach us nothing. Furthermore, even within the class of non-atheist non-BT
mystical doxastic practices, we are still very unlikely to find unanimity.
A more promising strategy for AP is to borrow from all mystical doxastic practices
in so far as they agree. If perennialism were correct, then this would be straightfor-
ward. The central insight of all mystical experience would be always and everywhere
the same. Our preference for contextualism complicates matters. Nonetheless, agree-
ment is still possible. Some abstract metaphysical claims do seem to be common to
most mystical traditions: the ultimate reality is simple, unsurpassable, worthy of
worship or respect or emulation, and so on. More importantly, even though mystical
traditions disagree radically about metaphysical or doctrinal details, they all recog-
nize correct mystical experience using very similar moral tests. All mystics defend a
very demanding self-transcending counter-intuitive morality. As this connection
between mysticism and morality was the main lesson of chapter 5, we can conclude
that religious diversity does not undermine that chapter’s central conclusions. AP can
still borrow moral insights from mysticism.
10
Immortality
The popular touchstone for all philosophies is the question, ‘what is their bearing
on a future life?’
[William James, quoted in Edwards (ed.), Immortality, p. 181]
If immortality is to give us an assurance or a hope of progressive improvement, it

can only be if we have reason to believe that the interests of spirit are so
predominant a force in the universe that they will find, in the long run, satisfac-
tion in the universe.
[J. McT. E. McTaggart, Some Dogmas of Religion, p. 135]
There are objections to the doctrine of pre-existence. But it seems to have been
invented with the good intention to save the honour of the Deity, which was
thought to be injured by the supposition of his bringing creatures into the world
to be miserable, without any previous misbehaviour of theirs to deserve it.
[Benjamin Franklin, quoted in Givens, When Souls Had Wings, p. 189]
The question of immortality seems to help AP against BT. Any adequate theodicy
needs immortality; while AP can remain neutral. Immortality is a controversial
commitment that BT needs and AP does not. If BT can prove immortality, then
honours are even; otherwise, this issue supports AP.
Unfortunately, things are not so simple. BT does need immortality. But perhaps
AP needs it too. BT can then turn the tables. Immortality is to be expected if there is
an omnipotent benevolent God, but without such a God it is vanishingly unlikely. Its
inability to support immortality thus counts against AP. And if atheism can survive
without immortality, then immortality weakens AP’s credentials relative to both its
main rivals.
This chapter is in two parts. Sections 10.1 and 10.2 address the metaphysical
demands of BT; while section 10.3 asks what morality demands. The conclusion of
sections 10.1 and 10.2 is that the demands of BT are excessive. Although the meta-
physical arguments themselves are complex and inconclusive, human mortality is the
most natural conclusion given our prior commitments. Indeed, one motivation for this
project is to ask what cosmic purpose might look like without human immortality.
A corollary of sections 10.1 and 10.2 is that immortality is very unlikely unless BT
is true. If AP is true, then we probably do not survive death. AP thus needs to avoid
IMMORTALITY 
the moral arguments of section 10.3, which argue that immortality is essential to
morality. I argue that, while our project must take such arguments seriously, the
metaphysical demands of BT outstrip those of morality. AP can satisfy the latter
while rejecting the former.
My secondary goal in this chapter is to illustrate the value of exploring neglected
intellectual traditions. Moral arguments for immortality seem strange to contempor-
ary analytic philosophers. But they were very prominent in the philosophical school
that immediately preceded our own. Several key arguments in part I drew on Idealist
themes—notably Leslie’s axiarchic plenitude in chapter 3 and Anselm’s Boethian
metaphysics in chapter 6. In this chapter, we turn our attention to British Idealism.
Although not fashionable today, Idealism dominated the philosophical scene to an
extraordinary degree in the late nineteenth and early twentieth centuries, and its
demise arguably owes more to intellectual fashion than to the force of its opponents’
critiques.1 When we seek metaphysical alternatives to contemporary global natural-
ism, Idealism is a natural place to start. And, as we’ll see, the Idealists’ moral and
metaphysical commitments have much affinity with AP. AP is not committed to
Idealism, and most British Idealists explicitly endorsed either BT or atheism. But AP
can fruitfully borrow from Idealism, and perhaps even mutually support it.
10.1 What Afterlife Does BT Need?

In this earthly life, some people’s lives are rendered not worth living by horrendous
evils inflicted by others, or by natural evils. An afterlife allows those evils to be
compensated, outweighed, or redeemed within the individual’s own life. If human
beings never survive death, the evils and inequalities of this world cannot be
reconciled with the existence of a loving, benevolent God. A BT God could create a
world where each creature has only one life, but only if every life was rich and
flourishing. And that is not a plausible description of our world.
All BT religions posit an afterlife; and all extant theodicies appeal to that afterlife.
But what must BT’s afterlife be like? Every religion provides many details about its
afterlife, along with many mysteries. What is essential to BT? I will argue that BT
needs personal survival of death, and individual pre-existence before birth, for all
human beings and all sentient animals. I therefore extend the traditional demands of
BT, which often limits itself to future survival for humans. The need for personal
survival flows from BT’s commitment to divine benevolence, an afterlife for animals
is needed to rebut chapter 8’s argument from animal suffering, and the case for pre-
existence draws on the liberal consequentialism introduced in chapter 8. My claims
1
For an excellent overview of British Idealism, and its intellectual dominance, see Mander, British
Idealism. On its demise, see Mander, British Idealism, ch. 15; and Skidelsky, ‘The Strange Death of British
Idealism’.
 THE CASE AGAINST BENEVOLENT THEISM
about BT’s commitments thus rest on the conclusions of earlier chapters. I make no
pretence of moral neutrality.
10.1.1 Does BT require personal survival?

As we saw in chapters 7 and 8, BT typically insists that, even if God is concerned for
the impersonal value of outcomes, God is also person-affectingly benevolent: seeking
what is best for each individual human being. And any benevolent God must at least
ensure that each person’s life goes reasonably well overall. It is not sufficient that a
greater good arise somewhere to outweigh horrendous evil. That good must com-
pensate the person who suffers the horrendous evil. If the greater good occurs post
mortem, then it must happen to the same person. Therefore, an adequate theodicy
requires that individual human beings survive death. To distinguish itself from AP,
BT needs personal survival. While this requirement is a commonplace of most BT
religions, it will prove very significant later on, especially during our exploration of
personal identity in section 10.2.
10.1.2 Does BT require immortality?

Philosophers often discuss, not the afterlife, but immortality. Does BT need an
afterlife of infinite duration?
One might think that immortality is both sufficient and necessary to erase hor-
rendous evil. Earthly suffering cannot create infinite disvalue. If we all enjoy an
afterlife of infinite positive value, then every human existence has the same infinite
value. No amount of earthly suffering reduces the total value of one’s existence. Even
horrendous evil is not bad for those who suffer it, and the argument from evil
collapses. Conversely, without immortality we cannot hope to enjoy infinite value,
and horrendous evil remains undefeated.
Unfortunately, this argument fails, for reasons familiar from the debate over
infinite utility.2 Any plausible aggregative principle for immortal lives must imply
that, if any two lives are identical at some times, and if one is better at all times when
they differ, then the latter is a better life. If x and y enjoy identical afterlives, and x’s
earthly life includes horrendous evil while y’s does not, then x’s overall existence is
worse than y’s. Horrendous evil is still bad for x.
Infinite duration alone does not eliminate horrendous evil. Redemption requires a
greater good that transforms the evil into something that is itself good. But then
redemption may also be available in a finite afterlife. As we saw in chapter 5,
transcendent goods cannot require infinite duration, because they are actually avail-
able in this life. A moment’s direct experience of God might cancel out any amount of
terrestrial suffering. Of course, not everyone does experience God in this life. This is
2
This solution to the infinite utility puzzle is from Vallentyne and Kagan, ‘Infinite Value and Finitely
Additive Value Theory’. See also Mulgan, ‘Transcending the Infinite Utility Debate’.
IMMORTALITY 
why BT needs an afterlife. But God could easily arrange for universal divine experi-
ence in a finite afterlife.
Immortality is neither necessary nor sufficient to eliminate horrendous evil.3 BT is
thus not logically committed to immortality, unless the latter is required by morality
itself. While most actual BTs do defend immortality, this particular commitment is
optional.4
10.1.3 Does BT need an animal afterlife?

Thus far, I have followed BT orthodoxy. I now depart from it quite radically in two
ways. My first departure concerns animals.5 I argued in chapter 8 that an adequate
theodicy must include an afterlife for all sentient animals, especially those who
suffer agony. Our reluctance to treat human beings as special, and our utilitarian
insistence on the significance of suffering, mean that redemption (or at least
compensation) must be available to non-human animals. And this typically does
not occur in this life.
The need to include animals is not metaphysically problematic. If God can ensure
human survival, why not do the same for animals? (One possible difference is that
God may be unable to offer animals redemption. We return to this difficulty shortly.)
But it does raise the metaphysical price of BT. In particular, this is one place where
the demands of BT exceed those of morality.
10.1.4 Does BT need pre-existence?

My second departure from standard BT is that an after life is not sufficient. BT must
also posit prior lives. In chapter 8 we objected, not only to the amount of evil in the
world, but also to its distribution. Suffering is distributed very unequally along
morally irrelevant lines. Many innocent people suffer horrendous evils, while many
3
Another argument for immortality is that it is essential to BT because it is essential for a meaningful
life. We return to this argument in section 10.3. If immortality is necessary for meaning, then AP will need
it just as much as BT.
4
Some opponents will argue that BT is committed to immortality. One possible argument here is that, if
human immortality is one of God’s options, and if an eternal life is better than a finite one, then any
perfectly benevolent God will make humans immortal. (BT has several possible replies. As we saw in
chapter 7, BT can deny that God’s perfect benevolence takes this maximizing form. Also, even if perfect
benevolence implies that God will choose immortality if it is available, it doesn’t follow that perfect
benevolence requires the possibility of immortality. A benevolent God could still create humans who could
only enjoy finite lives if that is the best possibility.) A related argument draws on Rowe’s argument that a
morally unsurpassable God must create unsurpassable lives. If every possible finite human life is surpass-
able by a better finite life, then God can only create infinite lives. BT thus would be committed to the
possibility of immortality. While these are genuine difficulties for BT, I set them aside in the text, to focus
on more general issues surrounding the afterlife.
5
Although an afterlife for animals is not Christian orthodoxy, it has its defenders in recent philosophy,
notably Dougherty, The Problem of Animal Pain. I discussed Dougherty’s theodicy briefly in chapter 8, and
return to it briefly in footnote 8 below. (I am grateful to an anonymous for referee for alerting me to
Dougherty’s book, which appeared only while I was making final revisions to my manuscript.)
 THE CASE AGAINST BENEVOLENT THEISM
guilty people enjoy very pleasant lives. If we have contra-divine free will (CDF), then
perhaps not even God can eliminate all horrendous evil. But God could eliminate
undeserved horrendous evil. In a just world, suffering would be distributed according
to desert, and only those who deserved to suffer would do so. If it contains
undeserved horrendous evil, then in this crucial respect our world is not what God
would create.
Here is how God could eliminate undeserved horrendous evil. Imagine two
otherwise identical worlds, R-World and S-World. In each, many people suffer in
ways that cannot be justified given the person’s behaviour in this lifetime. In S-World,
each individual lives only once. Much suffering is undeserved. In R-World, each
individual is reborn many times. One’s fate in each life depends on one’s actions in
previous lives. All suffering is deserved.
R-World is more just than S-World. And there is no other morally significant
difference. Both worlds contain the same pattern of lives. If desert has any value, then
R-World is better. God is traditionally defined as perfectly just and perfectly ben-
evolent. Any God who attaches any significance to justice for humans will prefer
R-World. And undeserved horrendous evil is presumably worse for the individual
than deserved horrendous evil. So any benevolent God will also prefer R-World.
R-World and S-World are two epistemic possibilities for our actual world. If God
created this world, and if R-World is possible, then we are living in R-World. There
are only three possibilities: either rebirth is actual; or rebirth is logically impossible; or
God does not exist. If rebirth is logically possible but not actual, then BT’s God does
not exist. BT must either defend the cycle of rebirth, or argue that it is logically
impossible.
I shall argue that BT cannot dismiss rebirth. It certainly seems absurd to claim that
rebirth is impossible for any possible rational creature. Surely God could create
rational beings who experience a cycle of distinct lives. BT must therefore limit its
impossibility claim to human beings: rebirth is impossible for creatures like us.
(Perhaps because the conditions for our continuing individual identity are tied to
this life.) As in our discussion of different kinds of freedom in chapter 8, BT now
faces an obvious objection: if rebirth is impossible for humans, won’t a benevolent
God create other rational beings instead? Taken together, our discussions of freedom
and rebirth suggest that God has three options: (a) create humans who suffer
horrendous evil and are not reborn; (b) create other rational beings who enjoy the
most valuable freedom that is not contra-divine (F2) and never suffer horrendous
evil; or (c) create other rational beings who are reborn and therefore never suffer
undeserved horrendous evil. Liberal critics of BT will insist that, if rebirth is not
possible, then a benevolent God would prefer not to create human beings at all.
Without rebirth, our world is simply too unjust. God will instead create creatures
who never perform evil, or different creatures for whom rebirth is possible. If this
liberal objection succeeds, then BT must defend the logical possibility of rebirth. But
once rebirth is possible, God will surely choose it.
IMMORTALITY 
Something akin to rebirth is thus a necessary component of any theodicy for

humans. It is also an excellent way to justify the agony of non-human animals, and
their limited capacity for flourishing. Animals suffer, and enjoy less rich lives, as a
result of their past misdeeds or unwise choices, rather than cosmic injustice. Our
cycle of rebirth will thus include both humans and other animals.
While BT theodicy needs something like rebirth, rebirth per se is not essential.
What theodicy actually needs is pre-existence: the claim that present humans (espe-
cially those who suffer horrendous evil) have lived before this life. The most obvious
mechanism for pre-existence is a cycle of rebirth in this world. But pre-existence has
also often been imagined without rebirth. (Consider the Platonic tradition, where the
soul’s natural home is a spiritual realm, and we descend to this terrestrial realm only
once.) This distinction between rebirth and pre-existence matters, because some
objections to rebirth do not rule out pre-existence per se. (In particular, as we’ll
soon see, while rebirth apparently conflicts with contra-divine freedom, pre-existence
supports such freedom.)
The argument that a just God would favour pre-existence is not new. It can be
found in all cultures where belief in rebirth is common. Pre-existence has often been
offered, and defended, as a theodicy.6 Nor is it unknown in the Western BT tradition,
where its most famous Christian proponent is Origen:
When He in the beginning created those beings which He desired to create . . . He created all
whom He made equal and alike . . . But since these rational creatures themselves . . . were
endowed with the power of free-will, this freedom of will incited each one either to progress
by imitation of God, or reduced him to failure though negligence. And this . . . is the cause of
the diversity among rational creatures, deriving not from the will or judgment of the Creator,
but from the freedom of the individual will.7
However, having been declared a heresy, pre-existence fell out of favour in our
philosophical tradition. I argue that, in light of modern liberal utilitarian values,
the time has come to reconsider that decision.
BT has several options here. First, BT can embrace my argument, and argue that
God has chosen pre-existence. I shall argue that this is the best BT response. Second,
BT could deny that pre-existence is possible for human beings, and then defend God’s
decision to create us rather than non-human rational beings who never suffer
undeserved horrendous evil. Third, BT could deny that pre-existence is desirable.
Because it is redeemed in the afterlife, undeserved horrendous evil is not bad for the
individual at all. God thus has no reason to eliminate undeserved horrendous evil.
6
Two excellent overviews are Obeyesekere, Imagining Karma; and Givens, When Souls Had Wings.
7
Origen, quoted in Givens, When Souls Had Wings, pp. 95–6. On Origen generally, see Coppleston,
History of Medieval Philosophy, p. 22; Pelikan, The Christian Tradition, vol. 1, p. 96, 151; Givens, When
Souls had Wings, pp. 91–8. Rebirth-based theodicies are also common among the Druze (Obeyesekere,
Imagining Karma, p. 315); various Shi’ite sects such as the Nuzayriyah and the Alevis (Obeyesekere,
Imagining Karma, p. 316); and in Mormonism (Givens, When Souls had Wings, p. 219).
 THE CASE AGAINST BENEVOLENT THEISM
We return to the possibility that pre-existence is impossible in section 10.2, as it

depends on one’s account of personal identity. We now ask whether BT can deny the
desirability of pre-existence. The idea that God redeems horrendous evil, rather than
merely compensating it, is a staple of BT apologetics, as we saw in chapter 8. Given
our limited understanding of the nature of possible goods, the possible forms that
future human life might take, and the powers available to God, we cannot rule out the
possibility of redemption.
If BT only offers a defence, then it is reasonable to cite the bare possibility of a
redemptive afterlife. But our BT seeks a theodicy. Redemption must be likely, and not
merely possible. The standard sceptical BT moves that support defence will under-
mine theodicy: how can we know that redemptive goods are (probably) available?
Our BT is also constrained by our prior moral commitments, especially a utilitar-
ian aversion to suffering, a liberal consequentialist aversion to avoidable injustice,
and a broader consequentialist refusal to overvalue freedom relative to suffering.
These commitments make it harder to claim, as BT must, that individual post-
mortem redemption entirely removes the disvalue of the unequal and unjust distri-
bution of pre-mortem suffering.
Our first departure from BT orthodoxy also make BT’s task more difficult.
Theodicy now has a broader scope, because it must offer redemption to animals.
We touched on this difficulty in chapter 8. Most extant accounts of human redemp-
tion involve the victim coming to understand her suffering, its justification, and its
place in her life as a whole. Is such understanding available to non-human animals,
especially those with limited cognitive capacities? Can God increase an animal’s
capacities in the afterlife, rendering it capable of redemption, without obliterating
its identity as this individual animal? On some accounts of identity, redemption will
be impossible for animals. Therefore, God will want to avoid undeserved animal
suffering. If God has created this world, God must have chosen pre-existence for
animals. And if the role of pre-existence is to link an animal’s present suffering to her
past decisions, then her previous lives must involve rational agency.8
Pre-existence reduces the comparative plausibility of traditional redemption stor-
ies, because it offers a richer alternative. The usual contrast, as we saw in chapter 8, is
between a world with redeemed horrendous evil, and one lacking horrendous evil
altogether. BT objects that the latter must lack greater goods such as contra-divine
freedom. By definition, a world without horrendous evil is very unlike the actual
8
For instance, Dougherty presents a theodicy for animals based on the Eastern Christian notion of
deification (Dougherty, The Problem of Animal Pain). In the afterlife, both animals and humans are raised
to a superhuman level of moral and spiritual awareness, enabling them to place their earthly suffering in
the wider context of God’s love for creation. AP need not deny that such post-mortem mental improve-
ment is one logical possibility—even though it creates serious difficulties for any materialistic story about
the continuity of animal identity. The real question is whether post-mortem deification represents God’s
best option overall. In particular, does undeserved horrendous evil play any useful role? As I suggested in
chapter 8, Dougherty’s argument at this point appeals to non-consequentialist moral principles that our
project reasonably rejects.
IMMORTALITY 
world. So BT can plausibly claim that this (allegedly better) imaginary world lacks
some actual good. By contrast, pre-existence offers the prospect of a world containing
all the goods of this world, including contra-divine freedom, as well as some
horrendous evil. If horrendous evil is essential to some greater good, then pre-
existence can accommodate that good. Instead of positing a better non-actual
world without horrendous evil, pre-existence allows the opponent of redemption to
offer an alternative interpretation of the actual world. BT’s task is now much harder.
If it seeks to avoid pre-existence, BT must now link its greater good to undeserved
evil. And BT must therefore insist that, if it turns out that actual horrendous evil is
deserved, that would somehow make things worse!
If BT does persuade us that undeserved horrendous evil is essential to some greater
good, the defender of pre-existence can change tack. A world without pre-existence is
now deficient because many people are deprived of that greater good—because they
do not suffer undeserved horrendous evil. A suitable cycle of rebirth could remove
this deficiency, enabling God to ensure that everyone suffers undeserved horrendous
evil in some lifetime or other. The opponent of BT can thus offer a compelling
disjunctive argument. Either undeserved horrendous evil can be redeemed, or it
cannot. If not, then God will not choose a world where anyone suffers undeserved
horrendous evil. This can hold true in the actual world only if humans enjoy pre-
existence. If redemption is possible, then God will choose a world where everyone
suffers undeserved horrendous evil. This can hold true in the actual world only if
humans are trapped in a cycle of rebirth. Either way, BT is committed to some
doctrine of pre-existence or rebirth.
Religious diversity nicely illustrates this pattern of argument. As we saw in
chapter 9, an adequate theodicy must also explain the existence of (very widespread)
religious error. If BT is true, then most people base their lives on false religious (or
non-religious) beliefs about the world and their place in it. A benevolent God would
only permit religious diversity if it were necessary for some greater good. For a
person-affectingly benevolent God, that good must be located within the individual’s
own life. Her false belief must (somehow) be good for her.9 Perhaps true religious
knowledge is better if it arises out of reflection on one’s own past errors. For most
people, this transformation never occurs in this life. Therefore, BT must posit a
posthumous religious conversion. However, as with horrendous evil, this creates a
dilemma for BT. If my current false beliefs are necessary for some greater good that
I will enjoy in the next world, doesn’t your (earthly) true faith then prevent you from
enjoying that very good? If so, wouldn’t we expect a truly benevolent God to ensure
that everyone enjoys the benefits of false religion? Current true believers must have
been unbelievers in a previous life.
9
An alternative is that the individual’s false beliefs are inextricably linked to something else that is good
for her, such as her freedom. However, in chapter 9, I rejected all attempts to explain religious diversity by
reference to human freedom, whether directly or indirectly.
 THE CASE AGAINST BENEVOLENT THEISM
If it still insists on denying the desirability of pre-existence, BT must argue that it

clashes with some other divine goal. The most likely objection is that, like other
proposals to avoid evil, pre-existence is inconsistent with CDF. To explore this
objection, we must ask what exactly pre-existence involves. Obeyesekere helpfully
distinguishes between ethicized and non-ethicized rebirth.10 Historically, the former
usually comes first. The cycle of rebirth is a purely natural phenomenon with no moral
dimension. In ethicized rebirth, by contrast, rebirth tracks desert so that everyone gets
what they deserve in each life. (Under non-ethicized rebirth, this is staggeringly
unlikely!) To eliminate undeserved horrendous evil, we need ethicized rebirth.
But now suppose human beings have CDF regarding horrendous evil. (If not,
theodicy fails anyway, as we saw in chapter 8.) A perfectly ethicized rebirth mechanism
must ensure that I get what I deserve in this life. But my earthly fate depends, in part, on
the actions of other human beings. To match my fate to my deserts, the rebirth
mechanism must perfectly predict others’ behaviour. If perfect predictions are possible,
then God can make them too. But if God can predict all human behaviour, then God
can choose between possible worlds—and thus avoid all horrendous evil. Perfectly
ethicized rebirth contradicts CDF. As theodicy requires CDF, it must reject rebirth.
BT cannot avoid pre-existence so easily. The defender of pre-existence has several
replies. One option is to reject CDF altogether. In chapter 8, we assumed that any
satisfactory theodicy needs CDF. But perhaps rebirth is an alternative to CDF. God
foresees all horrendous evil but does nothing to prevent it—because horrendous evil
serves some greater purpose and is always and everywhere deserved. When choosing
between worlds where all horrendous evils are deserved, God opts for a purely
deterministic world where some such evils are used to teach human beings valuable
lessons. This option is worth exploring, especially given the controversy surrounding
contra-divine freedom. However, it does face the obvious objection that, if God’s
options include deterministic worlds with no horrendous evil, then we need a
powerful reason why God chooses to allow any evil. Surely even deserved horrendous
evil is still bad in itself?
A less provocative option is to deny the conflict between CDF and pre-existence.
This is where the distinction between rebirth and pre-existence comes into its own.
BT often rejects rebirth because it suggests determinism. But pre-existence has no
such implication. Indeed, many philosophers have defended pre-existence precisely
because it best accommodates libertarianism. Prior choices in a purely spiritual
world can be entirely free of any causal determination or interference. We would
then be responsible for our fate in this life even if we do not currently enjoy
incompatibilist freedom. Even if theodicy must reject rebirth, it can easily accom-
modate pre-existence.11
10
Obeyesekere, Imagining Karma, p. 75.
11
The fact that agents in a deterministic universe might have enjoyed incompatibilist freedom in a
previous non-physical existence means that no physical evidence could ever decisively refute libertarianism.
IMMORTALITY 
If it is offered as an account of the actual world, then pre-existence cannot entirely

remove horrendous evil. Horrendous evil does occur. But pre-existence can eliminate
all undeserved horrendous evils. God can ensure that only those who deserve
horrendous evil end up in places where others might inflict horrendous evil on
them. Perhaps God creates many different cosmic regions within the actual world.
In regions like this one, agents have contra-divine freedom to inflict horrendous evil.
Other regions are like chapter 8’s liberal-w3, where contra-divine freedom is reserved
for choices between goods. God then ensures that only those who deserve horrendous
evil end up in regions like this. Some who deserve horrendous evil escape, due to the
free choices of others. But all who suffer horrendous evil get what they deserve. God
chooses unmerited freedom from evil over undeserved horrendous evil. Surely a
merciful God is preferable to an unjust one—at least, given our liberal consequen-
tialist commitments.
Another possible explanation is that our past deeds merited only the (small) risk of
horrendous evil. (Perhaps those who unequivocally deserve horrendous evil end up
somewhere much worse.) This alternative has two advantages. First, it allows God to
avoid any appearance of arbitrary punishment. Second, if different degrees of
culpability merit different risks of evil, that would explain the fact that some creatures
are (statistically) more likely to suffer evil than others.
The imaginative use of pre-existence could thus enable God to eliminate the worst
kind of horrendous evil, rather than merely reducing it.12 If there is a benevolent
God, then there is no undeserved horrendous evil.
Dialectical context is crucial here. No doubt a defence can be constructed without
pre-existence or rebirth. It is possible that God allows some undeserved horrendous
evil for some reason we cannot fathom. But given the value commitments of our
project, and the need for theodicy rather than defence, the most plausible BT story
will include pre-existence. And BT has no compelling reason to deny that God would
choose pre-existence. Some BTs will cite scripture. But appeals to revelation should
be set aside here, for two reasons. First, I argued in chapter 9 that we should accept
epistemic and moral parity between different religions. Once we do so, revelations
This provides BT with an additional reason to endorse pre-existence. It also suggests that compatibilist
opponents of free will theodicy have very strong reason to reject pre-existence. Fortunately, chapter 8’s
disjunctive argument from evil did not rely on compatibilism, and therefore it is not undermined by the
possibility of spiritual pre-existence.
12
Our present objection to any theodicy without pre-existence thus sidesteps an objection to the
argument from evil put forward by Peter van Inwagen (van Inwagen, The Problem of Evil). Van Inwagen
concedes that some possible worlds contain fewer evils than the actual world. However, he argues that
there is no least number of acceptable evils. Whatever God does, God could have created a better world
with fewer evils. It is impossible to create an unimprovable world, and God cannot be faulted for failing to
do the impossible. (Van Inwagen’s argument has obvious affinities with Swinburne’s argument that there is
no best possible world, which we discussed in chapter 7.) However, as I argue in the text, the acceptable
number of undeserved horrendous evils is zero. So God can be faulted if God creates a world containing any
such evils.
 THE CASE AGAINST BENEVOLENT THEISM
supporting pre-existence will counterbalance those against. Second, the existence of

early Christian proponents of pre-existence casts doubt on the (modern) claim that
this doctrine is contrary to the Christian scriptures. The rejection of pre-existence
owes more to fashion and prejudice than to either reason or revelation. The rehabili-
tation of pre-existence thus fits the spirit of our project. I shall assume that, unless
pre-existence is impossible, a benevolent God would prefer it.
I conclude that BT needs an afterlife involving individual survival for all sentient
animals, together with some form of pre-existence to eliminate or reduce undeserved
horrendous evil. It requires all these in addition to contra-divine freedom.
These are the metaphysical commitments of BT. Are they plausible? BT needs the
afterlife, not as an abstract logical possibility, but as a human reality. Our question is
not whether there could be creatures who survive death, but whether we survive.
Section 10.2 addresses that question.
10.2 Do We Survive or Pre-exist?

Ideally, we would settle questions regarding the afterlife on independent grounds,
and then use the results to adjudicate between BT and its rivals. Opponents will hope
that independent evidence rules out BT’s metaphysical commitments, while BT will
trust that it establishes them. Unfortunately, as we’ll see, things are not so simple. Our
evaluation of both empirical evidence and metaphysical argument is inevitably
coloured by our other philosophical commitments. My goal is mutual support, not
decisive refutation.
This section explores two philosophical sources of information about the afterlife:
empirical evidence (section 10.2.1); and the metaphysics of personal identity (section
10.2.2). My aim is to evaluate the metaphysical claims required by BT, as set out in
section 10.1. BT requires a cycle of rebirth (or its equivalent) for both animals and
humans. Is this a credible interpretation of the actual world?
One immediate complication is that, as we saw in section 10.1, BT could take a
different approach, arguing that pre-existence is impossible for humans, and there-
fore that it isn’t numbered among God’s options. Although I cast doubt on this
alternative in section 10.1, it remains a live option for BT. Accordingly, we must also
ask whether, if actual pre-existence proves elusive, BT could instead argue that pre-
existence is impossible. These questions of possibility come to the fore in section
10.2.2. We begin, however, with evidence of actual human pre-existence or post-
mortem survival.
10.2.1 Evidence for immortality
The question whether human beings either survive death or pre-exist is empirical. So
why not simply answer it empirically? We begin with empirical evidence for survival.
Direct empirical evidence for an afterlife includes near death experiences, seances,
and so on. Each piece of evidence still has its defenders and detractors. But
IMMORTALITY 
philosophers tend to disregard it all. In the late nineteenth and early twentieth
centuries, the empirical investigation of psychic research was a relatively mainstream
intellectual activity. The Society for Psychical Research, in particular, set rigorous
standards for its investigations, prompted in large part by Henry Sidgwick.13 Evalu-
ations of the evidence have always been mixed, even from sympathetic interpreters
such as C. D. Broad.14 Contemporary philosophers are typically extremely negative.
Hales sums up the current attitude very nicely: ‘Well-known philosophers, such as
William James, C. J. Ducasse, C. D. Broad, Antony Flew, and H. H. Price who take
seriously [paranormal] matters are regarded as respectable but slightly nutty uncles
whose excesses are to be indulged with a wink and a smile.’15 Johnston is more blunt:
‘Most reliable reviews of [the evidence collected by the Society for Psychical
Research] find it is mostly weak and inconclusive.’16
The contemporary philosophical rejection of evidence based on paranormal phe-
nomena does seem compelling. Our commitment to explore forgotten metaphysical
positions is not a commitment to embrace pseudo-science. Questions about near-
death experiences, telepathy, and the like seem straightforwardly empirical, and the
evidence has not been forthcoming. We can reasonably reject these possible argu-
ments for the afterlife.
This rejection of direct evidence is not confined to atheists. Many BTs agree that
the evidence is weak. They reply that we should not expect evidence of the afterlife. If
God miraculously recreates our bodies somewhere else, or transports our souls to
heaven, then what direct evidence of survival could we possibly expect? BT philo-
sophers are largely untroubled by the absence of direct evidence for the afterlife.
The main positive empirical evidence against survival comes from the striking
correlations between mental phenomena and the brain. Consider the impact of brain
injuries on personality, and the depressingly familiar impact of age and disease
on our cognitive capacities. If consciousness diminishes as the brain diminishes,
surely it ceases when the brain dies. If an elephant were to stand on my head, this
would both extinguish my mental life and crush my brain. It is hard to believe these
two simultaneous effects are unrelated.
Many atheists take this correlation argument to be obviously decisive.17 On the
other hand, the close correlation between mind and body is so obvious that defenders
of survival have always been aware of it; and they have had millennia to present
alternative explanations. We first note that the argument from correlation is not
a direct argument against the afterlife. Rather, it is primarily an argument for
13
For an excellent overview of the society’s activities, Sidgwick’s role in them, and their relevance to his
own philosophy, see Schultz, Henry Sidgwick, ch. 5.
14
Broad, The Mind and its Place in Nature, especially pp. 532–45.
15
Hales, ‘Evidence and the Afterlife’, p. 335.
16
Johnston, Surviving Death, p. 132.
17
See, e.g., Edwards, Immortality, pp. 160–1. Edwards traces this atheist argument in its modern form
back to Joseph Priestly as early as 1777.
 THE CASE AGAINST BENEVOLENT THEISM
materialism about human beings, and then indirectly an argument against survival.
Dualists are unimpressed by correlation, arguing that changes to the body only affect
the soul while it is connected to the body and that this tells us nothing about the fate
of the soul once that connection is severed. (Consider the famous Platonic image of
the body as the prison of the soul, where conditions in the prison affect the soul but
death sets it free.) Therefore, the argument from correlation does not enable us to
sidestep the debates about personal identity explored in the rest of this section.
A second weakness is that, even if it succeeds, the argument from correlation only
demonstrates that death is naturally the end of our conscious existence. It cannot rule
out the possibility that God miraculously ensures our survival. Indeed, it is hard to
see how any empirical evidence could rule this out. Yet, as we’ll see, miraculous
divine intervention is the mechanism favoured by many BTs, including Christian
orthodoxy.
Empirical evidence thus fails to settle the question of the afterlife. The positive
evidence is unconvincing, but BT can do without it. The argument from correlation
persuades many. But, as we’ve just seen, BT can sidestep it. Furthermore, our
suspicion of naturalist imperialism, and our willingness to countenance non-natural
explanations, considerably weakens the argument—as we’ll see when we discuss
dualism in section 10.2.2.1.
Could empirical evidence help us decide about pre-existence? The main direct
evidence for pre-existence comes from reports of past lives. As with direct evidence of
survival, the evidence here is highly contested. Some find it compelling, while others
deride it.18 Most Western philosophers fall into the latter camp. One obvious
limitation of the evidence, even if we accept it, is that it covers a tiny minority
of people. Even if some fortunate individuals remember some past lives, most of
us remember nothing. But a successful theodicy requires more or less universal pre-
existence.
Other evidence for pre-existence is indirect. Pre-existence is often introduced as
the best (or only) way to explain certain facts about human knowledge or behav-
iour. Givens notes that ‘Pre-existence has been offered to account for why we know
what we should not know, whether in the form of a Greek’s slave’s grasp of
mathematics, the moral sense common to humanity, or the human ability to
recognise universals.’19 He adds that ‘many philosophers have found in human
pre-mortality the necessary precondition for a will that is genuinely free and
independent’.20 (Given the role of CDF in theodicy, and its apparent conflict
with rebirth, this last argument is especially significant.) And McTaggart famously
defended pre-existence as offering the best explanation for significant facts about
18
A contemporary philosophical defender is Robert Almeder. See, e.g., Almeder, Death and Personal
Survival; and the exchange in Hales, ‘Evidence and the Afterlife’ and Almeder, ‘On Reincarnation: A Reply
to Hales’.
19 20
Givens, When Souls Had Wings, p. 5. Givens, When Souls Had Wings, p. 6.
IMMORTALITY 
our personal lives, most notably the phenomenon of deep personal attachment
based on comparatively brief acquaintance.21
The evidence against pre-existence either reprises the correlation argument, or
cites the lack of general memories of past lives. However, most proponents of pre-
existence are dualists, and therefore unlikely to be moved by either argument. For
the dualist, as we’ll see, mental activity does not depend on a physical brain, and
memory is not essential for personal continuity. Once again, we cannot evaluate
the empirical evidence without addressing philosophical questions about personal
identity.
Our evaluation of the evidence regarding pre-existence is inclusive, and our prior
commitments and conclusions point in different directions. On the one hand, our
project supports the general idea that much in the human world lacks a fully
deflationary natural explanation. Furthermore, our commitment to methodological
pluralism, and our response to religious diversity in chapter 9, should make us
open to the presuppositions of non-Western religion and philosophy. (Philo-
sophers in non-Western traditions do not regard reports of past lives as obviously
absurd.)
On the other hand, AP cautions against the caprice of leaping to explanations that
treat humans as metaphysically special. Is my knowledge really so impressive that it
cannot be explained without citing my individual pre-existence? Proponents of AP
could reasonably side with the sceptics here, and simply reject pre-existence
altogether. But our approach throughout, faced with empirical controversy, has
been to remain agnostic wherever possible.22 The evidence may count against
survival and pre-existence, but it certainly does not rule them out altogether. And
much depends on our account of personal identity, to which we now turn.
10.2.2 Personal identity and post-mortem survival

To judge whether I do, or even could, survive death, we must first decide which
possible future events would count as my survival. Whether I can persist after death
depends on how I manage to persist before death. And that, in turn, depends on what
I am. I begin with a brief survey of the prospects for survival and pre-existence on
each of the most common accounts of personal identity.
I shall argue that most credible accounts of personal identity neither guarantee
survival or pre-existence nor rule them out. However, not all accounts are equally
friendly to all forms of survival. While personal identity cannot settle the question of
survival, it is relevant to it.
Our exploration of personal identity is driven by several distinct agendas. First, BT
definitely demands that individual humans survive death. Therefore, AP would like
21
McTaggart, Some Dogmas of Religion, p. 121.
22
The AP agnostic is in good company in this particular case. Agnosticism about the empirical evidence
for an afterlife has been defended by J. S. Mill, A. C. Ewing, and Bishop Butler.
 THE CASE AGAINST BENEVOLENT THEISM
to show that survival is impossible, while BT often seeks an account of personal

identity where survival is inevitable (and thus involves no extra metaphysical com-
mitment). On the other hand, some BTs argue that survival is all but impossible
without miraculous divine intervention. BT could then turn the tables on AP, arguing
that BT has a decisive advantage because morality requires survival. We must
therefore ask several different questions regarding each account of personal identity:
does it make human survival impossible, inevitable, likely, possible, or possible but
only with divine intervention?
Things are even more complex for pre-existence. I argued in section 10.1 that BT
needs pre-existence. If so, BT needs an interpretation of personal identity where pre-
existence is (at least) possible; while AP hopes to demonstrate that pre-existence is
impossible. However, BT’s alternative strategy involves the impossibility of pre-
existence; to combat that strategy, AP wants pre-existence to be possible but
extremely unlikely. A further complication is that, for reasons that will emerge in
section 10.3, AP wants to separate the demands of morality from those of BT by
arguing that survival is possible without pre-existence. We thus have several ques-
tions for each interpretation of personal identity: does it make pre-existence impos-
sible, inevitable, likely, possible, or possible but only with divine intervention? And
does it allow for survival without pre-existence?
BT’s God has creature-affecting concern for all sentient or rational beings. There-
fore, the scope of both survival and pre-existence must be universal. BT must argue
that all creatures survive death. This broader context sometimes becomes crucial.
However, for ease of exposition I will mostly concentrate on human personal identity.
Three distinct notions of ‘impossible’ are in play here. The strongest claim is that
survival or pre-existence is not possible for any rational being. We can set this strong
claim aside, because there is no reason to doubt that God could create immortal
reincarnated beings. Our two remaining senses concern human beings. When
opponents of BT claim that pre-existence is impossible, they mean it is not a possible
interpretation of actual human experience. By contrast, when BT claims that God
could not create humans who pre-exist, this claim covers all possible human histor-
ies. The two claims are clearly distinct. For instance, if memory of past lives is
necessary for personal continuity, then the fact that people do not remember their
past lives might rule out actual pre-existence while leaving the possibility of human
reincarnation in alternate possible histories untouched. (Although, as we’ll see in
section 10.2.2.3, the relationship between memory and pre-existence is not quite so
straightforward.)
... Continuity of soul Suppose we are immaterial souls, distinct from our
bodies. Personal identity requires continuity of soul. The death of my body is merely
an external event in the life of my soul. Survival is eminently possible, as are pre-
existence and rebirth. There is no logical reason why the same soul should not be
reborn in different bodies, or exist indefinitely without any body at all.
IMMORTALITY 
Philosophers have long argued that dualism implies, not only survival, but also
immortality.23 A simple spiritual substance has no parts, and therefore cannot
dissolve. If they aim to establish a logically necessary connection, such arguments
are too ambitious. Souls can be created or destroyed, and God could easily ensure
that each soul lives only once. Dualism does not guarantee any afterlife. But it does
ensure that God’s options include both survival and pre-existence. It is no surprise
that dualism has been popular within BT, and especially among defenders of pre-
existence.24
On the other hand, dualism will not help BT’s other strategies. Dualism does
nothing to render pre-existence impossible. Nor does it render either survival or pre-
existence so unlikely that they require divine intervention. If dualism is correct, then
AP could reasonably posit an afterlife if morality demands it. Finally, while dualism
permits both pre-existence and post-mortem survival, survival doesn’t necessarily
imply pre-existence. Immortal souls could come into existence. Pre-existence thus
remains a separate metaphysical cost of BT.
... Materialism Suppose we are material beings, and our identity requires
continuity of the same physical body (or perhaps some privileged body part such
as the brain). This person cannot possibly live without this body, nor can he be
reborn in different bodies. As Johnston notes, for the materialist: ‘Bodies are stuck in
this life, unless something very weird is happening at death.’25
Atheist materialists argue that survival and pre-existence are now simply impos-
sible. But many Christian philosophers embrace materialism about personal iden-
tity.26 S. T. Davis even describes materialism as the orthodox Christian position.27
(One can be a materialist about the criteria for personal identity without being a
materialist tout court.) Survival is still logically possible, it just requires miraculous
divine intervention. (Perhaps something very weird is happening at death!) Christian
materialists argue that this fits the traditional Christian belief in a general physical
resurrection. Without God, survival is vanishingly unlikely. But God could recon-
struct my body after my death. Immortality is a gift from God, rather than our
natural state.
Atheist materialists object that, once a person is dead, even God can only reani-
mate her body. This does not restore the original person to life. Personal identity
23
By 1474, Ficino could list fifteen alleged proofs of immortality! (Copenhaver and Schmitt, Renais-
sance Philosophy, p. 153).
24
For instance, the most prominent philosophical defender of pre-existence in recent times was
McTaggart, who argued that the metaphysical considerations that favour future existence apply equally
to the past, and therefore that the dualist who embraces survival should also accept pre-existence
(McTaggart, Some Dogmas of Religion, p. 127).
25
26
See, e.g., van Inwagen, ‘The Possibility of Resurrection’; and Murphy, Bodies and Souls, or Spirited
Bodies.
27
Davis, Risen Indeed, pp. 47, 86.
 THE CASE AGAINST BENEVOLENT THEISM
requires continuity of a living body or brain. The person only continues if she never
actually dies. Christian materialists reply that God could satisfy even this require-
ment, although this may require some fancy footwork. (Van Inwagen explores the
possibility that God replaces each person’s body with a duplicate at the moment of
their apparent earthly death, and then revives the person elsewhere.28) Other BTs,
and many atheists, object that such deceitful manoeuvring is unworthy of a perfect
being. Why hide our true nature from us? As Johnston complains, Christian materi-
alism turns God into ‘a deceiving trickster’.29
Rebirth is also possible under materialism. But it too requires something miracu-
lous. Whatever governs the mechanism of rebirth (whether God or the impersonal
laws of karma), it could in principle ensure that the essential physical elements of my
body are combined again to form the essential physical elements of another body.
This would be no harder for God than ensuring survival once.
On the other hand, materialism does rule out spiritual pre-existence, because there
can be no personal continuity without physical rebirth. I suggested earlier that
spiritual pre-existence, as opposed to rebirth within this world, may be necessary
to reconcile pre-existence and CDF. If so, BT materialists must deny the compati-
bility of pre-existence and CDF, and then reject the former because it conflicts with
the latter.30 If we still wish to insist that BT needs pre-existence, we have several
options. First, we can accept that materialism, pre-existence, and CDF are not all
compossible, and conclude that, of the three, it is materialism that must go. BT needs
pre-existence and CDF. It doesn’t need materialism. We’ve already seen that dualism
also meets BT’s needs. (And there may be other alternatives as well.) Unless dualism
is impossible, God will prefer it to materialism. If a benevolent God has created us, we
should conclude that we are spiritual beings.
Another option is to draw on our earlier conclusion that, even if perfectly ethicized
rebirth is inconsistent with CDF, God can still avoid all undeserved horrendous evil
by not placing any innocent being at the mercy of beings who might inflict horren-
dous evil. Only the undeserving are reborn in cosmic regions like this! While this may
seem metaphysically extravagant, it is certainly not beyond the reach of an omnipo-
tent God. God could reconcile CDF, materialism, and pre-existence.31
I conclude that materialism is consistent with both survival and pre-existence, and
could satisfy BT’s metaphysical needs. Our next question is whether BT could change
strategies, and argue that materialism renders pre-existence impossible. This seems
28
Van Inwagen, ‘The Possibility of Resurrection’. For discussion, and further references, see Johnston,
Surviving Death, pp. 90–109.
29
30
This may be one reason why Christian orthodoxy, which endorses both materialism and CDF, rejects
pre-existence.
31
A more extreme alternative for materialist BT is to abandon CDF. If all actual horrendous evil is
deserved, then God has sufficient reason to create this world without CDF. As I argued in section 10.1.4,
this option is worth exploring, but also very problematic.
IMMORTALITY 
very unlikely. Recall the different notions of impossibility introduced in section

10.2.2. Perhaps BT’s materialist opponents could claim that the combination of
pre-existence and CDF is not a possible interpretation of actual human experience.
But the apparent conflict between CDF and pre-existence would not justify the
stronger BT claim that no humans could possibly enjoy pre-existence. God could
certainly arrange a system of material rebirth that perfectly matches life prospects to
past deserts. Materialism does not support BT’s alternative strategy.
On the other hand, materialism does make both survival and pre-existence
extremely unlikely unless God miraculously intervenes. Therefore, materialism does
make it very difficult for AP to endorse any afterlife. If materialism is credible, then
AP must deny that morality demands survival.
... Psychological continuity Locke famously identified personal identity, not

with continuity of some underlying substance (either spiritual or material), but
with psychological continuity, especially memory. I am the person who wrote the
first draft of this chapter because I remember writing it.
On the simplest psychological account, continuity of memory is both necessary
and sufficient for personal identity across time. We consider sufficiency first. If
memory is sufficient for personal identity, then God could easily ensure survival by
planting my memories in a new human body, an animal, a disembodied soul, or a
digital computer.
The sufficiency claim faces two well-known objections. The first concerns dupli-
cation or fission. Suppose two bodies step out of the teletransporter, each claiming to
‘remember’ what I did today. If memory is sufficient, then both are identical to me.
But then, by transitivity of identity, they must be identical to each other. But how can
two distinct people be identical? The sufficiency claim leads to paradox.
BT could reply that speculation about purely logical possibilities carries little
weight. To threaten the sufficiency condition, duplication must actually occur. And
God can easily ensure that no one ever divides. This reply will not convince analytic
philosophers, especially as the literature on personal identity is littered with specu-
lation about things that will never happen. But perhaps it does lessen the force of the
fission objection.
The second familiar objection, first raised by Butler against Locke, is harder to avoid.
To apply the sufficiency claim, we must distinguish true memories from apparent ones.
Suppose two people both believe they wrote the earlier draft of this book. Each believes
he is the person who wrote it. But only one of us actually did write it. The other is
mistaken. The obvious answer is that the real memory is that of the person who
actually did the thing. But this threatens circularity. We cannot use memory as a
criterion of personal identity if we can only validate memories after we have individu-
ated persons. Therefore, we must conclude that memory is not sufficient.
If memory is not sufficient, then we need additional criteria. We examine these in
section 10.2.2.4. On the other hand, if memory is sufficient, then survival is clearly
 THE CASE AGAINST BENEVOLENT THEISM
possible. (We consider the prospects for pre-existence shortly.) And despite its
difficulties, the idea that memory is sufficient for identity is very popular, especially
among those sympathetic to fantasies of digital immortality where humans are
‘uploaded’ into computers. This raises the possibility that, even if human beings
have not survived death hitherto, they may survive in the future.32
We now turn to our second question: is memory necessary for identity? If memory
is necessary, that raises no difficulties regarding survival. God can easily ensure that
our future lives contain memories of this life. But memory is a significant prima facie
problem for pre-existence, because the vast majority of people do not remember any
past lives. This implies that pre-existence is not a plausible hypothesis about actual
humans. (Or actual animals, unless their memories are much richer than ours!)
This objection to pre-existence may seem decisive. But we should recall that, like
the existence of horrendous evil, the fact that virtually no one remembers their past
lives is not news. Defenders of pre-existence have had millennia to explain this
obvious fact. They do so by denying that memory is necessary for identity. Indeed,
if the necessity claim is proposed as a conceptual analysis, then the very existence of
billions of believers in rebirth refutes it. Some dualists deny the significance of
memory altogether. But most defenders of pre-existence seek a more modest role
for memory, one that is compatible with the evidence. One common model is that
memories of past lives are recovered in some future life.33 An individual goes through
a long series of lives (L1, L2, L3, . . . , Ln). In the final life (Ln), all previous lives are
remembered. Earlier lives are analogous to a series of dreams: each unrelated to the
others, but all remembered by the single waking self. (This metaphor is especially apt
under Idealist, Buddhist, or Neoplatonic accounts where our final state is akin to
waking from the dream of our earthly life.) The fact that some individuals do seem to
remember past lives thus supports pre-existence; while failures to remember past lives
do nothing to undermine it. (After all, sporadic memory is common within this life—
yet no one thinks it rules out identity here.) This model provides enough personal
continuity to ground moral responsibility across lives. If BT is true, then, for all
anyone knows, this is the model that God has chosen.
If memory is sufficient for personal identity, and if we permit sporadic memory
within lives, then our new model ensures both survival and pre-existence. If memory
is necessary, then these metaphysical options at least remain open. I conclude that,
whether memory is sufficient or necessary, it poses no direct threat to BT’s demands.
However, memory does support AP over BT. If personal identity is linked to
32
Digital ‘immortality’ raises many intriguing moral and philosophical issues. I address some in
Mulgan, ‘Ethics for Possible Futures’ and Mulgan, ‘Moral Philosophy, Superintelligence, and the Singu-
larity’, and I hope to consider them at greater length elsewhere.
33
This model draws on both British Idealism (especially McTaggart) and traditional Buddhist accounts
of the life history of a Buddha or Arahant. (See, e.g., McTaggart, Some Dogmas of Religion; Williams,
Mahayana Buddhism; Obeyesekere, Imagining Karma, especially p. 161.) Another possible model is that
identity across lives is preserved by psychological continuity without continuity of memory.
IMMORTALITY 
memory, then pre-existence is clearly harder to establish than survival, without being
impossible. This strengthens AP’s claim (developed in section 10.3 below) that pre-
existence is a metaphysical extravagance that takes BT far beyond the legitimate
demands of morality. Conversely, if memory is sufficient for identity, then AP can
reasonably posit future survival even if only by artificial digital means. (Unlike BT,
AP need not claim that humans have ever survived death in the past, merely that they
might do so in the future.) We return to this debate in section 10.3.
... The further fact view If memory is necessary but not sufficient, then there
must be other necessary conditions for personal identity. As Parfit puts it, we need
both psychological continuity and ‘some further fact’.34
One option is to combine memory with continuity of either soul or body. I could
cease to exist either from loss of memory or from the destruction of my body or soul.
(This is one natural interpretation of the dualist model of pre-existence sketched at
the end of section 10.2.2.3.) A more agnostic suggestion, inspired by Parfit, is that
personal identity requires memory with the usual cause, where that cause could be
either physical or spiritual. These accounts all inherit the merits and difficulties of the
simpler spiritual and bodily continuity views discussed earlier.
A more radical proposal is that God’s will provides the further fact—either because
God is the usual (direct) cause of continuity of memory, or because God’s will is itself
a condition of identity. S. T. Davis suggests that X at t1 is the same person as Y at t2
(in part) because God wills this to be the case.35 The criteria of human identity thus
depend on God’s plan for us. If that plan involves resurrection, then those criteria will
reach beyond death. Opponents dismiss this view as ‘identity mysticism’.36 But BT
should not dismiss it out of hand, After all, why shouldn’t a creator—especially one
who creates ex nihilo—determine or decree what counts as the continued existence of
this thing? We often defer to human artists when the question arises whether this
painting today is the same work of art as that painting yesterday. Why not defer to
God regarding the identity of God’s creatures?
The divine will view nicely illustrates the impossibility of a neutral resolution of the
personal identity debate! Whatever its merits within BT, this view is a non-starter for
atheism—unless it leads directly to the no-self view discussed in 10.2.2.7. (If God is
necessary for personal identity, and there is no God, then there are no persons.)
Similarly, AP cannot accept a divine will interpretation of personal identity. As with
the divine command theories of ethics discussed in chapter 2, AP can agree that
divine will would have provided the best interpretation of personal identity. But as
God has no plans for us, this cannot be what actually grounds actual human
34
Parfit, Reasons and Persons, p. 210. (Parfit’s original further fact is in addition to both physical and
psychological continuity.)
35
Davis, Risen Indeed, p. 120. While God’s will could be a sufficient condition on its own, Davis defends
it only in conjunction with psychological continuity.
36
Zimmerman, ‘Criteria of Identity and the “Identity Mystics” ’.
 THE CASE AGAINST BENEVOLENT THEISM
continuity. AP could then either argue that something else grounds personal identity
(in the absence of divine will), or reject personal continuity altogether.
Divine will is a double-edged sword for BT. On the positive side, it certainly makes
pre-existence and survival available to God. On the negative side, divine will rules out
any suggestion that pre-existence is impossible. More generally, by broadening God’s
options, the divine will interpretation makes it much harder to argue that actual
human history was God’s best possible option.
... Identity in the Absolute Idealism is a broad movement, offering several

distinct accounts of identity and immortality.37 Some Idealists are dualists. Each of
us has a single distinct immaterial soul that survives death with its individuality
intact—usually in communion with God or other immortal spirits. This position has
the same general advantages and disadvantages as dualism in general.
But Idealism also offers new models of survival. Many Idealists defend a more
collective metaphysic, where individual souls live on in a single undifferentiated
spiritual unity (the Absolute, the world-soul, God). We must distinguish two forms
of undifferentiated survival: dissolution where previously separate souls merge into
the Absolute after death; and realization where each soul discovers that it was
(merely) an aspect of the Absolute all along.
Under realization, both survival and pre-existence are guaranteed. Each of us is,
and always has been, an aspect of something eternal. If I am the Absolute, then I have
always existed, and will always exist in future. The question is whether this counts as
personal survival, or whether what I come to realize is that there are no persons at all.
(We return to that worry in a moment.)
Under dissolution, pre-existence is also straightforward. The soul can exist separ-
ately before this life. But survival is problematic. If continuing as a separate soul is
essential to my present identity, then dissolution into the Absolute ends my exist-
ence. Idealists reply that, far from ending my existence, dissolution is the fulfilment of
the soul’s destiny. Identity with the Absolute marks, not the loss of my personality,
but a radical expansion of it.
A central philosophical issue within Idealism is whether it is consistent with BT.38
Most British Idealists endorsed BT. Indeed, many were devout Christians. But many
BT critics of Idealism object that undifferentiated unity obliterates a distinction
between God and individual creatures that is essential to BT—along with the equally
vital distinction between creatures themselves. Without distinct individuals, there is
no place for charity or love. BT thus rejects realization. The individual soul is created
by God, and therefore cannot be co-eternal with God. BT also rejects dissolution. Even
37
On the different approaches to immortality within British Idealism in particular, see Mander, British
Idealism, chs 5 and 11 (especially pp. 158–60, 179, 434).
38
On the tension between British Idealism and BT, see Mander, British Idealism, chs 5 and 11
(especially pp. 153–6, 178–9, 417–20, 435–8).
IMMORTALITY 
if it eventually attains union with God, the soul must retain its individual identity. By
obliterating individual personality, undifferentiated unity deprives us of any mean-
ingful personal survival, and undermines the very idea of a benevolent God.39
Undifferentiated unity is especially troubling if BT posits a person-affecting God.
And I argued in chapters 7 and 8 that BT’s God probably does have some person-
affecting motivations. I therefore proceed on the assumption that BT theodicy seeks a
more robust mode of individual survival than anything available under undifferen-
tiated Idealism. Undifferentiated unity may also seem to spell disaster for AP,
because personal survival is necessary for morality. However, as I argue in section
10.3.10, AP’s strongest reply here is that truly impartial consequentialist morality
does not demand personal survival. AP can endorse full-blown Idealism.
We have therefore found one account of personal identity that, while preserving
both survival and pre-existence in some form, seems unlikely to meet BT’s needs. We
now explore two others.
... Identity as concern Another popular account identifies my self with my

concerns. Bosanquet even presented this as the common-sense view: ‘No one will
much object to an identification of the self in the main with the things we mainly care
for.’40 The widespread use of thought experiments in this area lends support to this
account. It is very plausible to suppose that our reactions to imaginary tales primarily
reveal what we care about. In teletransportation, fission, or body-swap tales, I am
really asking not ‘Is it me?’ but ‘Do I care?’ If these thought experiments also reveal
our views about identity, then the two must be closely linked.41
A concern-based account can deliver survival. If I care about things outside myself
that continue after my death, then their persistence is my survival. And, if I care
about the right things, I may even enjoy immortality. One possibility, suggested by
Bosanquet’s own discussion, is that our unity with the Absolute arises, not via
realization or dissolution, but through identification. We cease to care about our
individual personalities, and identify only with the Absolute. J. S. Mill famously
defended a secular alternative, summed up in his memorable metaphor of living on
‘in the ongoing rush of mankind’. Mill draws explicitly on Comte’s thought that: ‘He
who has left great results acquires in others a subjective immortality, so that the work
of his life is perpetuated and even extended.’42
Mark Johnston has recently offered a new reinterpretation of Mill’s secular
immortality.43 Personal identity is constituted by a certain kind of concern for
39
See, e.g., A. E. Taylor’s critique of Bernard Bosanquet in Taylor, The Faith of a Moralist.
40
Bosanquet, The Value and Destiny of the Individual, p. 260.
41
The suggestion that our intuitive responses to body-swap or fission tales primarily reveal our moral
concerns is itself likely to appeal to moral philosophers rather than to metaphysicians. This is therefore
another place where one’s philosophical judgement depends on one’s priorities.
42
Comte, System of Positive Polity, quoted in Mill, ‘Auguste Comte and Positivism’, p. 342.
43
Johnston, Surviving Death.
 THE CASE AGAINST BENEVOLENT THEISM
future personalities. Most of us feel this special concern only (or primarily) for
those future personalities that occur in this body, or contain apparent memories of
this personality. But this is a contingent weakness, a moral failing. A good person
would feel the very same concern for all future personalities. And therefore she
would live on in all of them.
Any story about concern can play two distinct roles. It can be either a substitute for
personal survival, or an interpretation of it. The former is less controversial. I come to
the belief that personal identity is not what matters. I cease to worry about my
survival, and focus instead on some other goal. On a concern-based interpretation, by
contrast, I literally live on—whether in the Absolute, or in God, or in others. By
identifying with something outside myself, I live on in it. This more ambitious role is
explicitly what Johnston defends. He introduces his own account with this promise: ‘I
will explain how a good person quite literally survives death.’44
Adopting a less self-centred pattern of concern is no doubt commendable, and
I argue in section 10.3 that it may prove sufficient for both morality and meaning.45
But this will not satisfy BT. While the individual may cease to care about herself, a
person-affecting benevolent God does not. My loss of interest in my own survival will
not satisfy God’s desire to compensate me for the evils I suffer in this life. And, of
course, a new pattern of concern provides no consolation to most victims of
undeserved horrendous evil, as they die before they have any chance to become
less egoistic.
This last limitation is also fatal to any theodicy based on literal concern-based
survival.46 Johnstonian survival is not open to all innocent sufferers of horrendous
evil, because almost none are Johnstonian good people. There is no consolation here
for the child who suffers horrendous evil and then dies before she can radically
transform her patterns of self-concern. Animal victims of horrendous evil fare even
worse, as they are completely unable to adopt the good person’s pattern of concern.
Finally, concern-based survival doesn’t offer pre-existence. The good people of the
past live on through us, but we ourselves did not exist before. And therefore our
suffering is undeserved.
BT must therefore reject any interpretation of personal identity in terms of
concern, and defend a more objective account that offers universal opportunities
for survival and pre-existence. By contrast, AP can easily embrace a concern-based
interpretation, so long as it is willing to deny that personal identity in any more
traditional metaphysical sense is necessary for morality. AP can also treat concern as
44
45
I have my doubts whether Johnston’s ingenious solution really captures the attitude of the good
person. Does she extend her self-concern to others, or replace self-concern with some alternative concern
that does not include identification? The latter might be equally praiseworthy (more so, if it opens the
possible of genuine self-sacrifice), but it cannot deliver Johnstonian survival (Mulgan, ‘Review of John-
ston’s Surviving Death’).
46
Neither Mill nor Johnston attempts any such theodicy.
IMMORTALITY 
a substitute for personal identity, based on a recognition of the cosmic irrelevance of

individual persons.
... The no-self view Perhaps there is no continuity of persons. There is no

separate metaphysical entity that continues from moment to moment. There is only
the continuous flux of experiences or moments of consciousness. This view is
common among Buddhists, has been attributed to David Hume, and owes its
prominence in recent analytic debate to Derek Parfit.47 One might be drawn to the
no-self view by the failure of some other account. If you believe that personal identity
requires a continuous immaterial soul, and then decide that there is no soul, you
might conclude that there can be no enduring persons.
No-self views fall into two camps. Eliminativists deny there are persons; while
reductionists claim only that personal identity is reducible to, and no more valuable
than, some constituent relationship between experiences. Eliminativism rules out
survival of death, because it also rules out survival before death. Eliminativists
embrace either radical moral revisionism or fictionalism about persons. Neither is
likely to appeal to BT.
Reductionism is less extreme, but still problematic for BT. Reductionists typically
argue that personal identity is less morally significant than we are inclined to believe.
(This is famously Parfit’s conclusion.) This sits uneasily with BT’s traditional divine
concern for the fate of each individual human being. Davis speaks for many when he
says that ‘such [reductionist] options are not available to Christian philosophers
because Christian teaching entails that we (the very persons that we are) will survive
death’.48
Our discussion of theodicy in chapter 8 largely presupposed non-reductionism
about persons. I therefore assume that BT rejects the no-self view, even in its modest
reductionist form. As with undifferentiated unity, AP can embrace reductionism, if it
is prepared to deny that morality requires personal survival.
10.2.3 Settling the dispute over personal identity

Our exploration of possible interpretations of personal identity yields some definite
conclusions. BT must reject the no-self view, undifferentiated unity, and concern-
based accounts of personal identity; while AP can accommodate these. Dualism,
materialism, psychological continuity views, and further fact views all remain on the
table, with each raising its own distinctive issues for both BT and AP.
We might think our next step is to select the best metaphysical account of personal
identity, and of human nature, and then focus on its implications for survival and
pre-existence. If we define identity in terms of the Absolute or the individual’s
pattern of self-concern, or if we embrace the no-self view, then this would undermine
BT immediately.
47 48
Parfit, Reasons and Persons, pt 3. Davis, Risen Indeed, p. 116.
 THE CASE AGAINST BENEVOLENT THEISM
Unfortunately, this simple strategy is too swift, for several reasons. First, personal
identity is highly controversial. There simply is no philosophical consensus, nor any
hope that one will be reached soon. Second, the controversy over personal identity
depends upon the facts about human beings, and opinions thus differ according to
one’s view of what those facts are. Should the theorist of personal identity seek to
accommodate ‘facts’ about our fate after death? If so, what are those facts? Is it a fact
that some humans have veridical near-death experiences or memories of past lives?
Third, views about personal identity vary in tandem with other philosophical posi-
tions. At the dawn of analytic philosophy, the debate over immortality was often
between idealists and materialists. And today, while Christian materialism is a
significant position, most dualists are BTs. Fourth, personal identity has an inextric-
able moral element. What you think we are depends on why (if at all) you think our
lives matter. A philosopher’s views here vary with his or her normative views or
meta-ethical position. Fifth, and most problematically in our dialectical context,
disputes about personal identity and human nature often track disputes about the
existence of God. (Most obviously, of course, religious views shape people’s account
of the facts to be explained in the first place.)
These complicating factors all come together in the case of John Locke.49 Locke
largely invented the modern use of imaginary cases to test accounts of personal
identity. His famous tale of the prince and the cobbler who wake in one another’s
bodies is designed to motivate a psychological continuity account. But unlike his
modern followers, Locke’s own data—the fixed points that any successful account
must accommodate—include the Christian doctrine of the Last Judgement, where
God will hold people responsible for their past behaviour. Given God’s perfect justice,
the Last Judgement is only possible if the person who is punished is the person who
committed the original wrong. It must be possible for human beings to survive death
with their moral responsibility for past crimes fully intact.50
To the modern secular ear, this looks like a moral argument, where an account of
personal identity is designed (in part) to preserve attributions of moral responsibility.
On this interpretation, Locke is open to the charge that he relies on an outdated
retributivist morality, a charge made forcefully by William James as early as 1890:
‘The mere stream of consciousness . . . cannot possibly be as responsible as a soul . . .
To modern readers, however, who are less insatiate for retribution than their
grandfathers, this argument will hardly be as convincing as it seems once to have
been.’51 In its original context, however, Locke’s argument has quite a different
flavour. Facts about God’s justice are quite distinct from our need to make moral
49
I draw here especially on Johnston, Surviving Death; and Forstrom, John Locke and Personal Identity.
50
It is worth noting that section 10.2.2.3’s Buddhist/Idealist model—where the person only remembers
her past lives in her final incarnation—could also satisfy Locke, because each person’s final incarnation can
be held responsible by God for all her earlier actions.
51
Quoted in Edwards (ed.), Immortality, p. 182.
IMMORTALITY 
(or legal) judgements. Locke does not tailor his notion of person to preserve or justify
some existing social practice. Rather, the existence and justice of God’s judgements is
a revealed fact that any analysis of our metaphysical nature must respect—no less
than it must respect the (equally indisputable) fact that human beings die.
10.2.4 AP and the metaphysics of immortality

What does AP bring to these tangled metaphysical debates over personal identity and
the afterlife? In this section, I defend two conclusions. First, empirical evidence and
metaphysical argument do not settle questions of survival and pre-existence. Second,
faced with this uncertainty, it is self-aggrandizing to posit survival or pre-existence.
We saw in section 10.1 that the direct evidence for and against either survival or
pre-existence is inconclusive. The same is true of the metaphysical debate about
personal identity. The contemporary literature is dominated by intuitions about
(more or less) fanciful cases. Many philosophers are suspicious of this methodology,
especially when such intuitions are said to offer insight into metaphysical reality
rather than merely describing human moral practices. We seek to discover the truth
about life after death. Why think the intuitions of some metaphysician can do that?
Our prior commitments reinforce this scepticism, because the intuitions of analytic
metaphysicians are often driven by an underlying commitment to global naturalism.
If they were willing to countenance non-natural phenomena, and to deny that science
explains all human experience, then perhaps analytic philosophers would take
dualism and Idealism more seriously. For instance, consider the question whether
survival as a disembodied or reincarnated soul without memories of this present life
would satisfy our present desire for personal survival. Dualists and their opponents
simply have divergent intuitions here.52 Given the vast numbers of believers in
rebirth and pre-existence, should the modern philosophical descendants of Locke
be quite so confident that they have got this one right?53
52
McTaggart puts the disagreement well: ‘If, indeed, a man should say that he takes no more interest in
his own fate, after memory of his present life had gone, than he would take in the fate of some unknown
person, I do not see how he could be shown to be in the wrong. But I do not believe that most men would
agree with him, and to most men, therefore, the prospect of a continuance of valuable existence, even with
the periodical loss of memory, would still seem to be desirable’ (McTaggart, Some Dogmas of Religion,
p. 131).
53
Or consider McTaggart’s (currently very unfashionable) defence of dualism, based on the alleged
unavailability of natural explanations of human knowledge and friendship, and on his rejection of matter.
McTaggart concedes that belief in matter (namely, anything distinct from spirit) is ‘a hypothesis which
would render immortality incredible’, precisely because it supports the argument against dualism based on
correlations between mind and body (McTaggart, Some Dogmas of Religion, p. 101). ‘If the self was an
activity of the body, it would be impossible that it should continue to exist when the body had ceased to
exist. We might as well suppose, in that case, that the digestion survived the body as that the self did. But
the body, as we have now seen, only exists for the selves which observe it, and we cannot, therefore, reduce
any self to be an activity of its own body’ (McTaggart, Some Dogmas of Religion, p. 101). While few
contemporary BTs will want to rely on McTaggart’s elaborate Idealist metaphysic, or his bravura dismissal
of the evidence for correlation, perhaps we should not dismiss his defence of dualism and pre-existence
altogether.
 THE CASE AGAINST BENEVOLENT THEISM
Some accounts of personal identity are incompatible with conventional moral

realism. It might therefore seem that our commitment to moral realism will partly
resolve our uncertainty about personal identity, because AP must at least reject those
accounts. However, this is not the case. While some eliminativisits (and some Idealist
supporters of undifferentiated unity) would jettison morality along with human
persons, AP can avoid nihilism by adopting a less person-centred morality. BT is
committed to person-affecting benevolence, but AP is not. (We return to the moral
dimension in section 10.3.)
We therefore cannot help ourselves to a neutral account of personal identity, and
then use it to narrow down our options regarding the afterlife. Our universe is
ambiguous here. But now our aversion to caprice does offer clear advice. In the absence
of a compelling argument in favour of survival, we should not endorse any metaphys-
ically extravagant view. In their different ways, Christian materialism and BT dualism
both make humans very special—either because God goes to so much trouble to ensure
our survival, or because we are naturally so different from the material world around us.
As with our earlier discussion of CDF freedom in chapter 8, we should reject the
leap to our own metaphysical specialness. When evidence and argument run out, AP
objects that to insist that humans are special is a kind of caprice. BT responses to evil
often need both personal survival and incompatibilist freedom. For a defence, these
need only be bare possibilities. But theodicy requires probability. God’s purposes
could include personal immortality for human beings. (They could include any-
thing.) But this requires either the creation of disembodied souls radically different
from the natural universe, or constant divine intervention in the natural order of
death and decay. Why leap to either of these, when AP offers a more sober
alternative? Human beings die, and that is the end of us. This is the view that best
fits a non-human-centred cosmic purpose.
Pre-existence, one of our new requirements for BT theodicy, is especially troubling.
To posit either pre-existence or rebirth, without compelling evidence, is itself a moral
failing. Ethicized rebirth implies a just past. The less fortunate deserve their misfor-
tunes, and are entitled to neither assistance nor redistribution. A false belief in a just
past is very dangerous, as it leads us to fail to assist innocent victims of injustice. False
belief in ethicized rebirth illegitimately reduces our concern for the least fortunate. If
there were a compelling ‘neutral’ metaphysical case for pre-existence, then of course we
should believe it. But there is not. Rebirth is under-supported by evidence and
argument. In such epistemic circumstances, for well-off people to believe they ‘deserve’
their good fortune, while others deserve to suffer, is an extreme case of caprice. Belief in
ethicized rebirth is a leap of faith. And it is a leap too far. If BT requires such a leap, then
it is on shaky ground indeed. The risk of moral error is too great.
If pre-existence were logically impossible, then BT could deny that God will choose
it. But, as we’ve seen, this option is not tenable either. There is no plausible
interpretation of personal identity that both permits future survival and rules out
pre-existence. BT is stuck with the leap to past lives.
IMMORTALITY 
I conclude that, while logically possible, the metaphysical demands of BT regard-

ing survival and pre-existence are illegitimate leaps of faith. The fact that it must
make those leaps thus counts heavily against BT.
10.3 Moral Arguments for Immortality

One defence of BT is that, however extravagant its metaphysical requirements may
appear, they can be given an independent justification. But that justification is moral,
not metaphysical. Given our commitment to moral realism, we must take this
defence of BT very seriously.
Moral arguments for immortality have fallen out of favour in recent philosophy.
This is surprising, given the enormous interest in parallel arguments regarding
freedom or God. It is also unfortunate. Whatever their ultimate fate, these arguments
still have much to teach us about the possible relations between morality and
metaphysics.
We must address two BT claims. First, BT defends its own metaphysical commit-
ments, arguing that morality shares them. My reply is that the demands of BT outstrip
those of morality. In its second claim, BT objects that AP cannot satisfy the demands
of morality. Because AP’s failing here is linked to its strong moral realism, many
atheists can borrow this argument against AP. My reply is that morality is less
demanding than BT claims, and that AP can satisfy its actual demands. I begin with
a catalogue of moral arguments drawn from the vast literature on this topic.
10.3.1 A catalogue of moral arguments

The history of philosophy offers many moral arguments for immortality. These
include the following:
1. The cosmic justice argument. Morality tells us to play our part in making the
world just. We cannot adopt a goal unless we know that goal will [or might] be
achieved. Therefore, we must believe that the world will [or might] be perfectly
just. The world can only be just if there is an afterlife where God apportions
happiness to desert. Therefore, we must believe there is such an afterlife.54
2. The argument from natural desire, or the common consent of mankind. Histor-
ically, this was perhaps the most influential argument for immortality in
Western thought.55 Human beings have certain natural desires that can only
be satisfied if we are immortal; nature does nothing in vain; therefore, we are
immortal.
54
This argument, of course, is inspired by Kant’s notorious discussion of the Highest Good (see, e.g.,
White, Commentary, pp. 267–9).
55
The argument received its canonical statement from Aquinas, and was especially popular in the
eighteenth century, and among later British Idealists (see, e.g., Walker, Eighteenth-Century Arguments for
Immortality; and Taylor, The Faith of a Moralist, pp. 268–78).
 THE CASE AGAINST BENEVOLENT THEISM
3. The perfect virtue argument. The moral law obliges us to aim at perfect virtue.
Unlike the highest good, this moral goal is instantiated solely in my own life. But
perfect virtue is not attainable in this life. Indeed, it is not possible in any finite
human life, because it requires an infinite span of time. Therefore, the demands of
morality only makes sense on the supposition that I am immortal.56
4. The continuity of agency argument. A momentary isolated self cannot function
as a rational agent, and therefore cannot live a meaningful life. To deliberate,
I must always believe I have a future. Unless I am immortal, there will be some
point at which I no longer have a future. My life would then become meaning-
less. Therefore, every agent must act on the (perhaps unconscious) presuppos-
ition that she is immortal.57
5. The Sidgwick argument. Morality only makes sense if there is a perfect correl-
ation between self-interest and aggregate well-being. Such a correlation clearly
does not occur in this life. Therefore, morality requires an afterlife.58
6. The permanent impact argument. Unless I leave a permanent mark on the
world, my life is in vain. One day, it will be as if I had never lived. No matter
how long I live, and however much I achieve, my earthly achievements will
eventually fade into nothing. I can make a permanent impact only if I am
always present. My life is in vain unless I am immortal.59
7. The significant difference argument. Unless I make some significant difference
to the universe, my life is in vain. Given the vast (perhaps infinite) size of the
cosmos, no finite earthly contribution could ever be significant from the point of
view of the universe. An infinite contribution is required. But only an immortal
being could make an infinite contribution. Therefore, my life is in vain unless
I am immortal.
8. The union with the divine argument. The purpose of human life is to achieve
union with the divine. As the divine is eternal, only an immortal being can hope
to achieve such a union. A life that fails to achieve its purpose is meaningless.60
9. The vision of God argument. Only a vision of God can confer meaning on
human life. Such visions are not possible in this life, and perhaps are only
available to immortal spiritual beings.
56
This argument is also Kantian (see, e.g., White, Commentary, pp. 269–76; Sullivan, Immanuel Kant’s
Moral Theory, pp. 142–4).
57
This argument is inspired by Korsgaard’s (Kantian) critique of Parfit’s reductionism about personal
identity. Korsgaard, ‘Personal Identity and the Unity of Agency’. For related discussion, see Mulgan, ‘Two
Parfit Puzzles’; and Mulgan, Future People, ch. 3.
58
Sidgwick never explicitly formulates this argument. But something like it surely lies behind his
lifelong interest in psychic research (Schultz, Henry Sidgwick, ch. 5).
59
The next two arguments are often presented as interpretations of Tolstoy’s famous claim that
something can be worth striving for only if one faces no prospect of death (see, e.g., Metz, ‘The Immortality
Requirement for Life’s Meaning’).
60
Both this argument and the next one are often combined with the argument from the common
consent of mankind, as union with the divine and a vision of God are common suggestions for the essential
good that is missing from mortal life.
IMMORTALITY 
A full discussion of all these arguments would require (at least) another book.
I hope to write that book some day. For now, I focus on general features. While there
are obvious differences between the various arguments, I shall treat most of them as
instances of a single argument, whose central premise is that our life is meaningless if
we are not immortal. I call this the meaning argument. (Two exceptions are the
cosmic justice argument and the argument from the common consent of mankind.
I treat these two arguments separately.)
I have presented a moral argument with a theoretical conclusion: humans are
immortal. But Kantian moral arguments, of course, are typically practical rather than
theoretical. Their conclusion is not a claim about the world, but a leap of faith.
‘Practical argument’ and ‘moral argument’ are now often used interchangeably. To
avoid confusion, I use the latter as a general term for any argument defending
metaphysical claims on moral grounds. I reserve the former for those arguments
that, following Kant, disavow any claim to theoretical knowledge, and claim only
that, as deliberating agents, we must act on the presupposition that we are immortal.
Our project is sympathetic to practical arguments. Religious ambiguity requires
leaps of faith. We cannot dismiss an argument simply because it is (merely) practical.
If immortality is necessary to make sense of morality, then a leap of faith may be
legitimate. BT will then insist that the correct account of personal identity implies
that immortality is only plausible if BT is also true. The leap to immortality is also a
leap to BT, and therefore AP cannot make that leap. As we saw in section 10.2, AP
could deny BT’s premise, and reply that survival is plausible even without a benevo-
lent God. (As we saw several times in part I, AP need not be unsympathetic to
dualism.) However, AP’s strongest reply is that the leap to immortality is not
required by morality, and AP can make whatever lesser leap is necessary.
AP can defend itself against practical arguments for immortality. However, not all
traditional moral arguments are practical. Moral arguments often conclude that we
are immortal, not merely that we must assume that we are. In response to these non-
practical arguments, AP offers essentially the same reply: either immortality is
possible without BT’s God, or it is not implied by any moral premise that AP must
accept.
The rest of this chapter outlines this defence of AP. I begin by defusing the most
ambitious moral arguments: cosmic justice (section 10.3.2) and common consent
(section 10.3.3). I argue that both are only plausible if reinterpreted as meaning
arguments. Section 10.3.4 highlights the gap between the demands of BT and
morality, arguing that moral arguments cannot justify either human pre-existence
or survival for animals. Section 10.3.5 outlines the meaning argument, while section
10.3.6 explains why it matters to our project. Unlike many atheists—especially global
naturalists—AP cannot simply reject meaning arguments out of hand. Our project is
based on the significance of metaphysics for ethics, and on a commitment to robust
non-naturalist moral realism. AP morality must involve some connection to tran-
scendent value. Our AP is normative. Sections 10.3.7 to 10.3.10 explore various ways
 THE CASE AGAINST BENEVOLENT THEISM
for mortal humans to connect meaningfully to infinite cosmic values. This paves the
way for part III’s development of AP morality.
10.3.2 Does morality require a just world?

Cosmic justice is the most ambitious moral argument. If morality requires a perfectly
just cosmos, then it already makes identical metaphysical demands to theodicy.
Personal survival and pre-existence are no longer peculiar to BT. If they are necessary
for perfect justice, then morality needs them as well. But does morality require a
perfectly just world? I shall argue that it does not, but that there is some truth behind
this particular moral argument.
The cosmic justice argument has two obvious weaknesses. First, why should my
goal as an individual moral agent be a just world, rather than to make the world more
just? I can play my part even if I know that, because others will not play theirs, the
world will never be just. The non-compliance of others is a serious moral issue, but
we do not solve it by wishing it away. Second, even if perfect justice is my goal, I need
not believe that it will come about. I need only believe that the goal is possible. And it
may be ‘possible’ only in a very abstract sense. Perfect justice can still function as a
meaningful moral ideal, even if my knowledge of human nature teaches me that it is
vanishingly unlikely.
The cosmic justice argument is worth discussing, despite its obvious weakness,
because it contains two elements of truth. First, some moral theories do need a just
cosmos. Unsurprisingly, Kantian ethics is the most striking example. The protest-
ations of his atheist followers notwithstanding, Kant needs an afterlife to render his
austere non-consequentialism credible.61 But the obvious solution is to abandon
Kant. Other moral theories, such as consequentialism, do not need perfect justice.
Our project, of course, favours consequentialism over Kantianism. In the absence of
compelling metaphysical arguments, AP cautions both against the leap to a just
cosmos, and against any normative theory where the justice of the cosmos excuses
moral agents from taking responsibility for the predictable consequences of their own
actions. Of course, things would be very different either if we had independent
evidence of cosmic justice, or if Kantian ethics could be demonstrated a priori. But,
obviously, neither is the case.
The second grain of truth in the cosmic justice argument is that ordinary morality
is under threat if the world becomes too unjust. This is also the partial truth in
Sidgwick’s argument. Sidgwick’s original dualism of practical reason is overstated.62
Reason does not collapse into incoherence without a perfect correlation between duty
61
To cite the most obvious example: Kant needs an afterlife to justify his extreme attitude to truth-
telling. You should not lie to potential murderers because, if you tell the truth, you are not responsible for
the evil they do. It is not your job to make the world just. This is obviously more plausible if there is
someone else whose job it is (i.e. God). And any such story requires, inter alia, immortality.
62
For critique of Sidgwick, see, e.g., Schultz, Henry Sidgwick, ch. 4; Phillips, Sidgwickian Ethics.
IMMORTALITY 
and self-interest. Most contemporary moral philosophers regard the clash between
self-interest and aggregate well-being as a site of real moral conflict. Our moral lives
are structured by the tension between these two conflicting sources of moral
demands. While that conflict is difficult to resolve, it is not impossible.
However, although the Sidgwick argument is overstated, much contemporary
moralizing would flounder in a world that is very unjust, or where the gap between
self-interest and the common good is too great. (Our collective inability to even
acknowledge the moral sacrifices required by climate change is one striking
example.) But it is premature to describe this as a failure of morality or reason.
Rather, what breaks down in a severely unjust world is morality as ordinarily
understood. We must separate morality itself from our current moral conceptions.
As affluent citizens in modern liberal democracies, we tend to assume that, in normal
circumstances, it is possible to combine living morally and living well. We acknow-
ledge that, in exceptional or tragic circumstances, familiar moral duties may require
extreme self-sacrifice. But we treat those circumstances as exceptional, and as tragic.
Now suppose we come to believe that, without an afterlife, it is no longer possible for
a human being to live well. This would require a readjustment of our morality. But it
would not necessarily show that morality did not apply. Even if no possible human
life is worth living, some are still better or worse than others. Moral distinctions still
make sense. The resulting morality might be very demanding, but perhaps it should
be. This is grist to the mill of consequentialism, which favours a more austere
morality.63 As we saw in chapter 5, AP both welcomes this feature of consequential-
ism, and pushes the theory further in that direction. (We return to these themes in
part III.)
I conclude that, despite its appeal, the cosmic justice argument is implausible,
especially for those predisposed to consequentialism. As yet, BT has given us no
reason to doubt either that the demands of theodicy outstrip those of morality, or
that AP can satisfy morality’s legitimate demands.
A final option is to turn the cosmic justice argument into a meaning argument.
Without a just cosmos (or a coincidence between self-interest and morality), our
moral lives cannot be meaningful. The obvious reply is that there are alternatives that
do supply meaning. We consider those alternatives below.
10.3.3 From common consent to meaning

The argument from the common consent of mankind asserts that human beings have
desires that cannot be satisfied in this life, and that the existence of such desires
points to a life to come. The common consent argument is usually deployed by
traditional BT. However, if it were plausible, the argument would actually support
pre-existence, and perhaps also survival for animals. Anthropological evidence
63
I explore these issues in Mulgan, Ethics for a Broken World; Mulgan, ‘Ethics for Possible Futures’; and
Mulgan, ‘Utilitarianism for a Broken World’; and I hope to explore them further in future work.
 THE CASE AGAINST BENEVOLENT THEISM
suggests that, across all human societies, the most common belief about the afterlife is
a cycle of rebirth within the kin group, often involving nearby non-human animals.64
If our ‘natural’ desires were a reliable guide to our fate, then the real demands of
theodicy might well be justified. Sadly, the argument fails.
The obvious problem with this argument is that many desires are not satisfied. The
existence of a desire, however widespread, cannot provide evidence that its object
exists. As Aristotle pointed out, we cannot infer earthly immortality from the evident
fact that many people don’t want to die! The inference from desire to object requires
a teleological background, where ‘nature does nothing in vain’, and each ‘natural’
desire is assured of fulfilment. If we are designed by a benevolent God, then this is
very plausible. The argument from the common consent of mankind is very powerful
if we are already confident that BT is true. But without BT, the argument fails.
Evolution cannot replace a benevolent God. Even if a desire is selected by evolution,
that doesn’t prove that it is ever fulfilled. A desire to survive death could have many
beneficial consequences in this life.
Like the argument from cosmic justice, the argument from common consent is
most charitably reinterpreted as a meaning argument, where the desire for survival is
‘natural’ in the sense that its object is necessary for a flourishing human life.
10.3.4 A moral argument for pre-existence and animals

Following the literature, I have spoken of moral arguments for immortality, as
opposed to survival or pre-existence. The distinction between immortality and
post-mortem survival is secondary. Some moral arguments suggest an afterlife of
infinite duration, others do not. But pre-existence is more problematic. Any moral
argument is essentially forward-looking, and therefore favours future survival over
pre-existence. Could there be a moral argument for pre-existence? One possibility is
an ambitious cosmic justice argument, where I can only believe that the universe will
be just in the future if I can be confident that it has been just in the past. (And we saw
in section 10.1.4 that past justice implies pre-existence.) A second possibility is to
insist that, while morality only requires future survival, the only plausible afterlife
scenarios also involve pre-existence. So we get pre-existence for free. (Perhaps we are
convinced by McTaggart’s suggestion that the best dualist arguments for future
survival also yield pre-existence.65)
Neither option is very promising. We saw in section 10.3.2 that the cosmic justice
argument is not very credible, even when it is much less ambitious. And, in section
10.2, we saw that every account of personal identity leaves open the possibility of an
afterlife without pre-existence. In the absence of a compelling argument to the
contrary, AP will insist that, whatever else it may require, morality does not require
pre-existence. But BT does need pre-existence. Therefore, the metaphysical needs of
64 65
Obeyesekere, Imagining Karma. McTaggart, Some Dogmas of Religion, p. 115.
IMMORTALITY 
theodicy outstrip those of morality. This conclusion is sufficient on its own to

accomplish our first goal in section 10.3. BT cannot simply help itself to the demands
of morality. BT requires a separate (and extremely controversial) leap of faith.
Similar remarks apply to our other extension to BT theodicy to cover animals.
Perfect cosmic justice probably does require justice for non-human animals, includ-
ing those born before the evolution of humans. But that makes it much harder to
claim that morality requires perfect justice. Why does human morality have any
implications about the post-mortem fate of non-human animals? (It is revealing that
most arguments based on cosmic justice do not mention animals at all.66)
10.3.5 The meaning argument

The meaning argument was extremely popular around the turn of the nineteenth and
twentieth centuries, especially among the British Idealists.67 William James even
regarded this as the principal ground for belief in immortality among his contem-
poraries.68 However, its close association with British Idealism meant that the
argument fell foul of the trenchant attacks of Idealism’s naturalist critics, especially
Bertrand Russell.69 Here, as elsewhere, subsequent analytic philosophy sides with
Russell against his Idealist opponents. The meaning argument did get some attention
within analytic philosophy in the 1960s.70 But this was mostly negative. As Metz
sums it up: ‘The central criticism to emerge was that death intuitively cannot
undercut the worth of performing certain constructive actions.’71
The idea that immortality is necessary for meaning, and that death somehow
threatens to drain life of its significance, is commonplace. But what exactly does it
mean? I began with a catalogue of more specific arguments partly to flesh out the
elusive notion of meaning. Meaning goes beyond well-being, happiness, or value. The
meaningless life is not necessarily an unhappy one. It can indeed be very pleasant.
But it is pointless, empty, aimless. Meaning thus has an active, purposive, teleological
dimension. The meaningless life may be filled with pleasant diversions and distrac-
tions. But nothing is worth doing. The central projects of life seem absurd.
A second common component of meaning is epistemic. A meaningful life involves
self-knowledge—and an understanding of one’s nature, one’s place in the cosmos,
and one’s relation to objective moral truths. The unexamined life may be happy, but
it is not meaningful.
66
One exception, of course, is Dougherty’s theodicy for animals, discussed briefly in chapter 8 and in
footnote 8 above.
67
Mander, British Idealism, chs 4 and 10 (especially pp. 133–6, 356–90).
68
Edwards, Immortality, p. 181.
69
See, especially, the rhetorical flourishes of Russell, Philosophical Essays, p. 59 ff.
70
See, e.g., Flew, God and Philosophy, p. 105.
71
Metz, ‘The Immortality Requirement for Life’s Meaning’, p. 170. (See also Schmidtz, ‘The Meanings
of Life’.)
 THE CASE AGAINST BENEVOLENT THEISM
Meaning arguments can be more or less ambitious, along several dimensions.

I shall illustrate these using the divine union argument. The most ambitious argu-
ment holds that meaning is only found in union with the divine. All other human
activity is completely pointless. Divine union is necessary for meaning. More modest
arguments claim that divine union merely contributes to meaning, but is not
necessary to it.72 In our current dialectical context, such weaker moral arguments
are inadequate. It is not enough that life would be better, more valuable, or more
meaningful if we were immortal. A mortal life must be meaningless, not just less
meaningful. Unless we are immortal, our lives are pointless; and our moral choices,
decisions, and dilemmas are void of all significance. Immortality must be necessary
for meaning.
At the risk of oversimplifying, we can fit most meaning arguments into a simple
schema:
1. A meaningful human life requires a specific connection (C) to some specific
value (V); and
2. Only an immortal being can enjoy C to V.
Any meaning argument begins by identifying V and C, and establishing their
essential connection to meaning. It must then demonstrate that C is available to
immortal beings, and not to mortal ones. The first part of this demonstration is easily
overlooked, but it is crucial. General angst about the pointlessness of existence gives
us no reason to believe in immortality. If all human life is meaningless, then the
argument collapses. It is not enough to show that meaning is not possible if we are
mortal. We must also show that meaning is possible if we are immortal. Something
specific is missing in mortal lives that is found in (at least some) immortal ones.
Immortality may not be sufficient for meaning. (Perhaps there could be meaningless
immortal lives.) But immortality must put meaning within our reach.
10.3.6 Who is worried about meaning?

I shall now argue that the meaning question is only significant given several other
controversial philosophical assumptions. On many influential meta-ethical and
normative ethical views, either meaning is trivially available in ordinary mortal
human lives, or it is not available to anyone.
Suppose one is an internalist, in the sense made famous by Bernard Williams.73
Statements about reasons, values, meaning, well-being, what makes life worth
living, and so on, must connect to the agent’s motivations. The meaning
72
To further complicate matters, divine union could be meaningful in its own right, or it might be a
precursor to or a component of something else. Perhaps divine union is essential for one component of a
fully flourishing life, such as cosmic knowledge or cosmic harmony. That component, in turn, could be
either necessary for meaning, or sufficient, or neither. We return to these issues in chapter 12.
73
Williams, ‘Internal and External Reasons’. For further discussion, see section 1.4.4.
IMMORTALITY 
argument now turns on a psychological generalization: no one finds a mortal

existence satisfying. But this is simply false. Some people do find life meaningful
despite not believing in any afterlife. Of course, there are also people who find no
joy, value, or meaning in any human activity. But infinitely prolonging these
activities does not help—a point emphasized by Camus and Williams, among
many others.74
The meaning worry arises only for externalists, for whom statements about
reasons, values, meaning, well-being, and what makes life worth living, all refer to
objective normative facts that are, in principle, independent of psychological facts
about what actual human beings happen to prefer. It is no coincidence that propon-
ents of the meaning argument are typically externalists; as are those (like Sidgwick)
who, while rejecting the argument, have taken it seriously. If pressed, the consistent
internalist will simply deny that any life could be meaningful as the externalist uses
the term. There is no ‘objective’ normative property of meaningfulness. There is
therefore nothing lacking in human lives that an immortal life could contain.75
Internalists easily dissolve the meaning argument. But many externalists are
equally untroubled. For the nihilist, all talk of meaning presupposes non-existent
objective values. The absence of meaning is perfectly general, and immortality offers
no solution. The meaning argument has bite only for externalist moral realists.76 But
many moral realists need not worry either. Even if the two notions are distinct,
meaning is closely related to well-being. A meaningful life is a special kind of
flourishing life. For hedonists and preference theorists, all the goods that make our
lives worth living are straightforwardly available in this earthly life. We have pleasure
and desire satisfaction in abundance. Of course, many people have desires that
cannot be satisfied within their earthly lives, and many miss out on pleasure
altogether. But these mundane facts cannot ground an argument for immortality,
as an eternal life would only deliver more of the same.
If all mortal lives lack meaning, then meaning must go well beyond pleasure or
preference-satisfaction. Worries about meaning only arise if our theory of meaning is
objective. And one’s account of meaning is very unlikely to be less objective than
one’s theory of well-being. Worries about meaning are thus especially acute for
proponents of objective list theories of well-being, where a worthwhile human life
requires more than pleasure, the absence of pain, and the satisfaction of one’s present
desires. If what matters in human lives is not entirely subjective, then meaning can
easily be missing in this life.
74
See, e.g., Camus, The Myth of Sisyphus; Williams, ‘The Makropulos Case’. Indeed, Williams himself
goes further, and predicts that no human agent could find an immortal existence satisfying and still be
herself. For a critique of Williams, see Chappell, ‘Infinity Goes up on Trial’.
75
Cf. Parfit’s diagnosis of his dispute with Williams over external reasons (Parfit, On What Matters, vol.
2, pp. 433–9).
76
In so far as they can successfully mimic moral realism, non-cognitivists may face analogous problems.
But, as I said in chapter 2, I set non-cognitivism aside in this book.
 THE CASE AGAINST BENEVOLENT THEISM
Most moral naturalists need not worry about meaning, because they endorse
subjective accounts of well-being—namely hedonism or preference theory—and
are likely to offer similarly subjective stories about meaning. Should naturalists
who endorse an objective list theory be worried? The obvious answer is that they
should not. If moral properties are natural properties, then value and meaning must
be located within the natural world. Surely physical creatures in a world of natural
value don’t need immortality to make life meaningful?
Some threats to meaning do seem purely natural. Relative to the cosmos as a
whole, human beings seem so small that it is easy to think our puny endeavours are
irrelevant. Or the physical universe itself can seem insignificant, especially if it is finite
and doomed to extinction. But moral naturalists offer the same reply to both threats.
They simply deny that finitude renders either human existence or the cosmos itself
insignificant. Our entire universe will one day fade away. Naturalists agree that, in
Russell’s colourful phrase, ‘the whole temple of Man's achievement must inevitably
be buried beneath the debris of a universe in ruins’.77 But so what? If value only arises
within the physical universe, if moral facts are natural facts, then how could the
entire physical universe itself lack meaning? If concerns about meaning persist, then
duration is almost certainly not the problem—and thus immortality is not the
solution. If the nihilist worry bites in a finite physical universe, it will continue to
bite in an eternal one.
Moral naturalists need not fear the meaning argument. Its only remaining
target is that subset of moral objectivists who also endorse non-naturalism. The
physical universe does seem empty to the non-naturalist. But the non-naturalist’s
real worry has nothing to do with impermanence. It is rather that, if the material
world is all there is, then there is nothing to ground any value at all. This is a
worry that our project shares. But the answer is to posit something transcendent
that grounds mundane (physical) value. That something need not be the agent’s
own immortal soul. Once we distinguish the ground of value from its instances,
the transcendence of the former is perfectly compatible with the transience of
the latter.
The meaning argument now seems to have dissolved entirely. However, non-
naturalists may still wonder whether meaning can be found in finite human lives. If
morality is grounded in a realm of value that transcends the purely natural world,
then one might worry that the fleeting lives of finite physical creatures cannot count
for very much, because our mortal existence cannot connect to genuinely transcend-
ent values in any meaningful way. AP seems to exacerbate this worry, because its
cosmic values are especially distant from mortal human life.
I conclude that the meaning argument does raise important questions for exter-
nalist non-naturalist moral realists who endorse an objective list theory of human
77
Russell, Philosophical Essays, p. 59.
IMMORTALITY 
well-being. As these are our moral commitments, meaning arguments do threaten

our project. Many atheists can shrug off worries about meaning. AP cannot.
10.3.7 AP and the meaning of life

AP per se says nothing about personal immortality, either way. In principle, AP could
simply concede the moral argument, admit that human beings are immortal, and
defend itself (against both BT and atheism) on other grounds. However, this
response is less than ideal. Once we concede immortality, the prospects for theodicy
are very greatly improved. Its ability to deny immortality is thus a very significant
advantage that AP should only relinquish very reluctantly.
Nor is it clear that AP can simply accept immortality. As we saw in section 10.2,
some plausible metaphysical views suggest that survival and pre-existence are
extremely unlikely unless there is a benevolent God to ensure them. (For materialists
about personal identity, survival without divine intervention is all but impossible.) If
morality needs survival or pre-existence, then metaphysical considerations favour BT
over AP, because BT immortality is compatible with a wider variety of accounts of
personal identity. Indeed, the connection between immortality and BT is often built
into the moral argument itself. What confers meaning is immortality within a just
cosmos ordered by a benevolent deity.78 AP cannot borrow any moral argument that
takes this form.
Why not bite the bullet? Why not concede that a meaningful human life requires
immortality, and then conclude that no human life has meaning? Perhaps AP
should turn the meaning argument on its head. We are not immortal, and our lives
are meaningless. Therefore, we must reject BT! After all, AP must deny that we
enjoy one possible kind of meaning: namely, the value or meaning that comes from
being part of the cosmic purpose. However long they may last, our human lives
have no objective value.
As I noted in chapter 2, AP can combine realism about cosmic values with human
moral nihilism. Our moral talk presupposes a human-centred cosmic purpose; there
is no such purpose; therefore our moral talk is false. All human talk of value or
meaning is false. Perhaps this is the most natural interpretation of AP, and I explore it
further in part III. My main aim in this book, however, is to develop a version of AP
that enables human beings to find meaning in our lives—in light of the cosmic
purpose, and in spite of the fact that it is non-human-centred. This search is the task
of part III, which can be read as an extended reply to the meaning argument.
Any form of AP must recognize a gap between human beings and cosmic values.
A non-human-centred cosmic purpose must rule out some kinds of value, flourish-
ing, and meaning that would have been available under BT. (Our lives cannot derive
any meaning from God’s love for us, for instance.) If the meaning argument only
78
See, e.g., Metz, ‘The Immortality Requirement for Life’s Meaning’, p. 177; Taylor, The Faith of a
Moralist, p. 256.
 THE CASE AGAINST BENEVOLENT THEISM
cites those distinctively BT values, then it adds nothing to the case against AP. AP
already denies that this kind of meaning is essential to human life. AP then insists
that other, more modest values are still open to us even if we are mortal beings in an
AP cosmos. Perhaps these are sufficient for ‘meaning’, and perhaps not. But either
way, it is hubris to demand more. If human life is to be meaningful, it must do
without the grandiose moral and metaphysical pretensions of BT. To provide an
extra objection to AP, the meaning argument must show that immortality is essential
for some kind of meaning that is not itself directly tied to the desirability of a personal
connection to a benevolent God. The remainder of this chapter explores AP’s
response to two distinct challenges: that a mortal life cannot offer connections to
infinite value (section 10.3.8), and that mortal agents cannot make a significant
difference (sections 10.3.9 and 10.3.10).
10.3.8 Alternative connections to infinite value
The meaning argument not only posits an infinite transcendent value. It also insists
that each of us, as an individual, must instantiate that value. To connect with the
infinite value, I must be immortal. For AP, this insistence on personal instantiation is
the most natural place to object to the meaning argument.
There are many things one might find lacking in a transient, finite, purely
materialist universe—many reasons to believe that morality needs some transcendent
foundation. The global naturalist either rejects such arguments and finds the purely
physical universe entirely sufficient, or she responds to its deficiencies by embracing
nihilism. But my preferred formulation of AP combines a commitment to objective
values with dissatisfaction with the thought of a mundane purely material universe.
My Normative AP must therefore explain why its dissatisfaction does not force it to
choose between personal immortality and moral nihilism.
A finite mortal human life can connect to a transcendent value without instanti-
ating it, in several ways. First, cosmic values that transcend the spatio-temporal
realm, and are therefore eternal, might be necessary to ground, explain, or underpin
the value of all finite things. Without these eternal Platonic values, there would be no
finite value, and nothing would be worth doing. But it does not follow that each
instance of value must itself be eternal. Like any other finitely valuable thing, human
beings can resemble the transcendent value without sharing its eternity.
Second, the life of perfect virtue can function as an ideal for human lives, even it
has some features that humans cannot share. We can centre our moral life on
imitation of some perfectly virtuous superhuman moral exemplar, either real or
imaginary. (A proponent of AP might imagine a being whose life was significant
from the point of view of the universe, such as those introduced in chapter 7, and
then seek to emulate that being.) Even if perfect virtue requires immortality, it
doesn’t follow that any imitator must herself be immortal. (Christianity offers a
fruitful model here. I can seek to imitate Christ without the blasphemy of supposing
that his perfect virtue is possible for me.)
IMMORTALITY 
Third, without being immortal ourselves, we can still discover, study, contemplate,
worship, and strive to emulate, resemble, or experience the cosmic purpose, the
Platonic Good, or a personal God. AP rules out reciprocal relationships between God
and humans, such as mutual love. But it leaves open asymmetric connections.
Consider the suggestion that a meaningful human life requires a vision of God that
is only available in the afterlife. As we saw in chapter 5, AP replies that experiences of
God or cosmic purpose are available in this life.
Christians (and proponents of other BT religions) may object that the experiences
of mystics in this life only make sense against the broader story of God’s plans for
humanity. And that story essentially involves an afterlife. Today’s mystical experience
is a foretaste of the bliss of heaven, and is a fraudulent illusion unless heaven is real.
AP’s reply draws on chapter 9. Religious diversity forces us to reinterpret traditional
religious claims. AP offers its own (partly deflationary) reinterpretation: while
Christian mystics are in touch with some genuine supernatural reality, their more
human-centred pronouncements are distortions produced by a combination of
deference to religious tradition and the natural human tendency towards self-
aggrandizement. The postulation of a personal afterlife is one of those illusions.
The challenge for AP is to show that, even under this less ambitious reinterpretation,
these experiences still have real moral significance. As ever, this is a challenge for
part III.
Finite mortal humans are only capable of (at best) a very imperfect cosmic
understanding. AP can simply concede that perfect cosmic understanding requires
personal immortality, and therefore lies beyond the reach of mortal humanity. But
human meaning does not require perfect knowledge. In our small fallible way, we can
gain some insight into cosmic value and purpose; and our fleeting insights are
sufficient to give our lives some meaning.
10.3.9 Can I make a difference in an infinite world?

So far, AP’s reply to the meaning argument emphasizes contemplation and under-
standing. This sits uneasily with our consequentialist commitments. Consequential-
ists will object that, if our only connections to infinite value are passive, then our
human activities have no meaning. This is the worry behind the permanent impact
and significant difference arguments. We begin with the former. The standard reply
is that I could make a permanent impact by affecting something else that is
eternal.79 I might be mirrored, reflected, or remembered by something or someone
who continues indefinitely, and perhaps even outlasts the physical universe itself.
79
See, e.g., Metz, ‘The Immortality Requirement for Life’s Meaning’, p. 164. Another philosophical
reply is that, under a four-dimensionalist metaphysic, it is timelessly true that I have lived. (There is a
famous story of Einstein consoling the relatives of a deceased friend with the thought that, due to four-
dimensionalism, he was no less alive than ever.) I do not live at all times, any more than I live in all places.
But there is no time at which it is really as if I had never been. This suggests that the real worry is not
permanence, but significance. While this argument would be congenial to AP, I am wary of placing too
 THE CASE AGAINST BENEVOLENT THEISM
Perhaps traces of one’s existence remain even after the next Big Bang, or I might be
remembered eternally by God.80
AP can borrow this reply. An AP God might remember details that are not
themselves intrinsically worth knowing. Or perhaps God remembers some greater
whole, of which my life is an insignificant part. (This would complement the
accounts of real but cosmically insignificant value to be developed in part III.)
However, BT will reply that being remembered by God fits much more naturally
into a BT framework. Divine remembrance is inseparable from divine love, and
therefore not available to either AP or atheism. This kind of meaning is only open
to BT. At this point, AP’s response is clear. If permanent impact requires divine love,
then AP must deny that a meaningful life needs this kind of permanence.
Here the permanent impact shades into the significant difference argument. BT
must claim that non-permanent impact lacks significance. We saw earlier that unease
about the insignificance of the physical universe itself can be sidestepped by moral
naturalism, or solved (for non-naturalists) by positing something transcendent that
grounds mundane (physical) value. If the significant difference worry is distinct, it
must involve our insignificance relative to some actual instance of value. This could
be either the transcendent Good itself, or God, or an infinite cosmos, or perhaps even
a very large finite one. Nothing we do will make any difference because we are so
small. I am insignificant against the vastness of the physical universe. Immortality
would make me loom larger in the world.
BT’s worry is thus that finite contributions count for nothing in an infinite context.
This is the worry behind the infinite utility puzzle. We can therefore borrow the best
solution to that puzzle. As we saw in chapter 8, the best theory of aggregation will
conclude that finite ‘additions’ do matter even in infinite contexts.81 Furthermore, if
the meaning argument is practical, then a practical solution will suffice. And for
practical purposes, human morality is more interested in guiding our deliberations
than in constructing consistent theories of cosmic comparative value. We can
reasonably focus on the difference we make here and now, without worrying whether
we increase total cosmic value.82
AP can thus dissolve the significant difference argument even if there is some
actual infinite value. If the cosmos is finite, then the argument is much easier to
much weight on such a controversial metaphysical hypothesis. And I am aware that non-philosophers
invariably regard such arguments as evasions. (Were Einstein’s friend’s relatives really consoled at all?)
80
Divine remembrance could even constitute personal immortality. I continue to exist after death
simply because God remembers me. (This account of personal identity might especially appeal to some
Idealists.) God’s remembering X might be the further fact that constitutes X’s identity—perhaps both after
death and during life. So long as divine remembrance has no BT implications, AP could concede this
possibility, and then argue that survival by divine remembrance cannot meet the metaphysical demands of
BT theodicy. (For instance, divine remembrance doesn’t give us pre-existence or the redemption of
horrendous evil.)
81
See, e.g. Vallentyne and Kagan, ‘Infinite Value and Finitely Additive Value Theory’.
82
I draw here on Mulgan, ‘Transcending the Infinite Utility Debate’.
IMMORTALITY 
dissolve. Our contributions may be very small relative to the vastness of the cosmos,
but that does not mean they cannot matter to us.
AP must deny that our lives have cosmic significance. (Otherwise, morality would
require BT.) AP must therefore separate significance within a human life from
cosmic significance. But if that separation can be made, AP can defeat the meaning
argument.
Significant difference and permanent impact arguments do trouble AP. But this is
only because AP struggles to recognize human significance at all. This is the central
task for part III. The meaning argument highlights the importance of that task, but it
does not create any new difficulty.
10.3.10 Larger agency

Some will remain unconvinced. Surely, even in a finite cosmos, my life is too small to
matter. Why care about the difference I can make? Surely my individual agency is
insignificant, and meaning is only available to greater agents such as those discussed
in chapter 7. As we have just seen, AP can reply that human meaning does not
require larger agents. An alternative AP reply accepts that meaning requires some
larger agent, but then argues that this agent need not be my individual immortal self.
It might instead be some group agent of which I am a part.
Suppose I simply cannot believe that I will survive death. I cannot even entertain
that belief as a ‘postulate of practical reason’, whatever that means. Perhaps I agree
with Sidgwick’s caustic verdict on Kant:
I am so far from feeling bound to believe for purposes of practice what I see no ground for
holding as a speculative truth, that I cannot even conceive the state of mind which these words
seem to describe, except as a momentary, half-willful irrationality, committed in a violent
access of philosophic despair.83
But suppose also that I remain convinced that meaning requires continuity of agency
beyond my death. No future self is identical to my present self. But there are future
people who join with me in group action. Instead of taking my present action in
isolation, and despairing at my mortality, I ask what we together can do. I give my
present action meaning by playing my part in our best group action.
Group action is often criticized for its metaphysical extravagance. How can one
seriously attribute intentions or preferences to groups?84 But in our present context,
group agency is less metaphysically demanding than individual agency. I can believe
that present and future selves act as a group even if I am sceptical about my future
survival.
83
Sidgwick, The Methods of Ethics, bk 4, ch. 6, p. 507.
84
On the other hand, many recent proponents of group agency try to separate it from the metaphysical
excesses of traditional Idealism. (See, e.g., Gilbert, Sociality and Responsibility; List and Pettit, Group
Agency; and Tuomela, Social Ontology.) AP could easily appropriate these more modest models of
collective agency.
 THE CASE AGAINST BENEVOLENT THEISM
As well as delivering significance, group agency also defeats other meaning

arguments. Continuity of agency is now assured. Even at the moment of my death,
I can still meaningfully act, because my group has a future, even if I do not.85 Or
consider Sidgwick’s gap between morality and self-interest. If I redirect my attach-
ment to some larger whole, then my self-concern moves closer to impartial concern.
Of course, some conflict between self-interest and cosmic value remains. (I am
unlikely to identify with the set of all rational beings, and even my identification
with smaller groups is never absolute.) But, as we saw earlier, total reconciliation
between self-concern and morality is not essential to a meaningful life. Group
identification need only bring the Sidgwick gap down to a manageable level.
Group action is hardly uncontroversial, and raises more questions than it answers.
But it does provide a metaphysically parsimonious alternative to personal immor-
tality, highlighting once again the comparative modesty of morality relative to BT
theodicy. As obligations to future people become more pressing in our ethical lives,
and as ethical issues become more globalized and interconnected, group actions will
become ever more significant.86 While this can make individuals feel insignificant, it
may also teach us that collective action is the key to a meaningful ethical life.
BT may insist that no finite agency, however large, can replace the significance
available to immortal souls in heaven. But AP joins the moral naturalist in replying
that, if larger-scale human concerns cannot give our lives sufficient meaning to make
our human endeavours worth pursuing, then no plausible afterlife will suffice either.
I conclude that, if it can ground any human morality, AP can dissolve the meaning
argument for immortality. If human lives can be meaningful at all, then they can be
meaningful enough even if death marks the end of our existence.
10.4 Conclusion
Our long discussion of immortality has reached rather undramatic conclusions. BT
makes extravagant metaphysical demands, requiring a leap of faith that AP can
reasonably reject. While morality does make some demands, especially for those
committed to objectivist externalist non-naturalist moral realism, morality’s
demands are less than those of BT. AP can thus reasonably hope to satisfy the former
while dismissing the latter. The afterlife supports AP against BT, and does not
support atheism against AP. Along the way, however, AP has issued many promis-
sory notes, which must be redeemed in part III.
85
This response to the continuity of agency argument for personal immortality is modelled on the best
reductionist reply to Christine Korsgaard’s Kantian argument for belief in a continuing self, which
I develop in Mulgan, ‘Two Parfit Puzzles’; and Mulgan, Future People, ch. 3.
86
Mulgan, Future People, chs 7 to 9; Mulgan, Ethics for a Broken World; Mulgan, ‘Ethics for Possible
Futures’.
PART III
Ananthropocentric Purposivist
Morality
11
A Dialogue
Suppose parts I and II have convinced you that AP is true. What difference should
that make to your life, your values, or your moral theory? In part III, I explore the
implications of AP both within competing moral theories, and for the choice between
them. Part III is tentative and exploratory, even more so than the rest of the book.
Imagine that you discover, not only that the universe has some remarkable
property, but that this property explains the very existence of the universe. Would
this change your moral outlook? For the benevolent theist, cosmic purpose and its
source are clearly morally relevant. So the shift from BT to AP clearly would affect
one’s moral world view. But does the shift from atheism (or agnosticism) to AP have
the same effect? If you start from BT, is conversion to AP any different from
conversion to atheism? What does AP bring to human morality that atheism cannot?
The impact of AP on a person’s life will depend on both her prior commitments
and her reasons for embracing AP. To explore the diverse possibilities, I open our
discussion of AP ethics with a series of monologues delivered by seven imaginary
converts to AP. Collectively, they bring together the arguments of parts I and II, as
well as the main overarching themes of the book. The characters take turns to outline
their own primary reasons for taking AP seriously, and the impact it has had on their
lives.
COSMO: I’ve always been fascinated by the really big metaphysical questions.
Why is there something rather than nothing? Why is there any physical universe at
all? The ultimate questions are worth asking, they are intelligible, and they have
(as yet) no materialist or scientific answer. I’ve never been able to reconcile a
benevolent God with the evils of this world, so traditional theism has never been a
live option for me. But just accepting the cosmos as a vast brute fact doesn’t seem
satisfactory either. (While I wouldn’t go as far as Leibniz, I am sympathetic to the
following modest Principle of Sufficient Reason: it always counts in favour of any
theory that it leaves behind as few brute facts as possible.) So I’m left with AP. At the
moment, I’m agnostic between Leslie’s axiarchic Platonism and a non-human-
centred personal creator. I like the metaphysical simplicity of the former, but I find
the latter much easier to relate to my life. (Perhaps talk of God as a person is a
distorting but essential concession to our human frame of reference.)
How does AP impact on my life? Actually, it makes a big difference. The simple
fact that there might be a cosmic purpose, a reason for the existence of everything,
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
makes me want to know what that purpose and reason are. If there is some Parfitian
selector, then the central task of philosophy is to find it. I’m not entirely persuaded by
any particular more detailed account of AP, but I am very interested in any enquiry
that might shed light on the cosmic purpose. (To cite two examples, their relevance to
debates about cosmic purpose gives new significance to two topics I’ve long been
interested in: Big Bang cosmology and the possibility of an infinite universe.)
And I don’t think this is merely a matter of personal preference. Reflecting on the
cosmological argument, and why I find it persuasive, has shown me that, while
I wasn’t a great believer in objective values before, the philosophical package that
best makes sense of all the facts is one that posits both a cosmic purpose and cosmic
values, and some connection between the two. I’m still not really interested in the
meta-ethical squabbles between non-naturalism and moral supernaturalism.
(Though I know they interest some of you!) But both axiarchism and AP theism
support a much stronger moral realism, with much more robust values, than I would
have expected.
That’s the main impact of AP on my life. But it’s not the only one. The thought
that there might be cosmic values makes me take all my life choices more seriously.
I try to bring my life into harmony with those values. I know that sounds very vague
and New Age-y. And, of course, the cosmological argument alone doesn’t give very
much guidance, especially in its axiarchic interpretation. But even there the under-
lying thought is that creative values gave rise to this physical universe because it is
valuable. That suggests to me, however dimly, that a creative life centred around
genuine values is the way to go. And this idea is much clearer on a theist interpret-
ation of AP, where the origin of the universe lies in the actions of a creative person.
(Of course, I was already inclined to regard creativity as important. Otherwise
I probably wouldn’t find the cosmological argument compelling in the first place.
But AP does give my pre-philosophical values a new impetus.)
The cosmological argument may not give us much moral direction. But it does
bolster the more detailed moral advice that, I’m sure, you’re all going to offer!
FI-TU: I’ve always been a fan of evolution and of deflationary arguments. BT is too
anthropocentric for me, and I find the Darwinian critique of specific design argu-
ments compelling. But I’ve recently been drawn to AP by the fine-tuning argument.
Unlike COSMO, I don’t accept any Principle of Sufficient Reason. Not every fact has
an explanation. But some facts are so remarkable that they do cry out for explanation,
and then it is a very great defect if a particular world view cannot explain them.
I could accept some universes as brute facts, but I find it very hard to accept this
universe, which seems so remarkably fine-tuned for life, as a brute fact. And the
extant materialist explanations simply don’t work. Multi-verse stories don’t have
enough empirical support, and in any event they are defeated by Mellor’s probability
objection.
Like COSMO, I’ve never been directly interested in ethics. But reflecting on why
I find the fine-tuning argument so compelling suggests to me that I do regard its
A DIALOGUE 
intelligibility, its friendliness-to-life, and the fact that it is partially understood by

some of its inhabitants as remarkable and significant features of our cosmos.
Unlike the cosmological argument, the best teleological argument tells us quite a
lot about the content of the cosmic purpose. If we accept an argument that brings
together the best human features, cosmic features, and fine-tuning arguments, then
we learn that the potential for the emergence of conscious, reason-following, value-
perceiving agents is built into the fabric of the cosmos; that the cosmos is designed to
be governed by intelligible mathematical laws; and that it is arranged so that creatures
will emerge within it who can comprehend it.
Let me flesh this out a bit more. My starting point is the emergence of life within
the cosmos, especially conscious rational life that is both capable of recognizing and
responding to values, and able to begin to comprehend the cosmos itself. (At this
stage, our best evidence for the existence of life of that kind is humanity. But human
beings might turn out to be mere by-products or preconditions of the real cosmic
purpose, rather than essential elements of it. And, given the argument from evil,
I think that such secondary roles are the most we could possibly hope for.) The
cosmic coincidences are so striking that the obvious conclusion is that the cosmos
was fine-tuned so that life would emerge. This gives us I, U, and FL as significant
cosmic features. (Recall that these are the facts that our universe is intelligible, that it
is understood by some of its inhabitants, and that it is friendly-to-life.) Now recall the
distinction between global and local probability, and Mellor’s objection to multi-
verse explanations of the cosmic coincidences. To offer a superior explanation here,
we must posit an objectively significant order in the cosmos. Given our current
scientific models, the natural place to look for that order is mathematics. Our
universe is governed by complex mathematical laws, and their comparative math-
ematical complexity provides an ordering of possible universes. So M (the fact that
our universe is governed by precise and elegant mathematical laws) is another
putative cosmic value.
Another thing I have in common with COSMO is that AP has given me a renewed
interested in possible sources of cosmic knowledge. In my case, that primarily means
cosmology. (One obvious focus for current research is to evaluate the epistemic
credentials of the multi-verse, the main non-purposive alternative.) Many scientists
hope that some future Theory of Everything (TOE) will provide a full and satisfying
explanation for cosmic coincidences, and thus for all the remarkable features of our
cosmos. I share that hope. I see no reason to abandon science. But thinking through
the fine-tuning argument has taught me to distinguish between science and valueless
purposeless materialism. Any future TOE must overcome the deficiencies of current
materialist explanations. I’m inclined to think that, to do so, it will need something
like cosmic purpose or cosmic values.
Having rejected BT as too anthropocentric, I worry that all our attempts to capture
the cosmic purpose will turn out to be equally human-centred. The cosmic purpose
might be very different from anything we expect. In the meantime, however, I think
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
the fine-tuning argument does offer some (fallible) guidance. The actual cosmic
purpose may be quite different from M, I, U, and FL, but they remain our current
best guesses. As well as pursuing cosmic knowledge itself, I try to choose personal
projects that reflect those (tentative and fallible) cosmic values. As COSMO said, if
I hadn’t started with some sense that those things mattered, then I probably wouldn’t
have had any sympathy for any fine-tuning argument that seeks to explain them. But
I come out with a stronger commitment to those values that I had going in.
How exactly does AP affect my life choices? Well, first I try to embody U—to
understand both the universe and its purpose. All knowledge and understanding has
a new significance. Second, I look for abstract analogues of what the underlying
purpose might be, such as being governed by intricate mathematical laws, or com-
bining surface complexity with fundamental simplicity. And then I look for projects
that are like that. Again, this is re-enforcing values I already had. But that doesn’t
mean it makes no difference. All life choices are comparative. AP strengthens some of
my existing moral commitments but not others. I still take trivial pleasure or fun or
wealth into account, but I can’t see how any of these could be analogues of any
credible cosmic purpose. So I now give those factors less weight in my life. As most
life choices involving either weighing trivial pursuits against significant projects, or
weighing one significant project against another, this makes a very big difference.
One final role of AP is this. Exploring these issues has brought home to me the
limits of human understanding. AP raises the disturbing possibility that real cosmic
significance lies in knowledge that passes our understanding. But that would be to
trespass on SCALY’s territory.
SCALY: I’ve long found BT inadequate as an explanation of the scale of things.
Why would a human-centred God create such a vast cosmos? On the other hand, I’ve
always been sympathetic to the explanatory arguments for theism, especially the fine-
tuning argument. So a non-human-centred cosmic purpose strikes me as offering the
best explanation overall.
I start where FI-TU stops. Suppose the fine-tuning argument, or something else,
has persuaded us that purposeless materialism cannot explain our cosmos. That
leaves a choice between BT and AP. And then considerations of scale favour AP.
The argument from scale is much more speculative than other arguments, espe-
cially cosmological and teleological ones. This is partly because it has been compara-
tively ignored by philosophers. But the argument is also speculative in two other
ways. First, the very facts it sets out to ‘explain’ are themselves the objects of
speculation. The cosmological argument turns on the fact that there is something
rather than nothing; while fine-tuning explains why life has emerged. No one denies
that there is something or that life has emerged. The argument from scale turns on
facts we do not yet know. We know the universe is very large, of course, both spatially
and temporally. However, the most compelling argument from scale doesn’t rest on
the sheer physical size of the cosmos, but rather on the nature or distribution of its
inhabitants. And that is something we do not yet know.
A DIALOGUE 
The final sense in which the argument is speculative is that it draws on themes
from speculative fiction. Science fiction leaves some philosophers cold. But if we seek
to explore AP, then the literature and criticism of science fiction has much to teach
us, because non-human (and thus non-human-centred) perspectives have long been
a central preoccupation of that literature.
I think the key both to understanding the content of cosmic purpose, and to
answering the prior question whether there even is a purpose, is the search for
extraterrestrial life. If the universe is widely populated by a range of different rational
beings, and especially if these include beings who are intellectually or morally
superior to us, then AP is much more plausible than BT, and we can reasonably
hope to learn incomparably more about cosmic values from alien sources than we
could ever figure out for ourselves. At the other extreme, if we truly are alone, then
perhaps BT is credible after all. (Although other arguments, such as evil, may still rule
it out. And, as I’ll explain shortly, I think we have independent reason to discount the
possibility that we are alone.)
We need to consider several intriguing intermediate options. One is a universe
with infinitely many inhabitants. This possibility would favour AP for two reasons.
First, of course, if there really are infinitely many alien races, it is very unlikely that
none of them is better than us! An infinitely populated cosmos almost certainly
contains some very superior beings. Second, even if no non-human species were
clearly superior, the infinite cosmos would still favour AP. Or, at the very least, it
supports a view of cosmic purpose that is closer to AP than to BT. For ease of
exposition, let’s adopt a theist interpretation of AP. We can express the relevant
purpose as follows: God created the universe so that it would be home to (infinitely
many) rational beings, but the precise details of individual rational lives are irrelevant
to God. A universe without any human beings would have served just as well.
Another intermediate possibility is that the superior beings who truly understand
the cosmos might be our own (technologically enhanced or entirely digital) descend-
ants. Here, again, this leads to the borderland between AP and BT. Suppose human
beings are a step along the evolutionary ladder to truly significant beings. We can
separate two possibilities here. Our superior descendants might be produced by blind
biological evolution, acting without direct human intervention. Or they might be our
(deliberately created) digital descendants. (It is, of course, more likely that any truly
significant digital beings will be our distant descendants: machines designed by
machines designed by machines . . . designed by human beings.) In either case,
while human beings are indirectly relevant to cosmic purpose, our welfare is no
part of that purpose. (We could, of course, combine the two intermediate options.
Perhaps God has created infinitely many evolutionary chains, of which ours is merely
one. The success or failure of any individual chain is cosmically irrelevant. Even our
super-enhanced descendants don’t really matter.)
The discovery of intelligent aliens would be relevant to many other arguments for
AP. To cite two examples: we won’t appreciate the full force of the challenge to BT
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
from religious diversity until we have before us, not just a variety of familiar human
religious traditions, but also the vast array of as yet unimagined extraterrestrial
alternatives; and facts about alien science and religion will obviously force us to
re-evaluate arguments from fine-tuning or mysticism in ways we cannot now foresee.
Like both COSMO and FI-TU, AP re-enforces my own existing interests here. I’ve
always been fascinated by science fiction tales of aliens or infinite universes. AP
wouldn’t have got a grip if I didn’t have those interests already. But AP does give
those existing interests a much greater significance. The search for extraterrestrial life
is no longer an eccentric curiosity. It is now the central moral and metaphysical
question.
What if the search proves futile? Well, of course, this is an area where positive
proof is possible while negative proof is not. We will never know we are alone. Given
the size of the physical universe, I don’t think we can ever hope even to be confident
that we are probably unique. And in the absence of compelling evidence, I think it
would be a moral failing, a capricious grandiosity, to proceed on the assumption that
we are unique, or even that we might be.
The argument from scale highlights the need for an ethic of belief. The existence,
or otherwise, of extraterrestrials is not irrelevant to human concerns. But nor is it a
question we can answer at the moment. To borrow a phrase from philosophy of
religion, our current knowledge leaves our universe ambiguous regarding alien
intelligent life or future intelligent digital life. We need a response to that ambiguity.
AP appeals to me, in large part, because I’ve always felt that the one non-self-
aggrandizing response, the only humble attitude, is to proceed on the assumption
that we are not alone.
Elevating the search for extraterrestrial life is the main impact of AP. But it is not
the only one. Like COSMO and FI-TU, I try to imagine how the cosmic knowledge
I seek might transform my life. Obviously, once we do get alien cosmic knowledge it
will have a profound impact. Equally obviously, we presently have no real idea what
values, moral practices, or sources of moral insight superior aliens might reveal. So
the practical upshot is limited. But I do always ask myself: what might an alien think
of my decisions, my actions, my life? Of course, that question is available to the
atheist, or to the person who doesn’t believe in aliens. (Hume asked it, after all. His
ideal observer might as well be from another planet.) But it has a special resonance
for me, for two reasons. First, I don’t think my extraterrestrial ideal observer is
imaginary. She is not merely a theoretical construct or an idealized self. So the
question is real, even if I don’t know the answer. Second, I don’t regard extraterres-
trial observers as merely offering a different perspective on human affairs. I regard
them as genuinely ideal. Like COSMO and FI-TU, I think we can only explain all the
non-moral facts by positing objective cosmic values. The real superiority of my
imagined aliens is not intellectual or technical but moral. And their moral superiority
has two dimensions: the superior aliens matter and they have superior moral
knowledge. (Could the two come apart? Perhaps not. If, as FI-TU argued, U is part
A DIALOGUE 
of the cosmic purpose, then at least some of the universe’s morally significant
inhabitants must understand the cosmic values. And even if moral knowledge is
not essential for cosmic significance, a universe containing beings who matter could
well contain other inhabitants whose moral knowledge exceeds our own.)
My question is not about the quaint preferences of space-ship-piloting monsters. It
is this: what will the actual aliens who are in touch with the transcendent cosmic
values think of how I live? That is a sobering question. And only AP opens the space
where that question can be asked.
MYSTY: My background is different to that of COSMO, FI-TU, and SCALY. I began
with a commitment to BT, as a practitioner of one specific Christian mystical doxastic
practice. But then I started to worry about diversity, especially as I encountered
contemplative practitioners from other religions.
As someone within an actual religious tradition, who engages with actual mystical
texts, traditions, and experiences, I often find that philosophical discussions of
mysticism miss the point. Some deficiencies are factual. Philosophers often operate
with outdated models of what mysticism is. They focus on isolated experiences of
pure consciousness that anyone could have, as opposed to the disciplined activities of
professional contemplative elites embedded in socially established religions with
scriptures, authorities, and traditions. Mystical theology is far removed from Russell’s
‘man who drinks too little and sees God’!
I agree with Alston that we should treat mysticism as a series of established
doxastic practices, and try to measure them open-mindedly against the same stand-
ards as other established doxastic practices. I’m especially interested in comparisons
with sense perception, mathematics, and morality. Unsurprisingly, as someone
interested in both mysticism and morality, I was already sympathetic to Alston’s
doxastic pluralism. The naturalist project of distorting or reducing all human experi-
ence to fit a single mould has never held much appeal.
However, I’m much more worried about religious diversity than Alston is. Of
course, mystics have always confronted other religions. Religious diversity is nothing
new. And each tradition has its set of responses, my own included. But I don’t find
any traditional response fully satisfactory. If I recognize practitioners of mystical
practices in other religions with contradictory doctrines as my epistemic peers, how
can I be confident that I am right and they are wrong?
One response is to deny epistemic parity. But I find that implausible. When I read
the writings of mystics from very different backgrounds, or meet with other practi-
tioners, I recognize more similarities than differences, at least in terms of sincerity,
perceptiveness, discipline, and all the other marks of genuine mystic insight. (I’m
thinking here, not only of other Christian denominations, or even other Abrahamic
faiths, but also of contemplative traditions in Buddhism or Hinduism.)
A second response is to try to reconcile seemingly contradictory traditions. All the
various attempts at reconciliation, from perennialism to Hickian pluralism to anti-
realism, strike me as versions of atheist nihilism rather than genuine alternatives to it.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
Hick’s pluralism, in particular, has always struck me as an abandonment of trad-

itional religion, not an interpretation of it. (It doesn’t help that I’ve always found
Kantian metaphysics unappealing!)
A third response is to insist on an exclusivist interpretation of my own tradition.
But I can’t shake the feeling that exclusivism is arrogant. I agree with Plantinga that
theists are not obliged to abandon exclusivism in the face of diversity. But
I personally find exclusivism untenable.
A fourth response is the radical atheist solution, where all mystical experience is
dismissed as delusional. Should I simply dismiss my own mystical tradition as delu-
sional, or accept the crude deflationary explanations of Freudians, postmodernists,
Marxists, or evolutionary zealots? These all seem under-motivated to me, especially
given the links between mysticism and morality which I outline below.
A final response is quietism. But Alston’s alternative of ‘sitting tight’ in the face of
epistemic parity doesn’t work for me either. Religious diversity blurs the boundaries
between first-personal and third-personal perspectives on my own tradition. Epi-
stemic parity makes it hard to grant authority to my own direct experiences without
according similar authority to the testimony of other religious traditions. But then
I quickly end up with contradictory reports whose epistemic credentials are more of
less on a par.
I’ve been looking for a middle-road—a way to salvage some epistemic role for
mysticism in the face of religious diversity and epistemic parity. And AP fills that gap.
I like AP’s explanation because it preserves some epistemic role for mysticism, while
offering a new interpretation that explains epistemic parity. Our various mystical
traditions are misperceptions of something real, rather than mere delusions.
My interest in mysticism has always had a strong moral component. My own
tradition, like many others, emphasizes the moral credentials of the true mystic.
Moral improvement is an essential element in all plausible ‘over-rider systems’, in
Rowe’s phrase. And I think religious contemplation has been a source of genuine
moral knowledge—especially in relation to altruism, universality, and impartiality.
Unlike COSMO, FI-TU, and SCALY, who say they began their exploration of cosmic
purpose without explicit moral interests, my interest in mysticism has always been
driven, in large part, by moral dissatisfaction. What draws me both to religion in
general, and to contemplative traditions in particular, is a sense of the urgent
normative pull of moral demands that are independent of both my own desires
and any cultural practices or norms. And I don’t think we can make sense of that
urgency without insights from religion.
In meta-ethics, I’m attracted to moral supernaturalism: to the idea that mys-
tical experience is experience of cosmic values. I’ve never found anti-realist or
moral naturalist accounts of our moral practices convincing, and non-natural
moral facts seem too ungrounded. Moral supernaturalism fits my own experience
of the tight connection between contemplation and morality. I wouldn’t go so far
as to insist that all moral philosophers should become mystics, or even that they
A DIALOGUE 
should all acquire some minimal contemplative competence. (Although I don’t

think this suggestion is any more absurd than the idea that moral philosophers
need expertise in formal logic or game theory or neuropsychology or economics
or quantum physics.) But I do think moral philosophers should take very
seriously the possibility that mystics are moral experts—that some minority
doxastic practices might offer insight into cosmic values. Contemporary philoso-
phy vacillates between treating moral knowledge as impossible or illusory
(cf. nihilism), or as ubiquitous and unproblematic (cf. all appeals to common-
sense intuitions or ‘platitudes’). But perhaps moral knowledge is instead an
uncertain glimpse of a supernatural reality that is grasped only dimly and rarely
and only by a few.
Despite our different starting points, I do share FI-TU’s interest in the intelligi-
bility of the universe, and SCALY’s search for extraterrestrial modes of cosmic
knowledge. I’m inclined to interpret both in mystical terms. Truly valuable under-
standing, for me, is likely to resemble contemplation rather than physics. But perhaps
aliens won’t have our sharp divide between the empirical and the mystical. Like
FI-TU and SCALY, I think we should be very open-minded about the future of
‘science’. I especially reject the idea that the success of science somehow proves, or
even supports, materialism or atheism.
I’m also sympathetic to some other teleological arguments that FI-TU didn’t
mention, such as special features arguments based on consciousness and moral
experience. I think some features of human experience are very difficult to cash out
in purely material terms. I find it impossible to either accept consciousness itself as an
anomalous brute fact, or to buy any extant physicalist reduction of consciousness.
I guess that’s not surprising, given my suspicion of deflationary materialist explan-
ations of mystical experience in particular.
My main question, on a day-to-day basis, is how to reinterpret my own contem-
plative tradition, and especially its moral tests, in light of AP’s non-human-centred
cosmic purpose. My tradition provides internal resources here. Christianity already
has a strong commitment to self-transcendence—pushing us to seek a more other-
centred or God-centred perspective. So I try to see AP as another, rather unexpected,
step along a familiar road. (If we take on board the lessons of other contemplative
traditions, then perhaps that step is not itself without precedent.) The challenge, and
it is one we all face, is how to integrate AP’s austere cosmic values into a human life.
One question that interests me is the value of interpersonal human relationships,
which seem to have no cosmic value at all.
COSMO, FI-TU, and SCALY focused on personal projects that resemble, imitate,
or involve cosmic purpose. My tradition supports these idioms. But it also offers a
different ethical strand, based on interpersonal relations and moral exemplars. Our
ethical thinking, both religious and secular, often starts from a story of how some
actual moral individual responded in some actual human situation. (We call these
people ‘saints’.) The details of those stories needn’t concern us here. The question is
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
how to fit them into an AP framework. How can human moral responses to human
situations have any cosmic significance?
I answer this question by borrowing another theme from my own tradition. We
are taught to model our relations with one another on God’s relation to creation. This
still makes sense under AP. Of course, within AP, all talk of God as personal, and of
God’s motives or attitudes, must be analogical. These metaphors are concessions to
our limited human understanding, not claims about objective reality. But any
reasonably sophisticated BT already makes this analogical move. I don’t think it is
too much of a stretch to interpret AP by positing a being who responds positively to
cosmic value, and who therefore can be said to love the aspects of creation that
constitute the reason for creation. And like the fine-tuning and scale arguments,
mysticism can give us information about those reasons. (Although, I must confess
that, once we move beyond the interpretive matrix provided by specific religious
doctrines and texts, the lessons of mysticism are much more opaque and harder to
read than more straightforward evidence such as the cosmic coincidences that make
possible the emergence of life. But then again, who said that human interpretation of
cosmic values should be easy?)
Divine creation can provide a model for human life, not only in our general
attitude to creation, but also in our relations to one another. An ethical exemplar is
then someone who translates the divine attitudes into human terms. Of course, it is
very hard for us to tell who the right exemplars are. And the gap between human and
divine means that many very different human responses might be equally faint
echoes of the same divine original. But, again, this is not a new difficulty. Diversity
of human saints is also found in BT, especially (but by no means only) in those
traditions that are more ecumenical or open to other faiths.
Any AP morality is less definite than a traditional BT ethic. But AP still offers
a more realist interpretation of morality than any atheist alternative. Non-
supernaturalist exemplar theory offers nothing to unify our disparate examples and
nothing non-natural to ground their significance. (Appeals to ‘human nature’ in
contemporary virtue ethics have no normative force because they rely on an out-
moded teleological biology.)
AP provides moral ideals that are austere, impartial, and demanding. This gives
morality an objective push that is unavailable to the atheist. For those who feel
morality needs that push, this counts very heavily in favour of AP.
ONTO: My main interest is in the ontological argument. I have never found the
human-centred aspects of BT appealing, and theodicy doesn’t persuade me. But
the modal ontological argument has always fascinated me, and I think there is
something to it. The main stumbling block is to prove the possibility of a perfect
being while avoiding parodies. I’m exploring both practice-based presumptions of
possibility such as those offered by Pruss and Alston, and also direct experiences
of possibility drawn from Anselm’s own tradition of contemplation. In both cases,
I think the ontological argument needs prior evaluative commitments. Without
A DIALOGUE 
independent values you can’t decide between competing conceptions of perfection or

worship-worthiness.
Anselm’s argument needs a distinction between the merely conceptual thinking of
the fool and the richer understanding of the enlightened believer. The fool can think
that God doesn’t exist, while the believer realizes that this thought does not corres-
pond to any real possibility. (Just as the person who understands the word ‘chimera’
knows that, being a mythical beast, the chimera cannot exist.) I think the failure to
produce a compelling demonstration of God’s existence shows that understanding
cannot be attained by metaphysical or logical reasoning. Many analytic commen-
tators conclude that the ontological argument fails. I conclude instead that under-
standing must come from elsewhere. The ontological argument requires doxastic
pluralism. I am especially interested in mysticism and morality as sources of
understanding.
Anselm’s own text begins with an invitation to God to reveal the truth. For
Anselm, understanding is a gift from God. That points to mystical traditions. But a
divine gift might also take the form of modal or moral beliefs, or social practices.
Because I’m suspicious of the anthropocentrism of Anselm’s BT, I reinterpret his
‘divine gift’, not in individual terms, but in cosmic ones. Understanding is not God’s
gift to me (or to Anselm), but rather a gift to rational creatures throughout the
cosmos (or even a gift to the cosmos itself). This is analogical talk, of course. But, as
MYSTY said, all human talk about God is analogical to some degree.
There are two premises to any ontological argument. In Hartshorne’s neat formu-
lation, one premise tells us that God’s existence is either necessary or impossible,
while the other tells us that God’s existence is not impossible. Let’s call these
requirements: ‘conditional’ and ‘antecedent’. I doubt that logical argument alone
can establish either premise. There are infinitely many possible concepts of God.
Some meet the conditional requirement, others don’t. And more to the point, there
are too many different concepts that meet the conditional requirement. The long
history of parody shows that many different necessary beings are thinkable.
We need to pick the right concept of God. And here, I believe, we need to look to
the role of God in actual human lives. The obvious problem is religious diversity. Like
MYSTY, I’m not satisfied with any of the BT responses to diversity. (For instance,
I don’t think Pruss has any sound basis for insisting that Christian belief in a
necessary being plays a central role in making lives meaningful while Buddhist beliefs
do not.) So I think a necessary preliminary to any successful ontological argument is
a reinterpretation of religious experience or practice that admits a diversity of
conflicting traditions that all enjoy epistemic parity. Like MYSTY, I think AP is a
better bet here than either atheism or BT.
The second requirement is the antecedent one: God is possible. Again, I think we
must appeal to practice or experience here. That means we must consider the two
premises together. We seek an interpretation of some realm of human experience
(modal, moral, mystical) that gives a central role to necessity (or some weaker modal
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
notion such as Leftow’s worship-worthiness), but also points to the possibility of a

necessary being. That’s a tall order, but I don’t think it is absurd. And here too I think
AP breathes new life into this argument.
Unlike other arguments such as fine-tuning, and especially evil, philosophers tend
not to draw moral implications from the ontological argument itself. This argument
tells us that God exists, but we must look elsewhere for advice on how to respond to
God’s existence. I think this sharp divide between the metaphysical and the moral is
too simple. The ontological argument is very abstract, even more so than the
cosmological. You might think it wouldn’t have any real-life implications. But it
does have some. As COSMO said, the very fact that I believe there is some cosmic
purpose, even without any further knowledge, motivates me to seek cosmic know-
ledge, and gives all my life choices a new seriousness. But the ontological argument
also has a substantive impact. By focusing our attention on a very demanding ideal of
perfection, it pushes morality in a certain direction.
Within AP, I think the ontological argument has several distinctive moral roles.
The first is to emphasize modal elements: a perfect being is one whose existence and
qualities are not accidental, and whose power extends to all possible worlds. It is a bit
of a stretch to translate that into one’s own life, of course. (No human endeavour is
metaphysically necessary!) But a new reason to care about permanence, solidity, and
perfection can affect one’s major decisions, even if only at the margins.
It may seem vainglorious to try to bring transcendent perfection into my own life.
I think of it like this. Our ethical talk already has many metaphors based on
resemblance to a perfect ideal. If we take the ontological argument seriously, then
those metaphors may be taken literally. There is an obvious parallel with SCALY’s
discussion of alien moral observers or agents. Real perfection has greater normative
force than any imaginary ideal.
The ontological argument also increases the cosmic significance of modal logic and
modal explanation. (Just as, for the rest of you, your favourite arguments raise the
cosmic significance of cosmology, evolution, mysticism, or the search for extrater-
restrial life.) If modality is linked to real perfection, and especially if that perfection is
responsible for the existence of the cosmos, then exploring modality is no longer a
frivolous academic enterprise. Following Leibniz, we might even conclude that we
need to invoke God (or cosmic purpose) to explain modal facts, thereby linking
purpose and modality even more closely.
Another distinctive role is that the ontological argument supports a particularly
strong meta-ethical story about cosmic value. Analytic critics make fun of Anselm’s
unfashionable Boethian metaphysic with its degrees of reality. I do think we can
detach Anselm’s insight from his particular Boethian story. But no ontological
argument can dispense with transcendent objective values altogether. And those
values must yield a scale of goodness that we can use to rank both possible objects
and possible worlds. If AP is to borrow an ontological argument, then it is committed
to very robust independent values.
A DIALOGUE 
A final role is that the ontological argument reinforces mysticism. There is a

slippery slope between taking Anselm’s argument seriously and taking his contem-
plative tradition seriously. Anselm himself wrote the Proslogion to help his charges
become better monks. Once we grant the crucial significance of exploring the concept
of (moral) perfection, both Anselm’s formula (‘that than which none greater can be
conceived’) and its original context make much more sense.
Anselm’s example might also influence our reading of mysticism. I don’t have
MYSTY’s background in genuine mysticism, but I have tried Anselm’s procedure as
part of my own ethical decision-making. I seek a clear concept of a perfect being by
progressively imagining better possible beings, and then identifying and trying to
imagine away their imperfections. Even if a positive experience of a perfect being
never comes along, the process of conceptual clarification can be helpful. But without
the motivation of the ontological argument in the background, I doubt I would
bother.
EVE: My starting point is a set of liberal utilitarian moral commitments that make
it hard for me to accept any BT theodicy. That led me initially to atheism. But I’ve
come to think that a purposeless materialist universe is not a good match for my
moral commitments either. I’m a moral realist who finds both moral naturalism and
atheist non-naturalism unconvincing. So I’m drawn to either moral supernaturalism
or a model where non-natural moral facts mutually support the explanatory argu-
ments some of you favour.
My main personal interest is how AP morality might differ from both atheist and
BT ethics. I’m interested in two sets of moral resources: first, resources that are
available to BT and not to atheism, because they rely on God; second, resources that
are available to atheism and not to BT, because atheists don’t need to worry about
theodicy. AP can borrow both sets, because it recognizes cosmic purpose without the
need for theodicy.
I start with the distinctive resources of BT. Can AP borrow them? Of course, many
BT moral resources are not open to AP. There are no divine commands issued to
human beings, no divine beings becoming incarnate as humans, no special divine
plan for my life, and so on. These are the most obvious ethical resources of BT. But
they are not the only ones. God’s plan for creation might ground transcendent
cosmic values that can play any number of roles in our moral lives. We can
contemplate, study, admire, copy, resemble, or worship cosmic values, and use
them as a model for our projects, lives, or relationships. Atheists have values too,
of course, but they lack the solidity and normative force of the BT cosmic values that
AP inherits. As someone with consequentialist leanings, I find it very natural to think
of morality in terms of optimal responses to independently objective values. Like
MYSTY, I think atheists simply cannot do justice to the external normative pull of
moral demands.
What does AP add that I couldn’t get from BT? Here we turn to the distinctive
resources of atheist ethics. The big lesson here is what we can now reject: the
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
overvaluing of freedom, especially libertarian freedom, compared to suffering or

achievement; the need to posit an afterlife or to believe (against all the evidence)
that the world is just. It is not only that I find these things impossible to believe, but
I also think my ethical outlook is different (I would say ‘better’) without them. I find
AP liberating in the way that many people find atheism liberating. AP opens up a
more liberal this-worldly-focused ethic, but one that retains the full normative force
of divine purpose.
From my moral perspective, BT has two main ethical defects. The first is
its commitment to human metaphysical specialness. For any follower of Jeremy
Bentham, this is the height of caprice! Human specialness pervades BT thought,
and corrupts BT ethics in many ways. One striking example is the ethics of our
relations with non-human animals. I agree with van Inwagen that the discovery that
animal suffering long pre-dated humanity has great significance for natural theology.
Evolution has taught us the volume of animal suffering. Even the most extravagant
theodicy cannot redeem this evil. No individual animal can come to understand the
cosmic purpose of its own suffering. BT is thus driven to posit a moral gulf between
human beings and other animals. Their suffering cannot matter very much at all. As
a utilitarian, of course, I reject this sharp divide between humans and other animals.
Suffering is of central ethical significance, and their suffering can matter every bit as
much as our own.
Of course, AP has difficulty making moral sense of suffering. You might think it
fares worse here than BT, which at least accords cosmic significance to human
suffering. How can utilitarians, of all people, embrace a view where suffering doesn’t
matter at all?
This, for me, is the most difficult challenge of AP. I don’t think AP has a very
convincing answer—certainly not yet. But I also think both atheism and BT fare
worse. Or, at least, they face such severe problems that it is worth trying to get AP to
work. The deficiency of BT is that, to construct a theodicy, it must (in effect) both
deny the ethical significance of animal suffering and attach great importance to the
(ethically unimportant) divide between animals and humans beings. In terms of
moral consistency, I think this is actually worse than AP’s total indifference. The
problem for atheism relates, not to suffering per se, but to its inability to make sense
of any values at all, as we saw earlier.
AP’s story about suffering, both animal and human, has to involve the same
appeals to analogy or resemblance that ground any AP-based human morality.
I find this easier to explain in the case of pleasure, rather than suffering. Desirable
consciousness, in Sidgwick’s phrase, involves a human connection to objective
values. Perhaps pain is similarly related to cosmic disvalue. I agree there is a lot of
work to do here. But I think this is the route to explore.
A second example of the caprice of BT is its insistence that each human life matters
to God. A person-affecting divine morality reflects human metaphysical specialness.
Of course, that makes theodicy much more difficult. BT must now argue that each
A DIALOGUE 
person (I would say each ‘creature’) gets the best life he/she/it could possibly have
enjoyed. Even if it rejects this maximalist formulation, BT must still insist that every
person has a very good life containing no unredeemed horrendous evil. This seems
prima facie implausible, given what we know about the world. (There is a connection
here to BT optimism, which I will discuss soon.)
If we reject both a sharp divide between animals and humans, and a person-
affecting morality, then we can consider more impersonal stories about our cosmic
role. Perhaps, although our suffering has some moral significance, it is outweighed by
some greater cosmic good. This is a sobering thought, and no doubt a morally
troubling one. But it should not be unthinkable for the utilitarian. (Like SCALY,
I also think we need to explore intermediate positions between AP and BT, where we
have some cosmic role, but it is less significant or flattering than we thought.)
The other elements of human metaphysical specialness are more familiar. IMMY
will deal with immortality, life after death, and rebirth. That leaves free will. BT
requires very controversial metaphysical and moral claims about human free will. As
a utilitarian, whose morality is primarily forward-looking and value-based, I find
BT’s obsession with free will unappealing. (It reminds me of the worst excesses of
Kant.) BT makes metaphysical demands that go far beyond the comparatively
modest needs of our moral practices. (Or perhaps I should say, to be more accurate,
that the moral practices we need can do without extravagant free will claims. If we
have to jettison some of our nastier libertarian or retributivist leanings, then so much
the better.)
The main advantage of AP here is that, unlike BT, it can remain agnostic about the
metaphysics of freedom. AP also leaves us free to explore different stories about the
value or meaning of freedom. We need not insist that we have incompatibilist or
contra-divine freedom, or that valuable freedom must include the live option of
inflicting horrendous evil on innocent beings. We can be (rightly!) suspicious of self-
aggrandizing interpretations of our own ‘experience of freedom’, of metaphysical
pictures that make our moral choices causally unique in the universe, and of moral
theorists who insist that we need bad options to make good ones ‘truly’ valuable.
There may be a connection here to AP’s openness to extraterrestrial life. AP is
open to the discovery of new kinds of freedom, or new forms of autonomous liberal
social existence—whether in thought experiments, or in metaphysical theories about
ourselves, or in empirical findings about animals, or even in the lived experience of
aliens or intelligent machines. (Perhaps the creatures who enjoy the best freedom are
yet to be born!)
One possibility that especially interests me is that the true value of freedom lies in
choices between goods, not in the hackneyed BT choice between good and evil. Once
we recognize this point, we begin to see how much better our world could be.
This brings me squarely to the second failing of BT: its cosmic optimism. As a
consequentialist, of course, I think BT remains committed to Leibniz’s dictum that
this is the best of all possible worlds. I know many theists follow Adams and
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
Swinburne in watering this down. But unless it radically departs from divine
omnipotence, BT must insist that this world is very good and very just. I think this
is an insidious commitment. Aside from leading us to postpone the search for justice
to some hypothetical afterlife, it also distorts our views about justice to fit the way the
world actually is. It is no coincidence that, with some honourable exceptions, BT
regimes have tended to be very socially conservative. (And, of course, they would be
even more conservative if, as they should, they embraced reincarnation!) It is too easy
to shift from insisting that everyone will eventually get what they deserve to pro-
claiming that everyone deserves what they currently have. By contrast, AP allows us
to conclude that the cosmos is very unjust indeed, at least by any human standard.
There is a price, of course. Without theodicy, without the assurance of a human-
centred divine providence, I have no guarantee that any of my concerns has any real
cosmic significance, or that any of my plans will ever succeed. But that comfort isn’t
available to the atheist either. And perhaps that is just how life is.
IMMY: My background is different again. I start from the British Idealist thought
that a worthwhile life must be grounded in some transcendent eternal value. But I’m
also persuaded by Sidgwick’s critique of Idealist metaphysics, and I’ve come to regard
the BT insistence on individual survival as too self-centred. The moral argument
for immortality doesn’t convince me, but I do think there is something to it.
A meaningful life does need something that I feel the atheist cannot supply.
Given my Idealist background, I’m especially sympathetic to COSMO’s Platonic
axiarchism, to moral supernaturalism, and to MYSTY’s and ONTO’s linking of
mysticism and morality. But my distinctive interest is whether AP has an alternative
that can replace immortality or cosmic justice in BT’s ethical scheme. I agree that AP
can’t offer a guarantee that our lives aren’t meaningless. But I do think it offers more
hope than atheism. That is partly because, in common with many of you, I believe
both that moral nihilism is a more significant cause of despair than cosmic injustice
and also that AP can ground moral realism.
I’m also exploring more direct AP substitutes for personal immortality itself. My
starting point is the thought that drove me to AP in the first place. Isn’t it unduly self-
centred to focus on my personal survival? Shouldn’t it be sufficient if some larger unit
of either agency or concern continues after my death, or extends its reach beyond my
lifetime? AP doesn’t allow me to think of myself an eternal agent. But a larger finite
unit might have cosmic significance in a way that my individual life cannot. Con-
siderations of scale make me wonder whether even the whole human story is large
enough. Perhaps our best hope for cosmic significance is to think of our lives as (very
small) pieces in a vast cosmos-spanning response to cosmic value. A fanciful thought,
no doubt, but no more fanciful than the comforting tales of BT and Idealism down
the ages. And more hopeful than any purposeless materialist alternative.
Let me fill in a few more details. Like EVE, I’m worried that the strong metaphys-
ical demands of theodicy limit our exploration of the ethical possibilities. And I think
those commitments are both more demanding, and more controversial, than many
A DIALOGUE 
BTs realize. I’ve always been impressed by McTaggart’s purely metaphysical argu-
ments for pre-existence. So I don’t think we can dismiss rebirth or reincarnation out
of hand. But once pre-existence is on the philosophical table, modern liberal notions
of justice seem to demand it. Of all the possible worlds that are epistemically possible,
given what we know about this world, the most just ones are those where human
suffering is deserved because of our actions in previous lives.
Theodicy needs pre-existence. But—and for me this is a very big ‘but’—evidence
and argument simply are not sufficient to establish pre-existence. And to presume
pre-existence in the absence of compelling evidence is to risk very serious moral
error. (It amounts, in effect, to a presumption of guilt for those who suffer horren-
dous evil.)
There is a wider issue here. Reflecting on the rapid demise of British Idealism, and
other shifts in philosophical fashion, has taught me the ubiquity of reasonable
disagreement, especially within both metaphysics and ethics. I think one central
task of moral philosophy is not to resolve those disagreements, but to help us to live
with them. And here AP is much more flexible than BT.
Another site of reasonable philosophical disagreement is personal identity.
Because it is tied to both person-affecting divine morality and personal post-mortem
survival, BT focuses on criteria for individual survival and responsibility. (After all,
Locke’s seminal discussion was itself motivated by the fact that God will hold me
responsible on the Day of Judgement for my past acts.) I think both Idealism and
utilitarianism teach us, instead, that our patterns of concern can take many different
forms, and that they can (and perhaps should) come apart from our views about
identity and survival. As with the question of extraterrestrial life, AP is open to the
many different patterns explored by science fiction tales or utilitarian thought
experiments. The plasticity of our intuitive reactions to thought experiments involv-
ing teletransporters or digital copies of human agents should make us wary of giving
too much moral significance to our current views on these matters. Perhaps future
technology will offer new (literal) ways to live on in the ongoing rush of mankind.
While I don’t go so far as Mark Johnston, who seems to identify personal survival
with a certain kind of concern, I do think that our patterns of concern are of more
fundamental moral importance than our individual identity.
The two main moral arguments for immortality are based on cosmic justice and
meaning. I am sympathetic to the impulse behind those arguments, but we need to
reinterpret them in a less individualist, less self-centred way. Both arguments, I think,
are really about the avoidance of despair. And here I think nihilism is a greater threat
than either cosmic injustice or personal mortality. The real existential worry is not
that my plans will ultimately fade into nothing, or that good may not triumph over
evil. It is rather that no plan is worth pursuing or that there is no such thing as good
or evil. These existentialist worries are real, because nihilism is the natural morality
for any purposeless materialist universe. Conversely, I find that robust moral realism
goes a long way to meeting the psychological need for a just cosmos. Even if the
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
universe is not particularly just, it is nonetheless a place where questions of justice do

make sense. The tendency to assume that the latter requires the former—that there
can only be moral facts in a just cosmos—is another legacy of BT. And atheism
unconsciously shares this legacy. Only AP separates the two questions.
The search for meaning is conceptually separate from the demand for cosmic
justice. But they stem from the same impulse, and similar remarks apply. We need
some connection to genuine values. But the connection need not be individual
instantiation. Once we recognize this, AP offers the same resources as BT. There
are many ways for human beings, either individually or collectively, to relate in
valuable ways to cosmic values. We don’t need to be personally immortal.
Both the Idealist tradition and recent utilitarian thought highlight moral puzzles
about infinity. How can a finite life measure up to infinite values? How can I make
any difference to an infinite cosmos or an infinite population of rational beings?
The main lesson I take from the debate over infinite utility is that infinity does
not make any great difference. Even if my contribution to the whole is very small, or
even infinitely small, it doesn’t follow that my participation must fail to confer
meaning on my life. We should not trust our intuitions about large numbers, let
alone infinite ones.
The need for immortality is yet another site of reasonable disagreement. BT insists
that only a close personal connection to an infinite transcendent God is sufficient for
meaning; while ‘secularists’ argue that this world can provide adequate substitutes.
(This was a central point of contention between the British Idealists and their
utilitarian opponents.) If ‘secular’ implies the rejection of moral realism—and espe-
cially of both moral supernaturalism and non-naturalism—then I agree that secu-
larism is inadequate. But then AP itself is not secular. On the other hand, if a ‘secular’
world view can include cosmic purpose, objective value, and even a non-human-
centred God, then I believe it can confer meaning in the ways that the rest of you are
exploring.
12
Human Well-being
[For Anselm] rational desire will track the objective excellence of natures . . . We
can rank creatures in terms of their likeness to the Word.
[Sandra Visser and Thomas Williams, Anselm, pp. 62, 126]
Theism affirms that the world is such that living morally is not a mere ideal, but
amounts to living in harmony with how things are, at the most ultimate and
profound level.
[John Bishop, Believing by Faith, p. 226]
To the extent that anything is good . . . it is good for us to love it, admire it, and
want to be related to it, whether we do in fact or not.
[Robert Adams, Finite and Infinite Goods, p. 20]
A God who confronted man simply as exalted, distant, and strange, that is, a
divinity without humanity, could only be the God of a dysangelion, of a ‘bad
news’ instead of the ‘good news’. He would be the God of a scornful, judging,
deadly No. Even if he were still able to command the attention of man, he would
be a God whom man would have to avoid, from whom he would have to flee if he
were able to flee, whom he would rather not know, since he would not in the least
be able to satisfy his demands.
[Karl Barth, Evangelical Theology, p. 11]
The previous chapter informally introduced a range of possible ways to incorporate

AP into one’s life. The rest of part III explores the impact of AP on our moral theory.
How could AP matter to us? The same question arises within BT. Why should
God’s purposes and plans matter to me? BT could give crudely instrumental answers
based on fear of Hell or promise of Heaven. But BT philosophers are seldom content
with such mercenary connections. God’s plans and purposes should have intrinsic
moral significance for us.
We could regard God, cosmic purpose, or cosmic values as sources of moral
reasons that compete with our prudential, self-regarding aims. On this view, AP
doesn’t affect the substance of my well-being, but it does reduce its moral signifi-
cance. We return to this possibility in chapter 13. The more ambitious alternative is
that the moral reasons generated by AP’s cosmic purpose not only compete with my
flourishing, they also (partly) constitute it. The present chapter explores this second
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
approach, drawing on a long BT tradition where right relation to God is what makes
my life go well.
I begin, then, with human well-being. Utilitarians obviously need an account of
well-being. (If we seek to maximize human well-being, then we need to know what
well-being is.) And most consequentialists who depart from utilitarianism do so by
adding new values (notably either ecological values or distributive values such as
fairness) not by removing human well-being.1 So they too must consider well-being.
(We explore consequentialist theories that do ignore human well-being briefly in
section 12.3.4, and again in chapter 13.) But concern for well-being is not limited to
utilitarians and consequentialists. Every moral theory, and every human moral agent,
asks what makes for a flourishing human life. If AP affects well-being, its moral
significance is assured.
Well-being poses two distinct challenges to any AP morality. The more obvious is
that AP appears to deny that human well-being has any value. The second challenge
is that, even if AP can admit that human well-being matters, it is difficult to see how a
non-human-centred cosmic purpose could make any difference to our moral lives.
I begin with the first objection. As we’ll see, the best explanation of how AP allows
human value also explains how AP can influence human well-being. In particular,
I shall argue that AP supports an objective list theory of well-being against subjective
alternatives such as hedonism or preference theory.
Due to constraints of space, this chapter only gestures at AP’s impact on well-
being, rather than engaging in detailed examination of particular theories. (Hence the
paucity of specific references to the vast recent literature.) The chapter has two
specific goals. Throughout parts I and II, I have often appealed to the objective list
theory of well-being (OLT) introduced in chapter 1. OLT supports AP. My second
goal is to show that this support is mutual: AP in turn supports OLT. My first goal,
however, is simply to show that AP can make sense of human well-being, and that
it is not irrelevant to it. I begin with the most ambitious possible role for AP, but
less ambitious alternatives are also available, as set out in sections 12.3.2, 12.3.3,
and 12.3.4.
12.1 From Cosmic Purpose to Cosmic Value

Our first challenge is to demonstrate that AP can recognize the value of human well-
being. Prima facie, AP seems to render human lives worthless, because it undermines
all human-sized values. If it is to avoid nihilism and support an objective list theory,
AP must ground objective human values. At this stage of our inquiry, we assume that
1
For overviews of well-being, especially in the utilitarian tradition, see Griffin, Well-Being; Parfit,
Reasons and Persons, App. I; Mulgan, Understanding Utilitarianism, ch. 3; and the works cited in Mulgan,
‘Consequentialism’. Consequentialists who go beyond well-being include Hooker, Ideal Code, Real World
(fairness); and Attfield, Environmental Ethics (environmental values).
HUMAN WELL - BEING 
we are convinced that AP is true. We therefore believe in a non-human-centred

cosmic purpose. Our task is to get from that purpose to human values.
The argument from cosmic purpose to objective values that can feature in an
objective list theory has two main steps. The first is from cosmic purpose to cosmic
values, while the second is from cosmic values to human-sized values. We take each
step in turn.
By ‘cosmic values’, I mean simply objective values that are cosmically significant.
Cosmic values exist independently of anyone’s opinions, preferences etc., and play a
key role in the existence, nature, or purpose of the universe. As we’ll see, the precise
nature of that role varies, depending on whether the correct meta-ethic is moral
supernaturalism or non-naturalism. (In section 12.3.1, I introduce further distinc-
tions between objective values and cosmic ones.)
AP offers several different arguments from cosmic purpose to cosmic values. The
choice between them will largely depend on one’s meta-ethical position and other
philosophical commitments. I will consider five possibilities.
1. Moral supernaturalism delivers the strongest connection: identity. Moral facts

are supernatural. Facts about cosmic purpose are facts about cosmic value.
Cosmic purpose automatically delivers cosmic values.
2. Some arguments that deliver cosmic purpose also directly yield cosmic values.
One possibility, explored in chapters 3 and 4, is that we invoke purpose and
value together to best explain some phenomenon. This route is available to
moral non-naturalists who deny identity. Moral facts are distinct from meta-
physical facts. But the two go together.
In chapter 1, I introduced a distinction between Simple AP and Normative

AP. The former claims only that there is a cosmic purpose, the latter adds that the
cosmic purpose has normative significance for us. Our first two routes suppose that
the case for AP itself delivers Normative AP rather than merely Simple AP, and thus
gives us cosmic values alongside cosmic purpose. If our case for AP over atheism
relies on the arguments canvassed in part I, then this is very likely. (Consider the
number of times throughout this book that I have appealed to cosmic values.)
However, for the sake of completeness, suppose instead that our case for AP leaves
evaluative questions open. We have established cosmic purpose, but remain agnostic
about its connection to value. (Part I has only established Simple AP and not
Normative AP.) We would still have three distinct routes to cosmic values.
3. Some independent moral principle bridges the gap between cosmic purpose and
cosmic value. (Perhaps the reason for a universe’s existence itself constitutes a
source of value within that universe.)
4. Whatever our reason for endorsing AP itself, cosmic values could be the object
of a separate leap of faith. Suppose some metaphysical argument converts us to
AP, but leaves open all questions of value. We must now respond to our
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
evaluative uncertainty. As a Millian moral experiment in living, we might

explore a life built on recognition of cosmic values based on the cosmic
purpose. (Once we are convinced that there is a cosmic purpose, a life founded
on the significance of that purpose does at least seem worthy of exploration.)
5. Our final option is fictionalism. We accept that there are no cosmic values, but
we choose to ‘endorse’ them anyway.
I personally find this five-fold disjunction plausible. I doubt that anyone sympa-
thetic to AP will reject all five routes! If we find AP compelling, we are likely to also
endorse cosmic values.
12.1.1 Direct routes to cosmic value

One possible motivation for introducing cosmic purpose in the first place is to
ground moral facts, and thereby avoid nihilism. Everyone who finds both anti-
realism and moral nihilism unappealing must embrace moral realism. And, as
I noted in chapter 2, moral realists who are suspicious of both moral naturalism
and ‘weird’ sui generis moral facts must embrace moral supernaturalism. This set of
commitments is not uncommon. Many people find all extant non-supernatural
meta-ethical options unsatisfactory. Supernatural moral facts are then our only
way to avoid nihilism. If moral supernaturalism is true, then cosmic purpose simply
is cosmic value. If we accept the arguments of part II, then those cosmic values must
be non-human-centred. AP’s cosmic values are thus assured. AP must be normative.
The arguments of part I provide further support for moral supernaturalism. If we
seek an explanation of the existence and nature of the cosmos, an understanding of
mystical practice, or an account of the properties of a perfect being, then the simplest
answer is to posit an identity between cosmic purpose and cosmic value. Even if purely
meta-ethical considerations fail to convert us to moral supernaturalism, these sup-
plementary arguments may tip the balance. Consider just two examples, from
chapters 4 and 5 respectively. If we need cosmic values to explain the fact that our
universe is friendly-to-life, and we find non-naturalism implausible, then moral
supernaturalism is the only alternative to brute fact. And the most powerful explan-
ation of why successful mystical doxastic practices yield a moral transformation is that
mystical experience is experience of supernatural moral facts. These two examples
each support moral supernaturalism by raising the price of nihilism. Without moral
facts, we cannot explain the existence of the universe or the success of mysticism.
Our interest in part III is normative, not meta-ethical. What matters now is
whether AP delivers cosmic values. Whether those values are supernatural or non-
natural is secondary. In earlier chapters, wherever possible, I have tried to leave open
the choice between moral supernaturalism and non-naturalism. Suppose we prefer
the latter. The connection between cosmic values and cosmic purpose is now more
complicated. But the arguments in part I still support a close connection. Many
metaphysical arguments can be given an intellectualist reading as easily as a
HUMAN WELL - BEING 
voluntarist one. On the voluntarist interpretation, moral facts simply are supernat-
ural facts, and God’s creation brings moral facts into existence. On the intellectualist
interpretation, moral facts are distinct from facts about cosmic purpose. As we saw in
chapters 3 and 4, this interpretation facilitates two distinct modes of explanation. In
both our cosmological and fine-tuning arguments, claims about objective value can
feature either as premises or as supplementary conclusions. If we are already com-
mitted to objective values, then we can conclude that the cosmos was created in
response to those values. If we do not begin with such a commitment, then we can
offer the combination of objective values and a divine response to them as the best
explanation of S (there is something, when there might have been nothing) or FL
(our universe is friendly-to-life). Therefore, even if we don’t start with objective
values, we acquire them in the course of the argument. Either way, we leave the
cosmological or fine-tuning argument with a commitment to objective values.
(Mysticism further illustrates the resources available to the intellectualist. Mystical
experience could still produce a moral transformation if it involves, instead of direct
access to supernatural moral facts, either access to supernatural non-moral facts that
are clearly connected to moral ones (such as God’s intentions), or direct access to
non-natural moral facts themselves.)
AP per se is agnostic regarding the mechanism by which the universe comes into
being. But many arguments for AP do provide information about the source of the
cosmic purpose. And that information can further support cosmic values. In
chapter 6, and throughout part II, we combined AP with perfect being theology,
where the cosmos owes its existence to a creator God whose perfections include
moral ones. Any AP that incorporates such a creator automatically delivers cosmic
values. But even much weaker information about the source of cosmic purpose would
also help. The three most popular alternatives are axiarchism, voluntarist theism, and
intellectualist theism. Voluntarist theism is moral supernaturalism, where God’s will
creates moral facts. Axiarchism and intellectualist theism, by contrast, offer us
independent cosmic values which play an essential explanatory role. Either way, we
get cosmic values.
If we are convinced that AP is true, then it is quite likely that our journey has
already convinced us of AP’s cosmic values. At the very least, this places the burden of
proof on the person who, while she accepts AP, still denies cosmic values.
12.1.2 An independent route to cosmic value

My own view is that the best case for AP will give us cosmic values directly. But suppose
I am wrong. I shall now argue that we can still justify cosmic values.2 We need a bridge
from cosmic purpose to cosmic value. That bridge can be an independent moral
principle, a leap of faith, or a fiction. This section addresses the first option.
2
This section owes much to a discussion following my paper at the Oxford moral philosophy seminar
in 2006.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
For the sake of argument, we must first assume that a convincing case for AP has
failed to deliver cosmic values. One plausible candidate is a fine-tuning argument
where the universe exists because it is X, but we remain officially agnostic regarding
the value or significance of X.3 Suppose we accept such an argument. Can we then
find an independent moral argument to show that X is valuable? Is there a plausible
moral principle to take us from purpose to value?
The general principle ‘whenever X is a purpose, X is valuable’ is open to obvious
counterexamples. Imagine an explorer who uncovers a mysterious structure in a strange
land, and wonders what its purpose might be. She discovers it is a concentration camp
created by an evil fanatic, and its purpose is the systematic extermination of an ethnic
minority. Does this discovery give our explorer any reason at all to value the systematic
extermination of ethnic minorities? Surely not. Next, suppose we discover that the
entire universe was created by a malevolent deity whose goal is to inflict maximum
suffering on innocent sentient beings. Would this discovery give us any reason at all to
value the suffering of the innocent? Presumably not. But then, by parity of reasoning, the
discovery that the universe has purpose P also has no implications for value.
I agree that our explorer should not value the systematic extermination of ethnic
minorities, and that the discovery of a malevolent deity should not lead us to value
suffering. A proponent of AP could reply that, while an evil purpose generates some
(prima facie) values, these are always outweighed by stronger reasons not to value
suffering. (Just as the value of the sadist’s pleasure is always outweighed by the
disvalue of her victim’s pain.) But AP is on much stronger ground if it can avoid
this reply and agree that evil purposes generate no values at all.4
This concession is not fatal, because a general principle linking purpose and value
is far stronger than we need. AP need not claim that every imaginable cosmic
purpose would generate values. If we seek a narrower principle, we have two main
options: rule out evil purposes directly or confine our attention to the purposes of
universes. To simplify, I combine these two restrictions into one principle:
The Value of Purpose Principle (VPP). If U has purpose X then X is valuable if:
(i). the source of U’s purpose is not morally deficient; and

(ii). X is something we already have independent reason to value; and
(iii). U is a universe; and
(iv). U is our universe.
3
In chapter 4, I argued that any value-free fine-tuning argument must fail, because it cannot rebut
Mellor’s objection that it cannot ground objective probabilities. Indeed, this claim was central to our
argument for AP against the atheist multi-verse. Chapter 4 thus supports the conclusion of section 12.1.1,
that the case for AP will deliver cosmic values. However, in the present section, for the sake of complete-
ness, we must set our earlier scepticism aside and suppose that a value-free case for AP is possible.
4
We should note that AP itself could never generate a purpose that is evil (in the everyday sense)
because such a purpose is still a human-centred one. The objection is thus to AP’s hypothetical commit-
ments, not to its actual cosmic purpose.
HUMAN WELL - BEING 
To judge whether clauses (i) and (ii) apply to our universe, we need both
information about the cosmic purpose and judgements about value. We must already
have some confidence that X is valuable. This threatens circularity. But it might
instead be another case of mutual support, where the discovery that X is the purpose
of U reinforces an existing belief that X is valuable. We cannot hope to dispense with
considered moral judgements altogether. After all, without some moral judgements
we could not invoke VPP (or any other moral principle) in the first place.
Clauses (iii) and (iv) raise the possibility that value is universe-dependent.
A universe’s purpose generates values only within that universe. Other possible
universes might contain different values. While some philosophers assume that our
moral judgements extend to all possible universes, it is sufficient for any practical
purpose that they apply correctly to the actual world.5 VPP distinguishes between a
universe and the items within it. This distinction is not ad hoc. We saw several times
in chapters 3 and 4 that a similar distinction is a recurrent theme of cosmological and
fine-tuning arguments. And we must recall, again from chapters 3 and 4, that the
philosopher’s ‘universe’ contains everything that actually exists. Science fiction tales
where ‘our universe’ is created by malicious aliens fall outside VPP, as they merely
involve the creation of new things within a universe. VPP asserts that the creation of
a universe generates new values in a way that creation within a universe does not.
Clauses (iii) and (iv) are borrowed from BT. This idea of universe-dependent value
is a modern descendent of the voluntarist BT tradition. Pure voluntarism posits a
meta-ethical identity between moral facts and the divine will. VPP is a weaker
principle, because it uses a substantive ethical principle to link the two. This route
to cosmic values is thus a halfway house between voluntarism and intellectualism.
VPP is a conditional, not a biconditional. This is all we need. Our four conditions
are jointly sufficient for value. We can remain agnostic whether any of them is
necessary. Perhaps (i) and (ii) would suffice in the absence of (iii) and (iv), so that
non-deficient purposes enhance the value of non-universes. Alternatively, (iii) and
(iv) might suffice in the absence of (i) and (ii). Perhaps even a prime facie evil
purpose would create value within a universe.
How could we justify VPP? We could simply appeal to intuition or considered
moral judgement. Positing some connection between purpose and value does not
seem too far-fetched. I acknowledge that any attempt to derive objective values solely
from a (value-free) fine-tuning argument would be very ambitious. But my present
aim is more modest. I claim only mutual support. AP strengthens the independent
case for objective values, however strong that case might initially be. Our VPP is thus
5
One might worry that the argument in the text contradicts the (putative) supervenience of the moral
on the descriptive. However, as we saw in chapter 2, this would be a mistake. Even if facts about value do
supervene on descriptive facts, they must supervene on the set of all descriptive facts—and that includes
facts about cosmic purpose. Supervenience alone could never teach us that evaluative facts must be constant
across worlds with different purposes, or between purposive and purposeless worlds.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
quite modest. And it does have considerable intuitive appeal. (Witness the continu-
ing appeal of voluntarist accounts of ethics—both theist and atheist.)
AP could also offer a disjunctive ad hominem argument, designed to show that
VPP should especially appeal to those whose meta-ethical views lead them to reject
our more direct arguments. Recall our original commitment to avoid moral nihilism.
The atheist non-nihilist has only three options: moral naturalism, non-naturalism,
and non-cognitivism. Ideally, we want to retain moral realism. So we begin with the
atheist’s two realist options: naturalism and non-naturalism.
Contemporary moral naturalism typically appeals to naturalized teleology, in one
of two ways: either the ‘purposes’ of evolution replace those of God or natural law as
the ground of human values, or the naturalist endorses some subjective theory of
well-being where value resides in the agent’s pleasures or preferences.6 Of course,
moral naturalists typically deny that there is a cosmic creator. So they deny that VPP
applies. But given their teleological inclinations, naturalists should still accept VPP in
theory, and agree that the preferences of a cosmic creator would have some norma-
tive force.
Atheist non-naturalists can eschew teleology. But their methodology commits
them to accepting both intuitively plausible moral principles and non-natural values.
The non-naturalist atheist must agree that intuition sometimes justifies moral prin-
ciples. At the very least, this means that she cannot dismiss VPP out of hand.
If she is prepared to abandon moral realism, the atheist non-nihilist has a third
option. She can replace moral realism with non-cognitivism. But non-cognitivists
should find VPP especially appealing. Of course, actual non-cognitivists are typically
even more hostile to BT than moral realists. (Indeed, their rejection of moral realism
is often driven by their atheism, as we saw in chapter 2.) Most non-cognitivists
vehemently deny that there is a cosmic purpose. Once again, VPP does not apply.
However, on many non-cognitivist stories, human beings themselves somehow
create values. Surely the ability to create value is at least as plausible for a cosmic
creator. If she rejects VPP, and denies God any ability to create new values, the non-
cognitivist will struggle to consistently claim that humans create their own values.7
These ad hominem arguments are not decisive. Opponents of AP’s cosmic values
can still deny VPP. But any substantive moral principle is rejected by some philoso-
pher. Doubts about cosmic value will no doubt persist. The possibility of being
unmoved by values is ever present in moral philosophy. (Some philosophers doubt
that pain is bad, that achievement is good, or that Bob Dylan is better than Miley
Cyrus. This tells you something about philosophers, but probably not much about
6
See, e.g., Boyd, ‘How to be a Moral Realist’; Foot, Natural Goodness.
7
Many atheist meta-ethicists find God and cosmic purpose simply absurd. They will therefore reject
VPP’s antecedent as impossible, and regard any subsequent speculation as idle. However, as I argued in
chapter 2, this should make us wary of atheist opposition to VPP. If you think God is incredible, how
reliable are your speculations about what would follow if God did exist? We return to this question in
section 12.3.3.
HUMAN WELL - BEING 
value—and nothing about music.) I conjecture that many of those who continue to
deny that a cosmic purpose has moral significance either (a) still cannot take
seriously the possibility that there is a cosmic purpose; and/or (b) cannot see how
such a purpose could have meaning for them. At this stage of our inquiry, (a) is
simply a failure of imagination. The main aim of part III is to address (b).
12.1.3 An alternative independent principle

These proposed links between purpose and value will not convince everyone. A more
radical option for AP is to bypass purpose altogether and argue directly from The
universe exists because of X to X is valuable. This would give us the following
principle:
The Value of Explanation Principle (VEP). If the fact that U is X features in the best explanation
for the existence of U, then X is valuable if:
(i). X is not morally deficient; and
(ii). X is something we already have independent reason to value; and
(iii). U is a universe; and
(iv). U is our universe.
AP could replace VPP with VEP. Although AP itself insists on purpose as well as
value, it could defend the two claims independently. An even more radical alternative
is to dispense with purpose altogether and focus entirely on cosmic values. Consider a
modification to the fine-tuning argument for AP. The original argument of chapter 4
concerns the connections between three propositions:
1. The universe exists because of X.
2. X is the purpose of the universe.
3. X is valuable.
As we saw in chapter 4, opponents allege that AP stumbles at both the step from (1)
to (2) and the step from (2) to (3). Why does an explanation for existence yield
purpose? And why should purpose yield values? AP could avoid both stumbles by
arguing directly from (1) to (3). If a universe exists because of X, then X is valuable in
that universe—whether or not X is the ‘purpose’ of that universe. (Obviously, if it
were to dispense with cosmic purpose, AP would then have to change its name.) The
discovery of cosmic purpose matters mainly because it leads us to cosmic values. The
latter without the former would still be a very significant discovery.
12.1.4 A leap of faith

Suppose our case for AP leaves open the question of cosmic value and we reject
independent moral principles such as VPP or VEP that link cosmic purpose to
cosmic value. But also suppose we are not (yet) cosmic moral nihilists. While we
have no argument for cosmic values, we also have no knockdown argument against
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
them. Our universe is thus ambiguous regarding cosmic value. As ever, we must
decide how to respond to this new ambiguity. One option is a leap of faith that
invokes cosmic values alongside our non-human-centred cosmic purpose.
We must separate three distinct leaps. First, of course, AP itself could be an object
of faith. This is a possibility we have explored throughout this book. The available
evidence underdetermines our metaphysical and moral views, and then we endorse
AP because it is the least self-aggrandizing, most challenging, or most under-
explored possibility. Our original dilemma between atheism, BT, and AP calls for a
leap of faith, and AP is the most leap-worthy alternative. A second leap occurs when,
having embraced AP (for whatever reason), we go beyond evidence and argument to
posit cosmic values. The best response to ambiguity about the moral significance of
cosmic purpose is to endorse cosmic values that track that purpose. A third leap adds
human-sized values to accompany either cosmic values or cosmic purpose. Suppose
some metaphysical argument has persuaded us to endorse both AP and its cosmic
values, but we remain uncertain about AP’s human significance. We could then leap
to the claim that cosmic values do make a difference to a human life. (Of course, this
third leap only makes sense if cosmic values can make a human difference. We return
to that question in section 12.3.)
All three leaps are closely related. Our first two leaps often go together, because
cosmic purpose and cosmic value are very hard to separate. If our real interest lies in
value not purpose, then we might leap to cosmic purpose only because it is necessary
to underpin cosmic values. And the third leap often lies behind the other two,
because the impetus behind both cosmic leaps most likely comes from their impact
on human morality.
Our present interest is in the second leap: from cosmic purpose to cosmic values. Is
this leap ever reasonable? In general, any respectable leap of faith must meet two
minimal conditions. The first condition is familiar from Kant’s original practical
arguments. One cannot leap to what is known to be false or incoherent or extremely
improbable. We can only posit cosmic values if we have some intelligible idea of what
those values might be, and we are not convinced that cosmic nihilism is true. This first
condition can be met. Of course, many atheists insist that the very idea of cosmic value
makes no sense. But this is primarily because they also reject all talk of cosmic purpose,
Gods, divine plans, and other supernatural realities. At this stage of the argument, we are
accepting supernatural facts about cosmic purpose. It is then much harder to deny that
cosmic value is intelligible, unless one is a nihilist about all evaluative claims. But global
nihilism is an extreme position. And to deny the intelligibility of non-nihilist alterna-
tives would be even more extreme. Even if we ultimately reject moral supernaturalism
and non-naturalism, both are (at least) philosophically respectable positions. The
debate over objective (cosmic) values is a site of reasonable philosophical disagreement.
(I doubt that anyone who seriously denies this will have got this far through this book!)
Suppose the first condition is met: if there is a cosmic purpose, then cosmic value is
not absurd. This brings us to the second condition. A legitimate leap must make
HUMAN WELL - BEING 
available some great good that would otherwise be impossible. We cannot posit
cosmic values on a whim. They must make some significant difference to a human
life. That impact could come either via human-sized values (section 12.3), or through
moral reasons independent of the agent’s own well-being (chapter 13). Our first leap,
if we make it, must also meet this second condition. We cannot respond to ambiguity
by positing either cosmic purpose or cosmic values unless they affect our lives. Any
cosmic leap of faith thus presupposes the success of our later arguments. The primary
goal of part III is to show that the second condition is met. Cosmic values can make a
very significant difference.
12.1.5 Fictionalism
Fictionalism is similar in practice, though more metaphysically modest. Instead of
leaping to X, we posit X as a fiction. Leaps and fictions can both be defended as
Millian moral experiments in living. In both cases, we immerse ourselves in activities
premised on X, despite lacking sufficient evidence to believe that X is true. The
difference is that, while a leap of faith responds to genuine ambiguity, fictionalism
reacts to the (alleged) truth of nihilism. We resort to fiction when we are confident
that X is not true.
As with leaps of faith, we can separate three possible fictions. First, AP itself might
be a fiction. We know there is no cosmic purpose, but we pretend there is. Second,
cosmic values might be a fiction based on belief in AP. We believe there is a cosmic
purpose. While we are convinced that there are no cosmic values, we pretend that
there are. (Perhaps we are thoroughgoing normative nihilists who regard all evalu-
ative claims as false, but nonetheless we regard cosmic values as highly useful.) Third,
human values might be a fiction based on belief in both cosmic purpose and cosmic
values.
All three fictions are worth exploring. In each case, as with the leaps of faith, the
fiction must make a human difference. (Otherwise, why immerse oneself in it?) The
obvious model for our first fiction is the common fictional approach to conventional
religion. Even a committed atheist might worry that the loss of traditional faith will
undermine morality, and thus introduce the fiction of a divine commander to avoid
practical nihilism or social chaos. Of course, AP’s God cannot issue divine com-
mands. But suppose one believes the following three propositions:
1. There is no cosmic purpose;
2. Only a cosmic purpose could ground genuine objective values;
3. Without a commitment to objective value, human life has no meaning.8
Someone with these three beliefs might want to adopt the fiction that there is a
cosmic purpose. Yet suppose, because of scale or evil or diversity or some other
8
As we saw in chapter 10, meaning is one common justification for moral fictions. Alternatives to play
the role of our third claim include social cohesion, self-sacrifice, or psychological well-being.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
argument from part II, that she finds the very idea of a benevolent God so incredible
that she cannot ever posit BT as a fiction. Or perhaps she finds the moral costs of any
credible BT interpretation of the actual world too high—especially if it is adopted as a
fiction. (BT implies a universal cycle of rebirth where the disadvantaged deserve their
fate. As we saw in chapter 10, this belief will sanction great injustice if it turns out to
be false.) For such a person, an AP fiction might be her only hope.
We return to the third fiction in section 12.3.4. Our primary concern here,
however, is the second fiction, where we make up imaginary cosmic values to
match an actual cosmic purpose. This second fiction may have the same motivation
as the first. We are convinced that AP is true. But despite our belief in AP, we still find
the very idea of objective values incoherent. Yet we also become convinced that we
need cosmic values to ground human morality. Our cosmic values must therefore be
fictional. This is where fictionalism offers new resources that are unavailable to our
earlier routes to cosmic values. In particular, unlike a leap of faith, a fiction need not
be metaphysically credible. Indeed, many philosophers endorse fictionalism about
moral facts precisely because they think such entities are metaphysically implausible.9
Our first fiction presupposes the failure of the central arguments of this book, and
the second presupposes the rejection of our central moral commitments. But both
fictions are live options: alternative responses to the moral inadequacies of atheism
and BT. Furthermore, as we’ll see in more detail in section 12.3.4, these cosmic
fictions have moral resources that go beyond those available to any merely human
moral fiction.
These are our five routes from AP to cosmic values. As I said earlier, anyone
sympathetic to AP is likely to find at least one of them plausible. I conclude that, if we
accept AP, then we should believe in cosmic values. The real question is whether we
can combine our non-human-centred cosmic values with human ones. Before
addressing that question, however, we pause to ask what the cosmic values are.
12.2 What Do We Know about Cosmic Value?

Suppose we agree that, if there is a cosmic purpose, then it will give us cosmic values.
To learn more about the values, we must therefore ask what we know about the
purpose. AP doesn’t provide nearly as much information about cosmic purpose as
BT does. However, many arguments in parts I and II do provide some knowledge of
cosmic purpose. (Chapter 11’s imaginary dialogue was designed to illustrate some of
these informally.)
The central feature that distinguishes AP from BT is that the cosmic purpose is not
human-centred. This single radical departure from BT destabilizes any human
morality based on cosmic purpose. Any AP ethic must be very different from any
9
See, e.g., Joyce, The Evolution of Morality.
HUMAN WELL - BEING 
BT ethic. And a non-human-centred cosmic purpose also distinguishes AP ethics

from any atheist ethic. AP’s defining feature ensures a distinctive moral theory, and
seems to rule out any human ethic at all. We confront this challenge head on in
section 12.3.
However, our knowledge of cosmic values is not entirely negative. In the course of
our inquiry, we have also acquired positive knowledge both about the content of
those values and about potential sources of new knowledge. Here are some earlier
lessons about cosmic values:
1. In chapter 4, I argued that the best fine-tuning argument treats all of the
following as relevant to cosmic purpose: that the universe is governed by
mathematical laws; that it is intelligible; that it is (partly) understood; and
that it is friendly to the emergence of life. These are all objectively significant
features of our cosmos. Cosmic values would thus include mathematical order,
intelligibility, understanding, and life.
2. Even chapter 3’s cosmological arguments provide some content. If the best
explanation for the existence of a physical universe lies in the creative decisions
of a perfect God, then cosmic values will incorporate something analogous to
creativity or freedom.
3. If we combine chapter 5’s defence of mysticism with chapter 6’s ontological
argument, then we can conclude that mystical experience either provides
insight into the motivations of a perfect being who is also a cosmic creator,
or it offers direct knowledge of a realm of transcendent value. The morality
embedded in mystical traditions can thus be read as a (human-tinted) vision of
cosmic values. This suggests that the cosmic values include simplicity, impar-
tiality, transcendence of self, and so on.
4. Our everyday morality itself may also shed light on cosmic values. Chapters 2
and 5 explored the possibility that the best explanation for human moral beliefs
is that they offer some insight into the transcendent realm of objective (cosmic)
values. Given the vast gap between the cosmic purpose and our human facul-
ties, our moral knowledge will inevitably be much less secure than we ordinarily
suppose. But that doesn’t mean it is entirely mistaken.
5. We have also learnt much about the structure of cosmic values. In all fine-
tuning arguments, the startling facts to be explained include the fact that the
universe is governed by a small set of simple, unified laws that produce order
and complex life. Other arguments for AP also give a prominent place to unity
and simplicity. Mystics testify to the unity of the divine, while cosmological and
ontological arguments point to a single cosmic cause or a unified perfect being.
We could simply add unity and simplicity as distinct cosmic values. But it might
be better to say that unity and simplicity are structural features of all cosmic
values. A unified combination of other cosmic values is better than a chaotic
jumble.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
Our knowledge of the cosmic values is very sketchy, and it may forever remain so.
We will always want to know more. But even our current very sketchy knowledge is
sufficient for many purposes. Indeed, if our aim is merely to show that AP has some
moral significance, then it may be sufficient simply to show that there are some (non-
human-centred) cosmic values. Even this bare fact may have a profound impact on
human morality. Or perhaps it is enough to know that, whatever their real nature, we
can best make sense of cosmic values by interpreting them in terms of (say) M (our
universe is governed by precise and elegant mathematical laws), U (our universe is
understood by some of its inhabitants), FL (our universe is friendly-to-life), com-
plexity, simplicity, and creativity.
Furthermore, although our knowledge of cosmic values is very limited, we have the
resources to expand it. We have discovered several routes to cosmic knowledge that
are under-explored in contemporary moral philosophy, such as using cosmological
information to flesh out fine-tuning arguments, exploring possible non-human-
centred interpretations of existing contemplative traditions, and searching for extra-
terrestrial life to discover whether consciousness, sentience, or rationality play any
role in cosmic purpose. AP gives these existing epistemic endeavours a new signifi-
cance and urgency by linking them to cosmic values. For many of the participants in
chapter 11’s imaginary dialogue, this was the most significant impact of AP.
12.3 From Cosmic Value to Human Value

Suppose we agree that cosmic purpose yields cosmic values and that the cosmic
purpose is not human-centred. The obvious conclusion is that human well-being has
no objective value at all. Nothing about our lives matters. As I said in the introduc-
tion, this is AP’s most counter-intuitive implication. Many readers will think it is
sufficient reason to abandon AP altogether. (This is precisely what many readers of
earlier drafts have thought!) Can AP avoid this extreme conclusion? If not, what are
the prospects for human morality? I will explore three possibilities. (1) Human well-
being derives its objective value from a non-human-centred cosmic purpose (section
12.3.1). (2) Human well-being has value, but that value is not derived from the AP
cosmic purpose (sections 12.3.2 and 12.3.3). (3) Human well-being has no value, but
AP’s cosmic purpose can still provide some moral guidance (section 12.3.4).
12.3.1 Cosmic value as the ground of human value
I begin with the most ambitious option—an attempt to ground human values in AP’s
non-human-centred cosmic purpose. AP says that human beings are no part of the
cosmic purpose. AP also says that the cosmic purpose is objectively valuable. This
suggests that, if human well-being has any value, that value must be independent of
AP. In this section, I deny this. Even under AP, human well-being can still participate
in the same objective values as the cosmic purpose. The trick is to separate objective
values from cosmic values.
HUMAN WELL - BEING 
In this section, I present and defend one model designed to enable AP to recognize
the objective value of human well-being. I call this the two-tier model. Section 12.3.1.1
introduces the model and illustrates it using examples of cosmic values drawn from
earlier chapters. Section 12.3.1.2 demonstrates how AP uses the two-tier model both to
support the objective list theory and to transform it by emphasizing cosmic knowledge
and cosmic harmony. Section 12.3.1.3 addresses the hard cases for AP, arguing that the
model could capture the significance of human pleasure, pain, and desire. Finally,
section 12.3.1.4 defends the model. My aim is merely to sketch one possible solution to
AP’s central ethical problem. Other models could be developed, and AP is not wedded
to any details in the two-tier model. But I hope this section does enough to persuade its
critics that AP can recognize the value of human well-being.
... Introducing the two-tier model Imagine a continuous scale measuring the
objective value of objects within a universe.10 Suppose the valuable underlying
property is P. Any object that is P to any degree is objectively valuable to a
corresponding degree. However, there is a cosmic threshold on the scale, such that
values below that threshold have no impact on cosmic purpose.
This new model is easiest to explain within intellectualist theist AP, where an AP
God creates in response to independent objective values. When deciding which
cosmos to create (or whether to create at all), God responds to the goodness of
some property P. However, God is not interested in every possible instance of
P. God is only moved by significant instances of P. God chooses the universe that is
best regarding P for the purposes of divine creation. But this need not be the universe
with maximum P, as judged purely against the continuous scale that neutrally
measures P.11 Instead, the P-best world (for divine creation) is the one with the best
set of significant instances of P. The existence of instances of P whose value falls below
the cosmic threshold has no impact on the divine choice-worthiness of a possible
world. (This simply follows from the definition of that threshold.) These insignificant
P-instances play no role in God’s decision, and are therefore irrelevant to the cosmic
purpose. Nevertheless, they do have positive objective value, because they are
10
The argument in this section builds on earlier work of mine in a very different theoretical context:
Mulgan, ‘Two Parfit Puzzles’; Mulgan, Future People, especially ch. 3; Mulgan, ‘Utilitarianism for a Broken
World’. In the text, I offer a relatively informal presentation of my proposed model. A more precise
formulation would assign values to possible states of the universe, and the values of objects would then be
some function of the values of possible states containing those values.
11
Indeed, as we saw in chapters 7 and 8, Swinburne and others have emphasized in a different context
that there may be no greatest possible amount of P. The two-tier model thus offers AP an alternative reply
to Swinburne’s challenge. A related issue is whether the two-tier model is consistent with divine perfection.
Wouldn’t a perfect God respond to all values? AP’s reply, borrowing from BT theodicy, is that God’s
perfect response to values need not always involve maximization. Perhaps God responds (perfectly)
appropriately to some instances of value by ignoring them. (Of course, as we saw in chapters 7 and 8,
AP must deny that any legitimate divine non-maximization could allow BT to avoid the arguments from
scale and evil. AP could insist, for instance, that a benevolent God could not consistently regard human
horrendous evil as irrelevant!)
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
instances of P and P is the measure of objective value. The values underlying cosmic
purpose are thus more fine-grained than the purpose itself.
Within the two-tier model, AP must deny that human beings represent, constitute,
or contribute to significant instances of P. (Otherwise we have reverted to BT.) This is
consistent with the claim that human beings lack P altogether. But the former does
not imply the latter. We might instead be cosmically insignificant instances of P. Our
lives and collective achievements partake of the value that underlies the cosmic
purpose, but not enough to matter from God’s perspective. The lives, projects,
welfare, and concerns of human beings are a very pale reflection of the complexity
and beauty of the universe—or whatever other feature explains its existence. From
the divine point of view, this degree of resemblance is so inadequate that we are not
worth bothering about. However, from our point of view, our lives and concerns
represent the only values we can meaningfully promote.
A key feature of the two-tier model is that values play different roles in different
practical contexts. Our human choices are not God’s creative choices. Both respond
to the same values, but God’s threshold for creative significance is far higher than our
threshold for human relevance.
We need to separate two uses of the phrase ‘cosmic values’ that I have hitherto run
together. ‘Cosmic values’ could be either the values that God uses in creation or the
underlying objective scale of values. In our new model, the two can come apart. This
allows us to claim both that human beings are not cosmically valuable and that we are
objectively valuable. (Borrowing Sidgwick’s famous phrase, we might distinguish
between the ‘point of view of the universe’ and the perspective of divine creation.
The former provides the objective scale of values, while the latter captures only those
values that contribute to the cosmic purpose.)
To illustrate our two-tier model, we now consider some possible exemplars of
P. Suppose, as I argued in chapter 4, that understanding is a component of the cosmic
purpose. The universe exists (in part) so that its nature and purpose can be understood
by some of its inhabitants. Contra BT, I suggested in chapter 7 that human under-
standing falls below the threshold required for divine significance. From God’s per-
spective, the presence of human understanders adds nothing to the choice-worthiness
of this universe. An otherwise identical human-free cosmos would have been just as
good. However, the valuable property that is instantiated by true understanders is also
possessed (albeit to a lesser degree) by human understanders. Therefore, we can
reasonably conclude that, of two otherwise identical human lives, the one containing
greater (or more valuable) understanding is objectively better. The cosmic value of
understanding can thus provide practical prudential guidance to humans.
Our second example draws on the BT tradition that an appropriate relationship to
God is essential to a flourishing human life. Our model can capture the significance
of human–divine relationships in several distinct ways. First, some aspects of a
human life might directly resemble aspects of God’s divine creative life. Perhaps
human creative efforts are pale imitations of God’s creation, or humans’ relationships
HUMAN WELL - BEING 
with one another are pale imitations of God’s relations to other (cosmically signifi-
cant) beings. Or we might focus instead on human projects whose object is clearly of
cosmic significance. Imagine a person whose life is spent in contemplation of God or
in the search for knowledge of cosmic values. By definition, these projects have
(cosmically) valuable objects. The question is whether any human relationship to
those same objects can itself be valuable. In both cases, something human (our
creations or our contemplation) might register on the same scale of value that
provides God’s reasons for creation. As with understanding, these human connec-
tions to objective values are not significant to God. But God’s own relation to
objective values constitutes the paradigm of a valuable relationship—a paradigm
that human connections can mirror in different ways. On the one hand, a human
connection to cosmic values resembles God’s response to those same values. Here we
have the same object, but very different responses. On the other hand, human
projects whose objects are very unlike God’s objects may resemble divine creation
more closely in other respects. Our impact on objectively valuable objects is very
limited. We can only create very insignificant things. But in so far as we are creating
what is objectively valuable, we are in that respect closer to God. (We will see a
similar pattern repeated in our discussion of consequentialism in chapter 13.)
This example highlights the gap between divine choice and human significance. In
chapter 2, I endorsed Adams’s argument that moral supernaturalism is superior to
moral naturalism precisely because the gap between transcendent goods and mun-
dane objects creates a space for the critical stance. Ethical debate always remains
open, for the moral supernaturalist, because there is always room for reasonable
disagreement whether this or that mundane object better resembles the transcendent
good. AP widens the gap between divine creation and human life even further. Very
different aspects of our lives may (very distantly) resemble different aspects of the
realm of cosmic values in very different ways. Is passive contemplation of God’s
creative act a greater human good than active creation of something valuable but
cosmically insignificant? AP cannot give definitive answers to such questions. But
this inability is not a failure. It merely reflects the transcendence of objective value.
And AP still offers some guidance: we should avoid projects that don’t resemble
God’s relation to cosmic values at all.
Cosmic values like complexity or simplicity or mathematical elegance can also
feature in our two-tier model. Once again, many different kinds of resemblance are
possible. Here are two examples. One human project might have a cosmically
significant object, as when we study the mathematical laws underlying the cosmos;
while another project might itself resemble those cosmic values through its complex-
ity or elegance.12 It will often be very hard to say, in any particular case, which of two
12
Consider the familiar perfectionist thought that the best human life combines a variety of diverse
complex goals into a single unified whole (see, e.g., Hurka, Perfectionism).
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
human endeavours that faintly mirror cosmic purpose is more valuable. But both will
be more objectively valuable than projects with no resemblance to any objective value.
Our new model thus separates two traditional BT ideas: to be valuable is to be
loved by God and to be valuable is to resemble God. Of course, AP denies that God
loves us. But AP can cite resemblances between finite beings and the transcendent
value exemplified by God. Our welfare (and perhaps the fate of our planet) is a very
pale reflection of anything an AP God would care about. But it is still a genuine
reflection. Too pale for us to figure in God’s plan. But not so pale that our moral
concepts are completely unconnected to objective values.
... How AP supports objective list theory In chapter 1, I suggested that AP

mutually supports the objective list theory of human well-being (OLT). We can
now see why. Recall that, under OLT, the components of a good life extend beyond
pleasure and preference satisfaction to include achievement, knowledge, autonomy,
and so on. Items on the list have value over and above any pleasure they produce or
any desire they satisfy. A life devoted to eating mud or counting blades of grass is less
valuable than one devoted to mathematics or theology, even if the former contains
more pleasure or preference-satisfaction, simply because these intellectual endeavours
offer valuable achievements while mud-eating or grass-counting do not. To flourish, a
human being must incorporate objectively valuable activities or items into her life.
AP supports OLT by grounding objective values that are independent of human
beings, and that play a cosmic explanatory role. This makes it much harder to insist,
as opponents of OLT do, that such values only enter an agent’s well-being if she
happens to care about them. On the other hand, if there were no independent
objective values, then OLT would be very hard to motivate. If the object of my
project has no independent value, then how can the project itself have any value over
and above my pleasure or desire? (Once we reduce value entirely to pleasure and
desire, we have abandoned OLT.)
Any interpretation of AP that recognizes human values thus supports objectivism
about human well-being. Our two-tier model also supports pluralism about human
values. For all we know, there may be many distinct cosmic values. (There are
certainly many epistemically plausible candidates!) And even if there is a single
cosmic value, the human-sized analogues of that unified value will be many and
various. We cannot expect to discover a single human value. We will always need a
list of distinct objective human values. We need OLT.
Our two-tier model, together with the conclusions of earlier chapters, also sup-
ports many popular list items. Achievement, knowledge, human relations, creativity,
and freedom are all very plausible human analogues of cosmic values. AP thus
explains why these items enhance human well-being, and why they should appear
on our list.
As well as supporting OLT, AP may also alter it. More detailed exploration is likely
to transform our understanding of the relative value of competing human activities.
HUMAN WELL - BEING 
It is very unlikely that our pre-theoretical (or pre-AP) judgements perfectly track
degrees of resemblance to AP’s cosmic values.
One way that AP may significantly affect OLT is by changing the relative import-
ance of individual and collective human accomplishments. As we saw in chapter 10,
AP is sympathetic to the moral premise of the meaning argument for personal
immortality: it is only by connecting herself to valuable objects that transcend her
brief mortal existence that any human agent can bring meaning into her life. AP
rejects the inference to personal immortality, because the required connection to
larger-scale accomplishments need not be individual instantiation. The individual
human being can join with others to form a collective agent that does instantiate or
produce something of cosmic significance. Possible collectives include human soci-
eties spread across generations, humanity as a whole, or perhaps even the set of all
rational beings in the cosmos.
OLT already recognizes the role in individual well-being of large-scale endeavours.
The individual accomplishments of the scientist or artist or philosopher draw much
of their value from their contribution to larger intergenerational human enterprises.
But AP raises the importance of the collective dimension. At the extreme, participa-
tion in very large-scale collective actions may be the only possible valuable thing in
my life. If AP cannot value lower-level human activities at all, then my membership
of the set of all rational beings is all that matters. Even less extreme AP theories,
which can value human-sized achievements, will still balance the two levels of
achievement very differently than any atheist OLT, because larger-scale achieve-
ments participate more directly in, and more closely resemble, the ultimate source of
all value—the transcendent non-human-centred cosmic purpose.
Another illustration of AP’s potential impact is the place of cosmic purpose itself on
our list. As chapter 11’s imaginary dialogue demonstrated, many actual human
projects are already directed at cosmic purpose. Many people already seek to discover
the purpose of the cosmos, to serve or worship a higher power if there is one, or to share
knowledge of why the universe exists. (There is an obvious connection with our
previous illustration. The search for knowledge of the cosmic purpose, and appropriate
responses to it, is itself a classic example of a long-term collective accomplishment.)
Activities directed at the cosmic purpose can fall under existing generic list items,
such as knowledge or achievement. Discovering the truth about cosmic purpose is a
noteworthy achievement. Another popular list item is the appreciation of beauty. If
the universe exists because it is beautiful or elegant or complex, then a proper
appreciation of its beauty may require that one understand that explanatory con-
nection. (Just as a proper understanding of a work of art may require some know-
ledge of its creator’s intentions.)
The truth of AP might raise the comparative significance of these list items,
especially if one converts to AP from atheism. Knowledge of cosmic purpose is
more valuable, and the achievement of acquiring such knowledge is more impressive,
if there is a cosmic purpose—as opposed to one’s knowledge being purely negative.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
And cosmic beauty, though independently valuable, may be much more significant if
it holds the key to the universe’s very existence. The arguments of parts I and II might
fuel a new interest in both cosmology and theology.
Someone who converts to AP from BT probably already had an interest in cosmic
matters. But her conversion to AP might still give those interests a new urgency, as
she suddenly finds that she must rethink her picture of cosmic purpose—and she
must do so without many familiar BT sources of cosmic knowledge. Wherever one
starts, a conversion to AP thus makes a difference.
While some list theorists will be content to bring cosmic purpose under existing
generic items, a more ambitious route involves special cosmic list items. Virtually all
BT accounts of the flourishing human life explicitly include ‘an appropriate relation
to the divine’ (or something along similar lines) as a separate list item.13 Indeed,
many BT philosophers regard this as the ultimate—or perhaps even the only—list
item. An appropriate relationship to God is the primary goal of human life, and every
other human good is good only because it is an aspect of right relation to God.
We can separate two items here: cosmic knowledge and cosmic harmony. It is good
to know whether one’s universe has a cosmic purpose and to live in harmony with
any purpose that it has. Some lists may include cosmic knowledge without cosmic
harmony. After all, many atheists value cosmic knowledge. The truth of atheism is
not merely esoteric, nor is this just another piece of useful information. The ‘know-
ledge’ that there is no God plays a special role in the lives of many atheists. By
contrast, cosmic harmony can only feature on the lists of these who believe in cosmic
purpose. (Atheists can only endorse a conditional version: ‘Cosmic harmony would
be important if there were a cosmic purpose. But there isn’t.’) Belief in cosmic
purpose thus introduces a new item, as well as raising the significance of cosmic
knowledge. Our two-tier model supports the introduction of both cosmic knowledge
and cosmic harmony, because these are especially plausible candidates for resem-
blance to cosmic values.
Once we have acquired some positive cosmic knowledge, the search for cosmic
harmony can take many different forms. (We saw several of these in chapter 11’s
imaginary dialogue.) The pursuit of cosmic harmony might involve contemplation,
worship, or other attempts to bring cosmic purpose into one’s life. Each BT religion
offers its own list of harmony-based human goods, including specific relationships
with God and other religious goals. Of course, AP cannot borrow these BT lists
wholesale. But it does agree that we should seek both cosmic knowledge and cosmic
harmony, and that acquiring the former will guide and motivate our quest for
the latter.
13
For instance, John Finnis’s influential list includes the following under the title of ‘religion’: ‘the
establishment and maintenance of proper relationships between oneself (and the orders one can create and
maintain) and the divine’ (Finnis, Natural Law and Natural Rights, pp. 89–90).
HUMAN WELL - BEING 
An objective list theory is not merely a list of good things. Plausible OLTs also
include priorities between list items. The simplest priority is lexical, where one list
item is incomparably more important than another.14 (One common example is that
accomplishment is lexically superior to trivial pleasure: no amount of the latter could
compensate for the loss of any of the former.) Cosmic harmony or cosmic knowledge
might outweigh other list items, especially pleasure, desire-satisfaction, or mundane
knowledge. If we find lexicality too strong, we might say instead that some list items
have greater weight without being lexically superior. Perhaps cosmic knowledge is
particularly valuable, even though vast mundane knowledge is better than minuscule
cosmic understanding.
Other OLTs involve more complex relations between list items. Sometimes one
item structures the value of other items. Three common examples of such structuring
items are autonomy, pleasure, and experience. One can treat autonomy as a separate
contributory element among other values or as a precondition for the value of other
items. On the former view, autonomy increases the value of other items that would
still have some value in its absence. But on the latter view, knowledge and achieve-
ment do not contribute to my well-being at all unless I pursue them autonomously.
Similarly, OLT can co-opt the appeal of hedonism or preference theory by adding
experience, preference, or pleasure requirements, where other items add value to
one’s life only if one is aware of them, prefers them, or enjoys them.
For the atheist, cosmic harmony is impossible, and cosmic knowledge is (at most)
merely contributory. It is good to know one’s place in the universe, but one can live a
good life in blissful ignorance. From the atheist’s point of view, the believer’s personal
relations and scientific accomplishments still add value to her life—even if her
motivation for engaging in them is the mistaken belief that she is following some
divine plan. By contrast, for BT cosmic harmony is often the foundation for all other
good things in life. And cosmic harmony is impossible without cosmic knowledge. If
we want to give cosmic harmony a central role, we could say that an appropriate
relationship to God is a precondition of the value of other list items. ‘Living my life in
accordance with God’s plan’ is not an optional extra. Other good things can only
enhance my life if I see them in relation to God’s plan for my life, or recognize them
as flowing from God, or realize that all good things owe their existence to God, or
delight in them as signs of God’s love, or enter into whatever other connection the BT
philosopher deems to be essential.
The two-tier model offers a new interpretation of this familiar BT theme. As
I argued in chapter 2, AP agrees that atheists (and BTs) can lead flourishing and
morally good lives. Cosmic knowledge is not essential to a good life, nor is any cosmic
14
On lexicality in general, see Griffin, Well-Being, pp. 85–9; Mulgan, Future People, pp. 64–9. Lexicality
is especially important in consequentialist intergenerational ethics, as it features in one popular solution to
Parfit’s repugnant conclusion. Indeed, an examination of the role of lexical levels in that context was my
original motivation for the two-tier model. (We return to lexicality in chapter 13.)
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
harmony that requires cosmic knowledge. Ignorance of AP is no barrier to the good

life. However, cosmic harmony—in the form of resemblance to cosmic values—is the
metaphysical ground of all valuable human activities. (This resemblance is what puts
any item on the list in the first place!) So cosmic harmony is essential to human well-
being after all. And cosmic knowledge is essential if one wishes to truly understand
why one’s other projects are worth pursuing. This strongly suggests that cosmic
knowledge is both a distinct list item and an enhancer of the value of all other items.
(Any achievement is more significant if one knows what really makes it valuable.)
The two-tier model thus supports an OLT where both cosmic harmony and
cosmic knowledge play significant roles. By contrast, some other common list
items look less important under AP. In particular, human pleasure and preference
seem to be unrelated to cosmic values. Our model thus provides very strong
comparative support for OLT over its main rivals: hedonism and preference theory.
If human pleasure and preference-satisfaction have no objective value, then OLT is
the only game in town.
However, this defence of OLT comes at a price. I have said several times that AP’s
most counter-intuitive implication is that human suffering has no cosmic signifi-
cance. But it is even more counter-intuitive to suggest that pain (in particular) plays
no role in human well-being. The next section asks whether AP can avoid this
troubling implication.
... Three hard cases: pleasure, pain, and desire Most philosophers agree that
hedonism and preference theory capture something essential to human well-being.
Pleasure, pain, and desire matter. These three items are perhaps the most troubling
for AP. Especially troubling is the central hedonist thought that human suffering is
bad. In casual conversation, I have found that one of most common moral objections
to AP is that it implies that human pain and suffering have no disvalue.
If we endorse an objective list theory, then we agree that human well-being is not
reducible to pleasures and preferences. We saw in section 12.3.1.2 that the recogni-
tion of cosmic values reinforces this anti-subjectivism. The independent value of our
projects must count for something. But can it really be all that matters?
If our case for AP includes an argument from evil, then AP must insist that human
suffering has no cosmic significance. From the perspective of divine creation, a world
where innocent human beings suffer horrendous evil is no worse than one with no
horrendous evil at all. (Otherwise God would have chosen one of the latter worlds.)
But our new model allows us to still claim that human suffering has objective disvalue.
The key is our distinction between objective value and cosmic value. The former is
the underlying scale that determines what matters, while the latter reflects God’s
creative response to that scale. Our argument from evil relates to cosmic value, not
objective value. AP must deny that human suffering matters to God, but it need not
deny that it matters at all. AP can recognize an objective evaluative difference
between human pleasure and human pain.
HUMAN WELL - BEING 
We want a scale of objective value where human pleasure ranks above human
suffering, even though both fall below the threshold of cosmic significance. BT itself
provides several useful resources here. Cosmic values may include God’s own joy in
creation; or the joy that cosmically significant non-human rational agents experience
when they contemplate God or understand cosmic purpose or connect to genuine
values. (As we saw in chapter 7, these superior beings may be integral to AP’s cosmic
story.) God’s mental states, and even the mental states of cosmically significant
creatures, are no doubt very different from our own. Any resemblance between
human pleasure and divine joy will be very distant. For AP, as for many traditions
within BT, all talk of God’s motives or joys is analogical. But it is not too far-fetched
to suppose that when God successfully creates in response to cosmic values, or when
a cosmically significant creature attains cosmically significant understanding, the
result is more similar to our pleasure than to our pain. And that is all we need.
Human pleasure is very different from divine joy. But the former is closer to the latter
than human pain is. A happy human life is objectively better than an unhappy one.
Mysticism provides both a good illustration of this model and a good argument for
it. If, as I argued in chapter 5, the heightened and strange experiences of mystics are our
best guide to cosmic values, then they may also provide our best glimpse of what truly
valuable joy is like. And the fact that mystics typically report their experience as a kind
of joy supports the claim that, despite its strangeness, the experience of cosmic values is
closer to pleasure than pain. (If ‘pleasure’ seems too mundane an expression to apply to
mystical experience, then we might replace it with Sidgwick’s richer notion of ‘desirable
consciousness’.15 Surely mystical experience is more desirable than not?)
Preference is an easier case, though the argument is essentially the same. God’s
attitude to cosmic values is akin to desire. (Or, at least, God’s successful creation is
closer to human desire-satisfaction than to human frustration.) And cosmically
significant beings within the cosmos will also desire cosmic understanding, or
other activities that reflect the divine purpose. While it falls short of cosmic signifi-
cance, a human life of satisfied preferences is therefore objectively better than one
where all desires are thwarted.
The two-tier model can only provide limited support for pleasure and preference.
AP is very unlikely to support either hedonism or preference theory. Pleasure and
preference are not the only features of human life that resemble God’s attitudes to
cosmic values. Nor are they likely to be the closest resemblers. Other human states
or activities will be more valuable. But this should not worry the proponent
of OLT. Indeed, some austere list theorists already relegate pleasure, pain, and
15
Sidgwick’s desirable consciousness is a useful bridge between sophisticated hedonism and OLT, as it
enables the hedonist to take account of the content and structure of pleasure as well as its intensity and
duration. Perhaps AP could be developed using this Sidgwickian hedonism rather than OLT. I leave this
task for another day. (On Sidgwick’s hedonism, see Hurka, British Ethical Theorists from Sidgwick to Ewing,
pp. 195–201; Phillips, Sidgwickian Ethics, especially pp. 53–6.)
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
preference to very subsidiary roles, where they only count when the project in
question is independently valuable and they contribute to its success. Pleasure only
matters when its object is valuable; pain is only bad because it interferes with
autonomous rational achievement. While these are austere views, they are not
crazy.16 And this is precisely the sort of shift in perspective that we should expect
once we begin to take AP seriously. Our model easily accommodates these more
austere interpretations of OLT. If human pleasure and desire aim to mirror God’s
reasons for creation or the valuable attitudes of cosmically significant beings, then
they are most likely to do so when they accompany independently valuable projects.
Some will object that this explanation misses the point. Surely human suffering is
not bad because it is unlike God’s joy? Pain is just bad, period. AP has a choice of two
responses here, depending on its meta-ethical commitments. The intellectualist
agrees that human suffering has independent disvalue. God’s joy is not the measure
of objective value. That role is played by the independent objective scale of values.
God’s joy is merely one appropriate response to those values, and our aversion to
human suffering is another. By contrast, the moral supernaturalist does think that
our pleasure is good (only) because it resembles God’s creative joy. But her defence is
simply that this is no more or less absurd than any other moral supernaturalist claim.
Once we reject non-naturalism, something must make moral facts true. And if
something makes our pleasure good, then why not its resemblance to God’s joy?
... Defending the two-tier model Why should we believe this strange new
picture, rather than dismissing it as a cute technical trick? One answer is that, once
we accept the metaphysical case for AP, this new picture best makes sense of all the
appearances. Any moral realist who is convinced by both the fine-tuning argument
and the argument from evil must hold the following three views:
1. The cosmos is remarkably valuable.
2. The evils of this world are sufficient to show that it was not created by someone
who is concerned for human well-being.
3. Human lives and achievements are valuable in a way that generates genuine
reasons for human beings.
Each of our three competing metaphysical world views stumbles over one of these
claims. Atheism falls at the first hurdle. It cannot explain the existence of such a
remarkable cosmos. (I argued in chapter 2 that atheism also falls at the third hurdle,
because it cannot make sense of genuine reasons.) BT falls at the second hurdle. It
cannot explain the existence of evil. AP has difficulty with the recognition of human
16
Consider the following remark attributed to Sigmund Freud when, at the end of his life, he refused all
drugs except aspirin: ‘I prefer to think in torment than not to be able to think clearly’ (quoted in Griffin,
Well-Being, p. 8).
HUMAN WELL - BEING 
value. The two-tier model reconciles AP with that third claim, and thus opens up the
possibility of a new reflective equilibrium.
The argument from evil is central to AP’s interpretation of the two-tier model. As
one anonymous reader observed, BT could also endorse the model, and simply insist
that humans are above the threshold of cosmic significance. If our case against BT
rested only on chapter 7’s speculations about extraterrestrials, this response would be
reasonable. (Until superior aliens turn up, surely it makes more sense to suppose that
God does care about us?) But the failure of theodicy undermines any BT interpret-
ation of our model. If we were above the threshold of divine concern, then God would
not allow us to suffer horrendous evil. Therefore, if there is a threshold, we must fall
below it.
The two-tier model treats both divine creation and human moral life as practical
deliberative contexts, each involving its own distinctive values. Another argument for
this model is that the decision to treat well-being as a deliberative context is
independently fruitful elsewhere in moral philosophy. In particular, it helps make
sense of our obligations to future people, or so I have argued at length elsewhere.17
Our new model is therefore not an ad hoc innovation designed solely to salvage AP.
I have illustrated the two-tier model using intellectualist theist AP. But it could be
adapted to fit voluntarist AP. God’s will can give rise to a scale of values that is, in
some respects, more fine-grained that God’s own concerns. Our model does not fit so
easily into axiarchism, where there is no intermediary between goodness and exist-
ence. However, the model could be adapted to fit Leslie’s axiarchic plenitude, as set
out in chapter 3. We might postulate a lexical gap between human values and
whatever is genuinely cosmically valuable. The addition of flourishing human lives
can increase the value of a possible world, just as the addition of wretched lives
decreases it. But our fate is never sufficient to tip a world from positive to negative (or
vice versa). If the actual world is good enough to exist, it would still be good enough
to exist with no human inhabitants (or with much more miserable ones). Any good
world that contains humans would still have been good enough without us. The
threshold of cosmic significance is not sensitive to the presence of human beings. But
that doesn’t mean the world isn’t (slightly) better for containing us. Once again, we
would still have a place on the scale of objective value.
Axiarchism thus could borrow the two-tier model or something like it. However,
the model is better suited to intellectualist theist AP, and may therefore support that
interpretation of AP over its rivals.
I have defended my two-tier model as a true meta-ethical story. But it could also be
adopted as a practical postulate, a leap of faith, a useful fiction, or a theoretical
convenience. As moral realists, we have compelling practical reasons to find some
bridge between cosmic purpose and human value. We might invoke my new model,
17
Mulgan, ‘Two Parfit Puzzles’; Mulgan, Future People, especially ch. 3; Mulgan, ‘Utilitarianism for a
Broken World’.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
not as a true account of how things are, but simply as a demonstration that such a
bridge is possible. Other models are no doubt possible. I do not claim that mine is the
only conceivable one. But it does make sense.
The two-tier model is ambitious and speculative. Fortunately, it is not essential to
our case for AP. There are other less controversial ways for AP to impact on human
well-being, as we shall now see.
12.3.2 Cosmic purpose is not the only source of value
Nothing in AP per se implies that all human normativity must be grounded in the
cosmic purpose. We can easily imagine a pluralist AP morality that combines cosmic
values with independent human values. Even if we concede that human suffering has
no connection to cosmic purpose, that does not force us to conclude that suffering is
simply irrelevant for creatures like us.
Our project has been sympathetic to pluralism from the outset. So this new
pluralist approach makes sense. However, it faces two challenges. Pluralism threatens
to render cosmic values morally irrelevant and it may seem inconsistent with our
original case for AP. The charge of irrelevance is easily rebutted. If we combine
cosmic and human values, then the former can easily influence the latter. For
instance, the AP pluralist can still endorse the two-tier model sketched in section
12.3.1, with the caveat that the objective values that underlie God’s creative decisions
are merely one source of human normativity. These cosmically derived objective
values will then still weaken, supplement, or colour our human values. In particular,
if its non-human-centred cosmic purpose is one source of human normative reasons,
then AP reduces the moral pull of my own parochial concerns. This is not surprising.
Any value outside my own life has the same effect. If the egoist comes to recognize
other people’s welfare as a source of reasons for her, then that weakens the practical
pull of her own welfare. Similarly, if I recognize that (say) the pursuit of cosmic
knowledge is independently valuable, then this reduces the net force of my other
pleasures or desires. But this is as it should be. If I discover that I have hitherto
attached comparatively too much weight to one narrow set of reasons, then I should
adjust my priorities.
If AP can ground cosmic values, then it is very unlikely to leave our human values
untouched. (And, as I argue in section 12.3.3, AP’s cosmic purpose can also affect
human well-being even if we reject cosmic values altogether! AP is almost certainly
not irrelevant.)
A second possible worry about this new pluralist AP is whether the acknowledge-
ment of independent values undermines our meta-ethical argument for AP. If we
defend AP by arguing that it alone grounds genuine moral facts, then how can we
now consistently recognize values that are not grounded in cosmic purpose?
Moral pluralism does rule out full-blown moral supernaturalism, where all moral
facts are identical to supernatural facts. But it is consistent with non-naturalism, or
with a hybrid position that combines non-naturalism about human morality with
HUMAN WELL - BEING 
moral supernaturalism about cosmic values. As we have seen several times since
chapter 2, AP is also consistent with these weaker meta-ethical positions. While
moral supernaturalism certainly helps AP, AP is not wedded to it. AP can also have
considerable appeal for non-naturalists. Any moral realist who is not a moral
supernaturalist must already recognize sources of normativity that are not derived
from the cosmic purpose. Moral supernaturalism especially appeals to those who find
independent moral values or principles problematic. Conversely, our pluralist
approach will appeal to those non-naturalists who reject moral supernaturalism
because they are willing to countenance independent moral facts. Moral supernat-
uralism cites the implausibility of free-standing moral facts. But the proponent of
moral pluralism presumably has no general objection to such facts. So it is natural
that our present discussion is more sympathetic to non-naturalism than to moral
supernaturalism. A proponent of AP might, for instance, be convinced by the
cosmological, fine-tuning, and evil arguments; while at the same time finding non-
naturalism a perfectly adequate account of moral facts. This combination is not ad
hoc. It represents a perfectly consistent world view. Someone who finds moral
realism and externalism about reasons unproblematic is likely to find part I’s
explanatory arguments for AP all the more compelling. The non-naturalist could
also easily believe that the arguments of part I teach us about objective values at the
cosmic level, without thinking that AP is necessary to explain all moral facts.
I conclude that nothing essential to AP per se rules out human values that are
independent of cosmic purpose—and neither does any essential component of our
cumulative case for AP. Throughout the book, I have highlighted the most ambitious
AP where all value is intimately linked to cosmic purpose. But AP could be more
modest and yet still be a distinctive and provocative view.
12.3.3 Human value without cosmic values
We have just explored the possibility of combining cosmic values and human values.
A more radical alternative is to reject cosmic values altogether. Suppose the argument
of section 12.1 fails, and AP’s cosmic purpose is unconnected to objective values. In
this section, I will argue that AP can still impact on human well-being and still
support the objective list theory. I begin, however, with the theoretical option that is
least favourable to AP: a subjective account of well-being.
Suppose we identify well-being with preference-satisfaction. AP affects a person’s
well-being only if it connects with her existing motivations. Such connections will
exist for some individuals and not for others. However, as we saw in chapter 11’s
imaginary dialogue, a range of existing human motivations do connect to cosmic
purpose. Like pure mathematics or cosmology, the study of cosmic purpose is an
abstract enquiry that many people pursue for its own sake.
AP could also change a person’s preferences. In particular, as several of
chapter 11’s imaginary characters testify, the discovery of a cosmic purpose might
raise the comparative significance an individual attaches to her ‘cosmic’ desires.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
Knowledge about cosmic purpose is more interesting when it is positive rather than
negative. Conversion from atheism to AP could even ignite new desires to explore or
contemplate the cosmic purpose.
Many atheists disavow any interest in cosmic purpose. But can anyone really be
confident that her desires would remain unchanged if her metaphysical picture of the
world were turned upside down? (Would your total lack of interest in God’s plan for
your life survive a conversion to BT?) As we saw in our discussion of meta-ethics in
chapter 2, it is easy to say that a question doesn’t matter when you are already sure
you know the answer. (If someone thinks God is impossible, then his speculations
about how his preferences would be affected by a conversion to BT or AP must be
taken with a pinch of salt!)
Cosmic purpose can matter on a preference-based account of well-being. Similar
remarks apply to hedonism. This is hardly surprising, as pleasures often track
preferences. Many people enjoy discovering or contemplating the cosmic purpose,
and would take great pleasure in acquiring positive cosmic knowledge or in bringing
their lives into harmony with the cosmic purpose. As virtually every theory of well-
being gives some role to pleasure and preference, AP does matter.
Cosmic purpose thus has some role even within the most subjective accounts of
well-being. In theoretical terms, however, this is a rather superficial victory. Cosmic
purpose contributes to human well-being in the same way as mud-eating or football-
watching. It is something that some people happen to enjoy or want. The philosoph-
ically more interesting question is whether the person who lacks any interest in
cosmic purpose is making a mistake. And here both hedonism and preference theory
answer in the negative. If you are indifferent to cosmic purpose, then it is unrelated to
your well-being. AP is an optional extra.
If AP grounds objective values, and if those values generate objective human
values, then subjectivism is obviously mistaken, as we saw in section 12.3.1. But
subjectivism could still be mistaken even if there are no cosmic values. The trad-
itional case against subjectivism doesn’t cite cosmic values. Many defenders of the
objective list theory are atheists, after all. Belief in independent objective values is
distinct from belief in cosmic purpose, and many moral philosophers defend the
former without even considering the latter. The subjectivist puts all human endeav-
ours in the same category as mud-eating and grass-counting—one doesn’t have to
acknowledge cosmic harmony to object to that! AP adds to the case for OLT, but it is
not essential to it.
Suppose we endorse OLT for unrelated reasons, and then later come to believe in
AP. If the cosmic purpose doesn’t generate cosmic values, can its discovery still have
an impact on my well-being? First, of course, most lists make room for pleasure and
preference. So our earlier discussion applies to OLT as well. If AP alters my
preferences or enjoyments, then it affects my well-being. Second, as we saw in section
12.3.1.2, cosmic purpose falls under many generic list items such as knowledge,
achievement, and appreciation. This role for cosmic purpose does not depend on
HUMAN WELL - BEING 
cosmic values. Third, our list could include cosmic knowledge and cosmic harmony,
even if there are no cosmic values. (The appeal of these list items could be independ-
ent of any argument for AP.) If so, the discovery of AP will certainly influence my
thinking about well-being.
Cosmic purpose thus has some prudential impact, whatever one’s theory of well-
being. The discovery of AP would also support OLT over subjectivism. Again, this is
obvious if AP delivers cosmic values, as we saw in section 12.3.1.2. Without cosmic
values, the case for OLT rests on our considered judgements about human values.
And my own experience is that taking AP seriously does push my intuitions in a
more objective direction. It is easy for the atheist to dismiss cosmic harmony and
cosmic knowledge—or to relegate them to mere instances of generic (and optional)
preferences for achievement or knowledge. But the realization that the atheist’s life
might be founded on a mistaken view of the nature of the cosmos reduces the
plausibility of the subjectivist claim that cosmic knowledge or harmony is entirely
optional. Or so it seems to me.
AP almost certainly does deliver cosmic values. But even it is doesn’t, it is certainly
not irrelevant to human well-being.
12.3.4 Cosmic realist nihilism
We have been exploring ways that someone who believes in AP could also believe in
human values. As with cosmic values, our final option is to admit that there are no
human values, and then posit them as a useful fiction.
One possibility is thoroughgoing fictionalism, where human values are posited
alongside equally imaginary cosmic purposes or values. A more interesting inter-
mediate option is cosmic realist nihilism, introduced in chapter 2. Suppose we are
realists about cosmic values, but nihilists about human morality. Our case for AP has
yielded cosmic values. But we reject human moral facts as comforting self-
aggrandizing illusions. (Perhaps we are divine command theorists who infer
human moral nihilism from the absence of divine commands issued to human
beings.) Yet we also agree that, without a commitment to objective morality,
human life is seriously deficient. We need to construct a moral fiction to govern
and guide our lines. We could, as most fictionalists do, simply make something up—
or arbitrarily privilege some actual subset of present human concerns. If we seek a
less arbitrary foundation, we might look to human connections to cosmic value,
however distant those connections might be. Even if we deny the objective value of
any human resemblance to cosmic values, we might still defend the reality of that
resemblance. It could then be the foundation on which we construct our fictional
human values.
A full exploration of cosmic realist nihilism would require a complete moral
theory, which I will not develop here. (The consequentialisms explored in the next
chapter are one promising place to start.) But any fiction will begin with an account of
human well-being. And the cosmic realist nihilist has resources for fiction-building
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
that are not available to the conventional nihilist. Unlike the latter, she recognizes
the existence of objective cosmic values. The cosmic realist’s human moral fiction
is likely to be an objective list theory built on realist cosmic values.
The cosmic realist nihilist could be even more radical. She could reject all human
fictions, and simply deny that human well-being matters at all. This position is still
distinct from global normative nihilism, because its cosmic values are real. Even if
nothing we do in response to those values contributes anything of value to our lives,
the cosmic realist nihilist might still insist that some responses to cosmic value are
more appropriate than others. This would take us beyond concern for human well-
being altogether, and into the more austere realms explored towards the end of our
final chapter.
This concludes our exploration of human well-being. We have seen that AP can
recognize the value of human well-being in several different ways and that AP
mutually supports the objective list theory that featured in several arguments in
parts I and II. We have also seen that AP pushes OLT in a particular direction by
enhancing the role of cosmic knowledge and cosmic harmony. AP does not imply
human moral nihilism, nor is the discovery of AP irrelevant to our prudential
reasoning.
13
Ananthropocentric Purposivist
Moral Theory
The really hard question is whether the sort of maximal devotion to God
demanded by theism is compatible with love for anything finite at all.
[Robert Adams, Finite and Infinite Goods, p. 185]
When we do find ourselves not knowing what to do, it is probably because we are
weighing our own desires too heavily.
[Sandra Visser and Thomas Williams, Anselm, p. 205]
An evaluative judgement endorses a response to the pull of the good which is

there independently of the evaluation . . . We can identify this pull of the good as
God’s call to us.
[John Hare, God’s Call, p. 49]
What Scotus wants to argue is that God is the ultimate goal of all actions.
[Richard Cross, Duns Scotus, p. 23]
There are two very different sensibilities out of which moral discourse and even
entire moral theories arise. One is the idea that morality attracts. The other is
the idea that morality compels. The former focuses on value, the latter on
obligation . . . I want to investigate a theological virtue ethics in which morality
is driven by the attractiveness of good.
[Linda Zagzebski, Divine Motivation Theory, pp. xi–xii]
This chapter explores possible AP moralities. The chapter is in two parts. Sections
13.1 to 13.6 ask what AP can borrow from atheist ethics. Here, I focus on one moral
tradition: consequentialism. My original plan for part III was to argue that AP
mutually supports one particular consequentialist normative ethic. Part III is now
more eclectic, and its conclusions are more tentative. But I still think there are
important connections between AP and both value-based morality in general and
consequentialism in particular. The first part of this chapter explores those connec-
tions. Consequentialism is also the normative ethic I know best. Therefore, this is the
easiest way for me to illustrate the impact of AP on moral theory. A final reason to
concentrate on atheist consequentialism is that other moral theories are covered in
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
the second part of the chapter. Sections 13.7 to 13.10 ask what AP can borrow from
BT ethics. Here, instead of selecting one ethical theory, I canvass a wide range of
options. The aim of the chapter is to show how AP can develop distinctive and
plausible moral theories, drawing on a variety of different ethical traditions.
As in the last chapter, I have two goals here. In chapter 1, I introduced several
consequentialist commitments that played a central role throughout parts I and
II. Consequentialism supports AP. I must now show that this support is mutual. My
commitments regarding well-being, notably the objective list theory, were addressed in
chapter 12. In this chapter, I demonstrate how AP supports consequentialism.
However, although my cumulative case for AP rests on my consequentialist
commitments, AP itself is not wedded to any particular moral theory. My second
aim in this chapter is to illustrate how AP could borrow from a range of other moral
theories.
These two goals explain why the chapter’s treatment of different moral theories is
unbalanced, with atheist consequentialism dominating, BT consequentialism occu-
pying a more prominent place than one might expect, and non-consequentialism
playing a very secondary role. A full treatment of all moral theories would, of course,
take many books. I focus here on my own utilitarian tradition and on some themes
from BT ethics whose interplay with AP strikes me as particularly surprising or
noteworthy.
13.1 AP and Atheist Consequentialism

Most contemporary consequentialists are atheists. Utilitarianism, in particular, is
often linked to meta-ethical positions associated with atheism, notably moral natur-
alism or anti-realism. Indeed, many philosophers regard consequentialism as the
natural morality for atheists. Once we dispense with God, the promotion of value is
morality’s only rational basis.
Despite the affinities between atheism and consequentialism, I believe that atheists
cannot adequately justify consequentialism, and that AP can do much better.
Sections 13.1 to 13.6 argue that AP supports an austere value-based consequentialist
moral theory that is distinct from conventional (atheist) consequentialism.1 I will
concentrate on the distinctive substantive features of AP consequentialism. I begin,
however, with AP’s contribution to moral method. As we have seen in earlier
chapters, AP is suspicious of any ethical method that places too much weight on
our intuitions about particular cases, especially when these might reflect self-
aggrandizing caprice. AP therefore favours ethical methods based either on intuitions
about principles or on purely theoretical considerations.
1
My treatment of consequentialism in this chapter draws freely on Mulgan, The Demands of Conse-
quentialism; Mulgan, Future People; and Mulgan, Understanding Utilitarianism. As with well-being in
chapter 12, my goal here is to sketch the resources available to AP, rather than engage too closely with the
vast consequentialist literature. For further references to the main debates, see Mulgan, ‘Consequentialism’.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
One popular ethical methodology is reflective equilibrium, where the best moral
theory is the one that best matches our considered moral judgements. The classic
example is Rawls’s defence of his liberal egalitarianism.2 Within consequentialism,
one prominent recent example is Brad Hooker’s reflective equilibrium defence of rule
consequentialism. Hooker argues that ‘the best argument for rule consequentialism is
that it does a better job than its rivals of matching and tying together our moral
convictions’.3 (We return to Hooker’s rule consequentialism in section 13.5.)
AP morality is very unlikely to fit our existing considered moral judgements. As we
saw in chapter 12, even when it does accommodate human well-being, AP pushes
human values in a new and austere direction. And, as we’ll see in section 13.2, AP’s
total package of values is very different from those associated with any traditional
consequentialism. AP cannot plausibly claim equilibrium with our current moral
beliefs.
Any AP morality will be very counter-intuitive. If our test of moral truth is
coherence with our current moral intuitions, then AP cannot ground any plausible
morality. AP has three broad options here. The first is to reduce the gap between AP
morality and our intuitions. We saw one example in chapter 12. Opponents allege
that AP is committed to denying the value of human well-being altogether.
Chapter 12’s two-tier model allows AP to defend the objective value of human
well-being, while still denying that our lives have cosmic value. AP can thus respect
our moral intuitions about objective value. The present chapter offers several argu-
ments along similar lines, where AP morality comes closer to common sense than we
might expect.
However, this conciliatory strategy has its limits. AP’s main values (namely, its
cosmic ones) are neither those of common sense nor of the utilitarian tradition. Any
moral theory grounded in those values must diverge from our current ethical
thinking to some degree. (If it didn’t, then AP would be of little interest to moral
philosophers!)
A second strategy is to widen our notion of reflective equilibrium, so that we
now seek coherence between our moral beliefs and our other beliefs.4 If we bring in
our considered metaphysical views, then perhaps AP consequentialism is the new
reflective equilibrium we would reach once we adjust our moral ideas in light of a
non-human-centred cosmic purpose. This book is the first step towards that new
equilibrium: an attempt to imagine what moral intuitions we might have, once we
have been converted to AP.5
2
Rawls, A Theory of Justice.
3
Hooker, Ideal Code, Real World, p. 101. See also Hooker, ‘Rule-Consequentialism, Incoherence,
Fairness’, p. 29; Miller, ‘Hooker’s Use and Abuse of Reflective Equilibrium’.
4
On the varieties of reflective equilibrium, see Daniels, Justice and Justification.
5
In a similar vein, my other recent work imagines how our ethical intuitions might be transformed in a
variety of different possible futures (Mulgan, Ethics for a Broken World; Mulgan, ‘Ethics for Possible
Futures’).
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
The problem with this strategy, of course, is that belief in AP is not widespread (to
say the least). Any new wide reflective equilibrium must therefore be very speculative.
Who can say how our moral intuitions might adapt in light of AP? We cannot easily
construct a reflective equilibrium based on our future moral judgements. And in the
meantime, AP morality must start somewhere. We need to find some current moral
judgements on which we can reasonably rely.
Our considered moral judgements operate at various levels. In addition to intu-
itions about specific cases (either actual or hypothetical), we also have commitments
regarding moral ideals (fairness, impartiality, justice), general moral rules (torture is
wrong, generosity is good), or abstract features of morality (universalizability, super-
venience), and so on. For both Rawls and Hooker reflective equilibrium includes
these more abstract intuitions. This is why reflective equilibrium can be revisionary.
When our intuitions about particular cases clash with our moral ideals, it is often the
former that give way. Consequentialists typically have more faith in general prin-
ciples than particular intuitions. Consequentialist reflective equilibrium can thus be
very revisionary. Moral revision is especially likely under AP. Once we recognize a
non-human-centred cosmic purpose, our familiar moral ideals may push us in
surprising new directions.
Several arguments in part I concluded that our moral knowledge is not entirely
unreliable. In particular, chapter 5 argued that mystical experience provides some
insight into cosmic values. However, that defence of human moral knowledge has
two limitations: our knowledge is limited to very general moral principles or values
and it is distorted by our human-centred perspective. Our earlier discussion thus
supports an ethical method that relies on abstract principles or values rather than
particular cases.
A more radical alternative abandons our moral intuitions altogether, along with
reflective equilibrium. This option is also available to AP. After all, reflective equi-
librium is not uncontroversial. Critics object that coherence with current beliefs is an
inadequate test of moral truth. Moral philosophy should be open to the possibility of
radical error and the need for wholesale revision. Perhaps our current moral practices
are deeply flawed. Instead of privileging our moral judgements (whether particular or
general), we should begin with the most general facts about cosmic values, and then
follow the ethical argument wherever it leads.
This radical theoretical strand runs through the utilitarian tradition, alongside the
more conciliatory intuitive method.6 We have already endorsed this radical Ben-
thamite tradition several times.7 AP supports the shift from conservative intuition to
radical theory. If we are mistaken about the very nature of value, and about our
6
Prominent recent exponents of consequentialist extremism include Shelly Kagan, Peter Singer, and
Peter Unger. See also section 13.4.3.
7
I also defend it elsewhere in relation to ethical method (Mulgan, ‘Theory and Intuition in a Broken
World’).
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
significance in the cosmos, then our current ethical thinking is very likely unreliable.
Coherence with human intuitions is a strange test for cosmic morality.
The shift away from reflective equilibrium, and especially the rejection of par-
ticular intuitions, is not substantively neutral. It favours consequentialism over
non-consequentialism, and it favours particular options within the consequentialist
tradition. Objections to consequentialism often cite its extremely counter-intuitive
results. Once we disregard particular intuitions, those objections lose their force.
Many normative ethicists agree that reflective equilibrium is unsatisfactory, but
they retain it simply because they see no alternative. Unless we rely on our human
intuitions, how can we do moral theory at all? What favours one theory over
countless possible others? It is here, I believe, that AP comes into its own, especially
compared to atheist consequentialism.
It is impossible to do ethics without relying on some moral judgements. If
‘intuition’ covers all moral claims, then the search for a non-intuitive ethic is doomed
to fail. The real choice is between reliance on judgements about particular cases (X is
wrong; Y is good for humans) and reliance instead on theoretical principles and
abstract value claims (objective values provide moral agents with reasons to act;
possible world A is better than possible world B). When I speak of the rejection of
reflective equilibrium, I refer not to the abandonment of all moral judgements but to
the shift from particular intuitions to abstract principles. AP offers a justification for
this shift, because our particular moral intuitions are infected by our self-
aggrandizing concerns in a way that our abstract principles and values are not.
A theoretical defence of consequentialism must explain why we should promote
value. The contemporary literature is surprisingly short on arguments for conse-
quentialism.8 Most proponents treat consequentialism as prima facie plausible, and
then either argue against departures from pure consequentialism, or rebut objections
to it. I believe this approach blinds consequentialists to the fact that it is surprisingly
difficult to defend consequentialism within an atheist world view.
Atheist consequentialism’s difficulty stems from its limited and inadequate meta-
ethical options. Atheist moral nihilists are correct. To ground a revisionist morality,
we need objective values with considerable normative force. And neither moral
naturalism nor non-naturalism nor anti-realism can provide that normative force.
By contrast, I shall argue in this chapter that, by developing a world view where
objective values are linked to the very existence of the cosmos, AP can succeed where
atheist consequentialism fails. For AP, consequentialism reflects the divine (or
cosmically creative) point of view. This gives it a solidity the atheist cannot hope
to match.
By sidelining ethical intuitions and grounding objective values, AP supports
consequentialism. AP also pushes normative ethics in a variety of directions that
8
Mulgan, ‘Consequentialism’ lists the most prominent exceptions.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
favour consequentialism over its rivals. These substantive contributions are the main
theme of this chapter. Chapter 12 argued that AP mutually supports the objective list
theory, but also renders it more impersonal and austere. AP has the same impact on
consequentialism.
13.2 AP and Value

While AP is consistent with many different moral theories, all AP moralities will
share some common features. In particular, AP supports a close connection between
morality and value. We addressed the content of AP’s values in chapter 12. We now
explore their role in moral theory.
To begin with, AP supports the general idea that there are objective values. As we
saw in chapter 12, we may never be sure which objective values AP actually does
ground. (We return to that controversy at the end of this section.) But chapter 12’s
central lesson was that, if we accept AP, then we are very likely to endorse some
positive evaluative claims that are independent of human interests, preferences, or
concerns. AP thus supports any moral theory that gives a significant role to objective
values. AP correspondingly undermines the central justification for subjective mor-
alities such as existentialism, contractualism, or preference utilitarianism, as these are
all motivated by the alleged need to purge ethics of mythical ‘objective’ values.
This obvious implication of AP is very significant. The reluctance to countenance
any values beyond individual inclination, desire, or preference is a key barrier to
moral realism, and thus to any moral theory that posits genuine external reasons. By
removing that barrier, AP paves the way for revisionary morality.
A moral theory can admit objective values without making them central. But AP also
supports the more specific idea that morality is primarily concerned with appropriate
responses to value. (As we saw in part I, a common theme of many arguments for AP is
that the very existence of the cosmos itself results from a divine response to objective
values.) AP supports value-based morality, alongside objective values.
Contemporary value-based moral theories include forms of virtue ethics and
deontology, as well as consequentialism. In addition to its general support for
value-based morality, AP also favours two specifically consequentialist claims
about value. First, AP supports evaluative claims about possible states of the universe.
(This is especially true if our case for AP includes the cosmological and fine-tuning
arguments of chapters 3 and 4 or chapter 7’s argument from scale.) Such evaluative
claims are distinctive of consequentialism. All non-nihilist moral philosophers speak
of ‘value’ in some sense or other. But many opponents of consequentialism reject its
evaluations of states of affairs.9 Some argue that, while things can be better for this or
that person, or from this or that perspective, nothing can be simply better per se.
9
See, e.g., Foot, ‘Morality as a System of Hypothetical Imperatives’.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
Sidgwick’s ‘point of view of the universe’ is a mirage. Only individuals have points of
view, and each has her own. Against this scepticism about comparative evaluations,
AP’s cosmic values provide a real foundation from which competing states of affairs
can be compared. The universe does have a point of view.
Consequentialism doesn’t just posit these global evaluations. It also gives them
ethical priority. Morality consists in appropriate responses to the comparative value
of states of affairs. Right action makes the world better. AP explains why global values
take precedence. Ontologically, this is the primary locus of value. It is the value of the
universe as a whole that explains its existence. If morality mirrors the divine response
to values, then these are the facts about value that matter most.
By defending both the intelligibility and the priority of evaluations of states of
affairs, AP dissolves one key barrier to consequentialist theories based on responses
to such values. This is a classic case of mutual support. Philosophers who are sceptical
about such evaluations are also more likely to baulk at our explanatory arguments for
AP in the first place. But, as I have argued throughout this book, mutual support can
be a virtuous circle, not a vicious one.
Consequentialism’s final defining claim is that it favours one particular response to
value. Morality tells us to promote value. AP supports this claim as well. The central
thing we know about cosmic value is that it is to be promoted. Otherwise, cosmic
values could not feature in an explanation of the existence or nature of the cosmos.
For the axiarchist, cosmic value is itself directly efficacious. Goodness promotes itself!
For the intellectualist theist, God’s response to cosmic value is to promote it. God
creates the cosmos because it is good. For the voluntarist theist, things are more
complex. Cosmic value flows from God’s will, rather than being prior to it. But here
too promotion is especially salient. If God’s creation brings value into existence, then
the response to value that best resembles this divine original will be the creation of
additional value—i.e. the promotion of value.
AP offers no decisive proof of consequentialism. God may also respond to cosmic
values in many other ways. And there is a significant gap between God’s responses and
any human morality. Moral theories that deny objective value altogether, or reject
comparative global evaluations, or ignore responses to cosmic values, all sit uneasily
with AP. But AP could be combined with a non-consequentialist response to objective
values, and especially to comparative evaluations. (And, as we’ll see in section 13.5,
some non-consequentialist alternatives may have a natural affinity with AP’s actual
cosmic values.) However, AP does imply that promotion is at least one central divine
response, and therefore that consequentialism is a natural place to start.
AP’s emphasis on values thus supports many familiar consequentialist themes. AP
consequentialism is distinctive because AP’s values are distinctive. Most strikingly,
AP reverses the standard (atheist) consequentialist story about value. Due to its
utilitarian heritage, consequentialism begins with aggregate human well-being. Con-
sequentialist accounts of value differ along three dimensions: (1) What do they say
about individual human well-being? (2) How do they aggregate human well-being?
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
(3) Do they add other values? (Such as animal welfare, environmental value, aesthetic
value, distributive value, etc.) But every atheist consequentialist starts with the value
of individual human lives. By contrast, AP starts with cosmic values that ignore
human well-being altogether. AP may then add human-sized values, in the various
ways explored in chapter 12. But then again, it may not. AP morality could dispense
with human-sized value altogether. Furthermore, even if it does recognize the value
of human well-being, AP’s value theory is still distinctive, both because it offers a new
substantive account of human well-being where connections to cosmic values play a
much greater role, and also because it gives priority to cosmic values. AP conse-
quentialism cannot concern itself only (or even mainly) with promoting human
welfare as ordinarily understood.
13.3 AP’s Austere Morality

In addition to its emphasis on values, contemporary consequentialism has several
other distinctive features. Relative to its competitors, consequentialism is austere,
impartial, and future-oriented. Any morality based on AP values will share these
features. As well as supporting consequentialism, AP also pushes it further in these
familiar directions.
Consequentialism is notoriously austere. It requires agents to set aside their com-
paratively trivial personal whims, satisfactions, and projects, and devote themselves
instead to the promotion of aggregate human well-being. AP exacerbates this austerity,
because its values are even further removed from the individual’s immediate concerns.
Consequentialism also demands impartiality. The moral agent can give her own
interests, perspectives, relationships, or personal values no greater weight than anyone
else’s. From the point of view of the universe, each is to count for one, and none for more
than one. AP supports an even deeper impartiality, because its values suggest that
human beings are not only no more important than one another, but also no more
significant than anything else in the cosmos that (very dimly) resembles cosmic values.10
The consequentialist commitment to impartiality extends to temporal impartiality.
Future people are of central importance. Unless something goes drastically wrong in
the next few centuries, most future well-being will be enjoyed by people who do not
yet exist. A theme of the recent consequentialist literature is that, from this tempor-
ally impartial perspective, conventional morality gives far too little weight to the
interests of distant future people.11 This is one place where all consequentialists must
depart from conventional morality, especially if they insist on temporal impartiality.
10
I explore the connections between AP and impartiality in Mulgan, ‘How Should Impartialists Think
about God?’
11
The classic consequentialist attack on temporal partiality is Cowen and Parfit, ‘Against the Social
Discount Rate’. For discussion and references, see Mulgan, ‘Utilitarianism and our Obligations to Future
Generations’.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
As we saw in chapter 10, AP reinforces this focus on the future, because human
history can only hope to become cosmically significant over the long term. I suggested
in chapter 12 that the need to connect to intergenerational projects supports an
objective list theory with a strong emphasis on small contributions to very large
collective accomplishments. A theme of my own recent work is that the need to place
distant future people centre stage has many implications for moral theory.12 In
particular, we need an objective list theory to make sense of our intergenerational
obligations, and its ability to accommodate obligations to future people gives conse-
quentialism a potentially decisive advantage over its non-consequentialist rivals.
Once again, AP thus supports both objective list and consequentialism.
Austerity, impartiality, and a commitment to the future are general features shared
by all consequentialists. AP consequentialism has one final feature that is more
distinctive. Many consequentialist moral theories are monist: prescribing a single
response to a single value. (Everyone everywhere should always maximize aggregate
pleasure, to take one familiar example.) By contrast, AP favours a pluralist moral
theory. Our lack of reliable knowledge about the cosmic purpose, our recognition of
the gap between cosmic purpose and human values, and our uncertainty surrounding
the steps between the two, all render AP morality a very hesitant and uncertain
business.
A theme of this book has been that AP recognizes many dimensions of value
pluralism. It is worth recalling these:13
a. Pluralism about cosmic value. Our current best guess, based on the arguments
in parts I and II, suggests that there are several distinct ultimate cosmic values.
Chapter 4’s fine-tuning argument alone yields intelligibility, understanding,
friendliness-to-life, mathematical elegance, complexity, and simplicity.14
b. Pluralism about human resemblance. Even if the cosmic value itself is unitary,
human projects can resemble that value in many different ways across many
different dimensions. As we saw in chapter 12, there is no single best human
way to resemble cosmic value.
c. Pluralism about interpretation. Our ignorance of cosmic purpose adds a further
plurality of reasonable interpretations of both cosmic values and human resem-
blance. Even if there were a single best human approximation to a unitary
cosmic value, we would have no idea what it looked like.
d. Pluralism about human response. Different human responses to cosmic values
can resemble the best divine response to the same values in many different
12
Mulgan, Ethics for a Broken World; Mulgan, ‘Ethics for Possible Futures’.
13
Another kind of pluralism arises within rule consequentialism. This is pluralism about scope, where
different scopes (individual, group, species, all rational beings) may reasonably apply to different moral
projects. (We return to this pluralism in section 13.5.)
14
Perhaps these different values do all ultimately resolve into a single unitary value. If they do, however,
that would exacerbate our second and third kinds of pluralism.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
ways. Suppose we must choose between contemplating the real cosmic purpose
and promoting some pale imitation. Who can say which is closer to God’s
creative response to value?
e. Meta-ethical pluralism. At every level, there is reasonable disagreement whether
claims about cosmic value, human value, or human response are objective
moral facts, practical postulates, necessary fictions, or optional experiments in
living.
AP thus pushes consequentialism in a direction that some consequentialists find
uncongenial—just as its advocacy of objective list theory will discomfort hedonists
and preference utilitarians.
AP’s support for consequentialism is a double-edged sword. Consequentialism’s
distinctive features lead to familiar objections. By emphasizing the former, AP
exacerbates the latter. I will now illustrate these objections using the simplest form
of consequentialism. This is maximizing act consequentialism: the right act for any
moral agent on any occasion is the one that maximizes value. On the simplest AP
account of creation, this is how God decides. God chooses the best possible cosmos.
The act consequentialist says to each of us: ‘Go, and do likewise!’ I shall call the
resulting theory: ‘AP act consequentialism’ (APAC).
Objections to APAC fall into two main categories. One allegation is that APAC is
irrelevant. It is simply pointless to tell me to imitate God! This leads to two distinct
objections:
1. Epistemic inadequacy. I have no way of knowing in advance which action would
produce the best result. It is impossible, even in principle, to calculate the
impact of one individual’s actions on aggregate well-being. AP increases this
epistemic difficulty. Once we introduce cosmic values, or even human values
based on cosmic purpose, our ability to predict valuable consequences dimin-
ishes still further.
2. Impotence. My individual action makes no difference at all. This objection is
especially pressing for AP. How can one human being promote the cosmic
purpose? How can I make the cosmos any better than it would otherwise have
been? How can my actions make any cosmically significant difference?
Other objections complain that even if APAC does offer me guidance, it offers the
wrong guidance. Can a life devoted to maximizing non-human-centred cosmic values
be fulfilling, coherent, or morally plausible? Act consequentialism is very counter-
intuitive in three distinct ways.
3. Demandingness. Consequentialism requires the agent to continually sacrifice
herself whenever she can improve the lives of others.15 This is unreasonably
15
I address demandingness at length in Mulgan, The Demands of Consequentialism. For further
references, see Mulgan, ‘Consequentialism’.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
demanding. Even within the human realm, AP exacerbates these demands,

because its account of human well-being is more austere. (Even if my goal is
only to promote my own well-being, I may find myself having to sacrifice my
own pleasure or preference in pursuit of cosmically significant collective pro-
jects.) APAC also introduces a new kind of demand. If my goal is to behave
morally, then I may have to sacrifice, not only my own well-being, but all
human interests to serve some ‘higher’ cosmic good.
4. Alienation. Opponents reject both what consequentialism asks me to do and how
it asks me to see the world. The former is the demandingness objection. The latter
leads to Bernard Williams’s infamous ‘integrity’ objection.16 If I become a
consequentialist, then I will only ever see my own interests from the impartial
point of view of the universe. But this is neither possible nor healthy nor desirable.
AP threatens a new alienation. It forces me to see the world through non-human-
centred eyes. Even if human well-being has some value, it is very insignificant in
the cosmic scheme. (And if the main arguments of chapter 12 fail, then perhaps
I should see no value in human affairs at all!) If I can promote cosmic value, then
this non-human perspective will dominate my moral viewpoint. It is not clear
that I could see the world this way, and very doubtful that I should.
5. Immorality. Act consequentialism notoriously ignores all the permissions and
prohibitions of common-sense morality. In her pursuit of aggregate value, the
consequentialist agent must be prepared to lie, steal, murder, torture, and so on.
AP demands such immorality in the service of non-human-centred cosmic
values, even when no human good is served.
The main challenge for AP morality is to rebut these objections. They arise most
clearly for APAC, but any AP morality will face similar problems.
13.4 AP Act Consequentialism

Our five objections to consequentialism are all familiar, and APAC can draw on
familiar consequentialist replies. One response is to abandon act consequentialism,
while still retaining a broadly consequentialist moral theory. We explore that option in
sections 13.5 and 13.6. As we’ll see, AP’s most distinctive contributions to consequen-
tialist moral theory may come after we have left act consequentialism behind. We
begin, however, with the options available to the defender of APAC.
13.4.1 The epistemic inadequacy of APAC

The epistemic objection is well trodden.17 Our concern is what AP adds to the debate.
Most consequentialists concede that I cannot hope to know which act would actually
16
Williams, ‘A Critique of Utilitarianism’.
17
See, e.g., Lenman, ‘Consequentialism and Cluelessness’.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
maximize value. But no plausible moral theory could demand such knowledge. We
should shift from actual consequences to predictable expected ones. I cannot choose
the action that will maximize value. I can only be guided by my best guess. And if
I focus my attention close to home, then I can make confident predictions. I have no
idea how to maximize the well-being of all distant future people, but I can guess what
will maximize the welfare of my present community. If I relieve my neighbour’s
suffering, the distant consequences could be disastrous. But I probably will promote
expected value. (As we’ll soon see, this response to the epistemic objection also serves
to bring consequentialism closer to common-sense morality, by recommending that
agents focus on their nearest and dearest.)
This reply is controversial. But many consequentialists find it compelling. And my
present claim is only that, if it is does succeed, then AP can borrow this act
consequentialist reply. I do not know what will maximize value. But I can form
some subjective expectations, especially if I confine my attention to values that are
local in two senses: their objects are close to my own life and the values themselves
are human-sized rather than fully cosmic. I don’t know much about cosmic values or
the lives of distant rational beings, but I can be reasonably confident how to make my
own life go better. If my life has some objective value, then act consequentialism
offers some moral guidance. My distance from the cosmic values now becomes an
advantage. Because I have no idea how to advance those values, I can reasonably set
them aside.
One dimension of uncertainty is uncertainty about value. Even if we can predict
all possible consequences, we may not know which outcome would be best, because
we are not sure how to evaluate things. Uncertainty about value is familiar,
especially in intergenerational contexts where we struggle to judge what is best
for our descendants. AP exacerbates this uncertainty. Even if human affairs can
participate in objective values related to cosmic purpose, they do so very indirectly
and in many different ways. And our knowledge of cosmic purpose itself is very
sketchy.
As with uncertainty in general, the theoretical solution is straightforward, even if it
can be very difficult in practice. All we can do is rely on our current best guesses
about value. Under AP, those guesses are very speculative, as we saw in chapter 12.
But that is not necessarily an objection to APAC. Indeed, AP now offers a new
response to the epistemic objection. Although it increases our uncertainty, AP also
renders that uncertainty less objectionable. We find consequentialism epistemically
inadequate because we expect morality to offer clear simple guidance. This is a
reasonable expectation if morality is either a human construction or the creation of
a benevolent God. But if morality responds to cosmic values that lie far beyond our
ken, then we have no reason to expect reliable moral knowledge. Perhaps a barely
intelligible opaque ethical ideal is the best we can hope for.
These brief remarks will not convince everyone. Those who remain unconvinced
may wish to replace APAC with another, less epistemically demanding value-based
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
alternative, such as those outlined in sections 13.5 and 13.6, or with one of the non-
consequentialist alternatives explored in the second half of the chapter.
13.4.2 Is APAC impotent?

We turn now to the charge of impotence. Even if I know all the relevant facts,
including facts about value, APAC offers no guidance because my action makes no
difference at all. We encountered similar objections in chapter 10, where BT argued
that no mortal human agent can make a significant difference or a permanent
impact. We must consider two separate cases, depending on whether AP acknow-
ledges human values or not.
If we recognize human values, then APAC can borrow standard consequentialist
replies to impotence. We must first clearly separate the charge of impotence from all
epistemic worries. Suppose I do know the expected value of my available options. If
one option has greater epistemic value, then APAC tells me to perform it. I am
therefore not impotent. Of course, on a global scale, my contribution will be
comparatively very small. But that doesn’t make it irrelevant. If there are a million
people to be saved, and I can only save one, then it is no objection to APAC if it tells
me to save one person. This result looks insignificant, but it is not.
Impotence therefore only bites if all my options have equal expected value: if
I truly cannot make a difference. This will occur, not for very large finite populations,
but for infinite ones, where my action leaves the total sum of happiness unchanged.
However, as we saw in chapter 10, utilitarians argue that I am not impotent even in
infinite cases.18 Most utilitarian solutions to the infinity puzzles posit some alterna-
tive ranking of possible states of affairs, such that universe A can be better than
universe B even if both contain the same infinite amount of happiness. (For instance,
if everyone in A is twice as happy as everyone in B, then A is better than B.) The
search for a consistent alternative ranking is the main focus of the literature on
infinite utility. If that search succeeds, then APAC can borrow the result. Even if the
universe contains an infinite number of locations where objective value is present, my
actions can still produce a better universe, if they raise objective value at enough
locations.
Not everyone finds standard consequentialist replies to impotence persuasive.
Within consequentialism, rule consequentialists object that my individual contribu-
tion often is insignificant or irrelevant. We can only escape impotence by shifting to a
collective evaluation of consequences. As we saw in chapter 10, AP shares this worry.
But that is mainly because AP has doubts about the significance of individual human
lives considered in isolation. I argued in chapter 12 that AP could respond by making
the value of large-scale human endeavours a constituent of human well-being. But
perhaps it is more natural for AP to abandon human-sized values altogether.
18
This solution to the infinite utility puzzle is from Vallentyne and Kagan, ‘Infinite Value and Finitely
Additive Value Theory’.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
Accordingly, let us now suppose that AP denies the value of human well-being
altogether and recognizes only cosmic values. Impotence is now a serious threat. My
individual actions have no objectively valuable impact. I can only affect human
affairs, and human affairs have no objective value at all.
Faced with a value she cannot promote, the consequentialist has three main
options. First, she can bite the bullet. Consequentialism offers no guidance at all.
Every option is morally permitted, and we must choose on some non-moral basis.
While many consequentialists accept that morality doesn’t determine every choice,
few will go this far!
The consequentialist’s second response to impotence is to find something else that
she can promote. In chapter 12, we explored many different ways that human values
might enter AP morality. Perhaps the very fact that APAC collapses into impotence
without human values is our reason to posit those values. APAC could here borrow
from my own earlier contribution to the infinite utility debate.19 The alternative
rankings that allow us to distinguish between infinite populations can be interpreted
in two different ways. We could interpret them as objective accounts of aggregate
value. Even though A and B contain the same infinite total utility, A is better than
B. My contribution really does make the cosmos better. A modest alternative reading
is that, although I do not raise total cosmic value, I do raise local value. Suppose I am
a utilitarian, and I double the happiness of everyone on my planet. Because my
universe is infinite, total cosmic happiness remains unchanged. But I have increased
total local happiness. Because I cannot affect total cosmic happiness, local happiness
is the only thing I can promote.
The shift from global to local value is a departure from pure APAC. We abandon
the core consequentialist idea of maximizing overall value in favour of some nar-
rower value that we can promote. APAC now takes another step along the same road,
replacing genuine cosmic values with something else that individual humans can
influence. The difference, of course, is that our new replacement has no objective
value at all. However, as we saw in chapter 12, the introduction of human values is
not completely arbitrary. Even if human affairs have no objective value, they can still
resemble cosmic values to a greater or lesser degree. Flourishing human lives are
closer to cosmic purpose than impoverished ones. Not close enough to objectively
matter, but still closer. Once I realize that I cannot promote cosmic values, I could
aim instead to promote some (imaginary) value based on resemblance to cosmic
purpose. I am still responding to cosmic values, and in a recognizably act conse-
quentialist way.
As we saw in chapters 10 and 12, this reply may push us in the direction of larger-
scale, intergenerational human projects—as these come closest to cosmic values. Rule
consequentialists then insist that such projects are best promoted collectively, rather
19
Mulgan, ‘Transcending the Infinite Utility Debate’. The argument in the text also echoes chapter 12’s
treatment of values as deliberative constructs rather than moral foundations.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
than individually. (And, as we’ll see in section 13.5, AP rule consequentialism claims
to avoid fictional human values altogether, by linking human morality to the
collective promotion of real cosmic values.)
A third reply to impotence is to abandon promotion, and seek to respond to
cosmic values in a different way—such as contemplation, study, or worship. Rule
consequentialists often embrace alternative responses, and we explore them in
sections 13.5, 13.9, and 13.10. However, most act consequentialists would regard
this as an abandonment of consequentialism, not a reinterpretation.
I conclude that the impotence objection need not be fatal to APAC. If it can value
human well-being, then APAC can borrow standard consequentialist replies and is
therefore no worse off than atheist act consequentialism. If it cannot value human
well-being, then APAC may still avoid impotence by adopting fictional but not
arbitrary human-sized values based on cosmic values. (As we’ll see in sections 13.5
and 13.6, other replies to impotence are available to those consequentialists who are
willing to move beyond APAC.)
13.4.3 Is APAC too demanding?

Opponents of consequentialism have long accused it of being implausibly alienating,
demanding, or immoral. While AP takes these features of consequentialism to
extremes, it may also offer new resources to defend them.
APAC seems very counter-intuitive. It obliges me to give almost no weight to my
own interests or even those of humanity as a whole, and to violate many common-
sense moral prohibitions. I may find myself killing, lying, or abandoning personal
projects in pursuit of small increments to distant cosmic values. Is this a credible
moral theory?
If AP can rebut the epistemic and impotence objections, this success will unfor-
tunately exacerbate these new objections. AP’s advice is much more alien if I can
affect cosmic value, rather than having to settle for promoting some human-sized
proxy for real value!
All three objections raise similar issues for APAC. To avoid repetition,
I concentrate on the demandingness objection. Consequentialists offer many replies.
I shall argue that AP can borrow the best of these.20
One response is denial. Despite appearances, consequentialism does not make very
great demands. Consequentialism now benefits from its epistemic deficiencies. To
defeat the epistemic objection, we shifted from actual results to expected ones. I am
much better informed about the likely impact of my actions closer to home. If my
goal is to maximize expected human well-being, then my best course of action is to
focus on my nearest and dearest. Consequentialism is now much less likely to
demand that I sacrifice their welfare to save distant strangers.
20
This section draws especially on Mulgan, The Demands of Consequentialism, chs 1 and 2.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
Denial has its defenders. However, in an age of global information and compara-
tively efficient aid agencies, this is not a plausible reply to the original demandingness
objection. Even if distant aid is not efficient, there are always opportunities for self-
sacrificial do-gooding closer to home. By contrast, denial is a credible response to the
extra demands imposed by AP, especially its most alienating demands. In practice,
I am very unlikely to know that I could sacrifice humanity for a superior (cosmically
significant) alien race; or that, by inflicting great suffering, I could make the cosmos
more beautiful. (And surely it would be the height of caprice to even take these
possibilities seriously!) But if such sacrifices are impossible, then consequentialism
will never actually demand them. Given my ignorance of cosmic values, my best
consequentialist option is to focus on human well-being, perhaps with some shift
towards projects that vaguely resemble the cosmic purpose. Denial can thus remove
the additional counter-intuitiveness of AP act consequentialism.
Denial will remain controversial. One important theoretical question concerns the
moral significance of counterfactuals. Does it count against a moral theory that it
would offer unacceptable advice in some possible but non-actual situations? If I were
sure that the extinction of humanity would promote some (greater) cosmic value,
then consequentialism would demand this. Is this a decisive objection? Philosophers
disagree. But most would agree that, other things equal, our moral theory should
deliver acceptable advice in plausible counterfactuals. If so, denial is not a fully
adequate response. On the other hand, most moral philosophers don’t find counter-
factual demands as troubling as real ones. So denial does at least weaken the
demandingness objection.
A second way to reduce APAC’s demands is to remove the gap between self-
interest and impartial value. If a flourishing human life is a life devoted to the
common good, then such devotion involves no sacrifice of personal well-being.
Elsewhere, I have argued that, for standard human values, the gap between self and
others cannot plausibly be eliminated.21 Act consequentialism does demand consid-
erable sacrifices. However, AP’s non-human-centred human values offer this elim-
ination argument new life. If a close connection to large-scale human projects
directed at cosmic values is the only way my life can have any value, then a moral
theory that instructs me to devote myself exclusively to such projects, and abandon
all other activities, merely tells me to maximize my own well-being. If its values are
correct, then APAC is not really demanding at all. (On the other hand, we may feel
that this degree of commitment to non-human-centred human values is itself too
alienating. If so, we must seek an alternative solution.)
A third response to demandingness is to bite the bullet. Extremist consequentialists
insist that morality just is very demanding and alienating. My life is not objectively
more valuable than anyone else’s. So why shouldn’t I be obliged to sacrifice myself for
21
Mulgan, The Demands of Consequentialism, ch. 1. We also addressed similar issues in chapter 10, in
connection with Sidgwick’s dualism of practical reason.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
the greater good? I might find morality’s demands difficult or unappealing. But so
what? Psychological difficulty does not negate normative force.
Extremism strikes me as the best defence of APAC. AP strengthens this response
in several ways. If my own life has no objective value, or if other aspects of the cosmos
have much greater value, then a demanding morality becomes more compelling.
And, as I argued in section 13.3, AP supports impartiality between human beings.
Given our cosmic insignificance, and the many different ways that human lives might
dimly resemble cosmic purpose, we have no justification for treating one human
being as more valuable than any other. AP thus supports extremism by rendering the
ordinary demands of consequentialism less objectionable. (We could then borrow
from our previous replies to argue that AP introduces no additional demands.)
One key battleground between extremists and moderates is the methodological
role of intuitions. Extremists dismiss our moderate moral intuitions as products of
caprice. The fact that consequentialism seems too demanding is no objection to it. As
we saw in section 13.1, AP supports this suspicion of human moral intuitions. If true
moral facts are founded on a strange and unexpected cosmic purpose, then our self-
aggrandizing intuitions are a very unreliable moral guide. As the aversion to caprice
is a guiding theme of our project, it is no surprise that AP finds extremism congenial.
AP further supports extremism by providing alternative sources of moral know-
ledge that are independent of our (moderate) intuitions. One route here is meta-
ethical: if we derive our moral obligations directly from cosmic values, then their
credibility does not depend on their intuitive appeal. A second route is mystical. One
theme of chapter 5 was that mystics typically endorse a very demanding and
alienating morality. If mysticism offers moral insight, then morality probably is
very demanding. (As we saw in section 13.1, chapter 5’s defence of mystical moral
knowledge only covers general principles and abstract values. Chapter 5 thus reduces
the comparative force of the particular moral judgements that support a less
demanding morality.)
Many consequentialists, and most non-consequentialists, find extremism hard to
swallow. And AP does make consequentialism more extreme. On the other hand, it
also offers new justifications for extremism and new replies to objections. AP’s net
impact on extremism may thus be positive.
If extremism still seems an implausible guide to human action, we could treat it
instead as a moral ideal. We should aim for the extreme, even though we have little
hope of reaching it. This use of extremism is congenial to AP. As we saw several
times in earlier chapters, AP naturally pictures human knowledge as groping
towards some unattainable cosmic ideal. It then makes sense to see human
morality in the same light.
I conclude that APAC extremism is worthy of serious attention. While even more
extreme than atheist extremism, it also has additional resources. If AP is the most
plausible metaphysical picture of our cosmos, then APAC might be our most
appropriate moral response. However, for those who still find APAC too extreme
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
even in a non-human-centred world, more moderate consequentialisms are avail-

able, as we shall now see.
13.5 Rule Consequentialism

The most popular consequentialist response to demandingness is neither denial nor
extremism. It is restructuring. Many consequentialists concede that act consequen-
tialism makes implausible demands. But they reply that some other recognizably
consequentialist moral theory avoids those demands.
The most popular alternative to act consequentialism is rule consequentialism.
Rule consequentialism can initially be characterized as follows: an act is morally right
if and only if it is called for by a set of rules whose acceptance by everyone would
result in at least as good consequences judged impartially as any other. The funda-
mental moral questions are: ‘What if we did that?’, and ‘How should we live?’ My
version of rule consequentialism draws on Brad Hooker’s recent formulations.22 We
first seek an ideal moral code. Acts are then assessed indirectly: the right act is the act
called for by the ideal code. We imagine ourselves choosing a moral code to govern
our community. I operationalize rule consequentialism as follows. To evaluate ideal
codes, I ask what would happen if we (the present generation) attempted to teach a
given moral code to the next generation. This sets aside the cost of changing existing
moral beliefs, but factors in the cost of (for instance) trying to get a new generation to
accept a very demanding ethic. This gives us our ideal code. To evaluate actions, I ask
what someone who had internalized the ideal code would do in my situation.
Rule consequentialism is subject to many objections, and the literature is substan-
tial. Rather than rehashing old debates, I focus on places where AP either exacerbates
the problems of rule consequentialism, or offers it new resources to defeat objections,
or influences its content. Our guiding question is simple: can AP develop a plausible
human morality by replacing act consequentialism with rule consequentialism?
13.5.1 Why should we follow rule consequentialism?

Rule consequentialism faces a motivation problem. Why should we follow the ideal
code? Why should anyone follow the rules that generally promote value, rather than
directly promoting value individually? Opponents accuse rule consequentialism of
‘rule worship’. Why follow the ideal code when you could do more good by breaching
it? In other words, why follow rule consequentialism when it diverges from act
consequentialism?
As we saw earlier, Hooker argues that ‘the best argument for rule consequentialism
is that it does a better job than its rivals of matching and tying together our moral
22
Hooker, Ideal Code, Real World. I develop my version in Mulgan, The Demands of Consequentialism,
especially ch. 3; Mulgan, Future People, chs 5 and 6; and Mulgan, Ethics for a Broken World, ch. 7.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
convictions’.23 Rule consequentialism is preferable to act consequentialism because it

better fits our moral convictions. Consider demandingness. Rule consequentialism
makes only reasonable demands, as a relatively small percentage of the combined
income of the well off would be enough to meet all the needs of everyone on earth.
Rule consequentialism also seems fairer than act consequentialism, as it does not
require an agent to do more than her fair share in promoting the good, even if others
fail to do theirs. I have argued elsewhere that rule consequentialism unites a range of
distinct common-sense departures from act consequentialism, by incorporating
them into its ideal code.24
Hooker’s original reflective equilibrium defence of rule consequentialism is con-
troversial.25 AP creates two new problems for such a defence. First, cosmic rule
consequentialism could be very counter-intuitive. (For instance, it could instruct us
to bring about the cessation of human life, if that would best promote some non-
human cosmic value.) Second, as I suggested in section 13.1, coherence with human
intuitions is a strange test for cosmic morality.
However, all is not lost. We can construct a new two-tiered defence of rule
consequentialism. If we have a theoretical defence of consequentialism in general,
then rule consequentialism may then emerge as the most intuitively appealing (or
least unappealing!) variety of consequentialism. And, as we have just seen, rule
consequentialism does seem more appealing than act consequentialism.
Another option for AP is to abandon particular intuitions altogether, and offer a
theoretical defence of rule consequentialism. Elsewhere, I have suggested that we
should consider rule consequentialism as a departure from a pure consequentialist
ideal, a regrettable but inevitable concession to human finitude.26 Act consequen-
tialism is the moral ideal, but it is not appropriate for us, due to our limited
knowledge, limited powers of calculation and concentration, and limited sympathy.
Unlike an infinite, self-sufficient, perfect God, we are finite fallible social beings. For
us, the promotion of value can only be a collective enterprise. The best way that we
collectively can promote value is by following the rule consequentialist ideal code.
This model of rule consequentialism is very congenial to AP for two reasons. First,
AP makes the ideal real. The AP consequentialist ideal is God’s actual cosmic
creation in response to cosmic values. Second, AP’s ideal, while real, is also more
distant from us than any atheist ideal. We have already seen many times how difficult
it is for human beings to respond meaningfully to such an austere cosmic ideal. AP
forces consequentialists to make concessions. We must depart from APAC. And
23
Hooker, Ideal Code, Real World, p. 101. See also Hooker, ‘Rule-Consequentialism, Incoherence,
Fairness’, p. 29.
24
Mulgan, The Demands of Consequentialism, ch. 3; Mulgan, Future People, ch. 5.
25
See, e.g., Miller, ‘Hooker’s Use and Abuse of Reflective Equilibrium’; and the discussions cited in the
previous footnote.
26
Mulgan, Future People, ch. 5.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
once we move away from act consequentialism, rule consequentialism is the closest
approximation to the cosmic ideal that is available to fallible finite human agents.
Another way to bridge the gap between divine act consequentialism and human
rule consequentialism is to imagine, not a single creator God, but a group of creators.
Throughout this book, I have borrowed BT’s assumption of a unique personal
creator. However, as Hume famously observed, another possibility is a collaborating
group of creators.27 The consequentialist interpretation here is that a group of gods
bring the universe into being in response to cosmic values. On this model, our ideal is
cosmic rule consequentialism. We seek the ideal code that, when internalized by a
group of cosmic creators, would produce the best cosmos. This cosmic ideal code
would replace act consequentialism as the inspiration for our human morality.
BT consequentialism offers another reason to prefer rule consequentialism. In
section 13.8, I argue that AP can borrow BT divine command ethics, where human
moral obligations are commands issued by God. If that argument succeeds, it favours
AP rule consequentialism (APRC) over APAC, because rule consequentialism fits
more easily into a divine command model. Historically, BT utilitarians were typically
rule utilitarians. God’s call is issued to human beings living in society. Although the
Ten Commandments are addressed to each of us individually, their utilitarian
justification rests on the beneficial effect of our collective obedience. AP cannot
believe in divine commands issued to human beings, but rule consequentialism
could model a divine call to other rational beings. Or so I argue in section 13.8.
I conclude that AP is compatible with several credible defences of APRC. If rule
consequentialism is on the table for the atheist, then it remains a viable option
under AP.
13.5.2 Human well-being within APRC

Any rule consequentialism based on cosmic values will still be very alien. If we find it
too alien, we can add human-sized values. Chapter 12 considered several different
ways to add human values to AP’s cosmic ones. As well as borrowing those earlier
solutions, APRC has new resources here. In this section, I bring the various options
together.
1. Human-sized values only. We could simply stipulate a second departure from the
ideal of cosmic act consequentialism. In addition to a collective response to value,
we also replace cosmic values with human-sized values. The best human approxima-
tion of the cosmic ideal is human rule consequentialism. This brings us closer to
conventional (atheist) rule consequentialism. But APRC is still distinct. As we saw in
chapter 12, AP’s interpretation of human well-being can include both responses to
cosmic values and resemblances to such values. Therefore, the ideal code that
maximizes AP’s human values will be especially austere and outward-looking.
27
Hume, Dialogues Concerning Natural Religion.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
This option ignores AP’s cosmic values altogether. It is therefore unlikely to serve
as a complete AP moral theory. But we could combine human-sized rule consequen-
tialism with a more cosmically oriented act consequentialism, perhaps within an
overarching hybrid moral theory such as that discussed in section 13.6.
2. The two-tier model. Chapter 12’s two-tier model distinguishes objective values
from the values relevant to specific deliberative contexts. The scale of objective values
may be more fine-grained than the cosmic values that motivate God’s creative
decisions. If our ideal code maximizes objective value rather than cosmic value, and
if human lives have objective value, then the ideal code is sensitive to human
concerns. (The threat remains that human concerns will be swamped by cosmic
ones. We return to that threat in section 13.5.4.)
3. Human values as deliberative constructs. Elsewhere, I have distinguished two roles

that values play in rule consequentialism.28 We seek to promote our foundational
values, while deliberative values guide particular ethical decisions. We use founda-
tional values to select the ideal code, and then our deliberative values emerge within
that code. A moral code guides our deliberations, in part, by telling us which values to
adopt. My original example was the lexical level. (In chapter 5 of Future People, I argue
that, instead of building a single ad hoc lexical level into our foundational values, we
should treat lexicality as a feature of deliberation within the ideal code. This new
lexical level tells the ideal society’s inhabitants what they can reasonably insist on, and
what must be guaranteed to all.29) In the present case, our deliberative values could be
human, even though our foundational values are strictly cosmic. Our best collective
response to cosmic values involves the internalization of human-sized values. Human
moral agents can only promote cosmic value indirectly, by aiming at human-sized
analogues. Cosmically significant projects will then emerge, over many generations,
from our collective commitment to human well-being.
This is not ad hoc. Rule consequentialists often emphasize that the ideal code
adopts non-consequentialist ends. Human agents do not flourish if they regard
everything as a means to maximum aggregate well-being. We best promote our
foundational values if we don’t always aim to promote them.
I conclude that APRC can justify its recognition of human well-being in a number
of ways—either as a necessary departure from the cosmic ideal, or as a pale resem-
blance of cosmic values, or as an appropriate deliberative response to the underlying
scale of objective values.
28
Mulgan, Future People, ch. 5where I first developed the two-tier model deployed in chapter 12.
29
On lexicality in general, see Griffin, Well-Being, pp. 85–9; Mulgan, Future People, pp. 64–9. The
initial motivation for introducing a lexical level is to avoid Parfit’s repugnant conclusion. However,
lexical accounts of foundational values lead us into paradox. My solution allows us to retain a total
utilitarian account of foundational values without embracing the repugnant conclusion (Mulgan, Future
People, ch. 3).
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
13.5.3 The scope of rule consequentialism

Rule consequentialism says that the right action is the one that follows from the code
whose widespread internalization would produce the best consequences. One press-
ing question is: internalization by whom? Who will learn the various moral codes we
are comparing? We ask ‘what if everyone did that?’ Does this include everyone in my
society, my nation, my species, my planet, or my cosmos? Do we include future
people or limit ourselves to the present? What is the scope of rule consequentialism?
These are familiar debates within rule consequentialism.30 Some limit the ideal
code to one society or group. Others apply rule consequentialism globally, asking
what would happen if all human beings were taught the same ideal code. Broader
scope increases the distance from act consequentialism. For instance, local rule
consequentialism can be very demanding, especially if the local population is small
relative to the pool of potential beneficiaries. (Imagine an ideal code designed to
completely eliminate world poverty even when it is internalized only by the popu-
lation of Norway!) Rule consequentialists who seek a less demanding theory thus
typically favour broader scope. Their ideal code must be internalized by all human
beings.
AP supports broad scope. AP removes the main rationale for local rule conse-
quentialism, because differences between groups of people are very unlikely to have
any cosmic significance. AP also provides a positive motivation for broad scope. The
strongest reply to the impotence objection is that we can make a difference on a
cosmic scale. This reply requires the broadest possible scope. AP pushes us towards
ever-more ambitious collective goals—such as the preservation of our ecosystem, the
discovery of cosmic purpose, or the exploration of the universe.
Rule consequentialism increases the importance of the future. My individual
actions may have no long-term impact, but our collective actions surely do. (My
consumption of fossil fuels may not matter, but humanity’s consumption matters a
great deal.) Our choice of ideal code must consider future consequences. A trickier
issue is temporal scope. Do I ask what would happen if everyone did this now, or
should I follow the rules whose internalization by all human beings, from now on,
would maximize value?
AP favours the broadest temporal scope, because it prompts us to consider
collective agents that extend far into the future. (As we saw in chapter 10, such
agents may be essential to AP’s reply to BT’s meaning argument for personal
immortality.) Even if individual human lives are irrelevant, humanity as a whole
could contribute something cosmically significant. At this preliminary stage of
human history, we have little idea what that contribution might be. So perhaps the
ideal code first directs our attention to study and exploration, in the hope that distant
future generations will know how to make a cosmic difference. (Such a code will also
30
Hooker, Ideal Code, Real World, ch. 8; Mulgan, Demands of Consequentialism, ch. 3.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
favour the preservation of humanity and its resources. Distant future people cannot
promote anything if they do not exist!) APRC thus provides a useful corrective to our
natural tendency to assume that the purpose of humanity can be realized (or even
comprehended) in the present.
Rule consequentialism also offers a new solution to a familiar problem for
AP. What if not even the whole of humanity could make any difference to cosmic
value? We begin with the best rule consequentialist solution to the original infinite
utility puzzle. If the human population is infinite, then we should ask what would
happen if everyone in the infinite population followed some particular moral code.
A code that makes all sentient beings happy is better than one that made everyone
unhappy.31
Suppose instead that the infinite population includes non-human rational beings,
alongside a finite number of humans. (We discussed this possibility at length in
chapter 7.) Rule consequentialism will then go beyond human beings, and imagine
internalization by all beings capable of learning a moral code. After all, our fixation
on moral rules for human beings seems absurdly parochial from AP’s cosmic
perspective. AP removes the rationale for any restriction to human beings. There is
nothing significant about us.
Consider another example from chapter 7. Suppose the cosmos contains, in
addition to human beings, superior rational beings whose understanding of its nature
and purpose is itself a part of that purpose. (God has created this universe, in part, in
order that these superior beings will emerge within it.) Rule consequentialism now
recommends the ideal code whose internalization by all rational beings would
produce the best consequences. Ex hypothesi, some rational beings do have a
cosmically significant impact. (If nothing else, it matters whether or not they try to
understand the universe. So the ideal code will definitely instruct them to do that.)
Rule consequentialism now easily avoids the impotence objection. The collective
impact of all rational beings does matter.
In this scenario, it makes no cosmic difference whether human beings internalize
the ideal code or not. Our collective impact is not significant. The consequences are
the same whether rule consequentialism’s scope includes all rational beings or is
limited to cosmically significant ones. However, broader scope is preferable for two
reasons. First, it delivers a simpler moral theory. (As we saw in chapter 12, AP
reinforces the value of simplicity.) Second, if our objective values are more fine-
grained than our cosmic ones, then a broader rule consequentialism does produce
better results. Cosmically insignificant rational beings can still be objectively signifi-
cant. Recall our earlier decision to replace actual values with expected ones. Unless we
are certain that there is no objective value in the human realm, broader scope yields
31
Surprisingly, rule consequentialist discussions of infinite utility are rare, and the solution sketched in
the text is not one I have seen elsewhere.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
greater expected value. While it may be caprice to assume we are objectively valuable,
we cannot be certain that we are not.
APRC can thus conclude that X is the right action for me because it follows from
the code of rules whose internalization by everyone best promotes cosmic value, even
though X has no cosmic significance at all.
What are we to make of these tales of infinitely numerous or objectively superior
rational beings collectively promoting cosmic value? In chapter 7, I argued that AP
does give us good reason to believe that this extraterrestrial story is true. (Or, at least,
that we best avoid caprice by proceeding on the assumption that it might well be
true.) An alternative is to regard non-human rational beings as a moral fiction.
Perhaps we posit an infinite or cosmically significant rational community so that
we can construct a human-sized response to cosmic values. As we saw in chapter 12,
the resulting fiction is still closer to moral realism than any atheist fiction, because the
cosmic values at its foundation are real.
13.5.4 The content of the ideal code

APRC’s moral guidance depends on our knowledge of the content of the ideal code.
But rule consequentialism is even more epistemically optimistic than act consequen-
tialism. How can we possibly know exactly what code of rules would maximize
human well-being? As ever, AP exacerbates this difficulty, especially if it adds
infinitely many non-human rational beings, or superior aliens, or non-human cosmic
values.
Detailed knowledge of any ideal code is beyond us. If rule consequentialism insists
on such knowledge, then it already faces a decisive epistemic objection well before AP
enters the picture. Any credible rule consequentialism limits its epistemic demands.
The standard reply is that while we cannot discover all the details of the ideal code,
we can reasonably make some general predictions. One starting point is that things
go best if, other things equal, agents seek to promote the good. Act consequentialists
argue that the individual reason to promote the good exhausts the ideal code. Things
go best if everyone always maximizes the good. Rule consequentialism thus collapses
into act consequentialism. Rule consequentialists reply that, due to human calcula-
tive limitations and ineradicable partialities, act consequentialism is not the best
code. Things go best overall if people do not relentlessly seek to maximize. The ideal
code is not limited to the reason to promote the good. This debate within conse-
quentialism naturally highlights the differences between the two theories. But we
should not overlook their similarities. No rule consequentialist rejects the reason to
promote the good altogether. Any plausible ideal code must give that reason a very
prominent place. Rule consequentialism emphasizes benevolence, and good-
promotion in general, more than conventional morality does. In particular, the
ideal code will demand more in relation to famine relief and the distant future
than most other ‘moderate’ ethical theories.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
We do therefore have some knowledge of the ideal code: it includes a strong reason
to promote the good. Do we know more? We can start by asking why that single
consequentialist reason is not sufficient. What else will our code include?
Rule consequentialists often borrow from existing moral codes. Actual codes
would not have survived unless they promoted human well-being reasonably well.
We do better if we follow the tried and true rules of common-sense morality, rather
than either adopting act consequentialism or trying to figure out the ideal rules for
ourselves. Hooker himself even offers closeness to conventional morality as a tie-
breaker when our ability to independently predict consequences runs out.32 Rule
consequentialism is thus more conservative than act consequentialism.
AP challenges any moral conservatism. Conservatism sits uneasily with broad
scope and the introduction of cosmic values. Why suppose that our local moral code
is a good way for all humanity (let alone all rational beings) to promote the good? It is
one thing to suggest that human morality evolved to serve human ends, but quite
another to claim that it represents the best way for all rational beings to promote
cosmic value.
Hooker’s conservatism fits within his reflective equilibrium defence. But as we saw
in section 13.1, AP rejects any reflective equilibrium method that privileges common-
sense intuitions about particular cases. (As I noted in section 13.1, this is not the
wholesale abandonment of all moral judgements, but instead the replacement of
intuitions about particular cases with abstract principles and values that are poten-
tially much more radical.) AP thus undermines the main independent motivation for
conservatism. Just as we cannot use intuition to defend APRC in the first place, we
also cannot turn to common-sense morality to flesh out the ideal code. Rule
consequentialists must instead provide theoretical arguments, drawing lessons
about the ideal code from the nature of value or the human condition. I will briefly
discuss four examples. The ideal code will: emphasize obligations to future people;
include different rules for different groups of people; sanction responses to value
other than promotion; and make room for individual moral autonomy. I shall argue,
as ever, that AP can borrow, support, and transform these four rule consequentialist
commitments. I discuss the latter two examples in detail in sections 13.5.5 and 13.5.6.
This section deals with the first two.
We have already discussed consequentialism’s theoretical emphasis on future
people. In practice, act consequentialism can sidestep the distant future, because
my individual impact is very slight and almost impossible to predict. Rule conse-
quentialism cannot so easily avoid the future. Collectively, present people have a
very significant impact on future people. And, as we saw in section 13.5.3, AP
supports inter-temporal scope and large-scale intergenerational projects. We must
32
Hooker, Ideal Code, Real World, p. 32.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
consider both our impact on future people and their contribution to objective
values.33
Common-sense morality includes obligations to future people. But rule conse-
quentialism goes much further. If our ideal code has temporally impartial founda-
tions, then future people are overwhelmingly significant and rule consequentialism
must be revisionary—demanding great present sacrifices to promote the interests of
distant future people.34
Intergenerational ethics also provides an excellent example of the need for diversity
within the ideal code. As I suggested in section 13.5.3, within any large-scale inter-
generational project different generations will play different roles. Given our ignor-
ance of the cosmic purpose, our role is to explore and study so that later generations
will know what they should promote.
Diversity of roles is widespread in the ideal code. A human society where everyone
does the same thing will not maximize value. A theme of recent rule consequential-
ism is that a universal moral code may give different advice to different agents,
depending on their situation, circumstances, or inclinations.35 (Do we really want
doctors, mercenaries, public servants, and journalists to all internalize the very same
rules?) Opponents object that rule consequentialism cannot accommodate role
diversity, because the ideal rules must be the same for everyone. But this is a mistake.
A moral code need not be uniform in application. The same code can say different
things in different situations. The reflective doctor realizes that her rules of person-
alized attention to individual patients are not appropriate for the bureaucrat who
must allocate scarce resources within the health system. Rule consequentialism can
embrace diversity.
APRC introduces a new diversity. As well as diversity within a society, and across
generations, we might imagine diversity between rational species. If we share our
cosmos with superior alien agents, our common ideal code may say very different
things to different species. Diversity arises in two general contexts. First, for epistemic
and motivational reasons, the ideal code advises each person to focus on her own
concerns, because this is where she can most reliably do good. Analogously, an inter-
species code might instruct each species to focus on its own collective projects. (We
are unlikely to know how to promote the interests of aliens!) Second, the ideal code
will recommend specialization when different roles in a complex society require
distinct moral temperaments or priorities. This is particularly likely across species,
especially if some can have a cosmically significant impact while others (such as
33
My own interpretation of rule consequentialism introduces a third role for the future, because the
consequences of teaching a code to the next generation include the impact of our present teaching on the
behaviour of subsequent generations.
34
35
I explore the need for diversity within rule consequentialism in Mulgan, The Demands of Conse-
quentialism, pp. 98–103.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
humans) cannot. The ideal code may prescribe an active moral role to the former,
relegating impotent species to a role that is either contemplative or inward-looking.
(If we cannot promote cosmic values, we can only look after ourselves.)
APRC also supports diversity by supporting both non-promotion and pluralism,
to which we now turn.
13.5.5 Beyond promotion

Act consequentialism only allows one response to value. Agents must always and
everywhere promote value. Rule consequentialism defends many non-consequentialist
responses to value. Finite fallible agents often do better if, instead of seeking to
relentlessly maximize value, they respond to value in other, less direct ways. Rule
consequentialist agents will sometimes honour, study, or embody values rather than
relentlessly trying to maximize them.
AP supports an ideal code that is not fixated on promotion. Our ability to directly
promote cosmic values is hampered by both ignorance and incapacity. A theme of
this book is how little we currently understand the cosmic purpose and its connection
to value. We cannot promote what we do not understand. AP will therefore favour an
ideal code that advises careful study of cosmic purpose and cosmic values as a
necessary preliminary to any attempt at promotion. This ‘preliminary’ task may
take generations. Perhaps, for many individual rational agents, it will be the central
moral task of humanity during their lifetime. This is a problem for APAC, where
I consider only the predictable consequences of my present action. But rule conse-
quentialism is on familiar ground here. The ideal code already includes rules gov-
erning situations where individual agents or groups of agents cannot directly
promote value. Alternatives to promotion can enter the ideal code either as prelim-
inaries to future promotion, or as by-products of general dispositions that typically
promote value. For instance, if the ideal code tells all rational agents to care about
beauty, then an agent who has internalized that code will not be indifferent to beauty
simply because she cannot promote it.36
Our cosmic ideal code will include non-promotion. Even superior non-human
rational beings cannot always directly promote cosmic value. Even the best rational
beings may often do better by seeking to instantiate or admire cosmic value in their
own lives or communities, rather than vainly trying to maximize it across the cosmos.
Perhaps our best response to that ideal code is to follow those recommendations. For
superior beings, non-promotion is occasional; for us, it may be permanent. The ideal
code thus can offer human guidance, even if we cannot meaningfully promote cosmic
value at all.
36
The ability to consider responses other than promotion also enables rule consequentialism to sidestep
Robert Adams’s objection that consequentialism offers no guidance in hopeless situations, where agents
cannot even hope to promote the good (Adams, Finite and Infinite Goods, p. 225).
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
13.5.6 Consequentialist liberalism

As we saw in chapter 8, utilitarians have always had an ambivalent relationship to
human freedom. One authoritarian strand sacrifices freedom to the greater good—
forcing individual agents to maximize aggregate well-being at the expense of their
own autonomy, and favouring bureaucratic institutions that maximize material
welfare. This is the popular caricature of utilitarianism and it is not without a basis
in fact. But another strand, most famously associated with J. S. Mill, unites utilitar-
ianism and liberalism. Freedom is both an essential component of a flourishing
human life and the most efficient way to promote human happiness.
Contemporary rule consequentialism embraces this Millian liberal tradition.37 The
ideal code encourages individuals to make their own experiments in moral living, and
it leaves wide scope for individual moral choice. Given our nature, and the way that
our well-being is constituted, a liberal code promotes aggregate well-being much
better than any illiberal alternative. The ideal code doesn’t just support liberal politics
over despotic alternatives. It also supports moral freedom, where agents feel permit-
ted to choose among a range of competing actions, lifestyles, or experiments in living.
AP can support liberal rule consequentialism. This may seem surprising. After all,
if human well-being has no value, then neither does human freedom. However, if AP
is able to deliver any rule consequentialist morality, then it will support a liberal one.
AP enhances Mill’s famous argument for liberty based on pluralism about value.
There are many different values we can respond to, and many different valuable
responses. No one can combine all possible human goods in a single life. This is
partly due simply to constraints of time, but also because different valuable traits may
be incompatible. (Mystical contemplation and political activity cannot easily be
combined in a single life.) There is no single best way for human beings to live.
A moral code that tells everyone to do the same will inevitably omit many important
human goods. Furthermore, moral progress requires moral experiments in living. If
we all do the same thing, we will never learn to do better. AP supports this argument
because, as we saw in section 13.3, it recognizes many dimensions of value pluralism.
Contemporary rule consequentialism accommodates value-pluralism by building
it into the ideal code together with role diversity. Instead of a single set of prescrip-
tions for all human beings, we get a complex social framework with different rules for
different people. AP can borrow this solution, and take it in new directions. Some
agents will be naturally inclined to study cosmic purpose, while others are better
suited to following human-sized practical projects. The best human society—the
society that best promotes what is truly valuable—will include both. And if different
actions are best supported by different world views, then the best human society will
also include divergent metaphysical views as well. The ideal code will thus foster a
37
Hooker, Ideal Code, Real World, pp. 41–2; Mulgan, Future People, ch. 5.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
Millian liberal democracy, whose wildly different experiments in living are based on
incompatible metaphysical and meta-ethical foundations.
AP thus supports liberal rule consequentialism. But AP also changes liberalism. In
the conventional liberal ideal code, individuals choose which human goods to pursue.
Under APRC, individuals choose between competing approximations of the divine
response to cosmic values. Both codes provide ample scope for moral choice. But
they sanction very different choices. Liberty is not licence. No plausible ideal code
leaves all options morally open. AP may be both more permissive than conventional
rule consequentialism and more restrictive. AP’s code may permit acts that human-
sized codes forbid. Perhaps agents are allowed to sacrifice human values for cosmic
ones. (This is still less radical than APAC, where such sacrifices are demanded.)
Conversely, AP’s code may prohibit some frivolous human ‘goods’ that cannot
plausibly be thought to resemble cosmic values. (Here, the code does require the
sacrifice of human ‘values’ for cosmic ones, but only because the former are not really
valuable at all.) This difference will be especially stark if, despite the arguments of
chapter 12, AP cannot find a credible way to recognize the objective value of human
pleasures and preferences.
Conventional rule consequentialism defends liberalism by citing human fallibility
and human well-being. Human freedom is instrumentally valuable. Given our
limited cognitive abilities and motivations, we cannot reliably follow the reason to
promote the good. Human freedom is also intrinsically valuable. It is an essential
component of a flourishing human life. These human elements seem an insuperable
barrier to any AP liberalism. How can AP admit either that freedom is always good,
or that all rational beings are fallible?
AP can plausibly insist that freedom is a universal value for rational agents. Our
evidence about value comes from two sources: the human realm and cosmic purpose.
We know that freedom is good for us, if anything is. And the case for AP suggests
that freedom and creativity are also divine values. It is no great leap to assume that
freedom and creativity are good for superior rational beings whose connection to
cosmic values is much closer than ours.
Fallibility is more controversial. Perhaps truly cosmically significant rational
beings will lack our failings. But AP need not insist on universal fallibility. If some
rational beings are infallible, then the ideal code will offer them very different advice.
But its advice to us must take account of our fallibility.
The argument from fallibility may apply only to agents like us. However, argu-
ments based on pluralism will apply to all possible agents. Even an ideal code for
superior beings must somehow encourage a wide range of distinct experiments in
living. Either the ideal code specifies freedom as the universal route to valuable
pluralism; or the code offers diverse rules to deliver pluralism in ways that are
appropriate to the nature, limitations, and value of different agents. AP can remain
agnostic between these two alternatives. Whatever is best for other beings, freedom is
the only way for human beings to reap the benefits of pluralism.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
I conclude that APRC is one very promising avenue for future research in AP
moral theory. AP both supports rule consequentialism over act consequentialism and
transforms it. If we endorse AP, then our most promising moral outlook may be a
collective consequentialist one.
13.6 Hybrid Consequentialism

Act and rule are the most prominent forms of consequentialism. However, the
literature contains many other varieties. One that offers new resources to AP is the
hybrid view. The basic hybrid view is due to Samuel Scheffler.38 Act consequential-
ism goes wrong because it requires the agent to give her own well-being the same
weight as everyone else’s. The solution is an agent-centred prerogative, permitting the
agent to give disproportionate weight to her own interests in her practical deliber-
ations. This reflects the fact that, while her interests are not objectively more valuable,
they do loom larger from her personal point of view. More generally, any hybrid view
balances two sets of reasons: the impartial reason to promote the good and the
agent’s own personal reasons.
The hybrid view is a natural option for AP. Even if the reason to promote the good
does not cover them at all, human interests can still enter via the personal point of
view. Indeed, we could imagine a triple hybrid that balances (i) a cosmic reason to
promote cosmic values, (ii) an impartial (but species-specific) reason to promote the
human good, and (iii) a personal reason to promote my own good. The triple hybrid
view recognizes three legitimate perspectives: the point of view of the universe, the
perspective of humanity, and the personal point of view. (A simpler alternative might
instead treat the perspective of humanity as one aspect of my personal point of view
as a human being.)
Hybrid views face several theoretical challenges. How exactly do we balance the
competing reasons? Can we avoid unreasonable demands without being too lenient
elsewhere? Is the resulting theory still genuinely consequentialist? AP complicates
matters, especially if it introduces a triple hybrid view. But these complications
should be seen as challenges for further exploration, not as knock-down objections.
A more serious worry is whether AP can justify the initial departure from act
consequentialism. Why should we recognize either the personal point of view or the
perspective of humanity? One defence of any hybrid view is psychological necessity.
We are creatures who cannot help seeing the world from a perspective that privileges
our own interests and projects. But psychological necessary alone is not sufficient,
especially given the inevitable objection from caprice. Hybrid theorists also seek a
normative justification. For instance, utilitarians defend the personal point of view as
a vital aspect of human well-being. No human being can flourish unless her personal
38
Scheffler, The Rejection of Consequentialism. See also Mulgan, The Demands of Consequentialism, ch. 6;
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
perspective is developed and encouraged. Human well-being can only be promoted if

we recognize the significance, to each individual, of her own point of view.
AP can borrow this normative justification. Under theist AP, or any story where
cosmic purpose makes reference to creativity or consciousness or understanding,
personal points of view will capture something resembling cosmic values. If I strive to
promote creativity or consciousness or understanding in the cosmos as a whole, then
it is natural to take account of my own conscious creative standpoint. (And this may
be the only place where I can reliably promote those values.) Moreover, the human
point of view could be defended as an essential background to every human-sized
project. If AP can value anything human, then it can probably defend the individual’s
point of view.
The standard hybrid view gives both competing reasons equal normative and
meta-ethical standing. It balances the reason to promote the good against the agent’s
personal point of view without privileging either. A more radical alternative would
treat the cosmic reason as real, and add the human or personal reason as a (psycho-
logically necessary) fiction. In the final section of chapter 12, we explored the
possibility of introducing fictional human values based on AP’s real cosmic values.
If we find such a combination desirable, then a hybrid view would neatly separate
reasons based on fictional human values from those founded in real cosmic ones.
More baroque options are also available. My own work has explored combined
consequentialisms, where different consequentialist theories govern different moral
realms.39 My original suggestion combined an act consequentialist account of our
duties to distant strangers with a rule consequentialist story about obligations within
our local moral community—with the two components balanced by a revised agent-
centred prerogative. AP morality might instead combine rule consequentialism about
human-sized values with act consequentialism about cosmic value. Or we could
combine two different collective theories—perhaps so that all rational beings pro-
mote cosmic values and all human beings promote human values. I leave the details
for another day.
This concludes our exploration of what AP can borrow from atheist consequen-
tialism. Many issues remain unresolved, but we have seen that AP can develop
interesting new consequentialist theories. A full discussion of AP’s moral options
would now discuss atheist non-consequentialism in similar detail. In principle, any
non-consequentialist moral theory could be adopted by AP. However, in the
interests of space, we must be selective. Accordingly, we now ask instead what
AP can borrow from BT ethics. We have already seen that AP draws heavily on BT
meta-ethics, and chapter 12 explored BT contributions to AP’s interpretation of
human well-being. The rest of this chapter asks what AP can borrow from BT
normative ethics.
39
Mulgan, The Demands of Consequentialism, ch. 10; Mulgan, Future People, ch. 11.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
13.7 BT Consequentialism
We begin with BT consequentialism. This may seem a strange starting point.
Contemporary BT ethics is (almost universally) anti-consequentialist. Indeed, BTs
are often consequentialism’s most strident opponents. However, BT and consequen-
tialism are not necessarily diametrically opposed. Historically, some very influential
utilitarians endorsed BT. Consider just two examples. The most popular ethics
textbook in English universities in the early nineteenth century was William Paley’s
The Principles of Moral and Political Philosophy.40 Paley defends a theological
utilitarianism. God wants his human creatures to flourish, and therefore God insti-
tutes moral rules that best promote human well-being. Divine command grounds
rule consequentialism. My second example is R. M. Hare, one of the most prominent
philosophical utilitarians of the mid-twentieth century.41 Hare’s prescriptivism looks
like a model of atheist meta-ethics. But his preference utilitarianism owes much to his
own Christian commitments, and is perhaps best interpreted as a form of BT
consequentialism.42
AP cannot borrow either Paley or Hare wholesale. (Aside from BT itself, each has
other commitments we have already rejected—notably Paley’s pre-Darwinian biol-
ogy and Hare’s prescriptivist meta-ethic.) But both offer resources that are unavail-
able to atheist consequentialism. Paley offers a connection between consequentialism
and divine command. If AP can borrow the latter, it can then use it to support the
former. (We return to divine command in section 13.8.) And Hare’s notion of the
‘archangel’ who impartially takes on all preferences could provide a useful model for
the ethical role of superior (cosmically significant) rational beings, alongside Hume’s
‘ideal observer’ and other artificial moral role models. (As I argued in chapter 7, the
discovery of actual role models who respond to real cosmic values may provide new
impetus to ideal agent or observer theories such as Hare’s and Hume’s.)
Much BT opposition to consequentialism stems from two features of BT that AP
explicitly rejects: revelation and theodicy. We briefly consider each in turn. BT ethics
often rests on revelation, especially interpretations of scripture. AP obviously cannot
borrow this moral method, because AP rejects all revelation. However, this also leaves
AP better able to defend consequentialism. BT opposition to consequentialism is often
doctrinal: specific religious texts and traditions contain many non-consequentialist
moral injunctions. Such doctrinal objections will not worry AP.
Our second source of BT opposition to consequentialism is theodicy. As we saw
in chapter 8, one way to defend God’s decision to create this world is to deny that
God must create the best possible world. Many theodicies are explicitly non-
consequentialist. Because it needs no theodicy, AP thus removes another barrier
40
On Paley’s influence, see Schneewind, Sidgwick’s Ethics and Victorian Moral Philosophy, pp. 122–9.
41
See, e.g., Hare, Moral Thinking.
42
This interpretation is defended by John Hare in God and Morality and God’s Call.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
to BT consequentialism. Once we recognize that AP’s God is not interested in

human suffering, we can take more seriously the possibility that by creating this
world God has maximized value.43
AP can thus take elements of BT ethics in a more consequentialist direction. In
particular, as we saw in chapters 2 and 12, AP borrows the BT idea that value is
resemblance to God. AP could then extend this idea to cover morality as a whole.
Right action is whatever best resembles God’s creative act. (As ever, AP distinguishes
itself from BT both by endorsing different values and by widening the gap between
divine and human responses to value.)
This completes our discussion of what AP can borrow from consequentialism,
whether atheist or BT. We can reasonably conclude that BP does mutually support a
distinctive consequentialist moral theory. Our next question is whether AP can
borrow from non-consequentialist BT ethics. Most BT ethics is not explicitly conse-
quentialist. But BT ethics often is value-based and moral realist. AP’s broader moral
agenda thus has more in common with non-consequentialist BT ethics than with
much atheist ethics. These synergies are not surprising. After all, AP’s cosmic values
are modelled on BT, and AP meta-ethics mirrors BT’s division between intellectual-
ism and voluntarism. The rest of this chapter explores some surprising places where
AP can borrow from non-consequentialist BT ethics. Our first example is perhaps the
most surprising. In the next section, I argue that AP can borrow BT divine command
ethics.
13.8 Divine Command Ethics

One very popular foundation for BT ethics is divine command theory. This may
seem beyond our reach. Surely AP denies that God issues divine commands to
human beings? However, I shall argue that AP can borrow from this BT tradition.
Following Paley’s lead, we can base morality on God’s call to rational creatures.
No AP God would issue commands to human beings. But a non-human-centred
creator may well want some other rational beings to respond to cosmic value. Such a
God might call all rational beings to respond to cosmic values as they are able. We are
not the reason why this call is issued, but we can still respond to it.
To illustrate how AP could borrow from divine command ethics, I focus on one of
the founders of voluntarism: Duns Scotus.44 My account of Scotus’s philosophy is
43
As we saw in chapters 7 and 8, BT theodicy makes several distinct non-consequentialist moves—from
Adams’s argument that a gracious God need not create the best, to Swinburne’s argument that there is no
best world if we value human lives, to Plantinga’s emphasis on the significance of incompatibilist freedom.
By removing the need for theodicy, AP removes one primary BT motivation for all these other departures
from consequentialism.
44
My reading of Scotus draws especially on Cross, Duns Scotus; Kent, ‘The Moral Life’; Williams, ‘From
Meta-Ethics to Action Theory’; Kent, ‘Rethinking Moral Dispositions’; Hare, God’s Call, pp. 50–70; Hare,
God and Morality, pp. 89–114.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
selective and idiosyncratic. A full exploration of Scotus’s moral theory would take us
too far afield. But his basic ideas are very suggestive for AP, especially as we move
beyond consequentialism.
Scotus was arguably the most influential philosopher in the late medieval period.
His ethical ideas flowed to Kant via Crusius; and atheist echoes can be found today in
existentialism and non-cognitivism. Among contemporary moral philosophers,
however, Scotus is largely forgotten, eclipsed by his Benedictine nemesis Aquinas.
This is unfortunate. Scotus offers a window into an earlier, more Platonic mode of
ethical thinking, thus offering continuity with Anselm and the Neoplatonists. Given
the affinities between AP and Platonism, which we explored in chapters 3 and 6,
Scotus is an intriguing ethical source for AP.
The centrality of divine will shapes Scotus’s ethic. God’s most morally significant
feature is the divine will, not the divine intellect. Scotus is typically seen today as a
paradigm divine command theorist. But this is misleading. Scotus’s ethical theory is
more nuanced. Scotus’s discussion of the moral law focuses on the Ten Command-
ments. Following Christian tradition, he divides these into two tables. The first four
commandments govern our direct obligations to God and the other six govern our
relations with one another. The two tables have very different foundations. The first
four commandments are necessary in a way that the other six are not. Given God’s
perfection, God is necessarily to be loved above all else by every rational being. God
necessarily loves God, and God necessarily commands all created rational beings to
love God. Although our obligation to love God is based on divine command, its
ultimate origin lies in the prior moral perfection of the divine nature.
While the first table of the Decalogue is necessary, the second table is contingent in
the following sense: God could have issued quite different commands. Scotus’s
favourite example is theft. Stealing is wrong because God has forbidden it. But God
could have omitted this rule. The prohibition on theft presupposes property. And
human beings could live perfectly well together without anyone owning anything.45
God could have instituted moral rules that never mentioned property. Our obligation
not to steal depends entirely on divine command.
God’s will is thus arbitrary in the sense that nothing else determines it. But Scotus’s
God is not some capricious despot whose commands are random, unmotivated, or
without reason. God’s will for us is directed at our true end. God is under no necessity
to create at all, and certainly under no necessity to create creatures like us. However,
once God has created creatures like us, God’s perfect justice requires that God ‘create
us so that union with God is our final end’.46 Having created us, God will necessarily
issue some commands that lead us to that end. There is no unique set of essential
commands, because there is no single path to our final end. There are many possible
45
The historical background to Scotus’s discussion of theft is the Franciscan dispute with the papacy
over the poverty of Christ and the related question whether monks could own property.
46
Hare, God’s Call, p. 66.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
paths and many possible commandments. However, moral commands that are not
feasible paths to human fulfilment are not possible divine commandments. God’s
freedom is thus still constrained by God’s moral perfection. In contemporary terms,
Scotus is thus closer to Robert Adams’s modified divine command theory than to
pure voluntarism.47
Scotus’s divine command ethic seems an unpromising model for AP. By defin-
ition, AP has no God who issues commands to human beings or has any plans for us.
I will argue, nonetheless, that AP can borrow from Scotus.
The centrality of God in Scotus’s ethic certainly makes it very difficult to construct
atheist analogues, unless we are prepared to abandon moral realism and resort to
existentialism or non-cognitivism. This is one reason why Scotus is less popular
today than Aristotelians or natural law theorists, whose ideas translate more easily
into a secular idiom.
However, its very theocentrism makes Scotus’s ethic a good model for AP. AP
easily adapts Scotus’s account of the first four commandments. Even without any
command issued to us, God’s nature gives us a reason to love God. More controver-
sially, AP could also borrow Scotus’s account of the role of divine command here.
Following John Hare, we might replace divine command with the looser notion of
God’s loving call to all created beings. Both the cosmos and its creator are to be loved
or appreciated. This moral fact is built into the fabric of the cosmos in a way that
other creatures can read clearly, and we can only glimpse. As we saw in chapters 5
and 7, mystical experience might offer us insight into God’s call to other rational
beings. Instead of the definite human-directed revelations that Scotus himself
believed in, AP argues that human mystics dimly perceive a divine call that is not
really aimed at us.
AP can borrow divine commands based on necessary divine properties. It is
obviously much harder to borrow Scotus’s account of the second table of the
Decalogue, where God’s commands are contingent and arbitrary. (Without actual
commands, we have no moral guidance.) But even here AP may be able to borrow
something. Here are two possibilities. First, we might limit ourselves to negative
knowledge, ruling out moral rules that could not be suitable divine commands
because they would not fit our true end. This would yield a much more pluralist
morality than anything Scotus could countenance. But it would fit into section 13.5’s
discussion of pluralism within APRC. That earlier discussion also suggests a more
ambitious adaptation of Scotus. Recall the possibility of a divine ideal code that issues
different advice to different rational beings, or perhaps a code directed only at
superior beings. If the universe contains superior (cosmically significant) beings,
then God will issue commands to them that are designed to fulfil their (divinely
47
Adams, ‘A Modified Divine Command Theory of Ethical Wrongness’. On the medieval scene, the
more radical position was taken in the next generation, when Ockham notoriously argued that God could
command that God not be worshipped.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
relevant) end. Our explorations of cosmic purpose, our mystical practices, and even
our moral intuitions could then be interpreted as insights into that cosmic code.
Several other distinctive features of Scotus’s ethics are also congenial to AP. One is
his account of our final end. ‘What makes Scotus’s ethics so distinctive is that he
thinks natural happiness has nothing at all to do with morality.’48 Our aim, as John
Hare puts it, is to become co-lovers with God—to bring our will for ourselves into
line with God’s will for us. (Or, more precisely, into line with what God wills for our
wills.) This end transcends both our physical nature and our natural happiness.
Scotus thus belongs to the Platonic tradition where the soul’s true home lies in the
world of Forms, rather than the Aristotelian tradition where eudaimonia can be
found in this world. Scotus here follows Anselm, for whom: ‘To fulfil God's purpose
for us is to preserve our proper place in the order of the universe.’49
AP can borrow Scotus’s account of human flourishing in two ways. Perhaps this is
what constitutes the flourishing of cosmically significant superior beings, and there-
fore what grounds God’s actual commands to them. Alternatively, perhaps our own
final end is a suitably asymmetrical relation with God, as explored in chapter 12.
(Perhaps contemplation of God is essential to flourishing human lives.) Either way,
Scotus’s divine commands aimed at an otherworldly Platonic end provide a suitable
supernaturalist alternative to currently fashionable naturalistic stories about human
flourishing.50
Scotus himself is no consequentialist. In particular, he emphasizes motives, not
results. Like most medieval philosophers, Scotus was a monk whose primary audi-
ence was other monks. Their shared social world left little scope for individual choice
of actions, but much room for reflection on one’s real motives. (Do I pray, cook, or
serve others out of self-interest or from genuine love for others or God?) The
monastic routine naturally creates a preoccupation with the inner life. This preoccu-
pation with motives seems to leave Scotus’s ethic ill-suited to modern life, and may
partly explain his eclipse in contemporary secular ethics. However, in the broader
context of AP’s divine ideal code, a Scotian ethic of motives might emerge as
eminently suitable for creatures who cannot hope to make a cosmically significant
impact. Perhaps, given our inability to make a difference that matters, our moral
theory should take a more quietist form. (We explore one BT-inspired possibility in
section 13.10.)
Scotus also offers an alternative model for hybrid moral theories. This comes from
his picture of morality as a clash between two competing ‘affections’. Scotus borrows
48
Williams, ‘From Meta-Ethics to Action Theory’, p. 338.
49
Visser and Williams, Anselm, p. 202.
50
A further complication is that, for doctrinal reasons, Scotus himself must insist that true happiness is
simply unavailable in this life and can only be achieved in heaven as a gift of divine grace. True happiness
would thus be unavailable without both a BT God and a personal afterlife. We addressed a version of this
particular Scotian position in more general terms in our discussion of the meaning argument in section
10.3.
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
Anselm’s distinctive model of human motivation. Human beings have two main
inclinations: the affection for advantage and the affection for justice. One defining
feature of Scotus’s moral philosophy is his insistence on radical human freedom. For
both Anselm and Scotus, human freedom requires two competing inclinations.
Creatures with only one inclination always act mechanically. They have no will
and their ‘acts’ lack moral significance. Our human will is a power to choose between
advantage and justice.51
Scotus’s first inclination is directed at our own (individual) happiness. The inclin-
ation to justice is harder to pin down. It has been variously characterized as follows:
‘[its] primary object . . . is God’; ‘human beings must discern their place in the divine
order and choose to maintain it’; it ‘inclines the will to act in accordance with the
moral law irrespective of its connection with our happiness’; and it ‘is directed
towards what is good in itself, regardless of its relation to us’.52 Scotus’s ‘justice’ is
thus broader than our contemporary notion. An alternative label might be ‘charity’—
another term with very broad meaning for Scotus.
For Scotus, the inclination for justice covers all moral obligations, whether to God
or to other human beings. AP might instead treat all human affairs as aspects of
advantage and reserve ‘justice’ for our response to the divine transcendent good. Or,
as with Scheffler’s original hybrid view in section 13.6, we could imagine three
distinct inclinations: advantage, human justice, and cosmic justice. (For Scotus, the
BT God unifies the two justice inclinations. For AP, they come apart.)
Scotus’s view combines moral psychology, metaphysics, and moral philosophy.
His two affections are meant to be actual human inclinations, not merely theoretical
constructions. To fully borrow from Scotus, AP must therefore claim that we really
do possess these two (or three) motivations. This claim is not wholly implausible.
Both the affection for advantage and the affection for human justice are susceptible to
evolutionary explanation. But many human beings also possess an affection for
cosmic justice: a desire to bring our lives into line with what ultimately matters. AP
cannot borrow Scotus’s own explanation of the origin of this self-transcending
affection. (It is not deliberately implanted by God to lead us to our final end of
union with God.) But neither will any evolutionary explanation suffice. (As we saw in
chapter 5, the atheist finds it very difficult to offer plausible explanations of any
mystical experiences or aspirations.) AP instead treats our inclination to cosmic
justice as a by-product of some other divine project—a not so unlikely accident in
a cosmos willed into being by a perfect creator.
I conclude that Scotus’s ethic is a fertile source of new resources for AP morality.
This illustrates both AP’s ability to borrow from unpromising BT theories such as
51
This debate over human freedom is the site of the original controversy between intellectualism and
voluntarism (see, e.g., Kent, ‘The Moral Life’, pp. 235–42).
52
Hare, God and Morality, p. 254; Visser and Williams, Anselm, p. 193 commenting on Anselm’s
original distinction; Cross, Duns Scotus, p. 87; and Hare, God and Morality, p. 91, respectively.
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
divine command ethics and also its ability to breathe new life into currently neglected
philosophical traditions.
13.9 Divine Motivation Theory

Our next BT normative ethic is Linda Zagzebski’s divine motivation theory (DMT).53
Zagzebski combines meta-ethics and normative ethics. Our interest here is in the
latter, although the two are intertwined. Section 13.9.1 outlines Zagzebski’s DMT,
and compares it to its secular competitors, before section 13.9.2 asks what AP can
borrow from Zagzebski.
Zagzebski explicitly denies that a BT ethic should be based on divine law or
command, as our epigraph shows. While AP can borrow from divine command
theory, it clearly finds the rival tradition of attraction more congenial. An AP God (or
other divine reality) who does not care for us could still attract, even if the absence of
direct divine commands removes any real possibility of compulsion or call. Zagzebs-
ki’s alternative BT ethic is thus a promising model for AP.
Within BT ethics, Zagzebski is closest to Adams’s resemblance theory, which we
first met in chapter 2.54 Adams’s own normative ethic, however, mixes a divine
resemblance theory of excellence with a (modified) divine command account of
human rightness and wrongness. Zagzebski thus offers a better example of a purely
resemblance-based BT normative ethic.
13.9.1 What is divine motivation theory?

Zagzebski begins with the central moral importance of exemplars. Our moral learn-
ing begins by seeking to imitate exemplary persons. What most matters about these
exemplary individuals is their motivation. Good acts and desirable outcomes are
secondary. The primary focus of ethics is the motivation of the virtuous person.
Zagzebski argues that DMT best captures the perennial Christian ideal of the
imitation of God. ‘God is the paradigmatically good person . . . Human motives are
good insofar as they imitate the motives of God.’55 Indeed, Zagzebski’s God plays a
broader moral role. ‘All moral properties are ultimately grounded in the emotional
experience of God.’56
If God is the standard of perfect motivation, then we need information about
God’s motives. Zagzebski cites two sources: the divine act of creation and the
Incarnation. Motives are revealed most clearly in action. God’s most significant act
is the creation of the physical universe. As we saw in chapter 3, divine creation raises
53
Zagzebski, Divine Motivation Theory.
54
Adams, Finite and Infinite Goods, especially ch. 1. While Zagzebski distinguishes her DMT from
Adams’s resemblance theory, the precise details of their dispute need not concern us here.
55
Zagzebski, Divine Motivation Theory, p. 213.
56
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
many theological puzzles. If God is perfect and self-sufficient, what possible motive
could God have to create anything? Alternatively, if God creates because the world is
good, was God ever really free not to create? Zagzebski replies that God’s creativity is
part of God’s nature. God cannot create nothing. God’s nature is to create. But God
still chooses what to create. While it flows from God’s nature, God’s creative choice is
still free. Creation is part of God’s nature because it arises from God’s motives,
especially love. God can only manifest love by creating beings to love. In particular,
because God is a person, God wants to create other persons to love.
Zagzebski explicitly rejects consequentialism. The goodness of motives does not
derive from the goodness of the outcomes they produce or intend. Rather, the value
of outcomes depends on the metaphysically prior value of motives: ‘the outcome of
the Creation, the created universe, is good because it is the end of God’s motive in the
Creation.’57 God’s creation is not directed at some end that is independently desir-
able. ‘God did not create because he was trying to accomplish something.’58
Our moral task is to imitate God’s motives. ‘We imitate God by loving what God
loves.’59 And we learn what God loves by seeing what God has actually created. To
discover how we can best imitate God, we must examine the created universe.
Unfortunately, even the most thorough study of creation can never reveal the divine
motives in much detail. God’s creative act is not a complete model for human virtue.
We need an actual historical exemplar who meets three necessary conditions: (i) the
exemplar must be a person; (ii) we must have ‘vivid descriptions’ of their actions and
motives; and (iii) the exemplary person must stand in ‘a special relation . . . to God’.60
In Christianity, the person who meets all three criteria is Christ. As a person who is
identical to God, Christ obviously meets the first and third criteria. Furthermore, in
the New Testament, we possess the vivid descriptions that DMT requires.
This brings us to Zagzebski’s second source of moral knowledge: the Incarna-
tion. That God became fully human and dwelt among us is, of course, a central
doctrinal tenet of Christianity. But Zagzebski argues that its moral significance is
often underappreciated in Christian philosophical ethics. For Zagzebski, the Incar-
nation is ‘a central event in moral understanding’.61 Only by becoming a historical
person and living an exemplary human life can God show us how to imitate the
divine motivations in a human way. ‘The imitatio Dei is made possible by the
imitatio Christi.’62
57
58
59
60
Zagzebski, Divine Motivation Theory, pp. 239–40.
61
Zagzebski, Divine Motivation Theory, p. 231. Zagzebski thus sides with those, including Scotus, who
argue that ‘God would have become incarnate even if Adam had not sinned’ (Zagzebski, Divine Motivation
Theory, p. 232). God’s plan for creation, and especially for created humanity, always involved incarnation.
62
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
Divine motivation theory differs from secular analogues because God provides the
metaphysical ground for our human exemplars. Their goodness consists in their
resemblance to the divine person, and their motives are virtues because they resemble
God’s motives. Without God, we have nothing to anchor the human exemplars.
DMT is thus less arbitrary than secular alternatives. It is also more realist. ‘The
anchoring of good emotion in the emotions of God makes divine motivation theory
more clearly realist than motivation-based virtue theory would be without a supreme
exemplar.’63
Zagzebski thus follows a familiar BT critique of any naturalistic ethic, one that we
have endorsed many times before, especially in chapter 2. This is why we need not
consider atheist virtue ethics in detail. If we are interested in developing an AP virtue
ethic, then DMT is a much better model.
Given our earlier focus on atheist consequentialism, we should also compare DMT
to BT consequentialism. DMT and act consequentialism offer identical moral advice
if the human response that best resembles God’s motivations is to maximize the
good. This is especially likely if God’s response to cosmic value is itself solely
consequentialist. Zagzebski denies consequentialism at both levels. Her God is not
a pure consequentialist and her human virtues are not limited to consequentialism.
We have already seen that Zagzebski reverses the consequentialist priority between
outcomes and motivations. This implies that God’s motives cannot all be cashed out
in terms of the promotion of independent values. Zagzebski also rejects utilitarian
aggregations of individual well-being. Each person is an ‘incommunicable’ individ-
ual.64 Zagzebski argues that, while the incommunicability of persons may not be
explicit in earlier Christian writers, or in contemporary moral philosophy, it captures
something that is central both to our everyday moral experience and to the Christian
tradition. ‘When we say that persons are irreplaceable, what we mean is that when a
person is lost to the world, something of irreplaceable value is gone. It is not the
instance of a quality that is lost, but something else. I am using the term “incommu-
nicability” to refer to whatever that is.’65
Zagzebski supports her rejection of consequentialism by noting that the behaviour
of actual exemplars (notably Christ) cannot plausibly be described as the maximiza-
tion of any impersonal value. If these exemplars are the standards of human moral
excellence, then morality is not purely consequentialist.
13.9.2 Can AP borrow divine motivation theory?

AP can borrow some aspects of DMT. As we saw earlier, DMT is easier to borrow
than divine command theory. We can seek to imitate an AP God who issues no direct
63
64
Zagzebski borrows the term ‘incommunicability’ from John Crosby, who in turn ‘adapts it from an
aphorism of Roman law’ (Zagzebski, Divine Motivation Theory, p. 195).
65
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
commands to human beings. Zagzebski’s departures from consequentialism need not

bother AP. AP can agree that God’s motives are not consequentialist. AP is not
committed to a consequentialist God. God’s may have non-consequentialist motiv-
ations. If God is perfect, then presumably God has all good motivations, and
therefore God’s responses to cosmic value will include all good responses. AP can
also borrow Zagzebski’s reversal of outcomes and motives. Indeed, this reversal
might assist the cosmological and fine-tuning arguments of chapters 3 and 4. If the
ultimate explanation for the existence of this universe lies in God’s motives, then it is
much easier to conclude that there is a genuine cosmic purpose behind the universe.
DMT is distinctive because it offers a non-consequentialist foundation for ethics,
where our non-consequentialist responses are grounded in a divine response that is
itself non-consequentialist. As we saw in section 13.5, AP rule consequentialism
already concedes that our best response to God’s motivations is not promotion.
Indeed, AP could make DMT even more pluralistic. For Zagzebski, all human virtues
are (ultimately) grounded in the unified motivations of a single God. AP can borrow
this theist formulation. But it need not. Perhaps each human virtue is grounded in a
distinct ultimate cosmic value. (This is quite plausible if we admit a plurality of
cosmic values.) Different cosmic values might naturally call forth different
responses—some are to be promoted, others are more appropriately objects of
contemplation, and so on.
Finally, AP supports Zagzebski’s use of God’s creation to discern God’s motives.
(We have done the same throughout this book!) Of course, AP’s exploration delivers
a different, non-human-centred set of divine motivations. AP’s substantive list of
divine motivations will differ from Zagzebski’s. But her method is very congenial to
AP. Divine motives provide a solid bridge between the facts we invoke God to
explain, on the one hand, and the notions of cosmic purpose and cosmic value, on
the other.
Unfortunately, two aspects of DMT are harder to borrow. One is Zagzebski’s
insistence on the incommunicable value of human persons. AP could, of course,
simply deny that God is an incommunicable person. (Zagzebski’s analysis of per-
sonhood is controversial, and AP need not concede that God is a person at all.) But
AP can concede, if only for the sake of argument, that God is an incommunicable
person of unique and irreplaceable value. Whether AP can endorse the moral
significance Zagzebski attaches to the incommunicability of human persons is
another matter.
I agree that incommunicability is a fruitful interpretation of one integral part of
our common-sense view of the value of human persons. In principle, AP can
accommodate that value. Perhaps the AP account of human well-being defended
in chapter 12 will preserve this aspect of human personhood. This might complicate
AP morality and push it in a non-consequentialist direction. But that need not be an
insuperable barrier. Alternatively, AP could agree that cosmically significant persons
are incommunicable, but deny that humans are. As so often before, AP rejects BT’s
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
human-centred claims as self-aggrandizing caprice. An incommunicable personal

God will seek to create other uniquely valuable persons. And perhaps God has
created such persons. But we are not them. While the best morality for aliens
preserves their incommunicability, the appropriate morality for humans will be
resolutely consequentialist.
A second aspect of DMT that AP cannot borrow is Zagzebski’s use of the
Incarnation. No AP God would have any reason to become a human being. No
human exemplar is identical to God. AP must reject this key element of Zagzebski’s
moral epistemology. This seems to leave AP entirely reliant on divine creation, which
in turn seems unlikely to provide sufficient content for any recognizable human
morality. Fortunately, things are not so simple. While identity is the best way to get a
human exemplar who mirrors the divine motivation, Zagzebski herself admits that
the ‘special relation to God’ could be resemblance rather than identity. And other
special relations are also possible, such as experience, encounter, or knowledge of
God. DMT need not insist that Christ is the only human exemplar. There is no single
unique way to live a good human life.66
One natural place for AP to seek alternative human exemplars is within established
mystical doxastic practices. Perhaps mysticism provides insight into the divine
motivation, which then transforms the individual human being so that her motiv-
ations come to resemble those of the divine person. Indeed, perhaps this is how
knowledge of divine motivation—and hence of moral truth—first enters the human
world. This would explain the role of moral tests in mystical traditions, which played
a key role in chapter 5.
Zagzebski acknowledges that human exemplars come from a variety of religious
traditions. She lists Socrates and Buddha, along with Jesus, as non-negotiable exem-
plars. (Any moral theory that fails to recognize these three individuals as exemplars
of virtue is prima facie unacceptable.) In Christian divine motivation theory, one
particular human exemplar has a privileged role: namely, the incarnate Christ. AP
denies the Incarnation. And, as we saw in chapter 9, AP insists on epistemic parity
between different religions. So AP will treat the exemplars of different traditions
equally. But AP can still discriminate between different groups of exemplars across
traditions. In particular, those mystical exemplars who owe their resemblance to God
to their own experiential connection to supernatural moral reality may still play a key
epistemic role. They alone bring true moral knowledge into the human realm. (And,
in practice, it may turn out that the highest humanly possible degree of resemblance
to God can only arise from direct divine experience. Perhaps only those who see God
can ever resemble God.)
AP can thus borrow DMT, even though the result may be a less certain, more
pluralistic, and more unsettling ethic than Zagzebski herself defends. As with Scotus,
66
ANANTHROPOCENTRIC PURPOSIVIST MORAL THEORY 
AP can borrow from, and transform, even those BT ethical traditions that seem to
rely most heavily on humanity’s special relationship with God.
13.10 An Ethic of Pure Contemplation

Both DMT and rule consequentialism are likely to include the contemplation of
cosmic value as one option. This is a prima facie appropriate response to cosmic
values, and it is something human beings can do. In their different ways, all the
participants in chapter 11’s imaginary dialogue engaged in contemplation, study,
observation, or exploration of the cosmic purpose and the realm of objective value
it reveals.
A more austere alternative is an AP ethic that consists entirely of contemplation.
As we saw in section 13.5, this is one possible implication of APRC. Perhaps the ideal
code says: ‘When you cannot promote, contemplate.’ For superior rational beings,
this rule only applies on some occasions. Most of the time, they can promote value.
For ineffectual human agents, this rule always applies. Because we can never do
anything else, ‘contemplate cosmic value’ is the only thing the ideal code ever tells
us to do.
DMT can also recommend pure contemplation. Contemplation is a human virtue
if it mirrors one of God’s own responses to cosmic value. Contemplation is the only
human virtue if it is our only response to cosmic values that comes close enough to
any of God’s responses. DMT is more likely than APRC to deliver an ethic of pure
contemplation, because contemplation almost certainly is one of God’s responses to
cosmic values. Without something akin to contemplation, why would God bother to
create a good world?
The ethic of pure contemplation is not new. Similar thoughts are found in many
mystical traditions; in the more theocentric forms of Christianity; and even in our
own academic professional ethics, where the disinterested pursuit of knowledge often
shades into contemplation. Any BT ethic of contemplation is tempered by God’s
command that humans aid one another. In part III, we have explored various ways
that AP might ground its own inter-human obligations. Unlike BT, however, AP can
simply dispense with human-sized ethics entirely. BT cannot easily reduce ethics to
the contemplation of non-human cosmic values. AP can.
Chapter 12 explored the possibility that contemplating cosmic values enhances
one’s individual well-being. This suggests yet another route to pure contemplation.
The egoist who seeks only her own good might realize that she can only bring
objective value into her life by studying the cosmic purpose. Contemplation alone
promotes prudential value. Similarly, the altruist might seek to bring this objective
good to others. (Her personal ethic thus includes the promotion of contemplation
alongside contemplation itself.)
But perhaps nothing, not even contemplation, makes human lives go well. Our lives
lack objective value. (This is why we cannot promote value.) But not all reasons are
 ANANTHROPOCENTRIC PURPOSIVIST MORALITY
prudential, and some reasons are unrelated to anyone’s well-being. If moral supernat-
uralism is correct, and cosmic purpose is the only source of genuine external reasons,
then pure contemplation is the only thing we have any reason to do. We naturally feel
that something of great value is lost when a human life is entirely devoted to the
contemplation of the cosmic purpose to the exclusion of all human activities. But
perhaps the ultimate lesson of AP, as the most austere mystics down the ages have
always taught, is that what we have lost was never really valuable after all.
Bibliography
Adams, M., ‘Anselm on Faith and Reason’, in B. Davies and B. Leftow (eds), The Cambridge
Companion to Anselm (Cambridge: Cambridge University Press, 2004), 61–83.
Adams, M., Horrendous Evils and the Goodness of God (Ithaca, NY: Cornell University Press,
1999).
Adams, R. M., ‘Anti-Consequentialism and the Transcendence of the Good’, Philosophy and
Phenomenological Research, 67 (2003), 114–32.
Adams, R. M., Finite and Infinite Goods: A Framework for Ethics (Oxford: Oxford University
Press, 2002).
Adams, R. M., Leibniz: Determinist, Theist, Idealist (Oxford: Oxford University Press, 1994).
Adams, R. M., ‘A Modified Divine Command Theory of Ethical Wrongness’, in G. Outka and
J. R. Reeder (eds), Religion and Morality (Garden City: Anchor Press, 1975), 318–47.
Adams, R. M., ‘Must God Create the Best?’, Philosophical Review, 81 (1972), 317–32.
Albright, C. R., ‘Neuroscience in Pursuit of the Holy: Mysticism, the Brain and Ultimate
Reality’, Zygon, 36 (2001), 485–92.
Almeder, R., Death and Personal Survival: The Evidence for Life after Death (Lanham, MD:
Rowman & Littlefield Publishers, 1992).
Almeder, R., ‘On Reincarnation: A Reply to Hales’, Philosophia, 28 (2001), 347–58.
Almeida, M., and Judisch, N., ‘A New Cosmological Argument Undone’, International Journal
for Philosophy of Religion, 51 (2002), 55–64.
Almeida, M., and Oppy, G., ‘Sceptical Theism and Evidential Arguments from Evil’, Austra-
lasian Journal of Philosophy, 81 (2003), 496–516.
Alston, W., Beyond ‘Justification’: Dimensions of Epistemic Evaluation (Ithaca, NY: Cornell
University Press, 2005).
Alston, W., Perceiving God (Cornell: Cornell University Press, 1991).
Alston, W., ‘Religious Language’, in W. Wainwright (ed.), The Oxford Handbook of the
Philosophy of Religion (Oxford: Oxford University Press, 2005), 220–44.
Alston, W., ‘Response to Bonjour’, in H. D. Battaly and M. P. Lynch (eds), Perspectives on the
Philosophy of William P. Alston (Lanham, MD: Rowman & Littlefield, 2005), 85–94.
Alston, W., ‘Response to Quinn’, in H. D. Battaly and M. P. Lynch (eds), Perspectives on the
Philosophy of William P. Alston (Lanham, MD: Rowman & Littlefield, 2005), 167–70.
Alston, W., et al., ‘Symposium on Perceiving God’, Philosophy and Phenomenological Research,
54 (1994), 863–99.
Armstrong, K., The Great Transformation (London: Atlantic Books, 2007).
Attfield, R., Environmental Ethics (Cambridge: Polity, 2014).
Barth, K., Anselm: Fides Quarens Intellectum (Eugene, OR: Pickwick Publications, 1975).
Barth, K., Evangelical Theology: An Introduction (Grand Rapids: William Eerdmans, 1963).
Battaly, H. D., and Lynch, M. P. (eds), Perspectives on the Philosophy of William P. Alston
(Lanham, MD: Rowman & Littlefield, 2005).
Bergmann, M., ‘Sceptical Theism and Rowe’s New Evidential Argument from Evil’, Nous, 35
(2001), 278–96.
 BIBLIOGRAPHY
Bergmann, M., and Rea, M., ‘In Defence of Sceptical Theism: A Reply to Almeida and Oppy’,
Australasian Journal of Philosophy, 83 (2005), 241–51.
Bishop, J., Believing by Faith (Oxford: Oxford University Press, 2007).
Bishop, J., ‘Can There Be Alternative Concepts of God?’, Nous, 32 (1998), 174–88.
Bishop, J., and Perszyk, K., ‘The Normatively Relativized Logical Argument from Evil’,
International Journal for Philosophy of Religion, 70 (2011), 109–26.
Bonjour, L., ‘Sense Perception, Epistemic Practices, and Skepticism’, in H. D. Battaly and
M. P. Lynch (eds), Perspectives on the Philosophy of William P. Alston (Lanham, MD:
Rowman & Littlefield, 2005).
Bosanquet, B., The Value and Destiny of the Individual (London: Macmillan, 1923).
Boyd, R., ‘Finite Beings, Finite Goods: The Semantics, Metaphysics and Ethics of Naturalist
Consequentialism, Part I’, Philosophy and Phenomenological Research, 66 (2003), 505–53.
Boyd, R., ‘Finite Beings, Finite Goods: The Semantics, Metaphysics and Ethics of Naturalist
Consequentialism, Part II’, Philosophy and Phenomenological Research, 67 (2003), 24–47.
Boyd, R., ‘How to be a Moral Realist’, in G. Sayre-McCord (ed.), Moral Realism (Ithaca:
Cornell University Press, 1988), 181–228.
Bradley, D., ‘Multiple Universes and Observation Selection Effects’, American Philosophical
Quarterly, 46 (2009), 61–72.
Brakke, D., Demons and the Making of the Monk (Cambridge, MA: Harvard University Press,
2006).
Broad, C. D., The Mind and its Place in Nature (London: K. Paul, 1925).
Broadie, S., ‘Agency and Determinism in A Metaphysics for Freedom’, Inquiry, 56 (2013), 571–82.
Broome, J., Weighing Lives (Oxford: Oxford University Press, 2004).
Camus, A., The Myth of Sisyphus, trans. J. O’Brien (New York: Penguin Books, 2005).
Carlson, E., and Olsson, E., ‘Is Our Existence in Need of Further Explanation?’, Inquiry, 41
(1998), 255–75.
Carr, B., and Ellis, G., ‘Universe or Multi-Verse?’, Astronomy and Geophysics, 49 (2008),
2.29–2.33.
Chadwick, O., ‘Introduction’, in J. Cassian, Conferences, translated by C. Luibheid (New York:
Paulist Press, 1985).
Chang, R., Incommensurability, Incomparability, and Practical Reason (Cambridge, MA:
Harvard University Press, 1997).
Chappell, T., ‘Infinity Goes up on Trial: Must Immortality be Meaningless?’, European Journal
of Philosophy, 17 (2009), 30–44.
Collins, R., ‘Evidence for Fine-Tuning’, in N. Manson (ed.), God and Design (London:
Routledge, 2003), 178–99.
Collins, R., ‘Fine-Tuning Arguments and the Problem of the Comparison Range’, Philosophia
Christi, 7 (2005), 385–404.
Comte, A., System of Positive Polity (London: Longmans, Green and Co., 1875–7).
Copenhaver, B., and Schmitt, C., Renaissance Philosophy (Oxford: Oxford University Press,
1992).
Coppleston, F., A History of Medieval Philosophy (South Bend, IN: University of Notre Dame
Press, 1990).
Cowen, T., and Parfit, D., ‘Against the Social Discount Rate’, in P. Laslett and J. Fishkin (eds),
Justice between Age Groups and Generations (New Haven: Yale University Press, 1992), 144–61.
BIBLIOGRAPHY 
Craig, W. L., ‘Theism and the Origin of the Universe’, Erkenntnis, 48 (1998), 47–57.
Cross, R., Duns Scotus (Oxford: Oxford University Press, 1999).
Daniels, N., Justice and Justification: Reflective Equilibrium in Theory and Practice (Cambridge:
Cambridge University Press, 1996).
Danto, A., Mysticism and Morality (New York: Basic Books, 1972).
d’Aquili, E., and Newberg, A., The Mystical Mind: Probing the Biology of Religious Experience
(Minneapolis: Fortress Press, 1999).
Davey, K., and Clifton, R., ‘Insufficient Reason in the “New Cosmological Argument” ’,
Religious Studies, 37 (2001), 485–90.
Davies., P., ‘The Appearance of Design in Physics and Cosmology’, in N. Manson (ed.), God
and Design (Routledge, 2003), 147–54.
Davies, P., The Goldilocks Enigma (London: Allen Lane, 2006).
Davis, S. T., Risen Indeed: Making Sense of the Resurrection (Grand Rapids: William Eerdmans,
1993).
de Lazari-Radek, K., and Singer, P., The Point of View of the Universe: Sidgwick and Contem-
porary Ethics (Oxford: Oxford University Press, 2014).
Dilley, F., ‘A Finite God Reconsidered’, International Journal for Philosophy of Religion, 47
(2000), 29–41.
Dombrowski, D., Rethinking the Ontological Argument: A Neoclassical Theistic Response
(Cambridge: Cambridge University Press, 2006).
Donner, W., and Fumerton, R., Mill (Oxford: Blackwell, 2009).
Dougherty, T., The Problem of Animal Pain: A Theodicy for All Creatures Great and Small
(Houndmills: Palgrave Macmillan, 2014).
Driver, J., Uneasy Virtue (Cambridge: Cambridge University Press, 2001).
Eadmer, The Life of St Anselm, edited and translated by R. W. Southern (Oxford: Oxford
Edwards, P. (ed.), Immortality (Amherst, NY: Prometheus Books, 1997).
Edwards, P., Reincarnation: A Critical Examination (Amherst, NY: Prometheus Books, 2001).
Ellis, G. F. R., ‘Cosmology and Verifiability’, in J. Leslie (ed.), Modern Cosmology and
Philosophy (Amherst, NY: Prometheus Books, 1998), 119–26.
Everitt, N., The Non-Existence of God (London: Routledge, 2004).
Ewing, A. C., Value and Reality: The Philosophical Case for Theism (London: George Allen and
Unwin, 1973).
Filice, C., ‘The Moral Case for Reincarnation’, Religious Studies, 42 (2006), 45–61.
Finnis, J., Natural Law and Natural Rights (Oxford: Oxford University Press, 1980).
Fischer, J. M., Kane, R., Pereboom, D., and Vargas, M., Four Views on Free Will (Oxford:
Blackwell, 2007).
Fisher, A., ‘God, Good, and the Open Question Argument’, Religious Studies, 41 (2005),
335–41.
Flew, A., God and Philosophy (New York: Harcourt, Brace, and World, 1966).
Foot, P., ‘Morality as a System of Hypothetical Imperatives’, The Philosophical Review, 81
(1972), 305–16.
Foot, P., Natural Goodness (Oxford: Oxford University Press, 2003).
Forstrom, K. J. S., John Locke and Personal Identity: Immortality and Bodily Resurrection in
17th-Century Philosophy (New York: Continuum, 2010).
 BIBLIOGRAPHY
Geertz, C., Islam Observed (New Haven: Yale University Press, 1960).
Gellman, J., Mystical Experience of God: A Philosophical Inquiry (Aldershot: Ashgate, 2001).
Gilbert, M., Sociality and Responsibility (Lanham, MD: Rowman & Littlefield, 2000).
Givens, T. L., When Souls Had Wings: Pre-Mortal Existence in Western Thought (Oxford:
Oxford University Press, 2010).
Godfrey-Smith, P., ‘Not Sufficiently Reassuring: Review of Nagel Mind and Cosmos’, London
Review of Books, 35 (24 January 2013).
Goldschmidt, T. C., ‘The New Cosmological Argument: O’Connor on Ultimate Explanation’,
Philosophia, 39 (2011), 267–88.
Gregory, E., ‘Remembering the Poor: Duties, Dilemmas, and Vocation’, in J. Perry (ed.), God,
the Good and Utilitarianism: Perspectives on Peter Singer (Cambridge: Cambridge Univer-
sity Press, 2014), 192–208.
Griffin, J., Well-Being (Oxford: Oxford University Press, 1986).
Grover, S., ‘Cosmological Fecundity’, Inquiry, 41 (1998), 277–99.
Grunbaum, A., ‘Is Simplicity Evidence of Truth?’, Royal Institute of Philosophy Supplement, 82
(2007), 261–75.
Grunbaum, A., ‘A New Critique of Theological Interpretations of Physical Cosmology’, British
Journal for the Philosophy of Science, 51 (2000), 1–43.
Gustafson, J., Ethics from a Theocentric Perspective, two volumes (Chicago: University of
Chicago Press, 1981–4).
Guth, A., The Inflationary Universe (Reading, MA: Addison-Wesley, 1997).
Haidt, J., The Righteous Mind: Why Good People are Divided by Politics and Religion
(New York: Penguin Books, 2012).
Haight, D., and Haight, M., ‘An Ontological Argument for the Devil’, Monist, 54 (1970),
218–20.
Hales, S., ‘Evidence and the Afterlife’, Philosophia, 28 (2001), 335–46.
Hare, J., God and Morality (Oxford: Blackwell, 2007).
Hare, J., God’s Call: Moral Realism, God’s Commands, and Human Autonomy (Grand Rapids:
William Eerdmans, 2001).
Hare, R. M., Moral Thinking: Its Levels, Method and Point (Oxford: Oxford University Press,
1981).
Harmless, W., The Mystics (Oxford: Oxford University Press, 2007).
Hartshorne, C., Anselm’s Discovery: A Re-Examination of the Ontological Proof for God’s
Existence (La Salle, IL: Open Court, 1965).
Hartshorne, C., Omnipotence and Other Theological Mistakes (New York: State University of
New York Press, 1984).
Hartshorne, C., Wisdom as Moderation (New York: State University of New York Press, 1987).
Hasker, W., ‘Can God Be Free? Rowe’s Dilemma for Theology’, Religious Studies, 41 (2005),
453–62.
Hazen, A., ‘On Godel’s Ontological Proof ’, Australasian Journal of Philosophy, 76 (1998), 361–77.
Hebblethwaite, B., In Defence of Christianity (Oxford: Oxford University Press, 2005).
Heck, P. L., ‘Mysticism as Morality: The Case of Sufism’, Journal of Religious Ethics, 34 (2006),
253–86.
Hick, J., An Interpretation of Religion: Human Responses to the Transcendent (Houndmills:
Macmillan, 1989).
BIBLIOGRAPHY 
Hick, J., and McGill, A. C. (eds), The Many-Faced Argument: Recent Studies on the Ontological
Argument for the Existence of God (Houndmills: Macmillan, 1968).
Himma, K., ‘Finding a High Road: The Moral Case for Salvific Pluralism’, International
Journal for Philosophy of Religion, 52 (2002), 1–33.
Holder, R., ‘Fine-Tuning, Multiple Universes, and Theism’, Nous, 36 (2002), 295–312.
Hooker, B., ‘Rule-Consequentialism, Incoherence, Fairness’, Proceedings of the Aristotelian
Society, 95 (1994), 19–35.
Hooker, B., Ideal Code, Real World: A Rule-Consequentialist Theory of Morality (Oxford:
Oxford University Press, 2000).
Horne, J. P., The Moral Mystic (Waterloo: Wilfrid Laurier University Press, 1983).
Huemer, M., Ethical Intuitionism (Houndmills: Palgrave Macmillan, 2005).
Hume, D., Dialogues Concerning Natural Religion (Oxford: Oxford University Press, 2006).
Hurka, T., British Ethical Theorists from Sidgwick to Ewing (Oxford: Oxford University Press,
2014).
Hurka, T., Perfectionism (Oxford: Oxford University Press, 1993).
Hurka, T., Virtue, Vice, and Value (Oxford: Oxford University Press, 2001).
Huxley, A., The Perennial Philosophy (New York: Harper and Row, 1944).
Irwin, T., The Development of Ethics, three volumes (Oxford: Oxford University Press, 2011).
Jackson, F., From Metaphysics to Ethics (Oxford: Oxford University Press, 1999).
James, W., The Varieties of Religious Experience (New York: Longmans, Green & Co., 1902).
Johnston, M., Surviving Death (Princeton, NJ: Princeton University Press. 2010).
Jones, R. H., Mysticism Examined: Philosophical Inquiries into Mysticism (New York: State
University of New York Press, 1993).
Joyce, R., The Evolution of Morality (Cambridge, MA: MIT Press, 2007).
Kane, R., The Significance of Free Will (Oxford: Oxford University Press, 1996).
Katz, S. T., ‘Language, Epistemology and Mysticism’, in S. T. Katz (ed.), Mysticism and
Philosophical Analysis (Oxford: Oxford University Press, 1978), 22–74.
Katz, S. T. (ed.), Mysticism and Sacred Scripture (Oxford: Oxford University Press, 2000).
Kent, B., ‘The Moral Life’, in A. S. McGrade (ed.), The Cambridge Companion to Medieval
Philosophy (Cambridge: Cambridge University Press, 2003), 231–53.
Kent, B., ‘Rethinking Moral Dispositions’, in T. Williams (ed.), The Cambridge Companion to
Duns Scotus (Cambridge: Cambridge University Press, 2003), 352–76.
Korsgaard, C., ‘Personal Identity and the Unity of Agency: A Kantian Response to Parfit’,
Philosophy and Public Affairs, 18 (1989), 101–32.
Kragh, H., ‘Contemporary History of Cosmology and the Controversy over the Multi-Verse’,
Annals of Science, 66 (2009), 529–51.
Kripke, S., Naming and Necessity (Oxford: Blackwell, 1980).
Kukla, A., and Walmsley, J., ‘Mysticism and Social Epistemology’, Episteme 1 (2004), 139–54.
Leftow, B., ‘Anselmian Polytheism’, International Journal for Philosophy of Religion, 23 (1988),
77–104.
Leftow, B., God and Necessity (Oxford: Oxford University Press, 2012).
Leftow, B., ‘Individual and Attribute in the Ontological Argument’, Faith and Philosophy, 7
(1990), 235–42.
Leftow, B., ‘The Ontological Argument’, in W. Wainwright (ed.), The Oxford Handbook of the
Philosophy of Religion (Oxford: Oxford University Press, 2005), 80–115.
 BIBLIOGRAPHY
Leibniz, G. W., ‘On the Ultimate Origination of Things’, in G. W. Leibniz, Discourse on

Metaphysics and Other Essays, translated by D. Garber and R. Ariew (Indianapolis: Hackett,
1991), 41–8.
Leiter, B., and Weisberg, M., ‘Do You Only Have a Brain? On Thomas Nagel’, The Nation, 22
October 2012.
Lenman, J., ‘Consequentialism and Cluelessness’, Philosophy and Public Affairs, 29 (2000),
342–70.
Leslie, J., Immortality Defended (Oxford: Blackwell, 2007).
Leslie, J., Infinite Minds (Oxford: Oxford University Press, 2001).
Leslie, J. (ed.), Modern Cosmology and Philosophy (Amherst, NY: Prometheus Books, 1998).
Leslie, J., Universes (London: Routledge, 1989).
Leslie, J., Value and Existence (Oxford: Blackwell, 1980).
Levy, N., ‘Are We Agents at All? Helen Steward’s Agency Incompatibilism’, Inquiry, 56 (2013),
1–14.
Lewis, C. S., ‘Religion and Rocketry’, in C. S. Lewis, The World’s Last Night and Other Essays
(New York: Harcourt Brace, 1960), 83–92.
Lewis, D., On the Plurality of Worlds (Oxford: Blackwell, 1986).
List, C., and Pettit, P., Group Agency: The Possibility, Design, and Status of Corporate Agents
(Oxford: Oxford University Press, 2013).
Logan, I., Reading Anselm’s Proslogion: The History of Anselm’s Argument and its Significance
Today (Aldershot: Ashgate, 2009).
McGinn, B., The Foundations of Mysticism: Origins to the Fifth Century (New York: Crossroad,
1991).
McGinn, C., Ethics, Evil, and Fiction (Oxford: Oxford University Press, 1997).
McGrew, L. M., and McGrew, T. J., ‘On the Rational Reconstruction of the Fine-Tuning
Argument: A Response to Robin Collins and Alexander R Pruss’, Philosophia Christi, 7
(2005), 425–43.
McGrew, T., McGrew, L., and Vestrup, E., ‘Probabilities and the Fine-Tuning Argument:
A Sceptical View’, Mind, 110 (2001), 1027–37. (Reprinted in N. Manson (ed.), God and
Design (London: Routledge, 2003), 200–8.)
Mackie, J., ‘Evil and Omnipotence’, Mind, 64 (1955), 200–12.
Mackie, J., The Miracle of Theism (Oxford: Oxford University Press, 1982).
McMahan, J., ‘Wrongful Life: Paradoxes in the Morality of Causing People to Exist’, in
J. Coleman and C. Morris (eds), Rational Commitment and Social Justice (Cambridge:
Cambridge University Press, 1998), 208–47.
McTaggart, J., Some Dogmas of Religion (London: E. Arnold, 1906).
Mander, W., British Idealism (Oxford: Oxford University Press, 2011).
Manson, N. (ed.), God and Design (London: Routledge, 2003).
Matthen, M., ‘Thomas Nagel’s Untutored Reaction of Incredulity’, The Philosophers’ Maga-
zine, posted online 4 February 2013.
Matthews, G., ‘Anselm, Augustine and Platonism’, in B. Davies and B. Leftow (eds), The
Cambridge Companion to Anselm (Cambridge: Cambridge University Press, 2004), 32–60.
Mawson, T., ‘How Can I Know I’ve Perceived God?’, International Journal for Philosophy of
Religion, 57 (2005), 105–21.
Mawson, T., Belief in God (Oxford: Oxford University Press, 2005).
BIBLIOGRAPHY 
Mellor, H., ‘Too Many Universes’, in N. Manson (ed.), God and Design (London: Routledge,
2003), 221–9.
Metz, T., ‘The Immortality Requirement for Life’s Meaning’, Ratio, 16 (2003), 161–77.
Mill, J. S., ‘Auguste Comte and Positivism’, in J. M. Robson (ed.), The Collected Works of
J. S. Mill, vol. 10 (Toronto: University of Toronto Press, 1969), 261–368.
Mill, J. S., Considerations on Representative Government, in J. M. Robson (ed.), The Collected
Works of J. S. Mill, vol. 18 (Toronto: University of Toronto Press, 1977), 213–310.
Mill, J. S., On Liberty, in J. M. Robson (ed.), The Collected Works of J. S. Mill, vol. 18 (Toronto:
University of Toronto Press, 1977), 371–578.
Miller, D. E., ‘Hooker’s Use and Abuse of Reflective Equilibrium’, in B. Hooker, E. Mason, and
D. E. Miller (eds), Morality, Rules and Consequences (Edinburgh: Edinburgh University
Press, 2000), 156–78.
Miller, K., ‘Answering the Biochemical Argument from Design’, in N. Manson (ed.), God and
Design (London: Routledge, 2003), 292–307.
Miller, K., Finding Darwin’s God (New York: Harper Perennial, 2007).
Monton, B., ‘God, Fine-Tuning, and the Problem of Old Evidence’, British Journal for the
Philosophy of Science, 57 (2006), 405–24.
Mulgan, T., ‘Consequentialism’, in Oxford Bibliographies Online.
Mulgan, T., The Demands of Consequentialism (Oxford: Oxford University Press, 2001).
Mulgan, T., Ethics for a Broken World: Reimagining Philosophy after Catastrophe (Durham:
Acumen Publishers, 2011).
Mulgan, T., ‘Ethics for Possible Futures’, Proceedings of the Aristotelian Society, 114 (2014), 57–73.
Mulgan, T., Future People (Oxford: Oxford University Press, 2006).
Mulgan, T., ‘How Should Impartialists Think about God?’, in J. Cottingham, B. Hooker, and
B. Feltham (eds), Impartiality (Oxford: Oxford University Press, forthcoming).
Mulgan, T., ‘Mill for a Broken World’, Revue International de Philosophie, 272 (2015), 205–24.
Mulgan, T., ‘Moral Philosophy, Superintelligence, and the Singularity’, draft manuscript.
Mulgan, T., ‘Review of Johnston’s Surviving Death’, Analysis, 71 (2011), 755–65.
Mulgan, T., ‘Theory and Intuition in a Broken World’, in S.-G. Chappell (ed.), Intuition,
Theory, and Anti-theory in Ethics (Oxford: Oxford University Press, 2015), 151–66.
Mulgan, T., ‘Transcending the Infinite Utility Debate’, Australasian Journal of Philosophy, 80
(2002), 164–77.
Mulgan, T., ‘Two Parfit Puzzles’, in J. Ryberg and T. Tannsjo (eds), The Repugnant Conclusion:
Essays on Population Ethics (Dordrecht: Springer, 2005), 23–45.
Mulgan, T., Understanding Utilitarianism (Durham: Acumen, 2007).
Mulgan, T., ‘Utilitarianism for a Broken World’, Utilitas, 27 (2015), 92–114.
Mulgan, T., ‘Utilitarianism and our Obligations to Future Generations’, in D. E. Millar and
B. Egglestone (eds), The Cambridge Companion to Utilitarianism (Cambridge: Cambridge
University Press, 2014), 325–47.
Mulgan, T., ‘What is Good for the Distant Future? The Challenge of Climate Change for
Utilitarianism’, in J. Perry (ed.), God, the Good, and Utilitarianism: Perspectives on Peter
Singer (Cambridge: Cambridge University Press, 2014), 141–59.
Murdoch, I., The Sovereignty of Good (London: Routledge and Kegan Paul, 1970).
Murphy, N., Bodies and Souls, or Spirited Bodies (Cambridge: Cambridge University Press,
2006).
 BIBLIOGRAPHY
Nagel, T., Mind and Cosmos: Why the Materialist Neo-Darwinian Conception of Nature is
Almost Certainly False (Oxford: Oxford University Press, 2012).
Nozick, R., Anarchy, State, and Utopia (New York: Basic Books, 1974).
Nozick, R., Philosophical Explanations (Oxford: Oxford University Press, 1981).
Obeyesekere, G., Imagining Karma (Berkeley: University of California Press, 2002).
O’Connor, T., ‘Review of Rowe Can God Be Free?’, Notre Dame Philosophical Reviews, 8 April 2005.
Oppy, G., ‘Arguments from Moral Evil’, International Journal for Philosophy of Religion, 56
(2004), 59–87.
Oppy, G., Ontological Arguments and Belief in God (Cambridge: Cambridge University Press,
1995).
Ord, T., ‘Global Poverty and the Demands of Morality’, in J. Perry (ed.), God, the Good and
Utilitarianism: Perspectives on Peter Singer (Cambridge: Cambridge University Press, 2014),
177–91.
Orr, H. A., ‘Awaiting a New Darwin: Review of Nagel Mind and Cosmos’, The New York
Review of Books, 7 February 2013.
Parfit, D., On What Matters (Oxford: Oxford University Press, 2011).
Parfit, D., Reasons and Persons (Oxford: Oxford University Press, 1984).
Parfit, D., ‘Why Anything? Why This?’, London Review of Books, 20 (2) 22 January 1998), 24–7.
Parfit, D., ‘Why Does the Universe Exist?’, Times Literary Supplement, 3 July 1992, 3–5.
Pasnau, R., ‘Justified until Proven Guilty: Alston’s New Epistemology’, Philosophical Studies, 72
(1993), 1–33.
Pelikan, J., The Christian Tradition: A History of the Development of Doctrine, vol. 1 (Chicago:
University of Chicago Press, 1975).
Phillips, D., Sidgwickian Ethics (Oxford: Oxford University Press, 2011).
Pigden, C., ‘Identifying Goodness’, Australasian Journal of Philosophy, 90 (2012), 93–109.
Pike, N., Mystic Union: An Essay in the Phenomenology of Mysticism (Ithaca, NY: Cornell
Pitts, J. B., ‘Why the Big Bang Singularity Does Not Help the Kalam Cosmological Argument
for Theism’, British Journal for the Philosophy of Science, 59 (2008), 675–708.
Plantinga, A., God, Freedom, and Evil (Grand Rapids: William B. Eerdmans, 1977).
Plantinga, A., God and Other Minds (Ithaca, NY: Cornell University Press, 1967).
Plantinga, A., ‘Naturalism, Theism, Obligation and Supervenience’, Faith and Philosophy, 27
(2010), 247–72.
Plantinga, A., The Nature of Necessity (Oxford: Oxford University Press, 1978).
Plantinga, A., Warrant and Proper Function (Oxford: Oxford University Press, 1993).
Plantinga, A., Warranted Christian Belief (Oxford: Oxford University Press, 2000).
Plantinga, A., and Tooley, M., Knowledge of God (Oxford: Blackwell, 2008).
Pruss, A., ‘Fine- and Coarse-Tuning, Normalizability, and Probabilistic Reasoning’, Philoso-
phia Christi, 7 (2005), 405–23.
Pruss, A., ‘The Hume–Edwards Principle and the Cosmological Argument’, International
Pruss, A., ‘The Ontological Argument and the Motivational Centres of Lives’, Religious Studies,
46 (2010), 233–49.
Pruss, A., and Gale, R. M., ‘Cosmological and Design Arguments’, in W. Wainwright (ed.), The
Oxford Handbook of the Philosophy of Religion (Oxford: Oxford University Press, 2005), 116–37.
BIBLIOGRAPHY 
Pruss, A., and Gale, R. M., ‘A New Cosmological Argument’, Religious Studies, 35 (1999),
461–76.
Pruss, A., and Gale, R. M., ‘A Response to Almeida and Judisch’, International Journal for
Philosophy of Religion, 53 (2003), 65–72.
Pruss, A., and Gale, R. M., ‘A Response to Oppy and to Davey and Clifton’, Religious Studies, 38
(2002), 89–99.
Putnam, H., ‘The Meaning of “Meaning” ’, Minnesota Studies in the Philosophy of Science, 7
(1975), 131–93.
Putnam, R., American Grace: How Religion Divides and Unites Us (New York: Simon &
Schuster, 2010).
Quinn, P. L., ‘Is Alston’s Response to Religious Diversity an Overstated Case?’, in H. D. Battaly
and M. P. Lynch (eds), Perspectives on the Philosophy of William P. Alston (Lanham, MD:
Rowman & Littlefield, 2005), 146–66.
Rahner, K., ‘Teresa of Avila: Doctor of the Church’, in A. Raffelt (ed.), The Great Church Year:
The Best of Karl Rahner’s Homilies, Sermons, and Meditations, translated by H. Egan (New
York: Crossroad, 1994).
Railton, P., ‘Moral Realism’, The Philosophical Review, 95 (1986), 163–207.
Rawls, J., A Theory of Justice (Cambridge, MA: Harvard University Press, 1971).
Raz, J., ‘Incommensurability and Agency’, in J. Raz, Engaging Reason: On the Theory of Value
and Action (Oxford: Oxford University Press, 1999), 46–66.
Raz, J., The Morality of Freedom (Oxford: Oxford University, 1986).
Rees, M., ‘Other Universes: A Scientific Perspective’, in N. Manson (ed.), God and Design
(London: Routledge, 2003), 211–20.
Rice, H., God and Goodness (Oxford: Oxford University Press, 2000).
Roberts, A., The History of Science Fiction (Houndmills: Palgrave Macmillan, 2005).
Roberts, M., ‘A New Way of Doing the Best We Can: Person-Based Consequentialism and the
Equality Problem’, Ethics, 112 (2002), 315–50.
Rogers, K., Perfect Being Theology (Edinburgh: Edinburgh University Press, 2000).
Rorty, R., ‘Response to Charles Hartshorne’, in H. Saatkamp (ed.), Rorty and Pragmatism
(Nashville: Vanderbilt University Press, 1995), 29–36.
Rowe, W., Can God be Free? (Oxford: Oxford University Press, 2004).
Rowe, W., ‘The Problem of Evil and Some Varieties of Atheism’, American Philosophical
Quarterly, 16 (1979), 335–41.
Rundle, B., Why There Is Something Rather than Nothing (Oxford: Oxford University Press, 2004).
Russell, B., Philosophical Essays (London: Routledge, 2009).
Russell, B., Religion and Science (Oxford: Oxford University Press, 1935).
Sachs, O., Migraine: Understanding a Common Disorder (Berkeley: University of California
Press, 1985).
Scanlon, T., What We Owe to Each Other (Cambridge, MA: Harvard University Press, 1999).
Scheffler, S., The Rejection of Consequentialism (Oxford: Oxford University Press, 1982).
Schmidtz, D., ‘The Meanings of Life’, in L. Founer (ed.), If I Should Die: Life, Death and
Immortality (South Bend, IN: University of Notre Dame Press, 2001), 170–88.
Schneewind, J., The Invention of Autonomy (Cambridge: Cambridge University Press, 1997).
Schneewind, J., Sidgwick’s Ethics and Victorian Moral Philosophy (Oxford: Oxford University
Press, 1977).
 BIBLIOGRAPHY
Schufreider, G., Confessions of a Rational Mystic (West Lafayette, IN: Purdue University Press,
1994).
Schultz, B., Henry Sidgwick—Eye of the Universe: An Intellectual Biography (Cambridge:
Cambridge University Press, 2004).
Sedley, D., Creationism and its Critics in Antiquity (Berkeley: University of California Press,
2007).
Sen, A., Development as Freedom (Oxford: Oxford University Press, 2001).
Sidgwick, H., The Methods of Ethics, 7th edition (Indianapolis: Hackett Publishing Company,
1981).
Singer, P., Practical Ethics, 3rd edition (Cambridge: Cambridge University Press, 2011).
Skidelsky, E., ‘The Strange Death of British Idealism’, Philosophy and Literature, 31 (2007),
41–51.
Sober, E., ‘The Design Argument’, in N. Manson (ed.), God and Design (London: Routledge,
2003), 27–54.
Sober, E., ‘Ending Science As We Know It: Review of Nagel Mind and Cosmos’, Boston Review,
November/December 2012.
Southern, R. W., Saint Anselm: A Portrait in a Landscape (Cambridge: Cambridge University
Press, 1990).
Stace, W. T., Mysticism and Philosophy (Philadelphia: Lippincott, 1960).
Steward, H., A Metaphysics for Freedom (Oxford: Oxford University Press, 2012).
Sullivan, R., Immanuel Kant’s Moral Theory (Cambridge: Cambridge University Press,
1989).
Swinburne, R., ‘The Argument to God from Fine-Tuning Reassessed’, in N. Manson (ed.), God
and Design (London: Routledge, 2003), 80–105.
Swinburne, R., The Existence of God, 2nd edition (Oxford: Oxford University Press, 2004).
Swinburne, R., Is There a God? (Oxford: Oxford University Press, 1996).
Swinburne, R., Providence and the Problem of Evil (Oxford: Oxford University Press, 1998).
Swinburne, R., ‘Second Reply to Grunbaum’, British Journal for Philosophy of Science, 56
(2005), 919–25.
Taliaferro, C., ‘Review of G. Oppy’s Ontological Arguments and Belief in God’, Australasian
Journal of Philosophy, 75 (1997), 553–5.
Taylor, A. E., The Faith of a Moralist: Gifford Lectures Delivered at the University of St Andrews,
1926–1928 (London: Macmillan, 1932).
Taylor, M., nOts (Chicago: University of Chicago Press, 1993).
Tien, D. W., ‘Warranted Neo-Confucian Belief: Religious Pluralism and the Affections in the
Epistemologies of Wang Yangming (1472–1529) and Alvin Plantinga’, International Journal
for Philosophy of Religion, 55 (2004), 31–55.
Tooley, M., ‘Plantinga’s Defense of the Ontological Argument’, Mind, 90 (1981), 422–7.
Trakakis, N., ‘Is Theism Capable of Accounting for Any Natural Evil at All?’, International
Tuomela, R., Social Ontology: Collective Intentionality and Group Agents (Oxford: Oxford
Tyron, E., ‘Is the Universe a Vacuum Fluctuation?’ in J. Leslie (ed.), Modern Cosmology and
Philosophy (Amherst, NY: Prometheus Books, 1998), 222–5.
Underhill, E., Mysticism (New York: Dutton, 1911).
BIBLIOGRAPHY 
Vallentyne, P., ‘Utilitarianism and Infinite Utility’, Australasian Journal of Philosophy, 71

(1993), 212–17.
Vallentyne, P., and Kagan, S., ‘Infinite Value and Finitely Additive Value Theory’, Journal of
Philosophy, 94 (1997), 5–26.
van Inwagen, P., ‘How to Think about the Problem of Free Will’, Journal of Ethics, 12 (2008),
327–41.
van Inwagen, P., Metaphysics (Boulder, CO: Westview, 2002).
van Inwagen, P., ‘Modal Epistemology’, Philosophical Studies, 92 (1998), 67–84.
van Inwagen, P., ‘The Possibility of Resurrection’, International Journal for Philosophy of
Religion, 9 (1978), 114–21.
van Inwagen, P., The Problem of Evil (Oxford: Oxford University Press, 2006).
van Roojen, M., ‘Moral Cognitivism vs Non-Cognitivism’, Stanford Encyclopaedia of Philoso-
phy, updated 2013.
Verdeyen, P., ‘Un theologien de l’experience’, in Bernard de Clairvaux: Histoire, mentalism,
spiritualism (Colloque de Lyon-Citreaux-Dijon, 1992), 557–78.
Visser, S., and Williams, T., Anselm (Oxford: Oxford University Press, 2008).
Wainwright, W., Religion and Morality (Aldershot: Ashgate, 2005).
Walker, M. A., and Ćirković, M. M., ‘Astrophysical Fine Tuning, Naturalism, and the Con-
temporary Design Argument’, International Studies in the Philosophy of Science, 20 (2006),
285–307.
Walker, R. G., Eighteenth-Century Arguments for Immortality and Johnson’s Rasselas (Victoria:
University of Victoria Press, 1977).
Weinberg, S., ‘A Designer Universe’, in P. Kurtz (ed.), Science and Religion: Are They
Compatible? (Amherst, NY: Prometheus Books, 2003), 31–40.
White, L. B., Commentary on the Critique of Practical Reason, 2nd edition (Chicago: University
of Chicago Press, 1996).
White, R., ‘Fine-Tuning and Multiple Universes’, Nous, 34 (2000), 260–76. (Reprinted in
N. Manson (ed.), God and Design (London: Routledge, 2003), 229–50.)
Wielenberg, E., ‘A Morally Unsurpassable God Must Create the Best’, Religious Studies, 40
(2004), 43–62.
Wilken, R. L., ‘Religious Pluralism and Early Christian Theology’, Interpretation, 4 (1986),
379–91.
Williams, B., ‘A Critique of Utilitarianism’, in J. Smart and B. Williams, Utilitarianism For and
Against (Cambridge: Cambridge University Press, 1973).
Williams, B., ‘Internal and External Reasons’, in B. Williams, Moral Luck (Cambridge: Cam-
bridge University Press, 1981), 101–13.
Williams, B., ‘The Makropulos Case: Reflections on the Tedium of Immortality’, in
B. Williams, Problems of the Self (Cambridge: Cambridge University Press, 1973), 82–100.
Williams, P., Mahayana Buddhism (London: Routledge, 1989).
Williams, T., ‘From Meta-Ethics to Action Theory’, in T. Williams (ed.), The Cambridge
Companion to Duns Scotus (Cambridge: Cambridge University Press, 2003), 332–51.
Wittgenstein, L., Tractatus Logico-Philosophicus (London: Routledge, 1981).
Woit, P., Not Even Wrong: The Failure of String Theory and the Continuing Challenge to Unify
the Laws of Physics (London: Vintage Books, 2007).
Wolf, S., ‘Moral Saints’, Journal of Philosophy, 79 (1982), 419–39.
 BIBLIOGRAPHY
Zaehner, R. C., Mysticism, Sacred and Profane (Oxford: Oxford University Press, 1961).
Zagzebski, L., Divine Motivation Theory (Cambridge: Cambridge University Press, 2004).
Zimmerman, D., ‘Criteria of Identity and the “Identity Mystics” ’, Erkenntnis, 48 (1998),
281–301.
Zuboff, A., ‘One Self: The Logic of Experience’, Inquiry, 33 (1991), 39–68.
Index
Adams, M. 234 and the meaning of life 309, 316–22, 340–2,

Adams, R. 41, 56–7, 180, 202–6, 226–7, 343, 361, 381, 385, 388, 394
373, 399, 407, 410 and meta-ethics 59–62, 155, 307, 368–9, 382
afterlife 280–322 moral argument 33–4, 60–2, 85, 348, 368–9
and AP 280–1, 299, 303, 305–7, 309, 340–2 moral implications of 27–8, 156, 203, 243,
and atheism 280, 291, 295–6, 299, 307, 311, 334, 337–8, 378
309–10, 317, 340, 342 and moral knowledge 61–2, 156, 158, 279
and BT 281–90 and moral theory 373–415
animals 233, 281, 283, 290, 309, 311–13 and mysticism 131–2, 138–40, 156–60, 183,
evidence for 290–3 187–9, 279, 331–3, 414–16
and infinity 211, 252, 282, 283, 308, 312, 315, normative (defined 7) 7–8, 28–9, 79, 98, 105,
342; see also infinite utility; infinite value 109, 113, 131, 159, 163, 189, 196, 309,
moral argument 281, 304, 307–22, 340 318, 345
and personal identity 293–303 and objective list theory 344, 360–6,
Alston, W. 130, 131, 135–51, 155, 156–7, 165, 370–2, 381
182–3, 187, 261, 266, 271, 277–8, 331–2 and ontological argument 174–5, 181, 183,
ambiguity 187–9, 267, 334–7
philosophical (defined 9) 9–13, 17, 18, 50, and religious diversity 146, 156, 183, 186,
163, 269, 306, 353 264–8, 277–9, 319, 331–3, 414
religious (defined 9) 8–9, 13, 14, 17–19, 37, and religious pluralism 276–7, 279
42–3, 50, 56, 59, 69, 78, 163, 218, 266, and teleological arguments 100–1
269, 273–4, 309 and value-based morality 278–80
ananthropocentric purposivism (AP) Anselm 161–76, 179–89, 334–7, 343, 373, 406,
and afterlife 280–1, 299, 303, 305–7, 309, 408–9
340–2 on divine incomprehensibility 168–70, 173,
and argument from 179
evil 88, 220–5, 337–40, 364, 367 on the fool 168–72, 180, 335
scale 195–7, 200, 213–19, 328–31 on moral psychology 406, 408–9
and atheist consequentialism 374–403 and Neoplatonism 165–6, 172, 174, 177,
and BT ethics 392, 404–15 180, 406
and Buddhism 267–8, 279 as rational mystic 170–4
and Christian mystical practice 157–60 on thinking and understanding 168–70, 335
contrast with BT 2–4, 11, 60–1, 215–17, 329 argument from evil 8, 10–12, 16, 19, 87–8, 96,
and cosmic features argument 108–9 154, 159, 188, 193–4, 198, 201–2, 216,
and cosmological argument 82, 83, 88, 91, 218, 220–60, 261–3, 269, 270–2, 281–90,
95–7, 325–6 296, 302, 306, 325, 327, 329, 338–9, 341,
defined 1–5 357, 364, 366, 367, 369
and divine command ethics 405–8 animal suffering 225–33
and divine motivation theory 412–14 and AP 88, 220–5, 337–40, 364, 367
and extraterrestrials 200, 208, 211, 213–19, and atheism 220, 337
231, 236, 269, 329–31, 339, 365, 395–6 and BT 221–5
and fine-tuning argument 128–9, 326–8 and free will 233–60
and God 3–4, 7, 18, 61–2, 96–7, 104, 108, and religious diversity 214, 262–3
157–9, 187–8, 299–300, 320, 357–60, argument from scale 31, 91, 97, 128, 193–219,
367, 404–5, 410 221, 261, 268, 328–31, 334, 340, 353,
and human features argument 104–5 357, 378
and human well-being 16, 212–13, 224, and AP 195–7, 200, 213–19, 328–31
251–2, 258, 338–9, 343–72, 375, 385–6, and atheism 193, 195–200, 205–6, 210,
388, 392–3 212–15, 330
leap of faith 18–19, 309, 345, 347, 351–4, 367 and BT 194–6, 198
 INDEX
argument from scale (cont.) leap of faith 18–19, 224, 258, 267, 270–1,
defined 193–200 293, 306–7, 309–10, 313, 322
and divine creation 201–10 and meta-ethics 42, 44, 48–59, 79, 349–50
and extraterrestrials 210–19 metaphysical commitments 224, 229, 232,
and science 194–5, 198–9 238–40, 243, 251–2, 281–90, 299, 303–4,
atheism 306–7, 312–13, 322, 338–9
and afterlife 280, 291, 295–6, 299, 307, moral argument 33–4, 59–61
309–10, 317, 340, 342 and mysticism 131, 138, 152, 156, 158–9
and argument from and non-naturalism 48–50, 52–4
evil 220, 337 and ontological argument 162, 165, 187, 188
scale 193, 195–200, 205–6, 210, and pre-existence 285–90, 294, 296, 310, 312
212–15, 330 and religious diversity 146, 261–3,
and cosmological argument 68–9, 71–8, 268–77, 287
93–5 sceptical 197, 223–4, 249
defined 4, 8–14 and science 68–9, 85, 114–15, 125, 147,
and fine-tuning argument 109, 114, 118, 198–9
125–7, 129, 348 and specific design argument 100, 102
leap of faith 19, 267 Bentham, J. 16–20, 338
and morality 33, 36–7, 42, 47–53, 55–7, Bishop, J. 220, 222, 343
61–2, 79, 128, 150, 152, 156, 158, 187, Boethius 165–6
200, 271, 309–10, 317, 325, 331–2, 334, Bosanquet, B. 301
337–8, 350, 352, 361–3, 370–1, 373–8, Boyd, R. 40–2, 56
387, 396, 406–7 British Idealism 281, 298, 300, 307, 313,
and mysticism 130, 137, 140, 144–7, 152, 340–2, 346
156, 158, 188, 332, 409 Broome, J. 90
and ontological argument 163, 175, 178, Buddhism 4, 66, 134–5, 155–6, 173, 183, 186–7,
180–1, 183, 187 267–8, 273–4, 277, 279, 298, 303–4,
and religious diversity 261–8, 270–5, 277–9 331, 335
and science 4, 68–9, 100, 114, 125–7, 140, Butler, J. 297
146–7, 199, 333
and teleological arguments 100, 102 Camus, A. 315
axiarchism 74, 79, 81–98, 100, 104–7, 111–12, caprice 16–21, 26, 67, 76–7, 102, 104, 107–8,
116, 124–5, 127–9, 210, 215, 281, 325–6, 141–2, 163, 174, 211–12, 215–16, 219,
340, 347, 367, 379 224, 226, 232, 240, 243–4, 267, 293, 306,
and cosmic purpose 90–1 338, 374, 388–9, 396, 402, 414
and cosmological argument 83–91 epistemic 17–18
defined 81–2 substantive 18–20
and explanation 83–7 Carter, B. 110–11
Christian mystical practice 136–60, 175–6, 182,
Barth, K. 161, 162, 170, 343 186, 277–9
belief, ethics of 14, 16–20, 25–6 and AP 157–60
benevolent theism (BT) objections 140–51
and afterlife 281–90 rationality of 135–40
and argument from scale 194–6, 198 and sensory doxastic practice 143–8
and consequentialism 392, 404–5, 412 see also Christianity; mysticism
contrast with AP 2–4, 11, 60–1, 216–17, 329 Christianity 3, 21, 58, 62 133, 138, 146, 153–6,
and cosmic features argument 106, 109 165, 167–8, 179, 182–3, 185, 187–8, 211,
and cosmological argument 81, 83, 91–7 215, 229, 264, 268–78, 283, 285–6, 290,
defined 2–4 292, 295–6, 300, 303–4, 306, 318–19,
and dualism 49, 94, 144, 148, 294–5, 333, 335, 406, 410–15; see also Christian
304, 306 mystical practice
ethics 14, 27–8, 41, 55, 338–9, 354–5, c-inductive argument 65–6, 95, 195–6, 222
358–60, 362–3, 365, 374, 392, 403–15 Comte, A. 301
and evil 221–5 consequentialism 14, 15, 20–1, 25, 37, 79, 89,
and extraterrestrials 200, 210–16, 231–2 109, 172, 196, 201–11, 214, 223, 225,
and fine-tuning argument 109–23 228–33, 238, 244–51, 254, 256, 258,
finite 196–7, 209 260, 281, 286, 289, 301, 311, 319, 337,
INDEX 
339, 344, 359, 363, 371, 373–406, Davies, P. 69, 74, 99, 106, 110–11, 114, 123, 128
408, 411–15 Davis, S. T. 227, 295, 299, 303
act 382–94, 396–7, 399, 402–3, 412 Descartes, R. 137, 144, 162, 171, 225
and AP 374–403 divine command ethics 55, 56, 61, 299, 337,
atheist 374–80, 403–5, 412 371, 192, 404–10
and BT 392, 404–5, 412 divine motivation theory 410–15
demands of 20–1, 37, 153, 311, 382–3, Donner, W. 141
387–90 Dougherty, T. 225, 229, 233, 251, 283, 286
and future people 21–5, 89, 228, 363, 380–1, dualism 5, 49, 93, 94, 102, 144, 148, 292, 294–6,
397, 399 300, 303, 305–6, 309
and God 201–10, 229–33, 412–13
hybrid 403–3 Edwards, P. 71, 291
liberal 223, 235, 246–9, 256, 260, 281, 286, ethic of pure contemplation 415–16
289, 400–2 Euthyphro Dilemma 8, 52, 55–6
rule 375, 381, 385–7, 390–404, 413, 415 Everitt, N. 193–5, 199
cosmic coincidences 25, 73, 110–13, 116, 118, evil
120, 122, 125, 127, 128, 148, 327, 334 compensation 228–9, 252, 283
cosmic features argument 105–9, 111–14, 118, horrendous 225–6, 234–9, 242, 245–60,
121–3, 125–7, 129, 327 262–3, 269–72, 281–90, 296, 298, 302,
and AP 108–9 320, 339, 341, 357, 364, 367
and BT 106, 109 inequality of 257–8, 271, 283, 286
cosmic harmony 314, 357, 362–4, 370–2 moral 220–1, 225, 233–60
cosmic knowledge 262, 265, 271–2, 327–8, 330, natural 221, 233–4, 236, 251, 257, 281
333, 336, 356, 368, 370–2 redemption 228–9, 252, 282–3, 286–7, 320
cosmic purpose see also argument from evil
content of 97–8, 129, 159–60, 354–6 evolution 11, 31, 50–2, 54, 100–5, 110, 116, 125,
and cosmic values 344–54 147, 154, 159, 193, 214, 216–17, 227,
and science 68–9, 99, 101–5, 113–15, 118, 230–2, 241–2, 248, 263, 312–13, 326,
126–8, 198–9, 327 329, 338, 350, 409
cosmic values 10, 28, 46–7, 59–61, 99, 105, 113, of morality 11, 51–4, 101–5, 147, 154, 159
131, 156, 159–60, 174, 201, 208, 213–14, of mysticism 54, 159
219, 224, 265, 310, 316–20, 322, 326–34, and teleological arguments 100–5, 110,
336–7, 340, 342–72, 375–6, 379–84, 116, 125
386–9, 391–3, 395–7, 399, 401–5, Ewing, A. C. 83
412–15 explanation
content 354–6 deflationary 51–2, 54, 130, 147–8, 158, 264,
and cosmic purpose 344–54 267, 278–9, 293, 319, 326, 332–3
fictionalism 346, 353–4 personal 85–6, 93
and human well-being 356–72 extraterrestrials 108, 194, 213–19, 245, 329,
and objective values 357–68 330, 333, 336, 339, 341, 356, 367, 396
cosmological argument 9–10, 53, 65–100, 105, and AP 200, 208, 211, 213–19, 231, 236, 269,
107, 109, 111, 131–2, 144, 163, 165, 172, 329–31, 339, 365, 395–6
174–5, 189, 185, 216, 224, 240, 279, and BT 200, 210–16, 231–2
325–8, 336, 347, 349, 355–6, 369,
378, 413 fictionalism
and AP 82, 83, 88, 91, 95–7, 325–6 about cosmic values 346, 353–4
and atheism 68–9, 71–8, 93–5 about human morality 47, 371–2, 387,
and axiarchism 83–91 396, 403
and brute fact 75–80 fine-tuning argument 10, 53, 78–9, 90, 101,
and BT 81, 83, 91–7 109–29, 131–2, 172, 174–5, 189, 196,
and objective values 71, 79, 81–5, 88–98 198, 240, 326–8, 330, 334, 346–9, 351,
and science 69–74 355–6, 366, 369, 378, 381, 413
Craig, W. 72 and AP 128–9, 326–8
critical stance 41, 56, 359 and atheism 109, 114, 118, 125–7, 129, 348
vs brute fact 115–23
D’Aquili, E. 148 and BT 109–23
Darwin, C. 100–1 for multi-verse 123–8
 INDEX
fine-tuning argument (cont.) grace 202–4

and probabilities 111, 113, 115, 117–22, hiddenness 146, 270–1
125–7, 174, 196, 326–7, 348 and human freedom 234–9, 250, 258–60,
and science 113–15, 123–8 269–70, 284, 288
Finnis, J. 362 impartiality 157, 205, 257–8
free will theodicy (FWT) 16, 220, 233–60, and meta-ethics 33–4, 37, 42, 44, 52–6, 85,
269, 285 93–5, 152–3
freedom and modal facts 75, 178–9
animal 241–3, 247–8 motivation 81, 92–3, 111–12, 194–5,
compatibilism 236, 243–5, 247–53, 197–210, 214–17, 223–4, 248, 363–7,
255–8, 289 379, 410–14
contra-divine (CDF) 235–43, 245–7, 253, necessity 70, 93, 163, 175–6, 179–80
255, 258–9, 269, 284–90, 292, 296–7, perfection 80, 180, 182–3, 185, 347, 355, 406
306, 339 and personal identity 291–2, 295–300, 320
counterfactuals of 237–9, 242, 245, 247, 253, and religious diversity 262–3, 269–71, 287
255, 263 worshipworthiness 184–6
and God 234–9, 250, 258–60, 269–70, Godel, K. 181
284, 288 Great Pumpkin Objection 142–3
and incommensurability 143, 246–7, Gregory, E. 154
255–6, 259 Grunbaum, A. 67, 78
incompatibilism 224, 236–7, 239–48, 250, Gustafson, J. 3, 193
252–4, 269, 288, 306, 339, 405
and transcendent goods 251–3, 258–9 Hare, J. 373, 407–9
value of 16, 248–50, 255–8 Hare, R. M. 404
Freud, S. 366 Harmless, W. 134, 135
friendliness to life 47, 108–29, 132, 195–6, 199, Hartshorne, C. 161, 162, 165, 175–6, 179,
212–13, 268, 327–8, 346–7, 355–6, 381 196, 335
as brute fact 115–23 Hebblethwaite, B. 193, 198, 211–14, 218
explanation of 109–13, 123–4 hedonism 22–6, 43, 58, 225–6, 232–3, 247–9,
probability of 117–20, 125–6 315–16, 344, 363–5, 370, 382
significance of 115–17, 120–3, 126–9 Heidegger, M. 65, 69–70
future people 7, 21–5, 35–7, 43, 89, 90, 107, Hick, J. 261, 265–6, 271–9, 332
216–17, 231, 321–2, 330, 376, 375, Hooker, B. 344, 375–6, 390–1, 394, 397, 400
380–1, 384, 393–400 Huemer, M. 36, 48, 52
human features argument 101–5, 107, 109, 111,
Gale, R. 76 112, 113, 129, 327
Gaunilo 162, 167–9, 174, 180 and AP 104–5
Gellman, J. 130, 146–8, 151 Hume, D. 51, 71, 82, 254, 303, 330, 392, 404
God Hume-Edwards principle 69, 71–4
and AP 3–4, 7, 18, 61–2, 96–7, 104, 108,
157–9, 187–8, 299–300, 320, 357–60, immortality 282–3, 307–10, 318–19; see also
367, 404–5, 410 afterlife, pre-existence
benevolence 4, 203, 226, 239, 251–2, 257–8, impartiality
281, 283 moral 20–1, 153–4, 238, 254, 258, 260, 301,
creature-affecting 203–6, 211, 226–30, 322, 332, 334, 355, 376, 380–1, 383,
233, 237–8, 251–2, 258, 294 388–90, 402, 404
person-affecting 202–13, 223, 226, 229, temporal 21, 398
272, 282, 287, 301–2, 306, 338, 341 Incarnation 3, 61, 215, 272, 410–11, 414
and best possible world 87, 96, 194, 200–12, infinite population 209–11, 214–15, 217–18,
215, 218–19, 223, 225–7, 289, 357, 382, 329, 385–6, 395–6
404–5 infinite utility 209, 252, 282, 320, 342, 385–6,
as brute fact 80, 94 395–6
and consequentialism 201–10, 229–33, infinite value 211, 252, 318
412–13 infinity, and afterlife 211, 252, 282, 283, 308,
defined 2–4 312, 315, 342
experience of 58, 61–2, 134–6, 152, 173, intellectualism 55, 81–3, 91–6, 112, 346–7, 349,
186–7, 215, 266, 282–3, 308, 319 357, 366, 367, 379, 405, 409
INDEX 
intelligent design 101, 198 Mill, J. S. 16–17, 21–2, 25, 141, 222, 255–7, 270,
intuitionism, ethical 10, 23, 39, 47–50, 58, 76, 293, 301–2, 346, 353, 400–1
79, 89–90, 212, 342, 349–50, 374–7, 389, on experiments in living 17, 21, 270, 346,
391, 397, 408 353, 382, 400–1
on moral progress 21–2
Jackson, F. 34, 35, 41, 43, 44, 84 Miller, K. 101, 214
James, W. 132–5, 147, 280, 304, 313 modal facts 178–9, 336
Johnston, M. 291, 295, 296, 301–2, 304, 341 modal logic 162, 176, 336
Joyce, R. 34, 46, 51–4, 354 modal realism 67, 74–5, 124
Moore, G. E. 38–43, 55–8
Kant, I. 13, 21, 50, 153, 161–2, 167, 273, 276–7, moral argument
307–10, 321, 322, 332, 339, 352, 406 for afterlife 281, 304, 307–22, 340
knowledge for AP 33–4, 60–2, 85, 348, 368–9
moral 27, 33, 41–2, 48–50, 52–4, 58, 60–2, for BT 33–4, 59–61
88, 130–1, 138, 140, 150–60, 172, 179, moral realism 35, 38, 45–7, 52, 55, 59–60, 85,
197, 211–12, 218, 224, 240, 248, 257, 103–4, 152, 243, 245, 273, 306–7, 309,
262, 265, 268, 330–3, 355, 376, 384, 389, 315, 326, 340–2, 346, 350, 369, 378,
396–7, 407, 411, 414 396, 407
value of 24, 28, 43, 107, 262–5, 268, 271–2, moral supernaturalism 44, 52, 54–9, 148, 152,
287, 313–14, 319, 327–8, 330–1, 333, 154–5, 175, 264, 326, 332, 337, 340, 342,
336, 357, 359–64, 368, 370–2, 415 345–7, 352, 359, 366, 368–9, 416
Korsgaard, C. 308, 322 morality, austere 20–1, 26–8, 37, 58, 61, 129,
153–7, 213, 311, 333–4, 366, 372, 374–5,
leap of faith, 12–13, 18–19, 113, 163, 214, 378, 380–3, 391–2, 415–16
218, 224, 244, 258, 267, 269–71, 293, multi-verse 111, 113, 114, 117, 118, 120, 121,
306–7, 309–10, 313, 322, 345, 347, 123–8, 214, 218, 326, 327, 348
351–4, 367 atheist 125–6
AP 18–19, 309, 345, 347, 351–4, 367 and cosmic purpose 126–8
atheism 19, 267 explanation of friendliness to life 123–8
BT 18–19, 224, 258, 267, 270–1, 293, 306–7, objections to 125–6
309–10, 313, 322 Murdoch, I. 33, 56
Leftow, B. 184–8, 336 mutual support 14, 25, 31, 53, 60, 79, 85, 122,
Leibniz, G. 71, 73, 75, 87, 161, 178, 180–1, 183, 131, 132, 137, 144, 148, 151–7, 212, 247,
207, 325, 336, 339 281, 290, 337, 344, 349, 360, 372, 373,
Leslie, J. 33, 74, 83–4, 86–7, 110–11, 120–1, 124, 374, 378, 379, 405
126, 210, 218, 325, 367 mysticism 10, 17, 27, 42, 50, 53, 54, 58, 60, 62,
Lewis, C. S. 215 97, 108, 130–60, 162–6, 170–9, 182–3,
Lewis, D. 67, 74 186–9, 208, 213, 216–19, 240, 243–4,
Locke, J. 297, 304–5, 341 258, 261–8, 276–7, 279, 319, 331–7, 340,
Logan 162, 165–7, 180 346–7, 355, 365, 376, 389, 400, 407–9,
414–16
Mackie, J. L. 48, 234–5 and AP 131–2, 138–40, 156–60, 183, 187–9,
McGinn, B. 133 279, 331–3, 414–16
McGinn, C. 48, 52 and atheism 130, 137, 140, 144–7, 152, 156,
McMahan, J. 89 158, 188, 332, 409
McTaggart, J. 280, 292, 293, 295, 298, 305, and BT 131, 138, 152, 156, 158–9
312, 341 defined 135
Mellor, H. 117, 120, 124–6, 326–7, 348 deflationary explanations 146–8
meta-ethics 25–6, 33–62, 84, 95, 104, 144, and morality 27, 42, 50, 53, 54, 58, 62, 130–1,
152–7, 217, 259, 264, 304, 314, 326, 332, 139–40, 142–3, 151–7, 159–60, 172–5,
336, 345–6, 349–50, 366–70, 374, 382, 240, 258, 264–5, 268, 276, 332–3, 346–7,
401, 403–5, 410 355, 365, 376, 389, 407–8, 414–16
and AP 59–62, 155, 307, 368–9, 382 phenomena 132–5
and BT 42, 44, 48–59, 79, 349–50 and religious diversity 146, 150–1, 155–7,
and God 33–4, 37, 42, 44, 52–6, 85, 93–5, 159, 261–8, 276–7, 279
152–3 and science 146–8
Metz, T. 308, 313, 317, 319 see also Christian mystical practice
 INDEX
Nagel, T. 35, 52, 94, 99, 102–7, 140, 148 and objective values 162–4, 172, 174–5,
naturalism, global (defined 34) 33–8, 42–50, 52, 183, 187–9
54, 62, 68, 86, 102, 130, 156, 198, 243, parodies 167–70, 172, 174, 176–81, 183, 185,
281, 305 187–8, 334–5
moral 22, 33–6, 38–45, 47, 51, 55–9, 103–4, open question argument 38–40, 41, 43, 55–8
152, 172, 243–4, 316, 320, 322, 332, 337, Oppy, G. 163, 166, 172, 176, 234, 237
346, 350, 359, 374, 377 Ord, T. 154
Neoplatonism 165–6, 172, 174, 177, 180,
298, 406 Paley, W. 14, 100, 404, 405
Newberg, A. 148 Parfit, D. 20, 22, 24, 25, 27, 32, 49, 65, 80,
nihilism 82, 84–8, 116, 171, 202, 206, 211, 227–8,
cosmic realist 46–7, 59, 371–2 230, 238, 299, 303, 308, 315, 326,
global normative 45–7, 50–1, 372 363, 393
moral 27, 33–6, 38, 44–8, 50–1, 55, 59, Penrose, R. 69
103–4, 147, 152, 187, 200, 243, 245, personal identity 115–16, 227, 293–309, 312,
317–18, 340, 346, 350–1, 371–2, 377 317, 320, 341
non-cognitivism 34, 35–8, 39, 44, 45, 47, 51, 59, and afterlife 293–303
60, 104, 152, 315, 350, 406, 407 and God 291–2, 295–300, 320
non-identity problem 24, 116, 171, 202, 227–8 Persyzk, K. 220, 222
non-naturalism 25–6, 34, 35, 39, 44–55, 58–62, Plantinga, A. 44, 51, 105, 137–9, 140, 142–3,
68, 84–6, 93, 103–4, 130–1, 140, 148, 145, 165, 176, 179, 182, 184, 185, 196,
152, 155, 157, 172, 174, 178, 198, 243–4, 220, 226, 234–7, 239, 247, 269, 270,
264, 309, 316, 320, 322, 326, 332, 334, 275–8, 332, 405
337, 342, 345–7, 350, 352, 366, Platonic Forms 56, 57, 137 251, 258–9, 273
368–9, 377 Platonism 84–6, 165–6, 172, 174, 177, 180,
nothing 65–7, 69–71, 77–9, 87–90, 92, 107, 173 213, 285, 292, 298, 318, 319, 325, 340,
Nozick, R. 23, 66–8, 74, 78, 86, 240 406, 408
pluralism
Obeyesekere, G. 285, 288, 298, 312 doxastic 10, 136–137, 140–1, 143, 331, 335
objective list theory 22–6, 37, 43, 48, 56, 89, religious 265, 272–7, 279, 331–2; see also
107, 150, 249, 315–16, 344–5, 357, religious diversity
360–6, 369–72, 374, 378, 381–2 pre-existence
and AP 344, 360–6, 370–2, 381 animals 285–6, 290, 298, 309, 311–13
and cosmic harmony 314, 357, 362–4, 370–2 and BT 285–90, 294, 296, 310, 312
and cosmic knowledge 262, 265, 271–2, human 243, 280–312, 317, 320, 341
327–8, 330, 333, 336, 356, 368, 370–2 preference theory 22–6, 43, 58, 89, 315–16,
objective values 3, 7–8, 16, 19, 22, 24–7, 33–4, 344–5, 350, 360, 363–6, 369–71, 378,
37–8, 43, 45–6, 53, 59–60, 62, 71, 79, 382–3, 404
81–5, 88–98, 103, 105, 107–8, 113–14, principle of sufficient reason 75–7, 80,
116–18, 120–1, 125–31, 156, 158–9, 144, 325–6
162–4, 172, 174–5, 183, 187–9, 194, Pruss, A. 71, 74, 76, 181–8, 334–5
196–8, 200–1, 205, 207–8, 210–12, 244, Putnam, R. 156
264, 315, 317–18, 326, 336–8, 342, 345,
347, 349, 353–60, 364–71, 375, 377–9, Rawls, J. 13, 21, 375–6
384–6, 389, 393, 395, 398, 401, 415 Raz, J. 246, 255–6
and cosmic values 357–68 reasons
and cosmological argument 71, 79, 81–5, epistemic 45–7
88–98 external 20, 24–6, 28, 37, 48, 82, 315–16, 322,
and ontological argument 162–4, 172, 174–5, 337, 369, 378, 416
183, 187–9 internal 24, 26, 37, 48, 82, 314–15
ontological argument 161–89, 208, 267, rebirth, see afterlife; pre-existence
334–7, 355 reflective equilibrium 16, 367, 375–7, 391, 397
and AP 174–5, 181, 183, 187–9, 267, 334–7 Reid, T. 136–8, 140
and atheism 163, 175, 178, 180–1, 183, 187 religious diversity 61–2, 91, 97, 146, 150–1,
and BT 162, 165, 187, 188 155–7, 159, 183, 186–7, 214, 261–79,
modal 175–87 285, 287, 293, 319, 330–2, 334–5, 353
INDEX 
and AP 146, 156, 183, 186, 264–8, 277–9, theodicy 16, 19, 61, 197–8, 204, 223–35, 235,
319, 331–3, 414 262, 270, 272, 280, 283–6, 289, 301,
and argument from evil 214, 262–263 310–13, 320, 337–8, 340–1, 357, 367,
and atheism 261–8, 270–5, 277–9 405; see also free will theodicy (FWT);
and BT 146, 261–3, 268–77, 287 theory of everything 70–1, 113–15,
and God 262–3, 269–71, 287 126–7, 327
and mysticism 146, 150–1, 155–7, 159, transcendent goods 251–3, 258–9, 282, 309,
261–8, 276–7, 279 316, 318, 320, 331, 336, 340, 359
Rice, H. 84, 86
Rowe, W. 142, 146, 205–10, 220, 283, 332 universe
Russell, B. 130, 142, 261, 313, 316, 331 beginning 72–3
existence 47, 65–98, 113, 325, 328, 347
Scanlon, T. 35, 45 friendliness-to-life, see friendliness to life
Scheffler, S. 402–3, 409 infinity 72–4, 214, 218, 326, 330
Schneewind, J. 10 inhabitants 194, 199, 200, 203, 205–6,
Schufrieder, G. 161, 162, 170, 171 208–19, 231, 328–9, 331, 365, 395, 399,
science 401, 407–8
and argument from scale 194–5, 198–9 intelligibility 61, 100, 106–8, 114, 121–2, 125,
and atheism 4, 68–9, 100, 114, 125–7, 140, 129, 327–8, 333, 355, 381
146–7, 199, 333 mathematical complexity 82, 87, 100, 106–8,
and BT 68–9, 85, 114–15, 125, 147, 198–9 112, 114, 121–3, 125–9, 195, 327–8,
and cosmic purpose 68–9, 99, 101–5, 355–6, 359, 381
113–15, 118, 126–8, 198–9, 327 necessity 70–1
and cosmological argument 69–74 size 194–7
and fine-tuning argument 113–15, 123–8 understanding 61, 100, 106–8, 121–3, 125,
limits of 11, 17–18 129, 327–8, 355–6, 381
and mysticism 146–8 utilitarianism 15–28, 39–40, 79, 107, 153, 200,
Scotus, D. 180, 373, 405–11, 414 205, 209–12, 222–3, 225–6, 231–5, 244,
Sedley, S. 78 247, 249, 252, 255–7, 260, 270–1, 283,
selector 80–2, 84, 86–9, 92, 326 285–6, 337–9, 341–2, 244, 374–6, 379,
Sen, A. 257 385–6, 392–3, 400, 402, 404, 412
sensory doxastic practice 136–8, 140–1, ethics of belief 16–20
143–9, 278 theological 404
Sidgwick, H. 25, 244, 291, 308, 310, 311, 315,
321, 322, 338, 340, 358, 365, 379, 388 Visser, S. 166, 179, 343, 373
Singer, P. 23–4, 153 voluntarism 55, 81–3, 88, 94–6, 112, 347,
Steward, H. 240–3, 248 349–50, 367, 379, 405, 407, 409
suffering
animal 198, 217, 220–1, 223, 225–33, 249, well-being 19, 22–6, 37, 40, 42–3, 45, 48, 56, 58,
251, 302, 338 79, 89, 211–12, 224, 226, 229, 231, 238,
human 7, 16, 207, 221–5, 233–4, 236, 239, 248–9, 253, 256–8, 260, 308, 311,
247, 249, 251, 254–7, 262, 267, 271, 313–17, 343–72, 374–5, 379–80, 382–8,
282–4, 286, 302, 338–9, 341, 364–6, 368, 392–3, 396–7, 400–4, 412–13, 415–16
388, 405 and AP 16, 212–13, 224, 251–2, 258, 338–9,
supervenience 43–4, 58, 349, 376 343–72, 375, 385–6, 388, 392–3
Swinburne, R. 2, 4, 6–7, 65–6, 68, 76–8, 80–1, and cosmic values 356–72
85, 92–6, 102, 109, 139, 142, 146, 161, as resemblance to God 56–8, 360, 382
165, 188, 206–10, 222, 226, 289, 340, see also hedonism; objective list theory;
357, 405 preference theory
Williams, B. 24, 314, 315, 383
teleological argument 99–129, 163, 165, 189, Williams, T. 166, 179, 343, 373
198, 212, 216, 224, 327–8, 333 (see also Wittgenstein, L. 65, 67
cosmic features argument, fine-tuning
argument, human features argument) Zagzebski, L. 373, 410–15

Tim Mulgan - Purpose in The Universe - The Moral and Metaphysical Case For Ananthropocentric Purposivism-Oxford University Press, USA (2015)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Tim Mulgan - Purpose in The Universe - The Moral and Metaphysical Case For Ananthropocentric Purposivism-Oxford University Press, USA (2015)

Uploaded by

Copyright:

Available Formats

OUP CORRECTED PROOF – FINAL, 3/8/2015, SPi

Purpose in the Universe

Part I. The Case against Atheism

Part II. The Case against Benevolent Theism

Part III. Ananthropocentric Purposivist Morality

1.1 What is AP?

As we shall see in chapter 3, the step from simplicity to probability is not

Theism, Atheism, and AP

THEIST (1) not (2)

Not (1) (2) ATHEIST

AP borrows arguments, but not arguments.

Parts I and II explore many different arguments. No single argument is decisive

1.2 Ambiguity: Religious and Philosophical

anything. And—even more importantly—not all explanations are scientiﬁc. Indeed,

1.3 Morality and Metaphysics

If my moral commitments were idiosyncratic, the book would be of little interest.

1.4 My Moral Commitments

1.4.1 Utilitarianism and suffering

I begin with a very straightforward utilitarian commitment. Utilitarians emphasize

1.4.2 A utilitarian ethic of belief

anthropocentrism of privileging human beings over other sentient animals. But AP

1.4.3 Two pictures of morality

Nozick’s experience machine is often read as a decisive refutation of hedonism.

1.4.5 Utilitarian meta-ethics

1.5 The Moral Implications of AP

1.6 Plan of the Book

A comprehensive survey of contemporary meta-ethics is beyond the scope of this

2.1 Two Approaches to Meta-ethics

2.2 Against Non-cognitivism

metaphysically incoherent. Indeed, the obvious impossibility of moral facts is often

2.3 Against Moral Naturalism

2.3.1 Deﬁnitional moral naturalism

2.3.2 Non-deﬁnitional moral naturalism

Robert Adams uses this disanalogy to develop a semantic objection to Boyd.

tend to be hedonists or preference utilitarians. By contrast, a commitment to

2.4 Against Moral Nihilism

cosmic realist nihilism. However, this position is worthy of further exploration.

2.6 Benevolent Theist Non-naturalism

merely criticizing naturalist explanations, BT non-naturalists can offer their own

2.7 Moral Supernaturalism

Unlike moral naturalists, moral supernaturalists defend actual identity claims. No

2.8 A Moral Argument for Benevolent Theism

2.9 Can AP Borrow the Argument for Theism?

develop a non-human-centred account of moral perfection. I address that challenge

It is not how the world is, but that it is that is mystical.

3.1 The Ultimate Question

 THE CASE AGAINST ATHEISM

3.2 Does the Ultimate Question Make Sense?

 THE CASE AGAINST ATHEISM

supernaturalism. A genuinely scientiﬁc response to our ultimate question could

3.3 Three Non-purposive Answers

3.3.1 The instability of nothing

 THE CASE AGAINST ATHEISM

3.3.3 The Hume–Edwards principle

Hume’s response to Leibniz.

 THE CASE AGAINST ATHEISM

... Explanation in an eternal universe The cosmological argument is obviously

 THE CASE AGAINST ATHEISM

In an eternal universe, the Hume–Edwards principle applies. We can get an

3.3.4 Modal realism

3.4 Is S a Brute Fact?

3.4.1 The Principle of Sufﬁcient Reason

 THE CASE AGAINST ATHEISM