Anda di halaman 1dari 268

An Introduction to Metametaphysics

How do we come to know metaphysical truths? How does metaphysical

inquiry work? Are metaphysical debates substantial? These are the
questions which characterize metametaphysics. This book, the first
systematic student introduction dedicated to metametaphysics, discusses
the nature of metaphysics – its methodology, epistemology, ontology,
and our access to metaphysical knowledge. It provides students with a
firm grounding in the basics of metametaphysics, covering a broad range
of topics in metaontology such as existence, quantification, ontological
commitment, and ontological realism. Contemporary views are discussed
along with those of Quine, Carnap, and Meinong. Going beyond the
metaontological debate, thorough treatment is given to novel topics
in metametaphysics, including grounding, ontological dependence,
fundamentality, modal epistemology, intuitions, thought experiments,
and the relationship between metaphysics and science. The book will
be an essential resource for those studying advanced metaphysics,
philosophical methodology, metametaphysics, epistemology, and the
philosophy of science.

TUOMAS E. TAHKO is a University Lecturer in Theoretical Philosophy and a

Finnish Academy Research Fellow at the University of Helsinki. He is the
editor of Contemporary Aristotelian Metaphysics (Cambridge, 2012) and the
author of numerous articles in journals, including Mind, The Philosophical
Quarterly, Erkenntnis, and Thought.
An Introduction to Metametaphysics

University of Helsinki
University Printing House, Cambridge CB2 8BS, United Kingdom

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of

education, learning and research at the highest international levels of excellence.
Information on this title:

© Tuomas E. Tahko 2015

This publication is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.

First published 2015

Printed in the United Kingdom by TJ International Ltd. Padstow Cornwall

A catalogue record for this publication is available from the British Library

Library of Congress Cataloguing in Publication data

Tahko, Tuomas E., 1982–
An introduction to metametaphysics / Tuomas E. Tahko, University of Helsinki.
pages cm
Includes bibliographical references and index.
ISBN 978-1-107-07729-4 (hbk) – ISBN 978-1-107-43429-5 (pbk)
1. Metaphysics. 2. Ontology. 3. Knowledge, Theory of.

4. Science–Philosophy. I. Title.
BD111.T28 2015
110–dc23 2015022580

ISBN 978-1-107-07729-4 Hardback

ISBN 978-1-107-43429-5 Paperback

Cambridge University Press has no responsibility for the persistence or accuracy of

URLs for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.

Preface page ix

1 Why should you care about metametaphysics? 1

1.1 Metametaphysics or metaontology? 3
1.2 How to read this book 6
1.3 Chapter outlines 7
1.4 Further reading 11

2 Quine vs. Carnap: on what there is and what there isn’t 13

2.1 On what there is 15
2.2 Plato’s beard 18
2.3 Enter Meinong 20
2.4 External and internal questions 27
2.5 Language pluralism 35

3 Quantification and ontological commitment 39

3.1 The meaning of the existential quantifier 41
3.2 The existential quantifier and ontological
commitment 45
3.3 Quantifier variance and verbal debates 49
3.4 Beyond existence questions 57

4 Identifying the alternatives: ontological realism,

deflationism, and conventionalism 64
4.1 Ontological realism and anti-realism 65
4.2 Ontological deflationism 71
4.3 Towards extreme conventionalism 76
4.4 A case study: Sider’s ontological realism 83
4.5 Taking stock 90

vi Contents

5 Grounding and ontological dependence 93

5.1 Ontological dependence: a fine-grained notion 94
5.2 Identity-dependence and essential dependence 98
5.3 Is grounding ontological dependence? 104
5.4 Formal features of ground 106
5.5 Grounding, causation, reduction, and modality 112
5.6 Grounding and truthmaking 116

6 Fundamentality and levels of reality 120

6.1 The ‘levels’ metaphor 124
6.2 Mereological fundamentality 127
6.3 Further specifications: well-foundedness
and dependence 133
6.4 Generic ontological fundamentality 136
6.5 Fundamentality and physics 141

7 The epistemology of metaphysics: a priori or a posteriori? 151

7.1 A priori vs. a posteriori 152
7.2 Modal rationalism and a priori methods 155
7.3 The epistemology of essence 163
7.4 Modal empiricism and the status of armchair methods 167
7.5 Combining a priori and a posteriori methods 172

8 Intuitions and thought experiments in metaphysics 177

8.1 Specifying ‘intuition’ 179
8.2 Intuitions and experimental philosophy 185
8.3 Experience-based intuitions 188
8.4 Rational intuition 190
8.5 Scientific thought experiments 194
8.6 Philosophical thought experiments 197

9 Demarcating metaphysics and science: can metaphysics

be naturalized? 203
9.1 Autonomous metaphysics 206
9.2 Fully naturalistic metaphysics 211
Contents vii

9.3 The Principle of Naturalistic Closure

and the Primacy of Physics 217
9.4 Methodological similarities 225
9.5 Moderately naturalistic metaphysics 231

Glossary 236
Bibliography 243
Index 255

Metametaphysical issues, or methodological issues pertaining to

metaphysics, have been central in my work for about a decade. My disser-
tation was called The Necessity of Metaphysics – I have always been optimistic
about our ability to overcome the many methodological challenges that
metaphysical inquiry faces. Although my views regarding many specific
questions have changed over the years, my general attitude towards meta-
metaphysics has remained largely unchanged: I still think that realism is
worth defending, that modal epistemology is of particular methodological
importance in metaphysics, and that we cannot do metaphysics without
relying on at least some a priori reasoning, whatever the correct account of
the relationship between a priori and a posteriori turns out to be.
In recent years I have taught several courses in Helsinki on the topics of
this book. There’s no doubt that preparing for these courses and discussing
the material with my students has helped me to better articulate many of
the central questions of metametaphysics. Much remains to be done before
this young area reaches the conceptual clarity that one might desire, but
I hope that this book goes at least some way towards this goal.
I would like to express my gratitude to the following people, who read
and commented on the material of this book: Hanoch Ben-Yami, Francesco
Berto, Matti Eklund, Guglielmo Feis, Marcello Oreste Fiocco, Joachim
Horvath, Markku Keinänen, James Miller, Matteo Morganti, Donnchadh
O’Conaill, Olley Pearson, Paavo Pylkkänen, and Anand Vaidya. Their val-
uable feedback saved me from many errors and omissions; any remain-
ing errors are my own. I have discussed the material of this book with
too many people to list here as well as presenting the papers related to
the book at numerous seminars and conferences. I appreciate the feed-
back received at these events. I would also like to thank Hilary Gaskin at

"# $%# "&'# &()

x Preface

Cambridge University Press for support throughout the process of writing

this book, as well as an anonymous reader for helpful comments. My great-
est debt, however, will always be to my mentor and PhD supervisor, the
late E. J. Lowe. In Jonathan’s work I first discovered the metametaphysical
attitude that continues to guide my work today. Finally, most of the work
for this book has been made possible by various grants from the Academy
of Finland.
I have drawn on the following previously published and forthcoming
material, although it has been extensively reworked for the purposes of
this book. In addition, some material in Chapter 9 is based on joint ongoing
work with Matteo Morganti. I’d like to thank him for the permission to use
that material in this book.
‘A New Definition ofA Priori Knowledge: In Search of a Modal Basis,’ Metaphysica
9.2 (2008), pp. 57–68.
‘A Priori and A Posteriori: A Bootstrapping Relationship,’ Metaphysica 12.2 (2011),
pp. 151–164.
‘In Defence of Aristotelian Metaphysics,’ in T. E. Tahko (ed.), Contemporary
Aristotelian Metaphysics (Cambridge University Press, 2012), pp. 26–43.
‘Counterfactuals and Modal Epistemology,’ Grazer Philosophische Studien 86
(2012), pp. 93–115.
‘Boundaries in Reality,’Ratio 25.4 (2012), pp. 405–424.
‘Truth-Grounding and Transitivity,’Thought: A Journal of Philosophy 2.4 (2013), pp.
‘Boring Infinite Descent,’ Metaphilosophy 45.2 (2014), pp. 257–269.
‘Natural Kind Essentialism Revisited,’ Mind 124.495 (2015), pp. 795–822.
‘Ontological Dependence,’ in E. N. Zalta (ed.), The Stanford Encyclopedia of
Philosophy (Spring 2015 edn); see
entries/dependence-ontological/ (with E. J. Lowe).
‘Empirically Informed Modal Rationalism,’ in R. W. Fischer and F. Leon (eds.),
Modal Epistemology After Rationalism, Synthese Library (Dordrecht: Springer,
‘The Modal Status of Laws: In Defence of a Hybrid View,’ The Philosophical
Quarterly (forthcoming), doi:10.1093/pq/pqv006.
‘Minimal Truthmakers,’Pacific Philosophical Quarterly (forthcoming), doi:10.1111/
papq.12064 (with Donnchadh O’Conaill.)
1 Why should you care about

This introductory chapter deals with the motivation for studying

metametaphysics and its importance for metaphysics more generally. The
relationship between metametaphysics and metaontology is clarified, some
guidance for reading the book is given, and chapter outlines are provided.
In addition, the chapter contains suggestions for further reading, divided
between introductory material and more advanced material.
Since you’ve opened this book, it is probably safe to assume that you
have an interest in metaphysics. Perhaps you think that metaphysics is an
interesting area of philosophy and want to know more about it or maybe
you’re a student or a professional philosopher specializing in metaphys-
ics. Alternatively, you might be suspicious of metaphysics and its value or
contribution within philosophy (and outside it). Perhaps you think that
metaphysics is not a substantial area of philosophy because it focuses on
pseudo-problems or merely conceptual, linguistic disagreements. You may
be coming to philosophy from another discipline, such as the natural sci-
ences, and you might be suspicious of the methods of philosophy, especially
when compared with the rigour of your own discipline. Or perhaps you work
in a different area of philosophy, wondering how on earth metaphysicians
could possibly justify their outlandish claims about the structure ofreality…
All of the above attitudes are metametaphysical attitudes. Just as with
any kind of attitude, if you hold a metametaphysical attitude you ought
to be able to justify why it is that you hold it. The reason might be simply
because you haven’t seen much discussion about what metaphysicians are
really up to or of how they think they arrive at their various metaphysical
positions. If that’s the case, you’ve opened the right book. If you’re inclined
to be dismissive about the value of metaphysics or think that its methods
are spurious because you have read all the great metaphysicians and found
their work wanting in this regard, you’ve also opened the right book. In

2 Why should you care about metametaphysics?

contrast, if you consider metaphysics the heart of philosophy and can’t get
enough of it, if you enjoy comparing different theories and judging their
relative merits, then – you guessed it – you too have opened the right book.
The author of this book is a metaphysician working in the tradition that
is usually called analytic metaphysics. The analytic vs. continental distinc-
tion is not – the author feels – particularly helpful, but for want of a more
descriptive account, it should be made clear that this book is focused on
the analytic tradition. The author of this book also has a particular meta-
metaphysical attitude. This attitude is a type of ontological realism, which
we will look into in detail later in this book. But as a reader, you should be
aware that the author is biased in favour of certain types of realist meta-
physics and towards the view that metaphysics does have both intrinsic
value and an impact throughout philosophy and the sciences. This is not
an uncommon attitude amongst metaphysicians, but it certainly requires
justification. However, this is not a research monograph defending a par-
ticular position, so space will be given to various positions. Metaphysicians
are a defensive lot; they hold their metaphysical views dear and their meta-
metaphysical views perhaps even dearer, despite the fact that they don’t
always explicitly express the latter. So you will notice that the present
author sometimes takes a defensive attitude. Accordingly, this introduc-
tion to metametaphysics is ‘opinionated’ – someone with a more dismis-
sive attitude towards metaphysics would no doubt write a very different
account. In any case, since it is still much too early to talk about a fully
established set of metametaphysical views, despite certain clear patterns,
anyone writing a book on metametaphysics has to make some difficult
choices on how to lay out the various positions and indeed even what to call
them. Similarly, the precise area that a book on metametaphysics – intro-
ductory or otherwise – should cover is certainly open to debate. This book,
if anything, errs on the side of covering too much, since at times the reader
may feel that the discussion has turned to first-order metaphysics instead
of the promised meta-analysis of metaphysics. This is largely because it is
very difficult, impossible even, to discuss the various metametaphysical
issues without resorting to a battery of examples of first-order metaphys-
ical debates.
The reader will soon notice that there are two themes not obviously
included under metametaphysics, but discussed extensively throughout
this book. They are epistemology and (philosophy of) science. Although it
1.1 Metametaphysics or metaontology? 3

is true that these topics are not obviously metametaphysical in themselves,

it would be difficult to avoid them altogether when discussing metameta-
physics. The reason for this is quite simple. A central, perhaps the central
question of metametaphysics is: How do we acquire metaphysical know-
ledge? Here is an alternative formulation of essentially the same ques-
tion: How does metaphysical inquiry work? These are very clearly epistemic
questions, having to do with metaphysical knowledge. Science and its phil-
osophy enter the picture very quickly after these initial questions, for one
popular answering strategy to epistemic questions in metaphysics is that
metaphysical knowledge and inquiry have something to do with scientific
knowledge or inquiry. Of course, not everyone would accept this answer
and even if one does, difficult questions remain concerning the exact rela-
tionship between metaphysical and scientific knowledge. At any rate, most
metaphysicians today would readily propose that there is either some sort
of important parallelism or else some continuity between metaphysics and
science. At the same time, metaphysics is also one of the last frontiers of
philosophy where pure ‘armchair reasoning’ without any connection to
experimental methods may seem a perfectly acceptable method of inquiry.
So there is also a tension here, one that strongly divides opinions. Given all
this, it is difficult to see how any book concerning metametaphysics could
remain completely silent about epistemic or scientific matters – this one
certainly doesn’t.

1.1 Metametaphysics or metaontology?

Most readers interested in metametaphysics are no doubt familiar with

another, closely related term, namelymetaontology. The title of this book
is a conscious choice: we can distinguish between metametaphysics and
metaontology. The usage of these terms is not entirely standardized, but
roughly put, it could be said that metametaphysics is the broader of the
two terms. More precisely, metametaphysics encompasses metaontol-
ogy, but covers other issues as well. This type of distinction can also be
made between metaphysics and ontology. The term ‘metaphysics’ has an
Aristotelian srcin: according to the usual story, the ‘meta’ (‘beyond’, or
‘after’) refers simply to the fact that in certain collected works of Aristotle
some works appear after his works concerning physics. So ‘metaphys-
ics’ does not really refer to the content of these works, but rather their
4 Why should you care about metametaphysics?

srcinal relative locations. The term ‘ontology’, however, has a more

content-oriented Aristotelian srcin, as the Greek ουτα (onta) refers to
‘being’. So ontology is the study of being (or being qua being – being as it
is in itself, as Aristotle might add). Note however that Aristotle did not use
these terms; they have been adopted later on. Ontology emerges as a some-
what more well-defined, albeit extremely general, area of study, whereas
metaphysics is typically conceived as concerning reality or the structure of
reality, in an even more general sense. The distinction between metaphys-
ics and ontology is, however, vague at best, since many authors use the
terms interchangeably. Accordingly, similar vagueness affects the distinc-
tion between metaontology and metametaphysics.
But what is metaontology? The first systematic use of the term is usu-
ally credited to Peter van Inwagen’s 1998 article of the same title. 1 In van
Inwagen’s usage, the term ‘metaontology’ has Quinean connotations.
Quine considered the central question of ontology to be ‘What is there?’ –
something that we will discuss in Chapter 2. But van Inwagen points out
that if we wish to consider what it is that we are asking when we say ‘What
is there?’, this seems to go beyond ontological questions, hence metaontol-
ogy. Van Inwagen defines a fairly strict sense of the term:metaontology in
Quine’s sense concerns quantification and ontological commitment (these
will be discussed in more detail in Chapter 3). This turns out to be a fairly
narrow understanding of metaontology, but note that on this definition
the metaontological question could be different for someone other than
Quine, who might think differently of the task of ontology. In any case,
largely because of this srcinal usage of the term, metaontology is typically
understood in this relatively narrow sense. In passing, we might note that
one alternative understanding of the task of ontology, perhaps closer to
the Aristotelian line, would be to give a central position to so-called ‘formal
ontology’. This term of art does not refer to ontology conducted with for-
mal methods (although it could involve formal methods); rather, it refers to
the study of ontological form, which involves the structures and relations in
which ontological elements (such as objects) stand.2 More generally, ontol-
ogy understood in this fashion involves an examination of the categorical

Peter van Inwagen, ‘Metaontology,’ Erkenntnis 48 (1998), pp. 233–250.
The terminology has Husserlian srcins, see Barry Smith and Kevin Mulligan,
‘Framework for Formal Ontology,’Topoi 3 (1983), pp. 73–85.
1.1 Metametaphysics or metaontology? 5

structure of reality – a task which goes back to Aristotle’s Categories. One

contemporary example of the systematic study of ontological categories in
this sense is E. J. Lowe’s four-category ontology.3 Hence, it is not difficult to
see that ‘metaontology’ understood from this point of view could amount
to something quite different than when understood from the Quinean
point of view. Partly for this reason, the title of this book contains the
broader term, namely ‘metametaphysics’, for we wish to account for both
of these views.
One source of confusion regarding the term ‘metametaphysics’ may
derive from the fact that the best-known work containing the word in
its title – the 2009 Metametaphysics anthology edited by David Chalmers,
David Manley, and Ryan Wasserman– is by and large focused on the pro-
ject of metaontology as van Inwagen defines it (with some exceptions).4
Indeed, the subtitle of the anthology is ‘New Essays on the Foundations
of Ontology’. In fact, the terms ‘metaontology’ and ‘metametaphysics’ are
also often used interchangeably. But let us attempt to define the term ‘met-
ametaphysics’ as it is used in this book.

Metametaphysics =df The study of the foundations and methodology of


Here, ‘metaphysics’ is understood to encompass ontology, so metameta-

physics will also involve the study of the foundations and methodology of
ontology. Accordingly, metaontology is to be understood as a subspecies of
metametaphysics. Chapters 2 and 3, and to some extent Chapter 4, could
roughly speaking be said to concern metaontology in van Inwagen’s sense,
although they do not do so exclusively. Subsequent chapters (Chapters 5
to 9) concern the methodology of metaphysics in a much broader sense;
they also involve a great deal of epistemology. However, the reader is
advised to not put too much weight on these distinctions, as they are
indeed vague. The guiding thought in this book is to be inclusive and the
suggested definition of metametaphysics certainly allows this. Both terms,
‘metaontology’ and ‘metametaphysics’, are used in this book, roughly in
the sense suggested here; that is, metaontology refers primarily to the

E. J. Lowe, The Four-Category Ontology: A Metaphysical Foundation for Natural Science
(Oxford: Clarendon Press, 2006).
D. Chalmers, D. Manley, and R.Wasserman (eds.), Metametaphysics (Oxford University
Press, 2009).
6 Why should you care about metametaphysics?

study of existence, quantification, and ontological commitment, whereas

metametaphysics encompasses these areas and also broader issues in the
methodology of metaphysics.

1.2 How to read this book

This book is aimed at relatively advanced undergraduate and graduate

students with at least some prior knowledge of metaphysics and related
fields. However, being the first of its kind, it will also prove helpful to
professionals working for example on metaphysics, epistemology, or phil-
osophy of science. While some prior knowledge of metaphysics is assumed,
prior knowledge of metametaphysics is not necessary. An introductory
course in metaphysics ought to be sufficient to follow the book, at least if
the reader supplements this book with some of the further material ref-
erenced within it. The recommended background is an advanced course
in metaphysics and basic knowledge of philosophical logic, although for-
malism is kept to a minimum. It should perhaps be emphasized that the
reader is certainly advised to read some of the primary material referred
to in the book, for it is impossible to do justice to all the topics that we
will cover. Partly for this reason, a fairly extensive bibliography for an
introductory book is provided. Emphasis is given to some of the most
recent literature in metametaphysics, with the hope that even experts
in the field may find the book useful. The book also includes a glossary
with short definitions of some of the most important technical terms. The
glossary is not exhaustive and the reader is also advised to consult the
index for the full context of each term, but the glossary can be used as a
quick reminder.
There are no particularly important guidelines regarding the process of
reading this book. The book has been written with the assumption that
most readers will proceed from the beginning to the end and this is indeed
the advisable order for those not very familiar with the topics of the book,
but each chapter can certainly be read on its own. Typically, when some
prior knowledge of relevant concepts, views, or tools is assumed, this is
indicated in the text, with reference to the chapter where the concept/
view/tool was first introduced. More advanced readers should have no
trouble jumping ahead to topics that interest them. If the book is used for
a course in metametaphysics, the teacher may decide to pick individual
1.3 Chapter outlines 7

chapters to supplement other material. The same can of course be done

with a normal metaphysics course, as many introductory courses in met-
aphysics now contain lessons on the methodology and foundations of
One aspect worth mentioning here is the number of examples from the
sciences that the reader of this book will encounter. In many cases, certain
metaphysical positions are illustrated with examples from the natural sci-
ences, physics and chemistry in particular. It is assumed that most readers
will have some familiarity with many of the examples from previous meta-
physics or philosophy of science courses, but they are generally laid out in
such a way that no prior knowledge is necessary. There are a few excep-
tions, though. For instance, certain examples from fundamental physics
may be difficult to understand without any prior knowledge of physics,
even though they are not presented formally. However, in these cases, the
reader will not miss anything absolutely crucial if they decide to skip the
more detailed examples.
A final note on the system of referencing used. Full bibliographical detail
is provided in footnotes and also in the final bibliography. In each chapter,
the first reference includes the full bibliographical detail; later instances
use the short-title system.

1.3 Chapter outlines

A brief outline of each chapter is provided below. The purpose of these

outlines is to give the reader a general idea of the topics discussed in each
chapter. Note however that technical terms and various ‘isms’ are not
defined in the outlines, so the reader should not be too concerned about
being able to understand the relevant views – that’s what the chapters
themselves are for. Although each chapter can be read on its own, some of
them are thematically connected. This is the case especially with Chapters 2
and 3, and partly also Chapter 4. These three chapters focus on metaontol-
ogy as it was defined earlier, although no particular attempt is made to
stay strictly within metaontology. Chapters 5 and 6 are somewhat techni-
cal, as they introduce the metaphysician’s ‘toolbox’ – concepts and tools of
formal ontology that are used in metaphysics and metametaphysics. Both
chapters also apply these tools. Chapters 7 and 8 turn to epistemic mat-
ters, discussing the methods of metaphysical inquiry. Chapter 9 concludes
8 Why should you care about metametaphysics?

the book with a discussion of the relationship between metaphysics and

science, taking advantage of much of the material of the earlier chapters.

Chapter 2: Quine vs. Carnap: on what there is and

what there isn’t

The historical srcins of metametaphysics are typically traced back to the

debate between W. V. Quine and Rudolf Carnap in the 1940s and 1950s. In
this chapter, an overview of that debate will be given, but the historical
details and the srcinal context of the debate will not be the main subject.
One central topic is the status of existence questions such as ‘Do numbers
exist?’ Are such questions substantial or merely conceptual, to be settled
by linguistics rather than genuine metaphysics? Carnap was famously scep-
tical about the metaphysical import of such questions, arguing that there
is nothing substantial at stake when we ask such questions. The result-
ing view is a type of language pluralism, according to which we can choose
our ontological framework – our preferred language – liberally. Alexius
Meinong’s views on the matter and the problem of ‘Plato’s beard’ – dealing
with non-existence – will also be discussed. Carnap’s distinction between
internal and external questions is outlined and some of its modern applica-
tions discussed.

Chapter 3: Quantification and ontological commitment

This chapter continues to discuss existence questions, but the focus shifts
towards quantification: the status and meaning of the existential quantifier,
including its history and name, are discussed. In particular, the question of
the Quinean criterion of ontological commitment, according to which we
are ontologically committed to those entities that we quantify over, is criti-
cally examined, also with reference to its modern counterparts. Moreover,
the possibility of so-called ‘quantifier variance’ is discussed, as defended
by Eli Hirsch and opposed by Ted Sider, among others. Quantifier variance
is the thesis that there is no single (best) quantifier meaning. The thesis is
closely related to Hirsch’s view that ontological debates concerning physi-
cal objects are ‘merely verbal’. Finally, Kit Fine’s alternative metaontologi-
cal position, which attempts to undermine the importance of existence
questions, is discussed.
1.3 Chapter outlines 9

Chapter 4: Identifying the alternatives: ontological realism,

deflationism, and conventionalism

This chapter surveys various metametaphysical positions, some of which

have already been discussed in previous chapters. The main contenders

are ontological realism, ontological anti-realism, deflationism, and con-

ventionalism. It will become clear that some terminological clarification is
needed in order to correctly identify the various subspecies of these views.
The debate concerning quantifier variance between Hirsch and Sider will
be discussed again, but from a slightly different point of view. Sider’s ver-
sion of ontological realism will receive further attention and is considered
as a case study, with reference to an example from physics.

Chapter 5: Grounding and ontological dependence

It is time to introduce some further metaphysical tools: grounding and

ontological dependence. The notion of ‘ground’ stormed into contempor-
ary analytic metaphysics at the beginning of the twenty-first century, but
the roots of the notion go all the way to Aristotle. At its simplest, ground-
ing may be understood as ‘metaphysical explanation’. To be more precise,
when some x is grounded in some y, it is usually thought that y explains x.
On the face of it, grounding expresses a relation of ontological depend-
ence. Ontological dependence is a family of relations and different ver-
sions of dependence will be discussed in some detail. The question whether
grounding is indeed a version of ontological dependence or not will also be
examined. The formal features of ground and some related notions as well
as applications are outlined. These include causation, reduction, modality,
and truthmaking.

Chapter 6: Fundamentality and levels of reality

This chapter concerns the view that reality comes with a hierarchical struc-
ture of ‘levels’. This type of view has a long history and it remains very
popular. Our everyday experiences as well as scientific practice seem to
strongly support such a view, since scale is a major factor in both of them.
Usually, the reference to scale becomes apparent when talking about parts
and wholes – which are studied in mereology: we talk about subatomic
10 Why should you care about metametaphysics?

particles constituting atoms, atoms constituting molecules, and molecules

constituting everything we see around us. We can express this in terms of
ontological dependence, which is covered in Chapter 5: the wholes depend
for their existence on their parts. Fundamentality comes in when we ask
whether there is an end or a beginning to this hierarchical structure, or
equivalently to the relevant chain of dependence. Much of the discussion
in this chapter will concern the analysis of ‘metaphysical foundationalism’,
which states that there is an end to the chain of dependence, and ‘meta-
physical infinitism’, which states that chains of dependence can continue
infinitely. These views are also discussed with reference to physics.

Chapter 7: The epistemology of metaphysics: a priori or a posteriori?

The epistemology of metaphysics, which is the topic of this chapter, is a

broad area. The discussion starts from the a priori vs. a posteriori distinction,
which turns out to be more controversial than one might have thought.
Various options to clarify the distinction are considered. The bulk of the
chapter deals with modal epistemology: our knowledge of possibility and
necessity. This will be our case study of the epistemology of metaphysics.
Much of metaphysical knowledge seems to involve modal elements, so we
need an account of how we are able to acquire modal knowledge. The two
main competitors here are ‘modal rationalism’ and ‘modal empiricism’.
At first glance, they seem to reflect the a priori vs. a posteriori distinction
regarding the source of modal knowledge, but the situation is more com-
plicated than that, as ‘pure’ a priori or a posteriori knowledge appears to be
scarce. Therefore, a view according to which both types of knowledge are
needed becomes somewhat attractive. Such a view and its prospects are

Chapter 8: Intuitions and thought experiments in metaphysics

This chapter continues on epistemic themes. Intuitions and thought exper-

iments are considered important sources of metaphysical knowledge, but
there is much controversy surrounding them: how reliable are they as
sources of evidence? One problem is that often it is not clear what is even
meant by ‘intuition’. This chapter examines a variety of ways to under-
stand metaphysical intuitions and their role in metaphysical inquiry. These
1.4 Further reading 11

include ‘rational intuitions’, which are typically associated with a priori

faculties, and experience-based intuitions. The chapter also discusses the
relationship between thought experiments and intuitions as well as the
differences and similarities between philosophical and scientific thought

Chapter 9: Demarcating metaphysics and science:

can metaphysics be naturalized?

This chapter concerns the relationship between science and metaphysics.

There are a variety of options in this regard. One of them is that metaphys-
ics is autonomous and able to tell us something about the world on its
own, despite the complications introduced by, say, modern physics. On this
view, metaphysical inquiry into the fundamental structure of reality can
uphold scientific realism. Another option is at the other extreme: fully ‘nat-
uralistic’ metaphysics. According to this view, metaphysics cannot hope
to say anything about reality independently of science. There are also less
extreme versions of this view. One such version is due to James Ladyman
and Don Ross; the view can be outlined by discussing two principles: the
Principle of Naturalistic Closure and the Primacy of Physics . The upshot of these
principles is that metaphysics has a role to play, but it is severely limited,
primarily unificatory. The possibility of building a methodological bridge
between science and metaphysics is also examined: even if the subject mat-
ter of the two disciplines is distinct, perhaps there are some similarities in
their method? Finally, a more modest, reconciliatory view of the relation-
ship between science and metaphysics will be proposed: ‘moderately natu-
ralistic metaphysics’.

1.4 Further reading

Although at the time of writing this is the only textbook dedicated to

metametaphysics, there are certainly important discussions of meta-
metaphysics, metaontology, and the methodology of metaphysics in
various textbooks and collections of papers. Themes relevant to meta-
metaphysics are also discussed, sometimes extensively, in books dealing
with philosophical methodology or metaphilosophy more generally. The
following is a list of such material that, while far from complete, may
12 Why should you care about metametaphysics?

prove useful to the reader of this book. The material is divided between
introductory and more advanced material. Here the focus is exclusively
on books, but the reader is encouraged to browse the final bibliography
for further material.

Introductory material

Francesco Berto and Matteo Plebani, Ontology and Metaontology: A Contemporary

Guide (London: Bloomsbury, 2015).
Chris Daly, Introduction to Philosophical Methods (Peterborough, ON: Broadview
Press, 2010).
Alyssa Ney, Metaphysics: An Introduction (Abingdon: Routledge, 2014).
S. Overgaard, P. Gilbert, and S. Burwood, An Introduction to Metaphilosophy
(Cambridge University Press, 2013).
David Papineau, Philosophical Devices: Proofs, Probabilities, Possibilities, and Sets
(Oxford University Press, 2012).

More advanced material

A. R. Booth and D. P. Rowbottom (eds.), Intuitions (Oxford University Press, 2014).

Albert Casullo and J. C. Thurow (eds.), The A Priori in Philosophy (Oxford University
Press, 2013).
D. Chalmers, D. Manley, and R. Wasserman (eds.),Metametaphysics (Oxford
University Press, 2009).
Fabrice Correia and Benjamin Schnieder (eds.), Metaphysical Grounding:
Understanding the Structure of Reality (Cambridge University Press, 2012).
R. W. Fischer and Felipe Leon (eds.), Modal Epistemology After Rationalism, Synthese
Library (Dordrecht: Springer, forthcoming).
Matthew C. Haug (ed.), Philosophical Methodology: The Armchair or the Laboratory?
(Abingdon: Routledge, 2014).
Matteo Morganti,Combining Science and Metaphysics(New York: Palgrave Macmillan,
T. E. Tahko (ed.), Contemporary Aristotelian Metaphysics (Cambridge University
Press, 2012).
Timothy Williamson, The Philosophy of Philosophy (Oxford: Blackwell Publishing,
2 Quine vs. Carnap: on what there
is and what there isn’t

The historical srcins of metametaphysics are typically traced back to the

debate between W. V. Quine and Rudolf Carnap in the 1940s and 1950s.
In this chapter, an overview of that debate will be given, but we will
not dwell on the historical details or the srcinal context of the debate.
Roughly put, the whole metametaphysics debate is sometimes reduced to
the question of whether existence questions such as ‘Do numbers exist?’
have any deep meaning or, on the other hand, are merely conceptual,
to be settled by linguistics rather than genuine metaphysics. Carnap is
famously considered as having been sceptical about the metaphysical
import of such questions, arguing that there is nothing substantial at
stake when we ask them. The resulting view is a type of language plural-
ism , according to which we can choose our ontological framework – our
preferred language – liberally. 2 This, at any rate, is symptomatic of the
view of some contemporary philosophers who identify themselves as

Readers interested in a more historically oriented account may refer to Matti Eklund,
‘Carnap’s Metaontology,’ Noûs 47.2 (2013), pp. 229–249. As Eklund makes clear, the
typical narrative of the Quine–Carnap debate is highly controversial. But to simplify
matters, we will focus on the way that the debate has been presented in much of con-
temporary literature.
We use the notion of ‘language pluralism,’ following Eklund’s ‘Carnap’s Metaontology,’
for what is sometimes called ‘ontological pluralism’; compare with Matti Eklund,
‘Carnap and Ontological Pluralism,’ in D. Chalmers, D. Manley, and R. Wasserman
(eds.), Metametaphysics (Oxford University Press, 2009), pp. 130–156. This is to distin-
guish Carnap’s view from a different view which is, somewhat confusingly, also some-
times called ‘ontological pluralism’. This more recent usage of the term refers to the
idea according to which there are many ‘ways of being’. We will return to this issue (in
passing) in Chapter 6. See also Kris McDaniel, ‘Ways of Being,’ in Chalmers, Manley, and
Wasserman (eds.), Metametaphysics, pp. 290–319.

14 Quine vs. Carnap: on what there is and what there isn’t

(neo-)Carnapian. 3 As a result of the dismissive view of metaphysics asso-

ciated with Carnap, Quine has emerged as the somewhat unlikely hero
in defence of the possibility of metaphysics. Quine’s famous article ‘On
What There Is’ (see Section 2.1 ) is considered the definitive work against
the Carnapian view that ontological questions are a matter of (linguistic)
preference. 4 Quine’s article is commonly considered to have once again
made ontology a respectable discipline after the devastating effects of
logical positivism, whose proponent Carnap was. However, it should
quickly be pointed out that Quine’s attitude towards ontology in the
context of the present book would seem to be much more deflationary –
that is, leaning towards the view that ontological questions are not sub-
stantive – than the typical narrative of the debate between Quine and
Carnap might suggest. The various options in this regard are discussed
in Chapter 4. In any case, there are good reasons to think that the disa-
greement about ontology between these two figures was not as deep as
it is typically played out to be, one of these reasons being that Carnap
himself apparently did not think that there is a deep disagreement about
ontology between Quine and himself. Be that as it may, the influence of
their debate for contemporary metametaphysics cannot be denied: the
definitive Metametaphysics anthology edited by David Chalmers, David
Manley, and Ryan Wasserman , which contains many of the modern clas-
sics of the literature, focuses almost exclusively on Quine and Carnap,
with a few notable exceptions. 6
There are a number of interesting aspects in the contemporary literature
surrounding the Quine–Carnap debate. We cannot discuss all of them, but
perhaps the most central one is the role of existence questions in ontol-
ogy. Another is the role and interpretation of the existential quantifier. We
will postpone some of the relevant discussion until the next chapter, which
focuses on ontological commitment, as we will need some background

For a prominent example, see Eli Hirsch, ‘Physical-Object Ontology, Verbal Disputes,
and Common Sense,’ Philosophy and Phenom enological Research 70 (2005), pp. 67–97. Note
however that Hirsch limits his discussion to composite physical objects.
W. V. Quine, ‘On What There Is,’The Review of Metaphysics 2 (1948), pp. 21–38; reprinted
in his From a Logical Point of View (Cambridge, MA: Harvard University Press, 1980),
pp. 1–19.
Again, see Eklund, ‘Carnap’s Metaontology.’
Chalmers, Manley, andWasserman (eds.), Metametaphysics.
2.1 On what there is 15

information before we engage with the topic in more detail. That back-
ground will be provided in this chapter, but we will also get to engage with
some of the more controversial aspects of the Quine–Carnap debate, includ-
ing the contemporary incarnations of their respective views. This chapter
starts with a brief recap of the role of Quine’s famous article in the first
section and continues with a discussion of one of its central themes, the
problem of ‘Plato’s beard’, in the second section. After that, in the third sec-
tion, we make a small detour and consider the views of Alexius Meinong,
whose import on the metaontology debate is often ignored. We will see that
Meinong’s work is very much relevant both for the Quine–Carnap debate
and for a number of contemporary views. In thefourth section we will intro-
duce Carnap’s distinction betweenexternal and internal questions, which also
survives, in one form or another, in many contemporary views. From this,
(neo-)Carnapian language pluralism emerges, to be discussed in the fifth

2.1 On what there is

If ontology is the study of what exists, then it would seem that the answer
to the ontological question is not very difficult: everything exists. This
is the rather anticlimactic starting point of Quine’s ‘On What There Is’.
It might appear that this answer is immediately unsatisfactory. For one
thing, what are we to say of Sherlock Holmes and other such entities that,
on the face of it, do not exist, but nevertheless appear to be something?
This is a topic which we will pick up below, but before that we ought to
spend a moment discussing the underlying motivation of Quine’s ques-
tion. In particular, is it really the purpose of ontology simply to list all the
things that there are? Surely not: for instance, we are also interested in the
relationships between existing things. Yet, it would not be fair to Quine to
reduce his position to the over-quoted opening of his article. The import-
ant metaontological aspect of the Quinean position is the thesis that being
is the same as existence, that is, there are no things that do not exist. Peter
van Inwagen, who is a card-carrying Quinean, summarizes the thesis as

This thesis seems to me to be so obvious that I have difficulty in seeing

how to argue for it. I can say only this: if you think that there are things
that do not exist, give me an example of one. The right response to your
16 Quine vs. Carnap: on what there is and what there isn’t

example will be either, “That does too exist”, or “There is no such thing as

On this view, saying that ‘There are numbers’ is on a par to saying

‘Numbers exist’. Van Inwagen’s statement serves to illustrate the key prob-
lem that Quine focuses on, namely the status of things that supposedly do
not exist, but are nevertheless talked about or referred to. Aswe’ll see in the
next section, this problem of ‘nonbeing’ continues to puzzle metaphysicians.
Another important aspect of the Quinean view, which van Inwagen lists
as one of the Quinean metaontological theses, is that all things that exist do
so in the same fashion,univocally. So when we ascribe existence to material
objects such as tables and chairs and to abstract objects such as numbers or
sets, we supposedly mean the same thing by ‘existence’. The supposed argu-
ment in favour of this view is simple:it is the same thing, or at least close
enough, to say that unicorns do not exist and that the number of unicorns
is zero. This connection between existence and number, says van Inwagen,

points towards univocalism about existence. Note however that one could
consider the univocalism in question to concerncounting rather than num-
bers (and existence):If we say that the number of apples in the basket is five
and that eating an apple is one of your ‘five a day’ (portions of vegetables or
fruit), are we using the numbers in exactly the sameway?
A third point about ‘existence’, which we’ll mention only in passing
here, is that the Quinean typically considers the existential quantifier ‘ ∃’ to
capture the presumed univocal sense of existence. We will return to issues
surrounding quantification and the interpretation of the existential quan-
tifier in the next chapter, but for now we can take the Quinean picture at
face value. The upshot of this picture – the Quinean method – is that the

core questions of ontology can be answered with a simple formula:

(1) Take your best scientific theory and assume that what this theory says
is true.
(2) Translate the sentences of your theory into a formal language, typically
first-order predicate logic.
(3) The domain of (existential) quantification in your translated theory will
give you the ontological commitments of that theory.

Peter van Inwagen, ‘Meta-ontology,’ Erkenntnis 48 (1998), p. 235.
2.1 On what there is 17

Note the mention of scientific theories here; for Quine, the central task of
philosophy is to assess and assist scientific theory choice – to determine the
ontological commitments of scientific theories. Of course, this three-step
formula is certainly not the whole of Quine’s import on the topic, but
merely a rough simplification.
While we are on the topic of Quine’s ‘On What There Is’, a further issue
to bring into attention is that in the famous article in question, Quine was
perhaps more interested in defending his version of nominalism rather
than a particular methodological or metaontological point. Quine wanted
to show that when we say that there are red houses, red roses, and red
sunsets, we do not need to commit to the existence of any universal redness,
which all of these red things exemplify. This is simple to show by follow-
ing the method described above: when we translate ‘Socrates is mortal’
into a formal language, we get ‘∃x M(x)’, where ‘M’ stands for the predi-
cate ‘mortal’ and the value of variable ‘ x’ is ‘Socrates’. Here the domain of
the existential quantifier includes Socrates, which we are quantifying over
(in this formula, ‘Socrates’ is a bound variable). But we are not quantifying
over ‘mortal’ – to express a commitment to the universal ‘mortality’ we
would have to resort to second-order logic.8 Now, Quine himself wishes to
strongly resist this type of quantification over predicates, where the predi-
cate is understood as ranging over things such as properties or universals.
As we will see, Quine strives to keep his ontology as sparse as possible,
which means that quantification over such suspicious entities as univer-
sals, for instance, is to be avoided at all costs. Moreover, Quine has inde-
pendent reasons to avoid ascending to second-order logic – he famously
considered second-order logic to be set theory in disguise, or somewhat
more poetically, ‘set theory in sheep’s clothing’.9 What Quine meant by
this is that second-order logic is not logic, properly speaking, but rather a
branch of mathematics (namely a part of set theory). On Quine’s under-
standing, first- and second-order quantification turn out to have very dif-
ferent statuses. The first quantifies over objects, the existence of which is

For more on second-order logic, and to see how this could be done, see Herbert B.
Enderton, ‘Second-order and Higher-order Logic,’ in E. N. Zalta (ed.), The Stanford
Encyclopedia of Philosophy (Fall 2012 edn); see
W. V. Quine, Philosophy of Logic, 2nd ed. (Cambridge, MA: Harvard University Press,
1986), Ch. 5.
18 Quine vs. Carnap: on what there is and what there isn’t

not in any doubt. But the second purports to be much more powerful, as it
quantifies over the realm of universals, properties, concepts … The exist-
ence of these more abstract things can, however, be doubted, and Quine
himself was particularly doubtful about them. It should beadded though that
Quine’s reason to resist quantification over abstract entities derivesfrom his
view regarding the role of quantification itself, namely that quantification is
intimately linked to ontological commitment. More precisely, the quantified
sentences of a theory express the ontological commitments of the theory.
We have arrived at the heart of one of Quine’s most famous slogans: ‘To
be assumed as an entity is, purely and simply, to be reckoned as the value
of a [bound] variable’. This is, in sum, what Quine considers to constitute
ontological commitment. We will assess the Quinean criterion of ontological
commitment primarily in thenext chapter; let us now turn to the problem
of nonbeing, which continues to enjoy a central role in the metaontology

2.2 Plato’s beard

Besides an unwarranted commitment to universals, Quine was concerned

about what has come to be known as the Platonic riddle of nonbeing:

Nonbeing must in some sense be, otherwise what is it that there isnot? This
tangled doctrine might be nicknamed Plato’s beard; historically it has
tough, frequently dulling the edge of Occam’s razor.

The historical srcin of this puzzle is Plato’s Sophist, where the views of
Parmenides are discussed. It is proposed that to say something or to think
something is to speak or thinkof something. The idea is that when Plato
thinks or says that Pegasus does not exist, he must be thinkingof Pegasus.
Hence, there exists something that Plato is thinkingof, namely Pegasus! The
upshot, aptly summarized by Christie Thomas, is that ‘The problem of non-
being has the perplexing consequence that negative existence claims quite

Quine, ‘On What There Is,’ p. 13.
For further discussion on ontological commitment and Quine’s criterion, see also
Philip Bricker, ‘Ontological Commitment,’ in E. N. Zalta (ed.), The Stanford Encyclopedia
of Philosophy (Winter 2014 edn); see
Quine, ‘On What There Is,’ p. 2.
2.2 Plato’s beard 19

generally seem to require the being of the very objects they deny’.
razor is supposed to lead us towards a sparser ontology by denying the exist-
ence of unwanted entities, but the doctrine seems to be in trouble ifthat very
denial commits us to the existence of those unwanted entities. The sparse
ontology that Quine prefers is appropriately illustrated with his metaphor of
a ‘desert landscape’. The idea is exactly that of Occam’s razor, namely that we
should avoid making ontological commitments beyond those that are neces-
sary. For Quine, the desire for desert landscapes is closely connected with
the project of avoiding the apparent ontological commitments of ordinary
language by paraphrasing – expressing the problematic sentence in an onto-
logically non-committing manner, where appropriate.
Quine’s proposed solution to the riddle of nonbeing relies on a para-
phrasing strategy familiar from Bertrand Russell’s theory of descriptions.
Quine wants to steer well clear of the numerous alternative solutions that
he discusses, such as unactualized possibilities and impossible entities –
these would lead us towards a more bloated ontology instead of the
desertified one that he prefers. By employing Russell’s theory, Quine can
paraphrase problematic statements in such a way that no commitment to
non-existing things is required. For instance, the statement ‘The present
King of France is bald’. can be paraphrased as ‘Something is the King of
France and is bald, and nothing else is the King of France’. This solution
relies on Russell’s theory of definite descriptions. Here the definite description
is ‘The present King of France’, but note that this description works quite
differently from a proper name such as ‘Socrates’ in the sentence ‘Socrates
is bald’. The resulting logical forms for these sentences are thus also differ-
ent, which enabled Russell to solve various puzzles.14 Quine believed that
here we also have the tools to solve the Platonic riddle of nonbeing. On
Quine’s view names, descriptions or even predicates do not entail onto-
logical commitment; only quantifying over something does. So if you can
paraphrase away the quantification over non-desirables like unicorns or,
in Quine’s case, universals, then your theory can be maintained without

Christie Thomas, ‘Speaking of Something: Plato’s Sophist and Plato’s Beard,’ Canadian
Journal of Philosophy 38.4 (2008), p. 633. Thomas goes on to provide a historically
detailed analysis of this case, arguing that Plato himself need not be committed to
the existence of things like Pegasus, and hence: ‘Plato’s beard is not quite as tough on
Ockham’s razor as Quine and others have led us to believe,’ p. 667.
See Bertrand Russell, ‘On Denoting,’ Mind 14 (1905), pp. 479–493.
20 Quine vs. Carnap: on what there is and what there isn’t

a commitment to the existence of such problematic entities. Moreover,

Quine suggests that those who do wish to commit to such entities can very
easily do this: by saying that there is something (using existential quantifica-
tion and a bound variable), which Socrates and Aristotle have in common,
namely that they are mortal, we may express a commitment to the univer-
sal ‘mortality’. Similarly, we commit ourselves to the existence of numbers
by saying that there is something which is a prime number larger than a
million. The point that Quine insists on is that there is no other way of
expressing ontological commitments than this use of quantification and
bound variables.

2.3 Enter Meinong

Before moving on to Carnap, we would do well to spend a moment discuss-

ing the often ignored but important contribution of the Austrian philoso-
pher Alexius Meinong to the debate at hand. Meinong was active before
Carnap and Quine and is best known from Russell’s critical commentary,
but his influence is in fact very much present in Quine’s ‘On What There
Is’. Quine’s article discusses the views of two ‘fictional’ characters, McX and
Wyman, who bear remarkable similarity to John McTaggart and Meinong,
respectively. Wyman is the more radical of the two: Wyman, Quine says, ‘is
one of those philosophers who have united in ruining the good old word
“exist”’.15 Meinong is not in agreement with Quine about ‘there is’ or ‘there
are’ and ‘exists’ being univocal. For Meinong, there are some things, such
as Pegasus, that do not exist. But, of course, Quine does not want to say
that there are things that do not exist. Instead, Quine insists that there is no
such thing as Pegasus. So the debate is exactly about the Platonic riddle of
nonbeing: Meinong thinks that some things are merely unactualized – they
do not have the property of existence – but we can still say that, in some
sense, there are such things as Pegasus. One way to illustrate the disagree-
ment between Quine and Meinong, then, is to ask whether existence is
a real property – something that an object can either have or lack. This
distinction can be further analysed with the help of Meinong’s distinction
between Sein – the existential status of entities – and Sosein – the properties
(or ‘nature’) that an entity has. The distinction is at the root of Meinong’s

Quine, ‘On What There Is,’ p. 3.
2.3 Enter Meinong 21

principle of independence, which states, roughly, that an entity may have any
properties whatsoever, independently of its existence. According to this
principle, non-existent objects can have all the properties that we ascribe
to them: we can (truly) say that non-existent golden mountains are made of
gold and that round squares are round. For Quineans everything exists, for
Meinongians not everything exists.
Meinong himself probably considered existence itself to be ‘said in
many ways’, that is, to be equivocal rather than univocal, but many later
Meinongians would agree with Quineans about there being just one sense
of existence. So there are several different aspects to the debate between
Quine and Meinong on the one hand and contemporary Quineans and
Meinongians on the other. Similarly, where Quine rejects the distinction
between ‘there is’ and ‘exists’, Meinong himself might endorse this dis-
tinction and make use of it in addressing the problem of nonbeing, but
not all contemporary Meinongians would do so. For instance, Graham
Priest, whose views we’ll discuss in more detail below, would be closer to
Quine on this particular point even though his view can be called a type
of Meinongianism.16 Accordingly, we should keep in mind that Meinong’s
srcinal position, which is partly open to interpretation, may be different
from the position of some contemporary philosophers who have adopted
the label ‘Meinongian’. The same is no doubt true of Quine and ‘Quinean’.
But the core of the debate between these various versions of Quineanism
and Meinongianism is the disagreement about the status of existence as a
property that can be distinguished from the object itself – this is an issue
where Priest would be squarely in the Meinongian camp.
Leaving the historical details aside, we may distinguish between vari-
ous further positions regarding existence and quantification among con-
temporary Meinongians. For instance, some would consider existence to
be univocal, with Quine, but some would deny this. Some would distin-
guish between different senses of ‘being’ for existents and non-existents,
whereas some would deny this distinction and insist that non-existents
have no being whatsoever. Let’s borrow a passage from David Lewis to fur-
ther illustrate the various options that now emerge; here is Lewis on ‘con-
troversial entities’ that some would consider to exist and some would not:

Graham Priest, Towards Non-Being: The Logic and Metaphysics of Intentionality (Oxford
University Press, 2005).
22 Quine vs. Carnap: on what there is and what there isn’t

These controversial entities include past and future things, the dead who
have ceased to be and those who are not yet even conceived; unactualized
possibilia; universals, numbers, and classes; and Meinongian objects,
incomplete or inconsistent or both. An expansive friend of the entities
who says that all these entities exist may be called an allist. A tough
desert-dweller who says that none of them exist may be called a noneist.
At first glance, it might seem that Meinong is an allist on Lewis’s terminol-
ogy, since he says that thereare all sorts of controversial entities. But note
that Meinong does not claim that these entities
exist. This is in fact a key point
for Meinong: there are different ways for different kinds of entities to be. Some
entities exist, but some of them only ‘subsist’. This ‘subsistence’ is ascribed
to something that does not exist, that perhaps is outside of being altogether.
But for Meinong, the being or nonbeing of an entity is not part of its very
nature. Meinong has another principle that captures this idea:
the principle of
indifference. This principle says that an entity isindifferent to being by its very
nature, even though in any given case the entity is either in a state of being or
of nonbeing. However, the entity’s nature
may dictate its nonbeing:a round
square has a Sosein (nature) which dictates that it could never exist, but the
principle of indifference states that the entity’s nature does not include its
being or nonbeing. All this opens up a number of puzzles, some of which
would require discussing and interpreting Meinong much more thoroughly
than we can do here. Hence, we may be better off relying on some of the
interesting secondary literature that Meinong’s work has inspired.
In the secondary literature, Meinong sometimes appears to come closer
to a type of noneism, as described by Lewis. That seems at least half right;
or rather, as Richard Woodward puts it, ‘Noneists hold that Meinong was

half right: right to distinguish existent objects from non-existent ones, but
wrong to distinguish those non-existents that have being (the ‘subsistent’)
from those that do not’.18 Allists, on the one hand, are with Quine at least
to the extent that they hold ‘existence’ to be univocal. On the other hand,
as we saw earlier, Quine famously prefers desert landscapes when it comes
to ontology, whereas the allists embrace any and all of the controversial

David Lewis, ‘Noneism or Allism?’, Mind 99.393 (1990), p. 23.
Richard Woodward, ‘Towards Being,’ Philosophy and Phenomenologica l Research 86.1
(2013), p. 183. See also Tatjana von Solodkoff and Richard Woodward, ‘Noneism,
Ontology, and Fundamentality,’ Philosophy and Phenomenological Research 87.3 (2013), pp.
2.3 Enter Meinong 23

entities. In contrast, the noneists may be desert-dwellers when it comes to

existence, but that does not compel them to accept the Quinean doctrine
of ontological commitment, which emphasizes quantification. Instead, a
noneist could hold that we can quantify over non-existents without much
trouble.19 Hence, it seems that the question of being vs. existence is separa-
ble from the riddle of nonbeing, as suggested by Tim Crane: ‘whether there
is a distinction between being and existence is irrelevant to the truth of the
claim that there are things that do not exist’.20
Meinongians would say that controversial as many of the entities
mentioned by Lewis may be, that does not mean that they are unable
to have properties. For instance, Pegasus has a certain set of properties –
being horse-like and having wings in particular – but lacks the property
of existence. The Quinean objections to this idea are two-fold: either the
Meinongians have misunderstood the notion of existence altogether or,
if there is a genuine ontological disagreement here, then the Quinean
position is obviously correct, since the Meinongian position is nearly
self-contradictory. So the Quinean reaction is rather severe. Quine raises
the objection that the Meinongian doctrine of unactualized possibilities
is highly implausible, citing examples like the round square cupola on
Berkeley College or other such logically impossible entities. The Meinongian
will acknowledge that such entities are indeed impossible, but insists
that this only means that they could not exist, not that there are no such
(impossible!) entities. This is of course a significant price to pay, although
the idea of impossible entities – or entire impossible worlds – may not be
completely alien to us. 21 Moreover, on the face of it, the Meinongian line
enables us to show that anything whatsoever exists: we can take any set of
properties X and in accordance with the Meinongian principle of independ-
ence it appears that there could be an object with those properties. Take the
properties of ‘being golden’, ‘being a mountain’, and ‘existence’
– if there
is an entity with these properties, then we may conclude that a golden
mountain exists!

See Richard Routley, ‘On What There Is Not,’ Philosophy and Phenomenological Research 43
(1982), pp. 151–578 and Priest, Towards Non-Being. See also David Lewis’s commentary
on Routley in Lewis, ‘Noneism or Allism?’
Tim Crane, The Objects of Thought (Oxford University Press, 2013), p. 24.
For discussion, see for instance Daniel Nolan, ‘Impossible Worlds:A Modest Approach,’
Notre Dame Journal of Formal Logic 38.4 (1997), pp. 535–572.
24 Quine vs. Carnap: on what there is and what there isn’t

The Meinongian position has certainly not enjoyed as much support as

the Quinean position. This may be partly due to prejudice going back to
the seemingly devastating criticisms of Meinong’s position by Russell and
Quine. Yet contemporary Meinongianism is at least as sophisticated as con-
temporary Quineanism.22 In fact, Meinongian metaontology and related
positions, such as noneism, have only recently started to receive the atten-
tion they undoubtedly deserve. Both of the above-mentioned problems for
Meinongianism had already been raised by Russell, and they are indeed
genuine problems for Meinongians, but there have been some interesting
attempts to address them.23 One of these attempts, building on the work of
Terence Parsons, is based on identifying so-called ‘nuclear properties’ – a
privileged set of properties which is allowed in our set X of properties that
an entity could satisfy. Importantly, ‘existence’ is not among such nuclear
properties; rather, it is an ‘extranuclear’ property. The obvious problem
that follows is how to specify the nuclear/extranuclear distinction, so the
Meinongian is not home free yet. Richard Routley’s ‘characterization pos-
tulate’ makes a similar attempt, suggesting that objects have the character-
izing properties used to characterize them. But Routley’s approach likewise
suffers from arbitrariness, for it is not clear who is supposed to determine
these characterizing properties.
There are, however, further strategies. For instance, Priest revises
the nuclear or characterizing properties approach, proposing that the
non-existent objects have these properties in the possible (or impos-
sible) worlds where they exist. Further, some properties, such as
being a golden mountain, are ‘existence-entailing’. What this means
is that the golden mountain, for instance, does not actually exist, but
since the property of ‘being a golden mountain’ is existence-entailing,
the entity does exist in those worlds which contain entities having that
property. It is not entirely intuitive to determine which properties are

For a survey of the current debate, see Maria Reicher, ‘Nonexistent Objects,’ in E. N.
Zalta (ed.), The Stanford Encyclopedia of Philosophy (Summer 2014 edn); see http://plato.
See for instance, Francesco Berto, ‘Modal Meinongianism and Fiction:The Best of Three
Worlds,’ Philosophical Studies 152 (2011), pp. 313–334, and Terence Parsons, Nonexistent
Objects (New Haven, CO: Yale University Press, 1980), as well as the already cited work
by Routley and Priest.
Priest, Towards Non-Being, p. 86.
2.3 Enter Meinong 25

existence-entailing in this sense. For instance, Priest himself holds that

‘being round’ is not existence-entailing, whereas ‘being round or square’
might be. 25 Accordingly, this solution may suffer from some of the same
problems that the nuclear/extranuclear distinction does. Moreover, the
solution obviously violates Meinong’s principle of independence, since
it explicitly suggests that the properties that an object has may have
a bearing on its existence. While Priest’s solution may be an improve-
ment over Meinong’s and Routley’s solutions, it is still an open question
which properties should be considered as existence-entailing. This issue
is picked up by Crane, who is of the opinion that there may be signifi-
cantly more such properties than Priest suggests. Crane also thinks that
one can get rid of the possible world machinery, which admittedly com-
plicates some of the other proposed solutions.
On Crane’s view, it is the natures of things that determine which prop-
erties are existence-entailing, as these natures determine which condi-
tions must hold for predications to be true of those things. In contrast,
non-existent objects cannot have certain properties, such as being golden,
because it is in the nature of golden things that they exist. This entails,
as suggested above, that the supposedly non-existent golden mountain
does exist in some worlds, even though not in the actual world. However,
non-existents, such as Sherlock Holmes, can have certain properties
that are not existence-entailing, such as being fictional. This is a prop-
erty that the non-existent Sherlock Holmes can have even in the actual
world. Crane’s suggestion as well requires something further, namely an
account of what it is to have a property and how exactly the natures of
things determine which properties are existence-entailing. But we will
leave this issue aside here, as it would require heavier metaontological
machinery than we have at our disposal so far (some of that machinery
will be provided in subsequent chapters).
Whether or not Crane is able to overcome the srcinal Russelian
challenge, it is clear that his view is another interesting example of an
anti-Quinean position. Crane’s view does differ somewhat from both
the Meinongian and the noneist lines, though. Crane’s starting point is
that he wants claims about non-existents to come out as true, but this is

See Reicher, ‘Nonexistent Objects.’
Crane, The Objects of Thought, p. 60
26 Quine vs. Carnap: on what there is and what there isn’t

inconsistent with the standard approach to existence and quantification.

Let’s briefly consider one of his examples:

(S) Some characters in the Bible existed and some did not.27

(S) appears contradictory, because the domain of the quantifier ‘some’

includes both existents and non-existents. So it cannot come out as true
on the standard understanding of quantifiers like ‘some’. But Crane wishes
(S) to come out as true without interfering with the basic logic of quanti-
fication. Instead, Crane offers a reinterpretation of domains of quantifica-
tion, suggesting that they may include non-existents as well. But why think
that domains of quantification could not include non-existents in the first
place? Crane has a theory, based on the popular conception of domains of
quantification regarded as sets:

I suspect that resistance to the idea of domains ‘containing’ non-existent

objects comes from a certain metaphysical picture of what these domains

involve. The picture is that sets are collections of things and these
things must be real if we are going to understand how language hooks
up with reality at all. The appeal of thinking in terms of sets is that sets
are precisely defined entities, and have a clear criteria of existence and
identity. On this picture, all members of sets must exist, and so if domains
are understood in terms of sets, then domains containing non-existents do
not make any sense.28

Instead of conceiving domains of quantification as sets, Crane proposes

that quantifying over things is simply a way of talking about things, to have
them in our discourse. Crane’s interpretation of (S) is that to say ‘there are
biblical characters who did not exist’ should not be understood as assert-
ing the existence of these non-existent biblical characters. Rather, it is to
say that some biblical characters did not exist – the quantifier ‘there are’
should not mislead us. So existential sentences, on Crane’s view (reflecting
that of Louise McNally), are not used to assert the existence of anything,
but rather to introduce new referents into the discourse. The upshot is that
the existential claim ‘there are Fs which are G’ is not obviously equivalent
to the claim ‘there exist Fs which are G’, rather than simply ‘someFs are G’,

Ibid., p. 30. For a classic discussion relating to this example, see Saul Kripke, Reference
and Existence (Oxford University Press, 2013).
Ibid., p. 37.
2.4 External and internal questions 27

where the natural language quantifier ‘some’ is not necessarily ontologi-

cally committing.29
To conclude this tightly-packed section, it seems clear that we would
do well to take the Meinongian approach and its variations seriously.
Meinongianism has been undergoing something of a renaissance in recent
years and this trend can be expected to continue. We will now return to the
heart of the Quine vs. Carnap debate.

2.4 External and internal questions

Quine’s ‘On What There Is’ was partly aimed at Carnap, since Quine was
suspicious of Carnap’s commitment to properties, propositions, and mean-
ings. For Carnap, these are as unproblematic as numbers: there is nothing
wrong in our commitment to such entities. Carnap’s position relies on his
distinction between internal and external questions:

[W]e must distinguish two kinds of questions of existence: first, questions

of the existence of certain entities of the new kind within the framework, we
call them internal questions; and second, questions concerning the existence
or reality of the system of entities as a whole, called external questions. Internal
questions and possible answers to them are formulated with the help of
the new forms of expressions. The answers may be found either by purely
logical methods or by empirical methods, depending upon whether the
framework is a logical or a factual one.30

Internal questions have an absolute priority in Carnap’s system, because it

turns out that only internal questions are to be understood as factual. These
are questions such as ‘Are there carbon molecules larger than C60?’ to which
we can, sometimes quite easily, provide a true or false answer based on the
empirical methods available in our scientific framework. Questions like
‘What is the smallest prime number?’ would also qualify as internal, but in
this case the answer is analytically true or false. For instance, Carnap would
consider all mathematical truths to be analytically true, so true in virtue

Ibid., p. 46 ff. For an in-depth discussion of quantifiers such as ‘some,’ see also Hanoch
Ben-Yami, Logic and Natural Language (Surrey: Ashgate, 2004).
Rudolf Carnap, ‘Empiricism, Semantics, and Ontology.’ Reveue Internationale de
Philosophie 4 (1950); reprinted in his Meaning and Necessity: A Study in Semantics and Modal
Logic (University of Chicago Press, 1956), p. 206.
28 Quine vs. Carnap: on what there is and what there isn’t

of meaning, similarly to the famous ‘All bachelors are unmarried’. In con-

trast, external questions concern the framework itself: they are philosophi-
cal or ontological questions about the nature and, perhaps, justification of
the whole framework. We might think that external questions should be
raised when introducing a new framework, such as a specific scientific the-
ory. Hence, we could ask:‘Is quantum mechanics committed to the objec-
tive existence of the wavefunction?’ or something similar. Or, again, we
could ask: ‘Do numbers exist?’ Philosophers certainly ask such questions in
earnest, but Carnap considers this a mistake: ‘[T]he introduction of the new
ways of speaking does not need any theoretical justification because it does
not imply any assertion of reality’.31 The attitude that Carnap is advocating
here is liberal, but also dismissive. It is liberal in the sense that we are free
to adopt novel frameworks without a lengthy assessment of their theoretical
justification (but note that they may still need a pragmatic justification). In
this sense, external questions could be considered a matter of pragmatic
preference, without true/false answers. But Carnap’s approach can also be
seen as dismissive in the sense that the value of a philosophical analysis
of different frameworks and indeed the possibility of objective comparison
between their theoretical commitments seems to be undermined.
To fully understand Carnap’s approach, we ought to say something
about what is being meant by ‘framework’ here. This turns out be a tricky
issue. Scientific theories were mentioned in this connection, but Carnap’s
point seems more general: the system of numbers and the system of points
in space-time can also be understood as frameworks.32 Since the notion of
a framework is clearly supposed to apply to a very wide variety of things,
including both abstract and concrete things, we could perhaps under-
stand the notion of a framework as any linguistic system (or a language
fragment) that is constructed in order to speak about new kinds of entities.
This way a framework could apply just as well to the system of numbers
and the system of concrete things. The language of a scientific theory deal-
ing with novel entities is certainly one example of such a linguistic frame-
work (think about the names that scientists give to these entities and the
concepts that they introduce), but any novel ‘language’, regardless of its

31 Ibid., p. 214.
For further discussion on ‘framework,’ see Eklund, ‘Carnap’s Metaontology,’ pp. 231 ff.
2.4 External and internal questions 29

function, would presumably qualify. With the internal/external distinction

and the notion of a framework in place, it is not difficult to see that, at
least on one interpretation, Carnap’s view would seem to lead towards a
certain type of relativism: if the factivity of any existence claim can only be
assessed relative to a given linguistic framework, then the type of universal
existence claim that philosophers are often interested in loses much of its
value. Indeed, the most radical interpretation of this view suggests that
asking factual questions in the external sense – like philosophers perhaps
sometimes do – is completely misguided, devoid of cognitive content.33 But
it should be immediately noted that this type of framework-relativity does
not necessarily arise from the interpretation of frameworks as linguistic
systems. The relativism that does emerge may not be any more serious than
a type of linguistic relativism: there are different possible languages and
the truth-values of sentences are relative to languages. So, whether this
gives rise to any kind of scepticism about ontology, as Carnap’s work is
often considered to do, is controversial. Indeed, if Matti Eklund is correct,
the type of sceptical result regarding ontology that is sometimes associ-
ated with Carnap only goes through on a more radical interpretation of

On a second, relativist, understanding, Carnap’s ‘frameworks’ are not mere

language-fragments; instead, frameworks are like perspectives or outlooks.
On the relativist view, framework-relativity is not the trivial dependence
of meaning upon language, but the sentences ontologists are concerned
with are framework-dependent in some more radical sense: the propositions
the sentences express are not true or false absolutely but only relative to
frameworks. An internal question is a question of what is true relative
to some framework in this demanding sense of ‘framework.’ An external
question is a question of what framework is the correct one.34

Eklund’s view is that Carnap is not a relativist in this sense, but merely a
language pluralist, which is the view that emerges from the interpretation of
frameworks as linguistic systems or language-fragments. We’ll return to a
contemporary discussion of language pluralism in the fifth section.
Leaving Carnap exegesis aside, the internal/external distinction is not
just of historical relevance. The distinction has been applied and adapted

For discussion, see Eklund, ‘Carnap and Ontological Pluralism,’ p. 132.
Eklund, ‘Carnap’s Metaontology,’ p. 233.
30 Quine vs. Carnap: on what there is and what there isn’t

in various contemporary accounts, especially by those who consider

themselves (or whom others consider) broadly ‘neo-Carnapian’. Many
contemporary views in metametaphysics have been motivated by the
internal/external distinction; one such is Thomas Hofweber’s. 35 Although
Hofweber’s reading of Carnap’s distinction may be controversial, the view
that he develops has an interesting contrast with Carnap’s view. Hofweber
distinguishes an internal and an external reading of quantification . The
external reading is the standard domain-related understanding of quan-
tification, so when quantifiers are read in this sense, our focus is on the
domain of objects being quantified over, whatever it may be. The resulting
truth-conditions on this reading follow the standard semantics for quanti-
fiers – this is how quantifiers are normally used. But Hofweber thinks that
quantifiers are polysemous, semantically underspecified. That’s because
sometimes quantifiers have, in addition to the domain-related reading, an
internal reading, which concerns the inferential role of the quantifier. The
idea is that sometimes, in ordinary English, we mean to use quantifiers
in a different sense than the external reading suggests. Hofweber doesn’t
claim that this is a particularly common phenomenon, but there are,
he thinks, clear special cases where quantifiers must be read in another
sense. One such case is when we attempt to communicate under partial
ignorance – when we have some idea about what we are talking about,
but perhaps lack detail. In such a case, the domain-related understanding
of quantifiers may be inappropriate because we may ignorant about the
To illustrate Hofweber’s external/internal distinction, consider how
we might understand the statement ‘Everything exists’ – something that
Quine might utter. On the face of it, this statement is trivially true, if
‘everything’ includes all the things that exist. But we need to keep in mind
the problems concerning non-existents; there may be some things, such as
Sherlock Holmes, that do not exist, and hence not everything exists. So if
‘everything’ includes Sherlock Holmes, then the statement is false. These
two senses are supposed to correspond with the external and the internal
reading of quantification, respectively. The internal reading is more appro-
priate when we consider the possibility of non-existents – it’s not clear

See for instance Thomas Hofweber, ‘Ambitious, Yet Modest, Metaphysics,’ in Chalmers,
Manley, and Wasserman (eds.),Metametaphysics, pp. 260–289.
2.4 External and internal questions 31

whether they are in the domain of the quantifier or not. Hofweber illus-
trates the distinction further with statements such as ‘Someone kicked
me’, where the reading of the quantifier is clearly external – there is a
specific someone who kicked me and it doesn’t necessarily even matter if
I don’t know who it is. But consider: ‘There is someone we both admire,’
but I have forgotten who it is that we both admire. In this case the idea
is that the quantifier will not range over the whole world, even though it
could be anyone that we both admire. Why is that not enough? Well, if it
happens to be, say, Sherlock Holmes that we both admire, then it appears
that the quantifier would range over something that is not included in
the world at all. So while communicating under partial ignorance about
who it is that we both admire, the quantifier cannot be understood in the
external sense.
Hofweber goes on to apply the internal/external distinction to
the familiar question: ‘Are there numbers?’ On the internal reading,
the answer turns out to be a trivial ‘yes’. But Hofweber is not an advo-
cate of the trivialization of metaphysics. Indeed, he suggests that on the
external reading the question is not trivial, and furthermore, he thinks
that mathematics does not provide an answer to the external reading.
Accordingly, perhaps it is the external reading of existence questions that
metaphysics is concerned with: they are not (entirely) trivial and the spe-
cial sciences do not answer them. On the face of it, this of course con-
trasts with Carnap’s view – Hofweber, after all, considers himself to be
defending the possibility of metaphysics, albeit in a ‘modest’ sense. But
since we have seen that it is rather difficult to pin down Carnap’s view, it
is an open question how much of this is truly in contrast with Carnap. For
Hofweber, the external and internal readings have the same status, but
his account does agree with Carnap’s to the extent that internal readings
of quantified statements are very often trivial. Indeed, when it comes to
the question ‘Are there numbers?’ Hofweber considers the question to be
trivial on the internal reading, but that’s not what metaphysicians are
usually thought to be after when they ask this question. Carnap’ s judge-
ment about this is that the metaphysician asks a meaningless question,
but Hofweber has a little more to say. In particular, he considers number
words to be non-referential. Hofweber posits that number-talk does not
even aim to refer. If this is correct, then there is nothing in the world that
numerals pick out and hence no such things as ‘numbers’:
32 Quine vs. Carnap: on what there is and what there isn’t

Number words, just like any other words, can be used by particular
speakers with the intention to refer, and these speakers can succeed in
referring to something. I can use ‘two’ to refer to my biggest tomato plant,
and succeed. But I can’t use it or any other word to refer to the number
two (as this phrase is commonly used).36

More generally, Hofweber concludes that all that is left for metaphys-
ics to do is to determine whether, in regard to a given existence question,
internalism or externalism is true. If internalism is true, then there are
no entities of the type in question. If externalism is true, then metaphys-
ics has nothing to do with the answer. According to this view, the role of
metaphysics is modest indeed, even if it is supposed to be more ambitious
than according to Carnap.37
Admittedly, Hofweber’s use of the internal/external distinction has only
a remote connection to Carnap’s, but the distinction has been taken up
by many others as well. For instance, David Chalmers adapts the distinc-
tion and develops a new, related distinction: ordinary existence assertions
vs. ontological existence assertions.38 The former, as the name suggests,
are ordinary assertions such as ‘There are three chairs in this room’. The
latter are typical philosophical assertions, such as ‘There is a universal
“redness” shared by this red chair and that red table’. Chalmers illustrates
the distinction with the idea of an ‘ontology room’ – a special place where
it is acceptable to make ontological existence assertions and possible to
have debates about matters ontological. Outside the ontology room, only
ordinary existence assertions can be properly evaluated. The ontology
room has become a standard tool in metametaphysics, although it may
have only a loose connection to Carnap’s srcinal view. In any case, we

ought to look into this tool in some more detail, since we will encounter it
later on as well.
Before Chalmers, Cian Dorr proposed that the questions of ontology
could be conceived as being discussed in the special language of ontol-
ogy, call it Ontologese, which is more suitable for ontological debates than

Hofweber, ‘Ambitious, Yet Modest, Metaphysics,’ p.268.
For further discussion, see T. E. Tahko, ‘In Defence of Aristotelian Metaphysics,’ in
T. E. Tahko (ed.), Contemporary Aristotelian metaphysics (Cambridge University Press,
2012), pp. 26–43.
See David Chalmers, ‘Ontological Anti-Realism’ (p. 81), in Chalmers, Manley, and
Wasserman (Eds.), Metametaphysics, pp. 77–129.
2.4 External and internal questions 33

ordinary English.39 The scenario that Dorr entertains is that of several (iso-
lated) language communities, each one of them speaking English, but dif-
fering in one crucial respect. Let us focus on just two of these communities,
the universalists and the nihilists. These labels reflect the attitude of each
community regarding the ‘Special Composition Question’ – something of a
default test case in the metametaphysics literature.40

Special Composition Question (SCQ) Under what conditions do a

number of objects compose a further object?

The universalists answer this question by saying that a number of objects

always compose a further object, no matter how far apart or how different
they are; your nose and the Eiffel Tower compose a further object, so do the
Milky Way and Andromeda galaxies, and indeed all those four things. The
nihilists are at the other extreme, as they would say that a group of objects
never composes a further object: all there is are simple objects, so-called
‘mereological atoms’. SCQ is a question concerning mereology, the study of
parts and wholes. Between the mentioned extremes lies the view that we
can give a set of conditions under which a number of objects compose a
further object – one famous answer on these lines is van Inwagen’s, accord-
ing to whom only living things – lives – constitute composite objects in
the required sense.41 Our current interest, however, is not so much the
SCQ itself, but rather our understanding of the debate in general. Since
these different language communities differ in their attitude regarding the
question, we must ask: is their disagreement merely linguistic or is there a
substantial difference in their views about the structure of reality? In other
words, is there any genuine metaphysics involved?
We can already see how this discussion is connected to Carnapian
themes: the different language communities in Dorr’s example could be
interpreted in terms of Carnap’s ‘frameworks’. So what would Carnap
say about the example? He would presumably insist that each of the lan-
guage communities truly has its own language and the claims that the
See Cian Dorr, ‘What we Disagree about when we Disagree about Ontology,’ in
M. E. Kalderon (ed.), Fictionalism in Metaphysics (Oxford University Press, 2005),
pp. 234–286. The label ‘Ontologese’ is from Ted Sider, see for instance his ‘Ontological
Realism,’ in Chalmers, Manley, and Wasserman (eds.), Metametaphysics, pp. 384–423.
The modern formulation of the Special Composition Question originates from Peter
van Inwagen’s Material Beings (Ithaca, NY: Cornell University Press, 1990).
34 Quine vs. Carnap: on what there is and what there isn’t

universalists and the nihilists make in their respective languages hold true
in that language. Dorr calls this type of view ‘conciliatory’, because we can
conciliate the debate between the communities simply by translating their
respective claims about composition. How exactly this could be done is of
course another question, and one which Dorr has attempted to answer.
Others, such as Ted Sider, would disagree with the Carnapians about the
possibility of such a translation.42 We need not go into all the details of the
possible translation scheme here, but there is one aspect that we will have
to consider in some detail later, namely the possible variation of the mean-
ing of quantifiers between languages.43
Before we move on, let us use this debate to further illustrate the idea
of the ontology room. To understand this tool in the present context, we
might think that it is us, as metaphysicians, who constitute one language
community, and ‘folk’ – the woman and man of the street – who consti-
tute another language community. In fact, the previous example, SCQ, is
something that ‘folk’ may not have a very strong opinion about. In the
‘folk language’, both universalism and nihilism seem to make rather
strange claims. Folk will likely think that a number of objects compose
a further object whenever they seem to do so. There are tables and chairs,
cars and houses. It would seem utterly silly to deny that these composite
objects exist. Similarly, ‘objects’ like the combination of your nose and
the Eiffel Tower seem blatantly not to exist. But many metaphysicians take
these types of debates quite seriously. So we may ask: is the disagreement
between ordinary folk and metaphysicians a substantial one? Here is the
charitable reading that Dorr proposes:

What we debate in the ontology room is the question what there is strictly
speaking – what there really, ultimately is – what there is in the most fundamental
sense. Of all the many meanings a quantifier like ‘something’ might
have, one is special. This is the one in terms of which all the rest are to be
analysed; it is the one such that to find out what there is in
this sense would
be to fulfil the traditional metaphysical goal of comprehending reality
as it is
in itself. When we do ontology, our quantifiers bear these special meanings.

See Ted Sider, ‘Hirsch’s Attack on Ontologese,’Noûs 48.3 (2014), pp. 565–572.
For a classic account on this issue, see Eli Hirsch, ‘Quantifier Variance and Realism,’
Philosophical Issues 12 (2002), pp. 51–73.
44 Dorr, ‘What we Disagree about when we Disagree about Ontology,’ p. 250.
2.5 Language pluralism 35

The italicized words in this passage – really, ultimately, fundamental –

may not seem very helpful. Indeed, unless they are combined with an
account about the hierarchical structure of reality and the import of
the fundamental – the most ‘real’ – things, then they do not help us
make sense of the metaphysician’s task. The idea behind such talk
wil l be revisi ted in due course, primarily in Chapters 5 and 6, but we
wil l kee p coming across it repeatedly in thi s book. We can already
note one easy reply to this type of suggestion, which Dorr mentions
as well: one can question the idea that there is any ‘special’ sense of
quantification and insist that reality can be described equally well in
a number of languages. Such a resp onse can be motivated by Carnap’s
approach, since we’ve seen that Carnapians would consider the idea
of the ontology room to be a doomed attempt at aski ng external ques-
tions that are factive. Instead, the Carnapian would claim that the
meaning we ascribe to quantifiers is a pragmatic choice. Hence, if we
are debating the SCQ, we should choose the answer most appropriate
to our pragmatic goals. It seems likely that the upshot would then be
some kind of modest position, between the extremes of universalism
and nihilism .
For an ontological realist, whose approach the passage quoted from
Dorr reflects, this result is unsatisfactory. We will discuss this position in
more detail later, but in the remainder of this chapter we will focus on
the Carnapian approach. Proponents of the Carnapian approach would
embrace the pragmatic result rather than be disheartened by it. This has
resulted in metaontological theories that purport to be able to deal with
the type of language pluralism that the Carnapian approach would seem
to entail.

2.5 Language pluralism

In contemporary terms, Carnap could be described as a language plural-

ist . This is, perhaps, the most charitable interpretation of his view: there
are a variety of different frameworks – interpreted as languages or lan-
guage fragments – that we can adopt, and different existential assertions
come out as true in these frameworks because each language has its own
sense of ‘existence’. This claim is blatantly in conflict with the Quinean
36 Quine vs. Carnap: on what there is and what there isn’t

approach according to which existence is univocal. Indeed, one way

to understand the debate between Carnap and Quine – or, at any rate,
between neo-Carnapians and neo-Quineans – is that they disagree about
the interpretation of existence claims. More accurately, the disagreement
is sometimes taken to concern the interpretation of the existential quan-
tifier: whether it is univocal in the Quinean sense or whether it varies
from context to context. This debate falls under the rubric of quantifier
variance , which is a topic that we will discuss further in the next chap-
ter. For the time being, let us focus on the historical background and
the basic idea of language pluralism , without all the complications of the
contemporary debate.
We can illustrate the idea of language pluralism with the debate about
the ontological status of numbers, which has already made its appear-
ance in earlier examples. Platonists would answer affirmatively when asked
‘Do numbers exist?’ They would consider numbers to be abstract objects.
Many mathematicians seem at least implicitly to favour this type of view.
Nominalists would disagree, arguing that numbers do not exist (they might
think there are no abstract objects at all). Now, enter the language plural-
ist; she thinks that the Platonists and nominalists are both correct, rela-
tive to their respective frameworks. In the Platonist language the assertion
of existence regarding numbers comes out as true, but in the nominalist
language it comes out as false. So from the point of view of the language
pluralist, there is no real disagreement here. We saw a similar attitude
emerging in the last section with regard to the Special Composition
Question. It seems that the position holds quite generally. The view that
disagreements about ontology tend to be merely linguistic disagreements
rather than substantial ontological disagreements naturally follows – this
could be labelled as a type of deflationism about ontology. But it is not really
clear how we should understand the view, for there is a sense in which it is
perfectly trivial: the utterance ‘Numbers exist’ seems to come out as true
in some languages (the Platonist’s language) and false in some other lan-
guages (the nominalist’s language). This is something we already saw in
Hofweber’s analysis. Yet, unless we have an understanding about whether

Eli Hirsch, a kind of neo-Carnapian, is perhaps the best known proponent of quantifier
variance, e.g., his ‘Quantifier Variance and Realism’.
2.5 Language pluralism 37

‘exist’ (or ‘numbers’, for that matter) means the same thing in these differ-
ent languages, we are not saying anything very worthwhile. Surely, the
language pluralist wants to say something more than this. But what does
she want to say?
First, regarding the historical case, we should note that language plural-
ism does not fall straight out of Carnap’s distinction between internal and
external questions. We can accept this distinction without committing
ourselves to language pluralism. Still, the tool provided by the internal/
external distinction is certainly attractive for anyone who is inclined to
accept language pluralism. If we agree that all the external questions –
which ontology mostly deals with – are pragmatic, then it might seem
that our ontological disagreements quickly begin to dissipate. However,
this inference does not go through as it is. As Matti Eklund puts it, the
internal/external distinction needs to be complemented by the assump-
tion that there are ‘equally good’ languages to choose from. 46 In other
words, language pluralism follows only if one thinks that it genuinely
doesn’t matter – other than for pragmatic purposes – which language we
choose. But this makes language pluralism a much less attractive view, for
there are surely many questions that we would regard less sensitive to the
Carnapian analysis. For instance, scientific questions such as ‘Are there
tigers in Australia?’ are evidently not subject to the same type of vari-
ation between languages as, perhaps, the question about the existence of
numbers. It would be rather strange indeed to insist that it is a question of
which language we adopt whether there turn out to be tigers in Australia.
So it is difficult to see how the language where tigers exist in Australia and
the language where they do not could be ‘equally good’ in the required
sense. The upshot is that even if in ontology we ask questions that often
seem to turn on a choice of language, this does not appear to be true of sci-
ence or ordinary discourse in general. Nevertheless, language pluralism,
at least in some of its forms, requires the additional assumption regarding
‘equally good’ languages.
It turns out that getting to the bottom of the language pluralist’s posi-
tion is more difficult than it may first have seemed. On the one hand, the
position is threatened by trivialization. On the other, the initial force and

Matti Eklund, ‘Carnap and Ontological Pluralism,’ p. 140.
38 Quine vs. Carnap: on what there is and what there isn’t

appeal of the position may be lost unless it generalizes. There is more

to be said about language pluralism, but we will return to some of these
issues in later chapters, especially in Chapter 4, where different metam-
etaphysical positions are discussed in more detail. In the next chapter we
will go deeper into the metaontological issues in the background of the

Quine vs. Carnap debate, quantification and ontological commitment in

3 Quantification and ontological

In the previous chapter, we saw that quantification and ontological

commitment are central themes in the history of metaontology. The cen-
tral question is: Does quantifying over some entity entail ontological com-
mitment to the existence of that entity? The Quinean story is now familiar to
us: to be (an entity) is to be the value of a bound variable, that is, a variable
in the scope of the existential quantifier. In this chapter we will, among
other things, clarify this Quinean idea and examine some of the implica-
tions of the different views we have already outlined.
We begin with an analysis of the meaning – or interpretation – of the
existential quantifier. We’ve seen that W.V. Quine and Alexius Meinong
would disagree about whether there is a difference between ‘there is’ and
‘exists’. For Quine they mean the same thing:existential quantification is
univocal, and does not allow for variation in meaning. But could the disa-
greement between Quine and Meinong be explained simply by a difference
in their chosen interpretation of the existential quantifier? This might
seem to be a tempting explanation and it is not untypical to hear mention
of ‘Meinongian quantification’ as opposed to the ‘Quinean’ sort. To be sure,
there are different views about quantification and existential quantifica-
tion in particular. According to one such view, even the name of the quanti-
fier is misleading: it should be called the ‘particular quantifier’ instead. On
this view, the particular quantifier is not existentially loaded (and hence
should not be called ‘existential’). This type of view is defended by some
noneists (about whom we learned in Chapter 2). Recall, however, that for
contemporary Quineans and Meinongians the important difference is
whether existence may be regarded as a property – not all Meinongians
would resist the Quinean univocality thesis. Furthermore, some would con-
sider it a mistake to regard ‘exist’ as any kind of a quantifier at all (recall
the example: ‘Some characters in the Bible existed and some did not’). On

40 Quantification and ontological commitment

this sort of view, we should perhaps be focusing on ‘there is’ and its ilk
instead. We might then suggest that the debate about univocalism con-
cerns ‘there is’, which means something different from ‘exists’. 1 Some of
these issues are closely related to the problem of whether ‘exists’ should be
considered a predicate as opposed to a quantifier. We have discussed related
matters in some detail in Chapter 2 and seen that there are arguments on
both sides. Below we will get a little deeper into this issue, but despite a
recently active debate, it should be noted that the received view still takes
existential quantification quite seriously, leaving the view taking existence
as a predicate to the minority.2 We will also consider a more neutral way to
understand ontological commitment, making use of a distinction between
the ontological commitments of a sentence and the semantic machinery
assigning truth conditions to that sentence.
Later on, in the third section, we will return to the topic of quanti-
fier variance in its contemporary form. A brief overview of the ongoing
debate between Eli Hirsch and Ted Sider on this topic will be given,
although these two are by no means the only philosophers who have
contributed to the discussion. Finally, after we have considered quantifi-
cation and ontological commitment from several angles, the reader may
be pleased – or frustrated! – to discover that the emphasis on quantifi-
cation in metaontology may in fact be misguided, as argued by Kit Fine.
Fine’s view will be considered in the fourth section. The discussion so far
has assumed that the manner in which Quine understands the task of
ontology is at least broadly correct, even if we have heard from many dis-
senting voices already. This understanding is that ontological questions
are (primarily) quantificational questions. But this assumption has been
called into question; we will consider some reasons to think that the
most interesting ontological questions are in fact not quantificational.
So what else could these interesting ontological questions be? That will
be a recurring topic in later chapters, but one popular suggestion is

For a related discussion, see Hanoch Ben-Yami,Logic and Natural Language (Surrey: Ashgate,
2004); Hanoch Ben-Yami, ‘Plural Quantification Logic: A Critical Appraisal,’ Review of
Symbolic Logic 2.1 (2009), pp. 208–232.
For further discussion on ‘existence’ and especially on the issue of whether it should be
considered a predicate, see Michael Nelson, ‘Existence,’ in E. N. Zalta (ed.), The Stanford
Encyclopedia of Philosophy (Winter 2012 edn); see
win2012/entries/existence/ and the references therein.
3.1 The meaning of the existential quantifier 41

that the interesting questions concern what is fundamental – what are

the most basic building blocks of reality. The Quinean account does not
address this type of question, but others claim to be able to do so. Of
course, the debate goes on, since there are those who would take a (neo-)
Carnapian attitude towards these supposedly interesting ontological
questions. After all, the question about what is fundamental looks very
much like an external question in Carnap’s sense, as discussed in the
previous chapter : if the fundamental is assumed to concern the object-
ive, framework-independent structure of reality, then it would seem
to violate Carnap’s framework-dependence. Perhaps this isn’t neces-
sarily the case, though, since a Carnapian could understand questions
regarding fundamentality in a different manner, internal to a particular
(philosophical) framework. Be that as it may, one goal of the present
chapter – and this book in general – is to provide the reader with some
further tools to assess this convoluted debate.

3.1 The meaning of the existential quantifier

On the basis of the discussion in the previous chapter, we can summarize

two competing views about the meaning of the existential quantifier – or
simply the meaning of ‘exists’. On the Quinean view, ‘exists’ has only one
meaning, it is univocal. When we say that numbers exist or that cabbages
exist, we are using the very same existential quantifier, and its meaning is
the same in both cases. On one version of the Meinongian view, statements
like ‘There are things that do not exist’ express something different: we
might call it ‘Meinongian quantification’. So we can distinguish between
normal cases of ‘exists’ and cases where the entity under investigation
lacks the property of existence, but nevertheless has some sort of being. We
have examined the problem of non-existence in the previous chapter, so
it is not our primary concern in this chapter. But are there any other rea-
sons to doubt the Quinean doctrine regarding univocalism about ‘exists’?
Note that the debate could be understood on two different levels. On the
one hand, ‘exists’ may refer to the use of the term in our current, actual
language. On the other, we may consider which language is the ‘best’ for
our theoretical or pragmatic purposes. So, whether ‘exists’ is univocal in
the first sense and whether it is univocal in the second sense could be
regarded as separate questions. We will primarily focus on the first sense of
42 Quantification and ontological commitment

the question, but when we get to discussing quantifier variance, the second
issue becomes more pressing.
On the face of it, it may seem preferable to ascribe different meanings to
‘exist’ in different contexts. For instance, when we say that numbers exist,
we do not seem to mean it in quite the same sense as when we say that
some sort of material object exists. Accordingly, perhaps the meaning of
the existential quantifier could be thought to depend on the type of entity
that we are quantifying over? After all, numbers are abstract objects, so
their ‘existence’, whatever it amounts to, would not seem to be quite like
that of material objects. The natural conclusion to draw from this is that
‘existence’ simply means something else in these cases; ‘exists’ is equivocal
rather than univocal. Attractive as this line of thought may be, Quineans
would not find it persuasive. The mistake, they would insist, is thinking
that existence is some kind of activity – something that a thing does. Peter
van Inwagen, commenting on Gilbert Ryle’s argument to the effect that
existence is equivocal, takes just this line and specifies further what the
univocacy of existence amounts to:

The thesis of the univocacy of existence […] does not imply that existence
comes in species or that ‘existent’ is a ‘generic’ word like ‘colored’ or
‘sexed.’ This thesis does not imply that there are or could be ‘species’
words that stand to the generic ‘existent’ as ‘red and ‘green’ stand to ‘color’
and as ‘male’ and ‘female’ stand to ‘sexed.’3

It turns out that the issue is rather more complicated than it might have
seemed at first. Indeed, to clarify what the Quinean position amounts to,
we should take aboard the thesis that quantification commits us ontologically
in addition to the thesis that it is univocal.
One way to understand the idea of ontological commitment is to think
about the truth-conditions of sentences.4 To cash out the ontological commit-
ments of a sentence, we need to determine what kinds of constraints the

Peter van Inwagen, ‘Being, Existence, and Ontological Commitment,’ in D. Chalmers,
D. Manley, and R. Wasserman (eds.),Metametaphysics (Oxford University Press, 2009),
p. 486.
For further discussion on this and other ways to understand ontological commit-
ment, see Augustín Rayo, ‘Ontological Commitment,’ Philosophy Compass 2/3 (2007),
pp. 428–444; see also Philip Bricker, ‘Ontological Commitment,’ in E. N. Zalta (ed.),
The Stanford Encyclopedia of Philosophy (Winter 2014 edn); see
3.1 The meaning of the existential quantifier 43

sentence’s truth imposes on the world. However, an important caveat that

should be noted here is that the ontological commitments of a sentence can
be distinguished from the assignment of truth-conditions to that sentence.
This is an idea taken up by Augustín Rayo, who argues that the machin-
ery a semantic theory uses in specifying truth-conditions (e.g., sets) is not
itself a constraint on the world. In other words, even if sets are used in
specifying truth-conditions, this does not itself constitute a demand on the
world. Moreover, on Rayo’s view, when we take a familiar quantificational
claim such as ‘Numbers exist’, that is, ‘ ∃x(NUMBER(x))’, it will depend on
our semantic theory as to exactly what the sentence is committed to. Rayo
himself has recently developed an interesting metaontological position,
but here we will focus on his critical remarks.5 The target of Rayo’s critique
is the metaphysicalist, who may be making the type of error we have just dis-
cussed: the metaphysicalist holds that a type of correspondence must exist
between the logical form of a true sentence and the metaphysical structure
of reality. This can be illustrated with an example. In order for the sentence
‘Elizabeth reads’ to be true, there must be a certain ‘metaphysically privi-
leged’ structuring of reality into the constituents of the sentence. First, let
us suppose that what we are referring to with the sentence is the fact that
Elizabeth reads. In this case, the constituents of the fact could be Elizabeth
and the property of reading (although this may not be the only way to do
it!). The important thing is that there is a correct way of structuring reality,
which will take into account the metaphysically privileged deconstruction
of the fact into its constituents. If there are several ways of deconstruct-
ing the fact, the metaphysicalist typically adds that only one of them is
fundamental.6 These are all issues to which we shall return, but note that
the metaphysicalist seems to be committed to something that we saw Rayo
resist earlier, namely that the truth-conditions of the sentence ‘Elizabeth
reads’ would now appear to impose direct constraints on the world, via
the metaphysically privileged structuring. So Rayo’s point is that the meta-
physicalist may be employing a problematic philosophy of language, since
the truth of a sentence does not in fact need to impose the restriction on
the world that the metaphysicalist assumes.

See Augustín Rayo, Construction of Logical Space (Oxford University Press, 2013). For dis-
cussion, see Matti Eklund, ‘Rayo’s Metametaphysics,’ Inquiry: An Interdisciplinary Journal
of Philosophy 57.4 (2014), pp. 483–497.
This simplifies matters a bit; compare with Rayo, Construction of Logical Space, pp. 6–7.
44 Quantification and ontological commitment

Let us return to the quantificational claim, ‘Numbers exist’. In contrast

with Rayo, Quine’s view is that such first-order quantificational sentences
always carry a commitment to the objects being quantified over. But it is
important to see that this does not necessarily follow automatically. For
further illustration, consider the following two sentences:7

(1) There is a pig on the sofa.

(2) A pig is on the sofa.

Sentences (1) and (2) seem to express the same fact. But (1) seems to carry
with it more existential ‘weight’ than (2), since it is natural to interpret it
as expressing the existential quantifier, whereas (2) does not immediately
strike us as containing an existence claim. As Tim Crane notes, (2) does of
course entail the existence of a pig and would normally only be uttered by
someone who believes that a particular pig – the one on the sofa – exists.8
But reflecting Rayo’s line of thought, Crane points out that we ought to dis-
tinguish between the content of a sentence and what the content of the sen-
tence entails. We can interpret both (1)and (2) as involving quantification,
since ‘a pig’ may perhaps be considered to function as a quantifier. But this
does not mean that (1) makes an existential claim unless a semantic link
exists between existence and quantification. (2) of course entails that a pig
exists and normally a person uttering (2) would surely not deny a commit-
ment to the existence of pigs. However, the point of saying that (1) makes
an existential claim relies on the thought that (1) and (2) express the same
proposition. Crane accepts that (1) and (2) say the same thing and even that
they both involve quantification, but the point he is pressing is that just
the fact that a claim is quantificational does not make it an existence claim.
Not unlike Rayo’s suggestion, semantic theory would be needed to estab-
lish such a link and both Crane and Rayo seem to be sceptical of the exist-
ence of such a link.
Furthermore, even if Quine is correct about the existential commitment
of first-order quantification, we must keep in mind that his claim does not
extend to all quantificational sentences, such as those involving modality.
We might say that ‘It is possible that there are such things as dragons’, and

These examples are discussed in Tim Crane, The Objects of Thought (Oxford University
Press, 2013), p. 43 ff.
Ibid., p. 44.
3.2 The existential quantifier and ontological commitment 45

a standard way of specifying the truth-conditions of such sentences is by

including those possible dragons as values of variables. But it does not yet
follow from this that the world must contain possible dragons. Rather, as
Rayo insists, we can only conclude that the semantic theory used is a com-
mitment to such possibilia. So Rayo’s lesson is that we can talk about the
ontological commitments of a sentence and the ontological commitments of
semantic theory, but these should be distinguished from one another. We do
not need to work out all the implications of these remarks for the Quinean
criterion of ontological commitment, but since there clearly are alterna-
tives to the Quinean criterion, then how did it come to be the default view
in the first place?

3.2 The existential quantifier and ontological


It is not difficult to see why something called the existential quantifier

would be considered to commit us ontologically to the existence of what-
ever it is that we quantify over. But it would be a mistake to think that this
is just due to the name of the quantifier. To be sure, the Quinean analysis
does suggest that the first-order existential quantifier is ontologically com-
mitting, but the history of this existentially loaded reading of the quanti-
fier reveals that before Quine and Russell, such a reading was by no means
the default.9 To see that existential quantification is not obviously exist-
entially loaded, consider the following example, due to Graham Priest:
‘I thought of something that I would like to give you as a Christmas pre-
sent, but I couldn’t get it for you because it doesn’t exist’. 10 This seems to
be a perfectly legitimate instance of quantification, but it does not appear
to be existentially loaded. On the contrary, it is explicitly said that what
is being quantified over does not exist. Moreover, this is not a particu-
larly odd statement by any means. I might be thinking of Escher’s impos-
sible staircases, for instance, and be playing with the idea of getting a
miniature version of one as a Christmas present for you. In this case, not

For a historical overview, see Graham Priest, ‘The Closing of the Mind: How the
Particular Quantifier Became Existentially Loaded Behind Our Backs,’ The Review of
Symbolic Logic 1.1 (2008), pp. 42–55.
Graham Priest, Towards Non-Being: The Logic and Metaphysics of Intentionality (Oxford
University Press, 2005), p. 148.
46 Quantification and ontological commitment

only does the entity in question not exist, but it could not exist. So it seems
that there is a natural use for existentially non-loaded quantification. But
perhaps there are some theoretical reasons to prefer existentially loaded
Before modern logic emerged, there is little reason to think that quan-
tification was used in an existentially loaded sense. Of course, Aristotle
and the mediaeval logicians did not talk about ‘quantification’, but the
principle was already in use. The idea of quantifying over something fea-
tured in the practice of determining the domain of discourse. According to
this approach, ‘existential’ quantification does not involve any special
connection with existence. Rather, ‘existence’ becomes a technical term
which should not be considered as ontologically loaded. Indeed, some
nineteenth-century logicians, such as Keynes and Venn, were very careful
to point out that in the context of logic ‘existence’ should be understood
simply as a sort of placeholder for whatever it is that we are currently
investigating.11 It was only at the beginning of the twentieth century that
Bertrand Russell knowingly started using existentially loaded quantifica-
tion. In the previous chapter it was mentioned that Russell was interested
in Meinong’s work, but ended up reacting to it quite critically. In fact, if
Priest’s interpretation is to be believed, Russell seems to have decided to use
existentially loaded quantification and only started developing arguments
to that effect as an afterthought.12 The result was that Quine had a natu-
ral ally in Russell against Meinong. In any event, Quine’s rather rhetorical
case against Meinong in ‘On What There Is’ has been hugely influential
and, perhaps partly because of Russell’s earlier influence, it was accepted
without much quarrel. But as we have seen, there are those who remain
sceptical about the ontological weight of the existential quantifier.
What is the alternative? If existential quantification were not consid-
ered existentially loaded, then perhaps we would do better if we called it
the ‘particular quantifier’ instead, as is sometimes suggested. In that case,
the quantifier should be read as ‘for some’, where it is left open whether
the entity being quantified over exists or not. This is the type of view that
Crane ends up considering as well:

For details, see Priest, ‘The Closing of the Mind,’ p. 46.
Ibid., p. 51.
3.2 The existential quantifier and ontological commitment 47

[H]ow should we represent ‘some’ and ‘exists’ in a formal or logical

representation of natural language? If we want to account for our initial
data (e.g. sentences like [Some characters in the Bible existed and some did
not]), we have a choice. We could translate ‘some’ as ‘∃’ in the usual way.
But in this case, we should not understand ‘∃’ as ‘there exists’; we should

express existence in another way. Or we could translate ‘there exists’ using

‘∃’, in accordance with Quine’s claim that ‘existence is what the existential
quantifier expresses.’13 But in this case, we should not understand ‘∃’ as
‘some’; we need another quantifier symbol for ‘some.’14

It is up to us to decide which way to go. Some, like Priest, have in fact

introduced a novel symbol for ‘some’ to make it clear that it is not being
used in the Quinean existentially committing sense. But we face this deci-
sion only if we reject Quine’s criterion of ontological commitment and its
close association with quantification. Note also that if the Quinean criter-
ion, which associates ontological commitment with (scientific) theories, is
rejected, then it remains unclear what ontological commitment amounts
to. Rayo’s analysis gives us some clues as to what the options are, but there
are plenty of other ways to go.
Crane himself is partly sympathetic to a proposal due to Zoltán Gendler
Szabó, according to which ‘ontological commitment’ is to be explained
with reference to the attitudes of thinkers.15 More precisely, Szabó suggests
that for a thinker to be ontologically committed is for her to ‘believe in’.
This corresponds partly with the ordinary use of ‘believing in’ – not in the
sense that one may, say, not believe in violence, but in the sense that the
object of belief is part of one’s ontology. Someone might say, ‘I don’t believe
in violence’, but that doesn’t mean that this person believes there to be no
violence! That is, violence can be a part of one’s ontology regardless of a
more abstract non-belief in violence. For believing-in to correctly represent
the world, one must have a true conception of whatever it is that one does
believe in. If one believes in Martians in Szabó’s sense, then one stands
in a certain relation to the term [Martians], where the square brackets

W. V. Quine, ‘Existence and Quantification,’ in hisOntological Relativity and Other Essays
(New York, Columbia University Press, 1969).
Crane, The Objects of Thought, p. 47.
Zoltán Gendler Szabó, ‘Believing in Things,’ Philosophy and Phenom enological Research 66
(2003), pp. 584–611.
48 Quantification and ontological commitment

indicate what the relevant plural term refers to. To put this a bit more
clearly: to think that Martians are green is to stand in a certain relation to
the proposition <Martians are green>. Now, for a thinker’s representation
of [Martians] to be correct, Martians must exist and at least part of one’s
conception of them should be true (e.g., Martians being green). Moreover,
we can think of Martians even if they don’t exist, because thinking of them
is just to stand in a certain relation to a proposition about them. But it is
a further question whether the proposition, say, <Martians are green>, is
representationally correct, or in other words, true. Szabó’s summary of the
view is as follows:

[T]he content of thinking of Fs is [Fs], and that [Fs] is representationally

correct just in case Fs exist and the conception of Fs is true. Putting these
things together yields the following result: if Fs exist but the conception of
[Fs] is false, then someone who believes that Fs exist would be right while
someone who believes in Fs would be wrong.16

In this way, one can say that there are true propositions about Fs (where
Fs are, e.g., Martians), while denying belief in Fs. This is possible if the
term is representationally incorrect: if there’s nothing that fits the ordin-
ary conception of Fs. Yet, once it’s agreed that there aretrue propositions
about Fs, that would suggest that one should believe that there are Fs. This
seemingly contradictory (and admittedly a little confusing!) situation is
resolved in Szabó’s theory because one can assume that the ordinary con-
ception of Fs does not apply to Fs. For instance, one could believe that
there are Martians, but deny that they are anything like what the stereo-
typical story about little green men suggests. The upshot that Szabó draws
is that ontological commitment cannot be as simple as the Quinean idea
of theory acceptance, since one may think that even a true theory mis-
represents some of its crucial terms. On Szabó’s terminology this would
mean that one should not believe in some of the entities that the theory
is – on Quine’s criteria – existentially committed to. Of course, Quine does
not talk about ‘believing in’, at least not in this fashion, so strictly speak-
ing Szabó’s theory is just an alternative way of understanding ontological
commitment, one that a Quinean could readily dismiss. Alternatively, if
a Quinean states: ‘I believe that Pegasus flies’, this would not entail that

16 Ibid., pp. 606–607.

3.3 Quantifier variance and verbal debates 49

she ‘believes in’ Pegasus; we can instead give a paraphrase of the sentence
following the Russelian account of definite descriptions. But at any rate,
it appears that there are a variety of alternative views regarding quantifi-
cation and its links or lack thereof to ontological commitment, as well as
some reasons to think that even the term ‘existential quantifier’ is itself
misleading. Regardless, we will continue to talk about the ‘existential
quantifier’ in this book, for problematic as the term and its connotations
may be, this is certainly the standard usage. However, the reader is advised
to remain wary about these matters in what follows.

3.3 Quantifier variance and verbal debates

There is a popular strand of the quantification debate that we have not yet
properly discussed, even though it focuses exactly on the issue of whether
(existential) quantification is univocal or not: quantifier variance. Since
quantifier variance is one of the central topics of metaontology, this is also
an important reason for considering the debate between Carnap and Quine
to be central in metaontology. But we do not need to dwell on Carnap
and Quine any longer, especially because we saw that the contemporary
reconstruction of their debate is problematic at best. Nevertheless, the
issue concerning the univocalism vs. equivocalism of quantification is, by
many, considered to be at the heart of metaontology. For instance, here is
Sider’s view on the matter: ‘the central question of metaontology is that
of whether there are many equally good quantifier meanings, or whether
there is a single best quantifier meaning’.17 Sider himself is strongly of the
opinion that there is indeed a single best quantifier meaning. Note that, to
be precise, we must keep in mind that the thesis that there is a single best
quantifier meaning is in fact different from the thesis that quantification
is univocal: all that Sider’s position requires is that in the context of ontol-
ogy – when we speak Ontologese (see previous chapter) – quantification is
univocal. So there could be cases where the meaning of quantification is
context-dependent and hence equivocal, but those cases do not undermine
the fact that quantification is univocal with regard to ontological questions.

17 Ted Sider, ‘Ontological Realism,’ in Chalmers, Manley, and Wasserman (eds.),

Metametaphysics, p. 397.
50 Quantification and ontological commitment

In Sider’s terminology, quantification ‘carves at the joints’, which is to

say that it latches on to the structure of reality in a reliable manner – quan-
tification tells us something about how things really are in reality. 18 For
Sider, this is what ontological realism amounts to, so he effectively identifies
univocalism about quantification with ontological realism. But this usage
of ‘ontological realism’ is somewhat problematic, for one could deny that
quantification is univocal or that it occupies such a central position in
metaontology, but still hold that reality has an objective, mind-independent
structure. The latter is a more traditional, broader sense of realism, some-
times called metaphysical realism. Moreover, ontological realism is sometimes
taken to be the view that ontological debates are, by and large, substantial
rather than merely verbal or linguistic, but again, this view is independent
of the question regarding quantification.19 We will return to these options
in much detail in Chapter 4, where some terminological clarification
regarding different metametaphysical positions is provided. Accordingly,
for now it will be better to abstain from using the label ‘realism’ in any but
the broadest possible sense, namely in the sense that ontological facts are,
in some sense, objective.
If Sider and van Inwagen are some of the loudest contemporary propo-
nents of univocalism about quantification (although Sider prefers to talk
about ‘single best quantifier meaning’ and hence there is an important
difference between him and van Inwagen), then Eli Hirsch is no doubt the
most vocal defender of equivocalism. One might even get a sense that the
metaontology literature has been dominated by a dialogue on this very
issue between Sider and Hirsch. But by now the reader should be aware
that we are treading on difficult terrain. As our survey of the Quine vs.
Carnap debate in the previous chapter showed, it is not always easy to
determine the source of the disagreement – and if this is so in the case
of metaphysics, the problem is even more acute in the case of metameta-
physics.20 However, let’s give the stage to Hirsch before we pass judgement
about this debate.

For a defence of this view, see Ted Sider,Writing the Book of the World (Oxford University
Press, 2011).
For a survey of all three options, see C. S. Jenkins, ‘What Is Ontological Realism?’,
Philosophy Compass 5/10 (2010), pp. 880–890.
A case in point due to Gerald Marsh, ‘Is the Hirsch–Sider Dispute Merely Verbal?’,
Australasian Journal of Philosophy 88.3 (2010), pp. 459–469.
3.3 Quantifier variance and verbal debates 51

As we might expect from a proponent of equivocalism, Hirsch calls his

position ‘roughly Carnapian’, although he immediately distances him-
self from Carnap with regard to a few issues. 21 One of these issues is that
Hirsch fancies himself a realist, whereas he takes Carnap to subscribe to
anti-realism. The sense of ‘realism’ that Hirsch has in mind is realism in
the broadest sense described above, namely he does not think that our
subjective, linguistic choices determine what exists in the world. Hirsch
claims that our linguistic choices do, however, determine what we mean
by ‘existence’. If we then take the meaning of ‘existence’ to correspond
to existential quantification, it seems that the meaning of the existential
quantifier can vary from language to language. This is the phenomenon of
quantifier variance that we have already discussed in passing. As a point
of interest, note that on the basis of quantifier variance, Hirsch would also
disagree with Rayo’s metaphysicalist, because it is central to his view that
none of the candidate quantifier meanings is metaphysically privileged in
the sense that Rayo’s metaphysicalist might suggest.
Quantifier variance is the first important feature of Hirsch’s view, but
there is a second, albeit related feature of his view, namely the thesis that
some ontological debates are ‘merely verbal’. More precisely, Hirsch argues
that debates concerning the ontology of physical objects are merely verbal.
One example is the debate about composition, which we discussed briefly
in Chapter 2 – we will return to this debate below. What makes a debate
merely verbal?22 One way that Hirsch puts this is that both disputants
ought to agree that the other disputant speaks the truth in their respec-
tive language. Of course, when we are dealing with typical philosophical
debates, it is quite unlikely that the disputants do in fact think that their
opponent is speaking the truth, hence the ‘ought’. So, ‘merely verbal’ is
used in a sense which suggests that the debate is not genuine, contrary
to what the disputants think – it is not substantial. In its simplest form,
this is easy to understand: a debate between two people is merely verbal if

Eli Hirsch, ‘Ontology and Alternative Languages,’ in Chalmers, Manley, and Wasserman
(eds.), Metametaphysics, p. 231.
We will here focus on Hirsch’s account. The literature on this topic is expand-
ing rapidly though, see for instance Kathrin Koslicki, ‘On the Substantive Nature
of Disagreements in Ontology,’ Philosophy and Phenomenological Research 71 (2005),
pp. 85–105; David Chalmers, ‘Verbal Disputes,’ Philosophical Review 120.4 (2011),
pp. 515–566; C. S. I. Jenkins, ‘Merely Verbal Disputes,’Erkenntnis 79.1 (2014), pp. 11–30.
52 Quantification and ontological commitment

they agree on objective facts, but use a different language to speak about
them. If the same word means different things in different languages,
then this may give rise to a very simple merely verbal dispute. If apples
are ‘apples’ in your language and ‘oranges’ in someone else’s, then this
is likely to cause some confusion. But once you’ve successfully translated
their ‘orange’ into your ‘apple’, then the disagreement dissipates. Hirsch’s
claim is stronger, though, as he wants to say that some debates are merely
verbal quite generally. It’s not just that two people may have a merely ver-
bal dispute; two groups of philosophers holding competing views can also
have such a dispute.
It is plausible that these two features of Hirsch’s view, quantifier vari-
ance and verbal disputes, are logically independent.23 Someone may defend
quantifier variance without thinking that many ontological debates are
merely verbal. Equally, someone may think that many ontological debates
are verbal, but resist quantifier variance. Given that we have already dis-
cussed the complications related to the interpretation of quantifiers in
some detail, it is the second aspect of Hirsch’s view that deserves more
attention than the first here. We will return to quantifier variance and
related issues in Chapter 4.
Let’s get a little deeper into the verbal dispute issue. Hirsch makes some
important qualifications to this view, the most important of which is the

In my view, an issue in ontology (or elsewhere) is “merely verbal” in the

sense of reducing to a linguistic choice only if the following condition is
satisfied: Each side can plausibly interpret the other side as speaking a
language in which the latter’s asserted sentences are true.24

What Hirsch insists is that a debate can fail to be merely verbal only if
each side in a debate is genuinely charitable and still fails to interpret the
assertions of the other side as being true in their respective language. This
is, roughly, Hirsch’s principle of charity. But note that Hirsch is open to there
being debates that are substantial – that is, debates that do not reduce to a
linguistic choice. This is another aspect in which Hirsch considers himself

As proposed by Matti Eklund, ‘Review of Quantifier Variance and Realism: Essays in
Metaontology,’ Notre Dame Philosophical Reviews (2011); see
Hirsch, ‘Ontology and Alternative Languages,’ p. 231.
3.3 Quantifier variance and verbal debates 53

to part ways with Carnap, whom Hirsch takes to make the sweeping claim
that all ontological debates are merely verbal. Interestingly, Hirsch men-
tions the debate between Platonists and nominalists as an example of a
debate that he considers not to be verbal. Recall that in Chapter 2 we used
the debate between Platonism and nominalism as an example of a debate
which a language pluralist might consider to be merely linguistic. In view
of this, Hirsch’s position would seem to amount to a type of restricted
or moderate language pluralism. Let’s try to understand this view a little
The first issue to note is that a merely verbal dispute cannot just be one
in which each side makes true assertions in their respective language. For,
as Hirsch notes, there could be a merely verbal debate even in cases where
the sides do not make true assertions: ‘There may be a verbal dispute about
whether Harry is running around the squirrel or merely running around
the tree containing the squirrel, when in fact Harry is home in bed and the
runner is someone who looks like Harry’.25 Accordingly, it should be suf-
ficient to consider a debate to be merely verbal if each side agrees that the
other party speaks the truth in their respective language. It then becomes
a matter of linguistic interpretation to determine whether a given dispute
is indeed merely verbal. This would leads us towards the view that issues
in metaontology can sometimes be settled in terms of linguistic interpreta-
tion. Hirsch does not put it quite this way, but this is close to a view that
some philosophers working on the topic hold.
One of Hirsch’s primary examples of a merely verbal debate is the
one between perdurantists and endurantists. This is a classic debate about
the spatiotemporal parts of material objects: everyone would agree that
material objects have spatial parts, but do they also have temporal parts?
In other words, how do objects persist through time? Perdurantists hold
that material objects have temporal parts in addition to spatial parts and
hence that objects ‘perdure’. Endurantists hold that material objects do
not have temporal parts, which leads them to think that objects are wholly
present at any time that they exist; they ‘endure’. Due to related issues con-
cerning space and time, these views are versions of four-dimensionalism
and three-dimensionalism, respectively. On the perdurantist view, mate-
rial objects can be thought of as a type of ‘spacetime worm’, since all the

Ibid., p. 238.
54 Quantification and ontological commitment

different temporal parts are also, in some sense, parts of the object. In con-
trast, the endurantist view attaches no special significance to past or future
temporal parts of an object – the object is there at a given time and that’s
it. This may look like a perfectly normal ontological debate, presumably
with some connection to physics, but it has become something of a test
case in metaontology – primarily because some philosophers, like Sider,
take it to be a substantial debate whereas some, like Hirsch, consider it
merely verbal.
Much ink has been spilled trying to settle the matter, but we will con-
sider a slightly different approach. The reason for this is that the perdur-
antism/endurantism debate may not be a particularly good example of a
substantial ontological debate in the first place. Why? Because there are
plausible arguments to the effect that the theories are in fact metaphysically
equivalent, as defined by Kristie Miller:

[T]wo theories will be metaphysically equivalent if they are empirically

equivalent theories for which there is a non-arbitrary mapping of the

sentences of one theory onto the sentences of the other theory that
preserves the truth values of those sentences. 26

Miller goes on to argue that perdurantism and endurantism, at least

when combined with similar assumptions about composition and other
ancillary issues, are equivalent exactly in this sense. We will skip her case
for the empirical equivalency of the views, since this is more of an issue for
(philosophy of) science. More importantly for our present purposes, Miller
argues that there is no genuine difference in the ontological elements that the
two competing theories postulate (even if there may be a difference in how
they talk about these elements). We can understand the idea with the help
of a simple illustration.27 When we watch TV, we see movement and think
that we are perceiving motion. But in fact it is only an illusion of motion,
since the movement on the TV just consists of a rapid sequence of static
images. The upshot is that we may think of four-dimensional temporal
parts succeeding one another as creating a similar illusion of motion rather
than revealing anything about the true ontology of spacetime. Likewise,

Kristie Miller, ‘The Metaphysical Equivalence of Three and Four Dimensionalism,’
Erkenntnis (2005) 62.1, p. 92.
For further details, see E. J. Lowe and Storrs McCall, ‘3D/4D Controversy: A Storm in a
Teacup,’ Noûs (2006) 40, pp. 570–578.
3.3 Quantifier variance and verbal debates 55

the three-dimensional picture, which suggests that an object endures

throughout the period of time it moves, can be understood simply as a dif-
ferent perspective regarding the same ontological elements.
Entrenched proponents of perdurantism/endurantism may not be satis-
fied with this story, but we need not concern ourselves with this – it suffices
that we have established how controversial the example is. But if we were
to accept the metaphysical equivalency of perdurantism and endurantism,
then would that not support Hirsch’s view? Not all the way: most if not all
ontological realists would accept that some ontological debates are merely
linguistic, but Hirsch extends the claim to many other debates concern-
ing ordinary physical objects. For instance, Hirsch thinks that the debate
about when a collection of material objects composes a further object is a
prime example of a merely verbal debate. This debate about mereology and
composition is one that we briefly touched upon in the previous chapter,
in connection with van Inwagen’s ‘Special Composition Question’ (SCQ).
The two alternatives that we discussed were universalism and nihilism – the
first view takes it that a collection of material objects always composes
a further object whereas the second view holds that composition never
occurs.28 Hirsch specifies that ‘The dispute over nihilism that I view as
verbal occurs when both sides agree that there are simples and disagree
about whether there are composites’.29 In this case, there is perhaps more
at stake than in the perdurantism/endurantism debate, as the proponents
of nihilism will end up saying, for instance, that there are no such things as
tables (there are just ‘simples arranged tablewise’), whereas the universalists
must say that even something like the combination of your nose and the
Eiffel Tower composes a further object. Hirsch himself thinks that on a
charitable interpretation, the nihilists and universalists can agree, especially
because they already agree that in ordinary language we talk as if there
were things like tables and chairs but we don’t typically consider that any
odd combination of material objects should be considered a further object.
Of course, the proponents of these seemingly odd views can resort to
the methodological move that we also discussed in the previous chapter,
namely to insist that in the ‘ontology room’, the debate between nihilists

The SCQ originates in Peter van Inwagen’s Material Beings (Ithaca, NY: Cornell University
Press, 1990).
Eli Hirsch, ‘Physical-Object Ontology, Verbal Disputes, and Common Sense,’ Philosophy
and Phenomenological Research (2005) 70.1, p. 68, fn. 2
56 Quantification and ontological commitment

and universalists can be taken seriously. But there may be a better move
available, as Hirsch’s opponent only needs to provide some sort of plausi-
ble criterion for when composition occurs and then the debate turns on
whether everyone accepts that criterion or whether there are other, com-
peting criteria available. For example, one might suggest that composition
occurs only when the combination of the various candidate objects pos-
sesses novel causal powers over and above of the causal powers of the can-
didate objects on their own.30 This type of criterion is testable: we can study
each of the candidate objects to determine their individual causal pow-
ers and then study the presumed composite object that these candidate
objects compose. If the causal powers of the presumed composite object
are not merely a conjunction of the causal powers of the candidate objects,
then the criterion under consideration would suggest that the composite
object does indeed exist. Here we have a genuine difference between the
ontological elements of two competing theories, namely the set of causal
powers with which composite objects are endowed. Moreover, Hirsch’s
condition for a merely verbal debate is not obviously satisfied, because the
role of causal powers in the suggested theory is not something that can
easily be considered simply as a different language or a different way of
speaking of the same ontological elements. However, in fairness it should
be noted that the theory regarding the causal powers of composite objects
is certainly a controversial one – for one thing, it’s not clear which causal
powers we should focus on (some causal powers could be considered trivial
when it comes to composition).
Unfortunately, the task of settling the Special Composition Question –
or even whether it really is an ontologically substantial question – is far too
great for us to undertake here, so we will have to just take it as a lesson
in the convolutions of metametaphysics. The debate between Hirsch and
Sider is ongoing. In his most recent rejoinder to Hirsch, Sider argues that
‘Ontologese is supposed to be just like English except for the semantics of
quantifiers’.31 The idea is that one could suspend Hirsch’s requirement for
charity, at least in those contexts where quantification is involved. Here it
is worth noting again that we have been focusing on the status of quan-
tification in our actual language. If we instead consider the best possible

For one version of such a view, although restricted to persons, see Trenton Merricks,
Objects and Persons (Oxford University Press, 2001).
Ted Sider, ‘Hirsch’s Attack onOntologese,’ Noûs 48.3 (2014), p. 567.
3.4 Beyond existence questions 57

language for the purposes of philosophy/science, then one could in prin-

ciple agree with Sider – in the best language, ontological disputes are plaus-
ible, not merely verbal. But Hirsch (or someone else) might still insist that
our actual language just isn’t like this: we are not in fact using the best
possible language, Ontologese, but rather a much inferior one. In any case,
Sider is primarily putting forward a conditional claim: if quantification
‘carves at the joints’, argues Sider, then it is not subject to the type of vari-
ation in meaning that some other terms, like ‘chair’ – and perhaps even
composite objects in general – are. Sider thinks that the same goes for
certain scientific terms, such as ‘electron’. However, we cannot really do
justice to Sider’s positive project before we have a clearer idea about the
metaontological machinery he uses, such as the appeal to ‘joint-carving’.
But there is one reaction to this debate and its ilk that we still ought to
take into consideration in this chapter, namely Fine’s influential claim that
metaontology should not have existence questions as its primary focus.

3.4 Beyond existence questions

We have seen that many metaontological debates concern questions about

the existence of different types of objects. The debate about when a num-
ber of objects compose a further object is of this type as well, since we
are looking for existence-conditions for composite objects. But the most com-
monly cited example is surely the existence of numbers, which we have
already encountered repeatedly. The question ‘Do numbers exist?’ is a typi-
cally philosophical one, as it would rarely be asked outside the ‘ontology
room’. The metaontological issue is whether such questions, which may
seem trivial, are substantial questions at all. The number of papers written
on the subject is certainly substantial and some philosophers think that
we ought to reconsider the debate altogether. In the preceding discussion
we have assumed that the manner in which Quine understands the task of
ontology is at least broadly correct, even if there are several issues with the
details. This understanding suggests that ontological questions are quanti-
ficational questions. So when philosophers ask ‘Do numbers exist?’, they
mean to ask: ‘∃x(x is a number)?’
In what might be labelled a ‘neo-Aristotelian’ sprit, Fine has argued that
rather than focusing on the quantificational question about the existence
of numbers, the initial ontological question should be replaced with one
58 Quantification and ontological commitment

focusing on the nature of numbers. The core of this alternative approach is

that the question about the existence of numbers is indeed trivial, at least
in a sense, whereas there are still questions to settle regarding the onto-
logical status of numbers. The way Fine puts this is that the question is
about whether numbers are real:

The realist and anti-realist about natural numbers, for example, will most
likely take themselves to be disagreeing on the reality of each of the
natural numbers – 0, 1, 2, … ; and this would not be possible unless each
of them supposed that there were the numbers 0, 1, 2, … It is only if the
existence of these objects is already acknowledged that there can be debate
as to whether they are real (Quine’s error, we might say to continue the
joke, arose from his being unwilling to grasp Plato by the beard).32

Fine suggests, controversially, that existence should be considered as a

predicate: if we wish to ask whether integers exist, we should not formu-
late the question like this: ‘∃xIx?’, where ‘I’ refers to integers, but rather like
∀ ⊃
this: ‘ x(Ix Ex)?’, where ‘E’ is the predicate for existence – the question is
not whether some integer exists, but whether every integer does. Consider
Fine’s example concerning realism about integers vs. realism about natural
numbers. Fine takes it that realism about integers is intuitively the stronger
position, because it is committed to the existence of both the integers and
the natural numbers: if all the integers exist, then all the natural numbers
do as well, because integers contain the natural numbers. Realism about
natural numbers is clearly the weaker position, as it implies only a commit-
ment to natural numbers. But on the existence-quantificational construal it
turns out that realism about natural numbers is the stronger position: real-
ism about integers claims that there is at least one integer (negative or not),

whereas realism about natural numbers claims that there must be at least
one non-negative integer, specifically a natural number. So Fine claims that
the existence-quantificational account of ontological commitment gets the
logic of ontological commitment wrong. This would seem to be an artefact
of using the existential quantifier instead of expressing the thought that all
the integers or all the natural numbers exist.
As a possible rejoinder, Fine considers an appeal to a theory associated
with a kind of object F, call the theory ‘TF’. TF states F’s existence conditions.
32 Kit Fine, ‘The Question of Ontology,’ in Chalmers, Manley, and Wasserman (eds.),
Metametaphysics, p. 169.
3.4 Beyond existence questions 59

Now, if Fs are integers, then TF will state their existence-conditions, and it

might be thought that Fine’s analysis was simply too detailed – just a com-
mitment to TF will do if we want to express realism about integers, we need
not even know the exact content of the theory. But it’s not at all clear how
this would help, since both realists and anti-realists will surely believe in
the truth of TF, if it is a true theory to begin with.
The upshot of Fine’s analysis is that the existence-quantificational
approach really can’t tell realism and anti-realism apart. Instead, Fine takes
the realism/anti-realism debate to be concerned with which objects are
real, and in order to have this debate in the first place it must be assumed
that the objects under investigation exist. The joke about Plato’s beard in
the earlier quotation highlights just how different this approach is from
Quine’s, since Fine takes it that the great puzzle that Quine was preoccu-
pied with has a rather simple solution: realism in the usual sense must be
assumed at the outset.
By way of motivating Fine’s view, consider what follows if we agree that
answers to quantificational questions are often trivial. Surely, answers
to ontological questions had better not be trivial. Of course, there are
non-trivial questions that fit this pattern, such as ‘Does the Higgs particle
exist?’ This is a highly non-trivial and clearly important question. But since
this question has nothing to do with metaphysics, the worry arises that all
non-trivial existence question are answered by the special sciences, so what
is there for philosophy to contribute? If Fine is correct, there are plenty of
non-quantificational questions to address after we have settled the quan-
tificational ones. Fine’s judgement is that Quine’s approach to ontology is
based on ‘a double error’. On the one hand, Quine asks a scientific question
(numbers are indispensable for science) rather than a philosophical one; on
the other, Quine uses philosophical machinery (existence-quantificational
tools) to answer that question. This analysis applies also to the other cen-
tral questions of the metaontology literature that we have discussed, as
Fine observes:

I believe that the case of mereological sums and temporal parts has been
especially misleading in this regard. For the question of their existence has
often been taken to be a paradigm of ontological enquiry and, indeed, it is
this case more than any other that has given rise to the recent resurgence
of interest in meta-ontology. But the case is, in fact, quite atypical since it is
60 Quantification and ontological commitment

one in which the quantificational question is also philosophical and hence

is much more liable to be confused with the ontological question[.]33

If Fine is correct, we may have done metaontology something of a disser-

vice by having discussed the case of mereological sums and temporal parts
even to the modest extent that we here have. However, given the extensive
literature on these topics and the fact that these debates are by no means
at its end, it will be useful for anyone interested in metametaphysics to
have a basic grasp of the debates. Fine’s own analysis of the cases of mereo-
logical sums and temporal parts is as follows. At one point in history, equa-
tions such as ‘x2 = –1’ did not have a solution. But after the introduction
of complex numbers, we have been able to solve such equations. In other
words, we can extend the ‘domain of discourse’ to include such things like
complex numbers. Fine suggests that the upshot of this is that there really
is no substantive question as to whether there are complex numbers – we
could have decided not to introduce them, after all. Similarly, Fine thinks

that the existence of arbitrary mereological sums or temporal parts are

subject to a decision about domain extension. If we wish, we can extend the
domain of discourse in such a way that the mereological sum consisting of
your nose and the Eiffel Tower is included. But it makes little sense to ask
whether such strange objects really exist – the only substantive question
that remains is whether the domain can be extended in such a manner
(and it appears that the answer is ‘yes’). Hence, on Fine’s view, these ques-
tions do not appear to be particularly interesting or important questions
in metaphysics.34
What remains to be done in this section is to outline Fine’s positive
view. It was mentioned in passing that Fine thinks ontology should be pri-

marily interested in what is real rather than what exists. This type of move
is required, since when existence is taken as a predicate – as Fine pro-
poses – realism in the usual sense becomes trivial. So if we continue
to use the existential quantifier, we must introduce a ‘thick’ sense of
existence to distinguish realism and anti-realism. But instead of talk-
ing about ‘existence’, which would admittedly be rather confusing given
Fine’s approach, he proposes to use the term ‘real’ for this ‘thick’ sense
Fine, ‘The Question of Ontology,’ p. 159.
But for a recent attempt to defend the importance of existence questions in metaphys-
ics, see Chris Daly and David Liggins, ‘In Defence of Existence Questions,’ Monist 97.7
(2014), pp. 460–478.
3.4 Beyond existence questions 61

of existence. Accordingly, instead of being committed to the existence of

numbers, a realist about numbers in Fine’s sense would be committed
to numbers being ‘real’. A range of positions about the reality of ‘Fs’ can
thus be formulated, where ‘R’ is the reality operator:

 Thoroughgoing realism: ∀x(F x ⊃ Rx).

 Thoroughgoing anti-realism: ∀x(F x ⊃ ∼Rx ).

 An intermediate position where some Fs are real and some are not; G
being the dividing line: ∀x(F x ⊃ (R x ≡ Gx)).

In contrast, the quantificational account presents realism and anti-realism

as contradictories with no room for an intermediate position: ∃xFx vs.
∼∃xFx. Fine concludes that the realist and anti-realist about natural num-
bers disagree about the reality of each natural number, but this would not
be possible if natural numbers didn’t exist!
Tempting as Fine’s account may be, the ‘thick’ reality operator is obvi-
ously in need of further explication. Fine does make an attempt to clarify
the idea. First, he suggests that the reality operator R[…], where ‘…’ stands
in for a sentence, expresses what is ‘constitutive of reality’. Metaphysics
is concerned with how things stand in reality and thus this account helps
explain how ontology is part of metaphysics. The objects of our ontology
will be those that figure in ‘…’ – and this, Fine suggests, gives us a complete
ontology. We can’t ‘beef up’ ordinary quantificational claims into ontologi-
cal claims with R though, for instance, ‘R[∃xFx]’ – the prefixed claim is
neither necessary nor sufficient for a commitment to Fs:

It is not necessary since affirming the reality of an F is compatible with

denying the reality of an existential fact of the form ∃xFx on the grounds

that it is only the underlying particular facts which are real; and it is
not sufficient since affirming the reality of there being an F would be
compatible under a ‘bundle theory’ which only recognizes the reality of
general facts with denying the reality of any particular fact of the form F x.
And, similarly, it would seem, for any other prefixed claim.35

Fine’s suggestion is ingenious, but it may not be convincing to someone

who isn’t sympathetic to realism to begin with. At the very least, we need
to further clarify what realism itself amounts to, and we also need to hear
more about the interpretation of Fine’s reality operator. Fine’s work on

Fine, ‘The question of ontology,’ p. 173.
62 Quantification and ontological commitment

grounding may be of some help in this regard, but we will postpone discus-
sion of this notion until Chapter 5.36
Fine, we should note, is not alone in holding the view that metaphysics
should not focus on existence questions. This is also the view of Tim Crane,
who puts it as follows:

Recent philosophy has tended to express ontological or metaontological

debates in terms of quantification. […] If I am right in what I have argued
here [see sections 2.3 and 3.2 for Crane’s arguments], then this can’t
strictly speaking be the central question, if what is at issue is the meaning
of natural language quantifiers like ‘some’. Natural language quantifiers
are not ontologically committing, so metaontology cannot be about the
meanings of natural language quantifiers.37

In this regard, Crane is strongly opposed Sider’s suggestion, according to

which the central question of metaontology is whether quantification is
univocal or equivocal. This, as we saw, is indeed the central issue in Sider’s

debate with Hirsch. What, then, is Crane’s positive view about metaontol-
ogy? He is sympathetic to the line that Fine (and others, such as Jonathan
Schaffer) takes, namely that ontology should be concerned with what is
fundamental, what really exists. But as Crane notes, Sider’s view as well can
be translated into something similar: ‘the central question of metaontol-
ogy is whether there are many different kinds of being (or existence), or
whether there is one kind which is fundamental’. So Crane, Sider, Schaffer,
and Fine, as well as many others, would all be in broad agreement about
there being some kind of a privileged type of being – the fundamental type
as opposed to the derivative type – which is of special interest in metaphys-
ics. But this is a topic that we will postpone until Chapters 5 and 6. Note
though that someone like Hirsch would be unlikely to think that any sig-
nificant progress has been made here: why should we think that the debate
about what really exists is any less verbal than the debate about what exists?
Indeed, it’s not obvious that Fine and others have successfully undermined
this type of deflationary reaction, which suggests that many or even all onto-
logical debates are not substantial. This type of reaction comes in many

For further discussion, see also Kit Fine, ‘What Is Metaphysics?’, in T. E. Tahko (ed.),
Contemporary Aristotelian Metaphysics (Cambridge University Press, 2012), pp. 8–25.
Crane, The Objects of Thought, p. 51.
3.4 Beyond existence questions 63

varieties though and not all of these are compatible with Hirsch’s own pic-
ture – this topic is central in the next chapter.
Now that we have covered the topic of quantification and ontologi-
cal commitment from both a historical and a contemporary angle, it is
time to move into a more general topic, namely that of identifying the
different metaontological positions, such as realism, deflationism, and
4 Identifying the alternatives:
ontological realism, deflationism,
and conventionalism

This chapter examines various metametaphysical positions in a broader

sense than the two previous chapters. At this point in the book, we do not
yet have the full ‘toolbox’ of metametaphysics in use; but with a sense of
the Quine vs. Carnap debate, the role of quantification, ontological com-
mitment, and existence questions, we can put forward a rough classifica-
tion of views. In the first section we will discuss ontological realism and
ontological anti-realism. Ontological realism, in its various forms, is no
doubt the dominant position. However, although the majority of metaphy-
sicians call their view some kind of realism, we will see that the notion
of ‘ontological realism’ can be specified in a number of ways, not all of
which accurately reflect the views of those who claim to defend some form
of realism. Typically, ontological realism is traced back to Quine whereas
ontological anti-realism is thought to have Carnapian roots, but given the
discussion in the last two chapters, we’d better not make such associations
lightly – especially as it turns out that Quine and Carnap may not have dis-
agreed as fundamentally as it is sometimes suggested. In this chapter, the
historical context will be set aside, as the primary goal is terminological
The second group of views, discussed in the second section, is the range
of deflationary approaches to metaphysics. The notion of ‘deflation’ con-
trasts with ‘inflation’ – one way to understand this contrast is to think
of the ‘seriousness’ that is attached to ontological questions. Accordingly,
an inflationary view will ‘inflate’ the seriousness of ontological disputes,
claiming that there is truly something of substance at issue when we have
ontological disputes. This can go all the way up to the ‘full-blown’ – to
extend the analogy – ontological realism typically defended for instance
by ‘neo-Aristotelians’. Adeflationary view, in contrast, will ‘deflate’ the ser-
iousness of ontological debates. However, as we will see, characterizing

4.1 Ontological realism and anti-realism 65

these views in such terms is not so straightforward. Some deflationist

positions consider many of the typical ontological questions to be uninter-
esting or misguided, since there is generally an easy answer available.
These answers may draw on simple conceptual and/or empirical work. This
type of deflationism is sometimes labelled the ‘easy’ approach to ontology.1
But note that deflationism as such is not directly opposed to (all forms of)
realism – as we saw in the previous chapter, Eli Hirsch is one example of a
philosopher holding a type of deflationary position while also defending a
(modest) form of realism.
The third family of views to be discussed is opposed to realism: extreme
conventionalism, to be discussed in the third section. Extreme convention-
alism is explicitly defended only by very few philosophers, but it is not
untypical to see various philosophers going some way towards such a view,
or putting forward views which could be subjected to ‘slippery-slope’ argu-
ments that lead to extreme conventionalism.
In the fourth section, we will discuss one version of ontological realism
in more detail, primarily as an exercise in assessing the relative merits of
different metametaphysical positions. The focus here will be Ted Sider’s
version of ontological realism. The chapter concludes with a brief assess-
ment of the various options discussed.

4.1 Ontological realism and anti-realism

One might think that defining ‘ontological realism’ should be relatively

easy, at least if the notion is on a par with the notion of ‘realism’ without
the qualification ‘ontological’, or perhaps replaced with ‘metaphysical real-
ism’. Indeed, ‘metaphysical realism’, even though its precise formulation
is certainly much debated, can at least be defined at such a high level of
generality that most philosophers would be able to agree on its definition.
At the highest level of generality, ‘metaphysical realism’ could be under-
stood as the view that there is a mind-independent, objective reality. But
disagreement enters the picture immediately when the epistemic dimen-
sion is opened, that is, when we start to discuss how we can know anything
about that objective reality. When we move to metametaphysics, this epis-
temic dimension starts to cause disagreement even earlier. For instance,

See especially Amie Thomasson, Ontology Made Easy (Oxford University Press, 2015).
66 Identifying the alternatives: ontological realism, deflationism

we could attempt to understand ontological realism by comparing it with

ontological deflationism of the ‘easy’ type mentioned in the opening of this
chapter. The easy approach to ontology suggests that many or all ontologi-
cal questions are epistemically easy to answer, whereas ontological realism is
often, though not always, combined with the view (or rather, an attitude)
that many or all ontological questions are difficult to answer. This appears
to be primarily an epistemic issue. Alternatively, since some ontological
deflationists hold that many or all ontological disputes are non-substantial
or misguided, then ontological realism could be understood as the view
that many or all ontological disputes are substantial or (synonymously)
serious. But again, there is clearly an epistemic dimension to the question
of whether a dispute is substantial or non-substantial. Moreover, it should
be noted that there is no clear reason why one would be compelled to think
that all ontological disputes must fall into just one group. One might, as
an example, think that the dispute about numbers is not substantial, yet
consider the dispute about three-dimensionalism vs. four-dimensionalism
to be substantial. So if the notion of ontological realism is attached to the
question of substantiveness, then one may be an ontological realist about
some things and an ontological deflationist about others.
It turns out, then, that ontological realism is not so easily defined. But let
us set aside the contrast with ontological deflationism for a moment, since
we will discuss it further in the next section. In any case, it would seem
to provide only a negative definition of ontological realism. Can we define
ontological realism positively? One way to attempt this would be to focus on
the question of quantifier meanings, which was discussed in theprevious
chapter. On Sider’s view, ontological realism amounts to the vi
ew that there
is a ‘single best quantifier meaning’, that ‘carves at the joints’. In other
words, the meaning of the existential quantifier is not context-dependent,
or even if it is, there is a way to pick out a meaning which is the best for the
purposes of ontology. We discussed the debate between Sider and Hirsch
on this issue of quantifier variance in theprevious chapter, but there are
reasons to think that attaching the notion of ontological realism to the
view that there is a single best quantifier meaning is not the most natural
usage of the notion, as it is rather strict. For instance, those, such as the

Ted Sider, ‘Ontological Realism,’ in D. Chalmers, D. Manley, andR. Wasserman (eds.),
Metametaphysics (Oxford University Press, 2009), pp. 384–423.
4.1 Ontological realism and anti-realism 67

‘neo-Aristotelians’, who do not regard existence questions to be at the heart

of metaphysics would likely object to defining ontological realism strictly
in terms of there being a single best quantifier meaning. But it does not
seem unreasonable to think that there is something important that those
influenced by Sider’s idea that quantification carves at the joints and those
of the ‘neo-Aristotelian’ ilk are inagreement about. Presumably, they would
agree about there being substantive ontological questions, so they are both
opposed to (some forms of) ontological deflationism. But we’ve already set
aside the attempt to define ontological realism merely as the view that
opposes ontological deflationism. Perhaps there is a third sense of ontologi-
cal realism that captures the relevant agreement?
Unsurprisingly, the sense of ontological realism that we are moving
towards must be very broad, but this is only to be expected of a label which
ought to encompass a variety of views. Why ‘ought to’? Because it would
be unreasonable for a relatively small group of philosophers to ‘steal’ a
notion which, on the face of it, is quite general indeed. At the very least,
we should attempt to find a sense of the notion that the mentioned groups
might be able to agree about, since there is no doubt that they agree about
something that certain other groups of philosophers would resist. Now, one
issue is that the area of agreement might be so basic that there is little to
discuss here. This may be the result if we characterize ontological realism
as a type of ‘knee-jerk realism’, which Sider describes as the view that the
point of human inquiry is to conform itself to the world rather than to make
the world.3 For Sider, knee-jerk realism is the starting point of inquiry – it
is not separately argued for. But we don’t want ontological realism to come
out as a primitive assumption that can’t be argued for. Rather, it seems
that Sider’s ‘knee-jerk realism’ is metaphysical realism of a more traditional
sort, to be separated from a more specific usage of ontological realism. In
that case, the primitive assumption of metaphysical realism is something
that we are already familiar with from the previous chapter, via Kit Fine’s
argument to the effect that realism in ‘the usual sense’ has to be assumed.4
Hence, it seems that knee-jerk realism is too broad to capture the relevant
sense of ontological realism.

Ted Sider, Writing the Book of the World (Oxford University Press, 2011), p. 18.
Kit Fine, ‘The question of ontology,’ in Chalmers, Manley, and Wasserman (eds.),
Metametaphysics, pp. 157–177.
68 Identifying the alternatives: ontological realism, deflationism

The three candidate senses for ontological realism that have been
outlined so far correspond to a large extent with those discussed by Carrie
Jenkins under the labels ‘ontological disputes are serious’, ‘one best quanti-
fier meaning’, and ‘ontological facts are objective’, respectively.To clarify,
the third view suggests that there are some ontological facts and they are
objective (so one could reject this view either by claiming that there are
no ontological facts or by claiming that they are not objective). As Jenkins
argues, it is possible to hold various combinations of these three views,
even though it sometimes seems that they are assumed to stand or fall
together. Someone could very well have quarrels with the idea of there
being one best quantifier meaning while accepting that there are objective
ontological facts and that ontological disputes are generally serious (i.e.,
substantial). This goes some way against Sider, who seems to think that the
seriousness of ontological disputes and the rejection of quantifier variance
go hand in hand, so that to reject seriousness is to accept quantifier vari-
ance.6 As we saw in the previous chapter, Hirsch would likely disagree, but
we do not need to dwell on the quantifier variance debate here.
What is the upshot? Jenkins suggests, roughly in agreement with Karen
Bennett and David Chalmers, that the best terminological practicewould be
to use the label ‘realism’ for the most general of the three positions, namely
‘ontological facts are objective’.7 ‘Anti-realism’ will then be the rejection of
this view, either by claiming that there are no ontological facts orby claiming
that ontological facts are not objective. One motivation whichJenkins states
in favour of this terminological practice is that we already have labels for the
two other possible senses that could be attached to ‘ontological realism’. The
first case we discussed – having to do with the seriousness or substantive-
ness of ontological debates– can be captured with the notion of ontological
non-deflationism (or ‘inflationism’), to be discussed further in a moment.
The second case, ‘one best quantifier meaning’, can be captured in terms

5 C. S. Jenkins, ‘What Is Ontological Realism?’, Philosophy Compass 5/10 (2010),

pp. 880–890.
In fact, it’s slightly more complicated than this, as Sider’s case has to do with the
importance of structure and ‘joint-carving,’ but we will return to this in more detail in
the fourth section of the chapter.
7 Jenkins, ‘What Is Ontological Realism?’, p. 888; David Chalmers, ‘Ontological

Anti-Realism,’ in Chalmers, Manley, and Wasserman (eds.), Metametaphysics,

pp. 77–129; Karen Bennett, ‘Composition, Colocation and Metaontology.’ in Chalmers,
Manley, and Wasserman (eds.),Metametaphysics, pp. 38–76.
4.1 Ontological realism and anti-realism 69

of the debate on quantifier variance; that is, Sider sides with the ‘quantifier
invariantist’, who subscribes to the view that there is one best quantifier
meaning, whereas Hirsch sides with the ‘quantifier variantist’. Given these
labels, which are already in use, the notion of ‘ontological realism’ remains
distinct from the two other issues. Jenkins’s suggestion seems reasonable
and is consistent with the initial, broad usage of the notion that we adopted
in the previous section. Hirsch likewise wishes to defend ontological realism
in this broader sense, so the disagreement between Sider and Hirsch is not
about ontological realism understood broadly.
It seems that we have some good reasons to follow this terminological
practice. But one issue still looms large: as noted earlier, there is the risk
that, on this usage, ontological realism becomes an issue that we cannot
easily argue for – a knee-jerk view that must be accepted at the outset, if
at all. Related to this, consider Bennett’s reaction to anti-realism (Bennett’s
anti-realist claims that there is no fact of the matter about whether or not
there are Fs, for a given F):

I do not know how exactly to argue against it, and I am not entirely sure
what it means. ‘There are Fs’ might be vague or ambiguous in some way,
in which case the unprecisified sentence might not have a determinate

This reaction leaves open the possibility that one could accept there to
be a fact of the matter about whether or not there are Fs given an appropri-
ate precisification; that is, given that we are able to make it precise, in what
manner the relevant existence question is to be understood. Yet, one might
think that this is such a difficult task that we will never be able to precisify
the question in a satisfactory manner. Strictly speaking, this view would
not appear to be anti-realism, however, as it is effectively being said that
there at least could be a fact of the matter about whether or not there are
Fs – we just don’t know. Bennett defines a closely related view called epis-
temicism: there is a fact of the matter about whether or not there are Fs, and
disputes about it are not merely verbal, but we have very little to justify
our belief in either of the options.9 When laid out like this, epistemicism

Bennett, ‘Composition, Colocation and Metaontology,’ p. 40.
Compare this with epistemicism about vagueness, the view that there is a sharp cut off
for the applicability of vague predicates such as ‘is bald,’ but we cannot possibly know
what that cut off is.
70 Identifying the alternatives: ontological realism, deflationism

is committed to ontological realism – the idea that ontological facts are

objective – but claims that settling ontological questions may be very dif-
ficult, at least in some cases. This is not a very radical position, of course,
and would likely be accepted by most. But Bennett herself is attracted to an
even weaker version of this position, which is not committed to ontological
realism. This weaker version of epistemicism suggests only that there are
claims of the form ‘there are Fs’ that are not merely verbal, and further,
that there is very little justification for believing either that these claims
are true or that they are false. So this version of epistemicism is fully com-
patible both with ontological realism and anti-realism – it is a purely epis-
temic position.10
Bennett’s epistemicism (in either one of its versions) is certainly not our
first glimpse of the intimate connection between metametaphysics and the
epistemology of metaphysics, but here we see it in a very clear manner. It
turns out that many metametaphysical positions make implicit epistemo-
logical assumptions, for example regarding our ability to know whether a
given dispute is substantial and when it has been settled if it is substantial.
Bennett has something further of interest to say in this connection. She
focuses on two popular test cases in metaontology, one which we have
already discussed: the problem of material constitution (can two objects,
e.g., a lump of bronze and a bronze statue, spatiotemporally coincide?) and
Peter van Inwagen’s ‘Special Composition Question’ (when do a number of
objects compose a further object?). Bennett’s claim is that epistemicism is
a reasonable position regarding these popular problems. Here’s why: we
have already done more or less all the work that we can on these topics! 11
So Bennett’s idea is that since so many clever minds have worked on these
topics for so long and we still don’t seem to have a determinate answer,
it’s unlikely that an answer is forthcoming. Hence, we should adopt epis-
temicism regarding these problems. She does however admit that some-
one might still come up with a new, persuasive line of reasoning, which
would change our epistemic situation. But this does admittedly seem fairly

For a view which takes there to be no fact of the matter about whether or not there
are Fs, hence taking existence questions to lack truth-value, see Stephen Yablo, ‘Must
Existence-Questions have Answers?’, in Chalmers, Manley, and Wasserman (eds.),
Metametaphysics, pp. 507–525.
Ibid., p. 73.
4.2 Ontological deflationism 71

unlikely with regard to these two problems which have, after all, received
the attention of our leading metaphysicians for quite some time.
Note that all this is neutral with respect to ontological realism vs.
anti-realism, on Bennett’s intended reading. Since we have little reason
to go one way or the other regarding these problems, the debates are sim-
ply underdetermined, they take no stand on whether or not there really are
objective facts. Having said that, the anti-realist could get some mileage
from this situation: isn’t it grist to the anti-realist’s mill that we have not
been able to settle these debates? What other evidence could there even be
for anti-realism? Is it not unreasonable to require that the anti-realist is able
to demonstrate that there could be no objective fact of the matter? Perhaps
so, but this line of argument is unlikely to persuade the ontological realist,
since the realist does think that she is able to make at least some progress
regarding these problems – they are at the very least tractable. Of course, if
Bennett is correct, then these debates are considerably less tractable than
the ontological realist may have led us to believe, but this does not auto-
matically vindicate the anti-realist position. So if we follow Bennett, we
should perhaps remain agnostic about these debates, at least until further
evidence emerges.

4.2 Ontological deflationism

Bennett’s epistemicism – although it is non-committal about ontological

realism – rules out by definition the view that the ontological debates con-
cerning composition and spatiotemporal co-location are merely verbal
disputes. But if we cannot make any epistemic progress regarding these
debates, isn’t it more likely because these debates are merely verbal rather
than it just being exceedingly difficult to settle them? This, at any rate, is
a line of thought that an ontological deflationist could entertain. A quanti-
fier variantist like Hirsch could argue to this effect. However, since we have
already discussed the debate between Sider and Hirsch as well as Fine’s
reaction to this and similar debates in the previous chapter, we can now
focus on another branch of deflationism, due to Amie Thomasson. But first,
let us briefly consider ontological deflationism more generally: can we eas-
ily define it?
One attempt at a general definition of the core content of ontological
deflationism would be in terms of relative triviality: where most scientific
72 Identifying the alternatives: ontological realism, deflationism

discoveries involve finding out something new about the world, something
empirically tractable, ontological discoveries often seem to lack this type
of significance – indeed, it is not even appropriate to talk about ontological
discoveries. If we know all the properties of a lump of bronze, which hap-
pens to be in the shape of a statue, there seems to be little at stake when we
ask whether the lump and statue are two different things spatiotemporally
co-located or just one thing. The answer, one might think, reflects only our
semantic, conceptual, or linguistic preferences.
While the preceding is not exactly a definition of ontological deflation-
ism, it should at least make clear what motivates the view: deflationism
is the type of reaction that most of us engaged in metaphysics have heard
when trying to explain what it is that we metaphysicians actually do to
those not trained in the discipline (and indeed sometimes even to those
who are trained in it!). The issue is not, of course, quite as simple as this.
For one thing, we should note that the somewhat sceptical reaction of the
deflationist does not automatically lead to ontological anti-realism. Even
if the answers to some ontological questions are trivial or merely a matter
of semantic preference, it does not mean that there aren’t objective facts
concerning them. This is easy to see in the case of trivial answers, for we
can supposedly easily agree on what the fact is in these cases. The case of
semantic preference might seem to lead towards anti-realism, though – at
least the lack of any objective fact would perhaps explain why there is a dis-
agreement, say, about the co-location of the statue and the lump. But while
this type of semantic relativism is certainly compatible with anti-realism, it
does not entail it. This should already be clear from our discussion of epis-
temicism above, as we’ve seen that it is possible to remain neutral about
the ontological realism vs. anti-realism issue on epistemic grounds while
developing, say, a view about one’s semantic preference towards a given
ontological question. However, note again that Bennett’s epistemicism
rules out this type of semantic deflationism by definition; that is, her epis-
temicism assumes that (some or many) ontological disputes are not purely
verbal or relative to semantic preference.
So it is possible to separate ontological deflationism from the onto-
logical realism vs. anti-realism issue, but as a matter of fact, many onto-
logical deflationists, such as Hirsch and Thomasson, take themselves to
be committed to some version of ontological realism. One way to distin-
guish the type of ontological realism that deflationists may adopt and the
4.2 Ontological deflationism 73

type of ontological realism that we’ve seen Fine and Sider defend would
be to talk about lightweight and heavyweight realism, respectively, following
Chalmers.12 But since we’ve seen that it might be best to define ontological
realism simply as the view that there are objective ontological facts, this
distinction is perhaps not entirely helpful. Chalmers suggests that heavy-
weight realism comes with the additional commitment that ontological
questions are highly non-trivial, but this seems an unnecessary complica-
tion, given that we already have the notion of ontological deflationism (and
non-deflationism or inflationism), which corresponds with the lightweight
vs. heavyweight distinction. Accordingly, we will follow this more estab-
lished usage in what follows and leave the lightweight vs. heavyweight
distinction aside.
Despite its common-sense appeal, ontological deflationism faces an ini-
tial challenge: if we are to believe that many ontological questions – exist-
ence questions in particular – have a trivial answer, then how does the
ontological deflationist explain the fact that we often disagree about what
that answer is? A partial explanation might be that different people have
different semantic preferences, perhaps due to cultural norms and so forth.
Yet, philosophers typically claim to have arguments in favour of answer-
ing certain existence questions either positively or negatively. Indeed, one
might think that existence claims could never be trivial, unless they turn
out to be analytically true or false. But it is difficult to see how existence
claims could be analytically true or false. These types of considerations
partly motivate Chalmers to favour a type of anti-realism, but the same
considerations could of course motivate a version of non-deflationism as
well. Ontological deflationism is not quite so easily refuted though. As
we’ll see in a moment, Amie Thomasson provides a version of deflationism
whereby existence claims are not analytically true or false, but neverthe-
less turn out to have ‘easy’ answers. 13
Thomasson’s version of ontological deflationism does not rely on quanti-
fier variance, in contrast to Hirsch. This is one reason why we should distin-
guish between ontological deflationism andquantifier variance:one can be
a deflationist without being a quantifier variantist (whether the opposite is
possible, i.e., to be a quantifier variantist without being a deflationist, is more

Chalmers, ‘Ontological Anti-Realism,’ p. 78.
For the complete version, see Amie Thomasson, Ontology Made Easy. For a shorter over-
view, see her ‘The Easy Approach to Ontology,’ Axiomathes 19 (2009), pp. 1–15.
74 Identifying the alternatives: ontological realism, deflationism

controversial). On Thomasson’s view, ontological questions do have answers

(they are not moot), but they are not worthy of ‘serious’ debate because the
questions are typically not at all difficult to answer. So how do we answer
them? Usually the answer is reached with the help of a combination of con-
ceptual and empirical tools. Importantly, Thomasson regards these answer-
ing strategies to be so straightforward that they do not warrant philosophical
attention. Of course, this is not to say that allconceptual and empirical enquir-
ies are easy, but according to Thomasson there are reasons to think that the
typical philosophical existence questions that we have been discussing in this
book certainly fall into this group of ‘easy’ questions. One of these reasons,
or what at least motivates the reason, is metasemantic:
Thomasson holds
that there is a connection between ‘exists’ and ‘refers’ in such a way that
the truth-values of existence claims are a semantic matter (concerning ref-
erence), whereas existence claims themselves are ‘about’ the world. The real
work, then, is done at the level of semantics, which demands a theory about
the conditions under which our terms refer. A crucial part of Thomasson’s
approach is thus the project of detailing the conditions under which a term
may be properly applied, calledapplication conditions.
We need not go into all the intricacies of Thomasson’s metaseman-
tic approach, but it might be helpful to consider an example of how her
approach delivers the easy answers. Thomasson is mainly concerned with
sortals, so let us take a sortal such as ‘kangaroo’ as our example. In answer-
ing the question ‘Do kangaroos exist?’ we ought to start by considering the
application conditions for the term ‘kangaroo’. It seems clear, for instance,
that the conditions we normally associate with ‘kangaroo’ will be fulfilled
in a situation where we observe a large group of moderately sized brownish
mammals hopping on two well-developed legs in the Australian outback.
We might further engage in some empirical research to ensure that what
we are observing are indeed kangaroos rather than, say, wallabies, but this
can be quite straightforwardly determined, ultimately by a DNA analy-
sis if not otherwise. So it seems as if we can conclude, without any great
difficulty, that kangaroos exist. A similar strategy can be applied to most
sortals, such as ‘chair’ or ‘table’, and indeed to abstract objects and social
constructions such as ‘symphony’ and ‘corporation’.14 Ontological debates

For further discussion, see Amie Thomasson, Ordinary Objects (Oxford University
Press, 2007).
4.2 Ontological deflationism 75

concerning the existence of ordinary objects would thus seem to dissipate,

but surely there are cases where such easy answers are not forthcoming?
There are indeed cases in which Thomasson’s strategy does not provide
an easy answer. But she doesn’t regard this to be a flaw in the approach, but
rather an indication of an unanswerable question. On this view, any properly
formulated existence question can be answered, but the ‘proper formula-
tion’ requires that we are able to determine the relevant application condi-
tions. On Thomasson’s view, some of the problematic ontological disputes
involve terms for which application conditions have not been determined
and in fact cannot be determined, often because they involve a usage of the
term ‘object’ in such a way that application conditions for ‘object’ cannot
be determined (but note that it is controversial whether ‘object’ can be
understood as a sortal). For instance, when the nihilist about composition –
whom we encountered in Chapter 2 – says that there is no object composed
of the various parts of a table, but rather just particles arranged tablewise,
she might have different application conditions for ‘object’ in mind than
the universalist. More generally, an ontological dispute may turn out to be
merely verbal if the disputants have different application conditions in
mind (cf. quantifier variance).
The upshot of Thomasson’s easy approach is that things which exist
according to common sense, such as kangaroos, tables, chairs, and symph
ies, can easily be determined to exist, whereas dragons and unicorns do not
exist, because they lack sufficiently well-defined application conditions. For
instance, we could never complete the empirical part of the test required to
determine whether something fulfilling the application conditions for the
term ‘dragon’ exists, for there is no DNA test for dragons! Note however that
there may be application conditions for various fictional objects as well,
including fictional ‘dragons’. But for the purposes of the example we may
assume that on one, non-fictional understanding of ‘dragon’, a part of the
application conditions for the term would involve that dragons are some
sort of an animal. Yet, according to Thomasson there is nothing particularly
interesting, philosophically speaking, about these existence claims.
Thomasson suggests that the deflationism resulting from the ‘easy’
approach, whereby most existence questions can be straightforwardly
answered with the combination of conceptual and empirical tools, is
‘Carnapian’. This is partly because the focus is on language: existence
claims must be made using a language and once the relevant application
76 Identifying the alternatives: ontological realism, deflationism

conditions for the terms in that language have been settled, the exist-
ence claims are ‘internal’, more or less in the Carnapian sense. We do not
need to return to the complications involved with the interpretation of
the Carnapian position and the notion of ‘framework’, which were dis-
cussed extensively in the preceding chapters, but it is not difficult to see
that Thomasson’s approach could be appealing to those with Carnapian
sympathies. The view does have a clear point of connection with language
pluralism, as discussed in Chapters 2 and 3. Be that as it may, we now have
a fairly good idea about what ontological deflationism along Thomasson’s
line amounts to, if true: many or most ontological disputes, at least the
ones concerning existence claims of the usual type, are relatively easy to
settle. In some cases it may turn out that the dispute is not even substan-
tial, if the disputants have been using different application conditions for
the same term. But one advantage that Thomasson’s approach seems to
have is that even in these cases it is not all that difficult to come to an
agreement once the relevant application conditions are explicitly stated.
Of course, this still leaves the cases where it just isn’t possible to specify
application conditions in a satisfactory manner – Thomasson’s judgement
about these cases is that they are ‘unanswerable’, questions that on a closer
look turn out to be pseudo-questions. 15 But what if most of ontology turns
out to be confused in this manner? What if there really are no genuine
ontological questions, just pseudo-questions? What if we can carve up the
world any which way we like? Then we would end up with a more extreme
type of ontological deflationism: full-blooded conventionalism. This is the
view that we will examine in the next section.

4.3 Towards extreme conventionalism

Extreme conventionalism is not usually listed as one of the main contend-
ers amongst metametaphysical positions, nor is it typically discussed in
much detail separately from ontological anti-realism. But there are some
good reasons to discuss the view separately. For one thing, even though
anti-realism may naturally go with extreme conventionalism, it is implied
by it only if combined with some further assumptions. To see this, we will

See Amie Thomasson, ‘Answerable and Unanswerable Questions,’ in Chalmers,
Manley, and Wasserman (eds.),Metametaphysics, pp. 444–471.
4.3 Towards extreme conventionalism 77

have to lay out the position in more detail. First, let’s focus on a version of
conventionalism that starts from the apparent subjectivity of our manner
of classifying reality. A realist would think that even though some of our
classificatory efforts are indeed guided by mind-dependent reasons, such as
different pragmatic considerations, the majority of them are nevertheless
mind-independent. In other words, there really are such things as apples,
cats, mountains, and stars, and they have what we might call natural bound-
aries, something that separates them from other things. The realist would
consider the boundaries of these objects and kinds to ‘carve nature at the
joints’. It is not always easy to state the exact identity conditions of vari-
ous things that we believe to have natural boundaries, but the extreme
conventionalist thesis is that there are no such mind-independent identity
conditions and that all our efforts to determine the boundaries of things
are subjective.16 On this basis, it may seem that anti-realism immediately
follows, but this is not quite right, for it would be possible to maintain
that even if we are fundamentally unable to overcome our conventions
and hence state mind-independent identity conditions for anything what-
soever, this could simply be due to our psychological limitations. Thus, one
might adopt something along the lines of Bennett’s epistemicism, with-
holding judgement about realism vs. anti-realism. Perhaps this would be a
pyrrhic victory, though; honest ontological realism surely requires that we
can, at least sometimes, reach correct judgements about the structure of
reality, to carve at the true joints of reality.
What makes extreme conventionalism a metametaphysical position
rather than just a first-order metaphysical position? Well, one upshot of
extreme conventionalism is that many of our ontological debates turn out
to be non-substantial. For instance, the debate about composition to which
we keep returning would, on the conventionalist analysis, be merely due to
a clash between differentconventions about composition. Moreover, there is
nothing objective in the world that settles this clash between conventions,
for they have been adopted for some mind-dependent reason, such as a cul-
tural norm or a pragmatic choice. So the view at hand has metametaphysi-
cal implications that are very similar to quantifier variance, but the motive

For an entertaining construal of this type of view, see Achille C. Varzi, ‘Boundaries,
Conventions, and Realism,’ in J. K. Campbell, M. O’Rourke, and M. H. Slater (eds.),
Carving Nature at Its Joints: Natural Kinds in Metaphysics and Science (Cambridge, MA: MIT
Press, 2011), pp. 129–153.
78 Identifying the alternatives: ontological realism, deflationism

is very different. Indeed, we could discuss conventionalism under the rubric

of deflationism as well, distinct from both Hirsch’s and Thomasson’s ver-
sions, but it is perhaps better to further distinguish it from these deflation-
ist views, as the roots of the views are somewhat different.
What could possibly motivate extreme conventionalism? One argumen-
tative strategy relies on the fact that some boundaries are clearly not nat-
ural, but artificial: borders of countries are one obvious example. We may
also express the artificial/natural boundary distinction in terms of fiat and
bona fide boundaries, following Barry Smith and Achille Varzi.17 It is import-
ant to recognize here that the fiat/bona fide distinction applies equally to
the physical boundaries of objects and to the objects themselves: according
to the realist, the physical boundary of an apple, for instance, is a bona fide
boundary, and the individual apple is a bona fide entity. However, this is
exactly what the extreme conventionalist questions: when we look at the
apple closely enough, it is clear that, far from the smooth boundary that
it appears to have, we are in fact dealing with a very loose arrangement of
molecules, and further, with a swarm of subatomic particles. Moreover,
the debate concerning composition further supports the conventionalist’s
line of thought: when Tibbles the cat eats some fish, at what point does the
fish become a part of Tibbles? Which criteria we apply, the extreme con-
ventionalist will argue, is ultimately a matter of fiat; Tibbles may continue
to exist, but his identity conditions are not mind-independent. Varzi takes
this line of thought to its logical conclusion:

Consider the debate on unrestricted composition. There is no question

that we feel more at ease with certain mereological composites than
with others. We feel at ease, for instance, with regard to such things

as the fusion of Tibbles’ parts (whatever they are), or even a platypus’s

parts; but when it comes to such unlovely and gerrymandered mixtures
as trout-turkeys, consisting of the front half of a trout and the back half
of a turkey, we feel uncomfortable. Such feelings may exhibit surprising
regularities across contexts and cultures. Yet, arguably they rest on
psychological biases and Gestalt factors that needn’t have any bearing on
how the world is actually structured. 18

Barry Smith and Achille C. Varzi, Fiat
‘ and Bona Fide Boundaries,’ Philosophy and
Phenomenological Research 60.2 (2000), pp. 401–420.
Varzi, ‘Boundaries, Conventions, and Realism,’ pp. 144–145.
4.3 Towards extreme conventionalism 79

Varzi’s analysis may help us to get our feet back on the ground: is there
really any connection between how the world is structured and our evalu-
ation of things such as trout-turkeys? Varzi argues that there might not be,
since even though we initially feel uncomfortable about strange hybrids,
we have nevertheless welcomed, for instance, a variety of genetically
manipulated plant-hybrids, such as orange-mandarins and peppermint.
Indeed, our intuitions and feelings of discomfort should not be relied on
if we hope to determine the actual structure of reality, for it is true that
we are biased in our evaluations. Once again, we would appear to be on
our way towards a position not unlike the deflationism emerging from
language pluralism. But this time the view arises from the simple realiza-
tion that our judgements are thoroughly influenced by our psychological
biases. However, even though Varzi makes a commendable effort to enter-
tain the extreme conventionalist line of thought, it does become less plaus-
ible when we move from everyday examples and cases such as the debate
over composition to scientific examples.
Consider the Standard Model of particle physics and its spectacular suc-
cess in producing accurate predictions. The Standard Model gives us a list
of fundamental particles that do not seem to be subject to the type of argu-
ment that Varzi presents; that is, we can ‘look’ at electrons as closely as
we want, but if electrons are truly fundamental particles, then we will not
discover any substructure, anything that might cause us to think that there
are various candidates as to how the boundaries of the electron are to be
determined. Of course, if it turns out that electrons are not fundamental,
then we might have to look closer into their substructure to find truly fun-
damental particles – if there are any (we will return to the question of fun-
damentality inChapter 6). Moreover, there are various ways to interpret the
relevant physics, but none of these interpretations would seem to support
extreme conventionalism. Even views according to which there is strictly
speaking no particles, but merely some structure or relations, as certain
versions of structuralism hold, would fall short of extreme conventionalism.
Although the world might turn out to be merely a lump of dough on these
views, it would not be an amorphous lump, as the lump would have internal
structure, some ‘joints’ according to which we may structureit.19

For further discussion, see Tuomas E.Tahko, ‘Boundaries in Reality,’Ratio 25.4 (2012),
pp. 405–424. For a brief discussion of the notion of ‘amorphous lump,’ see Michael
80 Identifying the alternatives: ontological realism, deflationism

Accordingly, extreme conventionalism turns out to be a difficult to

view to hold consistently. But perhaps a slightly less extreme version of
conventionalism is more plausible? One version is due to Alan Sidelle,
whose conventionalism starts from modality, but ultimately extends also
to objects: Sidelle thinks that if modal properties are conventional, then so
are objects.20 Sidelle’s conventionalism may not be as extreme as the one
that Varzi entertains, but its metametaphysical implications are severe. We
cannot do justice to Sidelle’s detailed arguments here, but what we can do
is consider a test case and determine its metametaphysical implications.
According to conventionalism about modality, what is possible or necessary
is so because of the way we think about the world; our modal judgements
are based on conventions. So how does a conventionalist about modality
treat typical cases of metaphysically interesting modal truths, such as the
case of ‘Water is H2O’, which is commonly considered a metaphysical a
posteriori necessity? Sidelle puts it as follows:

[E]ach necessary a posteriori truth should be seen as derived from a

combination of an analytic principle of individuation that has empty
spaces to be filled in by empirical findings and a particular empirical
finding that of itself carries no modal weight. For example, in the case of
water’s being necessarily H2O, the analytic principle might be ‘Nothing
counts as water in any situation unless it has the same deep explanatory
features (if any) as the stuff we call ‘‘water,’’’ and the empirical fact, which
makes the result a posteriori, is that the deep explanatory feature of the
stuff we call ‘water’ is being composed of H2O.21

On Sidelle’s view, each example of the necessary a posteriori can be

divided into an a priori principle concerning the type of empirical fact
required to generate a necessary truth – the deep explanatory features –
and the empirical discovery of the relevant fact. Therefore, ‘the modal
force’ of a posteriori necessities comes from a priori principles, which Sidelle

Dummett, Frege. P hilosophy of Language, 2nd edn (Cambridge, MA: Harvard University
Press, 1981), p. 577.
For the connection between conventionalism about modality and objects, see Alan
Sidelle, ‘Modality and Objects,’ The Philosophical Quarterly 60.238 (2010), pp. 109–125.
Alan Sidelle, ‘On the Metaphysical Contingency of Laws of Nature,’ in T. S. Gendler
and J. Hawthorne (eds.), Conceivability and Possibility (Oxford University Press, 2002),
p. 319.
4.3 Towards extreme conventionalism 81

claims to be analytic principles that are true in virtue of our linguistic con-
ventions. There is an interesting complication to this story: what if the
conventions themselves had been different, for instance, what if the con-
vention that guarantees the necessity of ‘Water is H2O’ had been different?
Then it would seem that either water might have failed to be H2O or water
might not have necessarily been H 2O.22 This issue has one especially inter-
esting metametaphysical implication concerning the analysis of changing
conventions. If conventionalism (about modality) is true, then metaphysi-
cal debates concerning modality will presumably have different outcomes
when associated with different conventions. That is, if something like
Sidelle’s story is correct and ‘the modal force’ of necessary truths comes
from conventions, then a change in those conventions ought to have an
influence on the modal truths. Yet, as Sidelle argues, this is not quite so
To illustrate the problem, let us stipulate that the term ‘water’ refers
to the macrophysical, superficial qualities of water, such as being a clear,
tasteless liquid, etc. (let us ignore the various complications with such
phenomenal qualities for the moment). We can now imagine a counter-
factual scenario where another liquid with a different microstructure,
say, XYZ, as in Hilary Putnam’s traditional ‘Twin Earth’ scenario, falls
under the extension of ‘water’ due to having the same superficial quali-
ties. Now it seems that in the counterfactual scenario it would be true to
say that ‘There is water which lacks hydrogen’, but Sidelle insists that we
could not translate this sentence directly to our normal usage of ‘water’,
as there must be different rules governing the use of ‘water’ in effect
here. However, this failure of translation is not as radical as using the
sound ‘water’ (or its equivalent in other languages) for something quite
different, like overcooked spaghetti (Sidelle’s example!), since the usage
of ‘water’ in the counterfactual scenario is, after all, very much like our
use of ‘water’, given that we are referring to the superficial qualities that
XYZ also has.
What is the upshot? Well, it appears that there are different conventions
governing the use of ‘water’ for us and for the Twin Earth folk. So we cannot

This type of complication concerning conventionalism, about modality and pos-
sible strategies to address it, is discussed in Alan Sidelle, ‘Conventionalism and the
Contingency of Conventions,’ Noûs 43.2 (2009), pp. 224–241.
Ibid., pp. 233 ff.
82 Identifying the alternatives: ontological realism, deflationism

simply translate their true statement ‘There iswater which lacks hydrogen’ to
our equivalent statement, which is false.Thus, the difference in conventions
does not have any implications towards the essence of water after all. But
according to Sidelle, this doesnot undermine conventionalism, for it demon-
strates that we could have used thesame term in very similar circumstances
to generate different truth values for the same counterfactual and modal sen-
tences. What is crucial in this account is that Sidelle thinks we would not
have been making a metaphysical mistake despite the different truth-values.
Thus, the conventionalist can argue that
regardless of whether or not there are
mind-independent modal facts, it is the conventions themselves that gener-
ate ‘the modal force’; either there simply are no modal (or essentialist) facts
or if there are, then we can use ‘water’ just as well to refer to one such fact or
the other, without making any mistake (of a metaphysical sort). In the latter
case, we would end up with what Sidelle calls ‘metaphysical universalism’:

The Conventionalist notes that sincewe could have chosen– metaphysically

speaking – more or less any feature toguide our counterfactual usage, and
this then would have generated other, but parallel, true (and false) modal
sentences, then this story is going tohave to generate a sort of metaphysical
universalism, according to which whatever conventions we adopted, there
would have been a ‘mirror’ so-individuated entity or property to which these
rules determined reference.24

According to Sidelle, the fact that varying conventions fail to have modal
implications is due to a failure of translation, thus at the level of language
rather than ontological fact. Although, as Sidelle admits, if ‘metaphysical
universalism’ is true, then there is still a sense in which the realist would
be correct in saying that the source of modality is in the world – since there
are modal facts after all. But Sidelle thinks that this would still be compat-
ible with a type of conventionalism about modality, since any debate about
modal facts would nevertheless be answered with reference to our conven-
tions rather than the objective facts.
This compact presentation of Sidelle’s conventionalism about modality
is too brief to enable us to fully articulate the upshot of the position, but
for the purposes of our overview of various metametaphysical positions in
this chapter, we can already clearly see what kind of implications this line

Ibid., p. 235.
4.4 A case study: Sider’s ontological realism 83

of thought would have. Interestingly, one of these is, once again, a type of
epistemicism: Sidelle ends up accepting, although reluctantly, the possibil-
ity of a position according which there could be objective modal facts, even
though he insists that they would be inert when it comes to actual debates
concerning modality.

4.4 A case study: Sider’s ontological realism

As we saw earlier in this chapter, Sider understands ontological realism

somewhat differently (that is, much more strictly) from the broad sense
that we have adopted. But looking into his view in somewhat more detail
than we have done so far might be useful, for Sider’s work in this regard
has been very influential and anyone interested in metametaphysics is
likely to encounter his ideas repeatedly. Moreover, now that we have a
clearer idea about the various metametaphysical positions, we can apply
our understanding to Sider’s theory and assess it in the light of some of the
competing views. In his recent book, Sider sets up an ambitious project
in favour of realist metaphysics. 25 Sider’s central task is to defend realism
about structure , which he connects with the notion of fundamentality : ‘a fact
is fundamental when it is stated in joint-carving terms’.26 We will postpone
an analysis of fundamentality until Chapter 6, but all we need to know at
the moment is that, for Sider, fundamentality is a metaphysical notion,
which he uses ‘more or less interchangeably’ with ‘joint-carving’ and ‘part
of reality’s structure’. He wants to defend the view that some notions are
fundamental in the sense that there is a perfectly fundamental description of
reality expressible by terms that carve perfectly at the joints. Note also that
Sider is committed to a type of ‘levels’ of reality view, as he thinks that
there are degrees of fundamentality and hence degrees of joint-carving –
perfect joint-carving being at the fundamental level. All this talk about
levels and fundamentality will be clarified in Chapter 6, but let us try to
make some progress with regard to the notion of ‘joint-carving’, which is
often used in these contexts and which we have already come across in
this book.

Sider, Writing the Book of the World.
Ibid., p. vii.
84 Identifying the alternatives: ontological realism, deflationism

The srcin of the notion of ‘joint-carving’ is Plato’s dialogue Phaedrus,

which talks about dividing things by classes and natural joints. 27 The idea
could be described as follows. If one thinks that there is some structure
in reality, as Sider clearly does, then one way to describe this structure is
in terms of the ‘joints’ of reality. If we hope to talk about this structure,
to pick out these joints, then we must have some terms or concepts that
‘latch on’ to – that is, accurately describe – the joints. In other words, there
are some concepts that ‘carve’ nature or reality at its joints. Moreover, as
noted already, some concepts supposedly do this carving ‘perfectly’. What
this amounts to is a more difficult question, since we have not yet laid
out the full metametaphysical ‘toolbox’, but the driving idea is that some
concepts capture something about the most basic structure of reality.
The role of perfectly joint-carving terms in Sider’s theory is complex, but
here we are primarily interested in their role in securing the substantivity
of metaphysical questions. Sider regards it as a ‘near enough’ sufficient
condition for substantivity that a debate is cast in perfectly joint-carving
terms.28 Why only ‘near enough’? Because there may be multiple joints in
the vicinity (more on this later), or the joint-carving candidate may be of
the ‘wrong sort’:

The line through the Urals, though physically distinguished, isn’t

geographically or politically distinguished, and therefore seems irrelevant
to the boundaries of Europe. The physically significant line is indeed a joint
in nature, but it’s the wrong sort of joint in nature, relative to a dispute
over ‘Europe.’29

Sider goes on to offer a suggested answer as to what qualifies as the ‘right

sort’ of joint-carving candidate – and we will look into this answer in what
follows. But he does not think that this is a major problem, since he decides
to mostly ignore this complication. Sider’s central concern is to build a case
in favour of (what he calls) ontological realism, and against deflationism.
Sider’s definition of ontological realism, which we will call ‘Siderian onto-
logical realism’ to distinguish it from the use we adopted earlier, is roughly
as follows:

Plato, Phaedrus, translated by R. Hackforth (Cambridge University Press, 1972),
Sider, Writing the Book of the World, p. 71.
Ibid., p. 48.
4.4 A case study: Sider’s ontological realism 85

Siderian ontological realism (S-OR) Ontological realism says that the

crucial terms in ontological questions carve perfectly at the joints.

What are the crucial terms? Reflecting our earlier discussion, Sider
takes them to be quantificational . In fact, he states that ‘the thesis that
quantifiers carve at the joints is the best way to defend the substantivity
of ontological questions’. Ontological assertions are often of the form
‘∃xFx ’, or, ‘There are Fs’. Thus, the first-order existential quantifier is a
crucial term, and carves at the joints. If F also carves at the joints, then
the question of whether there are Fs is ‘guaranteed’ to be substantive; so
there is a tight connection between S-OR and quantification. In particu-
lar, Sider hopes to defend S-OR against deflationism by arguing that the
first-order existential quantifier carves (perfectly) at the joints. However,
Sider’s project is primarily methodological, as he is not concerned about
what the Fs are. Hence, we could attempt to reconstruct Sider’s key claim
as the following conditional claim about a fundamental language:

Fundamental language (FL) If L is a fundamental ( joint-carving)

language, then there are some perfectly joint-carving Fs that we can
successfully quantify over in L.

Accordingly, Sider is fallibilist about the details of fundamental ontol-

ogy, but he does insist that the language of our best science is a likely
candidate for a fundamental language – this assertion is evidence of Sider’s
naturalism; he is interested in laying out the ontological foundations of
scientific inquiry. But a point of interest, then, is what we can say about the
perfectly joint-carving Fs. If science is truly a fundamental language, then
it should be able to reliably quantify over Fs. We should at the very least

have some idea about how this happens. This could be done by pointing
out some Fs that carve perfectly at the joints. If no F carves perfectly at the
joints, then we have no cases where the debate over the existence of Fs is
guaranteed to be substantive, and indeed no clear sense as to what the sup-
posed fundamentality of the existential quantifier amounts to. Fortunately,
Sider gives us an example of a perfectly joint-carving candidate term,
namely ‘electron’ (or ‘being an electron’). To establish this, Sider ought to
show us that ‘electron’ latches on the ‘right sort’ of joint and hence can be

Ibid., p. 96.
86 Identifying the alternatives: ontological realism, deflationism

shown to capture a perfectly fundamental aspect of reality rather than a

mere semantic (i.e., conventional) bias.
Sider acknowledges that in ordinary language, for example when quan-
tifying over things such as tables and chairs, we are often not dealing with
perfect joint-carving. But he claims that perfectly joint-carving terms are
available when dealing with fundamental ontology – perhaps when quan-
tifying over subatomic particles like electrons. He does however admit that
there’s a caveat even with examples from fundamental physics:

When a term like ‘mass’ is introduced in physics, it’s intended to stand

for a fundamental physical magnitude, and so if there’s a joint-carving
property in the vicinity then that property is meant by ‘mass’, even if it
doesn’t quite fit the physicists’ theory of ‘mass.’31

This comes from a chapter of Sider’s book that deals with reference mag-
netism. The idea is that highly joint-carving candidate terms like ‘mass’ or
‘electron’ latch on to the joints of reality even if their usage, namely the
physicists’ theory of ‘mass’, does not quite reflect the actual joint-carving
terms ‘mass’ and ‘electron’. Recall that there can be multiple joints in
the vicinity of a perfectly joint-carving candidate, but any candidate suf-
ficiently close to a joint-carving one will do as long as there is a genuine
joint-carving property or object in the vicinity. So Sider’s view seems to be
that any characteristic feature that we actually use to identify electrons can
turn out to be mistaken. All that is required is that there is a joint-carving
category that is an approximate fit for actual usage.
Let us reproduce Sider’s key example, which focuses on the property of
being an electron.32 We have a theoretical term,E, which is intended to carve
at the joints. We also have a single element of structure, e (understood as a
feature of reality rather than a concept), which carves at the joints and is
associated with E by the relevant disputants (in an ontological or a scientific
debate). However,e only needs to satisfy enough of the ‘core theory’ typically
associated with E. By ‘core theory’ Sider means the defining characteristics
of E. In the case of ‘electron’, Sider lists mass, negative charge, and ‘being a
subatomic particle that orbits nuclei’ as part of the core theory. Note that
although Sider connects e with meaning, the issue is not conceptual; here
e would reflect the feature of reality expressed by the property of being

31 32
Ibid., p. 32. Ibid., pp. 48–49.
4.4 A case study: Sider’s ontological realism 87

an electron. Sider is interested in genuine, mind-independent joints in real-

ity, and e is supposed to pick out one such joint, whereas the core theory
ensures that E captures a sufficient number of the defining characteristics
of e to reliably latch onto it, in accordance with reference magnetism. This
model can be applied, for instance, to a debate about whether there are any
electrons in a certain space-time region, R. Since enough of the core theory
associated with ‘electron’ is satisfied by the property of ‘being an electron’,
we seem to have a substantive dispute in Sider’s sense. Sider concludes that
‘To have a substantive dispute about electrons, there must remain enough
common ground about what electrons are so that all disputants can be
regarded as talking about the same thing’.33
We could attempt to revise the core theory about electrons to produce
a more charitable reading of Sider’s example. Plausibly, the core theory
involves properties like having half-integer spin, and being governed by
the Pauli Exclusion Principle and Fermi–Dirac statistics. 34 But one might
still ask: what guarantees that there is a joint-carving category that is even
an approximate fit for the actual usage of ‘electron’ if we cannot point to
a single feature of electrons that is necessarily included in the core theory?
This is not a question that Sider considers, as he is committed to full-blown
externalism. For Sider, it does not matterwhich features of electrons are part
of the core theory, as long as all the disputants can be regarded as talking
about the same thing. This is an external constraint, rather than one based
on the intrinsic features of electrons. But others might worry that unless
we have some previous information about what electrons are – that is,
which features are necessarily included in the core theory – then we have
no means to determine whether the disputants are indeed talking about
the same thing.
In what remains of this section, we will discuss certain examples from
physics in order to put Sider’s theory in action. Readers not comfortable
with this level of scientific detail may safely move on to the next section,
although the essential content should be understandable without any
background in physics.

Ibid., p. 49.
The Pauli Exclusion Principle states that no two fermions can occupy the same quantum
state (and so have the same four quantum numbers) at the same time. Fermi–Dirac
statistics apply to (systems of) particles with half integer spin, i.e., fermions.
88 Identifying the alternatives: ontological realism, deflationism

The purpose of bringing in some actual science is to demonstrate how

difficult it may be, at least sometimes, to determine whether the dispu-
tants in a scientific debate are really talking about the same thing. Recall
that Sider’s example of a substantive question is whether there are any
electrons located in a certain space-time region, R. But suppose R consists
of a so-called ‘Bose–Einstein condensate’ (BEC).35 Strangeness ensues: since
all the participating particles are in the same quantum state, they cannot
be described by Fermi–Dirac statistics (we must use Bose–Einstein statistics
instead) and (rather strikingly) they are not subject to the Pauli Exclusion
Principle, which we included in the charitable interpretation of the ‘core
theory’ of electrons. So we are dealing with particles that have integer
instead of half-integer spin; that is, bosons as opposed to fermions. One of
the many peculiar aspects of these BECs is that if electrons, which individu-
ally are fermions, participate in them, they must come in tightly bound
electron pairs, thus producing the required integer spin and behaving like
bosons instead of fermions. Interestingly, physicists themselves are debat-
ing about whether such tightly bound electron pairs – so-called Cooper
pairs36 – should be regarded as bosons or not:

[W]hether or not the tightly bound electron pairs can really undergo
the BEC directly, is an old but fundamental (and nontrivial) problem,
particularly as we know that the bound electron pairs are not exactly

We do not have to concern ourselves with all the relevant scientific

details, but what we can already take from this is that the question about

35 A Bose–Einstein condensate is a phase of matter which emerges at extremely low

temperatures. In a Bose–Einstein condensate, an unlimited number of bosons can be
grouped together in a single (lowest) quantum state. The theoretical possibility of such
a condensate was predicted by Bose and Einstein in the 1920s; the first condensate
was produced in 1995. This can be achieved for instance by cooling down rubidium
atoms. For more details, see the popular article by E. A. Cornell and C. E. Wieman, ‘The
Bose-Einstein Condensate,’ Scientific American 278.3 (1998), pp. 40–45.
Named after the Bardeen–Cooper–Schrieffer (BCS) theory of superconductivity. In fact,
what we are talking about here are strictly speaking fermionic condensates , in which a
pair of fermions are bound together, as in a Cooper pair, and then, acting like a boson,
can form a Bose–Einstein condensate.
Gang Su and Masuo Suzuki, ‘Towards Bose–Einstein Condensation of Electron
Pairs: Role of Schwinger Bosons,’International Journal of Modern Physics B 13.8 (1999),
p. 926.
4.4 A case study: Sider’s ontological realism 89

whether there are any electrons located in R is getting rather compli-

cated! But is the question substantial in Sider’s sense? Even this is not
clear: whether we should consider a tightly bound electron pair to be a
boson might very well be a conventional matter, perhaps based on a merely
semantic decision by physicists. Or it might not. The question is whether
we are dealing with genuine, joint-carving natural kinds to begin with –
such as bosons and fermions – and what constitutes a member of these
kinds. If ‘boson’ and ‘fermion’ carve at the joints, then this question would
seem to be substantial. However, if the distinction between bosons and
fermions is itself conventional – if it doesn’t pick out any genuine joints in
reality – then the question about whether Cooper pairs are bosons is simi-
larly a matter of convention.
Physicists are unlikely to be very alarmed about any this, since there
clearly is determinate fermion-like behaviour and boson-like behaviour,
which we can describe statistically in accordance with the above-mentioned
Fermi–Dirac and Bose–Einstein statistics. So fermions and bosons can gen-
erally be distinguished in scientific contexts. But cases like the Cooper
pair instigate some doubts regarding this distinction. Physicists are happy
to take advantage of the peculiar behaviour of Cooper pairs (in super-
conductors, for instance), but it’s at best a curiosity to them whether the
fermion–boson distinction carves at the joints in Sider’s sense. At any rate,
there may be no clear core theory or common ground that would settle the
matter. One interpretation of this situation is that according to Siderian
ontological realism (S-OR), we are dealing with a non-substantive question.
Let’s put the physics aside for now and return to the philosophical
analysis. The case of BECs does not show that the more general question,
‘Are there electrons?’ is non-substantive in Sider’s terms – that would be
a worrying result. But given the lack of a determinate core theory about
electrons, it does become more difficult to test Sider’s theory. The rele-
vant test is supposed to be that every candidate meaning for ‘electron’ that
approximately fits the actual usage of the term, mutatis mutandis, provides
the same answer to the question ‘Are there electrons?’ But whether the
previous scientific example passes this test is unclear. We might compare
this with another popular example that Sider discusses, srcinally from
Bennett.38 We may ask: ‘Is that a martini?’ of a particular appletini. This

Bennett, ‘Composition, Colocation and Metaontology.’
90 Identifying the alternatives: ontological realism, deflationism

question, as Sider as well would admit, is non-substantive. But it could be

maintained (as Sider does) that this question is not about the concept of a
martini. Presumably, the question ‘Are there martinis?’ is nevertheless sub-
stantive, because every candidate meaning for ‘martini’ that approximately
fits the actual usage of the term, mutatis mutandis, provides an answer in
the affirmative – appletinis notwithstanding. Accordingly, one might think
that Bose–Einstein condensates are like appletinis, a non-standard case
that physicists may argue about at bars, but not a substantive ontological
While this analogy is initially appealing, it has a clear problem: ‘martini’
is surely not a candidate for a perfectly joint-carving term in the first place!
In the case of martinis, it is of little consequence whether all the disputants
can be regarded as talking about the same thing – they may have very dif-
ferent conceptions about what constitutes a martini, even to the extent
that they cannot be regarded as talking about the same thing. But this is
hardly a concern for S-OR. The situation is different with ‘electron’, which
is supposedly a good candidate for a perfectly joint-carving term. The point
is that the debate about BECs and Cooper pairs presumably should count as
a substantive question, whereas the debate about appletinis should not.
If ‘electron’ really is a perfectly joint-carving term, then we should strive
to establish a determinate core theory about electrons, part of which will
involve their behaviour in any number of odd cases, such as BECs. This is
no mean task.

4.5 Taking stock

What would a ‘serious’ ontological realist make of all the above? It is

unlikely that she would be too discouraged, as ontology was never sup-
posed to be easy. However, the ontological realist (in the broad sense) is
now in a rather peculiar position: on the one hand common sense is on the
realist’s side, but on the other the burden of proof may nevertheless be on
the realist. Why is that? Because despite our best efforts, we do not seem to
have much in terms of secure or established knowledge of matters meta-
physical. In other words, it is relatively easy for various deflationists, con-
ventionalists, and even anti-realists to point to ontological debates where it
is not at all clear that anything of substance is at issue. Moreover, in some
cases where it might appear to be possible to present easy arguments that
4.5 Taking stock 91

might settle ontological debates, there is also the threat that it is too easy
(cf. Thomasson’s ‘easy’ approach) to do so – too easy for the serious onto-
logical realist at any rate.
Even if it isn’t reasonable to conclude that the burden of proof is now
on the ontological realist, it does seem clear that the realist owes us a more
fine-grained story about how, exactly, we come to know metaphysical
truths. For the most part, metaphysicians are quite happy to pursue the
questions they are interested in without an in-depth analysis of the tools
they use to do so, but this arguably leaves them too vulnerable to the defla-
tionist’s critique. Indeed, since the rise of metametaphysics, metaphysi-
cians have been much more aware about the pressing need to thoroughly
survey the epistemological foundations of the discipline, including its
‘toolbox’. At least partly because of this, certain new areas of research have
in recent years gained a substantial boost. Among them are the topics of the
next two chapters: grounding, ontological dependence, and fundamental-
ity. As we’ll see, the metaphysician’s toolbox has been supplemented with
some very promising tools due to these areas of research. One central issue
has been to shift our interest towards what is fundamental rather than
what is derivative: if the status of composite objects, for instance, is sub-
ject to such an intense debate without there being much hope of settling
the Special Composition Question (see Chapter 2), then perhaps we should
focus on what it is that grounds the existence of composite objects – what
they depend on for their existence. The next two chapters will address such
Besides driving efforts to expand the metaphysician’s toolbox, prob-
lems emerging from metametaphysics have, to a certain extent, shifted
attention towards purely epistemic issues. At least, this would seem to
be the case once we have made sure that a given debate is not merely lin-
guistic. In other words, we also need conceptual clarification to ensure
that the language of metaphysics is accurate. Assuming that we can do
this, research concerning epistemic issues in metaphysics becomes cen-
tral. Of course, epistemology lives a life of its own, but there has recently
been a growing interest in such research – also by metaphysicians them-
selves and not just epistemologists. It may not be appropriate to speak
of ‘the epistemology of metaphysics’ as a subdiscipline of its own quite
yet, but the various topics to be discussed later in this book – modal
epistemology, the role of intuitions in philosophy, and the relationship
92 Identifying the alternatives: ontological realism, deflationism

between science and metaphysics – are linked precisely by the press-

ing need to establish a more rigorous epistemology of metaphysics.
Naturally, these efforts are of interest especially to the ontological real-
ist, for many deflationists and anti-realists would consider the project of
the epistemology of metaphysics to be hopeless, unnecessary, or simply
impossible. But since this is a project that is now receiving ample atten-
tion, it appears as if the ontological realists have taken the challenge
seriously and are indeed focusing on the (epistemic) foundations of their
discipline with an unforeseen enthusiasm. The rest of this book is dedi-
cated to this very project.
5 Grounding and ontological

The notion of ‘ground’ stormed into contemporary analytic metaphysics

at the beginning of the twenty-first century, 1 but the roots of the notion
go all the way back to Aristotle. At its simplest, grounding may be under-
stood as ‘metaphysical explanation’. To be more precise, when somex is
grounded in some y, it is usually thought that y explains x. Moreover, the
status of y is generally thought to be somehow prior to that of x – grounding
is typically understood to express priority between things. For instance, we
might say that the members of a set are prior to the set itself; the existence
of the set is grounded in its members. Or to take a more concrete example,
the existence of any given composite object is grounded in the existence
of its parts. For instance, we might suggest that the existence of any given
water molecule is grounded in the existence of hydrogen and oxygen atoms.
Somewhat more controversially, we might also say that mental states are
grounded in physical states. One thing that each of these explanations has
in common is that they arenon-causal. The existence of hydrogen and oxy-
gen atoms, for instance, does not cause the existence of water molecules.
Accordingly, grounding is often called ‘metaphysical explanation’ exactly
to distinguish it from ‘causal explanation’.
In each of the mentioned examples there appears to be some sort of
dependence relation between the grounded thing and the grounding thing
or things. Quite generally, this would seem tobe ontological dependence, which
is the other primary topic of this chapter. In fact, one question that we will

The definitive work is Kit Fine, ‘The Question of Realism,’ Philosophers Imprint 1
(2001), pp. 1–30, but for more recent discussion, see especially F. Correia and
B. Schnieder (eds.), Metaphysical Grounding: Understanding the Structure of Reality (Cambridge
University Press, 2012); see also R. L. Bliss and K. Trogdon, ‘Metaphysical Grounding,’ in
E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Winter 2014 edn); see http://plato.

94 Grounding and ontological dependence

consider is whether grounding is just (a variety of) ontological dependence.

The latter comes in a number of different forms and one goal of thischapter
is to familiarize the reader with different senses of ontological dependence.
This chapter is the most technical in the book, but formal notation will
only be used for clarification: all definitions appear first in natural language.
A basic introduction to typical formal features of grounding and ontologi-
cal dependence will be presented, but the reader is not expected to master
all of this material; the main emphasis is on gaining a basic understanding
of the notions in order to be able to apply to them to various philosophical
problems. We will start from a basic introduction to the notion of ontologi-
cal dependence and some of its primary uses; this will occupy the first two
sections. We will then move on to consider the relationship of grounding
and ontological dependence in the third section. A somewhat more tech-
nical presentation of the various formal features of ground will follow in
the fourth section, although formal notation will once again be avoided.
The last two sections of the chapter concern applications of grounding to
various other topics as well as related, more established notions, namely
causation, reduction, modality, and truthmaking.

5.1 Ontological dependence: a fine-grained notion

It is not uncommon to see the notion of ontological dependence used in a

rather coarse-grained manner, given that it encompasses a family of rela-
tions. For instance, we often see claims such as:

(1) ‘Sets ontologically depend on their members’.

(2) ‘Electricity ontologically depends on electrons’.
(3) ‘God doesn’t ontologically depend on anything’.
While all of the above no doubt express some important type of depend-
ence relationship, they are also clearly quite different from each other. In (1),

Additional recommended resources on ontological dependence include Fabrice Correia,
‘Ontological Dependence,’ Philosophy Compass 3 (2008), pp. 1013–1032, Tuomas E. Tahko
and E. J. Lowe, ‘Ontological Dependence,’ in E. N. Zalta (ed.), The Stanford Encyclopedia
of Philosophy (Spring 2015 edn); see
dependence-ontological/; and Kathrin Koslicki, ‘Varieties of Ontological Dependence,’
in Correia and Schnieder (eds.), Metaphysical grounding, pp. 186–213.
5.1 Ontological dependence: a fine-grained notion 95

we mean that a set {x, y, z} could not exist if its members, namely
x, y, z, did
not exist. The type of dependence in question rigid
is existential dependence,
to be clarified in a moment. (In fact, there is another sense of dependence
at work in (1)as well, namely identity-dependence, but we will return to this
example later on, in the next section.) In (2), we seem to have in mind a more
general kind of dependence:there could not be electricity if there were no
electrons. So existence of electricity depends on the existence of particles of
a specific kind, electrons. This second typeof dependence is also existential,
but to separate it from the rigid dependence in (1), we may callgeneric
it exis-
tential dependence. In (3)we are instead referring to the ontological independence
of God. Presumably, God does not depend for her existence on anything, by
her very nature. In other words, it is part of the essence of God that she is
ontologically self-sufficient. We might call this
essential independence, in con-
trast to essential dependence.
A family of notions is beginning to emerge. However, we should formu-
late each notion somewhat more precisely. The first thing to note in defin-
ing ontological dependence is the modal-existential element in dependence
claims. For instance, we’ve said that a set cannot exist unless its members
do. So there is a sense in which the existence of a set necessitates the exist-
ence of its members. Indeed, it is common to talk, for example, about rigid
existential necessitation as synonymous with rigid existential dependence
and generic necessitation as synonymous with generic existential depend-
ence. Typically, statements of ontological dependence are thought to refer
to metaphysical modality (rather than, say, conceptual or logical modality). 3
This is primarily because they concern matters that are broader than
just conceptual or logical; the ontological independence of God being a
case in point. Besides God, substances and perhaps fundamental particles are
often considered to be entities that do not depend for their existence upon
anything else.
We can now give an initial definition of rigid existential dependence:4

The distinction between different kinds of modality is a matter of some debate, but
metaphysical necessity is typically understood as a broader kind of necessity than con-
ceptual or logical necessity, while logical necessity is strictest. In other words, a meta-
physically necessary truth would be something that is necessary in virtue of something
other than the definitions of concepts and the laws of logic.
This and the following non-formal definitions are taken from Tahko and Lowe,
‘Ontological Dependence’.
96 Grounding and ontological dependence

Rigid existential dependence (RXD) x depends for its existence

upon y =df Necessarily, x exists only if y exists.

Why ‘rigid’? Because there is no flexibility here: the existence of a given

x requires the existence of that very y.5 It could not be something a little bit
like y, something falling roughly in the same category, for instance; it must
be y. The definiens in (RXD) is equivalent to ‘Necessarily, if x exists, then y
exists’, so that according to (RXD) the existential dependence of x upon y
amounts to the strict implication of y’s existence by x’s existence. We have
mentioned one example of rigid existential dependence, namely that sets
ontologically depend on their members (more precisely, a set depends rig-
idly on the very members it has, i.e., any change in a set’s members will
change the set itself). Another, although more controversial, example is
a particular person depending for his existence on his parents, or, more
precisely, on the particular sperm and egg that he srcinates from. This
example is of course related to the essentiality of srcin .6

However, (RXD) clearly fails to capture the intuitive sense of depend-

ence in some cases. Consider a living organism. A living organism would
appear to depend for its existence upon its parts, such as cells. But we also
know that a living organism may survive a change of any of its cells, pro-
vided that the change is not disruptive. Such an organism must certainly
have parts such as cells if it is to exist, but which objects those parts are is
inessential – and consequently it is not the case that it depends for its exist-
ence, in the sense defined by (RXD), upon any one of those parts. But it is
possible to define another sense of existential dependence in which it is true
to say that a composite object depends for its existence upon its proper
parts; a generic notion of existential dependence, defined as follows:7

Generic existential dependence (GXD) x depends for its existence upon

Fs =df Necessarily, x exists only if some Fs exist.

Composite objects are existentially dependent objects in the sense of

(GXD), since they require the existence of proper parts (set F as ‘proper

The srcin of the term ‘rigid’ is Saul Kripke, Naming and Necessity (Harvard University
Press, 1980).
As discussed in Kripke, Naming and Necessity.
The labels (RXD) and (GXD) are to separate eXistential dependence from eSsential
dependence – see below.
5.1 Ontological dependence: a fine-grained notion 97

part of x’ in (GXD)). The important difference between the rigid and the
generic cases is that (RXD) refers to a specific object whereas (GXD) only
requires that at least some Fs exist. Another example, mentioned earlier,
where (GXD) would seem to capture the correct sense of dependence is (2),
‘Electricity ontologically depends on electrons’.
Let us now turn to a formal presentation of these notions. We use
the sentential operator ‘ ☐’ for metaphysical necessity, the one-place
predicate ‘ E’ for existence, and the two-place sentential operator ‘ →’ for
material implication. Following this notation, we can formalize rigid
existential dependence (RXD) and generic existential dependence (GXD)
as follows:
(RXD) ☐ (Ex → Ey)

(GXD) ☐ (Ex → ∃yFy)

(RXD) can be read as ‘ x rigidly depends for its existence on y’, or alter-
natively ‘x rigidly necessitates y’. In (GXD), we have added the existential
quantifier ‘∃’ as well as the general term F to express the thought that ‘x
generically depends for its existence on something being an F’, or alterna-
tively ‘x generically necessitates F’. The important difference between the
rigid and the generic cases is that (RXD) refers to a specific object whereas
(GXD) only requires that at least some Fs exist. We would now have the
tools to formalize (1) and (2), but note that there are cases where further
tools are required. Consider:

(4) ‘Children ontologically depend on their parents’.

On the face of it, what we mean in (4) is that if parents x and y had not
existed, then their child z could not have come into existence. This looks
like a case of rigid existential dependence, but it is clear that once z has
been conceived/born, her father/mother can go out of existence without
any effect on her own existence. At that point, there is only past rigid exis-
tential dependence. For cases such as this, we would require temporally
relativized versions of (RXD) and (GXD), but we will omit these compli-
cations here.8 In any case, we now have an initial understanding of the
modal-existential analysis of ontological dependence.

The specific formalization is a voluntary exercise. See Correia, ‘Ontological Dependence,’
p. 1016 for further details.
98 Grounding and ontological dependence

But what about (3)? It does not seem to fit this analysis neatly. Moreover,
there are other famous examples where the modal-existential analysis
would seem to break down, or not to be sufficiently fine-grained. Consider
the relationship between Socrates and the temporally extended event
or process that was his life.9 If we attempt to formulate this relationship
in terms of (RXD), it will turn out that Socrates’s life depends rigidly for
its existence upon Socrates. But surely Socrates’s existence also depends
upon his life! At the same time, we may state numerous things that are evi-
dently true of the life of Socrates, yet are not true of him, and vice versa (for
example, his life was long, but came to an abrupt ending, while Socrates
himself was snub-nosed). The upshot is that Socrates and his life surely
cannot be identical. It seems thus that there are certain difficult questions
which the modal-existential analysis may not fully address, at least not in
the simple form that we have presented it. This warrants us to consider
more fine-grained notions of dependence, such as identity-dependence and
essential dependence.

5.2 Identity-dependence and essential dependence

Until quite recently, it was common to think that ontological dependence

can be fully characterized in modal-existential terms, as we have done
above. One obvious reason for this is that if one adopts the usual ‘modal-
ist’ analysis of essence, essential dependence will collapse into a form of
modal-existential dependence.10 But there is an alternative way to formu-
late (some varieties of) ontological dependence, which is needed if essence
is not analysed in modal terms. 11 One motivation for developing a non-modal
conception of ontological dependence is that the modal-existential analysis
appears to be too coarse-grained for some cases. We have already men-
tioned one example, from E. J. Lowe, but the most well-known examples
have been made famous by Kit Fine.12 Consider, for instance, what the
modal-existential account entails in the case of necessary existents. Take

The example srcinates from Tahko and Lowe, ‘Ontological Dependence.’
For a classic account, see Ruth Barcan Marcus, ‘Essentialism in Modal Logic,’ Noûs 1.1
(1967), pp. 91–96.
See Kit Fine, ‘Essence and Modality,’ Philosophical Perspectives 8 (1994), pp. 1–16.
Kit Fine, ‘Ontological Dependence,’ Proceedings of the Aristotelian Society 95 (1994), pp.
5.2 Identity-dependence and essential dependence 99

Socrates and the number 2, for example. Given that numbers necessarily
exist, it is necessarily the case that the number 2 exists if Socrates does. But
presumably we do not want to say that Socrates depends upon the number
2, or indeed on most necessary existents that you might put in the place of
the number 2. Yet, the modal-existential account makes everything depend
upon every necessary existent, which seems like the wrong result.
Admittedly, those who defend a modal-existential analysis of onto-
logical dependence could insist that it applies only to contingent objects.13
Simons makes this type of qualification by focusing on concrete enti-
ties, hence excluding necessary existents by definition; he also excludes
self-dependence. Simons calls the resulting notion of dependence weak rigid
dependence, but a stronger notion,strong rigid dependence, is also defined – the
latter is a special case of the former. 14 One example of weak rigid depend-
ence as defined by Simons would be a particular water molecule depending
for its existence on a particular oxygen atom. In the case of strong rigid
dependence, the dependent object cannot be a proper part of the object
it depends upon. So object x is strongly rigidly dependent on object y if x
depends for its existence on y and y is not a proper part of x. One example
of strong rigid dependence defined thus would be a particularized property
(these are often known as ‘tropes’ or ‘modes’) depending for its existence
on a substance, e.g., the particular redness of an apple depending for its
existence on the apple. In addition to these rigid notions, Simons defines
corresponding notions of (weak and strong) generic dependence. But while it
is possible to avoid some of the challenges raised for the modal-existential
account with these qualifications, they do nevertheless leave room for an
alternative account of ontological dependence that could also be applied
to necessary existents. Alternatively, a proponent of the modal-existential
account could simply bite the bullet and insist that every contingent entity
does rigidly depend for its existence on necessary existents. One reason to
do so would be the ability to get by with a sparser battery of ontological
tools – a consideration motivated by parsimony.
The modal-existential analysis of ontological dependence can thus be
developed further and it can perhaps overcome some of the problems
that are typically associated with it. But there are areas where a more

See Peter Simons, Parts: A Study in Ontology (Oxford: Clarendon Press, 1987), p. 295.
Ibid., p. 303.
100 Grounding and ontological dependence

fine-grained notion would seem to be of use. It should be noted though

that much of the contemporary literature in defence of a non-modal,
fine-grained analysis operates in a ‘neo-Aristotelian’ framework which typ-
ically assumes some ‘non-modalist’ version of essentialism. 15 Accordingly,
there is an obvious rift between the modal-existential analysis and the
essentialist analysis – one that we cannot fully bridge here. In any case, it
is good to keep in mind that the notion of ontological dependence itself
does not immediately force one to make a commitment in this regard,
even though some of its applications may entail such a commitment.
As a point of entry to the idea that there could be notions of depend-
ence not easily analysable in terms of the modal-existential account,
consider the fact that not all forms of dependence seem to involve a
requirement for existence at all. Indeed, as we already saw in the case of
(3), ‘God doesn’t ontologically depend on anything’, it seems that some-
thing beyond mere existential independence is being exp ressed. Instead,
one might say that God would not be the being that she is if she were
not ontologically independent by her very nature. The notion of essen-
tial dependence , which involves requirements for identity or essence , may
better express what God’s supposed ontological independence is about.
In other words, it is an essential property of God that she is ontologically
independent. It is not quite straightforward to define essential depend-
ence, although a formal definition will be given below. But before that,
we ought to get a better picture of what it means to say that an object
depends upon something for its identity ; that is, we should clarify the
relation of identity-dependence .
Note that the notion of ‘identity’ at play here is not the one symbol-
ized with the ‘equals’ sign, namely ‘=’. Rather, we mean ‘identity’ in
the sense of what a thing is, or which thing of a certain kind a thing is.
To say that the identity of x depends on the identity of y is to say that
which thing of its kind y is metaphysically determines which thing of its
kind x is. For instance, then, the identity of a set is metaphysically deter-
mined by the identities of its members. Identity-dependence expresses
the determination of the individuality of objects in terms of the indi-
viduality of other objects. Thus, the identity-dependence of a set upon its
members is a consequence of the fact that the Axiom of Extensionality

E.g., Koslicki, ‘Varieties of Ontological Dependence.’
5.2 Identity-dependence and essential dependence 101

functions as the criterion of identity for sets. We are now in a position
to give a definition of identity-dependence:

Identity dependence (ID) x depends for its identity upon y =df There is a
two-place predicate ‘F’ such that it is part of the essence of x that x is related
by F to y.

We can exemplify (ID) by letting x be {z} and y be z, in which case we

have that {z} depends for its identity upon z, because there is a two-place
predicate – namely ‘being a member of the singleton set’ (also known as
the unit set function) – such that it is part of the essence of { z} that it is the
singleton set of z. But, of course, for (ID) to express a usable notion, a full
account of the notion of ‘essence’ would be required – and this is some-
thing that we will not attempt to provide here. However, various attempts
to construct such an account have been made, for instance by Fine and
Lowe.17 A central idea behind both their accounts is the Aristotelian notion
of real definition. The real definition of an entity in this context could be
interpreted simply as a proposition which states the essence of the entity.
In this sense, a thing’s essence may be said to constitute its identity, when
one uses the word ‘identity’ in this distinctive manner to speak of a thing’s
identity, rather than using it to speak of the identity relation. Seen in this
light, identity-dependence as defined by (ID) is simply a species of essential
dependence, that is, a way in which the essence of a certain thing is deter-
mined by a relation in which it stands to another thing. So let us now move
to essential dependence. As with existential dependence, we can define a
rigid and a generic notion of essential dependence (sometimes called essen-
tial existential dependence), as follows. First, rigid essential dependence:18

Rigid essential dependence (RSD) x depends for its existence upon y =df
It is part of the essence of x that x exists only if y exists.

Note that, whereas two distinct entities plausible cannot be identity-

dependent upon each other, it plausibly is possible for each of two distinct

The Axiom of Extensonality is part of the standard Zermelo–Frankel formulation of set
theory; it states that any two sets with exactly the same members are equal.
Kit Fine, ‘Logic of Essence,’ Journal of Philosophical Logic 24 (1995), pp. 241–273;
E. J. Lowe, ‘Two Notions of Being: Entity and Essence,’ Royal Institute of Philosophy
Supplement 83.62 (2008), pp. 23–48.
These definitions follow Tahko and Lowe, ‘Ontological Dependence.’
102 Grounding and ontological dependence

entities to depend essentially for its existence on the other. Moreover, if x

depends essentially for its existence upon y, then x also rigidly depends for
its existence upon y; that is, essential dependence entails rigid existential
dependence ((RSD) → (RXD)) – but not, of course, vice versa. Similarly, we
can define generic essential dependence, as follows:

Generic essential dependence (GSD) x depends for its existence upon

y =df It is part of the essence of x that x exists only if some F exists.

(GSD), unsurprisingly, operates in a similar fashion with regard to (RSD)

as (GXD) does to (RXD). We can also express both of these notions formally,
using Fine’s notation. First, we need to introduce a new operator ‘☐x’, which
behaves like the more familiar ‘☐’, but should be read as ‘it is part of the
essence/nature of x that …’.19 With the help of this new operator, we can
introduce formal definitions of rigid essential dependence and generic
essential dependence, which correspond to (RXD) and (GXD):
(RSD) ☐x(Ex → Ey)

(GSD) ☐x(Ex →∃yFy)

(RSD) can be read as ‘it is part of the essence of x that it exists only if y
does’, whereas the natural reading of (GSD) is ‘it is part of the essence of x
that it exists only if something is an F’. It is important to note that while
it is uncontroversial that all essential truths about x are necessary truths
about x, the converse – the claim that all necessary truths about x are essen-
tial truths about x – has been challenged by Fine and Lowe.20 The challenge
is important in this connection because if all necessary truths about x were
also essential truths about x, then (RSD) and (GSD) would simply reduce to
(RXD) and (GXD) (recall that they can be understood in terms of metaphysi-
cal necessitation). In other words, essential dependence is of genuine interest
only to those who are non-reductionist about essence, that is to those who
adopt a non-modal conception of essence as defended by Fine and Lowe.
We can find slightly different formulations of essential dependence in
the literature. For instance, on Kathrin Koslicki’s re-construal of Fine’s
essentialist account, we get: 21

Fine, ‘Logic of Essence.’
See Fine, ‘Essence and Modality’; E. J. Lowe, ‘What is the Source of our Knowledge of
Modal Truths?’,Mind 121 (2012), pp. 919–950.
Koslicki, ‘Varieties of Ontological Dependence,’ p. 190.
5.2 Identity-dependence and essential dependence 103

Constitutive essential dependence (EDC) An entity x ontologically

depends on an entity (or entities), y, only in the case that y is a constituent
(or are constituents) in x’s essence.

This account relies on the notion of constitutive essence, which is developed

in Fine’s work; as Fine puts it: ‘we may take x to depend upon y if y is a
constituent of a proposition that is true in virtue of the identity of x or,
alternatively, if y is a constituent of an essential property of x’.22
It should be noted that even though the notions of essential dependence
defined above are no doubt more fine-grained than the modal-existential
notions, there may be reasons to think that even they are not sufficiently
fine-grained for all purposes (as argued by Koslicki). To illustrate, con-
sider Fine’s well-known discussion of Socrates and the singleton set that
has Socrates as its sole member. To use Fine’s terminology, we could say
that it is part of the constitutive essence of Socrates’s singleton set that it
has Socrates as its sole member, whereas it is not part of the constitutive

essence of Socrates to be the sole member of Socrates’s singleton set. But

as Koslicki points out, this is really just to say that Socrates’s singleton set
ontologically depends on Socrates whereas Socrates does not ontologically
depend on Socrates’s singleton set.23 The relevant notion of dependence
appears to be built into Fine’s notion of essence and, similarly, the notion
of essence assumed by Fine (and many other ‘neo-Aristotelians’) is already
built into (EDC).
Before we conclude our discussion of essential dependence, a more gen-
eral issue regarding the essentialist vs. modal-existential account should
be mentioned. It appears that those who are willing to buy into a suitably
fine-grained notion of essence will find the modal-existential account of

ontological dependence far too coarse-grained. Yet, those who are not sym-
pathetic to the relevant primitive notion of essence and would rather ana-
lyse essence in terms of modality would insist that the modal-existential
analysis is quite sufficient, and indeed that essential dependence collapses
into modal-existential dependence. A further issue is that there is some dis-
agreement amongst those who think that a more fine-grained analysis than
the modal-existential account is needed about how the relevant notion of

Fine, ‘Ontological Dependence,’ p. 275.
Koslicki, ‘Varieties of Ontological Dependence,’ p. 195.
104 Grounding and ontological dependence

essence is to be constrained. For instance, Koslicki regards Fine’s proposi-

tional notion of essence according to which there is little or no distinction
between essence and real definition as overly restrictive. Koslicki identifies
the source of this restrictive conception of essence, which is also present
in Lowe’s work, to be the focus on essences as individuating – this is the
type of view regarding essential dependence, namely identity-dependence,
which we have been focusing on here. In Koslicki’s alternative picture,
essences ‘must do more than individuate the entities whose essences
they are; and real definitions must do more than state conditions which
uniquely identify and delineate the entities under consideration at every
time and in every world in which they exist’.24 However, this is not the place
to settle the controversy regarding the essentialist vs. the modal-existential
account; we will return to the relationship between essence and modality
in Chapter 7.

5.3 Is grounding ontological dependence?

We have already made use of the notion of ‘ground’ and it seems clear that
it has something to do with ontological dependence. But is grounding just
a variety of ontological dependence? At the outset, we can assume that if
grounding were to be understood as a type of ontological dependence, it
would be some sort of explanatory dependence. The idea that whatever does the
grounding also somehowexplains what is being grounded is a crucial part of
the notion’s appeal. Relations of ontological dependence often seem to have
a similar type of explanatory role, but the link toexplanation is weaker:even
though the existence of water depends on the existence of hydrogen and
oxygen, it does not seem to be the case that the existence of hydrogen and
oxygen explain the existence of water. Rather, what explains the existence of
water is the ability of hydrogen and oxygen atoms to form molecules (even
though this is rather simplified). So itseems that not all relations of ontologi-
cal dependence can be grounding relations in the usual
We need something stricter than just ‘an explanatory role’ to identify
grounding – otherwise we would end up with a much too liberal notion,
for we may regard a number of loosely connected things explanatory in
some very loose sense. For instance, we might say that the fact that Smith

Ibid., p. 200, fn 13.
5.3 Is grounding ontological dependence? 105

murdered Jones is explained by certain events in Smith’s childhood, but

a more direct explanation might be Smith’s desire to rob Jones. Certain
events in Smith’s childhood may help us understand why Smith has murder-
ous desires, but it’s not clear that they serve to ground the fact that Smith
murdered Jones.
One suggestion that may help to make grounding more precise would
be to focus on priority. At the beginning of this chapter, the example was
given that if the members of a set are prior to the set itself, then the exist-
ence of the set is grounded in its members. So the grounding entities are
prior to – or more fundamental than – the grounded entities. This is another
sense in which grounding would seem to detach from certain types of
ontological dependence, as a purely modal understanding of dependence
(without any claim to priority) is also possible. This is of course the sense
which we defined above in terms of necessitation: even if x rigidly neces-
sitates y, this does not entail that y must be ontologically prior to x. At
any rate, it would be odd to say that parents are ontologically prior to
their children! An essentialist notion of dependence, as familiar from Fine
and Lowe, seems a more promising tool if we hope to express the rele-
vant sense of priority. 25 But we do not need to decide the matter here, as
pluralism is a perfectly viable option as well: there is certainly a use for
all of these different notions of dependence. Nevertheless, there may be
an independent motivation for finding a systematic link between onto-
logical dependence and grounding, as defining one in terms of the other
would be more parsimonious than having two primitive notions.26 In any
case, even if we could define one notion in terms of the other, it seems
that there are aspects of ontological dependence that are not captured
by all accounts of grounding as well as aspects of grounding that are not
captured by all accounts of ontological dependence. Accordingly, for the
time being it is advisable to keep the notions apart, especially since there
are some further, formal differences that need to be taken into account.
To these we now turn.

See Koslicki, ‘Varieties of Ontological Dependence,’ for a thorough discussion of the
essentialist notion of dependence.
26See Fabrice
Correia and Benjamin Schnieder, ‘Grounding: An Opinionated
Introduction,’ in Correia and Schnieder (eds.), Metaphysical Grounding, pp. 1–36.
106 Grounding and ontological dependence

5.4 Formal features of ground

A more precise treatment of ground requires taking into account a num-

ber of general specifications as well as certain formal features. The gen-
eral specifications fall into four groups: the first concerns the distinction

between full and partial grounds, the second concerns the distinction
between mediate and immediate grounds, the third concerns the dis-
tinction between weak and strict grounds, and the fourth concerns the
so-called ‘operational’ vs. ‘relational’ approaches to grounding.27
The full/partial ground distinction is fairly straightforward: if P and Q
together fully ground the conjunction P & Q, then P and Q are each a partial
ground for the conjunction P & Q. One reason to adopt this distinction is
that it is often difficult to determine what the full ground of something
is. For instance, we might say (to simplify) that the phenomenon of turbu-
lence is partially grounded in the rapid variation of pressure and velocity
in a region, but the complexity of the phenomenon and our limited under-

standing of it make it difficult to determine what would constitute the

full ground of turbulence. In fact, partly for the same reasons, turbulence
is typically treated statistically. There is nothing particularly controversial
about the full/partial ground distinction, except perhaps the question of
whether one of them is prior to the other. Fine is of the opinion that full
ground is certainly prior, and partial ground should be defined in terms
of full ground.28 The simple reason that Fine states for this preference is
that the conjunction and disjunction of P and Q both have the same partial
grounds, namely P and Q, but in the case of the disjunction, P ∨ Q, both P
and Q are also full grounds. To appreciate this difference, we must make
reference to full grounds. In general, when we talk about grounding, we
mean full grounding, but as we will see below, sometimes it is important
to specify that we mean only partial grounding (and not only for epistemic
reasons, as in the case of turbulence).

For discussion of the first three distinctions as well as a number of other distinc-
tions, see Kit Fine, ‘Guide to Ground,’ in Correia and Schnieder (eds.), Metaphysical
Grounding, pp. 37–80. We will not cover all of the relevant distinctions here. For dis-
cussion of the operational/relational approaches (the latter is sometimes also called
‘predicational’), see Kelly Trogon, ‘An Introduction to Grounding,’ in M. Hoeltje,
B. Schnieder, and A. Steinberg (eds.),Varieties of Dependence (Munich: Philosophia Verlag,
2013), pp. 97–122.
Fine, ‘Guide to Ground,’ p. 50.
5.4 Formal features of ground 107

The distinction between mediate and immediate grounds is also

straightforward. Immediate grounds are such that they do not need to be
‘mediated’ by any other grounding relations. For instance, the immediate
grounds of the conjunction P & Q are simply P and Q. But if we were to add
another conjunct, such as in (P & Q) & R, we would say that the entire con-
junction is only mediately grounded in P, Q, and R, since we first have to
consider the immediate ground of (P & Q) and the immediate ground of R
before moving on to the ground of (P & Q) & R. In this way, the full ground
of the more complex conjunction is mediated through other grounding
relations. Mediate ground can be defined in terms of immediate ground, as
all the mediate grounds can be arrived at by analysing immediate grounds,
like we just did in the case of (P & Q) & R. Fine finds this distinction impor-
tant, because it gives us a natural way to obtain a ground-theoretic hierar-
chy. The immediate ground of a given truth always follows immediately
at the next (lower) level of the hierarchy, and via the chaining of mediate
grounds, we can arrive at all the partial grounds for that truth at lower
levels of the hierarchy. Considerations such as this will prove to be crucial
for certain interesting applications of grounding.
The distinction between weak and strict ground is somewhat more com-
plicated – and there is some disagreement about it – but the idea can be
summarized as follows: a strict ground for a given fact always occurs at
a lower level in the explanatory hierarchy than the fact itself, whereas a
weak ground occurs at the same level.29 This means that some facts could
even be weak grounds for themselves! However, it is somewhat more dif-
ficult to find illustrative examples of weak grounding. One potential exam-
ple of weak grounding might be Jack’s being Jill’s sibling, which explains
Jill’s being Jack’s sibling, and also the fact that Jack and Jill are a pair of
siblings. Here Jack’s being Jill’s sibling and Jill’s being Jack’s sibling occur
at the same level of explanatory hierarchy. We do not need to dwell on this
case, though, as it turns out that the notion of strict ground is generally
of much more interest to us. In fact, given all these distinctions, we may
identify the focus of much of the grounding literature: typically when we
talk about grounding we have in mind a strict partial ordering.30 This bit of

For an analysis of weak ground, see Louis deRosset, ‘What is Weak Ground?’, Essays in
Philosophy 14.1 (2013), Article 2.
This has been labelled the ‘orthodox’ approach by Michael Raven, ‘Is Ground a Strict
Partial Order?’, American Philosophical Quarterly 50.2 (2013), pp. 191–199.
108 Grounding and ontological dependence

set-theoretic jargon captures three important formal features of ground –

irreflexivity, transitivity, and asymmetry– which we will discuss in more
detail below. Grounding understood as a strict partial order produces the
type of hierarchical chain of explanation that we have already been refer-
ring to implicitly in this book.
Let us now move to a slightly different issue, namely the distinction
between operational and relational approaches to grounding. Some propo-
nents of grounding take it to be a relation, whereas some prefer the oper-
ational approach. The relational approach is perhaps easier to understand
as it considers grounding expressions to be simply predicates that we use to
state certain relations that obtain between entities. On this view, an import-
ant question is what kind of entities the relata are, that is, an account of
the grounded and the grounding entities is required. Candidates include,
for instance, facts and states of affairs– in this chapter we primarily refer
to facts as the relata. The operational approach, which is preferred for
instance by Fabrice Correia and Kit Fine, is ontologically neutral concern-
ing the relata.31 As Correia puts it, ‘it should be possible to make claims of
grounding and fail to believe in facts’.32 On this view, the notion of ground is
to be expressed by means of a sentential operator, ‘because’ being a natural
candidate. We often say that p‘ because q’, and the idea of the operational
approach is that grounding is expressed by the ‘because’ operator in such
cases. One motivation for this view is that the grammatical form of many
grounding claims seems to point towards an operational approach. When
I say, ‘I took an umbrella because it was raining’, we see thatp and q can
be taken as sentences, and the ‘because’ operator takes these sentences to
make a further sentence.33 There are no obvious constraints as to the onto-
logical category denoted byp and q in these sentences, which is why the
operational approach can remain neutral about the existence of facts or
other typical candidates for the relata in grounding statements.
The ontological neutrality of the operational approach may be considered
an advantage, but it also has its costs:the approach loses a natural point

For further details, see Fabrice Correia, ‘Grounding and Truth-Functions,’ Logique &
Analyse 211 (2010), pp. 251–279.
Ibid., p. 254.
The example ‘I took an umbrella because it was raining’ may look like a case of caus-
ation rather than (or in addition to) grounding. This may indeed be the case. We will
return to the relationship between grounding and causation below.
5.4 Formal features of ground 109

of connection between grounding and ontological dependence, as it does

not seem to account for the idea ofexplanatory dependence. If explanation is
thought to be a dependence relation (between facts), then grounding is a
natural choice for describing this relation. This would perhaps strengthen
the case towards the claim that grounding is a specific variety of ontological
dependence, namely explanatory dependence between facts. However, it
should be noted that the distinction between the operational and relational
approaches should be understood as concerning the mostbasic notion of
ground, for it is possible to ‘translate’ statements of ground formulated in
terms of one approach to the other, given certain assumptions.
As mentioned above, grounding is commonly considered to be con-
strained by three formal features: asymmetry, irreflexivity, and transitivity.
These features are widely but not universally agreed upon. 35 The relevant
definitions are as follows, although note that the first two define positive fea-
tures rather than the negative ones which are taken to hold for grounding.

Symmetry A relation R is symmetric if and only if: if x is related by R to y,

then y is related by R to x.

E.g., ‘being siblings’.

Reflexivity A relation R is reflexive if and only if: R is self-relating, i.e.,

everything bears R to itself.

E.g., ‘being self-identical’.

Transitivity A relation R is transitive if and only if: if x is related by R to y,

and y is related by R to z, then x is related by R to z.

E.g., ‘being taller than’.

Since grounding is understood to be a priority relation, it seems clear that

it cannot be symmetric or reflexive. Ifx were to ground y and y to ground x,
then we would clearly have a loop– a vicious one if we are hoping to explain
one thing in terms of another. Similarly, ifx were to ground itself, then it

For further details, see Correia, ‘Grounding and Truth-Functions.’
For some challenges, see C. S. Jenkins, ‘Is Metaphysical Dependence Irreflexive?’,
The Monist 94.2 (2011), pp. 267–276; Jonathan Schaffer, ‘Grounding, Transitivity, and
Contrastivity,’ in Correia and Schnieder (eds.), Metaphysical Grounding, pp. 122–38;
Tuomas E. Tahko, ‘Truth-Grounding and Transitivity,’ Thought: A Journal of Philosophy 2.4
(2013), pp. 332–340.
110 Grounding and ontological dependence

would appear to be ontologically self-sufficient. Perhaps there are such onto-

logically self-sufficient entities, but we do not typically talk about them as if
they were self-grounding, but rather as being fundamental:that is, primitive
in the sense of not needing to be grounded at all. Accordingly, grounding as
it is usually understood is asymmetric and irreflexive. But what about tran-
sitivity? To give an example in terms of grounding, consider sets once again.
We might say that the existence of a set is grounded in the existence of its
members, and if its members include other sets, its existence is grounded
in the existence of their members. In this way, a natural chaining of ground
emerges, all the way back to the fundamental, ungrounded entities. In the
next chapter we will consider the implications of such grounding chains in
more detail, but the possibility of this type of chaining, and hence transitiv-
ity, is typically considered central for grounding.
Note that these formal features apply only to a subset of the distinctions
that were introduced above. For instance, because immediate grounds are
always direct grounds and mediate grounds indirect, it follows that only
the notion of mediate ground is transitive. Also, since a given fact may
be a weak ground for itself, only the notion of strict ground is irreflexive.
Accordingly, only grounding understood as a strict partial ordering ‘auto-
matically’ satisfies transitivity, irreflexivity, and asymmetry.
These formal features are not only relevant for grounding, of course.
They also provide us with a further method for distinguishing between
different types of ontological dependence. For instance, rigid existential
dependence is reflexive and (therefore) not asymmetric. But the type of
‘non-self-sufficiency’ that we are interested in when it comes to grounded
entities arguably requires irreflexivity and asymmetry. However, we can
also define an asymmetric relation of rigid existential dependence, which
holds when x rigidly necessitates y but not vice versa. This type of one-way
rigid necessitation is certainly of some interest, but it may not express the
type of non-self-sufficiency that we have in mind in the case of grounding.
To see this, consider the following. If Socrates is a contingent existent and
the empty set is a necessary existent, then Socrates one-way rigidly neces-
sitates the empty set. But surely, the existence of Socrates is not derivative
of the existence of the empty set!36

On this, see Fine, ‘Ontological Dependence.’
5.4 Formal features of ground 111

Even though asymmetry, irreflexivity, and transitivity are widely agreed

to be among the core formal features of ground, they have not gone unchal-
lenged. Considering some of these challenges may help us to understand
grounding better, so it is worthwhile to think about them. Here we will
focus on transitivity. Consider the proposition ‘There exists a particular
bottle of beer, b’.37 We might state the following grounding claims regard-
ing this proposition:

(i) The fact that a particular bottle of beer, b, exists is partially grounded
in the fact that b has a stable macrophysical structure.
(ii) The fact that b has a stable macrophysical structure is partially
grounded in the fact that certain fundamental laws of physics hold.

Both of these grounding claims may be considered plausible. Regarding

(i), we wouldn’t think that the bottle of beer exists if it didn’t have roughly
the sort of properties – all of them macrophysical – that we expect from a
bottle of beer, such as being rigid (assuming that it is a glass bottle), and
containing a tasty, usually alcoholic and thirst-quenching beverage. All of
the macrophysical properties of b are grounded in certain microphysical
properties, the causal features of the constituent microphysical particles in
particular, which together produce a stable macrophysical structure. This
is not to say that all necessary conditions are grounds, but (i) seems as
plausible a candidate as any. As to (ii), the grounding claim could refer to a
number of physical laws or principles. We do not need to go into the details
here – the stability of macrophysical objects clearly requires microphysical
stability and (ii) may be considered to refer to whatever physical laws the
stability of matter requires.38 So, there are good reasons to think that certain
fundamental physical laws at least partially ground many claims concerning
macrophysical stability.
Now, by transitivity, weget:

(iii) The fact thata particular bottle of beer

, b, exists is partially grounded in
the fact that certain fundamental laws of physicshold.

See Tahko, ‘Truth-Grounding and Transitivity’ for further discussion of this example.
For instance, we could refer to the Pauli Exclusion Principle, which states that two fer-
mions in a closed system cannot be in the same quantum state at the same time. This
principle is sometimes said to be responsible for the space-occupying behaviour of
matter, as it ‘prevents’ atoms from collapsing. We encountered this principle in pass-
ing in Chapter 4 and will refer to it again in Chapter 6.
112 Grounding and ontological dependence

It seems that (iii) states a part of the ‘ultimate ground’ of the existence
of b. This is perfectly in accordance with the usual treatment of ground-
ing – we only need the transitivity of grounding to get this result. But at the
same time one might think that (iii) is a rather odd grounding claim. The
reason for this is that we do not immediately see how fundamental laws of
physics are relevant for the existence of a particular bottle of beer. The clash
of intuitions at hand here– between a certain type of intuitive relevance
and ontological ground– may highlight a deeper issue concerning ground.
The issue is that of epistemic as opposed to ontological relevance. The notion
of ground is generally considered to be ontological, in which case the clash
of intuitions described here should not cause much concern, at least insofar
as the opposition to claims like (iii) is based only on considerations of epi-
stemic relevance. So if one asks whether there’s any beer left and someone
answers that there is a bottle of beer in the fridge, one would not consider
it relevant that the macrophysical stability of the bottle depends, in some
sense, on certain fundamental laws of physics. Considerations such as these
may prompt us to question whether ground is truly a single ontological rela-
tion or are there instead several distinct notions of ground, perhaps some
of them ontological and some epistemic. If this were the case, then it might
seem that grounding is unable to do all the work it is often claimed to do.
We turn to an overview of some this work, but we will briefly return to the
issue concerning distinct notions of ground towards the end of the chapter.

5.5 Grounding, causation, reduction, and modality

The title of this subsection suggests that a lot of ground will be covered.
This is true, but we will not need to spend too much time on any given
issue.40 The reason is partly because we are about to move to a rather

Besides Tahko, ‘Truth-Grounding and Transitivity,’ see for instance Jessica Wilson,
‘No Work for a Theory of Grounding,’ Inquiry 57.5–6 (2014), pp. 1–45 for discussion
regarding the work that grounding is supposed to do in different contexts. For further
discussion regarding the formal features of ground, see Gonzalo Rodriguez-Pereyra,
‘Grounding is Not a Strict Order,’ Journal of the American Philosophical Association
For a more in-depth discussion of many of the topics of this subsection, see for
instance Gideon Rosen, ‘Metaphysical Dependence: Grounding and Reduction,’ in
B. Hale and A. Hoffman (eds.), Modality: Metaphysics, Logic, and Epistemology (Oxford
University Press, 2010), pp. 109–135.
5.5 Grounding, causation, reduction, and modality 113

controversial area, so there are no clear standard views to be presented. For

instance, take the relationship between grounding and causation. We have
been avoiding this issue so far, but it is quite typical to describe grounding
as non-causal explanation. Similarly, although many of the examples might
tempt one to think that the grounded entity reduces to whatever it is that
grounds it, this is not how grounding is typically understood. The same
goes for modality: grounding does not seem to be just modal superveni-
ence, one thing necessitating another. The following illustrative example
from Fine concerning the relationship of the physical and the mental
serves to summarize all these aspects of grounding:41

It will not do, for example, to say that the physical iscausally determinative
of the mental, since that leaves open the possibility that the mental has a
distinct reality over and above that of the physical. Nor will it do to require
that there should be an analytic definition of the mental in terms of the
physical, since that imposes far too great a burden on the anti-realist. Nor
is it enough to require that the mental should modally supervene on the
physical, since that still leaves open the possibility that the physical is itself
ultimately to be understood in terms of the mental.

Fine’s suggestion, unsurprisingly, is that the right question about the rela-
tionship of the physical and the mental is whether the latter is grounded in
the former. Let us now look at each of the mentioned aspects of grounding
in some more detail.
As Fine suggests, even if physical states causally determine mental states,
this does not mean that mental states could not have a separate existence
over and above the physical. However, this does make an assumption
about what physicalism involves – one that not everyone would necessarily
accept. The assumption is that physicalism rules out the possibility that the
mental is separate from but depends on the physical. 42 To establish phys-
icalism in this sense, a stronger connection between physical and men-
tal states should be demonstrated, such as grounding. From this, we can
infer that grounding is, in some sense, stronger than causation – although

Fine, ‘Guide to Ground,’ p. 41.
For relevant background concerning physicalism in philosophy of mind, see for
instance Daniel Stoljar, ‘Physicalism,’ in E. N. Zalta (ed.), The Stanford Encyclopedia of
Philosophy (Spring 2015 edn); see
114 Grounding and ontological dependence

it should be mentioned that this is not universally accepted. Jonathan

Schaffer, for instance, has argued that grounding is ‘something like meta-
physical causation’.43 If this were correct, it would have certain important
ramifications: Schaffer himself argues that just as there are counterex-
amples to the transitivity of causation, there are counterexamples to the
transitivity of grounding as well.44 However, one major challenge for the
position that grounding is just a type of (metaphysical) causation is that
many of the examples we have considered would appear to involve a type
of explanation that does not seem to be purely causal. Another challenge is
that grounding seems to be possible between entities of a purely abstract
nature or between abstract and concrete entities, such as sets and their
members, whereas causation is, on the face of it, a relation that concerns
only physical entities. While some attempts are currently in progress to
characterize grounding as a variety of causation, for the time being this
remains a marginal view.
What about reduction: is there a connection to grounding? If we ask
whether the mental reduces to the physical, how does that compare to the
grounding question? There are some good reasons to think that these are
not one and the same question.45 However, one problem in settling this issue
is that reduction, just like grounding, allows for a variety of different interpret-
ations. This can be naturally illustrated with the example at hand, namely
the relationship between the mental and the physical.proponent
A of strict
reductive physicalism might (but does not have to) subscribe to the view that
mental and physicalpredicates or concepts are interchangeable in such a way
that the mental predicate applies if and only if the physical predicate applies.
This type of strict reductive physicalism may not be very popular in contem-
porary philosophy of mind, but if reductionism is understood in these terms,
then it is clearly distinct from grounding. Another alternative is to conceive
of reduction asidentity.46 In that case we would automatically lose irreflexivity

43 Schaffer, ‘Grounding, Transitivity, and Contrastivity,’p. 122.

But for a defence of the ‘orthodoxy’ and against the possibility of counterexamples
to the transitivity and irreflexivity of grounding, see Raven, ‘Is Ground a Strict Partial
For discussion, see Rosen, ‘Metaphysical Dependence: Grounding and Reduction,’
especially pp. 122 ff.
For a defence of the view that reduction should be conceived as identity, see Paul Audi,
‘A Clarification and Defense of the Notion of Grounding,’ in Correia and Schnieder
(eds.), Metaphysical Grounding, pp. 101–121.
5.5 Grounding, causation, reduction, and modality 115

and asymmetry, since identity is reflexive, symmetric, and transitive. So, it

seems that grounding would again turn out to be a different relation.
However, we do not necessarily need to interpret reductionism in terms
of identity. There are at least two further alternatives that have been enter-
tained in the grounding literature: reduction characterized in terms of
fundamentality and reduction characterized in terms ofessence.47 Roughly put,
the first view suggests that when two facts explain the same thing, but one
does so in more fundamental terms, then the less fundamental explanation
reduces to the more fundamental. The notion of fundamentality will be
examined in detail in thenext chapter, but for now it will be sufficient to
note that the more fundamental terms would be thosethat are further in the
grounding chain, that is, towards the beginning of the hierarchy. The
fundamental terms would then be at the very bottom of the hierarchy. The
second, essence-based characterization suggests that if we have an account
of the essence of, say, water in terms of its microstructure, 2HO, then we
might propose that being waterreduces to being H2O. This view (or at least
this example) certainly runs into some problems, some of which we will
consider in subsequent chapters, but on this conception of reduction, the
link to grounding could perhaps be maintained:if water is reduced to its
microstructure, then we may explain what it is to be water in terms of 2HO.
This results in what has been called agrounding–reduction link: If q reduces to
p, then p grounds q.48 In sum, our judgement about the grounding–reduction
link will depend heavily on how we prefer to understand reduction.
In this connection, we may anticipate the discussion in Chapter 7 regard-
ing a proposed link between metaphysical modality and essence, suggested
by Fine, Lowe, and others: they argue that all metaphysical necessities are
grounded in essence, even though a given metaphysically necessary truth
may not be an essential truth about any one thing. One way to understand
this proposal is in terms of the grounding–reduction link, namely that
metaphysical modality reduces to essence. But this would certainly require
understanding reduction in terms not of identity, but of something else, as
some metaphysically necessary truths may require several distinct essences
to ground them.

For an overview ofthese options, see Trogdon, An
‘ Introduction to Grounding.’
This type of characterization can be found in Rosen, ‘Metaphysical Dependence:
Grounding and Reduction.’
116 Grounding and ontological dependence

Finally, we should briefly examine the relationship between grounding

and modality. The two options here are that grounding is either necessary
or contingent.49 It is perhaps more popular to think about grounding as
necessary: if x (fully) grounds y, then x modally entails y. The modal con-
tent would likely be interpreted as metaphysical modality (since ground-
ing is understood as metaphysical explanation), that is, if x (fully) grounds y,
then x metaphysically necessitates y. Or, in other words, grounding holds
across metaphysically possible worlds. The connection to ontological
dependence is obvious here, since we saw that at least certain types of
dependence can be characterized in terms of metaphysical necessitation.
If grounding were contingent, then it seems that we would lose this type
of link to ontological dependence. Most philosophers working on ground-
ing assume that grounding is necessary, but systematic treatments of the
issue are still scarce. We do not need to engage with all the details here.
Instead, we might consider an example, building on the passage quoted
from Fine. Fine claims that to understand physicalism, it is not enough
to require that the mental modally supervenes on the physical. This, he
continues, leaves open the possibility that the physical could still be some-
how understood in terms of the mental. Such a result may seem peculiar,
but it is perfectly consistent with the modal picture being proposed: the
physical could (metaphysically) necessitate the mental, but this still leaves
open what grounds the physical itself, and it is possible that it is the men-
tal that somehow helps to explain the physical. So it appears that meta-
physical necessitation is not sufficient for grounding, but if grounding is
necessary, then grounding of course entails necessitation. Hence, if the
physical grounds the mental, then the physical also metaphysically neces-
sitates the mental.

5.6 Grounding and truthmaking

This final section of the chapter deals with an important issue, the
relationship between grounding and truthmaking. The truthmaking litera-
ture is somewhat older than the modern grounding literature, but the issue
is complicated by the fact that some of the early grounding literature was

For a much more comprehensive discussion of this issue, see Kelly Trogon,
‘Grounding: Necessary or Contingent?’, Pacific Philosophical Quarterly 94 (2013),
pp. 465–485.
5.6 Grounding and truthmaking 117

also truthmaking literature.50 Truthmaking is often characterized roughly

as follows. Some entity x makes it true that p, where x is a truthmaker and p
is a truthbearer, typically a proposition. The relation between a truthmaker
and a truthbearer is thought to be the truthmaker relation. The litera-
ture on truthmaking is vast and there is not all that much agreement, for
instance, about what kind of things could be truthmakers (they could be,
e.g., ‘states of affairs’ or just ‘entities’). What interests us are the attempts
to clarify the truthmaking relation, sometimes said to be the in virtue of
relation: we say that p is true in virtue of x. An influential attempt to clar-
ify this ‘in virtue of’ relation is in terms of truth-grounding: some entity x
grounds the truth of some proposition p if p is true in virtue of the exist-
ence of x. So, the entities which ground the truth of p are p’s truthmakers.
The idea here is that the ‘in virtue of’ relation could be defined as the con-
verse of grounding: x grounds p if and only if p is true in virtue of x. If this
is correct, then grounding would seem to have an important role to play in
truthmaker theory. To be more precise, on an understanding of grounding
as a relation between facts, we may define truth-grounding as follows:

Truth-grounding If the fact that x exists helps ground the fact

that p is true, then x helps ground the truth of p, i.e., x is a (partial)
truth-ground for p.

However, it is highly controversial whether this type of connection

between truthmaking and grounding in fact holds. One reason for the con-
troversy is that truth-grounding violates an important feature of traditional
truthmaking theory, namely that truthmaking is generally understood to
be a cross-categorical relation between an entity or entities and a truth, not
merely a relation between facts, as wehave taken grounding to be.51 Friends
of truth-grounding could of course insist that it is a mistake to conceive of
truthmaking as cross-categorical and argue in favour of a unified account of
explanation in terms of grounding, but it is unclear whether it is possible

In particular, see Gonzalo Rodriguez-Pereyra, ‘Why Truthmakers?’, in H. Beebeeand
J. Dodd (eds.), Truthmakers: The Contemporary Debate (Oxford University Press, 2005), pp.
17–31, and Benjamin Schnieder, ‘Truth-Making Without Truth-Makers,’ Synthese 152.1
(2006), pp. 21–46. However, the contemporary truthmaking literature dates from as
early as Kevin Mulligan, Peter Simons, and Barry Smith, ‘Truth-Makers,’ Philosophy and
Phenomenological Research 44.3 (1984), pp. 287–321.
On this, see David M. Armstrong, Truth and Truthmakers (Cambridge University Press,
2004), p. 5.
118 Grounding and ontological dependence

to maintain enough of the srcinal motivation for truthmaker theory if this

move is made.52
Many grounding theorists are in fact sceptical about the link between
grounding and truthmaking. Fine, for instance, is very much opposed to
such a link.53 At best, Fine thinks, we could point to a necessary connection
of the following type: if x grounds p, then the fact x will be a truthmaker
for p. But such a link would not be very informative, for it might not hold
to the other direction. Even if we discover that some propositionp is made
true by some fact x, it might turn out thatx is not a sufficient groundfor p.
The shaky link between truthmaking and grounding serves to highlight
one final and important question about grounding, which we have not dir-
ectly addressed yet: is grounding univocal? In other words, are there sev-
eral different types of grounding or are all genuine examples of grounding
analysable in terms of one common notion of ground? Univocalism about
grounding is commonly assumed without argument.54 To be precise, uni-
vocalism as we are here understanding it suggests that ‘grounding’ should
be considered in terms of a single notion of dependence, which may or
may not be analysable in terms of other notions, depending on the type of
univocalism. We have been treating grounding as univocal so far, but the
case of truthmaking in particular might drive some towards distinguishing
between different types of grounding. This view comes with certain risks,
however, as there are those who would regard non-univocalism to be evi-
dence of the incoherence of the whole notion of ground and hence driv-
ing us towards scepticism about grounding.55 However, given that there is
disagreement even about some of the formal features of ground, it could
be that we can arrive at a better sense of grounding if we consider it a fam-
ily of relations (if it is a relation!), like ontological dependence. That way
we could define different subspecies of grounding even with distinct for-
mal features, while they would all fall under the general notion of ground.
Such an approach has its appeal, but the question then becomes how to

Again, see Tahko, ‘Truth-Grounding and Transitivity.’
Fine, ‘Guide to Ground,’ pp. 43–46.
For instance, Trogdon assumes univocalism in his ‘An Introduction to Grounding,’ sug-
gesting that it i s a reasonable starting point. Trogdon also associates this type of univo-
calism with Rosen.
For a sceptical response, see Chris Daly, ‘Scepticism About Grounding,’ in Correia and
Schnieder (eds.), Metaphysical Grounding, pp. 81–100.
5.6 Grounding and truthmaking 119

specify the general notion of ground. If any given example of grounding

is an example of some subspecies of the general notion, but the different
examples vary to a large extent, then it is not clear why we should consider
them to be subspecies of a general notion of ground rather than distinct
notions altogether. Indeed, if this route is taken, then the general notion of
ontological dependence might serve us better.
But we should not end with such a sceptical note. Grounding has quickly
become a central area of research in analytic metaphysics and it is no
doubt one of the most intriguing themes in metametaphysics. At its best,
it promises to give us a thorough understanding of metaphysical explan-
ation. Even failing that, the numerous applications of grounding certainly
claim to do a lot of metaphysical work. As we have already seen in passing,
one the most important applications both of grounding and ontological
dependence concerns fundamentality and the hierarchical structure of
reality. This is the topic that we turn to in the next chapter.
6 Fundamentality and levels
of reality

This chapter concerns the view that reality comes with a hierarchical struc-
ture of ‘levels’. This view has a long history and it remains very popular.
Our everyday experiences as well as scientific practice seem, prima facie, to
strongly support such a view, since scale is a major factor in both of them.
The existence of a scale suggests that there is some way to structure reality
and this idea is naturally combined with the thought that this structure
can be described in terms of ontological dependence – a tool which is now
at our disposal. The relevance of scale becomes apparent when we consider
parts and wholes, which are studied in mereology: we talk about subatomic
particles constituting atoms, atoms constituting molecules (and other
superatomic structures), and molecules constituting everything we see
around us. To express this in terms of ontological dependence we might
say that a whole depends for its existence on its parts (although not everyone
would agree that this is the correct direction of the dependence relation,
as we will see). Fundamentality comes in when we consider whether there
is an end to this chain of dependence: do we ever reach the smallest parts?
That is, is there a fundamental, ‘bottom level’, or does the hierarchical
structure of reality continue ad infinitum? The received view has long been
that there indeed is a fundamental level that everything else ‘stands on’.
The fundamental level is usually thought to be at the smaller end of the
spectrum: atomism suggests that there are certain (subatomic) indivisible
simples, particles that are fundamental or ontologically independent. But
we must immediately note that the fundamental level must not necessarily
be at the bottom, the smaller end – the fundamental end could also be at
the top; that is, the universe as a whole could be considered fundamental,
to be prior to its parts.
We will look into all of these options in much more detail below. But
before that, it should be mentioned that some recent work on this topic,

Fundamentality and levels of reality 121

taking heed of contemporary physics, has gone some way towards refuting
the idea of ‘levels’ altogether. Without the hierarchical view introduced
by the levels metaphor, talk about a fundamental level seems problematic.
So before we can get started, we will have to clarify a number of issues,
such as the levels metaphor itself.
Further problems are introduced when we try to make sense of the pos-
sibility that there could be no fundamental level at all. While future phys-
ics may help to clarify the issue, it is not clear that any amount of empirical
work will conclusively settle it. An important question here is whether it
is possible that the levels not only go on ad infinitum, but are also infin-
itely complex. So far we have seen that each consecutive level brings with
it certain new properties, but could it turn out that at some point it’s just
‘turtles all the way down’? This is the idea that there is an infinity of levels
but the same structure repeats and hence we would have infinity without
complexity.1 But before we attempt to make sense of these ideas, we should
get clearer about the state of the art regarding fundamentality.
The cone in Figure 6.1 represents the hierarchical structure of reality,
with the smallest thing(s) at the bottom and the universe as a whole at

Figure 6.1

This type of repetitive structure has been coined as ‘boring’ in Jonathan Schaffer, ‘Is
There a Fundamental Level?’, Noûs 37 (2003), pp. 498–517. For further discussion on
this type of boring infinite descent, see Tuomas E. Tahko, ‘Boring Infinite Descent,’
Metaphilosophy 45.2 (2014), pp. 257–269.
122 Fundamentality and levels of reality

the top. Why a cone? Because it reflects the idea that there is a scale to the
structure of reality. Another reason: it seems natural to think that there
are fewer different kinds of things at the smaller end of the spectrum. The
Standard Model of particle physics currently postulates that there are 61
elementary or fundamental particles (if we count particles and their cor-
responding antiparticles as well as the various colour states of quarks and
gluons). In contrast, the larger end of the spectrum may be considered
to encompass all the different kinds of things there are in reality – per-
haps even infinitely many different kinds. The lines at each end of the
cone represent the hypothetical bottom and top levels. The dotted sections
represent the possibility of infinite descent and infinite ascent, beyond the
hypothetical bottom and top levels.
Four options immediately arise regarding the hierarchical structure of

(1) Closed (i.e., the chain of dependence terminates) at both ends of

the cone.
(2) Open at the top, but not at the bottom.
(3) Open at the bottom, but not at the top.
(4) Open at both ends.

If both ends of the cone are closed, then we might consider either one
of them as fundamental. If only one end is closed, then it appears that only
the closed end could be considered fundamental, whether it is the top or
the bottom. This follows directly from the idea that at the fundamental end
the chain of dependence must terminate. The first three options enable dif-
ferent varieties of metaphysical foundationalism, which, in its broadest sense,

states simply that there is a fundamental level. The fourth option, how-
ever, cannot support metaphysical foundationalism, as there is neither a
top nor a bottom level. In this case, only some sort of metaphysical infinitism
would seem to be available. However, this description is somewhat simpli-
fied, as we have not yet said much about what ‘fundamentality’ amounts
to. We will do so in due course.
There are three questions that this chapter is primarily concerned with:

1. Is there a fundamental level of reality, at either end of the spectrum?

2. If there is, what is ‘fundamental’ about it?
3. Supposing there is such a level, how can we know what it is like?
Fundamentality and levels of reality 123

Many philosophers working in metaphysics may be inclined to defend

a positive answer to the first question, or at least they are likely to give it
a shot. It would be quite a task to conclusively settle the first question –
that is not the aim of this chapter – but it will be worthwhile to research
whether there is any hope of settling it in the first place. But it is perhaps
the second, methodological question, as well as the third question which is
closely related to the second, that are the most interesting, and they have
indeed received increasing attention in recent literature.
Given the recent explosion in the literature concerning fundamentality
on the one hand and the relevant scientific issues on the other, it is not
possible to present a comprehensive survey. It is, however, necessary to
discuss both scientific and philosophical issues in order to study each of
the three questions. We might start with a critical view by James Ladyman
and Don Ross,2 who argue that reality is not organized into levels in the
first place and that we have some good reasons to think that there is no
fundamental level. However, before the discussion can rightly begin, two
issues must be clarified. First, is there a plausible interpretation of the
‘levels’ metaphor? Secondly, what is the relevant understanding of ‘fun-
damentality’? Both of these issues will receive a preliminary treatment
in the first section of this chapter, where we will also use some of the
tools that we learned about in Chapter 5. In the second section we will
consider some arguments in favour of and against so-called mereological
fundamentality , which is the usual or at least a very common understanding
of ‘fundamentality’.3 After this, the discussion gets somewhat more con-
troversial and technical, and the reader is advised to refer to cited material
where needed or simply skip some sections of the chapter unless inter-
ested in the more speculative content.
The third section attempts to specify some of the more technical
assumptions behind fundamentality, especially one notion which is often
relied upon when trying to specify fundamentality, namely well-foundedness.
This is a notion with set-theoretical srcins. We will consider the role of
well-foundedness in this discussion and see if it can indeed be of use in
clarifying the key idea behind fundamentality. We will also take advantage

See J. Ladyman and D. Ross (with D. Spurrett and J. Collier), Every Thing Must Go (Oxford
University Press, 2007), especially pp. 4, 53–57, and 178–180.
See for instance Schaffer, ‘Is There a Fundamental Level?’
124 Fundamentality and levels of reality

of the discussion in Chapter 5 to identify what type of ontological depend-

ence is relevant in the present discussion. With these specifications in
place, in the fourth section we will attempt to define a broader under-
standing of ‘fundamentality’, namely generic ontological fundamentality. One
motivation to search for a more generic understanding is that mereological
fundamentality faces some serious challenges. This takes us to a discussion
of an alternative understanding of the notion of ‘part’. Finally, an analysis
of evidence from physics regarding fundamentality will follow in the fifth
section. This section is obviously heavy on the physics and may be safely
skipped by readers who do not have a specific interest in the issue.

6.1 The ‘levels’ metaphor

Our inquiry begins with an analysis of the ‘levels’ metaphor: what do

we mean when we say that reality is organized into levels? In philoso-
phy, the classic account of a hierarchical, layered understanding of real-
ity comes from Paul Oppenheim and Hilary Putnam.4 They suggest that
levels of reality are based on science and that the hierarchy is manifested
by a reduction: macrophysical phenomena will be reduced to microphys-
ical phenomena as a part of a general physicalist reduction, and the ‘lev-
els’ metaphor describes this layered, reductive structure. Specifically, the
entities of a higher level of reality can be reduced to their parts – entities of
a lower level of reality – and in fact are already contained in the lower level.
Therefore, all entities, as they ultimately reduce to the fundamental level,
are already contained in it. For example, a biological organism occupies
a relatively high level of reality. We can explain the biological processes
of the organism in terms of lower-level phenomena such as biochemical
reactions, reducing the biological organism to its cellular structure. But
we can go even further, reducing the cellular structure to molecular struc-
ture and the chemical reactions to quantum chemistry, taking into account
the atomic and subatomic constituents of the molecules in the organism’s
cells. Before long, we will have reduced the biological organism to funda-
mental physics, presumably reaching the lowest level of reality.

Paul Oppenheim and Hilary Putnam, ‘Unity of Science as a Working Hypothesis,’ in
H. Feigl et al. (eds.), Concepts, Theories, and the Mind-Body Problem, Minnesota Studies in
the Philosophy of Science (Vol. II, pp. 3–36) (Minneapolis: University of Minnesota
Press, 1958).
6.1 The ‘levels’ metaphor 125

While this description may appear plausible at first sight – and no

doubt there are plausible reductions of this type – the Oppenheim–Putnam
(henceforth OP) account has been almost universally rejected. The primary
worry is the built-in reductionism of the account, for it is very difficult
indeed to provide a reductive principle which would explain all macro-
physical phenomena in terms of fundamental physics. One could perhaps
appeal to future physics here, suggesting that eventually we will be able to
account for everything in terms of fundamental physics, but any argument
that has to appeal to future physics will suffer from the obvious problem
that we do not yet have a clear idea what future physics will look like. In
any case, before such a reduction can be demonstrated, there is no reason
to just assume that it is possible. There are also other concerns: Ladyman
and Ross reject this picture because of its commitment to atomism – the
thesis that all matter is constituted of indivisible ‘atoms’, familiar from the
pre-Socratic philosopher Democritus. This is a view which they deny.5
All in all, the worry is that the OP account imposes much too strong a
requirement: it requires that each level of reality corresponds to a theory
in such a way that we can start, say, from psychology, and proceed all the
way to microphysics, via biology and chemistry, reducing each discipline to
the theory on the next level down in the hierarchy. It is easy to see that this
requirement is extremely difficult to fulfil. Despite tremendous attempts,
completely reducing psychology to, say, neurochemistry has not been
achieved. This is the case even with seemingly more promising reductions,
such as the reduction of chemistry to physics. One problem here is that it
would first have to be specified what kind of reduction we even have in
mind. For instance, can we account for the notion of ‘chemical substance’
in quantum mechanics and would this constitute an appropriate reduc-
tion?6 This is far from straightforward, and certainly not as straightforward
as the OP account suggests. However, more recently there has been a surge
of alternative accounts concerning the ‘levels’ metaphor, some of which
may fare better.
Some of the recent accounts of levels are indeed non-reductive, and also
deny the commitment to atomism that Ladyman and Ross find troublesome.

See Ladyman and Ross, Every Thing Must Go, p. 47.
For discussion, see Jaap van Brakel, ‘Chemistry and Physics: No Need for Metaphysical
Glue,’ Foundations of Chemistry 12 (2010), pp. 123–136.
126 Fundamentality and levels of reality

One such approach comes from Alexander Rueger and Patrick McGivern.7
They suggest that the hierarchy of reality should be understood as a
sequence, not of entities, but of behaviours of entities. These behaviours,
they claim, are not ordered according to spatial part–whole relations as
in the OP account – this is what helps to avoid reductionism. This type of
account might even be motivated by current physics:

When physicists talk about levels, they often do not have in mind a
mereological ordering of entities. Instead, what they describe is best
understood as a stratification of reality into processes or behaviours at
different scales. To describe a system’s behaviour at a particular scale,
we first specify a set of equations that represent the relevant features of
the system at that scale. It is then the solutions to those equations – for
instance, the integration of an equation over some time interval – that
describe the behaviour of the system on that scale. Note that ‘behaviour’ is
understood very broadly here as the distribution of properties of a system
over space and/or time.8

As Ladyman and Ross are opposed to the ‘levels’ metaphor in general, they
would presumably not be satisfied with this alternative approach. They are
not alone in questioning the metaphor.9 Moreover, the notion of ‘scales’
used by Rueger and McGivern would seem to do much of the work of the
srcinal ‘levels’, so it is not entirely clear that this account manages to
avoid the mereological connotations of levels. There are also those who
deny that there is a fundamental level, but accept the levels metaphor in
some form.10
But why exactly do Ladyman and Ross abandon the levels metaphor –
what is so problematic with the mereological ordering? One of their major

concerns appears to be that the classic accounts simply assume atomism

and take granularity to be the primary criterion for distinguishing levels.
However, if something along the line suggested by Rueger and McGivern
is feasible, then the atomistic assumption may not be necessary for the

Alexander Rueger and Patrick McGivern, ‘Hierarchies and Levels of Reality,’ Synthese
176 (2010), pp. 379–397.
Rueger and McGivern, ‘Hierarchies and Levels of Reality,’ p. 382.
For further discussion, see John Heil, From an Ontological Point of View (Oxford: Clarendon
Press, 2003), Ch. 2.
E.g., Schaffer, ‘Is There a Fundamental Level?’; Andreas Hüttemann and David
Papineau, ‘Physicalism Decomposed,’ Analysis 65 (2005), pp. 33–39.
6.2 Mereological fundamentality 127

levels metaphor. Another concern that Ladyman and Ross have is that the
mereological ordering is not supported by current physics. They contend
that the levels metaphor understood as a mereological structure ordered
by part–whole relations fails as there is no good evidence for mereological
atomism coming from physics.11 We’ll look into their objection in more
detail below. But as we will see in due course, mereological atomism is not
built-in to the notion of fundamentality or ‘levels’. Let us now try to specify
the sense of ‘fundamentality’ at play in this discussion.

6.2 Mereological fundamentality

We have already been talking about mereological atomism and its con-
nection to fundamentality, but we have not yet defined this conception of
fundamentality. Here is a simple definition:

Mereological fundamentality (MF) The world is organized into

mereological levels and the fundamental level is at one end of the

mereological scale.

Note that (MF) has two parts: the mereological hierarchy, and the idea
that fundamentality is a thesis about the fundamental mereological level.
(MF) comes in two primary forms, depending on which end of the mereo-
logical scale is considered fundamental. An additional commitment that
those in support of (MF) may have is that the entities at the fundamental
level have the highest degree of ‘reality’. This type of view about ‘degrees
of being’ can be traced all the way to Aristotle, but it has proven some-
what difficult to explicate this view and it remains unclear what the link
is between such a view and recent work on fundamentality.12 Metaphors
abound, but typically the idea is that x is fundamental or ontologically inde-
pendent in this sense if and only if nothing grounds x.13 The notion of onto-
logical independence is familiar to us from Chapter 5, but we can now
make some further use of it.

Ladyman and Ross, Every Thing Must Go, pp. 53–57.
For a recent attempt to clarify this link, see Kris McDaniel, ‘Degrees of Being,’
Philosophers’ Imprint 13.19 (2013). McDaniel argues that there is indeed such a link and
that we can (and should) make use of it.
See Jonathan Schaffer, ‘On What Grounds What,’ in D. Chalmers, D. Manley, and
R. Wasserman (eds.),Metametaphysics (Oxford University Press, 2009), p. 373.
128 Fundamentality and levels of reality

Furthermore, although the above definition of (MF) does not state it,
(MF) is generally combined with the idea that the there is an asymmetric
ontological dependence relation from one end of the scale to the other.
It is the direction of this dependence which divides proponents of (MF)
into pluralists and monists , although we should note that pluralism and
monism are views independent of (MF) and hence variations of plural-
ism and monism without (MF) are possible. 14 When combined with (MF),
the pluralists hold that the direction of the dependence is from the
larger to the smaller, resulting in mereological atomism – this is, per-
haps, the standard view. The monists think that the parts are depend-
ent on the whole and hence that there is only one fundamental entity,
namely the universe. One proponent of the monist view is Jonathan
Schaffer, who considers only substances to be fundamental, and further,
that there is exactly one substance, namely the cosmos. On this under-
standing, a substance is a basic or fundamental, ontologically independ-
ent entity. This is roughly how Aristotle might characterize the notion
of ‘substance’. It is the cosmos, conceived as a substance, which is prior
to its (arbitrary) parts. Schaffer’s Spinoza-inspired priority monism , how-
ever, does not assume mereological atomism. In fact, one of Schaffer’s
arguments in favour of priority monism is that it is compatible with the
possibility of atomless gunk , whereby matter is infinitely divisible ‘gunk’
and matter or objects have no smallest parts; there are no mereological
simples. The question of gunk is of some importance here, as those com-
mitted to (MF) + pluralism and the idea that the dependence relation
goes from the larger to the smaller will struggle to accommodate the
infinite divisibility of matter entailed by a ‘gunky’ ontology – we will
return to this shortly.
Even though (MF) combined with mereological atomism is something
of a default view when it comes to fundamentality, it is also the most
resisted: as we saw above, Ladyman and Ross strongly oppose the view.
Most opponents of (MF) are worried about the additional commitment to
mereological atomism. The specific worry that Ladyman and Ross have
is directed precisely against mereological atomism: none of the various
atomistic conceptions can stand the test of contemporary physics. The

For further discussion on the monism/pluralism issue as well as an alternative account
of fundamentality, see Kelly Trogdon, ‘Monism and Intrinsicality,’Australasian Journal of
Philosophy 87 (2009), pp. 127–148.
6.2 Mereological fundamentality 129

phenomenon to which they appeal is famous: ‘quantum entanglement’.

Since Ladyman and Ross consider the srcinal formulation of the idea by
Erwin Schrödinger to be difficult to improve upon, let us quote it here
as well:

When two systems, of which we know the states by their respective

representatives, enter into temporary physical interaction due to known
forces between them, and when after a time of mutual influence the
systems separate again, then they can no longer be described in the same
way as before, viz. by endowing each of them with a representative of its
own. I would not call that one but rather the characteristic trait of quantum
mechanics, the one that enforces its entire departure from classical lines
of thought. By the interaction the two representatives [the quantum states]
have become entangled.15

The details regarding the phenomenon of quantum entanglement can be

set aside here. What is important is that the question boils down to one

regarding individuality. If the atomistic conception of fundamentality were

correct, then we would expect to find some individuals, some mereological
atoms which are, in the relevant sense, independent. But given the phe-
nomenon of quantum entanglement, it looks as if the best candidates for
individuals, the elementary particles mentioned in the Standard Model,
do not behave as we would expect on the basis of the philosophical dis-
cussion.16 The core of the critique by Ladyman and Ross is that there is
nothing in fundamental physics that corresponds with the atomistic idea
of ‘simples’ – instead you will find a very complex arrangement of systems
relating to each other in ways that we can only model probabilistically,
such as the one described by Schrödinger in the passage above. However,
in fairness it should be noted that the interpretation of these issues is still
controversial – eighty years after the publication of Schrödinger’s paper.
For instance, building on the work of the physicists David Bohm and Basil

Erwin Schrödinger, ‘Discussion of Probability Relations Between Separated Systems,’
Proceedings of the Cambridge Philosophical Society 31 (1935), p. 555; quoted in Ladyman and
Ross, Every Thing Must Go, p. 19.
However, the inference from quantum entanglement to the failure of individuality
and hence mereological atomism can be questioned. For a discussion of this issue,
see Mauro Dorato and Matteo Morganti, ‘Grades of Individuality. A Pluralistic View of
Identity in Quantum Mechanics and in the Sciences,’ Philosophical Studies 163 (2013), pp.
130 Fundamentality and levels of reality

Hiley, interpretations of quantum entanglement that do not drop individu-

ality have been and are still being developed.17
But why is the atomistic conception of fundamentality so popular in the
first place, if we know that contemporary physics is unlikely to support the
picture? Sometimes a non-atomistic conception of fundamentality is con-
sidered. Ned Markosian discusses such a conception, but quickly dismisses
it, because abandoning mereological atomism would have unwanted impli-
cations for the proponent of fundamentality – the choice would then appear
to be between (MF) combined with atomism, and non-fundamentality. 18
One of the unwanted implications mentioned by Markosian is that it would
supposedly be highly implausible to say that nothing is ‘maximally real’ –
assuming that it is the fundamental level that has the highest degree of
reality.19 What Markosian has in mind is that if the world were ‘gunky’ and
there would be no fundamental mereological level, we would never reach
the highest degree of reality – metaphysical grounding would, as it were,
never ‘bottom out’.
A related line of thought is discussed by Ross Cameron.20 Cameron resists
the idea that there could be infinite chains of ontological dependence, also
because of some non-intuitive implications. Applying the idea to the case
of composition, Cameron notes that in a gunky world, composition would
never get ‘off the ground’. In other words, if complex objects are onto-
logically dependent on their mereological parts, then composition never
‘bottoms out’ in gunky worlds. The intuitive result, according to Cameron,
is that complex objects are not possible in gunky worlds. But such a strong
result may be alarming, for it would rule out the possibility that we live in
a gunky world, at least if we accept the plausible idea that complex objects
are ontologically dependent on their parts. Indeed, the intuitive appeal of

17 For a very recent example, see Paavo Pylkkänen, Basil J. Hiley, and Ilkka Pättiniemi,
‘Bohm’s Approach and Individuality,’ in A. Guay and T. Pradeu (eds.), Individuals Across
the Sciences, Ch. 12 (Oxford University Press, 2015).
See Ned Markosian, ‘Against Ontological Fundamentalism,’ Facta Philosophica 7 (2005),
pp. 69–84.
However, degrees of reality do not necessarily have to be associated with mereological
Ross P. Cameron, ‘Turtles All the W ay Down: Regress, Priority and Fundamentality,’
Philosophical Quarterly 58 (2008), pp. 1–14.
6.2 Mereological fundamentality 131

this approach has been questioned.21 There are various approaches that one
could take in order to resist the result and we will consider one of them in
detail below, but let us first briefly examine the intuition at stake. Cameron
and many others writing about fundamentality and the idea that compos-
ition couldn’t ‘get off the ground’ without a fundamental level typically
acknowledge that this appeal to intuition hardly constitutes an argument in
favour of fundamentality. An obvious problem here is that our ‘folk’ intui-
tions may not be a reliable source of evidence when it comes to matters
such as the constraints on the fundamental structure of reality – some-
thing that we will discuss further in Chapter 8. One complication concerns
the way in which we think about the structure of space-time. If space-time
itself is made up of zero-dimensional space-time points and hence has no
internal structure, no smaller parts, then this ‘pointy’ space-time would
seem to rule out a certain type of infinite regress by its very nature: surely
those space-time points themselves do not depend for their existence on
anything and hence are, in one sense, fundamental? On the other hand,
it’s standard to think that any given collection of these space-time points
composes a space-time region. These regions are identified in terms of their
composite space-time points. This discussion is related to various issues
in the metaphysics of space and time, but what matters to us now is that
there are two different ways to conceive of the relationship between the
points and the regions; as John Hawthorne puts it:

In antiquity it was common for philosophers to admit the existence of

points only as derivative entities and, relatedly, to think of facts about
points as derivative from more fundamental facts about extended objects.
On this conception, points exist only derivatively, as limits of lines, and

surfaces without thickness exist only derivatively as limits of bulky objects.

Nowadays, however, it is common for metaphysicians to hold both that
space-time regions are less fundamental than the space-time points that
compose them and that facts about the intrinsic character of space-time
regions are less fundamental that the facts about the intrinsic character of
space-time points and facts about how the space-time points are arranged.

See for instance Matteo Morganti, ‘Dependence, Justification and Explanation: Must
Reality Be Well-Founded?’,Erkenntnis 60.3 (2015), pp. 555–572.
John Hawthorne, ‘Three-dimensionalism vs. Four-Dimensionalism,’ in T. Sider,
J. Hawthorne and D. W. Zimmerman (eds.), Contemporary Debates in Metaphysics
(Oxford: Blackwell Publishing, 2008), p. 264.
132 Fundamentality and levels of reality

So there is at least one open question regarding the relative fundamen-

tality of the regions and the points. The relevant intuition that Cameron
and others are appealing to may presuppose the view that points are more
fundamental – this is of course what Hawthorne identifies as the more
popular view, but there is also the empirical question regarding the struc-
ture of space-time. General relativity (GR) treats space-time as ‘pointy’ and
hence may support the view at hand, but there are other options as well.
For instance, if we apply quantum field theory to this debate, the result
could very well be quite different – but this is something that we have
to leave aside here. In fact, Ladyman and Ross also point out that, as is
well known, the ‘conflict’ between (GR) and quantum mechanics is par-
ticularly problematic when considering small scales, such as the supposed
space-time points.23 In any case, there is also a separate metaphysical ques-
tion of how composition works for space-time regions. In particular, if
we forget about the ‘pointy’ conception and just ‘zoom in’ to a spacetime
region, then what would be problematic in assuming that we encounter
infinitely many smaller regions, without ever reaching the supposedly fun-
damental space-time points? Here, the srcinal intuitive pull of the failure
of composition to get ‘off the ground’ may be much weaker, unless it is
simply assumed that it is built in to the notion of composition that every-
thing is made up of fundamental, ‘pointy’ objects.
Now, to sum up the discussion so far: infinite chains of ontological
dependence are typically considered problematic, but the primary motiva-
tion for finding them problematic is that they would have non-intuitive
results, such as causing problems for an intuitive view about complex
objects. This has led some, like Cameron, to reject the possibility of the
actual world being gunky, as there clearly are complex objects. But it is
not obvious that infinite sequences of ontological dependence in them-
selves are necessarily problematic – we did not encounter any reasons to
think so in the discussion concerning ontological dependence in Chapter 5
and there may be ways to interpret composition that have no such conse-
quences. Moreover, what Cameron and many others seem to be concerned
about are infinite sequences of mereological dependence.24 This is dependence

Ladyman and Ross, Every Thing Must Go, p. 19.
Compare with Jaegwon Kim, Essays in the Metaphysics of Mind (Oxford: Oxford University
Press, 2010), p. 183. Kim prefers ‘mereological dependence’ instead of the term ‘mereo-
logical supervenience’ and contrasts it with causal dependence.
6.3 Further specifications: well-foundedness and dependence 133

between a whole and its parts, and it is, of course, the sense of dependence
that seems to be underlying (MF). One upshot of all this is that there is a
wide agreement about the incompatibility of pluralism + (MF) + gunk (or
metaphysical infinitism more generally). Recall that pluralism in this con-
text refers to the idea that there is a plurality of fundamental objects (or
kinds) at the fundamental level – a view that goes naturally together with
mereological atomism. Most seem to be inclined to drop gunk and keep
pluralism + (MF). But recall also that one could drop pluralism and adopt
monism instead, as Schaffer does – Schaffer’s priority monism suggests
that the universe as a whole is prior to its parts. In fact, Schaffer considers
it an argument in favour of his priority monism that it is able to accom-
modate gunky ontologies. But we’ve also seen reasons to doubt (MF) above;
the preceding considerations would seem to undermine at least one moti-
vation for (MF), because one could accept that mereological dependence
cannot be infinite while denying that the fundamental level consists of
mereological simples. Given this, as well as the forceful objections against
(MF) from physics by Ladyman and Ross (and others), the present status of
(MF) is in doubt. But rather than try to pass judgement about any of this, we
should open up the assumptions behind (MF) and contrast it with a more
general sense of fundamentality.

6.3 Further specifications: well-foundedness

and dependence

We saw in the previous section that (MF) concerns mereological depend-

ence. But the notion of fundamentality can be associated with another
type of dependence – we already know from Chapter 5 that ontological
dependence is in fact a rich family of notions. The idea that some notion
of ontological dependence will help us make sense of the fundamental
level of reality or of levels of reality in general came up with our discus-
sion of Cameron’s view. Following this discussion, we might say that if
there is a fundamental level, it could be defined in terms of ontological inde-
pendence, that is, the existence of anything fundamental cannot depend
on something else. This idea is commonly expressed with reference to
the notion of well-foundedness; we will now turn to its analysis. The notion
has its srcins in set theory and the axiom of foundation, which states that
a set cannot contain an infinitely descending sequence (of membership).
134 Fundamentality and levels of reality

In slightly different terms, the axiom of foundation requires that every

non-empty set S contains an element that is disjoint from S; as a conse-
quence, a set cannot be its own element. The upshot of the axiom of foun-
dation is well-foundedness, which can be applied to any order (which we
have generally called a ‘chain’ above), such as proper parthood. Something
x is a proper part of y when x and y are not identical, but y contains x in its
entirety. This is really just the intuitive conception of parthood, but the
qualifier ‘proper’ is needed to rule out reflexivity, that is, something being
a part of itself. To put this in terms of the notion of grounding, which is
typically used in this context, we could say that everything must be fully
grounded in something fundamental.25 With this in place, we can define
well-foundedness as follows:

Well-foundedness (WF) An order is said to be well-founded if every

non-fundamental element of the order is fully grounded by some
fundamental element(s).26

If an order is thought to be well-founded, then infinite descent is ruled

out in that order. As we have seen, an atomless, gunky ontology does
not satisfy (WF), as in such an ontology we have an infinite descending
proper parthood order. However, although in this context it is the clas-
sic notion of parthood that has received the most attention, the notion of
well-foundedness is not tied to parthood. In the next section we will specify
a more general understanding of fundamentality, but for the case of (MF)
it is exactly well-foundedness with regard to proper parthood that helps us
to formulate the idea of a fundamental level.
Note that (WF) is typically not considered to rule out chains that are infin-
ite just so long as they terminate eventually, even if it would take an infinite
number of steps (or an infinite amount of time) to reach the fundamental
level. To see this, consider the idea that the universe is infinitely large,
without a ‘top level’, but having a foundation, say, as required by (MF) (i.e.,
mereological atomism). This entails that an infinite chain directed towards
the mereological atoms is possible without violating (WF): there is a fun-
damental level, it just takes infinitely many steps to reach it! This type of

For the distinction between full and partial ground, see section 5.4.
There are various other possible formulations of well-foundedness, for a recent survey,
see Scott Dixon, ‘What Is the Well-Foundedness of Grounding?’, Mind (forthcoming).
6.3 Further specifications: well-foundedness and dependence 135

idea is partially analogous to the familiar Zeno paradoxes, such as the race
between Achilles and a tortoise. As the story goes, Achilles gives a head start
to the tortoise, but in the time that it takes Achilles to reach the starting
point of the tortoise,p1, the tortoise has already reached pointp2. So Achilles
must reach point p2, but since he does not move instantaneously, the tortoise
has reached point p3 in the meanwhile; this way it seems that Achilles will
never be able to overtake the tortoise. Of course, we know very wellthat this
sequence, although infinite, does terminate– Achilles will indeed overtake
the tortoise. The case of infinite chaining is somewhat different in the sense
that it would take an infinite amount of ‘time’ to reach the fundamental
level, so it is not directly analogous. But as long as there is a fundamental
level, infinity can perhaps be dealt with. This idea can be seen in Schaffer’s
work, as well as in Karen Bennett’s.27 For instance, Bennett requires that a
well-founded chain must not be infinite at the fundamental end, even though
well-foundedness is compatible with an infinity of grounded entities. A chain
can be infinite and well-founded, that is, grounded in something fundamen-
tal. As Ricki Bliss puts it:‘A finite grounding chain is well-founded buta well-
founded grounding chain need not be finite’. The upshot is that (WF) only
requires that the grounding chain eventually terminates, even if it takes
infinitely many steps to reach the fundamentallevel.
Given that we are now laying the foundations for a notion of funda-
mentality broader than (MF), we also have to characterize a notion of
ontological dependence that would be able to maintain the type of hier-
archical chain of dependence or order that could explicate it. We are look-
ing for asymmetric sequences of ontological dependence that could start,
for instance, from macrophysical objects and terminate at a fundamental
level. It seems that only a rather general sense of ontological dependence
will do the trick. One possibility which would appear to be sufficiently gen-
eral would be a sequence of generic existential dependence:29’

Generic existential dependence (GXD) x depends for its existence upon

Fs =df Necessarily, x exists only if some Fs exist.

See Schaffer, ‘Is There a Fundamental Level?’, pp. 509–512; Karen Bennett, ‘By Our
Bootstraps,’ Philosophical Perspectives 25 (2011), p. 34.
Ricki L. Bliss, ‘Viciousness and the Structure of Reality,’Philosophical Studies 166.2 (2013),
p. 416.
Originally introduced in section 5.1.
136 Fundamentality and levels of reality

Where F is a general term: necessarily, x exists only if another object of the

type F exists. (GXD) refers to an object’s existence requiring the existence
of an object of a certain sort rather than a specific object (as opposed to
rigid existential dependence). For instance, any given w
ater molecule cannot
exist unless hydrogen atoms exist. Or, more precisely, any given water mol-
ecule cannot exist unless the physical constraints required for the existence
of hydrogen atoms are fulfilled, which in turn requires the possibility of a
certain electron configuration, among other things. Since we are typically
interested in the ontological basis of macrophysical objects, it is perhaps
more natural to think about dependence on a specificrange of values for the
fundamental physical constants, or on certain physical principles
– we will
consider these requirements in the final section of this chapter.
With (WF) and (GXD) in place to specify a more general sense of hier-
archy, compatible with the mereological hierarchy of (MF) but also with
other ways to understand ‘levels’, we can move on to specify a generic
sense of fundamentality.

6.4 Generic ontological fundamentality

The term ‘generic ontological fundamentality’ is not established in the lit-

erature, even though the idea of a more general sense of fundamentality in
addition to the mereological version (MF) is widely recognized. Typically,
discussion about fundamentality focuses on versions of (MF). One reason
for this is simply the historical dominance of atomism. But if we are open
to the possibility of conceiving the hierarchical structure of reality in terms
of something other than the mereological scale and hence mereological
dependence, then we must make room for different conceptions of funda-
mentality as well. As we saw in the previous section, the underlying notions
of well-foundedness and ontological dependence do enable this. However,
rather than attempting to list all the possible ways to understand levels and
the relevant chains of dependence that structure them, we will here try to
define a very general sense of fundamentality, which could capture (MF) but
also many, if not all, other conceptions of fundamentality. We define this
general conception of fundamentality as follows:

Generic ontological fundamentality (GOF) The world is organized

into levels of ontological elements and the fundamental level consists of
ontologically minimal elements.
6.4 Generic ontological fundamentality 137

This definition is only provisional and it is obviously in need of further

explication, as nothing has yet been said about ‘ontological elements’ in
general and ‘ontologically minimal elements’ in particular. But since (GOF)
is supposed to be able to express (MF) as well, we can already guess at
least one thing that could take the place of ‘ontological elements’ in (GOF),
namely mereological elements. In that case, we would get the familiar
hierarchy of mereological levels and the ontologically minimal elements
would be mereological atoms – at least if (MF) is combined with pluralism.
Of course, (GOF) is just as compatible with pluralism as it is with mon-
ism, since there would be just one ontologically minimal element, the uni-
verse. But what else could the ontological elements be? That would seem to
depend on three things. First, what kind of hierarchy are we interested in?
Secondly, what kind of dependence structures that hierarchy? Thirdly, how
is the hierarchy manifested in the world, or in other words, what is the
relevant physics? All of these are difficult questions, especially the last one.
If the critique of (MF) by Ladyman and Ross that we discussed in section
6.2 is correct, then much of the relevant work in determining the correct
answers to these three questions will come from physics.
However, even if we will eventually need more input from physics, this
does not mean that we can’t say anything more at the moment. At the
very least, we can try to get a clearer idea about what is required from
ontologically minimal elements – the theoretical background of ‘ontologi-
cal minimality’. To get an initial idea, the ontologically minimal elements
could be compared to the idea of ‘minimal truthmakers’, familiar from
David Armstrong:

If T is a minimal truthmaker for p, then you cannot subtract anything from

T and the remainder still be a truthmaker for p.30

A minimal truthmaker for a given proposition can be understood as the

smallest or least encompassing portion of reality that fully grounds the
truth of that proposition. This analogy between minimal truthmakers and
ontologically minimal elements suggests that the fundamental level con-
sists of the least encompassing portion of reality. In usual accounts of min-
imal truthmakers, the least encompassing portion of reality is understood

David M. Armstrong, Truth and Truthmakers (Cambridge University Press, 2004),
pp. 19–20.
138 Fundamentality and levels of reality

in terms of parthood, which would naturally take us back towards (MF)

rather than (GOF). But there is another way to understand it, which we
might compare with what Ted Sider calls ‘ideological’ fundamentality, tak-
ing the notion of ‘ideology’ from Quine. 31
Sider uses the notion of fundamentality interchangeably with
‘joint-carving’ and ‘part of reality’s structure’. He specifies that his con-
ception of fundamentality is ideological rather than mereological or prop-
ositional. This can be further illustrated in terms of grounding: Every
non-fundamental truth is grounded by some fundamental truth. It should
be noted though that Sider distances himself from certain popular treat-
ments of ground, which we discussed in Chapter 5. In contrast to Schaffer,
Sider is not committed to entity-grounding , which points towards (MF).
Indeed, Sider claims that his ideological fundamentality is compatible with
infinite mereological descent – a gunky ontology. Similarly, Sider claims
to be able to accommodate infinite propositional descent, for instance a
proposition about a gunky object having a certain mass is true in virtue of
a proposition about the masses of its parts going one level down, and that
proposition in turn is true in virtue of a proposition about the masses of
the object’s parts on a further level down, and so on, ad infinitum.
So how should we understand Sider’s ideological fundamentality? He
specifies that some notions are fundamental in the sense that there is
a perfectly fundamental description of reality expressible by terms that
carve perfectly at the joints – the ‘joint carving’ metaphor was initially dis-
cussed in Chapter 4. Note also that Sider is committed to a type of levels
of reality view, as he thinks that there are degrees of fundamentality and
hence degrees of joint-carving – perfect joint-carving being at the ideo-
logical fundamental level. The connection between Sider’s approach and
(GOF) is that the ontologically minimal elements can also be understood
in terms of perfect joint-carving: a given term is part of the fundamen-
tal level if there is no other term that describes reality more minimally.
Note however that Sider is primarily interested in which concepts carve
at the joints (as discussed in Chapter 4). This brings with it a linguistic

For an account of minimal truthmakers, see Tuomas E. Tahko and Donnchadh
O’Conaill, ‘Minimal Truthmakers,’ Pacific Philosophical Quarterly (2015 [online], forth-
coming). For the details of Ted Sider’s ‘ideological fundamentality,’ see his Writing the
Book of The World (Oxford University Press, 2011); see also section 4.4 of this book.
6.4 Generic ontological fundamentality 139

connotation, which we may wish to distance ourselves from given the

present goal.
Sider’s picture goes some way towards clarifying the idea of ontological
minimality, but there may be a more neutral way to avoid the commitment
to (MF), with the help of the analogy concerning minimal truthmakers and
parthood. We might try to understand the notion ofparthood itself in a
more liberal manner. Arecent attempt to do this comes from Kit Fine, who
develops an alternative to the classic notion of ‘part’:

When one object is a part of another, there is a sense in which it is in the

other – not in the sense of being enclosed by the other, as when a marble
is in an urn, but more in the sense of being integral to the other. When
parts are in question, it is also appropriate to talk of a given object being
composed of or built up from the objects that it contains.32

Fine suggests that all manner of things, from sentences to symphonies

to sets, are composed of other things. Accordingly, he proposes to define
the general relation of part in terms of the operation of composition: ‘The
parts of an object are the object itself, or its components, or the compo-
nents of the components, and so on’.33 A question that remains open even
if we were to adopt this more liberal sense of parthood is whether infinite
chaining of these components is subject to similar ‘non-intuitive’ conse-
quences as the more traditional notion is. This does not follow automati-
cally though: it would seem possible that the components of a composite
object one level down are integral to the object, but if the components going
down one level further are no longer integral to it, then the infinite regress
causing the typical non-intuitive consequences wouldn’t obviously follow.
So it turns out that there are ways to understand parthood which do not
necessarily imply the non-intuitive consequences that have been associ-
ated with infinite chains of mereological dependence. This opens up the
possibility of interpreting ontologically minimal elements quite openly
indeed: the smallest, minimal ‘parts’ of reality do not need to be mereo-
logical elements at all, they can be anything that count as components, such
as structures, relations, objects, or whatever. This may enable us to over-
come at least some of the complications regarding the tension between a
classic, object-oriented metaphysics such as mereological atomism and the

Kit Fine, ‘Towards aTheory of Part,’Journal of Philosophy 107.11 (2010), p. 560.
Ibid., pp. 567–568.
140 Fundamentality and levels of reality

type of view defended for instance by Ladyman and Ross, according to which
reality is fundamentally relational or structural.34
Now that we have a clearer idea about what fundamentality amounts
to, let us briefly look at one outlier: the possibility of a repetitive version
of metaphysical infinitism – the view that the grounding chain does not
terminate. The idea is that the same structure could repeat infinitely.
This is an idea also entertained by Schaffer; he calls this type of infinite
descent ‘boring’, or in other words, the same structure keeps repeating –
it’s ‘turtles all the way down’. 35 In other words, there is no novelty in the
structure after a certain point. The part of the structure that repeats could
be of any length, as long as it starts anew at some point. What makes this
possibility interesting is that such a repeating structure might even be
enough to satisfy (GOF). To see this, consider the idea of ontologically
minimal elements again, as introduced with the definition of (GOF).
The infinite descent of a repetitive structure could allow for an onto-
logically minimal description in the sense that a description of the repeti-
tive part only needs to be supplemented with an instruction to continue as
before, for instance ‘the world stands on four elephants, the four elephants
stand on a turtle, the turtle stands on two camels, the camels stand on
four elephants, the four elephants stand on a turtle … and repeat ad infin-
itum’. No other terms than these four elephants, a turtle, and two camels
can be introduced that would describe reality more minimally – they carve
perfectly at the joints and hence could be understood to constitute the
fundamental level in the sense of (GOF). But we are treading on very con-
troversial terrain here – this option is not yet established in the literature,
even though it is receiving more and more attention.36
Summarizing, it appears that much remains to be done in order to spec-
ify the range of options with regard to fundamentality. The generic notion

For further discussion of this type of structuralist view and fundamentality, see for
instance Kerry McKenzie, ‘Priority and Particle Physics: Ontic Structural Realism as a
Fundamentality Thesis.’ British Journal for the Philosophy of Science 65 (2014), pp. 353–380;
Steven French, The Structure of the World: Metaphysics and Representation (Oxford University
Press, 2014).
Schaffer, ‘Is There a Fundamental Level?’, p.505. See also Tahko, ‘Boring Infinite
See for instance Matteo Morganti, ‘Dependence, Justification and Explanation: Must
Reality Be Well-Founded?’; Matteo Morganti, ‘Metaphysical Infinitism and the Regress
of Being,’ Metaphilosophy 45.2 (2014), pp. 232–244; Ricki Bliss, ‘Viciousness and Circles
of Ground,’ Metaphilosophy 45.2 (2014), pp. 245–256.
6.5 Fundamentality and physics 141

(GOF) sketched here is designed to be as liberal as possible, in order for the

various possible conceptions of fundamentality to be captured with one
general notion, but it must be stressed that there is not yet an established
usage for such a notion.

6.5 Fundamentality and physics

We have already discussed some of the limitations that contemporary

physics appears to impose on our discussion of fundamentality. But the
idea of a fundamental level is certainly present in contemporary phys-
ics, even if it is rarely discussed in much detail. This is of course partly
because of empirical limitations. For instance, the physicist David Bohm
discusses the possibility of an ‘infinity of levels’, stating that current
physics is entirely compatible with this possibility. He seems to think
that this is a purely empirical matter (this is what motivates Schaffer’s
liberal attitude towards the idea as well). Bohm makes one general
observation, concerning the requirement of ‘autonomy and stability’ in
modes of being, which suggests that regardless of whether there is infin-
ite descent or not, it is clear that the very idea of a ‘thing’ is intimately
connected with some kind of stability and autonomy of the thing (object,
entity, process …) in question – leaving aside the concern about indi-
viduality that we came across when discussing the critical points about
fundamentality from Ladyman and Ross. Perhaps we can say something
more about this requirement of stability, but it will require going into
some detail about the relevant physics. However, the reader should not
be alarmed if some of the details seem obscure, as this is all very specu-
lative. Since this is not a physics textbook, we will not define all the
relevant scientific terminology, but will instead assume that interested
readers have some familiarity with the topic. Those who do not wish to
engage with the remaining material may safely skip this last section of
the chapter.
Our starting point is very simple. For there to be any macrophysi-
cal objects, the forming of such objects must be possible. Consider the
minimal conditions for the possibility of stable macrophysical objects.

37 David Bohm, Causality and Chance in Modern Physics, 2nd ed. (London: Routledge,
1984), p. 95.
142 Fundamentality and levels of reality

A natural way to spell out these minimal conditions is in terms of the

physical laws that govern the forming of macrophysical objects. For
instance, it is necessary for the existence of (most) stable macrophysical
objects that molecules can bind together to form molecular complexes,
albeit some covalently bonded networks of carbon atoms such as dia-
monds or graphite lattices are examples of macrophysical, extremely
large molecules. At any rate, it is necessary for any macrophysical object
that atoms are able to form bonds to create stable molecules, and fur-
ther, that subatomic particles are able to form stable atoms. So what ena-
bles atoms to form bonds and subatomic particles to form atoms? Well,
the binding of molecules and atoms is dependent on the electron config-
uration of individual atoms, which in turn depends on the energy levels
of specific electrons and is moderated by the Pauli Exclusion Principle
(PEP). 38 Similarly, the manner in which subatomic particles form atoms
is dependent on the individual charges of subatomic particles, namely
the negative charges of the electrons and the positive charges of the
protons, where each proton consists of three quarks which make up the
total charge of the proton.
At this point it should be noted that although we have been referring
to subatomic particles , there is no reason to assume mereological atom-
ism here, bearing in mind the discussion of individuality in the second
section and the further point later on regarding the tension between a
classic, object-oriented metaphysics such as mereological atomism and
the type of view defended by Ladyman and Ross, according to which
reality is fundamentally relational or structural. All we need is a micro-
physical arrangement which enables the possibility of macrophysical
objects. Accordingly, whether or not we view electrons as particles with
some sort of individuality, we know that their behaviour is subject to
(PEP). More specifically, if we have two identical and indistinguishable
electrons, the wavefunction for the system of those two electrons must
be antisymmetric. As an example, let us consider a case of covalent

The Pauli Exclusion Principle states that no two identical fermions, such as electrons,
can have all the same quantum numbers (e.g., spin, angular momentum, energies of
electrons) at the same time. This means, for instance, that if two electrons in an atom
occupy the same orbital and hence have the same energy, then they must differ with
regard to another quantum number, namely spin.
6.5 Fundamentality and physics 143

bonding in a hydrogen molecule. For a covalent bond to form between

two hydrogen atoms, the complete wavefunction of the system, which
combines the spin and spatial wavefunctions, must be antisymmetric.
Accordingly, if the spin wavefunction is symmetric, then the spatial
wavefunction must be antisymmetric. However, only a symmetric spa-
tial wavefunction will lead to bonding, as we need an attractive force
between the hydrogen atoms, enabling them to share a pair of valence
electrons. Hence, since (PEP) requires the complete wavefunction to be
antisymmetric, we know that the spin wavefunction must be antisym-
metric for a bond to form. We also know that a third hydrogen atom
cannot bond with the two-atom hydrogen molecule, as it would neces-
sarily have an antisymmetric wavefunction with one of the hydrogen
atoms, and would therefore be repelled. As should be clear from this
description, the Pauli Exclusion Principle is crucial for the process of
covalent bond forming.
In the case of ionic bonds, such as the bond between sodium and chlorine
in sodium chloride molecules, (PEP) is responsible for the repulsive force
between Na+ and Cl ions. As the ions come closer and the wavefunctions of

their electrons start to overlap, (PEP) requires that these electrons cannot
be in the same quantum state. The situation is resolved by a change in the
energy levels of the electrons so that no two identical electrons occupy the
same quantum state. The change in the energy levels of electrons requires
energy and results in the repulsive force, typically called Pauli repulsion. This
repulsive force prevents the ions from coming any closer together. The
result is a stable sodium chloride lattice. Again, (PEP) is absolutely central
for this bonding process.
Since (PEP) governs both the bonding behaviour of atoms and the elec-
tron configuration of individual atoms, it is crucial for the emergence of
stable macrophysical objects. It is sometimes said that (PEP) is responsible
for the space-occupying behaviour of matter, as it prevents atoms from col-
lapsing together: the electrons must occupy successively higher orbitals to
prevent a shared quantum state and hence not all electrons can collapse
to the lowest orbital. Incidentally, it is (PEP) that explains why subatomic
particles can behave in a manner which is so different from macrophysi-
cal objects. The principle is key to understanding why fundamental phys-
ics cannot be viewed as a network of ‘microbangings’ – something that
144 Fundamentality and levels of reality

Ladyman and Ross accuse metaphysicians of doing in their search for ‘genu-
ine causal oomph’.39 The ‘microbangings’ model of fundamental physics is
indeed not very fruitful, but the idea of levels of reality and of a fundamen-
tal level does not need to be intimately connected with that model – this is
what we are in the process of showing here.
We’ve seen that macrophysical objects depend for their existence on
certain principles. These principles themselves depend on a distribution
of fundamental physical constants that fall within a specific range. The
dependence here is transitive: ultimately macrophysical objects depend for
their existence on the distribution of fundamental physical constants, as
will become clear in what follows.
One important fundamental physical constant associated with the sta-
bility of matter (and hence the existence of macrophysical objects) is the
elementary charge, namely 1.6021892 × 10−
coulombs. This is the charge
of a proton, whereas an electron has a negative charge of equal strength.
Interestingly, the charges of all other freely existing subatomic particles that
have a charge are either equal to or an integer multiple of the elementary
charge. Quarks, which are the constituents of protons, have charges that
are integer multiples of one third of the elementary charge, but they are
not freely existing as individiual particles. The total charge of the atom is of
course neutral. The picture gets somewhat more complicated when details
about the underlying fundamental forces are introduced; for instance, the
nucleus holds together in virtue of the strong nuclear force , which overpow-
ers the repulsive forces between the (net) positively charged quarks that
make up protons. (Quarks can form quark–antiquark pairs – mesons – and
combinations of three quarks – baryons, such as protons and neutrons –
which are collectively known as hadrons.) What is interesting for us is
whether fundamental physical constants, such as the elementary charge,
are physically necessary for the stability of matter. In other words, if the
fundamental constants had been different, would atoms still be stable?
It has been suggested that at least some of the fundamental physical
constants do, or at least could vary over time, specifically the fine-structure
constant (sometimes called the electromagnetic force coupling constant),

Ladyman and Ross, Every Thing Must Go, p. 4. By ’microbangings’, Ladyman and Ross
mean something like what has traditionally been called ’billiard-ball physics’, i.e., rad-
ically simplified physics in which particles are imagined to be tiny spheres bouncing
off each other.
6.5 Fundamentality and physics 145

which characterizes the strength of the electromagnetic interaction, and

the electron-to-proton mass ratio. However, there are experimental limitations
in determining whether fundamental physical constants do indeed vary
over time: observations commonly entangle a certain set of constants, or
assume that certain constants do not vary. It is also worth noting that if
all the fundamental physical constants were to vary together so that their
ratios would remain the same, it would presumably be impossible to estab-
lish this observationally.40
If we hope to establish some limitations concerning the possible vari-
ation of fundamental physical constants so that this variation would not
jeopardize the stability of matter, then we should direct our attention
towards the values of the constants relative to each other. It is plausible
that if all the constants were to change so that their proportions would not
change, then the stability of matter might not be in jeopardy. However,
if we change just one of the constants, this would be much more likely
to cause problems. Take the fine-structure constant, α: it is a dimension-
less constant which is expressed in terms of other physical constants,
namely the elementary charge, the electric constant, the Planck constant, and
the speed of light; the numerical value of α 1 is just over 137. A dimension-

less constant serves our purposes very well, as a change in a dimension-

less constant implies that the proportions between constants have changed.
The electron-to-proton mass ratio, β, is another good example. Indeed, had
either one of these fundamental physical constants been of a slightly differ-
ent value, then macrophysical phenomena would not be possible:

For example, if we were to allow the ratio of the electron and proton
masses β = me/mN and the fine structure constant α to change their values

(assuming no other aspects of physics are changed by this assumption –

which is clearly going to be false!), then the allowed variations are very
constraining. Increase β too much, and there can be no ordered molecular
structures because the small value of β ensures that electrons occupy
well-defined positions in the Coulomb field created by the protons in the
nucleus. If β exceeds about 5 × 10 3 α2, then there would be no stars. If

modern grand unified gauge theories are correct, then α must lie in the

For further discussion regarding fundamental physical constants especially as they are
related to laws of nature, see Tuomas E. Tahko, ‘The Modal Status of Laws: In Defence
of a Hybrid View,’The Philosophical Quarterly (2015 [online], forthcoming).
146 Fundamentality and levels of reality

narrow range between about 1/180 and 1/85 in order that protons not
decay too rapidly and a fundamental unification of non-gravitational forces
can occur. If, instead, we consider the allowed variations in the strength of
the strong nuclear force, αs, and α then roughly αs < 0.3α1/2 is required for
the stability of biologically useful elements like carbon. 41

It should be emphasized that it would not be physically possible to vary

individual constants in the simplistic manner imagined in this scenario: a
change in one constant would have ramifications throughout the range of
physical constants, as Barrow notes.42 However, it does seem clear that fun-
damental physical constants must fall within a specific, fairly narrow range
to enable the emergence of macrophysical objects.
What does all this tell us about the fundamental level? The answer
depends on what we take to be the ontological requirements for the
described stability of matter. Note that, on the face of it, we can only make
inferences about the actual world based on actual physics. If we could pro-
duce an argument to the effect that this stability is not possible without a
fundamental level, then we would have a good case at least for the exist-
ence of a fundamental level in the actual world – or, perhaps, any world
where macrophysical objects exist. Yet it seems that no amount of empiri-
cal research could establish that a fundamental level is metaphysically neces-
sary. So unless there is some a priori argument available to the effect that
the actual structure of reality is necessary, then we ought to draw infer-
ences only regarding the actual world. We do however have a fairly good
idea concerning the actual requirements for the forming of macrophysical
objects. This brings us back to the physical requirements regarding the
sequence of generic existential dependence (GXD), which were mentioned

in passing in the third section. One way to interpret these requirements

is exactly in terms of the type of constraints that have been listed above:
namely a list of requirements for the stability of matter, including physical
principles such as the Pauli Exclusion Principle and a limited fluctuation
of the fundamental physical constants. As the quotation from Barrow dem-
onstrates, macrophysical objects are extremely fragile and the range for

John D. Barrow, ‘Cosmology, Life, and the Anthropic Principle’, Annals of the New York
Academy of Sciences 950 (2001), p. 147.
This is not the only factor that we simplify for the purposes of illustration. For a more
rigorous account, see for instance Marc Lange, ‘Could the Laws of Nature Change?’,
Philosophy of Science 75.1 (2008), pp. 69–92.
6.5 Fundamentality and physics 147

the appropriate values of physical constants is narrow. Such precision, one

might think, would not be possible with full-blown infinity.
The philosophical idea here is similar to one expressed by John Heil,
who suggests that ruling out infinite complexity is a necessary constraint for
scientific theorizing; this leads him to deny actual infinities altogether.43
The approach outlined here is more liberal, as the generic sense of funda-
mentality that we defined in the previous section (GOF) allows for actual
infinities, provided that they are of the ‘boring’ sort. Since boring infinite
descent requires that there is no novelty in the structure after a certain
point, it appears that there is no need for infinite complexity either. Hence,
rather than infinitism per se, it may be that it is only infinite complexity that
we need to rule out. There is a worry though: if it cannot be ruled out a pri-
ori that the lower-level structure involves infinite complexity, it is not clear
whether anything more can be said. Certainly, the resulting ontological
picture is easier to swallow without infinite complexity, and this seems
to be what motivates Heil to abandon actual infinites as well, but the evi-
dence from physics does not settle the matter conclusively.
We could attempt to motivate the idea further. Take for instance the ele-
mentary charge, which was briefly discussed above. Quarks have charges
that are integer multiples of one third of the elementary charge, and if
they had internal structure then this structure would have to add up to the
appropriate charge as well. Now, if we continue this process indefinitely,
then eventually the charges of the constituent parts would approach zero.
Perhaps this oddity can be overcome, but how? At the very least, it is plausi-
ble that some as-yet-unknown forces that bind these near-chargeless parts
together would have to be postulated. Since the complex sequence goes on
infinitely, this would give rise to infinitely many fundamental forces. It is
the dream of physicists to unify all the fundamental forces, but this may
not be possible if there are infinitely many of them. Accordingly, it is not
surprising that many physicists find the idea of infinite descent unattrac-
tive. Here is a recent example:

In the last century, there have been repeated discoveries of underlying

structure. Moving from macroscopic objects, to atoms, to components
of the nuclei, to quarks, it has been demonstrated repeatedly that the
differences between supposedly fundamental particles are, in fact,

John Heil, The Universe As We Find It (Oxford University Press, 2012), pp. 38 ff.
148 Fundamentality and levels of reality

merely consequences of the composite structure of underlying reality.

It only seems a natural progression that such an approach of looking
for underlying structure be used to explain the particles of the standard
model. Attempts towards this end, dubbed preon models, met with many
obstacles, but still there was something deeper that presented itself as a

difficulty. The difficulty is that, as such a process does not have an end, we
can continue to suppose that below the currently understood structure
is another set of more fundamental particles. This idea quickly becomes
unappealing at a philosophical level, or even a practical level, as the
question then becomes ‘What could make it end?’44

Such speculation is certainly no more conclusive than the philosophical

discussion so far, but it is interesting to note that there are physicists whose
intuitions about the matter match those of many philosophers. However,
these intuitions have not gone unchallenged in physics either. For instance,
the physicist Howard Georgi, drawing on the issues surrounding renor-
malization in quantum field theory, suggests that effective quantum field
theories could ‘go down to arbitrarily short distances in a kind of infin-
ite regression’.45 However, Georgi adds that the mostprobable scenario is
one in which ‘at some very short distance the laws of relativistic quantum
mechanics break down and an effective quantum field theory is no longer
adequate to describe the physics’.46 He does not seem to have considered,
or else has rejected, the possibility that the infinite regress could be benign
because of being ‘boring’, as was discussed earlier. What we might take
from Georgi’s line of thought is that whatever happens when we approach
‘very short’ distances, it must be something that is able to support the laws
of relativistic quantum mechanics. With some effort, this may be consid-
ered to reflect the requirements that we associated with (GXD).
A further, speculative account that could perhaps accommodate infinit-
ism in physics comes from the Nobel Prize winner Hans Dehmelt. Dehmelt
proposes an intriguing quark/lepton substructure based on the model of
the triton – the nucleus of hydrogen’s radioactive isotope tritium (3H):

Jonathan Hackett, ‘Locality and Translations in Braided Ribbon Networks,’Classical and
Quantum Gravity 24 (2007), p. 5757.
Howard Georgi, ‘Effective Quantum Field Theories,’ in Paul Davies (ed.),The New Physics
(Cambridge University Press, 1989), p. 456.
6.5 Fundamentality and physics 149

I propose to extend the triton substructure scheme to an infinite number

of layers. Below the four layers listed above [up to subquarks], they contain
higher order dN subquarks, with N = 5 → ∞. In each layer the particles are
not identical but resemble each other in the same way as quarks and
leptons do, with masses varying as much as a factor 10 8. In an infinite

regression to simpler particles of ever increasing mass, they asymptotically

approach Dirac point particles.47

Up to N = 3, the level of electrons, Dehmelt’s model is motivated by cur-

rent physics, but it is speculative from N = 4 onwards, where electron sub-
structure is postulated. Importantly, it appears that the particles in each
new layer – as N increases – will be held together by new, stronger and
shorter-range forces (compare with the strong nuclear force, which eas-
ily overpowers the electromagnetic repulsion in the nuclei of atoms). As
N approaches infinity, we must also postulate infinitely many, arbitrarily
stronger and shorter-range forces. While this may be relatively unproblem-
atic mathematically, the ontological costs may be problematic.
Notice also that on the face of it, the model would appear to involve
mereological levels all the way down, although Dehmelt does not discuss
it in these terms. However, it is open to interpretation what exactly hap-
pens at the point when we reach N = 5 → ∞ . One option would be to inter-
pret the mereological chain as terminating at this point and the further
chain being ‘boring’ (that is, repetitive) and satisfying (GXD) in the sense
discussed earlier.
What about the ontological cost of introducing new forces at each level?
The worry here is that if ‘boring’ infinite descent requires an infinite num-
ber of novel forces, its plausibility as a candidate for something that could
satisfy (GOF) may diminish. In fact, Dehmelt’s model would also appear
to require the introduction of new exchange particles at each level, which
is perhaps even more problematic for (GOF). Consider the strong nuclear
force again: it binds protons and neutrons together in the nucleus, over-
powering the electromagnetic repulsion between protons. Like all the fun-
damental forces, the strong nuclear force is an exchange force, that is, it
involves the exchange of one or more particles. At the level of protons and

Hans Dehmelt, ‘Triton,… Electron,… Cosmon,…: An Infinite Regression?’, Proceedings
of the National Academy of Sciences 86 (1989), p. 8618. See also Tahko, ‘Boring Infinite
Descent’ and Morganti, ‘Metaphysical Infinitism and the Regress of Being.’
150 Fundamentality and levels of reality

neutrons, the strong nuclear force relies on the exchange of mesons, but it
is merely a residual force of the strong force (that is, the colour force) that
binds together positively charged quarks. In the latter case, the relevant
exchange particles are gluons. A full examination of the strong force would
require delving into quantum chromodynamics, but all we are interested
in here is the apparent need for independent exchange particle(s) at each
level where the force is active (e.g., mesons and gluons).
The core of the issue is this: if any physical model with infinite des-
cent requires postulating infinitely many new forces and corresponding
exchange particles, can such a model ever be well-founded? Ultimately,
that depends on our understanding of a viable fundamental level given
the complications of contemporary physics – which is something that has
not been examined in much detail so far. So as should be obvious from the
preceding discussion, research on these issues is still very young, and also
rather challenging because of the limited empirical data. But one thing is
clear: the search for a fundamental level will no doubt remain a central
puzzle in metametaphysics and physics alike for the foreseeable future.
7 The epistemology of metaphysics:
a priori or a posteriori?

Throughout this book, we have encountered various issues that fall within
the purview of epistemology. We have seen that the substance, significance,
and progress of many a debate depends on the success of our attempts to
gain knowledge about the subject matter of that debate. In the two preced-
ing chapters, we have acquired various further tools to express more pre-
cisely what it is that we are trying to gain knowledge about, but these tools
are of limited help when it comes to the epistemic side. Indeed, the dis-
cussion has quickly turned to matters so abstract that it is difficult to even
evaluate the relative merits of different views. So it is high time to address
the epistemic issues that loom large over the metametaphysics literature.
These issues are not commonly considered to be a part of metametaphys-
ics per se, but rather just a part of first order metaphysics or epistemology.
The present author holds a different opinion: in many cases the epistemic
part of a debate in metametaphysics is so intimately connected with the
ontological part that it is simply impossible to distinguish between metam-
etaphysics ‘proper’ and epistemology. Hence, we cannot avoid delving into
epistemic issues when we pursue questions in metametaphysics; they are a
central part of the methodology of metaphysics.
One of the central questions here is whether metaphysical knowledge
is gained by a priori or a posteriori means, or by some combination thereof.
If a posteriori elements are involved, then the obvious question that follows
is to what extent are these elements due to science; that is, what is the
relationship between science and metaphysics? However, this question is
so broad and important that we will dedicate a chapter of its own to it,
namely Chapter 9. But the a priori vs. a posteriori distinction itself ought to
be discussed briefly. We will do so in the first section of this chapter.
In no other area is the epistemic issue more pressing than in questions
that deal with modal truths. We have already seen plenty of examples

152 The epistemology of metaphysics: a priori or a posteriori?

regarding modal truths in metametaphysics, such as the speculative dis-

cussion regarding the (physical or metaphysical) necessity of a fundamen-
tal level in Chapter 6. The second, third, and fourth sections of the present
chapter focus on examining the potential sources of this type of modal
knowledge; that is, we focus on modal epistemology. Of course, modal know-
ledge does not exhaust metaphysical knowledge, but it is a particularly
important and illuminating case and hence serves as an appropriate case
study. We can divide the accounts of the source of modal knowledge into
two rough categories: modal rationalism and modal empiricism. As the names
suggest, the first is a group of views that attempts to explain modal know-
ledge primarily in terms of a priori resources, whereas the latter empha-
sizes a posteriori resources. A further complication involves the relationship
between essence and modality, as one currently popular account takes
essence to be ontologically prior to – or reducible to – modality. The ques-
tion arises: does the ontological priority of essence entail epistemic pri-
ority as well? We will discuss this issue in the third section. As we will
see, distinguishing the a priori and the a posteriori also causes complica-
tions, for it is quite difficult to point out examples of ‘pure’ a priori or
a posteriori knowledge. In fact, much of the fourth section of this chapter
will be spent in trying to clarify what the status of ‘armchair’ methods in
metaphysics is.
The fifth section takes stock of the preceding debate between modal
rationalism and modal empiricism. In the course of our discussion, we will
see that there is strong pressure towards a type of hybrid view: if the dis-
tinction between a priori and a posteriori knowledge is not as straightfor-
ward as is often assumed, then we’ll have to consider alternative ways of
formulating the debate. On the positive side, this may prove to be a meth-
odologically interesting development, especially if it allows us to make pro-
gress with regard to the primary issue as well, namely modal epistemology.

7.1 A priori vs. a posteriori

A priori and a posteriori knowledge are typically defined as knowledge

acquired by non-experiential and experiential means, respectively. This
distinction is known to be problematic, but since philosophers will no
doubt continue to use it (and we will do so as well), we ought to do our best
to clarify it. Consider two rather obvious problems, discussed for instance
7.1 A priori vs. a posteriori 153

by Laurence BonJour: the problem of how to define ‘experience’ and how

the a priori is supposed to be ‘independent’ of it.1 In the first case the prob-
lem is how to determine the correct scope of experience. Do mental pro-
cesses count as experience? How about mathematical or philosophical
reasoning that relies on certain learned patterns? Should only perceptual
information count as experiential? How does memory fit in with all of this?
The second problem involves issues concerning concept acquisition as a
precondition for a priori knowledge, but also independence of experience,
namely whether a proposition is indefeasible by experiential information.2
The basic underlying problem here is the following. On the one hand, if
we are too strict in how we define independence of experience, then noth-
ing whatsoever will count as a priori knowledge and hence the distinction
breaks down. On the other hand, if we are more liberal about what counts
as non-experiential, then the distinction becomes vague and we might
have cases where it is equally correct to say that some knowledge is a priori
or that it is a posteriori. However, as we will see in what follows, the labels
of ‘a priori’ and ‘a posteriori’ are currently used in such a broad fashion that
we are effectively already in a situation where the distinction has become
rather vague. This is not a reason for giving the distinction up altogether
though, for we can still identify certain characteristics that point towards
apriority or aposteriority and hence, at least arguably, we have some use
for the distinction.
Note that besides knowledge, we talk about justification and reasoning .
There is an extensive literature on a priori justification, which we will not
engage with here.3 Justification concerns the process of how a person’s
beliefs are justified. The same belief could be justified by a posteriori means
for one person and by a priori means for another. Moreover, a belief might
not be true even if it is justified. But not all issues concerning justification
are directly relevant for us here, as we are primarily interested in how

Laurence BonJour, In Defense of Pure Reason (Cambridge University Press, 1998),
pp. 7–11.
For further discussion regarding these and other problems, see Tuomas E. ahko, T ‘A
New Definition of A Priori Knowledge: In Search of a Modal Basis,’Metaphysica 9.2 (2008),
pp. 57–68, and Tuomas E. Tahko, A‘ Priori and A Posteriori: A Bootstrapping Relationship,’
Metaphysica 12.2 (2011), pp. 151–164.
For one influential account, see Albert Casullo, A Priori Justification (Oxford University
Press, 2003).
154 The epistemology of metaphysics: a priori or a posteriori?

it is possible to distinguish between apriority and aposteriority in the

abstract. Regarding reasoning, one might ask whether it makes sense to
talk about anything except a priori reasoning, since reasoning is presuma-
bly a (deductive) mental process. Hence, if one arrives at a belief purely by
reasoning, then the belief would appear to be justified by a priori means.
But perhaps it is not altogether impossible to conceive of something
called ‘ a posteriori reasoning’: generalizing from empirical observations
by inductive inference would seem to be a form of ‘reasoning’ distinct
from the deductive process we typically refer to by ‘reasoning’. So empiri-
cal generalization of this type could perhaps be called justification by ‘ a
posteriori reasoning’. In any event, these are not issues that we have to set-
tle here, but the reader is urged to keep in mind that there are different
ways to approach the topic.
Given that the distinction between a priori and a posteriori knowledge /
justification / reasoning / inquiry / methods, and so on does not appear
to be particularly sharp, we will resort to stipulation. Specifically, we
will stipulate that inquiry or justification involving deductive reasoning is a
characteristic of apriority, whereas the involvement of empirical elements is,
naturally enough, a characteristic of aposteriority. Sometimes it is not
perfectly clear whether some process of inquiry does involve the men-
tioned characteristics, but this is a question of a different level. What is
more important is that very often, if not always, a process of inquiry will
involve both characteristics. Typically, when any characteristics of aposte-
riority are present, philosophers are inclined to call the resulting piece
of knowledge a posteriori (e.g., the case of a posteriori necessities). But we
have already noted that this is likely to be too strong a condition, for abso-
lutely nothing might be totally ‘unpolluted’ by a posteriori characteristics,
and hence the a priori vs. a posteriori distinction would simply break down
altogether. Some might welcome this result, but it should be a last resort,
given that the notion of apriority features heavily in metaphysics. Thus,
in such cases it may be better to conclude that a priori and a posteriori ele-
ments are intertwined to such an extent that we simply can’t classify such
a piece of knowledge one way or another. Yet, this does not mean that we
couldn’t talk about the different characteristics of apriority and aposteri-
ority that are present in a given case. If one aspect dominates, say, a given
process of justification, then we might say that the resulting belief was
justified primarily by a priori means. But since we have acknowledged that
7.2 Modal rationalism and a priori methods 155

the distinction is unlikely to be sharp, then not that much hangs on this
issue in the first place.
In sum, the reader should be aware that the approach taken in this
book holds there to be a place for the notion of apriority, even if it is
sometimes very difficult to determine whether a process or a proposition
is a posteriori, a priori , or both. Given this, some mention should be made of
the possibility of a more deflationary approach: one might think that no
useful purpose is being served by the a priori vs. a posteriori distinction. For
instance, Timothy Williamson has recently argued that since experience
plays a role that is more than purely enabling but less than strictly evidential
both in clear cases of a priori knowledge (or justification) and in clear cases
of a posteriori knowledge, the distinction really does not have much theo-
retical significance.4 Often, it is the notion of apriority which comes out as
the underdog when such challenges are posed, probably because empiri-
cal elements have such a strong standing in contemporary philosophy.
Many of these challenges derive from logical empiricism, Quine’s denial
of the analytic/synthetic distinction, or more recent radical empiricism. It
is not clear that these challenges capture the subtlety of the intertwined
conception of apriority and aposteriority that we have stipulated above.
But there have been many recent challenges concerning the coherence
or significance of the notion of apriority, so they should be taken seri-
ously. Albert Casullo has recently made an effort to address some of these
challenges, although he is also concerned about the problems regarding
the experiential/non-experiential distinction.5 However, discussing these
issues in more detail here would take us too far from our primary topic,
even though ultimately any account of the epistemology of metaphys-
ics that resorts to these notions will have to face the difficult questions
regarding the a priori vs. a posteriori distinction that we have outlined here.

7.2 Modal rationalism and a priori methods

Let’s start by distinguishing two approaches:

Timothy Williamson, ‘How Deep is the Distinction Between A Priori and A Posteriori
Knowledge?’ in A. Casullo and J. C. Thurow (eds.), The A Priori in Philosophy (Oxford
University Press, 2013).
Albert Casullo, ‘Four Challenges to the A Priori–A Posteriori Distinction,’ Synthese (2013
[online], forthcoming).
156 The epistemology of metaphysics: a priori or a posteriori?

Permissive modal rationalism (PMR) suggests that our epistemic access

to modality generally involves rationalist, a priori elements.

Strict modal rationalism (SMR) suggests that modal epistemology is

entirely a matter of a priori reasoning.

Modal rationalism encompasses intuition- and conceivability-based
approaches of the type defended, for instance, by George Bealer 7 and
David Chalmers8, and also the essence-based account of E. J. Lowe.9
Bealer defends an intuition-based account, whereas Chalmers defends
a conceivability-based account as a part of a broader rationalist picture.
Lowe’s approach is slightly different, since he thinks, following Kit Fine,
that metaphysical modality is grounded in essence – we will return to this idea
in a moment.10 An upshot of this view is that the epistemology of modal-
ity becomes the epistemology of essence, as discussed in more detail in
the third section of this chapter. According to Lowe, the epistemology of
essence is an a priori process of understanding what an entity is (or would be).
In fact, Bealer also talks about ‘understanding’, but his usage differs from
Lowe’s: Bealer focuses on the possession of concepts whereas Lowe seems
to have a more general type of understanding in mind, albeit he does not
specify it very carefully.11 The literature on modal epistemology, however,
is enormous, and these are just a few prominent examples. In fact, because
of the size of the literature, it would require a series of books to fully
engage with the extant views. Thus, we will instead just take a small but
hopefully illuminating peek into the debate, with special attention to the

6 For an excellent selection of articles dealing with conceivability in particular, see

T. S. Gendler and J. Hawthorne (eds.),Conceivability and Possibility (Oxford University
Press, 2002).
See for instance George Bealer, ‘Modal Epistemology and the Rationalist Renaissance,’
in Gendler and Hawthorne (eds.), Conceivability and Possibility, pp. 71–125; and his ‘The
Origins of Modal Error,’Dialectica 58.1 (2004), pp. 11–42.
8 David Chalmers, ‘Does Conceivability Entail Possibility?’, in Gendler, and Hawthorne
(eds.), Conceivability and Possibility, pp. 145–200.
E. J. Lowe, ‘Whatis the Source ofour Knowledge of Modal Truths?’,Mind 121 (2012), pp.
Kit Fine, ‘Essence and Modality,’ Philosophical Perspectives 8 (1994), pp. 1–16.
For a more systematic theory of understanding and essence, see Anand Vaidya,
‘Understanding and Essence,’ Philosophia 38 (2010), pp. 811–833. Vaidya’s account, how-
ever, may not be compatible with the type of (realist) essentialism that Lowe hopes to
defend, so it represents a somewhat different approach to the topic.
7.2 Modal rationalism and a priori methods 157

methodological consequences.12 We’ll start from the conceivability-based

account and continue with the essence-based account, taking into account
how they relate to (PMR) and (SMR).
The idea that conceivability might be a guide to possibility has a long
history, but we will not dwell on its Cartesian roots here. Rather, we will
focus on one particular issue concerning the contemporary usage of con-
ceivability. But first, the claim obviously needs qualification: what kind of
conceivability and what kind of possibility are we talking about? Typically,
we are interested in whether conceivability might entail metaphysical possi-
bility, whereas it is taken as granted that conceivability entails at least con-
ceptual possibility. In other words, if you can conceive of a scenario x, then x
must at least be a scenario that is not conceptually contradictory. Hence, in
its simplest form, conceivability refers to our ability to coherently imagine
or conceive the scenario in question. Note, though, that strictly speaking we
should perhaps distinguish conceivability and imaginability, even though
most authors use them interchangeably.13 There are various reasons for
this, one of them being the fact that ‘imaginability’ typically has a very
strong connection to our ability to visualize something, so it is quite tightly
associated with just one sensory faculty. Conceivability, however, is typic-
ally understood in a broader fashion, not so tightly connected with vision.
Sometimes further requirements are added as well, for instance that we
should be able to imagine or conceive the relevant scenario as if it were
actual, that is, to imagine an alternative history or a metaphysically pos-
sible world that could have been actual. The theory of two-dimensional modal
semantics develops this line of thought, but this is a topic that we will omit
here, since a detailed presentation would take us quite far from our main
Given our restricted focus, we do not need to discuss all the intricacies
of the various modal rationalist accounts. Rather, we might concentrate
on those particularly problematic cases where seemingly metaphysically

For a selection of state-of-the-art articles on the topic, see R. W. Fischer and F. Leon
(eds.), Modal Epistemology After Rationalism, Synthese Library (Dordrecht: Springer,
For an argument to the effect that conceivability and imaginability should be dis-
tinguished, see Marcello Oreste Fiocco, ‘Conceivability, Imagination, and Modal
Knowledge,’ Philosophy and Phenomenological Research, 74.2 (2007), pp. 364–380.
For an overview of two-dimensional modal semantics, see David Chalmers, ‘The
Foundations of Two-Dimensional Semantics,’ in M. Garcia-Carpintero and J. Macia
158 The epistemology of metaphysics: a priori or a posteriori?

necessary truths such as ‘Gold is the element with atomic number 79’ are
to be reconciled with conceivable yet metaphysically impossible scenarios
such as ‘Gold might have turned out not to be the element with atomic
number 79’. One reaction to this issue is to say that it is always conceiv-
able that things might have been otherwise, while another would be to
insist that conceivability is restricted by the current a posteriori framework.
A proponent of (SMR) might favour the first option, whereas (PMR) is more
naturally combined with the restricted use. Plausibly, we can distinguish
between different types of conceivability here, as Chalmers and Stephen
Yablo, among others, have done.15 However, no generally accepted conven-
tion about the use of conceivability in such cases exists despite the vast lit-
erature, and in fact it may even be possible to be a strong modal rationalist
and accept that conceivability is always restricted – this would seem to be
a coherent view if it is emphasized that we cannot simply step out of our
current epistemic situation and so must always reason within certain ‘a
posteriori bounds’. To put the distinction between the two possible reactions
more precisely, the first issue is whether we should fix conceivability in
terms of how the world might be before we have any a posteriori knowledge,
limited only by a priori consideration, or how the world might be given the
a posteriori framework. Here we can immediately see that the relationship
between a priori and a posteriori elements comes into question.
On the basis of the above discussion, a further possibility of a dismissive
reaction emerges: if there is any sense in which a posteriori elements are
required to get the modal rationalist view off the ground then it would not
seem to be rationalism as it is traditionally understood. This type of dismis-
sive reaction would lead to either (SMR) or a flat denial of modal ration-
alism. However, this may not be an entirely fair reaction, for we should

(eds.), Two-Dimensional Semantics: Foundations and Applications (Oxford University Press,

2006), pp. 55–140.
Chalmers, ‘Does Conceivability Entail Possibility?’, and Stephen Yablo, ‘Is Conceivability
a Guide to Possibility?’, Philosophy and Phenomenological Research, 53 (1993), pp. 1–42.
Chalmers thinks that there might be up to eight types of conceivability; it is his dis-
tinction between primary and secondary conceivability that reflects the issue at hand: S
is primarily conceivable when it i s conceivable that S is actually the case, and it is sec-
ondarily conceivable when S conceivably might have been the case (p. 157). We will set
aside the complications regarding ideal conceivability, which Chalmers also discusses,
i.e., the case of an ideal conceiver with perfect rational capabilities.
7.2 Modal rationalism and a priori methods 159

presumably allow for at least some a posteriori elements without having to

conclude that the epistemic framework we are dealing with is empiricism.
For instance, we clearly learn about gold – that gold is the element with
atomic number 79 – through empirical means, but even once we know
this, the necessity of the proposition ‘Gold is the element with atomic num-
ber 79’ may only be derived by a priori means, as the modal rationalist
would have it. Similarly, although we learn about the concept ‘element’ via
our senses, a posteriori, this type of concept acquisition may be considered a
part of our background knowledge rather than speaking against (SMR). So
a charitable reading would seem to be that both (PMR) and (SMR) are valid
candidates and the question will hang on something other than whether
there is any empirical exposure whatsoever.
But if we understand conceivability in this charitable, more restricted
sense, taking into account the a posteriori framework, then it seems diffi-
cult to reconcile it with conceivable yet metaphysically impossible scenarios.
Given our knowledge about, say, gold, namely that it is the element with
atomic number 79, it ought to be impossible even to imagine gold not being
an element with that atomic number without making radical changes in
the a posteriori framework. So this might push us back towards understand-
ing conceivability in the unrestricted, a priori sense, as then it appears that
conceivability is a necessary requirement for (knowledge of) possibility.
This is because conceivability could then be interpreted simply as an initial
requirement for a lack of contradictions in the imagined possibility. Yet,
while this may save a sense of modal rationalism, it does appear to be a
weaker thesis than we were initially looking for.
Perhaps, in addition to the requirement of a lack of contradictions in
conceivable scenarios, we could give a corresponding negative require-
ment for conceivability? This could be something like the following: every-
thing not ruled out by a priori reasoning is conceivable. On the face of it,
this does seem to help at least a little, as plausibly a priori considerations
rule out contradictory propositions, so this qualification for conceivability
encompasses the previous requirement as well. The focus of the question
then moves to the epistemic: if conceivability is a prima facie guide to pos-
sibility in this sense, could it be a useful tool in modal epistemology? Of
course, given that there is an overlap between what is conceivable and
what is metaphysically possible, but also between what is conceivable and
what is metaphysically impossible, the previous overlap may not entail a
160 The epistemology of metaphysics: a priori or a posteriori?

reliable epistemic link between conceivability and metaphysical possibility.

In fact, there are infinitely many conceivable metaphysically possible sce-
narios (albeit that in the case of non-ideal conceivers such as us mere mor-
tals, this is subject to limitations in cognitive capacity). But there are also
infinitely many conceivable metaphysically impossible scenarios! To see this,
take any metaphysically necessary truth, such as ’Gold is the element with
atomic number 79’, and start generating metaphysically impossible sce-
narios where gold is the element with atomic number 80, or 81, or 82, or …
If we hope to use conceivability as a reliable guide to metaphysical
possibility, we should somehow be able to distinguish between meta-
physically possible and metaphysically impossible scenarios and hence
restrict our focus somehow. One of the most influential attempts to
do this is due to Williamson, who discusses conceivability as a part of
his counterfactual-based account of modal epistemology. 16 Note that
Williamson attempts to explain our knowledge of modality in terms of
our knowledge of counterfactuals and wishes to stay neutral about the epi-
stemic status of our modal judgements, so we are not suggesting that his
view is necessarily a version of modal rationalism. Nevertheless, he does
have some interesting things to say about the issue at hand. Figure 7.1 illus-
trates what the problem is:


Conceivable and

Conceivable and

Figure 7.1

Timothy Williamson, The Philosophy of Philosophy, (2007, Oxford: Blackwell Publishing),
especially Ch. 5.
7.2 Modal rationalism and a priori methods 161

Everything outside the circle is inconceivable and everything inside it is

conceivable – this is a relatively unproblematic distinction. But the prob-
lematic distinction is inside the circle, between what is conceivable but
metaphysically impossible and what is conceivable and metaphysically
possible. How are we able to divide the circle?
On Williamson’s line, the key is that we need to take advantage of our
background knowledge. In other words, Williamson thinks that our ability
to conceive of counterfactual scenarios, such as gold failing to be the ele-
ment with atomic number 79, must be restricted by at least some empirical

If we know enough chemistry, our counterfactual development of the

supposition that gold is the element with atomic number 79 will generate
a contradiction. The reason is not simply that we know that gold is the
element with atomic number 79, for we can and must vary some items
of our knowledge under counterfactual suppositions. Rather, part of
the general way we develop counterfactual suppositions is to hold such
constitutive facts fixed.17

This would already seem to concede that modal rationalism needs to

be supplemented with some empirical knowledge. But if most of the work
in modal epistemology is done with the help of our ability to entertain
counterfactual suppositions – that is, our ability to conceive of alterna-
tive scenarios – then perhaps this would still be acceptable to some modal
rationalists. Williamson at least provides an answer to the problem at
hand, as we can dismiss the conceivable but metaphysically impossible
scenario of gold having some other atomic number than 79 with the help
of ‘constitutive facts’ – background knowledge that we do not vary when
considering counterfactual scenarios. The manner in which the method
reveals that we must hold constitutive facts fixed is based on the idea that
any counterfactual supposition which fails to hold such a fact fixed will
generate a contradiction. This enables us to distinguish between conceiv-
able metaphysical impossibilities and metaphysical possibilities.
Promising as it is, one obvious problem with this proposal is that we
should somehow be able to know which facts are constitutive. Cases such
as ‘Gold is the element with atomic number 79’ may be rather good can-
didates for this analysis, as there are no obvious alternatives for what gold

Williamson, The Philosophy of Philosophy, p. 164.
162 The epistemology of metaphysics: a priori or a posteriori?

could be, but unless we have a more general story about the class of such
constitutive facts, then we have not yet discovered a reliable method. In
fact, Williamson does not even attempt to come up with such a story, as
he believes that we do not need to know which facts are constitutive; it suf-
fices that we know some constitutive facts.18 To see that it is important for
Williamson’s method that we hold the correct facts fixed, consider what
would happen if we held some non-constitutive facts fixed. What would hap-
pen is that we would end up ruling out (as metaphysically impossible) cer-
tain possibilities that we wish to include: say, if we held it fixed that Earth
is the third planet from the Sun, we would end up ruling out the possibility
that the planet Venus never formed, which is surely possible! Similarly, if
we fail to hold some constitutive fact as fixed, say that elements are defined
by their atomic number (if that is indeed the case), then we would errone-
ously include metaphysical impossibilities, such as gold failing to be the
element with atomic number 79. So it does seem that something more
needs to be said about constitutive facts.
However, attempts to come up with an analysis of constitutive facts have
been made; one candidate, not surprisingly, is that constitutive facts of the
relevant type concern essences. If Williamson’s account were to be supple-
mented with a satisfactory analysis of constitutive facts in terms of essences,
then we would seem to have a fairly strong case for modal rationalism.
But note that we would also lose part of Williamson’s srcinal motivation,
namely the possibility of a uniform account– an account that explains our
knowledge of modal truths in terms of a single rational faculty, namely the
ability to handle counterfactuals. So if essences are needed to undergird the
account, then why not go all out for an essence-based approach?
According to the essence-based approach, which was popularized by Fine,
essentialist facts can be understood as non-modal constitutive facts (which
entail modal facts).19 Now, if we have some prior epistemic access to these
constitutive, essentialist facts, then we can address the previous concern. For
instance, if we consider atomic number to be essential for elements, then
the atomic number of gold being 79 is a constitutive fact that we know inde-
pendently of Williamson’s counterfactual analysis, hence enabling adistinct

For further discussion, see Sonia Roca-Royes, ‘Conceivability and De Re Modal
Knowledge,’ Noûs 45.1 (2011), pp. 22–49, and Tuomas E. Tahko, ‘Counterfactuals and
Modal Epistemology,’Grazer Philosophische Studien 86 (2012), pp. 93–115.
Fine, ‘Essence andModality.’
7.3 The epistemology of essence 163

source of constitutive facts that could serveto restrict conceivability as well.

Yet it should be immediately noted that this may only push the problem one
iteration further, for now we ought to present an account of the epistemology
of essence.

7.3 The epistemology of essence

What we are now interested in is the process of coming to know modal facts
via our knowledge of essentialist facts. Fine says less about the epistemic
dimension than the ontological one, but his account of the ontological rela-
tionship between essence and modality has been hugely influential. On the
epistemic side, Lowe has done much of the pioneering work. Both Lowe’s
and Fine’s understandings of essence follow a ‘neo-Aristotelian’ line whereby
the essence of an entity is its
real definition. This idea is not easily elaborated on,
but one way to understand real definitions is to take them to be expressed by
propositions which tell us what a given entity is orwould be – we can also state
the real definitions of things that are non-existent.Lowe considers essence
to be prior to existence both ontologically and epistemically. In Lowe’s own
words: ‘instead of trying to explicate the notion of essence in terms of that
of modality, as on the Kripkean account of essence, the very reverse needs
to be done’.22 Note that unless we assume that essence can be understood in
non-modal terms, the area of the epistemology of essence would seem to col-
lapse into modal epistemology proper, as the essentialist truths would simply
be a proper subset of the modal truths. But if essence can be understood in
non-modal terms, then the question about our epistemic access to essential-
ist truths is very pressing indeed. Note further that Lowe holds that essences
cannot be entities in themselves, on pain of infinite regress. For if every entity
had an essence (as Lowe also holds) and essences were entities, then essences
themselves would have to have essences as well, and so on
ad infinitum.

20 For further discussion, see Tuomas E. Tahko, ‘Empirically-Informed Modal Rationalism,’

in Fischer and Leon (eds.), Modal Epistemology After Rationalism, Synthese Library
(Dordrecht: Springer, forthcoming).
See Lowe, ‘What is the Source ofour Knowledge of Modal Truths?’, p.935.
E. J. Lowe, ‘Essence vs. Intuition: An Unequal Contest,’ in A. R. Booth and
D. P. Rowbottom (eds.), Intuitions (Oxford University Press, 2014), p. 264. See also
E. J. Lowe, ‘Two Notions of Being: Entity and Essence,’ Royal Institute of Philosophy
Supplements 83.62 (2008), pp. 23–48. In the Kripkean account of essence, essential prop-
erties are defined simply as necessary properties.
164 The epistemology of metaphysics: a priori or a posteriori?

How do we come to know the propositions that express essences – the

real definitions – according to Lowe? The core of Lowe’s suggestion is that
to grasp an entity’s essence is simply to understand what the entity is:

To know something’s essence is not to be acquainted with some further

thing of a special kind, but simply to understand what exactly that
thing is. This, indeed, is why knowledge of essence is possible, for it is a
product simply of understanding – not of empirical observation, much
less of some mysterious kind of quasiperceptual acquaintance with
esoteric entities of any sort. And, on pain of incoherence, we cannot deny
that we understand what at least some things are, and thereby know
their essences. 23

Even though this passage may not convince everyone about the possibil-
ity of our knowledge of essences, it does at least make clear what Lowe’s
take on the issue is, namely that the epistemology of essence is a sim-
ple a priori process of understanding what an entity is (or would be). On

Lowe’s view, this is the basis of all modal knowledge, including a posteriori
To see how essence and metaphysical modality come apart on this view,
consider the difference between two geometrical examples, as discussed
by Lowe:

(E1) An ellipse is the locus of a point moving continuously in a plane in

such a fashion that the sum of the distances between it and two other fixed
points remains constant
(E2) An ellipse is the closed curve of intersection between a cone and a
plane cutting it at an oblique angle to its axis greater than that of the
cone’s side.24

Lowe suggests that (E1) gives us the generating principle of ellipses – and
hence their essence – whereas (E2) states merely a necessary property
of ellipses. We can understand properties such as (E2) once we know
the generating principle of ellipses, but Lowe insists that (E1) is not
contained in (E2); we need an understanding of the essences of cones
as well as ellipses before we can produce (E2). This is how essence and

Lowe, ‘Two Notions of Being: Entity and Essence,’p. 39.
Lowe, ‘What is the Source of ourKnowledge of Modal Truths?’, p.936.
7.3 The epistemology of essence 165

metaphysical necessity come apart: we may state numerous necessary

properties that ellipses have, but only something like (E1) will give us
their essence. It would be tempting to conclude that all metaphysically
necessary truths are true in virtue of the essence of something or other,
but Lowe suggests instead that ‘any essential truth is ipso facto a metaphysi-
cally necessary truth, although not vice versa: there can be metaphysically
necessary truths that are not essential truths’. 25 However, the view needs
to be qualified further, as this is not to say that (E2) would not also hold
in virtue of the essence of something – it just does not hold strictly in vir-
tue of the essence of ellipses. Lowe thinks that (E2) holds in virtue of the
essences of ellipses and cones together . So following this line of thought,
all metaphysical necessities are grounded in essence, even though a given
metaphysically necessary truth may not be an essential truth about any
one thing.
The structure of the epistemic process, then, is as follows:

(1) We know that the essence of x is expressed by proposition p (which

states the real definition of x).
(2) We know that if p expresses the essence of x, then p is metaphysically
(3) We know that p is metaphysically necessary.

(2) is trivially true on the Finean analysis of essence, so it is (1) that does the
real work here. To put all this to together, consider the following example,
which features one of the tools we also learned about in Chapter 5, namely
essential dependence:

Consider the following thing, for instance: the set of planets whose orbits
lie within that of Jupiter. What kind of thing is that? Well, of course, it is a
set, and as such an abstract entity that depends essentially for its existence
and identity on the things that are its members – namely Mercury, Venus,
Earth, and Mars. Part of what it is to be a set is to be something that
depends in these ways upon certain other things – the things that are its
members. Someone who did not grasp that fact would not understand
what a set is.26

Ibid., p. 938.
Lowe, ‘Two Notions of Being: Entity and Essence,’p. 37.
166 The epistemology of metaphysics: a priori or a posteriori?

So Lowe thinks that in many cases knowing how a thing is related to

other things is central to our knowledge of what a thing is. But note
that Lowe thinks that for the set to depend essentially for its existence
and identity on its members is only a part of what it is to be a set. In
general, knowing the complete essence of a thing may not be a very sim-
ple affair – perhaps it is almost impossible – but knowing a part of a
thing’s essence is often sufficient for talking or thinking about it com-
prehendingly, and for being able to distinguish it from other things.
This is what seems to be the basis of Lowe’s idea of understanding .
While Lowe’s line of thought seems to hold some promise, we have
hardly been provided with a full epistemology of essence so far. One
complication emerges immediately when we try to apply this account
to concrete objects: just how much of a thing’s essence do we need to
grasp – to understand – in order to have a sufficient picture of its existence
and identity conditions? In particular, is it plausible that this will be an a
priori process rather than a primarily empirical one in the case of things
such as gold and other natural kinds? Lowe’s picture remains incom-
plete in this regard. Because of these difficulties, a more straightforward
solution may be simply to abandon strict modal rationalism (SMR), as it
was defined in the previous section. If all or part of our modal knowl-
edge is acquired by some less powerful means than the conceivability-
and essence-based accounts would have it, then perhaps the risk of
over-generating possibilities and hence running the risk of including
metaphysically impossible scenarios can be avoided altogether – this
might still be compatible with some versions of permissive modal ration-
alism. Moreover, once empirical information is allowed in the picture,
it may be easier to account for modal facts concerning concrete objects.
In any case, there is an obvious motivation to examine alternative epis-
temic strategies in more detail; the recent interest in ‘modal empiricism’
is symptomatic of the difficulties that we’ve encountered with modal
rationalism. 27

Note however that attempts to develop different versions of the essence-based account
(although not necessarily modal rationalist in the strong sense) continue. For a very
recent attempt, see Bob Hale, Necessary Beings: An Essay on Ontology, Modality, and the
Relations Between Them (Oxford University Press, 2013), Ch. 11.
7.4 Modal empiricism and the status of armchair methods 167

7.4 Modal empiricism and the status

of armchair methods

The term ‘modal empiricism’ is not yet established in the literature, but
since we are looking into an approach that contrasts with modal rational-

ism, calling it modal empiricism would seem to be a natural28choice; the

notion has previously been used for instance by Carrie Jenkins. Roughly
speaking, modal empiricism encompasses views that suggest our modal
knowledge derives primarily from experience, or, to put it somewhat dif-
ferently, that experience is what ensures the reliability of our modal knowl-
edge. The latter is close to what Jenkins suggests:experience provides an
epistemic grounding for our concepts, the basis of our conceptual abili-
ties. Hence, somewhat surprisingly, the reliability of conceivability itself
is based on experiential knowledge! Williamson’s account, one aspect of
which we discussed above, has a similar starting point. It starts from the
evolutionary basis of our ability to entertain counterfactual suppositions

and proceeds to assess the reliability of conceivability on this basis. More

precisely, Williamson reduces our capacity for modal knowledge to our
capacity for assessing counterfactual conditionals, which he claims to have
an evolutionary basis; it has been extremelyuseful for us as a species to be
able to entertain counterfactual suppositions (and to do so successfully).
For instance, if a tiger threatens you, your survival may depend on your
ability to predict the tiger’s behaviour on the basis of itspossible behaviour.
This leads Williamson to conclude that the worries regarding conceivability
and especially the over-generation of possibilities that we discussed in the
previous section fail to acknowledge the evident successes of conceivability
in more mundane scenarios: ‘Once we recall its fallible but vital role in
evaluating counterfactual conditionals, we should be more open to the idea
that it plays such a role in evaluating claims of possibility and necessity’.
Thus, it seems that Jenkins’s and Williamson’s accounts– and many others
as well – have elements both of modal rationalism and modal empiricism.
One reason for this peculiar situation – the fact that many recent
accounts attempt to develop an empiricist basis for conceivability or other

C. S. Jenkins, ‘Concepts, Experience and Modal Knowledge,’ Philosophical Perspectives 24
(2010), pp. 255–279.
Williamson, The Philosophy of Philosophy, p. 163.
168 The epistemology of metaphysics: a priori or a posteriori?

rational capabilities – is that, historically and still today, there is an excep-

tionally strong consensus that experience can only be a guide to what is
actual. So it seems only natural that experience has to be supplemented
with something if it is to give us access to matters relating not only to the
actual, but also to the merely possible or the necessary. This situation also
tells us something important about the methodological issues surrounding
modal epistemology. Initially, we hope to explain a given phenomenon,
such as our apparent ability to grasp modal truths, by resorting to a single
explanation – we attempt to develop a uniform account. But if it turns
out that such an account faces insurmountable difficulties, as strict modal
rationalism perhaps does, the first reaction is not to drop uniformity but
to attempt a reduction of the srcinal account. This is not to suggest that
the accounts we have mentioned in this section, namely Jenkins’s and
Williamson’s, have necessarily developed in this manner. The debate con-
cerning modal epistemology as a whole, in any case, certainly shows signs
of such a development, for attempts to reduce our ability to conceive and
imagine into other, supposedly less mysterious sources of knowledge are
numerous. The driving idea is that, other things being equal, a uniform
account of modal epistemology is preferable to a fragmented account.
We arrive at a somewhat problematic result: we’ve attempted to dis-
tinguish between modal rationalism and modal empiricism roughly in
terms of the a priori vs. a posteriori distinction. But in both cases the situa-
tion is certainly not as simple as the a priori vs. a posteriori distinction (sup-
posedly) is. An additional caveat, which we have not yet discussed, is the
fact that if we were to defend a uniform account of modal epistemology,
then all our modal knowledge would have to be explained by resorting to
just one of these areas of knowledge. On the face it this seems implausi-
ble, because at the very least our knowledge of modal facts concerning
abstract objects would appear to require some knowledge that could plau-
sibly be labelled ‘ a priori’, whereas knowledge about modal facts concern-
ing concrete objects plausibly requires empirical knowledge. But perhaps
some alternative understanding of the a priori vs. a posteriori distinction
would enable us to make progress? For instance, Jenkins suggests that
although empirical knowledge is crucial in our efforts to gain knowledge
of the mind-independent world, the crucial elements of that knowledge
retain ‘many characteristic features of the a priori: it is knowledge of con-
ceptual truths through conceptual examination alone; no tests need to be
7.4 Modal empiricism and the status of armchair methods 169

conducted beyond the examination of one’s concepts, so it can be secured

in the armchair (by people whose concepts have previously been suitably
grounded) rather than the laboratory’.30 So Jenkins, like Williamson, is
inclined to play down the a priori vs. a posteriori distinction, as it appears
that in many cases the process of inquiry that leads to modal knowledge
involves elements that could readily be called either a posteriori or a priori.
This is certainly a valuable methodological lesson.
Before we examine the consequences of this lesson, let us pause for a
moment, since here we seem to be at the core of an important methodo-
logical issue. Here’s one way to put the puzzle. We philosophers primarily
engage in ‘armchair’ philosophy, just as Jenkins suggests above. 31 We read
philosophical texts and arguments, think about them in our armchairs,
and report our findings in journals and at conferences. So, on the face of
it, our philosophical work seems to lean strongly towards the a priori. Yet,
at the same time we believe that we can make progress in philosophy and
make interesting (and true!) observations about the world . These include
but are of course not limited to the judgements we make about modality.
Perhaps particularly troubling in this regard are our judgements about
natural kinds; for instance, the famous judgement that gold is, neces-
sarily, the element with atomic number 79, or that water is necessarily
H2O. These judgements are philosophically of the utmost importance, but
philosophers do not typically consult chemists or other scientists when
making these judgements, regardless of their empirical content. We do
all this without conducting experiments in the laboratory, and often even
without any in-depth knowledge about such experiments.
There are several ways that one might react to this puzzle. They include
a sceptical reaction, which we will set aside here (cf. the discussion on con-
ventionalism in Chapter 4), a type of reliabilism regarding our a priori tools
(perhaps along the lines of Lowe’s work discussed briefly above, i.e., that
it’s generally quite unproblematic and easy to grasp essences a priori), and
conceptualism, which emphasizes conceptual analysis as the primary area
of philosophical work (we can see elements of this in Jenkins’s account of
modal epistemology). A further reaction, proposed by Daniel Nolan, is to

Jenkins, ‘Concepts, Experience and Modal Knowledge,’ p. 266.
This puzzle and some of the discussion that follows is adapted from Daniel Nolan, ‘The
A Posteriori Armchair,’ Australasian Journal of Philosophy 93.2 (2015), pp. 211–231.
170 The epistemology of metaphysics: a priori or a posteriori?

look at our armchair methods from a slightly different perspective: perhaps

they’re not a priori at all, but a posteriori. What Nolan means by this is that
our armchair methods generally involve the senses playing a role that is
not merely ‘enabling’ – a role that is something more than a necessary part
of the acquisition of concepts et cetera. So what more can senses do when
it comes to the armchair? Nolan identifies four possible tasks, but we’ll just
look at two of them. The first is assembling and evaluating commonplaces. This
task amounts to the analysis of such stories as the one about (the statue
of) Goliath and the Lumpl, a piece of clay of which Goliath is made. 32 The
debate about material constitution can be introduced with the help of such
commonplaces. Nolan’s idea in this regard is simple enough: that there are
statues made of clay and that such statues can be smashed without destroy-
ing the clay are, according to Nolan, pieces of a posteriori knowledge. But
the real interest of the suggestion is that philosophers can make surpris-
ing ‘discoveries’ on the basis of such commonplaces, which are seemingly
available to everyone on the basis of a posteriori knowledge. For instance,
the realization that Goliath and Lumpl have different persistence condi-
tions has important consequences for metaphysical debates about compos-
ition; indeed, this might be considered to go some way towards answering
Peter van Inwagen’s Special Composition Question, which we have already
discussed extensively in Chapters 2 and 3.
While Nolan is no doubt correct about the involvement of a posteriori
methods in our assessment of cases such as the one concerning Goliath and
Lumpl, there is still room to argue that the relevant philosophical work is
nevertheless a priori. For one might insist that we need to have some grasp
of the kind of thing that statues and lumps are before we are in any posi-
tion to draw the philosophically important conclusions. And if one follows
something like Lowe’s line of thought on how we come to know at least a
part of the natures of different kinds of things, it would seem that the pro-
cess starts from a priori inquiry (ignoring the possible ‘enabling’ role that
a posteriori knowledge may have). So this task may perhaps be read in two
different ways, depending on one’s prior commitments.
Another armchair task that Nolan discusses is applying theoretical vir-
tues. By ‘theoretical virtue’, Nolan means things like internal consistency,

For the srcinal example, see Alan Gibbard, ‘Contingent Identity,’ Journal of Philosophical
Logic 4 (1975), pp. 187–221.
7.4 Modal empiricism and the status of armchair methods 171

external coherence, simplicity, explanatoriness, fertility, unificatory

power, and other such comparative as well as internal virtues. 33 The need
for such a task is of course to assess the respective virtues of different
theories, which may help us to choose between them; this task thus con-
sists primarily of comparative judgements. It must be noted immediately
that it is typical in the literature discussing theoretical virtues to consider
them as a priori.34 Nolan protests that this is just an assumption, because
the process of ‘applying theoretical virtues’ is rarely elaborated on. He sug-
gests that the role that theoretical virtues play is primarily epistemic: it’s
epistemically better to accept a theory that better satisfies theoretical vir-
tues. Of course, it is another question why, exactly, the theoretical vir-
tues that we usually cite, such as simplicity, are epistemically better. One
explanation is simply that such theories are more likely to be true. But
other attempts to justify theoretical virtues can be made; Nolan mentions
that they can sometimes be justified with reference to other theoretical
virtues. For instance, unificatory power may often promote further sim-
plicity, and unifying two theories may also increase explanatory value.
Hence, the appeal to unificatory power could be justified in terms of these
other virtues, provided that they are considered valuable. Perhaps a more
direct justification could sometimes be drawn from predictive success: it
has often turned out that a simpler theory makes more accurate predic-
tions, which is at least a pragmatic consideration that speaks in favour of
simplicity. So Nolan would suggest that this type of comparative assess-
ment between various theories can be conducted in the armchair since it
relies on established evidence rather than any direct empirical work. But
it will be left to the reader to judge whether this constitutes a reason not
to regard this type of comparative assessment as a priori work, as Nolan
Regardless of what we conclude from Nolan’s analysis of armchair
philosophy, if we abandon the dogma that modal knowledge must be
secured purely by a priori or a posteriori inquiry, perhaps we can make fur-
ther progress. It is to the credit of the pioneering modal empiricist views

For some further discussion regarding theoretical virtues, see Chris Daly, Introduction
to Philosophical Methods (Peterborough, ON: Broadview Press, 2010).
For example, see L. A. Paul, ‘Metaphysics as Modeling: The Handmaiden’s Tale,’
Philosophical Studies 160.1 (2012), pp. 1–29.
172 The epistemology of metaphysics: a priori or a posteriori?

that this lesson is clearly taken seriously. Whether or not the result-
ing view is, properly speaking, modal empiricism is partly just a termino-
logical matter. What seems to be needed, in any case, is some sort of a
hybrid view.

7.5 Combining a priori and a posteriori methods

There are at least two reasons to think that both in the area of modal
epistemology, which we have been focusing on, as well as in the epis-
temology of metaphysics more generally, there is a need to employ both a
priori and a posteriori methods. The first reason is that the area of research
encompasses entities of many different kinds and it is plausible that we
cannot acquire knowledge of all these different kinds of things in exactly
the same manner. In particular, metaphysics investigates both concrete
and abstract entities. When we are dealing with abstract entities, such

as sets, it is difficult to see how any judgements we might make about

the nature of sets could be anything but a priori , since we have no direct
contact with sets (even if we may have direct contact with the members
of sets). In contrast, when our research concerns concrete things – say,
instances of natural kinds such as water – it seems that we simply have
no alternative but to resort to empirical research in order to learn about
their properties.
The second reason to think that some kind of a ‘hybrid’ view is likely to
be correct is due to the issues we discussed in the first section of this chap-
ter and towards the end of the previous section:the a priori vs. a posteriori
distinction itself would appear to allow for some vagueness. Hence,
a priori

and a posteriori elements in metaphysical inquiry may be intertwined to the

extent that it is quite difficult if not impossible to take them apart at all.
This, at any rate, is the possibility that we will briefly explore in this section.
As a first pass, let us briefly return to Lowe’s view regarding modal episte-
mology, which we classified under modal rationalism. Lowe’s view, at least at
first glance, appears to be uniform:he holds that our access to modal knowl-
edge is purely a priori. However, it should be noted that Lowe himself appar-
ently never used the label ‘modal rationalism’ for his view. In fact, Lowe is
very critical of the other modal rationalist approaches, such as intuition- and
conceivability-based approaches. He argues that a view taking intuitions as
evidential in metaphysics, quite generally, is
‘ misguided and
7.5 Combining a priori and a posteriori methods 173

leads inexorably to an anti-realist conception of metaphysical claims’.
will return to the role of intuitions as a source of metaphysical knowledge
in Chapter 8, but for now let us focus on Lowe’s positive proposal. We saw
that it is based on grasping essences, understanding what a thing is. But how
exactly does this happen if not with the help of intuitions or conceivability?
That is, what other purely a priori sources of knowledge are there? Positing
a distinct rational faculty, that ofunderstanding, is unlikely to be very help-
ful in answering this question, as we concluded earlier. Yet Lowe makes an
important qualification to how he conceives of this process:
he holds that
our inquiry (into essence and modality at least) should notbe consideredcom-
pletely independent of experience, but rather as proceeding in a ‘cyclical man-
ner, by alternating stages ofa priori and a posteriori inquiry’.36 On this point,
the present author is sympathetic to Lowe’s suggestion, having called this
type of ‘cyclical’ process a ‘bootstrapping relationship’ betweena priori and
a posteriori elements.37 The bottom line is that we should befallibilists about a
priori knowledge just as we are about empirical knowledge, and thus deny the
existence of ‘pure’ a priori or a posteriori knowledge. If any work at all is left
for the notion of apriority, it seems that we must qualify our conception of it
and move beyond the naïve Cartesian account ofinfalliblea priori knowledge.
In order to see how a hybrid account of metaphysical inquiry might
work, let us look at the classic example of water and H2O. 38 Suppose
that ‘water’ designates a genuine, mind-independent natural kind.
A genuine natural kind must have a determinable set of identity and
existence-conditions and we should generally be able to state them.
Whether water in fact is a genuine kind (or whether there are any such
kinds) is in fact open to debate, but since water is the locus classicus , it
serves our purposes well. According to the Kripke–Putnam framework,
we know that samples of water are made up of H2O molecules. 39 If it is

Lowe, ‘Essence vs. Intuition: An Unequal Contest,’ p. 256.
Ibid., p. 257.
Tahko, ‘ A Priori and A Posteriori: A Bootstrapping Relationship.’
The following discussion is adapted from Tuomas E. Tahko, ‘Natural Kind Essentialism
Revisited,’ Mind 124.495 (2015), pp. 795–822.
See Saul Kripke, Naming and Necessity (Cambridge, MA: Harvard University Press, 1980)
and Hilary Putnam, ‘The Meaning of “Meaning”’ (1975), reprinted in his Mind, Language
and Reality: Philosophical Papers, Vol. 2 (Cambridge University Press, 1979), pp. 215–271.
Here we will not attempt to be entirely faithful to the views of Kripke and Putnam, but
will instead focus on the received view based on their earlier work, here labelled the
Kripke–Putnam framework.
174 The epistemology of metaphysics: a priori or a posteriori?

also the case that water has its actual microstructure essentially – even
though empirical work is needed to determine what individual samples
of water are made up of – then ‘Water is H 2O’ is a metaphysically neces-
sary a posteriori essentialist truth. That water does have its actual micro-
structure essentially is usually considered to be knowable a priori , but
we will see that this assumption must be clarified. Accordingly, the core
of the Kripke–Putnam framework is that the combination of an essential-
ist a priori truth about a given natural kind essence and empirical, a
posteriori information about the microstructure of that natural kind are
needed to establish such metaphysically necessary theoretical identity
Typically, one would conclude that since empirical information is
needed to establish such metaphysically necessary truths, the process of
inquiry is, by and large, a posteriori. After all, the content of the a priori part
here is very simple indeed, it only amounts to saying that if water is H2O,
then this is necessarily so. This a priori part is quite generally taken to be
entirely unproblematic, since it is thought that natural kinds such as water
are obviously defined by their intrinsic properties, which is to say their
microstructure. But once we start to unpack this assumption, it turns out to
be rather more complicated. For it must be recalled here that our initial
familiarity with water (and other natural kinds), is not via its microstruc-
ture, but rather via its macroscopic, phenomenological properties, such as
its boiling point and its ability to dissolve other compounds. So it seems
that we need an additional assumption to complement the a priori element
to get the desired result, namely the metaphysical necessity of the theoreti-
cal identity statement ‘Water is H 2O’. This assumption is that the following
principle is true:

Microstructural determination (MD) Microstructure determines the

macroscopic properties of chemical substances.

Naturally, (MD) is thought to apply to cases such as water. This determi-

nation is manifested by chemical properties, that is, properties of a chemi-
cal substance that typically become evident in chemical reactions such as
oxidation (these are purely chemical properties), or else relate to the chemi-
cal structure of the substance (these are sometimes referred to as physi-
cal properties). Thus, using philosophical terminology, we might define a
chemical property as a property of a chemical substance in virtue of which
7.5 Combining a priori and a posteriori methods 175

the substance can undergo chemical reactions. The question that emerges
now is: what is the epistemic status of (MD)? For it is this principle that
would seem to do most of the work in securing the metaphysical neces-
sity of ‘Water is H2O’ rather than the relatively simple a priori principle we
started with – or indeed even the empirical discovery that water is in fact
H2O. Moreover, (MD) also seems to be a core part of the criteria of identity for
chemical substances.
Here we at the heart of the ‘hybrid’ approach: it turns out that the rele-
vant principle – that microstructure determines the macroscopic proper-
ties of chemical substances – is not easily established either by ‘pure’ a
priori or empirical inquiry. Why is this? Let’s attempt to unpack what it
means that a chemical property is a property of a chemical substance in
virtue of which the substance can undergo chemical reactions. This should
ring a bell for those who recall our discussion of the ‘in virtue of’ rela-
tion in Chapter 5. Take a concrete example: electronegativity – the abil-
ity of an atom or a functional group of a molecule to attract electrons.
For the purposes of understanding how microstructure and macroscopic
properties are assumed to be related according to the picture at hand,
tracking the source of this ability is important. In the case of electro-
negativity, the ability of an atom to attract electrons is influenced by its
nuclear charge. Atoms with a higher electronegativity attract valence
electrons more strongly; hence the distance from the atom’s nucleus to
its valence electrons tends to be shorter. There is a straightforward way
in which electronegativity is related to the microstructural properties
of the substance, to its nuclear charge in particular. So in this case it
seems that we do have a reasonably good idea about how electronega-
tivity and microstructure are related. However, electronegativity is itself
something that is not directly observed, and so it is not macroscopic in
the sense that chemical properties such as boiling point, solubility, and
flammability are. We do of course have a pretty good idea about some of
the microstructural properties that are correlated with properties such as
these, but correlation is not determination and hence not enough to estab-
lish (MD).
The problem can be highlighted by asking whether we are really deal-
ing with physical or chemical properties: it used to be (and often still is)
common practice to consider properties such as boiling point to be phys-
ical properties, not necessarily connected with the chemical properties of
176 The epistemology of metaphysics: a priori or a posteriori?

a substance, rather than grouping both sets together under ’chemical

properties’ as is now often done. Moreover, if microstructure deter-
mines chemical properties, then it seems that these properties should
be reducible to microstructure. But this is certainly a problematic assump-
tion, one that requires an argument.
The case of water and H2O was thought to be a simple combination of a
straightforward a priori principle and an empirical discovery, rendering the
relevant metaphysical necessity an a posteriori truth. But the upshot of our
discussion seems to be that there is a third part to the story, (MD), which
underlies the bridging of the srcinal a priori and a posteriori parts. What is
the epistemic status of that principle? This still remains unclear. Certainly,
it is a controversial principle, so at the very least we can conclude that we
are not dealing with a simple a priori truth. But since current empirical
data does not establish the truth of (MD) either, it appears that it cannot be
a purely empirical principle either. Yet, the principle has remained largely
unchallenged in mainstream metaphysics (albeit it has been repeatedly
challenged by philosophers of chemistry).
None of this entails that metaphysical inquiry quite generally proceeds
as it has apparently done in the case of ‘Water is H 2O’. However, given
the prominence of the example, this should at least make us alert to the
possibility of such ‘hidden’ premises, and hence also about drawing quick
conclusions about the epistemic status of given pieces of philosophical
knowledge. In the next chapter, we will look further into the epistemic
tools available to the metaphysician.

40 For further details, see Tahko, ‘Natural Kind Essentialism Revisited.’

8 Intuitions and thought
experiments in metaphysics

This chapter discusses two general, closely related epistemic tools used in
metaphysics: intuitions and thought experiments. As we saw in Chapter 7,
both of these tools make an appearance in modal epistemology, which was
our primary case study in the epistemology of metaphysics. But since intui-
tions and thought experiments are used in philosophy more broadly – not
just in metaphysics – they deserve a treatment of their own. Note how-
ever that we will be focusing on the usage of these two tools in metaphys-
ics. Metaphysicians routinely cite intuitions as prima facie evidence for a
given view, but sometimes the appeal to intuitions seems to serve an even
stronger role – especially in cases where there do not appear to be any
other epistemic tools at our disposal. Thought experiments likewise often
rely on an intuitive reaction, such as in the case of Hilary Putnam’s Twin
Earth scenario (discussed in Chapter 4): we are presented with a thought
experiment, but instead of an empirical test the relevant test seems to be a
test of intuitions. On the face of it, thought experiments are thus a type of
‘intuition pump’, as discussed by Daniel Dennett.1
Given that the literature on intuitions spans the whole discipline of
philosophy, we cannot engage with the topic in anything like a thorough
manner.2 Instead, we might focus on the status of intuitions with regard to
metaphysical thought experiments in particular. Perhaps a better under-
standing of how intuitions work will also help us to better understand

Daniel Dennett, Intuition Pumps and O ther Tools for Thinking (New York: W. W. Norton &
Co, 2013).
For a general overview, see Joel Pust, ‘Intuition,’ in E. N. Zalta (ed.), The Stanford
Encyclo pedia of Philosophy (Fall 2014 edn); see
fall2014/entries/intuition/ . A recent anthology of state-of-the-art articles that may
prove useful is A. R. Booth and D. P. Rowbottom (eds.), Intuitions (Oxford University
Press, 2014).

178 Intuitions and thought experiments in metaphysics

the role of a priori knowledge in metaphysics. One reason for addressing

intuitions and thought experiments in the same chapter is that, as Darrell
Rowbottom puts it, it is a prima facie attractive view that intuitions are to
thought experiments as perceptions or observations are to experiments
(and hence play an evidential role comparable to experiments).3 While ini-
tially attractive, this view does of course have obvious challenges: although
we have clear standards of what counts as a good experiment, repeatability
being one of the primary ones, intuitions are not, on the face of it, restricted
by any such standards. So to get any mileage out of this idea, we ought at
least to attempt to set some criteria for what counts as a ‘well-formed’ intu-
ition analogously to those for a respectable experiment. To this effect, we
need to be more precise about what is meant by ‘intuition’, for the term is
certainly not used in an entirely unambiguous fashion in philosophy. This
will be the task of the first section in particular, where two conceptions of
intuition are examined – these are the ‘common-sense conception’ and the
‘a priori conception’.
In the second section, we will very briefly discuss the study of intui-
tions in so-called ‘experimental philosophy’. This is strictly speaking not
of great relevance in metametaphysics, but we will consider a potential
structural issue that may influence discussions of intuitions in metaphysics
as well. The third section introduces a further specification regarding intu-
itions, which is of special importance in metaphysics: ‘experience-based
intuition’. These are typically based on the phenomenological aspects of
experience. In the fourth section we will consider a version of the ‘ a priori
conception’ of intuition in more detail, drawing especially on the work of
George Bealer.
The fifth and sixth sections concern thought experiments in science and
metaphysics, respectively. While we are of course primarily interested here
in the philosophical role of thought experiments, they are used not just
by philosophers but also by scientists – some of the most famous thought
experiments are familiar from the work of Einstein and Galileo, for
example. So one question is: do philosophical thought experiments differ
from scientific thought experiments? The answer might depend on what
the thought experiment is supposed to test. In the case of science, some of

Darrell P. Rowbottom, ‘Intuitions in Science: Thought Experiments as Argument
Pumps,’ in Booth and Rowbottom (eds.), Intuitions, p. 119.
8.1 Specifying ‘intuition’ 179

the most famous thought experiments have actually been recreated in the
laboratory (e.g., the Einstein–Podolsky–Rosen thought experiment regard-
ing quantum entanglement, which we’ll discuss in the fourth section), but
in philosophy this would not always be possible. Certainly, many thought
experiments in epistemology have been recreated, most famously the
Gettier cases – they are quite easy to recreate.4 But there are many meta-
physical thought experiments we probably would not like to see realized,
at least not the ones involving zombies! (Of course, philosophical zombies
are nothing like Hollywood zombies, but some might regard qualia-lacking
copies of ourselves rather scary …) However, as it will turn out, the appar-
ent differences between scientific and philosophical thought experiments
may not be as drastic as it might first seem.

8.1 Specifying ‘intuition’

Let us focus on two candidate precisifications for the term ‘intuition’ as it

is used in philosophy; call them the ‘common-sense conception’ and the
‘a priori conception’.5 These are by no means the only candidate meanings,
but they do capture a substantial portion of the uses of ‘intuition’ in phil-
osophy and in metaphysics in particular. The first, common-sense concep-
tion, is fairly straightforward: intuitions in philosophy operate like ‘folk’
intuitions generally seem to do, namely as initial, pre-theoretical reactions
to various scenarios. In philosophy, this applies specifically to philosophical
thought experiments, but ‘folk’ intuitions may of course be applied more
broadly. We might take Putnam’s famous Twin Earth thought experiment
as our example, bearing in mind that a full analysis of the scenario ought
to take into account the discussion in Chapter 7 regarding ‘Water is H2O’.6
The Twin Earth thought experiment could be summarized as follows. Let

Gettier cases are examples of justified true belief (JTB) which we would not intuitively
consider as knowledge, hence challenging the classic JTB analysis of knowledge.
These and other conceptions of ‘intuition’ are discussed in more detail in C. S. I.
Jenkins, ‘Intuition, “Intuition”, Concepts and the A Priori,’ in Booth and Rowbottom
(eds.), Intuitions, pp. 91–115.
For the srcinal Twin Earth scenario, see Hilary Putnam, ‘The Meaning of “Meaning” ’
(1979), reprinted in his Mind, Language and Reality: Philosophical Papers, Vol. 2 (Cambridge
University Press, 1979), pp. 215–271. For further discussion regarding the analysis of
the scenario, see Tuomas E.Tahko, ‘Natural Kind Essentialism Revisited,’ Mind 124.495
(2015), pp. 795–822.
180 Intuitions and thought experiments in metaphysics

us imagine that the chemical properties of water are reproduced by some

molecular structure other than H2O – say, XYZ. This could perhaps happen
in another possible world or in a remote location of our own universe –
either option introduces some complications, but we can set them aside
for the purposes of this discussion. The question is: should we consider
this substance, XYZ, to be water? The typical intuitive reaction is that XYZ
is not water. The empirical details of the thought experiment are rarely
discussed, at least by metaphysicians. But the role of empirical knowledge
regarding the existence of metaphysically possible microstructures that
could replicate the chemical properties of water is certainly debatable,
partly for reasons that we discussed in the previous chapter. Chemistry will
presumably be of some help here, but it is often thought that we also need
metaphysical a priori work to determine what is metaphysically possible. In
later work Putnam himself expressed serious doubts about extending the
Twin Earth thought experiment across metaphysically possible worlds and
even said that the question about the possible variation of the laws of phys-
ics with regard to water ‘makes no sense’.7 Setting Putnam’s worries aside,
it is quite common to take the result of our intuitive reaction to the Twin
Earth thought experiment to be that water has its actual microstructure by
metaphysical necessity. But isn’t this an overly strong result to arrive at on
the basis of our ‘folk’ intuitions?
Here we arrive at the primary worry regarding the common-sense con-
ception of intuition. Insofar as intuitions are pre-theoretical reactions like
the one described in the case of the Twin Earth thought experiment, their
evidential role cannot be on a par with that of observations in regular
experiments. For one thing, the setting and interpretation of the thought
experiment itself will cause complications: in the Twin Earth thought
experiment the modal status of the scenario is particularly problematic,
as Putnam’s own later reaction makes clear. Moreover, as we discussed in
Chapter 7, the intuitive reaction seems to be based on assuming the truth
of an underlying, more controversial a priori principle, namely (MD) (which
was introduced in Chapter 7): microstructure determines the macroscopic
properties of chemical substances. But once we have identified all such
complications, our reaction to thought experiments may very well change.

Hilary Putnam, ‘Is Water Necessarily H2O?’, in J. Conant (ed.), Realism with a Human Face
(Cambridge, MA: Harvard University Press, 1990), p. 70.
8.1 Specifying ‘intuition’ 181

So there are reasons to think that pre-theoretical intuitions understood in

the lines of the common-sense conception should not serve an evidential
role. Of course, this does not mean that intuitions understood in this fash-
ion could not serve a different role, but it would have to be significantly
weaker than we might desire if intuitions are to be considered an impor-
tant tool in the epistemology of metaphysics.
Are there reasons to think that ‘intuition’ can be better specified in
terms of apriority? A feature of intuitions that might at least initially point
towards the a priori conception is that intuitions are typically considered
immediate.8 According to Carrie Jenkins’s analysis, this could mean two
different things. First, intuitions can be immediate in the sense of being
direct or non-inferential – or, at any rate, if there is some kind of an infer-
ence underlying an intuition, it is not explicit and, perhaps, not conscious.
Jenkins considers this type of view about intuitions to be quite common,
especially among epistemologists. There is, however, an obvious vague-
ness to this understanding of immediacy, for it is not clear what kind of
a role is played by the purely psychological factor of not being aware of any
explicit inference. Jenkins distinguishes this psychological criterion for
immediacy from the stronger requirement that immediacy means that
there is no underlying inference, not even an implicit, non-conscious one.
She does not further engage with this issue, but we might note that the
stronger requirement seems, perhaps, too strong, for if even non-conscious
inference isn’t allowed, then it might be impossible for a person to know
whether her immediate, intuitive reaction to something truly is immedi-
ate in the required sense. If this has an influence on the epistemic value of
intuitions, then we do not seem to have made much progress.
The second sense of immediacy that Jenkins identifies is a more liberal
one: intuitions are immediate in the sense of being (or at least seeming to
be) obvious, spontaneous, natural, compelling… These features are more
straightforwardly psychological (orphenomenological, relating to how the sub-
ject perceives the intuitive process), for better or worse. We may be in a risky
area here. If we start from the idea that intuitions ought to or could count
as evidence, then special attention should be given to the question whether
intuitions counting as evidence depend on phenomenological factors. An ini-
tial reason for concern is that unless we are very confident about our ability

See Jenkins, ‘Intuition, “Intuition”, Concepts and the A Priori,’ p. 94.
182 Intuitions and thought experiments in metaphysics

to identify the relevant phenomenological aspects that are associated with

intuitions in ourselves, then they may be unlikely to produce a sufficiently
reliable method. This is not to suggest that intuitions couldn’t be
fallible (more
on this shortly). But unless we have some further input on the degree of
confidence regarding our ability to identify the relevant phenomenological
aspects, then the process seems fallible on two counts. These are our ability
to identify something as an intuition in the first place and the reliability of
intuitions themselves. If we lack criteria of reliability on both counts, then
the use of intuitions might become hopelessly unreliable. But before this
sceptical challenge is deemed to refute intuitions as evidence, it ought to
be noted that similar considerations might just as well be applied to
tion: whether a perceptual state counts as evidence or not likewise seems to
depend on phenomenological factors, at least tosome degree. Sometimes we
have phenomenological reasons to think that a certain perception could be
unreliable. For instance, if one has blurred vision caused by watery eyes, this
would be an immediate phenomenological reason to lower the reliability of
visual evidence. So it is not obvious that intuitions are relevantly different
from perceptions in this regard.
Elijah Chudnoff has recently argued to the effect that intuitions share
many of the important epistemic and phenomenological aspects of sen-
sory perception.9 There is at least one important difference as well, though;
according to Chudnoff, ‘sensory perception includes concrete reality and
excludes abstract reality’, while ‘the subject matter of intuition includes
abstract reality and excludes concrete reality’. 10 This clearly sets intuition
and perception apart, even if they do share important factors. We will not
attempt to settle the abstract/concrete question here though. Chudnoff
himself emphasizes that what’s important is not the stronger claim that
intuition excludes concrete reality, but that it includes abstract reality. This
is enough to distinguish sensory perception and intuition.
Let us return to other general conditions for intuition. If immediacy with-
out further qualification seems too crude to characterize intuitions on its
own, then perhaps some additional characterization will help. One possibil-
ity is to look for a conceptual or linguistic basis, whereby intuitions are asso-
ciated with conceptual analysis, linguistic competence, or understanding.

Elijah Chudnoff, Intuition (Oxford University Press, 2013).
Ibid., p. 11.
8.1 Specifying ‘intuition’ 183

It’s fairly easy to see where this type of view might srcinate from. For
instance, when presented with propositions that are analytically true
– true
in virtue of the meanings of the concepts involved– we seem to have a rele-
vantly immediate reaction to them. Consider the proposition ‘All vixens are
female’. Since vixens are female foxes, the concept of ‘female’ would appear
to be contained in the concept of ‘vixen’. So we might say that it is purely
by understanding what the word‘vixen’ means that we know ‘All vixens are
female’ to be true. An interesting question is whether this involves an infer-
ence of some sort. Certainly, in most cases we would consider such cases to
be obvious in the sense discussed earlier, so any inference involved would
likely be implicit. But it is a matter of debate whether conceptual analysis
in general could be understood as free from inference in the relevant sense.
In any case, there is clearly a limitation with the view that intuitions are
closely connected with conceptual or linguistic competence, as very often
intuitions are used in connections where there seems to be more at stake
than conceptual or linguistic issues. This is arguably alsothe case in the Twin
Earth scenario, even though it should be noted that Putnam was srcinally
interested in the semantic aspects of the scenario, namely whatwe mean by
‘water’. But since the upshot of the scenario even when considered on purely
semantic terms is supposed to be thatmeanings are fixed externally (it’s the
microstructure of water that counts), it appears that the intuitive reaction
to the scenario is richer than just a grasp of something like the truth of ‘All
vixens are female’. Indeed, Ernest Sosa emphasizes that focusing on purely
linguistic features in connection with intuition is not sufficient. Rather, we
should be more interested in the underlying process ofunderstanding, which
Sosa specifies as follows:

Fundamental, intuitive rational beliefs are based atleast on understanding

of the propositions believed […]It is not, however,just the understanding of
a proposition, whatever its content, that gives a proper basis for believing it.
Otherwise, it would also constitute a basis for believing its negation, which
must be equally well understood. […] What sufficesis rather the being
understood (shared by a proposition and itsnegation pretty much equally)
along with the specific content of that very proposition

Ernest Sosa, ‘Intuitions: Their Nature and Probative Value,’ in Booth and Rowbottom
(eds.), Intuitions, pp. 36–49.
184 Intuitions and thought experiments in metaphysics

Given that Sosa includes the content of propositions as a crucial part of the
basis of intuitive rational beliefs, it would seem that the type of ‘intuition’
at work in the Twin Earth scenario could be salvaged, for it seems pos-
sible to build enough information (about the microstructure of water, for
instance) into the content of propositions. However, this certainly shifts
the focus from pure conceptual or linguistic ability towards something
richer. If we need both understanding and some further grasp of the con-
tent of propositions to reach intuitive rational beliefs, then the immediacy
criterion might once again be questioned.
A recent extensive study of the use of intuitions in philosophy, by
Herman Cappelen, focuses on this type of conceptual competence view
about intuitions, noting their specific phenomenology and epistemic
status.12 Cappelen suggests that intuitions understood in this fashion are
commonly considered to be used extensively as evidence in contemporary
analytic philosophy. But Cappelen’s key claim is that these types of intui-
tions are in fact not used as evidence, contrary to what many contemporary
analytic philosophers think. This calls for further reflection on the meth-
ods actually in use in contemporary analytic philosophy.
While Cappelen’s project is certainly interesting, we cannot engage
with it in detail here. By way of justifying this omission, we have already
seen some shortcomings in the conception of intuitions according to
which they are based on conceptual competence. In his commentary on
Cappelen’s book, David Chalmers has argued to the same effect, noting for
instance that even if in some areas of philosophy intuitions seem to stem
from conceptual competence, there are other areas where the intuitions
relied upon are clearly not similar, such as moral intuitions and other nor-
mative intuitions.13 It is more plausible (even if not uncontroversial) that
conceptual competence is central for linguistic intuitions, which are natur-
ally of interest in philosophy of language. More to the point, there may
be specific reasons to avoid this conception of intuitions when it comes to
metaphysical inquiry. For unless we have prior reasons to think that con-
cepts reliably mirror reality, then we have little reason to think that intui-
tions based on conceptual competence will provide reliable evidence of

Herman Cappelen, Philosophy without Intuitions (Oxford University Press, 2012).
David J. Chalmers, ‘Intuitions in Philosophy: A Minimal Defense,’ Philosophical Studies
171.3 (2014), pp. 535–544.
8.2 Intuitions and experimental philosophy 185

the structure of reality.14 Admittedly, this requirement may already be too

strong: is it even possible to have good reasons to think that our concep-
tual framework, quite generally, mirrors reality? Since we can’t very well
check this by resorting to some other framework, it looks as if the burden
of proof here should be on the sceptic. In other words, it may be acceptable
to assume that, by and large, our concepts do reflect reality unless there are
good reasons to think that they do not. In some cases, such good reasons
undoubtedly exist and indeed it would be a move comparable to the scep-
tic’s to assume that our concept always reliably mirror reality. Accordingly,
establishing some criteria to evaluate the reliability of concepts in this
sense without assuming either of the extreme views would seem like a
reasonable way to make progress. Attempts to engage in this type of work
certainly exist; we will discuss some relevant work below.

8.2 Intuitions and experimental philosophy

Before we move on, let us briefly note one interesting aspect related espe-
cially to the study of linguistic or conceptual intuitions (and of special rel-
evance from the point of view of epistemology). A movement which we
have not yet discussed is so-called experimental philosophy or ‘x-phi’.15 This
is an area which focuses on, among other things, testing popular philo-
sophical thought experiments and the intuitions that they generate in the
general population, especially transculturally. Among popular test cases
are the Gettier cases, having to do with intuitions regarding the concept
of knowledge. But there is also a growing literature on experimental phi-
losophy applied to metaphysics and to causation in particular. 16 Debate

However, for a defence of the view that experience provides an epistemic ground-
ing for our concepts and hence the needed mirroring, see C. S. Jenkins, ‘Concepts,
Experience and Modal Knowledge,’ Philosophical Perspectives 24 (2010), pp. 255–279.
15 For a survey of experimental philosophy, see for instance J. Horvath and
T. Grundmann (eds.), Experimenta l Philosophy and its Critics (London: Routledge, 2012).
A very recent analysis focusing especially on the source of intuitions is Helen de
Cruz, ‘Where Philosophical Intuitions Come From,’ Australasian Journal of Philosophy
See for instance David Rose and David Danks, ‘Causation: Empirical Trends and Future
Directions,’ Philosophy Compass 7.9 (2012), pp. 643–653. Another recent area of focus,
which we’ll discuss in more detail below, is temporal experience, see L. A. Paul,
‘Temporal Experience,’Journal of Philosophy 107.7 (2010), pp. 333–359.
186 Intuitions and thought experiments in metaphysics

about the merits and usefulness of experimental philosophy is heated. One

central issue concerns the difference, or lack thereof, in ‘folk’ intuitions
as opposed to the intuitions of professional philosophers. In other words,
what kind of weight should we put on studies focusing on ‘folk’ intuitions?
Also, does it matter if there is transcultural variation? An obvious concern
is also the setup of the experiments themselves, since it may be especially
difficult to describe philosophical thought experiments to non-experts
without biasing the interpretation one way or the other. Further, for some-
one who does not put much weight on intuitions in the first place, the
project of experimental philosophy may seem to be beside the point. For
instance, why would it matter whether there are, say, several notions of
causation at play in folk ‘intuitions’ if the relevant research is to be done
primarily with reference to fundamental physics? Indeed, it is difficult to
even discuss the topic of experimental philosophy unless we have some
more general agreement about the method of metaphysics, such as that
‘folk’ intuitions do and/or should play a role in it.
Because of these difficulties, we might fare better in the present context
by focusing on the general role of intuitions in philosophical methodology.
To this end, a recent note by Anand Vaidya is of special interest, since he
purports to have identified an element that underlies much of the experi-
mental philosophy literature; he calls it the Dependency Thesis (DT):

(DT) Experimental inquiry that aims at the discovery of variation in

intuitions about the application of philosophical concepts essentially
depends on non-experimental intuitions about the application of
philosophical concepts. And non-experimental intuitions are indispensible
[sic] for experimental inquiry into the reliability of intuitions about the

application of philosophical concepts.17

If Vaidya is correct about there being certain ‘non-experimental’ intui-

tions regarding the application of philosophical concepts, then those
intuitions are likely applicable to any philosophical debate concern-
ing philosophical concepts. As Vaidya points out, (DT) itself is a meta-
physical thesis about the relation between intuitions and inquiry and
hence it appears to be of special relevance from the point of view of

Anand Vaidya, ‘Intuition and Inquiry,’ Essays in Philosophy 13.1 (2012), Article 16.
8.2 Intuitions and experimental philosophy 187

The crucial notion in (DT) is ‘non-experimental intuition’. Vaidya gives a

formal definition of this notion, but an informal sketch suffices here; the
central idea is that non-experimental intuitions are something that one
must eventually appeal to in order to establish a criterion for determining
whether a (philosophical) concept is shared by two groups. Now, in the
setting of experimental philosophy, the two groups are likely to be, say,
Europeans and Asian, or some other culturally or geographically separated
groups. But if we are to apply the idea more generally, we can think of
the two groups simply as proponents of two different philosophical views.
A good example might be an anti-realist about something and a realist
about the same issue. We can immediately bring to mind several examples
discussed in earlier chapters that could be analysed in this fashion, such
as the debate between nihilists and universalists about mereological compos-
ition discussed in Chapters 2 and 3, or the debate between perdurantists and
endurantists in Chapter 3. What we are interested in is whether these two
groups share the philosophical concept(s) relevant for the debate, such as
the concept of composition.
The upshot of Vaidya’s analysis is as follows. If we hope to make pro-
gress in any debate possibly involving a variation of intuitions regarding a
given concept, we need to distinguish two possibilities:

(i) The proponents of two competing views both possess the relevant con-
cept and disagree about its application in a given case.
(ii) The proponents of two competing views do not both possess the rel-
evant concept and their disagreement is actually due to applying differ-
ent concepts in the given case.

One could attempt various strategies in distinguishing between these

cases, such as conducting an experimental study about the relevant con-
cept to generate the criterion for its usage or looking at a distinct theory
which proposes such a criterion.18 But each of these options, as suggested
by Vaidya in the case of the concept of ‘knowledge’, is likely to involve
an appeal to some further non-experimental intuitions. If this is indeed
the case, then intuitions are pervasive in metaphysical inquiry as well.
Note that here the proposed role of intuitions is not evidential, but rather

For discussion of such attempts with regard to the concept of causation, see Rose and
Danks, ‘Causation: Empirical Trends and Future Directions.’
188 Intuitions and thought experiments in metaphysics

something that is used to ensure that a debate is substantial, that is, non-
linguistic. How this is to be done, exactly, is another matter. In any case,
one consequence of this is that intuitions seem to come into play in meta-
physics whenever there is a possibility of a variation in the possession of
relevant concepts between proponents of competing views. But let us now
move to a more direct usage of intuitions in metaphysics.

8.3 Experience-based intuitions

The use of intuitions in metaphysics has not been studied as widely as their
use in epistemology and philosophy of language, for instance. One reason
for this may be the fact that many of the relevant intuitions do not neatly
fit into the specifications that we discussed above. For instance, metaphysi-
cal inquiry into the nature of time and space would typically start from
our experience of time and space rather than from, say, the analysis of the
concepts of ‘time’ and ‘space’. A very recent survey of the usage of such
experience-based intuitions has been made by Jiri Benovsky. 19 Benovsky
analyses a number of case studies regarding the use of experience-based
intuitions in metaphysics, ultimately drawing the conclusion that the intu-
itions in use – if they are to be called that – are by and large based on our
(varying) phenomenological experiences, which are unable to stand their
ground in the face of scientific scrutiny. The upshot of his analysis is that
we ought to not take such ‘intuitive data’ as serious metaphysical evidence.
But let’s take a look at one of Benovsky’s case studies so that we can decide
for ourselves.
Take the case of temporal experience with regard to the judgements we
make about the A-theory and B-theory of time. 20 Very briefly, the A-theory
amounts to the view that tense properties, meaning those involving time,
are irreducible – that is, ‘yesterday’, ‘last year’, ‘a million years ago’, and so

Jiri Benovsky, ‘From Experience to Metaphysics: On Experience-based Intuitions and
their Role in Metaphysics,’ Noûs (2013 [online], forthcoming).
This is not the place to introduce the reader to the details of these theories, but for
a compact introduction, see Ned Markosian, ‘Time,’ in E. N. Zalta (Ed.), The Stanford
Encyclopedia of Philosophy (Spring 2014 edn); see
spr2014/entries/time/. The locus classicus in this area is J. M. E. McTaggart, ‘The Unreality
of Time,’ Mind 17 (1908), pp. 457–473; reprinted in R. Le Poidevin and M. McBeath
(eds.), The Philosophy of Time (Oxford University Press, 1993), pp. 23–34.
8.3 Experience-based intuitions 189

on are fixed by the passage of time and correspond to genuine properties;

call these A properties. The B-theory, in contrast, denies the reality of tense
and holds that all these properties involving time are reducible to two-place
relations such as ‘a day earlier than’, ‘a year earlier than’, ‘amillion years
earlier than’; call these B relations. Each theory can be supported with vari-
ous metaphysical (and scientific) arguments, but most people would agree
that the A-theory has a certain intuitive advantage because on the face of
it only the A-theory can account for the passage of time. Temporal passage
is of course a key part of how we experience the world, so it would appear
that the intuitive support for the A-theory is derived from the phenomenol-
ogy of time, that is, temporal experience.21 The role of this phenomenologi-
cal element is larger than one might initially think: it is at least arguable
that if we did not perceive change in our surroundings or even just in our
thoughts, then time would not seem to be passing. The force of this point is
easy to see when we consider how subjective our perception of the passage
of time is: time can appear to pass very slowly or very fast depending on
external factors such as changes (or lack thereof) in our environment, or
depending on internal factors such as stress or chemically induced states
(i.e., drugs). Indeed, our sense of how quickly the time passes does not seem
to be very reliable at all.
Building on the point that change typically involves movement, Benovsky
observes that these phenomenological aspects regarding the passage of
time have peculiar consequences:

Take the case of the hour hand on a mechanical watch:it moves so slowly
that we just do not perceive it as moving. We can of course observethat
it has moved, if we look at it after some time, but such a “perception” is

no experience of movement at all. Thus, we have a case where there is

movement, but we do not experience it as such, simply because the hour
hand’s continuous movement is far too slow for us to be able to perceive
it. Indeed, what we realize here, is that our capacities to notice change and
movement have a lower limit and that anything that moves too slowly will
not be registered by our perceptual system as moving.22

Similarly, there are cases of ‘apparent motion’, where weperceive some-

thing as moving without there being any real motion at all – various visual

See Paul, ‘Temporal Experience.’
Benovsky, ‘From Experience to Metaphysics,’ p. [5].
190 Intuitions and thought experiments in metaphysics

illusions are a good example. Since our sense of the passage of time is
closely linked to change and hence motion, this seems once again to high-
light the unreliability of the phenomenology of time perception.
What is the upshot? Well, since the phenomenology related to our
experience of time is clearly very badly biased and often misleading, any
intuitive support that it provides for a general metaphysical theory about
the nature and reality of tense is surely in serious doubt. Based on this
and other examples, Benovsky states that we can draw no metaphysical
conclusions from the nature of our experience. So this purports to be a
strong case against using experience-based intuitions as evidence – or even
as heuristic tools. There are fairly obvious metametaphysical implications
to be drawn from this, one of them being that at least certain types of intui-
tions are likely to be ‘polluted’ by the phenomenological framework of our
experience. If this is the case, we need to seriously reconsider the value of
data drawn from these intuitions. But perhaps there are different types of
intuitions? If so, such intuitions might be able to avoid being polluted in
this sense. One possible avenue are pure ‘ a priori intuitions’, which would
appear to be more directly tied to our cognitive or rational abilities.

8.4 Rational intuition

It is not untypical to propose a strong connection between intuitions and

rationality.23 The result is an intriguing characterization of the a priori con-
ception of intuitions, according to which our belief in certain propositions
is justified by rational intuition rather than by the senses, introspection, or
memory. A good example is Bealer’s work: he proposes that we ought to
distinguish between the ‘a priori intuitions’ used in philosophy and the
‘physical intuitions’ used in science.24 If intuitions were to provide us with
similar warrant to a priori knowledge more generally, then their evidential
role would clearly be different from that of empirical warrant. Indeed, on
Bealer’s line, something that we intuit, philosophically, seems necessary to
us. This further tightens the connection between apriority and intuition.

E.g., George Bealer, ‘Intuition and the Autonomy of Philosophy,’in M. DePaul and W.
Ramsey (eds.), Rethinking Intuition: The Psychology of Intuition and Its Role in Philosophical
Inquiry (Lanham, MD: Rowman and Littlefield, 1998), pp. 201–240; Laurence BonJour,
In Defense of Pure Reason (Cambridge University Press, 1998).
Bealer, ‘Intuition and theAutonomy of Philosophy,’p. 165.
8.4 Rational intuition 191

Yet, it also produces an immediate limitation: if what we intuit is neces-

sary, then the subject matter of intuitions must be the realm of necessary
truths, for instance analytic, conceptually necessary truths, such as the
earlier ‘All vixens are female’. But it can’t be as simple as this, for surely
intuitions are often also fallible , in which case we might sometimes think
that what we’ve intuited is necessary (and a priori), but it is in fact false.
This is how it should be, though: one can be justified a priori in believing
that something is necessary, on the basis of an intuition for instance, but
that justification could later be defeated, for instance in light of empirical
Bealer has put forward a number of arguments to the effect that it is
incoherent not to consider intuitions as evidence. 25 He pitches them against
‘Quinean’ empiricism, which is committed to three principles: empiricism,
holism, and naturalism. The core of their combination is that experience/
observations form the prima facie basis of a person’s evidence, so that a
theory is justified if and only if it is the simplest complete theory that justi-
fies that evidence, and the natural sciences comprise the simplest theory.
The first of Bealer’s arguments, the Starting Points Argument, is very simple;
it points out that even empiricists rely on intuitions when following justifi-
catory procedures. A likely reply would be that even if intuitions are relied
upon initially, they do not participate in the justificatory part. So intuitions
could serve as a guide in the formulation of theories (the discovery part), but
not in their justification. Yet, Bealer points out, if the intuitions involved
in the starting points of the theories – the discovery part – were not reli-
able, then surely they would also influence the justification part. Hence,
even the empiricist must admit that the intuitions involved must be able
to serve as evidence.
Bealer’s second argument to this same effect, the Argument from Epistemic
Norms, runs as follows. We have certain epistemic norms that we routinely
use in our justificatory procedures. If the empiricist wishes to rule intui-
tions out as evidential, then it should be shown that intuitions fail when
it comes to this standard assessment of justificatory procedures. More pre-
cisely, Bealer argues that to abandon some source of prima facie evidence,
it should be demonstrated that this piece of evidence fails to satisfy the

George Bealer, ‘The Incoherence of Empiricism,’ Aristotelian Society Supplementary Volume
66 (1992), pp. 99–138.
192 Intuitions and thought experiments in metaphysics

‘three cs’: consistency, corroboration, and confirmation. As an example of a clear

failure of all three cs, consider prediction made with the help of tea-leaf

First, to the extent that we have looked, we find no particular consistency

among the tea-leaf readings made by a single person. Second, a person’s
readings are not corroborated by other people. Third, there is no pattern
of confirmation of the tea-leaf predictions or other tea-leaf claims by our
experiences, observations, and intuitions. Indeed, there is a pattern of
disconfirmation by these sources of prima facie evidence.26

In contrast, Bealer argues that intuitions manage to satisfy all three cs.
First, if you consider your intuitions, it is likely that they are largely a con-
sistent set. But even if some of your intuitions do conflict each other, you
will likely be able to come up with a more complete description of what
is happening in the problematic scenario in such a way that the apparent
conflict of intuitions can be resolved. Secondly, even though there may be

cases where your intuitions conflict with those of someone else, who there-
fore appears to fail to corroborate your own intuitions, there are also plenty
of clear test cases where our intuitions corroborate each other extremely
strongly. For example, in the areas of elementary logic and mathematics,
intuitions are widely shared. But even in more complex cases, such as the
Twin Earth thought experiment, the intuitive reaction to the scenario is
widely shared, at least when specified further. So it may be that often the
initial conflict in intuitions is simply due to an under-described scenario.
Finally, our intuitions seem to be very rarely disconfirmed by direct empir-
ical evidence, since the judgements we make on the basis of intuitions are
often quite independent of experience and observation.
One might of course further challenge that while some intuitions may
satisfy the three cs, it is nevertheless clear that not all intuitions fully sat-
isfy them. But perhaps this is acceptable to a moderate rationalist – to use
Bealer’s term. After all, if we endorse the fallibility of intuitions, we had
better acknowledge that in some cases they fail. It is enough to maintain
that intuitions can, at least sometimes, be used as evidence. Bealer himself
attempts to make a somewhat stronger case, but here we will set aside his
further discussion, largely because it is aimed primarily at the ‘Quinean’

Bealer, ‘The Incoherence of Empiricism,’ p. 110.
8.4 Rational intuition 193

empiricist, whereas we are interested here in a more general picture of the

epistemology of metaphysics– one that may be at odds with ‘Quinean’ empiri-
cism to begin with. But it is worth mentioning that this is a topic which con-
tinues to be discussed. Chudnoff, who has developed an alternative to the
type of understanding-based account of intuitions on which Bealer, Sosa,
and others have elaborated, also challenges such empiricism. An important
part of his argument, mentioned in passing earlier, is that intuitions concern
abstract matters, whereas sensory perception concerns concrete matters. The
empiricist, it is suggested, struggles to explain the former:

Our abstract knowledge encompasses knowledge about the necessary,

normative, infinite, and abstract. Sensory perception gives us knowledge
about the contingent, non-normative, finite, and concrete. And it is not
clear how memory, testimony, and deductive, inductive, and abductive
inference could work up knowledge about the contingent, non-normative,
finite, and concrete that sensory perception gives us into knowledge about
the necessary, normative, infinite, and abstract.27

This may remind the reader of issues discussed in Chapter 7 regarding the
relationship between a priori and a posteriori reasoning. Chudnoff makes a
related point: even if the empiricist has trouble accounting for the abstract,
the ‘apriorist’ also has trouble explaining how any ‘pure’ form of reasoning
alone could form the epistemic basis for any beliefs. It would seem that
some pre-existing background knowledge is always required. Chudnoff
himself proposes that intuitive reasoning could be the missing link: it
‘injects’ a priori reasoning with some content – this is a crucial element in
Chudnoff’s account of intuitions as ‘intellectual perceptions’. 28 However,
it will be left up to the reader to decide whether ‘pure’ empiricism can be
maintained when it comes to metaphysics. We have certainly seen that the
question is considerably more difficult than it may have initially seemed.
To finish this section, let us take a brief look at a recent attack on
intuition-based philosophy of the type suggested here, by James Ladyman
and Don Ross.29 This attack is based on something that we mentioned ear-
lier: the apparent difference between philosophical armchair intuitions

Chudnoff, Intuition, p. 14.
Ibid., p. 15.
J. Ladyman and D. Ross (with D. Spurrett and J. Collier), Every Thing Must Go (Oxford
University Press, 2007), pp. 10–15.
194 Intuitions and thought experiments in metaphysics

and those used in science. Ladyman and Ross give a number of examples
of metaphysical armchair intuitions which seem to be blatantly incorrect
from a scientific point of view. Indeed, it is easy to find such examples, in
metaphysics and science alike – just consider the fact that Newton devoted
more time to alchemy than any other area of research and apparently
thought that this work was highly important. So they emphasize the point
that intuitions are clearly not a reliable method of inquiry, but they don’t
think that this entails that armchair reasoning is completely worthless. For
they admit that it is often said that a particularly good physicist has ‘sound
physical intuition’. But here the use of the word ‘intuition’ is supposedly
different, as it refers to ‘the experienced practitioner’s trained ability to see
at a glance how their abstract theoretical structure probably – in advance
of essential careful checking – maps onto a problem space’.30 Ladyman and
Ross further distance the intuitions of metaphysicians from the intuitions
of scientists by pointing out that the former are often taken as evidence
whereas the latter are only heuristically valuable. On this view, intuitions
should at best constitute only prima facie evidence: careful study, or in some
cases empirical research, is required before they can be accepted. So it may
indeed be that intuitions in philosophy and science are used somewhat
differently and perhaps this speaks against them as reliable evidence in
philosophy. But perhaps we should not put too much weight on this differ-
ence until a more comprehensive study of how intuitions are in fact used
in science is seen.31 We will now change the topic a bit and look into the
method of generating intuitions in the first place, namely thought experi-
ments; it may be illuminating to start by considering exactly how thought
experiments are used in science.

8.5 Scientific thought experiments

To continue on from the previous section, it is worth noting at the out-

set that thought experiments are perhaps the most promising example
of shared ground between science and philosophy, since in certain areas
of natural science, such as physics, thought experiments have a long

Ibid., p. 15.
31 For what may very well be a first systematic attempt at such a study, see Jonathan
Tallant, ‘Intuitions in Physics,’ Synthese 190 (2013), pp. 2959–2980.
8.5 Scientific thought experiments 195

history. The relationship between science and metaphysics more gener-

ally is a topic that we will focus on in Chapter 9, so we can here limit our
focus on thought experiments in particular. The view that scientific and
philosophical thought experiments are indeed similar is rather popular,
at least among philosophers. But rather than attempt to list the poten-
tial similarities between them, we might instead consider what could
possibly make philosophical and scientific thought experiments differ-
ent. Some reasons to think that they are different have been put forward
by David Atkinson – a physicist. 32 Atkinson’s main point is that thought
experiments which do not lead to real, empirical experiments, are not
as valuable as the ones that do. Since philosophical thought experi-
ments typically do not lead to experiments while many scientific ones
do, this seems like an obvious difference between them. But note that
Atkinson does not consider only philosophical thought experiments to
be poor ones, but also a number of scientific thought experiments, such
as Galileo’s famous thought experiment about falling bodies in response
to Aristotle’s theory.
In Galileo’s thought experiment, we are to imagine two objects of dif-
ference masses connected via a string. This system of two objects should
be dropped from, say, a tower. If we assume that heavier objects fall faster
than lighter ones – as Aristotle’s theory of gravity predicts (objects fall at a
speed relative to their mass) – then the string connecting the lighter object
to the heavier one will quickly tighten, with the heavier object pulling the
lighter one down and the lighter one slowing down the heavier object. But,
of course, since the system considered as a whole is certainly heavier than
either one of the objects considered on their own, the system as a whole
should fall even faster. This produces a contradiction and should lead us
to conclude that one of our assumptions must be false. Now, this particu-
lar thought experiment did supposedly lead to a real one, where Galileo
dropped two balls of different masses from the Leaning Tower of Pisa in
order to demonstrate that they would fall at the same rate instead of at a
rate depending on their relative masses. This was to prove Galileo’s own
theory while falsifying Aristotle’s.

David Atkinson, ‘Experiments and Thought Experiments in Natural Science,’ in
M. C. Galavotti (ed.), Observation and Experiment in the Natural and Social Sciences, Boston
Studies in the Philosophy of Science 232 (Dordrecht: Kluwer, 2003), pp. 209–225.
196 Intuitions and thought experiments in metaphysics

Without dwelling on the historical details too much, Atkinson’s com-

plaint concerning Galileo’s thought experiment is that there is in fact
nothing inconsistent in Aristotle’s srcinal idea, contrary to what Galileo
claimed. Aristotelian theory suggests that the time it takes for a body to
fall is inversely proportional to its weight and this theory does hold some
truth when the body is falling in a liquid, such as water. So Atkinson sug-
gests that Galileo perhaps misread Aristotle and, moreover, presented
his thought experiment of the imagined inconsistency as a polemical
device. Whether or not this judgement is correct, we do know that even if
Aristotle’s reasoning was consistent, his account of gravity is nevertheless
unsatisfactory. Furthermore, this hardly tells us anything about the actual
process by which Galileo reached his conclusion about falling bodies. Thus,
even though Galileo’s thought experiment, as we know it, might not quite
establish everything that Galileo thought it did, it is nevertheless able to
rule out one possible way to conceive of gravity. In this regard, one might
think that Galileo did indeed produce a fine thought experiment, even if it
is based on the assumption that the medium is air rather than, say, water.
We might compare the situation with Newtonian mechanics: we know
that it breaks down in situations where the speed of light is approached,
for instance. But this obviously does not reduce the value of Newtonian
mechanics in more limited contexts, for the theory successfully rules out
certain possibilities regarding motion (e.g., a body cannot accelerate unless
there is a force acting upon it).
Consider another famous thought experiment in science, which
Atkinson considers to be a good one: the Einstein–Podolsky–Rosen (EPR)
thought experiment. The EPR thought experiment attempts to explain
away the ‘spooky action at a distance’ phenomenon (as Einstein called it)
of quantum entanglement. If we measure, say, the spin of an electron in a
quantum system which consists of two electrons travelling to different
directions, this apparently has an immediate effect on the other electron in
the system, although the two electrons are seemingly independent of each
other and could even be miles apart; thus the ‘spooky action at a distance’.
Einstein, Podolsky, and Rosen explained the phenomenon by introducing
so-called ‘hidden variables’: there must be something more to reality than
the standard quantum theory suggests, and this accounts for the strange
results. Now, the EPR thought experiment did lead to real experiments, but
its conclusion is in fact incorrect. John Bell’s later work is considered to
8.6 Philosophical thought experiments 197

have corroborated the so-called ‘Copenhagen interpretation’ of quantum

mechanics, which does not postulate hidden variables. The details are his-
tory, but the value of the EPR thought experiment would seem to be that
it produced a testable prediction – something that philosophical thought
experiments almost never do. On a related point, Atkinson claims that
string theory (or its modern version, ‘supersymmetry’) – a complex theo-
retical framework attempting to explain the observed elementary particles
in terms of the quantum states of one-dimensional ‘strings’ – is an exam-
ple of a bad thought experiment. The reason for this is the most common
complaint against string theory, namely that it seems that we can never
achieve the energy required to test the theory empirically, and hence it
will not lead to empirical experiments. It’s not entirely clear that this is
true, since some attempts to derive testable predictions from string theory
have been made, but we can leave this aside here. Let us instead consider
some more general indications of a bad thought experiment – in particu-
lar, can philosophical thought experiments ever overcome their obvious

8.6 Philosophical thought experiments

Since the previous section began with Atkinson’s analysis of scientific

thought experiments, we could begin this section by considering the two
indicators which, according to Atkinson and Jeanne Peijnenburg, reveal
that a thought experiment is a bad one – their prime examples are fam-
ous philosophical thought experiments.33 These indicators are contradictory
conclusions and conclusions which beg the question. As an example of the first
one, Peijnenburg and Atkinson mention the Doppelgänger thought experi-
ment which produced a heated debate in the philosophy of mind. Your
Doppelgänger is a molecule-for-molecule physical duplicate of you, pro-
duced, say, by a freak chemical reaction generated by a lightning strike and
swamp gas.34 The question is of course whether your physical duplicate is
also mentally identical to you. Opinions are divided, with one side insisting

Jeanne Peijnenburg and David Atkinson, ‘When Are Thought Experiments Poor Ones?’,
Journal for General Philosophy of Science 34.2 (2003), pp. 305–322.
This ‘swampman’ thought experiment srcinates in Donald Davidson, ‘Knowing One’s
Own Mind,’ Proceedings and Addresses of the American Philosophical Association 60.3 (1987),
pp. 441–458.
198 Intuitions and thought experiments in metaphysics

that of course the duplicate is also mentally identical and the other side
claiming that something would be missing – perhaps the duplicate would
be a phenomenological ‘zombie’.35 These two very different intuitive reac-
tions are of course untestable, except perhaps by another thought experi-
ment. The upshot is that the Doppelgänger thought experiment produces
irredeemably contradictory conclusions, even among experts.
In fact, the same thought experiment serves to illustrate the other
indication of a bad thought experiment, according to Peijnenburg and
Atkinson. They claim that the contradictory conclusions in this case are
caused by question-begging premises: the thought experiment is meant to
explain our intuitions about the mental and the physical, but these intui-
tions are also the cause of the contradictory conclusions. The worry, then, is
that the thought experiment relies on the very intuitions that it is designed
to produce and it certainly gives us a limited idea about the methodology
behind thought experiments. Take once again the EPR thought experi-
ment, which apparently did not make the correct prediction, although it
produced a real experiment (quite a bit after the actual thought experi-
ment was introduced). It seems thus that the EPR thought experiment
was good only because of the contingent fact that Bell happened to find a
way to test it empirically. One question that emerges is how long we are
supposed to wait for a potential empirical experiment before we deem a
thought experiment to be a bad one? This is a concern that Daniel Cohnitz
has also raised.36 Indeed, unless we can define what constitutes a thought
experiment, we might not even recognize the work that thought experi-
ments do (in science and philosophy). Peijnenburg and Atkinson refuse to
attempt this:

Since we are preoccupied with the difference between good and bad,
we do not feel the need to state exactly what thought experiments are;
after all one can distinguish good from bad theories, or thoughts, or
experiments without being able to define what exactly theories, thoughts
or experiments are.37

See David Chalmers, The Conscious Mind (Oxford University Press, 1996), p. 95.
Daniel Cohnitz, ‘When are Discussions of Thought Experiments Poor Ones?
A Comment on Peijnenburg and Atkinson,’ Journal for General Philosophy of Science 37.2
(2006), pp. 373–392.
Peijnenburg and Atkinson, ‘When Are Thought Experiments Poor Ones?’, p. 306.
8.6 Philosophical thought experiments 199

Yet, one might think that the value of a thought experiment, its ‘goodness’
or ‘badness’, is closely tied to what thought experiments are. If we set aside
the purely pragmatic criterion – that a thought experiment is only valu-
able if it produces a real experiment – then we ought to try to make some
progress in this regard. Note that even if we were to look for a purely prag-
matic criterion, philosophical thought experiments will clearly require a
different criterion than the one proposed by Peijnenburg and Atkinson.
As a starting point, we may derive a hint from our earlier discussion: it
looks as if a thought experiment can be valuable while failing to corre-
spond with actual reality; that is, thought experiments by themselves do
not need to be a reliable guide towards how things are in the actual world.
To put this in terms of an example, just take the EPR thought experi-
ment: although the thought experiment is perfectly consistent, it turned
out not to correspond with actuality (since it was refuted by Bell’s experi-
ments). Naturally, thought experiments that do not correspond with the
actual world in the relevant sense might not be very interesting, at least
beyond the purely pragmatic value of urging someone to falsify them
experimentally. But the goal of philosophical thought experiments is
clearly different – it would seem that it is enough if the thought experi-
ment describes a (metaphysically) possible scenario. Now, it should be
immediately noted that one area of debate with regard to many philo-
sophical thought experiments, such as the one involving Döppelgangers, is
exactly whether they are possible – or indeed even conceivable . Setting this
issue aside and assuming that most philosophical thought experiments
do succeed in describing a possible scenario, it remains a separate issue
whether this scenario is true, or in other words, corresponds with actual-
ity. Insofar as a thought experiment succeeds in this regard, it is typically
left to intuitions or further argument to settle the truth of the matter (but
we have of course discussed various problems regarding the use of intui-
tions as evidence). In any case, the suggestion that thought experiments
deal with possibility brings us back towards the discussion of Chapter 7,
namely modal epistemology.
Since we have discussed modal epistemology in great detail already, we
do not need an in-depth discussion here. But it may be worth noting that
mere conceivability is unlikely to be enough. The reason for this was dis-
cussed in Chapter 7, but can be found already in Roy Sorensen’s classic
account of thought experiments: ‘We may open a modal inquiry with a
200 Intuitions and thought experiments in metaphysics

casual appeal to what we can imagine, but we cannot close it’.38 In other
words, room for error always remains. But even if we sometimes conceive
of impossibilities, Sorensen suggests that more often than not we neverthe-
less manage to conceive of a possible scenario. Conceivability presumably
does capture various (metaphysical) possibilities and it is plausible that all
possibilities are, at least in principle, conceivable. The sceptic may com-
plain that since there is no clear way to ensure that something we conceive
is genuinely possible, this overlap can be regarded as worthless. But as
Sorensen notes, drawing an analogy with memory, this type of scepticism
may be too extreme to be an interesting thesis (it is near self-refuting).39
If conceivability is not enough, what are our options? An appeal to
conceptual knowledge or conceptual analysis might be tried, and often is.
This is the approach that Frank Jackson takes. 40 Jackson is specifically
interested in philosophical thought experiments, such as Putnam’s Twin
Earth thought experiment, but he does consider applying the concep-
tual framework to scientific thought experiments as well. For instance,
he considers Galileo’s thought experiment about falling bodies in com-
parison to the Twin Earth thought experiment and draws the following

We should not be too surprised at thought experiments revealing facts

about the empirical world. Detective stories make us familiar with the idea
that reconstructing ‘in our minds’ what would have been involved in the
butler doing it may reveal that he could not have done it. This is surely
very different from the Twin Earth thought experiments. They do not lead
us to revise our views about what Earth is like, or indeed what Twin Earth
is fundamentally like.41

We do not need to return to the Twin Earth thought experiment here.

Instead, we could question Jackson’s distinction between philosoph-
ical and scientific thought experiments. Consider one of the most fam-
ous philosophical thought experiments, from Jackson himself: the case

Roy Sorensen, Thought Experiments (Oxford University Press, 1992), p. 41.
For further discussion on conceivability, see Marcello Oreste Fiocco, ‘Conceivability,
Imagination and Modal Knowledge,’ Philosophy and Phenomenological Research 74 (2007),
pp. 364–380.
Frank Jackson, From Metaphysics to Ethics: A Defence of Conceptual Analysis (Oxford: Clarendon
Press, 1998).
Jackson, From Metaphysics to Ethics, pp. 78–79.
8.6 Philosophical thought experiments 201

of Mary, the colour scientist. 42 As we will recall, Mary is confined into a

black-and-white-room and learns every last thing about colour from books.
Does Mary learn anything new when she actually sees a new colour, say,
red? The expected intuitive reaction is to answer ‘yes’, which is supposed
to show that physicalism is false. But how are we to classify this thought
experiment, on Jackson’s criteria? According to Peijnenburg and Atkinson,
the case of Mary is an example of a bad thought experiment, comparable
to the Twin Earth thought experiment.43 However, according to Jackson’s
criteria, it would seem that the Mary thought experiment is closer to a sci-
entific thought experiment, as it is supposed to show that physicalism is
false and thus it clearly tells us something about the world – an indicator
of a scientific thought experiment in Jackson’s terms (never mind the fact
that this is a very controversial result!). Or take Sorensen, who considers the
Mary thought experiment to be a counterexample to the view that thought
experiments are appeals to ordinary language – Jackson’s own thought
experiment would then seem to speak against the conceptual approach
to philosophical thought experiments, just like his reaction to the Twin
Earth thought experiment.44 All this suggests that the distinction between
scientific and philosophical thought experiments may break down when
we look into the details.
Let’s take a step back. Is there anything that is obviously shared between
philosophical and scientific thought experiments? If we consider the exam-
ples that we have discussed, there is at least one thing that surfaces: thought
experiments in both disciplines would seem to make modal commitments.
When we construct thought experiments we are, quite generally, inter-
ested in different possible scenarios which are somehow constrained. The
constraint is typically expressed in terms of what could explain the current
empirical data in such a way that it fits a certain theoretical paradigm (such
as in the EPR thought experiment) or in terms of the existing theoretical
paradigm when we encounter novel empirical data, such as in the Twin
Earth thought experiment. The EPR thought experiment attempts to main-
tain the key elements of the classical framework and the intuitions that
support it (such as locality) in the face of quantum entanglement and the

Frank Jackson, ‘What Mary Didn’t Know,’ The Journal of Philosophy 83 (1986),
pp. 291–295.
Peijnenburg and Atkinson, ‘When Are Thought Experiments Poor Ones?’, pp. 309–310.
Sorensen, Thought Experiments, p. 94.
202 Intuitions and thought experiments in metaphysics

non-classical characteristics involved in the process of measuring quantum

states (such as spin). In contrast, the Twin Earth thought experiment oper-
ates within the theoretical framework in which chemical substances are
considered to be defined by their microstructure and tests our intuitions
in a scenario where novel empirical information challenges this frame-
work (by introducing XYZ, which behaves like water even though it has a
different microstructure). There are of course various more complicated
elements present in both thought experiments, but both of them clearly
require a modal element – indeed one might say that the modal element is
the very core of both thought experiments.
It is another question entirely how we delimit the space of possible sce-
narios in such a way that it enables us to focus on interesting thought
experiments instead of blatantly silly ones (e.g., those involving magic, as
in Disney fairy tales, etc.). It is obviously not enough to insist that the sce-
nario should be internally consistent, since even many fantastical thought
experiments manage to do that without having any scientific or philo-
sophical use or plausibility (although the line of demarcation is admittedly
vague). These are questions that we have considered, to some extent, in
Chapter 7, but more work in this area remains to be done. Unfortunately,
we cannot finish the work here. Instead, we will move into a more in-depth
discussion regarding the relationship between science and metaphysics,
now that we have found at least one interesting area of potential overlap.
Another tool that may provide some further insight in this regard is the
process of modelling, which we will discuss in Chapter 9.
9 Demarcating metaphysics
and science: can metaphysics
be naturalized?

Debate concerning the relationship between metaphysics and science

has been intense for a long time. It was an especially important topic in
the work of Carnap and Quine, given the hostility of the Vienna Circle
and logical positivism/empiricism more generally towards metaphysics. 1
However, it should be noted that what the logical positivists meant by
‘metaphysics’ is quite far removed from the type of analytic metaphysics
that we have been concerned with in this book. Here’s a succinct example
from Carnap:

Logical analysis, then, pronounces the verdict of meaninglessness on any

alleged knowledge that pretends to reach above or behind experience. This
verdict hits, in the first place, any speculative metaphysics, any alleged
knowledge by pure thinking or by pure intuition that pretends to be able
to do without experience. But the verdict equally applies to the kind of
metaphysics which, starting from experience, wants to acquire knowledge
about that which transcends experience by means of special inferences […].2

We can see that the type of metaphysics Carnap has in mind does not align
very neatly with the difficulty of distinguishing a priori and a posteriori ele-
ments discussed in Chapter 7, nor with the specification of intuition in
Chapter 8. Indeed, in the paper quoted above, Carnap discusses authors
such as Hegel and in the 1957 ‘Remarks by the author’ added to the English
translation he explicitly adds that he had in mind the systems of Fichte,

1 For an overview, see Richard Creath, ‘Logical Empiricism,’ in E. N. Zalta (ed.), The
Stanford Encyclopedia of Philosophy (Spring 2014 edn); see
2 Rudolf Carnap, ‘The Elimination of Metaphysics Through Logical Analysis of Language,’
trans. by Arthur Pap, in A. J. Ayer (ed.), Logical Positivism (New York: The Free Press, 1959),
p. 76; srcinally published in Erkenntnis 2.1 (1931) as ‘Überwindung der Metaphysik
durch logische Analyse der Sprache,’ pp. 219–241.
204 Demarcating metaphysics and science: can metaphysics be naturalized?

Schelling, Hegel, Bergson, and Heidegger. So, while it’s unlikely that the
logical positivists would be entirely happy with contemporary analytic
metaphysics, the primary target of their criticism was metaphysics of a
quite different type.
The background of the movement was the departure of various sub-
disciplines from the remit of philosophy. Since by the beginning of the
twentieth century there were already established subdisciplines of phys-
ics, mathematics, biology, psychology, and so on, it was not clear what
was left for philosophy to do. One perhaps fairly understandable reac-
tion to this was that all that’s left is ‘metaphysics’, something completely
non-empirical, based on a privileged access to some quite different type
of evidence than that to which the empirical sciences appeal. This led to
the mystification of philosophy, which now seemed quite distant from the
various empirical sciences that went from triumph to triumph. The logical
positivists, then, were extremely suspicious of the type of a priori reasoning
that would clearly have to be the basis of anything that was left for philoso-
phy to study. Of course, as we have seen in the previous chapters, the situ-
ation is not quite as simple as that, even if the strict demarcation between
metaphysics and science were correct: there are still ways to accommodate
a posteriori methods in metaphysics.
The question about the demarcation between metaphysics and science
may not be quite as pressing as it was when the logical empiricists were
active, but this doesn’t mean that the problem has been solved. Far from
it: the development of modern physics and especially quantum mechanics
have made it clear that the world may in fact be stranger than even the
wildest metaphysical speculations have dared to suggest. As Tim Maudlin
puts it:

Empirical science has produced more astonishing suggestions about the

fundamental structure of the world than philosophers have been able to
invent, and we must attend to those suggestions.3

This generates a novel worry about the status of metaphysics: if the world
is truly as strange as science suggests, then how could metaphysics ever be
able to say anything informative about it independently of the sciences?
The likely answer is that it cannot, but this is already something that

3 Tim Maudlin, The Metaphysics Within Physics (Oxford University Press, 2007), pp. 78–79.
Demarcating metaphysics and science: can metaphysics be naturalized? 205

all or at least most contemporary metaphysicians would accept. Indeed,

it would be a mistake comparable to logical empiricism to pursue meta-
physics completely in isolation of empirical science. But some would argue,
especially in view of all the strange puzzles in physics, that this goes both
ways: science needs metaphysics as well.
In this chapter we will examine this delicate relationship between meta-
physics and science in more detail. We will start from a more classical pic-
ture: metaphysics is autonomous and able to tell us something about the
world on its own, despite the complications introduced by modern physics.
On this view, metaphysical inquiry into the fundamental structure of real-
ity can uphold scientific realism: the categories, natural kinds, and other
classifications introduced in metaphysics do, by and large, ‘carve at the
joints’. In the second section we will move to the other extreme: fully nat-
uralistic metaphysics. According to this view, metaphysics cannot hope to
say anything about reality independently of science. We will lay out this
view in its most extreme form, which effectively becomes eliminativist: the
only possible metaphysics is really just science. This type of view is perhaps
closest to logical empiricism, but attempts to avoid some of its caveats. The
third section continues in the same vein, but introduces further qualifica-
tions regarding the relationship between science and metaphysics. Here,
the focus is especially on fundamental physics. We will examine two prin-
ciples, which have recently been broadly discussed in the literature: the
Principle of Naturalistic Closure and the Primacy of Physics.4 The upshot of these
principles is that metaphysics still has a role to play, but it is severely lim-
ited, primarily unificatory. We will also critically assess these principles.
The fourth section examines the possibility of building a methodological
bridge between science and metaphysics: even if the subject matter of the two
disciplines is distinct, perhaps there are some similarities in their method?
Finally, in the fifth section, a more modest, reconciliatory view of the rela-
tionship between science and metaphysics will be proposed: ‘moderately
naturalistic metaphysics’. According to this view, metaphysics does have
a fairly autonomous position, but it must take heed of the latest results in
science. Similarly, science can benefit from metaphysical analysis, since

4 These principles srcinate in J. Ladyman and D. Ross (with D. Spurrett and J. Collier),
Every Thing Must Go (Oxford University Press, 2007).
206 Demarcating metaphysics and science: can metaphysics be naturalized?

scientific theories must be interpreted, and this interpretation can be seen

partly as the task of metaphysics.

9.1 Autonomous metaphysics

There are plenty of metaphysicians who still pursue metaphysical topics

with little or no reference to scientific work. This may be more justifiable
in some areas of metaphysics than in others, but it does no doubt require
some justification, given that metaphysics supposedly deals with the struc-
ture of reality and science arguably gives us the best tools to gain accurate
knowledge about reality. One strategy is to focus on the role of metaphysics
as conceptual analysis. On this view, metaphysics is concerned with clarifying
certain important notions and primarily based on linguistic considerations.
Metaphysics would thus be, by and large, an a priori discipline employing
armchair methods, and hence largely independent of a posteriori elements.
There is more to be said about the relationship between conceptual analysis
and science – for instance, if metaphysics analyses scientific concepts, then
there would seem to be an important area of overlap between the disci-
plines. It could also be argued that, rather than being autonomous, meta-
physics conceived as conceptual analysis is derivative: science defines the
relevant concepts, metaphysics simply clarifies them. We have discussed
this type of approach in passing, in Chapters 7 and 8, but we will leave it
aside here. Instead, we will examine a more traditional approach attempt-
ing to secure the autonomy of metaphysics from science – perhaps the most
traditional. The roots of this approach go back to Aristotle:

There is a science which investigates being as being and the attributes

which belong to this in virtue of its own nature. Now this is not the same
as any of the so-called special sciences; for none of these others deals
generally with being as being. They cut off a part of being and investigate
the attributes of this part – this is what mathematical sciences for instance
do. Now since we are seeking the first principles and the highest causes,
clearly there must be some thing to which these belong in virtue of its own

5 Aristotle, Metaphysics, trans. W. D. Ross, revised by J. Barnes (Princeton University Press,

1984), Bekker page numbers 1003a22–28.
9.1 Autonomous metaphysics 207

In Aristotle’s time, mathematics had already been demarcated from philos-

ophy, at least to a certain extent. Aristotle himself was of course a ‘natural
philosopher’ in its true sense; he also engaged in some empirical research
(although it is debatable as to how much) and wrote about biology, phys-
ics, and many other ‘special sciences’. But it is metaphysics that Aristotle
considered to be the most general science. Thus, Aristotelian metaphysics
is the study of being as it is in itself (being qua being), whereas the special
sciences investigate only a part of that being.
It may seem strange to start a discussion about the relationship between
metaphysics and science from Aristotle, but it is useful to keep in mind
that philosophy and science were not always as obviously distinct as they
may now appear to be. To a certain extent, the contemporary defence of
various forms of ‘naturalism’ is an attempt to get back towards a situation
not unlike the one seen in ancient philosophy, where natural science was
still more or less a part of philosophy. No one, or hardly anyone, wants to
defend a conception of philosophy or metaphysics that is non-naturalistic –
that would almost be like saying that it is magic! For instance, even many
contemporary proponents of dualism might consider their view to be per-
fectly compatible with (some forms of) naturalism. One could say there
has been a certain inflation of the term ‘naturalism’. Accordingly, we will
avoid using the term to define views, except when it is qualified further. 6
The closest that we can get to a definition of ‘naturalism’ that takes into
account of all the different uses is perhaps something like ‘scientifically
minded’, or ‘scientifically respectful’. Or even just a negative thesis: ‘not
directly incompatible with contemporary science’. It’s clear that some-
thing as vague as this is not particularly helpful in demarcating different
But let us return to ‘autonomous metaphysics’. Note that we can
now say that if metaphysics is autonomous, this does not entail that it
is non-naturalistic. What it does entail is that the subject matter of met-
aphysics must be, to a certain extent, different from that of natural sci-
ence. But different in what sense? If we follow Aristotle’s cue, it would
appear that metaphysics is the science that examines being ‘in virtue of
its own nature’. So rather than focusing on any single detail of ‘being’ – or

6 For discussion of various forms of naturalism in philosophy, see Jack Ritchie,

Understanding Naturalism (Stocksfield: Acumen, 2008).
208 Demarcating metaphysics and science: can metaphysics be naturalized?

reality – autonomous metaphysics studies it as a whole. How is this differ-

ent from fundamental physics? Take the Standard Model of particle physics,
which lists the particles that are currently thought to be fundamental, the
‘building blocks’ of reality. Fundamental physics studies the properties of
these particles, the relations between them, and the fundamental forces
that interact with them. But this is not, of course, quite what metaphysi-
cians are doing – certainly not if metaphysics is taken to be completely
autonomous from science. Instead, metaphysics in the present sense may be
conceived as studying reality in an even more fundamental sense than funda-
mental physics does. This type of metaphysics is not interested in listing the
various fundamental particles: fermions, bosons, … Rather, it is interested
in listing the most basic categories that the fundamental ‘building blocks’
belong to. This is metaphysics of a ‘neo-Aristotelian’ ilk, the study of the
fundamental ontological categories: particulars, universals, tropes/modes,
and so on. An important part of autonomous metaphysics is to determine
how many of these fundamental ontological categories there are.7
Although the study of ontological categories is a good example of how
autonomous metaphysics can proceed, this doesn’t mean that one couldn’t
pursue such research in connection with more ‘scientific’ metaphysics. For
instance, many metaphysicians working on these questions would readily
revise their ontological picture in response to new scientific discoveries.
One example of such a revision could be the attention that monism – the
view that the world is, in some sense ‘one’ – has received in recent years.
Jonathan Schaffer, who is an industrious defender of a type of monism,
takes ‘quantum emergence’ to be a strong argument in favour of monism.8
The argument runs as follows:

(1) The whole possesses emergentproperties (due toquantum entanglement).

(2) If the whole possesses emergent properties, then the whole is prior to
its parts.
(3) Therefore, the whole is prior to its parts.

7 For instance, a defence of the view that there are four such categories can be found
in E. J. Lowe, The Four-Category Ontology (Oxford: Clarendon Press, 2006). For various
arguments to the effect that there should be fewer or more categories than four, see
the articles in Tuomas E. Tahko (ed.), Contemporary Aristotelian Metaphysics (Cambridge
University Press, 2012).
8 Jonathan Schaffer, ‘Monism: The Priority of the Whole,’ Philosophical Review 119 (2010),
pp. 31–76.
9.1 Autonomous metaphysics 209

This results in priority monism, which we discussed in Chapter 6. An assess-

ment of the argument would require looking into the relevant physics, but
all that matters for us now is that, as a matter of fact, much of what could
initially be labelled as ‘autonomous metaphysics’ is not quite as autono-
mous as may first seem, since input from contemporary science is taken
into account. But could this relation be symmetric? That is, could ‘autono-
mous’ metaphysics also influence science? This certainly seems to be a
hope that some metaphysicians have. The influence may not be seen as
direct, though, but instead as laying the foundations of scientific inquiry
more generally or in terms of structuring scientific thought in some way.
The idea is to develop a metaphysical foundation for natural science – to
explain why science is possible and indeed successful in the first place. 9
However, at this point, more ‘scientifically-minded’ philosophers tend
to get outraged: surely it is not our place as philosophers to tell scientists
how their discipline should be, or even can be pursued! But that is not
of course what even the more ambitious metaphysicians have in mind,
exactly. A more charitable reconstruction would be to say that metaphys-
ics may be considered to open up novel areas of research, perhaps by giv-
ing birth to new special sciences. This approaches the idea expressed in
the earlier Aristotle quote: metaphysics studies being in itself and in the
process of this study, new areas – parts of being – may be discovered. The
special science that studies that new part of being is no longer metaphys-
ics, but since metaphysics encompasses all of being, there is no part of
it that wouldn’t be of some relevance for metaphysics as well. There are
plenty of potential examples of philosophy giving birth to new special sci-
ences. Besides Aristotle’s work, which paved the way for physics, biology,
and other special sciences that have been established for a long time, we
could mention psychology and cognitive science as more recent examples.
The exact process by which these special sciences came to be is of course
quite complicated, but it’s quite clear that many questions being pursued
in various special sciences today were once questions that were pursued by
philosophers (or, perhaps more accurately, philosopher-scientists).
Let us conclude this section with a more specific example of how meta-
physics may be understood to be pursued autonomously of science. The

9 Again, see Lowe, The Four-Category Ontology; the subtitle of the book is A Metaphysical
Foundation for Natural Science.
210 Demarcating metaphysics and science: can metaphysics be naturalized?

case of material composition, which we have encountered repeatedly in

this book, serves as a fine example. Recall the case of a statue and the lump
of clay that composes it. According to a popular argument, even though
the statue depends on the lump of clay for its existence, the statue and the
lump are not numerically identical because they have different persistence
conditions.10 The lump of clay cannot survive the removal of a chunk of its
constituent matter, while the statue can survive the loss of some chunk of
it, such as an arm – we still regard the ancient Greek statues as statues even
when they are evidently missing some pieces, like arms and noses. So the
statue can depend for its existence – by metaphysical necessity – on the
lump, without the statue and the lump being identical. A metaphysician
might analyse the situation as follows: A statue is essentially statue-shaped
because it falls under the sortal statue. So the statue cannot survive, say,
being remoulded into a pot. This suggests that objects such as a lump of
clay and a statue can share all their material properties and their space-time
region. But the ‘common sense’ intuition is that two things with the same
properties in the same place are surely identical. The metaphysician is then
faced with the task of reconciling common sense – which a great many
philosophers wish to vindicate – with the metaphysical upshot of the above
None of this, of course, has much to do with science, but is rather based
on common-sense intuitions and a priori reasoning (cf. Chapter 8 for fur-
ther discussion of intuitions and a priori inquiry). In the teeth of those
philosophers who are more sceptical about autonomous metaphysics, this
type of philosophizing struggles to earn merit:

[Autonomous] metaphysics proceeds by attempts to construct theories that

are intuitive, common-sensical, palatable, and philosophically respectable.

The criteria of adequacy for metaphysical systems have clearly come apart
from anything to do with the truth. Rather they are internal and peculiar
to philosophy, they are semi-aesthetic, and they have more in common
with the virtues of story-writing than with science.11

This statement is obviously polemical, but it does raise a valid con-

cern: without a clearer picture of what counts as a good metaphysical
10 See for instance L. A. Paul, ‘The Context of Essence,’ Australasian Journal of Philosophy 82
(2004), pp. 170–184.
11 Ladyman and Ross, Every Thing Must Go, p. 13.
9.2 Fully naturalistic metaphysics 211

system, autonomous metaphysics may lose its claim to truth. Attempts to

establish better criteria for assessing metaphysical systems have of course
been made, and we have already discussed some of these attempts, such as
the nature of a priori inquiry. But it might seem easier and more sensible to
just give up on a completely autonomous metaphysics and instead attempt
to ‘naturalize’ metaphysics. Let us now give the stage to those who think
that this is the only way to salvage anything of metaphysics – if the result-
ing ‘scientism’ is to be even called ‘metaphysics’.

9.2 Fully naturalistic metaphysics

The view that we will now consider could be considered to go as far as

rejecting metaphysics altogether. The most influential defence of this
type of ‘scientism’ is no doubt Bas van Fraassen’s, so we will take his
view as a prominent example. 12 Van Fraassen calls his view ‘the empiri-
cal stance’. The term ‘scientism’ is often used in a pejorative sense,
but in recent literature it has been adopted by proponents of a more
scientifically-minded metaphysics. 13 The core of van Fraassen’s approach
can be summarized with the following three challenges to traditional
metaphysics. First, van Fraassen argues that th e remoteness of metaphys-
ical questions from empirical considerations makes them useless, even
if not quite meaningless in Carnap’s sense. Science und ergoes a constant
testing and even falsification, but metaphysics claims to seek the truth.
However, metaphysics is arguably never able to reach the position of
establishing whether anything it says is actually true or false. Empirical
science can justify and ground its own method in virtue of its practi-
cal relevance. Metaphysics, instead, turns out to be a merely formal or
conceptual exercise, without practical or scientific relevance. This is the
upshot of van Fraassen’s first critical point.
Secondly, van Fraassen argues that metaphysical questions are
irredeemably context-dependent and hence lack well-defined ‘answering
strategies’. One example of this is the question ‘Does the world exist?’,
which van Fraassen dismisses as extremely simple, pointing out that

12 Bas C. van Fraassen, The Empirical Stance (New Haven, CT: Yale University Press, 2002).
13 Ladyman and Ross, Every Thing Must Go, Ch. 1.
212 Demarcating metaphysics and science: can metaphysics be naturalized?

metaphysicians have nevertheless spent a lot of time on it. This type of

epistemic challenge is more or less what the previous chapter was attempt-
ing to address.
Thirdly, van Fraassen objects that metaphysics accounts for ‘what
we initially understand [in terms of …] something hardly anyone under-
stands’. 14 This undermines the explanatory value of metaphysics; the
contrast is especially striking when compared to science. The upshot is
thoroughly eliminative : we really ought to eliminate it from the discourse
Note that all this is coming from a scientific anti-realist . Van Fraassen’s
position is that while science may seek the truth, its primary goal is to
establish results of some practical or instrumental importance – some-
thing that will be of use. In other wor ds, in science it is not a catastrophic
result if a theory turns out to be false, as long as it was of some use. And
simply the fact that a scientific theory survived to the stage where it
got to be falsified suggests that it has some value, some heuristic use.
Consider Newton’s impor tant work in physics. We know that Newtonian
physics is not exactly true, and it produced (and in fact continues to pro-
duce!) many false beliefs. But there seems to be no question about the
fact that Newton’s work (in physics, but presumably not in alchemy)
was extremely useful if not necessary for scientific progress. By contrast,
many metaphysical theories do not even have the merit of being falsifi-
able , so it turns out that metaphysics is completely detached from the
scientific model.
A metaphysician, it seems, typically cannot claim any pragmatic or cogni-
tive utility for her falsified theory; although van Fraassen himself emphasizes
empirical adequacyrather than pragmatic utility:this is a requirement for a
certain type of isomorphism between what a theory says about observable
phenomena and the model (a family of structures) that the theory describes.
Van Fraassen points out that the failure of metaphysics to be properly falsifi-
able is particularly problematic for those contemporary metaphysicians who
strive to uphold some version of naturalism (which is most of them). But does
van Fraassen think that there is anything left for philosophy to do after this
devastating deconstruction? If there is, one thing is clear:philosophy should
not be based on dogma. Instead, van Fraassen thinks that philosophy, or

14 Van Fraassen, The Empirical Stance, p. 3.

9.2 Fully naturalistic metaphysics 213

philosophical views, should be conceived asstances: ‘A philosophical position

can consist in a stance (attitude, commitment, approach, a cluster of such–
possibly including some propositional attitudes suchas beliefs as well)’.15
We can see that to conceive of philosophical views as stances is not easily
combined with the type of metaphysical realism that we have assumed to
be the default position throughout much of this book. Or, to put it another
way, van Fraassen’s proposal would lead us to reject what James Ladyman
and Don Ross call ‘strong metaphysics’, which is more or less the type of
metaphysics that all non-deflationary metaphysicians pursue.16 This is
because a stance appears to have a strongly subjective element; it is an atti-
tude rather than a doctrine. We will look into the ‘weak metaphysics’ that
Ladyman and Ross propose themselves in a moment, but let us first give
van Fraassen a chance to persuade us to ‘convert’ to his preferred stance, the
empirical stance. One might think that ‘convert’ is inappropriate here, sug-
gesting that the view is more like a religion. But in fact this is exactly how
van Fraassen puts it: ‘Being or becoming an empiricist will then be similar
or analogous to conversion to a cause, a religion, an ideology, to capitalism
or to socialism, to a worldview’.17 Given this, it’s clear that van Fraassen
does not aim to lay out the empirical stance in terms of philosophical argu-
ments like the ones we have been discussing so far. Instead, the empirical
stance is primarily defined by a negative attitude: rule out a dogmatic atti-
tude, reject explanation for the sake of explanation – that is, where no test-
able, (empirical) evidence-based explanation is forthcoming – and do not
postulate explanations just for the sake of psychological satisfaction.
While this may seem like a healthy attitude, it does bear a risk: if we do
not postulate anything, then we may lack the motivation to develop new
empirical methods as well. In other words, sometimes postulating even a
false explanation could have a theoretical utility. The positive part of the
empirical stance is a simple endorsement of scientific virtues, namely that
we should take a view only as seriously as evidence warrants. Let us pause
here for a moment, for many readers will no doubt be familiar with exam-
ples from science which are not fully compatible with the empirical stance.
A good example is string theory:

15 Ibid., pp. 47–48.

16 Ladyman and Ross, Every Thing Must Go, p. 60.
17 Van Fraassen, The Empirical Stance, p. 61.
214 Demarcating metaphysics and science: can metaphysics be naturalized?

Depending on who one listens to, string theory has either already led us
a considerable distance down the road to a complete theory of quantum
gravity, or it has achieved absolutely nothing that counts as physics rather
than mathematics.18

The basic problem with string theory is that it has made little in terms of test-
able predictions – exactly the same issue that people like van Fraassen find
problematic when it comes metaphysics. Now, one reaction to this is to con-
sider string theory just as ‘bad’ as ‘strong metaphysics’. Indeed, the physicist
Lee Smolin reacts against string theory as strongly as van Fraassen does against
metaphysics; Smolin is especially troubled by the fact that string theory has
received vast amounts of funding in the last few decades, at the cost of leav-
ing research on alternative theories largely unfunded. 19 We do not need to

dwell on the politics of thesituation, but it is worth noting that thecelebrated

empirical sciences are apparently at least sometimes subject to problems simi-
lar to those encountered in metaphysics, according to van Fraassen.
How could a metaphysician with less eliminative tendencies reply to
van Fraassen’s challenge? Recall that one of van Fraassen’s points is that
scientific theories are ‘initially understood’. This flies in the face of a point
that was briefly entertained in Chapter 7: a scientific theory cannot be
‘understood’ unless it is interpreted, and the metaphysician may insist that
interpretation requires tools coming from outside of science: the interpre-
tation of any theory T cannot be provided within T, for otherwise it should
itself be interpreted, giving rise to a vicious regress (here T can be consid-
ered science as a whole, on a par to the empirical stance). When it comes
to areas like quantum mechanics, even most physicists would acknowl-
edge that philosophical questions cannot be avoided in the interpretation
of the relevant theories. Before those tools are applied, at most one has
the sort of instrumental ability that we saw van Fraassen to push earlier.
Indeed, these instrumental abilities may be satisfactory to scientists insofar
as their immediate goals are considered (to produce practical applications,

18 Ladyman and Ross, Every Thing Must Go, p. 168.

19 For a popular account, see Lee Smolin, The Trouble with Physics (London: Penguin
Books, 2006).
20 For further discussion, see Craig Callender, ‘Philosophy of Science and Metaphysics,’
in S. French and J. Saatsi (eds.), The Continuum Companion to the Philosophy of Science
(London: Continuum, 2011), pp. 33–54. One good example might be the ’many worlds’
9.2 Fully naturalistic metaphysics 215

to formulate and test theories), but cannot be satisfactory insofar as science

must be understood as mapping the way the world is.
Now, if metaphysics turns out to be necessary for interpreting scien-
tific theories, it also appears plausible that the concepts and categories
typical of metaphysics – the notions contained in the metaphysical vocab-
ulary that we use for interpreting scientific theories – are not (necessar-
ily) obscure. As a matter of fact, since metaphysical vocabulary has been
and is used to interpret scientific theories quite successfully, the argument
from obscurity could be resisted with reference to examples, such as, say,
the significant philosophical work that has gone into interpreting general
relativity.21 Moreover, metaphysical analysis appears in this respect closer
to the common-sense approach than its scientific counterpart. For instance,
is the notion of a universal, say, any more obscure than that of a Higgs boson –
taking into account the fact that the recent ‘discovery’ of the boson in ques-
tion relies on certain somewhat arbitrary criteria for ‘discovery’, commonly
used in particle physics?22 The answer is by no means obviously affirmative
unless one equates clarity with measurability – measurability to a certain
degree, that is. But, again, this is not what one normally intends by ‘under-
standability’ and ‘clarity’ – not even in an openly anti-realist context such
as van Fraassen’s. If concepts such as substance, relation, property, and the
like are at least no less understandable than scientific concepts – and are
in fact employed to interpret scientific theories – then van Fraassen’s chal-
lenge could perhaps be addressed.

interpretation of quantum mechanics, see Simon Saunders, Jonathan Barrett, Adrian

Kent, and David Wallace (eds.), Many Worlds? Everett, Quantum Theory, and Reality (Oxford
University Press, 2010).
21 For a survey, see Thomas A. Ryckman, ‘Early Philosophical Interpretations of General
Relativity,’ in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Spring 2014 edn);
22 Particle physics has an accepted definition for what is sufficient to claim a discov-
ery: 5.0 sigma significance, i.e. a level of certainty up to five standard deviations.
In statistical terms, this means a probability of less than one in a million that the
observed phenomenon was produced by something other than the postulated Higgs,
namely statistical fluctuation. The reason for this type of talk is of course that the
Higgs cannot be observed directly. Rather, we observe decay products, such as pho-
tons, which could be produced by a number of phenomena that have to be ruled out.
This also makes it quite clear that there is always an aspect of fallibility in these types
of results.
216 Demarcating metaphysics and science: can metaphysics be naturalized?

Note that the type of metaphysics that van Fraassen is primarily react-
ing against is ‘Quinean’, or at least that is how he sees the project: ‘The
genuflection toward science among Quine’s heirs has all too often been
toward a naive caricature purveyed by the past generation of philosophers
at whose knees Quine himself learned it’.23 Accordingly, he does not dis-
cuss or analyse the various more fine-grained formulations of metaphysics
and ontology that we have been pursuing in this book. In fact, a propo-
nent of ‘strong metaphysics’ might even concede van Fraassen’s point if
it is applied to those who attempt to strengthen the position of metaphys-
ics with an appeal to its continuity with science – E. J. Lowe for one cer-
tainly does concede the point. 24 Indeed, Lowe suggests that van Fraassen is
absolutely right when it comes to the assertion that metaphysicians claim
continuity with science by appealing to a shared met hodology, namely infer-
ence to the best explanation. Van Fraassen’s point, as we have seen, is that
metaphysics fails, catastrophically, to uphold the standards of science in
this regard. But this critique only goes through if metaphysics is indeed
conceived as continuous with science in this sense, something that Lowe
ascribes to ‘false friends’ of metaphysics:

[T]he fault lies here not with metaphysics as such, properly conceived, but
only with those of its false friends who mistakenly seek to enhance its
credit by assimilating its task to that of empirical science. Metaphysics and
empirical science are not ‘continuous’ with each other in any sense which
implies that they have the same goals and methods, or that metaphysics
is just the extension of empirical science to questions of greater generality
than any that are addressed by the so-called ‘special’ sciences. Rather,
when both are conducted fruitfully, metaphysics and empirical science
exist in a symbiotic relationship, in which each complements the other.25

The somewhat surprising upshot, then, is that metaphysics, ‘properly con-

ceived’, as Lowe puts it, can not only withstand van Fraassen’s challenge,
but can also agree that it has correctly identified a fatal flaw in one popular
approach to metaphysics. Certainly, Lowe and other proponents of ‘seri-
ous metaphysics’ owe us a story about how metaphysics and science are
related, if they’re not continuous, but in the above passage we already have

23 Van Fraassen, The Empirical Stance, p. 11.

24 E. J. Lowe, ‘The Rationality of Metaphysics,’ Synthese 178 (2011), p. 101.
25 Lowe, ‘The Rationality of Metaphysics,’ pp. 101–102.
9.3 The Principle of Naturalistic Closure and the Primacy of Physics 217

a hint about what the answer might be. We will return to this issue in the
fourth section of this chapter, but first we will consider another scientifi-
cally minded view, which attempts to provide a ‘naturalized’ conception of

9.3 The Principle of Naturalistic Closure

and the Primacy of Physics

The two principles in the title of this section come from Ladyman
and Ross. 26 They have been discussed extensively in recent years and it
would be impossible to do full justice to this discussion here. We will
instead focus on certain important aspects of the debate, relevant espe-
cially from the point of view of metametaphysics. It has already become
clear that Ladyman and Ross share some aspects of van Fraassen’s criti-
cism of metaphysics; the polemical first chapter of their book has spurred
a remarkable reaction not only from metaphysicians, but also from philos-
ophers of science. Since we ended the previous section with a quote from
Lowe, we might start this one by considering a challenge that Ladyman
and Ross pose for him and anyone else who pursues what has been pejo-
ratively labelled ‘neo-scholastic metaphysics’. Crucially, Ladyman and
Ross attempt to generalize their critique of contemporary metaphysics
beyond the broadly ‘Quinean’ picture, with which van Fraassen seems
to be primarily concerned. For instance, they attempt to challenge met-
aphysicians’ use of intuitions and a priori inquiry. However, as we saw
in Chapter 8, these issues are fairly complicated. Ladyman and Ross tar-
get one admittedly popular approach, according to which metaphysical
a priori inquiry is more or less just conceptual analysis. They appear to
suggest that Lowe and indeed most contemporary metaphysicians adopt
‘the familiar methodology of reflecting on our concepts (conceptual
analysis)’ for doing metaphysics. 27 Ladyman and Ross are quite correct
to ask how conceptual analysis could possibly reveal anything about the
structure of reality, but this is (again!) a concern that many metaphysi-
cians share – Lowe among them. So perhaps this critiquing aspect of the

26 Ladyman and Ross, Every Thing Must Go, Ch. 1.

27 Ibid., p. 16.
218 Demarcating metaphysics and science: can metaphysics be naturalized?

project to naturalize metaphysics is not as fine-grained and charitable as

one might hope.
For Lowe, metaphysics is the study of possibility – the realm of a priori
inquiry is metaphysical modality. On this view, metaphysics studies what
sorts of things there are in the world (fundamentally) and how they are
related, how reality is structured. The relevant method is a priori inquiry
into what sort of things are possible, followed by a partly empirical study
regarding which of those things are possible together. Only on this basis,
argues Lowe, can we determine what is actually the case. Ladyman and
Ross are aware of this more fine-grained type of approach, but they are
suspicious of the modal element: they deny that a priori inquiry can reveal
what is metaphysically possible. Indeed, we have seen in earlier chapters
that much work remains to be done in the area of modal epistemology,
which appears to be absolutely central for the type of view under consid-
eration now. But evidently some promising work in this area does exist.
We have discussed it extensively, but Ladyman and Ross make no real
effort to engage with it. They do however point out that metaphysicians
have often erred in their modal judgements, for instance with regard
to non-Euclidean geometry, deterministic causation, non-absolute time,
and so on. This is of course true, but hardly surprising: no one has sug-
gested that modal inquiry is easy or infallible. While the infallibility of a
priori inquiry may have been a doctrine of Cartesian metaphysics, it is
most certainly not a doctrine of contemporary analytic metaphysics, as
we clearly observed in the previous chapter . Hence, it is indeed true that
metaphysicians, like scientists, make mistakes. Kant may have held that
non-Euclidean geometry is impossible, but physics soon showed that not
only is it possible, but actual. The lesson that we should take from this
is that a more charitable picture of contemporary metaphysics ought to
acknowledge the possibility of revision – not unlike in the case of empiri-
cal results, which can be revised in the light of new empirical data. But
regardless of the negative project put forward by Ladyman and Ross (as
well as many other ‘scientifically-minded’ philosophers), they also propose
an extremely interesting and influential methodological framework – we
may call it the ‘scientistic stance’, which is a label they adopt with refer-
ence to van Fraassen’s idea of stances. 28

28 Ibid., p. 64.
9.3 The Principle of Naturalistic Closure and the Primacy of Physics 219

The core of the scientistic stance is the search for unification among sci-
entific theories on the basis of physics. In order to achieve such unification,
we should finally outline the two principles mentioned in the title of this
section: the Principle of Naturalistic Closure (PNC) and the Primary of Physics
Constraint (PPC). Here are short formulations of both principles:

(PNC): Any new metaphysical claim that is to be taken seriously at time t

should be motivated by, and only by, the service it would perform, if true,
in showing how two or more specific scientific hypotheses, at least one of
which is drawn from fundamental physics, jointly explain more than the
sum of what is explained by the two hypotheses taken separately. 29

(PPC): Special science hypotheses that conflict with fundamental

physics, or such consensus as there is in fundamental physics, should be
rejected for that reason alone. Fundamental physical hypotheses are not
symmetrically hostage to the conclusions of the special sciences. This,
we claim, is a regulative principle in current science, and it should be
respected by naturalistic metaphysicians.
These principles are meant to capture the negative project – to (partly)
justify the critical remarks we’ve discussed above – as well as ground the
positive project. It is the first principle especially that has received critical
attention, but both principles have certainly been challenged. Ladyman
and Ross are careful to note that they do not wish to go as far as the logical
positivists did with their verificationism – the view that only empirically veri-
fiable statements are meaningful. The logical positivists cast metaphysics
(among other things) aside altogether and did not even attempt to apply
their principles to metaphysics, in order to salvage something from it. In
contrast, Ladyman and Ross believe that the above principles can be applied
to metaphysics, to produce a fully naturalistic picture. There is certainly
some plausibility in prioritizing science like this. This much is made clear
by the various examples involving ‘pseudo-science’ discussed by Ladyman
and Ross. Much of the strength of the ‘pseudo-science’ critique relies on the
claim that from the point of view of the layman (and perhaps the typical
philosopher) fundamental physics is commonly viewed as a network of
‘microbangings’. Indeed, talk of ‘particle’ physics may suggest that funda-
mental physics deals with solid particles and applies Newtonian mechanics

29 Ibid., p. 37. 30 Ibid., p. 44.

220 Demarcating metaphysics and science: can metaphysics be naturalized?

to them, but this is quite far from reality. Nature does not play billiards!
But the question is whether this is enough to fully motivate principles like
(PNC) and (PPC).
Take (PNC) first. It would seem to reduce metaphysics merely to the task
of unifying scientific hypotheses. Many metaphysicians would think (and
have thought) that this misconstrues the methodology of metaphysics. For
instance, in a review of Ladyman and Ross’s book, Cian Dorr puts it as

The whole approach of Every Thing Must Go reflects an exaggerated sense

of the importance of argument in metaphysics, and a corresponding
underestimation of the difficulty of merely crafting a view coherent and
explicit enough for arguments to get any grip.31

The question arises: is it appropriate to evaluate metaphysical argumenta-

tion with the standards of natural science, as Ladyman and Ross appear to
do? Insofar as the methodologies of science and metaphysics are different,
it would seem to be inappropriate to apply the same standards to them.
Moreover, there are specific metaphysical questions which do not seem
to be hostage to science in the manner suggested by (PNC). For instance,
Katherine Hawley has argued that questions regarding the philosophy of
time can hardly be settled purely on the basis of physics; her case study is
the relationship between the special theory of relativity (STR) and presentism,
the view that only what is present exists. 32
According to STR, there are different frames of reference, which influence
our answer when faced with questions about the simultaneity of two dis-
tant events. But because STR does not privilege any one frame of refer-
ence over another, it also does not dictate whether or not two events occur
simultaneously. Since those philosophers of time who defend presentism
claim that only what is present exists, they must say that only those events
that are simultaneous with the present exist. Yet, STR suggests that this
question is open in a way in which the question of what exists is surely
not open. And here we would seem to have an automatic refutation of the
metaphysical theory of presentism based on science – a nail in the coffin of

31 Cian Dorr, ‘Review of Every Thing Must Go,’ Notre Dame Philosophical Reviews (2010); see
32 Katherine Hawley, ‘Science as a Guide to Metaphysics?’, Synthese 149.3 (2006): 451–470.
9.3 The Principle of Naturalistic Closure and the Primacy of Physics 221

autonomous metaphysics. But there are various ways in which presentists

could attempt to respond (and have responded). Some of them concern the
interpretation of STR, but there are other strategies as well. Hawley men-
tions three possible reactions:

First, we might accept that presentness is frame-dependent, accept

that existence cannot be frame-dependent, and thus reject presentism.
Second, we might accept that presentness is frame-dependent, insist
that only what is present exists, and thus conclude that existence is
frame-dependent. This second option is too relativistic (in a bad way)
for almost everyone, and may undermine the intuitions that lead
initially to presentism. Third, we might accept that existence cannot be
frame-dependent, insist on the truth of presentism, and thus conclude
that there is a privileged frame of reference, one which escapes notice
in STR. Simultaneity in that privileged frame is absolute simultaneity,
and events absolutely simultaneous with my typing now are absolutely
present. Positing a privileged frame of reference does not compel us
to adopt presentism, for we might argue that what is absolutely past
and future is also real. But the third option permits us to be presentists
without conceding that existence is frame-dependent. 33

So it appears that depending on other theoretical choices, STR is in fact

compatible with various views about the nature of time, including pres-
entism. However, it should be noted that Einstein specifically developed
STR to account for the absence of a privileged frame, so the onus would
appear to be on the presentists. Moreover, STR does, of course, entail some
constraints on what we can do in the metaphysics of time, but it does not
appear to immediately solve the metaphysical puzzle about time – there is
some room for interpretation regarding frame-dependence.
Take another example: in Chapter 6 we saw that input from physics is
surely needed if we hope to make progress regarding the debate about a
fundamental level, but what if the level is beyond the reach of all physic-
ally possible measuring devices? For instance, if some of the fundamental
particles suggested by the Standard Model do in fact have internal struc-
ture, the amount of energy needed might be so vast that it would be phys-
ically impossible to build a particle accelerator large enough to ever find
out about the constituents of these particles. Ladyman and Ross make

33 Hawley, ‘Science as a Guide to Metaphysics?’, pp. 465–466.

222 Demarcating metaphysics and science: can metaphysics be naturalized?

several claims that have the appearance of these types of metaphysical

issues, which we have been discussing throughout this book. For example,
they argue that ‘we should not interpret science – either fundamental
physics or special sciences – as metaphysically committed to the exist-
ence of self-subsistent individuals’.34 Now, this is an interesting, substan-
tive claim, but without resorting to the full metaphysical ‘tool box’ that we
partially laid out in Chapters 5 and 6, it is quite difficult to make sense of
it. Indeed, it is not even clear that this claim can be reconciled with (PNC),
for without further analysis its unificatory power remains questionable.
Hence, one might ask, how different is this from ‘neo-scholastic’ meta-
physics after all?
A final example, which we will not consider in detail: quantum mechan-
ics seems to have all sorts of implications towards the metaphysics of iden-
tity and individuation, but it may not help us to decide which metaphysical
theory of identity is ultimately the correct one.35 This is because of the pos-
sibility of underdetermination of metaphysics by physics; that is, the empir-
ical evidence we have is compatible with a range of metaphysical views
and hence does not help in deciding between them. 36 This, according to
Hawley, appears to be what has happened in the case of presentism and the
special theory of relativity.
With this spirited reaction in defence of metaphysics out of the way,
it must be granted that Ladyman and Ross have done important work to
deconstruct arguments in metaphysics that seem to be backed by little
more than ‘pseudoscience’. In this regard, their second principle, (PPC), is
initially easier to accept. Note however that the requirement of ‘consen-
sus’ in the second principle is likely to cause controversy and many would
insist that the principle should be supplemented with a fallibilist attitude
towards such a ‘consensus’. Obviously, we need to be careful with examples
drawing from science and even more careful if input from metaphysics to

34 Ladyman and Ross, Every Thing Must Go, p. 119.

35 For details, see for instance Mauro Dorato and Matteo Morganti, ‘Grades of Individuality.
A Pluralistic View of Identity in Quantum Mechanics and in the Sciences,’ Philosophical
Studies 163 (2013), pp. 591–610.
36 For further discussion on underdetermination, see Steven French, ‘Metaphysical
Underdetermination: Why Worry?’, Synthese (2011) 180, pp. 205–221; see also Steven
French, The Structure of the World:Metaphysics and Representation (Oxford University Press,
2014), especially Ch. 2.
9.3 The Principle of Naturalistic Closure and the Primacy of Physics 223

science is proposed. But the cost of accepting both of the principles pro-
posed is that metaphysics becomes simply a ‘handmaiden’ of science. Note
also that we have omitted entirely a discussion of the positive proposal put
forward by Ladyman and Ross, a view they call ‘ontic structural realism’
(OSR). We encountered the structuralist view briefly in Chapter 6, but have
not properly engaged with it. This is partly because the details of the the-
ory are less important from a metametaphysical point of view than from
the point of view of first-order metaphysics or philosophy of science, where
(OSR) is an important contender, and partly because we could not do justice
to (OSR) here.37 We might note, though, that this ontological picture has
been challenged, among other things, on the basis of an oversimplistic pic-
ture regarding ontological dependence: one version of (OSR) suggests that
objects depend for their identities on the structures to which they belong,
so the structure is more fundamental than the objects – hence the title of
Ladyman and Ross’s book, Every Thing Must Go. Now, this looks like a claim
of identity-dependence, but as we’ve seen in the discussion in Chapters 5
and 6, there are several difficult questions to address before we can accur-
ately formulate and use this notion of dependence – a task which Ladyman
and Ross do not undertake.38
Setting the details of the ontological picture aside, there is another
important methodological challenge that can be derived from the two prin-
ciples, (PNC) and (PPC). This concerns a core area of metaphysics – one that
we discussed in considerable detail in Chapter 7 and also earlier – namely
modality. We already mentioned the challenge regarding modal epistemol-
ogy, namely that philosophers have often wrongly regarded something as
possible or necessary. More generally, one might worry that it is unclear
what the conceptual space that metaphysics is supposedly concerned with
is like, and it is sensible to think that it is ultimately physical modality
that determines what we regard as metaphysically possible, necessary, or
impossible. In other words, following (PPC), a scientifically minded philoso-
pher could argue that all modality reduces to physical modality and hence
any judgements that we make about what is possible or necessary should
have physics – or science more generally – as their source.

37 For a classic overview of structural realism, see James Ladyman, ‘What is Structural
Realism?’, Studies in History and Philosophy of Science Part A 29 (1998), pp. 409–424.
38 For further discussion, see, e.g., Donnchadh O’Conaill, ‘Ontic Structural Realism and
Concrete Objects,’ The Philosophical Quarterly 64.255 (2014), pp. 284–300.
224 Demarcating metaphysics and science: can metaphysics be naturalized?

Craig Callender has raised a challenge not unlike this one, pointing
out that the focus on metaphysical modality is suspicious precisely because
metaphysicians have claimed that this is an area on which science can say
very little.39 Callender thinks that the focus on metaphysical modality is
one of the primary reasons for the disconnect (or even hostility!) between
contemporary metaphysics and philosophy of science. The rift could be
put as follows. Science concerns the actual world and makes claims about
what is physically possible and necessary, but metaphysicians want some-
thing stronger: they want to know what is metaphysically necessary, say,
whether the actual laws of physics hold across all possible worlds. Indeed,
the debate about the laws of nature is one of the central topics of con-
temporary analytic metaphysics and it is perhaps not surprising that those
philosophers who have scientific training or interests find this debate frus-
trating. Who cares about whether salt dissolves in water in some remote
possible world!40 Well, as Callender points out, this is not really the right
way to think about the matter:

I submit that the basic problem with some metaphysics today is the
idea that the philosopher and scientist doing ontology are performing
fundamentally different and separate jobs. The metaphysician’s picture
that the scientist works in the lab, discovering the actual world’s features,
while the metaphysician discerns the wider universe of possible, isn’t
right. The error is thinking that the science of the actual world doesn’t
affect what one thinks is possible or impossible. The history of science and
philosophy amply displays that what we think is possible or impossible
hangs on science. Or going in the other direction, the error is thinking
that modal intuitions are reliable if they are not connected to a systematic
theory of a large domain, one possessing many theoretical and empirical

39 Callender, ‘Philosophy of Science and Metaphysics,’ p. 40. Similar concerns are raised
in Maudlin, The Metaphysics Within Physics.
40 For discussion regarding laws of nature and this very example, see for instance
Alexander Bird, Nature’s Metaphysics (Oxford University Press, 2007). See also Tuomas
E. Tahko, ‘The Modal Status of Laws: In Defence of a Hybrid View,’ The Philosophical
Quarterly (2015 [online], forthcoming).
41 Callender, ‘Philosophy of Science and Metaphysics,’ p. 43–44.
9.4 Methodological similarities 225

This is, by all accounts, a fair assessment, one that the present author
would readily agree with. But there is one more issue worth mentioning
before we move on, having to do with the relevant interpretation of modal
claims. While Callender is correct in that the exclusive focus on irreducible
metaphysical modality is suspicious, the thought that the modal space
that scientists deal with is somehow absolutely well defined does not fol-
low. Even a proponent of ‘strong metaphysics’ need not commit to the
irreducibility of metaphysical modality. Indeed, as we saw in Chapter 7,
on the essentialist framework of Fine and Lowe metaphysical modality
reduces to essence, albeit it is unlikely that a scientifically minded phil-
osopher would find the appeal to irreducible essence any more satisfying.
Nevertheless, the point is that (semi-)autonomous metaphysics does not
need to commit to any single piece of doctrine, such as irreducible meta-
physical modality. The scientifically minded philosopher only needs to
insist on the necessity of the conceptual and methodological ‘toolbox’ of
metaphysics for performing certain tasks, such as the interpretive task we
have now encountered several times. Note that this immediately makes
room for a more moderate form of naturalism about metaphysic s – some-
thing that we will discuss in more detail in the last section of this chap-
ter. But first, let us take a look at the potential methodological similarities
between science and metaphysics, for it seems that Callender is calling
exactly for a closer study of such similarities, instead of focusing on the
evident differences.

9.4 Methodological similarities

Whilst rushing through the various comments and critical remarks con-
cerning the relationship between science and metaphysics we have, hope-
fully, made at least some progress. There is clearly quite a bit of chatter
between scientists (or rather, ‘scientifically-minded’ philosophers) and
metaphysicians and despite some extremists, there is a rough consensus
that the two cannot be entirely independent of each other. But if there is
some overlap between the disciplines, we still ought to ask: does this over-
lap concern the methods or the subject matter of science and metaphysics?
One recent and potentially helpful discussion on just this issue is due to
L. A. Paul, who suggests that metaphysics and science have effectively the
226 Demarcating metaphysics and science: can metaphysics be naturalized?

same methodology, but distinct subject matters.42 The core of Paul’s pro-
posal can be summarized with the following passages:

The ontological account describes the metaphysically prior categories

and constituents of the physically fundamental entities, and in this sense
describes features of the world that are more fundamental than those of
natural science.43

This proposal takes advantage of the metaphysical ‘toolbox’ that we are by

now quite familiar with: Paul suggests that the subject matter of metaphys-
ics is ontologically prior to that of science. She thinks that this ontological
priority is reflected by conceptual priority:

The fact that the subject matter of metaphysics can be ontologically prior
to the subject matter of science is reflected in the fact that many concepts
of metaphysics are conceptually prior to the concepts of science. […] There
is no way to make sense of the central concepts of classical field theory or
quantum chromodynamics without using a concept of property.44

Note that Paul is making a strong claim here: she appears to suggest that we
cannot even understand or interpret science without resorting to notions
which are distinctly philosophical, or require a philosophical analysis. It
is indeed relatively uncontroversial that the notion of a property is prior to
that of, say, electric charge: electric charge is a specific case of property.
In this sense, metaphysical concepts do appear more fundamental than
scientific concepts. But why should we think that this priority holds quite
generally? Moreover, this in itself may not be sufficient to establish that
metaphysical inquiry has a ‘privileged’ role to play when it comes to study-
ing the fundamental structure of reality, as one could interpret Paul to be
suggesting. In general, one might have doubts about whether conceptual
priority is a convincing criterion or indication of ontological priority. There
are at least two reasons to doubt that there is such a strong link between
them. First, it is often very difficult to even determine when one concept
is prior to another – the order of the acquisition of concepts being of little
help. Secondly, even if a specific sense of priority and dependence among

42 L. A. Paul, ‘Metaphysics as Modeling: The Handmaiden’s Tale,’Philosophical Studies 160

(2012), pp. 1–29.
43 Paul, ‘Metaphysics as Modeling: The Handmaiden’s Tale,’p. 5.
44 Ibid., p. 6.
9.4 Methodological similarities 227

concepts is granted, work is required to establish which of the seemingly

more fundamental linguistic or conceptual categories latch onto objective
structures of reality. Paul does however have something more to add in this
regard: she seems to have in mind an explanatory link, broadly understood.
Given that we can formulate this type of link more precisely by using the
‘toolbox’ of metaphysics, this type of explanatory link could perhaps be
described in terms of grounding or explanatory dependence, producing the
type of ontological priority that Paul was looking for.
However, it appears that we would need something more than this. The
mere fact that we use certain concepts in scientific theories and explana-
tions does not directly entail that we should be ontologically committed
to their existence in anything more than a heuristic sense. The reason for
this is that certain presuppositions concerning the ontological structure of
reality are likely to be at work in every explanation; and different pre-
suppositions might well lead to different explanations and different onto-
logical commitments . This clearly blocks the inference at hand, from the
use of certain concepts in explanation to the objective existence of those
ontological elements. In other words, what guarantees that the relevant
concepts accurately reflect the structure of reality? This is a more difficult
question to answer, so let us set it aside here in order to focus on another
important aspect of Paul’s proposal, having to do with the methodology of
Recall that Paul considers metaphysics and science to have, by and large,
the same methodology, but distinct subject matters. We have now seen
that the case concerning subject matter is far from settled, but what about
the case of methodology? At the core of Paul’s account is the claim that
both science and metaphysics rely on modelling. Specifically, Paul suggests
that – as they would appear to be for scientists – models are the metaphys-
ician’s primary tool of theory-forming. The categories of entities involved
in metaphysical models are different from those used in science (e.g., one
would talk of properties or substances rather than of particles or genes),
but metaphysics and science both model parts of reality nonetheless. Both
science and metaphysics use a priori reasoning to infer to the best explan-
ation, which helps us choose between empirically equivalent models. Now,
we’ve already noted that the method of inference to the best explanation
may be problematic, but let us set aside those worries for the moment.
Additionally, Paul suggests that both in science and in metaphysics the
228 Demarcating metaphysics and science: can metaphysics be naturalized?

usual theoretical virtues of simplicity, ontological parsimony, elegance,

explanatory power, and fertility may be used to evaluate models. In Paul’s
view, then, metaphysics and science are primarily demarcated by their sub-
ject matter, not their methodology.
The idea of metaphysics as modelling is initially attractive, but in order
for it to work in the area of metaphysics, we need some account of how to
choose between empirically equivalent models. In scientific contexts, models
can typically be tested in a relatively straightforward manner. Not so in
metaphysics. So how do we assess models? How do we decide which of
two models is better if they differ in no way with regard to the empirical
predictions that they make – if they even make any? Paul’s strategy appeals
to inference to the best explanation based on quantification over the best
available theories – somewhat reminiscent of Quine (although Paul does
not mention Quine). This viewpoint is, as we have seen, problematic: Does
it provide a reliable link between one’s metaphysical theory and the actual
structure of reality, or merely a pragmatic criterion? In general, one could
attempt to apply the above-mentioned theoretical virtues to justify model
selection. However, it is not obvious that, say, simpler or more elegant sci-
entific theories are more likely to be true, even if it were the case that
they are more explanatory, easier to understand, or pragmatically valuable.
These are all issues of intense dispute in the philosophy of science. In sci-
ence, however, empirical data can be brought to bear on the evaluation of
one’s conjectures – something to which metaphysics cannot or at any rate
typically does not have access in its own right. This naturally represents a
potential point of fundamental demarcation between science and meta-
physics. Indeed, Ladyman immediately replies, re-emphasizing something
that we have already come across: since in metaphysics truth is all that
matters (metaphysics has no practical applications) but theoretical virtues
are not obviously truth-conducive, the fact that metaphysical models can only
be assessed on the basis of their theoretical virtues suggests that the pro-
posed methodology is under threat.45
Even if this issue concerning empirically equivalent models could be
settled, there are aspects of the ‘metaphysics as modelling’ suggestion that
we ought to discuss in more detail. For this purpose, Paul has helpfully

45 James Ladyman, ‘Science, Metaphysics and Method,’ Philosophical Studies 160.1 (2012),
pp. 31–51.
9.4 Methodological similarities 229

provided us with a number of examples on how exactly this method can

be used and applied. Thought experiments are one particularly promising
avenue in this regard. For the process of modelling, two further tools are of
importance: abstraction and idealization. For instance, when examining the
connection between two distinct events, Socrates’s death and his drinking
hemlock, we may abstract away from the complex details of the actual
world in order to identify a counterfactual dependence relation between
these events. Paul suggests that by abstracting away superfluous details, we
can conclude that such counterfactual dependence is sufficient for causa-
tion. Hence, Socrates’s death was caused by his drinking the hemlock. The
process of abstraction is important, Paul says, because in actuality super-
fluous features may ‘muck up’ the real-world facts relating to dependence.
However, the model we get by abstraction is sufficiently isomorphic to the
real-world case for a reliable connection, or so the story goes. If this were
correct, the suggested parallel between science and metaphysics would
indeed hold: we get to identify the fundamental features of reality in both
cases via abstraction and idealization. This account seems promising, but it
can certainly be challenged. For it relies on an underlying account regard-
ing the role of counterfactuals in metaphysical modelling.
In Chapter 7 we have already pointed out certain problematic aspects of
this account, with regard to Timothy Williamson’s theory of modal epis-
temology.46 And Paul makes use of Williamson’s account explicitly.47 Here
is a brief recap of the problem we discussed earlier. The account does not
give us reliable means to explain the background knowledge, Williamson’s
‘constitutive facts’, needed to get the account started. Constitutive facts
are things that we must hold fixed across metaphysically possible coun-
terfactual scenarios, and knowledge of such facts is needed to secure trad-
itional examples of metaphysical necessities. But how is one to determine
which facts are constitutive? If one’s evaluation of counterfactual scenarios
depends on what is held fixed, it obviously follows that there is more to
metaphysical model-selection than counterfactual reasoning: at the least, it
also includes a non-counterfactual determination of the fixed background.
The upshot is that counterfactual supposition – at least on its own – does

46 Timothy Williamson, The Philosophy of Philosophy (Blackwell Publishing, 2007),

especially Ch. 5.
47 Paul, ‘Metaphysics as Modeling: The Handmaiden’s Tale,’p. 23.
230 Demarcating metaphysics and science: can metaphysics be naturalized?

not provide a complete story about our epistemic access to metaphysically

necessary facts, as knowledge of constitutive facts must precede the coun-
terfactual account itself.48
Consider the following example. Paul argues that a counterexample to
the claim that causation is necessarily a relation of counterfactual depend-
ence between events would require finding a metaphysically possible world
with a case of causation between events not exhibiting counterfactual
dependence. But how can we (fail to) find such a world – thus vindicating
the counterfactual account of causation – if whether it exists at all depends
on what we take to remain fixed across possible worlds? Accordingly, the
problem that immediately follows is that if causation is indeed neces-
sarily a relation of counterfactual dependence between events, then it is
exactly one of those precious metaphysical necessities that should be held
fixed across metaphysically possible counterfactual scenarios. According
to a Williamson-style analysis, failing to hold a constitutive fact fixed in
a counterfactual supposition entails a contradiction. But the question at
hand concerns whether it is a (metaphysically necessary) constitutive fact
that causation is a relation of counterfactual dependence between events.
In either case we are going to get unexpected results in our counterfac-
tual suppositions. Specifically, on the one hand we will get a contradiction
where there should be none if we mistakenly hold it fixed that causation is
a relation of counterfactual dependence between events, and on the other
hand we will not get a contradiction where there should be one if it is a
constitutive fact and we fail to keep it fixed. In other words, the method
can produce reliable results only if we know which facts to hold fixed in
the first place.
Summarizing, Paul clearly identifies an important parallelism between
scientific and metaphysical methodology: they both employ inference to
the best explanation and criteria for theory choice based on theoretical vir-
tues; and they both work with abstract and idealized models of reality. Her
suggestion that metaphysical models are in some sense more fundamen-
tal than scientific models is also intriguing. However, it remains open to
debate how exactly metaphysical models connect to scientific models, and
how (if at all) theoretical virtues can be reliable guides for theory choice in
48 On this, see Tuomas E. Tahko, ‘Counterfactuals and Modal Epistemology,’ Grazer
Philosophische Studien 86 (2012), pp. 93–115.
9.5 Moderately naturalistic metaphysics 231

9.5 Moderately naturalistic metaphysics

In what remains of this chapter, we will examine a more moderate pos-

ition, attempting a reconciliation between science and metaphysics. This
position has important similarities with the one proposed by Paul, but
also with Lowe’s more ambitious metametaphysical position. Recall that
we discussed Lowe’s position in connection to autonomous metaphysics –
Lowe considers metaphysics and science to differ both in terms of their
subject matter and with respect to methodology. However, by combining
elements from Paul’s and Lowe’s proposals, we can outline a position that
may improve on both of them. We do not need to rehearse Lowe’s position
in detail here, for the relevant aspects have been covered earlier in this
chapter and in previous chapters. The key aspect that we need to recall
from Lowe’s work is that, as he puts it, metaphysics can be characterized
as ‘the science of the possible’.50 Moreover, we should recall the remarks made
in Chapter 7 regarding the relationship between a priori and a posteriori
inquiry, namely that there are reasons to think that both of these are needed
when pursuing metaphysics, and that they can be considered to proceed in
a manner of ‘bootstrapping’ – that is, by alternating in a way that renders
talk of ‘pure’ a priori or a posteriori knowledge inert. We will also consider
whether it is possible to address one of the shortcomings we identified in
Paul’s account, namely finding a way to choose between models that goes
beyond a comparison of theoretical virtues. The suggestion, which is per-
haps rather surprising, is that it is at least in some cases possible to test
metaphysical hypotheses empirically, albeit in an indirect sense. This can be
done by applying such hypotheses to the interpretation of our best scientific
theories, as has been suggested already.
Lowe thinks that metaphysical a priori arguments are the source of
the type of constitutive or essentialist knowledge that the counterfactual
approach familiar from Williamson – which we encountered in the previ-
ous section – seems to struggle with. Further, Lowe holds that metaphysical
modality is grounded in essence. But what does this have to do with scien-
tific knowledge? Well, Lowe’s view is that we have a priori access to modal
knowledge concerning ontological categories – this epistemic assumption

49 For a recent example of a view on these lines, see Matteo Morganti, Combining Science
and Metaphysics (New York: Palgrave Macmillan, 2013).
50 Lowe, ‘The Rationality of Metaphysics,’ p. 100.
232 Demarcating metaphysics and science: can metaphysics be naturalized?

is the core of the methodology of Lowe’s metaphysics – and he specifies

that the relevant epistemic process is not based on intuitions or thought
experiments, but rather on direct a priori access to essentialist facts which
ground the modal truths. It is central to this account that such essentialist
knowledge precedes empirical knowledge about which ontological categor-
ies are actual. However, and this is a crucial point, Lowe maintains that
our epistemic access to essentialist knowledge is relatively unproblematic
and doesn’t in fact require any specific philosophical treatment. An illus-
trative example that Lowe provides concerns the case of transuranic elem-
ents: many of them were only synthesized after their possible existence
was determined by non-empirical means. With the help of Mendeleev’s
periodic table, chemists have been able to predict the existence of a number
of yet-to-be-discovered elements and to make highly accurate predictions
about their properties. Later on, we were able to synthesize these elements
and verify that they indeed had the predicted properties. Lowe proposes
that this process would not have been possible without a prior grasp of the
essences of these transuranic elements.51 But simply understanding what
would qualify as a transuranic element of a certain type was sufficient for
defining the relevant categories. This easily generalizes, seemingly lending
support to Lowe’s view of metaphysical knowledge of essences.
If Lowe is correct, it seems that a priori reflection was sufficient to predict
the characteristics of the yet-to-be-seen elements; and that these predicted
characteristics were essential features of those elements. But it must be
acknowledged that this prediction reflected (what we take to be) objective
features of the world only because, for the known elements of the table, we
had prior empirical acquaintance with them. Lowe’s idea, then, is that some a
priori idea of what is possible and what is not is certainly needed to even start
to structure empirical data conceptually. The type of modal rationalism that
Lowe puts forward, as we’ve seen, infers from this that our grasp of essences
is prior to the gathering of empirical data and input. This may be plausible
enough in some cases – Lowe’s most forceful examples do not concern nat-
ural kinds, but abstract objects such as sets. In these cases, Lowe’s take on
essences might well be a defensible approach.But since knowledge of natural
kinds (understood as explaining what is shared by the concrete members of

51 See E. J. Lowe, ‘Two Notions of Being: Entity and Essence,’ Royal Institute of Philosophy
Supplements 83.62 (2008), pp: 23–48.
9.5 Moderately naturalistic metaphysics 233

a kind rather than as pointing to abstract universals) requires input from the
empirical sciences, it appears that the process of coming to know essences
cannot be completely a priori, at least not in all cases. In fact, unless some
account is given of this empirical element Lowe’s point runs the risk of being
reduced to an uncontroversial but relatively unimportant claim about our
understanding of concepts (rather than grasp of ‘real’ essences).
What remains of Lowe’s proposal, then? Can his suggestion that meta-
physics is an exploration of asui generis, fundamental space of possibilities be
saved in spite of the shortcomings just pointed out? To maintain that we have
a priori access to essence even in the case of material objects such as natural
kinds, Lowe would have to adopt the view that the essential features of mate-
rial objects are, by and large, accessible to us via a simple reflective process.
This does not compel Lowe to claim that the full essence is to be reached so
easily, but we ought to be able to access enough of the essence to be able to
separate one object from another. Hence, if we try to separate water from
other superficially similar liquids, it would not be enough to say that, for
instance, water is a transparent liquid (in ‘normal’ conditions, i.e., in room
temperature etc.) because there are several other chemical substances, such
as hydrogen peroxide, that have the same (essential) features. But consider
where Lowe ends up:

[O]ur natural classifications do not need to be, and in factshould not be,
forced into a single, all-embracing taxonomic scheme. Realdivisions in
nature are reflected by our natural classificatory schemes, but they are often
divisions at different levels, allowing for agood deal of cross-classification.
So, to revert to an earlier example, thereis nothing wrong in saying that, for
some purposes, diamond, graphite and charcoal may be regarded as different
kinds of substance[.]52

So diamond, graphite, and charcoal may be regarded as different kinds

of substance for some purposes, even though they can be regarded as the
same substance, namely carbon, from a microstructural point of view. On
the face of it, the passage above may even seem to invite conventionalism,
but this is surely not what Lowe has in mind. Nevertheless, it does leave his
view open to a conventionalist challenge, for who is to say which of our clas-
sificatory schemes is supposed to reflect the real divisions in nature, if any?

52 E. J. Lowe, ‘Locke on Real Essence and Water as a Natural Kind: A Qualified Defence,’
Aristotelian Society Supplementary Volume 85 (2011), pp. 16–17.
234 Demarcating metaphysics and science: can metaphysics be naturalized?

Let us see where we can get with this. What we are effectively looking
for is a model, more or less in Paul’s sense, which would give us the tools
to assess cases like elements, explaining our ability to make predictions
about them in advance of observations. So consider carbon. Diamond and
graphite, which Lowe mentions, are two of the several allotropes of car-
bon – pure forms of the same element that differ in structure. But there
is of course a reason behind the fact that we typically classify different
allotropes as members of the same kind, despite the significant differences
in their chemical properties. The reason has its source in the fact that des-
pite their many differences, the allotropes of carbon can be subjected to
the same type of chemical reaction with the same type of result. This reac-
tion is oxidation: If one burns diamond, graphite, or charcoal, the result
is always the same, namely pure carbon dioxide. If any of these allotropes
were compounds of different elements instead, the burning would result in
some impurities. As it happens, this is exactly how Lavoisier, the famous
eighteenth-century chemist, discovered that diamond is indeed an allo-
trope of carbon. Lavoisier heated a diamond in a glass jar until it disap-
peared and observed that the weight of the jar had not changed, hence
concluding that the diamond must have been made of carbon to produce
the carbon dioxide gas present in the jar after heating. 53
We can conclude that when modelling elements, we ought to take into
account at least two things. The first is the ability of elements to form allo-
tropes and their ability to form compounds with other elements (which
can also take several forms; that is, polymorphism in general must be taken
into account). The second is the survival of something essential to the elem-
ent in all of the different forms that it can take. More precisely, it is in
virtue of the essence of carbon that carbon atoms are capable of forming
allotropes with varying crystalline structures, and it is also in virtue of the
essence of carbon that the vast range of forms that carbon can take in a var-
iety of allotropes and compounds still share some aspect of the elemental
form of carbon. What these aspects are is a question of chemistry and we
need not go into a lot of detail here. 54 But we could perhaps understand the

53 For further discussion, see Tuomas E. Tahko, ‘Empirically-Informed Modal Rationalism,’

in R. W. Fischer and F. Leon (eds.), Modal Epistemology After Rationalism, Synthese Library
(Dordrecht: Springer, forthcoming).
54 But see Tuomas E. Tahko, ‘Natural Kind Essentialism Revisited,’ Mind 124.915 (2015),
pp. 795–822.
9.5 Moderately naturalistic metaphysics 235

chemical properties of compounds to be a result of the interaction of the

elements present in those compounds, which would point towards micro-
structural essentialism. If this were correct, it would have to be the case
that something survives in the causal processes that elements undergo
when forming compounds – and also allotropes.
This all-too-brief example gives us some clues towards combining ele-
ments from Paul’s and Lowe’s proposals into an empirically informed analy-
sis of metaphysical-cum-scientific inquiry. Since the model we just examined
makes some predictions about how future scientific inquiry could proceed,
it is in a sense at least indirectly testable: that is, it is not entirely immune
to, and indifferent towards, empirical input. This type of indirect testability
could be used to build a bridge between scientific and metaphysical mod-
els and steer clear of the objection that, since metaphysics is not empiri-
cally relevant and yet aims at the truth, metaphysics should be dismissed.
Relatedly, the systematic application of metaphysical concepts and hypoth-
eses (taken from the space of possibilities in accordance to Lowe’s sugges-
tion) with a view to interpreting our best current science seems to provide
at least one criterion for going about answering metaphysical questions and
selecting between metaphysical models. What gets selected, in particular,
are those among the latter that are not only the best in terms of theoretical
virtues but also most suited, all things considered, for making sense of the
best available science. This, together with the strategy based on the modal
constraints for scientific theorizing emerging from Lowe’s proposal – the
idea that a preliminary exploration of possibility space is necessary for sci-
entific inquiry to even get started – is tantamount to saying that, in spite of
its (relative) autonomy, metaphysics is not a merely conceptual or linguistic
activity, fully distinct from science. The upshot would be what might be
labelled ‘moderately naturalistic metaphysics’.
It will be left to the reader to assess these options – and others, for our
discussion has hardly covered all the options. But perhaps it is reasonable
to conclude this book by noting that contemporary analytic metaphysics
can go at least some way towards meeting the challenge of combining sci-
ence and metaphysics, and indeed demarcating them.

allism and noneism: Competing views regarding controversial entities such as

past and future things, merely possible things, universals, numbers, and
so on. Allism is the view that all such things exist, whereas noneism states
that none of them exist.
application conditions: The conditions under which a term may be properly
atomism: The theory, srcinating from Democritus, that physical objects con-
sist of simples, indivisible particles. See also mereological atomism.
conceptual analysis: A methodological position emphasizing the linguistic or
conceptual clarification of the notions we use in philosophy and/or science;
typically a priori.
conceptual necessity: Necessary in virtue of the definitions of concepts (plus
the laws of logic), e.g., ‘All bachelors are unmarried’.
constitutive facts: Elements of background knowledge that are held fixed in
counterfactual suppositions. A plausible example of constitutive facts are
metaphysically necessary facts concerning the essences of entities.
contingent: In contrast to necessary; could have been otherwise. For instance,
a proposition is metaphysically contingent as opposed to metaphysically
necessary if it is true in some metaphysically possible worlds and false in
some other metaphysically possible worlds.
conventionalism: The view that some or all aspects of metaphysical inquiry are
based on linguistic, psychological, or societal conventions rather than genu-
ine distinctions in reality.
counterfactual supposition: Counterfactual suppositions concern non-actual
matters. If p holds in the actual world, what would follow if not-p?
deflationary, deflationism: The view that ontological questions (universally or
in some specific area) are not substantive, but are rather based on confu-
sion, linguistic choice, or something else.

Glossary 237

emergence, emergent: The idea that complex systems and patterns can arise or
emerge out of simple interactions; the emergent properties of a complex
system may not be reducible to the parts of the system.
endurantism and perdurantism: Competing views regarding temporal parts.
Perdurantists hold that material objects have temporal parts in addition

to spatial parts and hence that objects ‘perdure’; objects persist by having
temporal parts. Endurantists hold that material objects do not have tempo-
ral parts, which leads them to think that objects are wholly present at any
time that they exist; they ‘endure’. See also three- and four-dimensionalism.
entity: Something existing, an object, although notnecessarily a material object.
essence: The essence of an entity can be expressed through its identity and
existence conditions; the properties which make it the very thing it is.
These properties are necessary for the entity: it couldn’t exist without
them. Sometimes essence is explicated in terms of real definition, some-
times in purely modal terms. Typically, essences are not considered to be
entities themselves.
essentialism: The view that (at least some) things have essences, which make
them the very things or kinds of things that they are. One version of this
view is natural-kind essentialism, which postulates essences for natural kinds,
such as water.
external questions: Originating from Rudolf Carnap. External questions are
questions that go beyond established frameworks; they concern the appli-
cability of a framework to reality. According to Carnap, these are typically
questions that philosophers (metaphysicians) ask, but Carnap regards
them as meaningless. See also internal questions.
essential dependence: A form of ontological dependence whereby an entity x
depends essentially for its existence upon an entity y, in the sense that it is
part of the essence of x that x exists only if y exists.
existential dependence: A form of ontological dependence whereby an entity x
depends upon an entity y for its existence, in the sense that x couldn’t exist
if y didn’t exist.
experience-based intuition: Intuition derived from experience; concerning the
phenomenology of the subject matter of the intuition rather than concep-
tual analysis.
fallibilism, fallible: The view that a given position is subject to revision; the
position is fallible.
four-dimensionalism: See three- and four-dimensionalism.
238 Glossary

fundamental, fundamentality: Something , more basic than,

ontologically independent
or prior to everything else. It is often thought that all non-fundamental things
must ontologically dependon or be grounded in some fundamental things.
ground, grounding: A metaphysical priority relation distinct from causation.
For instance, the members of a set are prior to the set itself; the existence

of the set is grounded in its members. Grounding is often considered to

obtain between facts, i.e., if fact x is grounded in fact y, then y metaphysi-
cally explains fact x.
gunk: The idea of atomless gunk in mereology is that everything has a proper part;
there are no mereological atoms.
identity-dependence: A form of ontological dependence whereby an entity x
depends upon an entity y for its identity, in the sense that x wouldn’t be the
entity that it is or the kind of entity that it is if y didn’t exist.
infinite regress: A relationship transmitted through an infinite number of
terms in a series or chain, where there is no last term that terminates the
chain. See also metaphysical infinitism.
internal questions: Originating from Rudolf Carnap. Internal questions are
questions within a certain framework. They make sense once the frame-
work is established and adopted. For instance, internal questions can be
questions of varying difficulty in a scientific framework. But general ques-
tions regarding what exists in a framework are trivial, so when philos-
ophers (metaphysicians) ask general questions, they presumably do not
have internal questions in mind, or so Carnap thought. See also external
joint-carving: When a predicate carves at the joints, it is considered to correspond
with some structure in reality; it picks out a distinction which is genu-
ine rather than conventional. If a predicate carves perfectly, then it picks out
something fundamental.
language pluralism: The view, of Carnapian srcins, that we can choose our
(linguistic) framework liberally.
logical necessity: Necessary in virtue of the laws of logic, e.g., ‘It either rains or
does not rain’.
mereology: The study of parts and wholes, or parthood, and closely related
mereological atomism: The view that there are mereological atoms, simples that
have no further proper parts and that everything is ultimately composed of.
metaphysical foundationalism: The view that there is a foundation to reality,
a fundamental level that everything else ultimately depends on. On this
view, chains of dependence must terminate.
Glossary 239

metaphysical infinitism: The view that there is no foundation to reality, no fun-

damental level that everything else ultimately depends on. On this view,
chains of dependence could go on ad infinitum. See also infinite regress.
metaphysical necessity: Necessary in virtue of the essences of entities, e.g.,
‘Water is H2O’.

metaphysical realism: The doctrine that there is a mind-independent

external world.
modal, modality: Modality, or modal terms, describe what could be or must be
the case, i.e., they deal with possibility and necessity.
modal empiricism: A family of views emphasizing empirical elements in modal
modal epistemology: The study of how we come to know modal truths.
modal rationalism: A family of views emphasizing a priori, non-empirical elem-
ents in modal epistemology.
monism: The view, famously associated with Spinoza, that there is only one
independent thing or substance. A more moderate version popularized
by Jonathan Schaffer, priority monism, suggests that there is only one
fundamental thing.
natural kind: When science or philosophy picks out a grouping that reflects the
structure of the world that is independent of the interests and actions of
humans, it has picked out a natural rather than an artificial kind.
natural necessity: Also known as nomological necessity and physical necessity.
Necessary in virtue of the laws of physics or laws of nature.
naturalism: A group of views prioritizing natural science; the idea that natural
sciences exhausts our area of inquiry and natural phenomena are all there
is. Various different, weaker, and stronger versions of the view exist.
nihilism: The view that a collection of material objects never composes a fur-
ther object; composition never occurs. A possible answer to the Special
Composition Question.
nomological necessity: Also known as natural necessity and physical necessity.
Necessary in virtue of the laws of physics.
non-experimental intuition: A specific type of intuition associated with experi-
mental philosophy. Refers to an intuition to which the designer of an
experiment studying the intuitions of two populations directly appeals or
which must eventually be appealed to in order to establish a criterion for
determining whether two populations share a common concept.
noneism: See allism and noneism.
Ontologese: The special language of ontology, more suitable for ontological
debates than ordinary English.
240 Glossary

ontological dependence: A family of relations referring to various forms

of metaphysical, non-causal dependence relations between entities.
Identity-dependence, essential dependence, and existential dependence are forms
of ontological dependence.
ontological realism: In this book, the notion is used in a very general sense, refer-

ring to the view that ontological facts are objective or mind-independent.

But the reader should be aware that the notion is used in various more
precise (and controversial) senses in the literature.
perdurantism: See endurantism and perdurantism.
physical necessity: Also known as natural necessity and nomological necessity.
Necessary in virtue of the laws of physics.
possible worlds: Often thought to be just a façon de parler – a way of speaking
about modal terms, a heuristic tool – or according to modal realists, like
David Lewis, to refer to real alternative worlds (although not accessible
from the actual world).
principle of charity: The principle according to which a debate can fail to be
merely verbal only if each side in a debate is genuinely charitable and
still fails to interpret the assertions of the other side as being true in their
respective language.
principle of independence: A principle introduced by Alexius Meinong, which
states that an entity may have any properties whatsoever, independently
of its existence.
quantifier variance: The view that there is no uniquely best ontological lan-
guage for describing the world; suggests that many (but not necessarily all)
of our ontological disputes are merely verbal or conventional.
rational intuition: A conception of intuition as a special rational faculty, empha-
sizing its role in a priori judgements; our belief in certain propositions is
justified by rational intuition rather than by the senses, introspection, or
real definition: An Aristotelian notion associated with the nature or essence of
a thing. To give a real definition of some entity is to state its existence and
identity conditions.
reduction, reductionism: When one thing x reduces to another thing y, then
everything about x is in some sense already contained in y and can be
expressed by talking about y. Reductionism applied to two fields of study
more generally is often thought to undermine the distinction between
them, e.g., the attempt to reduce consciousness to neurophysiology.
reflexivity: A relation R is reflexive if and only if R is self-relating, i.e., every-
thing bears relation R to itself, e.g., ‘being self-identical’.
Glossary 241

relativism: May be applied to various areas. Relativism concerning a given sub-

ject matter suggests that judgements made about the subject matter are
relative in some sense, e.g., relative to linguistic decisions or psychological
Special Composition Question (SCQ): The question: Under what conditions do

a number of objects compose a further object? Introduced by Peter van

Inwagen, according to whom the answer to the question must be formu-
lated in non-mereological terms.
structuralism: A group of views favouring structure-oriented ontology in con-
trast to object-oriented ontology, e.g., we focus on ‘fields’ rather than ‘par-
ticles’. One version of this view is ontic structural realism.
substance: An Aristotelian notion referring to something independently exist-
ing, fundamental, or basic.
symmetry: A relation R is symmetric if and only if: if x is related by R to y, then
y is related by R to x, e.g., ‘being siblings’.
theoretical virtues: A group of features that a scientific or philosophical the-
ory may have, usually assumed to make the theory preferable because it
is more likely to be true or more likely to be useful. Theoretical virtues
include simplicity, explanatory power, internal consistency, empirical adequacy,
and so on.
three- and four-dimensionalism: Competing views regarding temporal parts.
On the four-dimensional view, a material object can be thought of as a
type of ‘spacetime worm’, since all the different temporal parts are also,
in some sense, parts of the object. In contrast, the three-dimensional view
attaches no special significance to past or future temporal parts of an
object – the object is there at a given time and that’s it. See also endurant-
ism and perdurantism.
transitivity: A relation R is transitive if and only if: if x is related by R to y, and y
is related by R to z, then x is related by R to z, e.g., ‘being taller than’.
truthmaking, truthmakers: Truthmaker theory concerns the idea that when
something x makes it true that p, then x is a truthmaker for p. In other words,
p is true in virtue of x. A variety of different views about the truthmaker
relation, the nature of truthmakers such as x as well as the nature of truth-
bearers such as p have been entertained.
universalism: The view that a collection of material objects always composes a
further object. A possible answer to the Special Composition Question.
vague, vagueness: Something that lacks sharp boundaries. A term or predicate is
vague when there is no sharp cut-off between cases where the term applies
and cases where it doesn’t apply, e.g., ‘is bald’. Some consider vagueness to
242 Glossary

always be merely linguistic, some think that there could also be vagueness
in the world, that is, metaphysical vagueness.
well-foundedness: An order is said to be well-founded if every non-fundamental
element of the order is fully grounded by some fundamental element(s).
The notion of well-foundedness may be used to explicate metaphysical

Aristotle. 1984. Metaphysics (W. D. Ross, trans.; revised by J. Barnes). Princeton

University Press.
Armstrong, D. M. 2004. Truth and Truthmakers, Cambridge University Press.
Atkinson, D. 2003. ‘Experiments and Thought Experiments in Natural Science,’ in
M. C. Galavotti (ed.), Observation and Experiment in the Natural and Social
Sciences, Boston Studies in the Philosophy of Science 232. Dordrecht: Kluwer,
pp. 209–225.

Audi, P. 2012. ‘A Clarification and Defense of the Notion of Grounding,’ in

F. Correia and B. Schnieder (eds.), Metaphysical grounding. Cambridge
University Press, pp. 101–121.
Barrow, J. D. 2001. ‘Cosmology, Life, and the Anthropic Principle,’ Annals of the
New York Academy of Sciences950, 139–153.
Bealer, G. 1992. ‘The Incoherence of Empiricism,’ Aristotelian Society Supplementary
Volume 66, 99–138.
1998. ‘Intuition and the Autonomy of Philosophy,’ in M. DePaul and W.
Ramsey (eds.), Rethinking Intuition: The Psychology of Intuition and Its Role in
Philosophical Inquiry. Lanham, MD: Rowman and Littlefield, pp. 201–240.
2002. ‘Modal Epistemology and the Rationalist Renaissance,’ in T. S. Gendler
and J. Hawthorne (eds.), Conceivability and Possibility. Oxford University Press,
pp. 71–125.
2004. ‘The Origins of Modal Error,’ Dialectica 58.1, 11–42.
Bennett, K. 2009. ‘Composition, Colocation and Metaontology.’ in D. Chalmers,
D. Manley, and R. Wasserman (eds.), Metametaphysics. Oxford University
Press, pp. 38–76.
2011. ‘By Our Bootstraps,’ Philosophical Perspectives 25, 27–41.
Benovsky, J. 2013. ‘From Experience to Metaphysics: On Experience-based
Intuitions and their Role in Metaphysics,’ Noûs, (online) (forthcoming).
Ben-Yami, H. 2004. Logic and Natural Language. Surrey: Ashgate.

244 Bibliography

2009. ‘Plural Quantification Logic: A Critical Appraisal,’ Review of Symbolic Logic

2.1, 208–232.
Berto, F. 2011. ‘Modal Meinongianism and Fiction: The Best of Three Worlds,’
Philosophical Studies 152, 313–334.
Berto, F. and M. Plebani. 2015. Ontology and Metaontology: A Contemporary Guide.

London: Bloomsbury.
Bird, A. 2007. Nature’s Metaphysics. Oxford University Press.
Bliss, R. L. 2014. ‘Viciousness and Circles of Ground,’Metaphilosophy 45.2, 245–256.
2013. ‘Viciousness and the Structure of Reality,’ Philosophical Studies 166.2,
Bliss, R. L. and K. Trogdon. 2014. ‘Metaphysical Grounding,’ in E. N. Zalta (ed.),
The Stanford Encyclopedia of Philosophy (Winter 2014 edn), see http://plato.
Bohm, D. 1984. Causality and Chance in Modern Physics, (Second edn). London:
BonJour, L. 1998. In Defense of Pure Reason. Cambridge University Press.
Booth, A. R. and D. P. Rowbottom, 2014. (eds.). Intuitions. Oxford University Press.
Bricker, P. 2014. ‘Ontological Commitment,’ in E. N. Zalta (ed.), The Stanford
Encyclopedia of Philosophy (Winter 2014 edn), see
Callender, C. 2011. ‘Philosophy of Science and Metaphysics,’ in S. French and
J. Saatsi (eds.), The Continuum Companion to the Philosophy of Science. London:
Continuum, pp. 33–54.
Cameron, R. P. 2008. ‘Turtles All the Way Down: Regress, Priority and
Fundamentality,’ Philosophical Quarterly 58, 1–14.
Cappelen, H. 2012. Philosophy without Intuitions. Oxford University Press.
Carnap, R. 1931. ‘The Elimination of Metaphysics Through Logical Analysis of
Language’ (A. Pap, trans.), in A. J. Ayer (ed.), Logical positivism. New York: The
Free Press, 1959, pp. 60–81; srcinally published in Erkenntnis 2.1 as
‘Überwindung der Metaphysik durch logische Analyse der Sprache,’
pp. 219–241.
1950. ‘Empiricism, Semantics, and Ontology.’ Revue Internationale de Philosophie
4; reprinted in his Meaning and Necessity: A Study in Semantics and Modal Logic.
University of Chicago Press, 1956.
Casullo, A. 2003. A Priori Justification. Oxford University Press.
2013. ‘Four Challenges to the A Priori–A Posteriori Distinction,’Synthese, (online)
Casullo, A. and J. C. Thurow 2013. (eds.), The A Priori in Philosophy. Oxford
University Press.
Bibliography 245

Chalmers, D. 1996. The Conscious Mind: In Search of a Fundamental Theory. Oxford

University Press.
2002. ‘Does Conceivability Entail Possibility?’, in T. S. Gendler and J. Hawthorne
(eds.), Conceivability and Possibility. Oxford University Press, pp. 145–200.
2006. ‘The Foundations of Two-Dimensional Semantics,’ in M.

Garcia-Carpintero and J. Macia (eds.), Two-Dimensional Semantics: Foundations

and Applications. Oxford University Press, pp. 55–140.
2009. ‘Ontological Anti-Realism,’ in D. Chalmers, D. Manley, and R. Wasserman
(eds.), Metametaphysics. Oxford University Press, pp. 77–129.
2011. ‘Verbal Disputes,’ Philosophical Review 120.4, 515–566.
2014. ‘Intuitions in Philosophy: A Minimal Defense,’ Philosophical Studies 171.3,
Chalmers, D., D. Manley, and R. Wasserman 2009. (eds.). Metametaphysics. Oxford
University Press.
Chudnoff, E. 2013. Intuition. Oxford University Press.
Cohnitz, D. 2006. ‘When are Discussions of Thought Experiments Poor Ones?
A Comment on Peijnenburg and Atkinson,’ Journal for General Philosophy of
Science 37.2, 373–392.
Cornell, E. A. and C. E. Wieman. 1998. ‘The Bose–Einstein Condensate,’ Scientific
American 278.3, 40–45.
Correia, F. 2008. ‘Ontological Dependence,’ Philosophy Compass 3, 1013–1032.
2010. ‘Grounding and Truth-functions,’ Logique & Analyse 211, 251–279.
Correia, F. and B. Schnieder. 2012. ‘Grounding: An Opinionated Introduction,’
in F. Correia and B. Schnieder (eds.), Metaphysical Grounding. Cambridge
University Press, pp. 1–36.
2012. (eds.). Metaphysical Grounding: Understanding the Structure of Reality.
Cambridge University Press.
Crane, T. 2013. The Objects of Thought. Oxford University Press.
Creath, R. ‘Logical Empiricism,’ in E. N. Zalta (ed.), The Stanford Encyclopedia of
Philosophy (Spring 2014 edn), see
Daly, C. 2010. Introduction to Philosophical Methods. Peterborough, ON: Broadview
2012. ‘Scepticism about Grounding,’ in F. Correia and B. Schnieder (eds.),
Metaphysical Grounding. Cambridge University Press, pp. 81–100.
Daly, C. and D. Liggins. 2014. ‘In Defence of Existence Questions,’ Monist 97.7,
Davidson, D. 1987. ‘Knowing One’s Own Mind,’ Proceedings and Addresses of the
American Philosophical Association 60.3, 441–458.
246 Bibliography

de Cruz, H. 2015. ‘Where Philosophical Intuitions Come From,’ Australasian

Journal of Philosophy 93.2, 233–249.
Dehmelt, H. 1989. ‘Triton,… Electron,… Cosmon,…: An Infinite Regression?’,
Proceedings of the National Academy of Sciences 86, 8618–8619.
Dennett, D. 2013. Intuition Pumps and Other Tools for Thinking. New York: W.

W. Norton & Co.

deRosset, L. 2013. ‘What Is Weak Ground?’, Essays in Philosophy 14.1, Article 2.
Dixon, S. ‘What Is the Well-Foundedness of Grounding?’, Mind (forthcoming).
Dorato, M. and M. Morganti. 2013. ‘Grades of Individuality. A Pluralistic View
of Identity in Quantum Mechanics and in the Sciences,’ Philosophical Studies
163, 591–610.
Dorr, C. 2005. ‘What We Disagree about When We Disagree about Ontology,’ in
M. E. Kalderon (ed.), Fictionalism in Metaphysics. Oxford University Press, pp.
2010. ‘Review of Every Thing Must Go,’ Notre Dame Philosophical Reviews, seehttps://
Dummett, M. 1981. Frege. Philosophy of Language, (Second edn). Cambridge,
MA: Harvard University Press.
Eklund, M. 2009. ‘Carnap and Ontological Pluralism,’ in D. Chalmers, D.
Manley, and R. Wasserman (eds.), Metametaphysics. Oxford University Press,
pp. 130–156.
2011. ‘Review of Quantifier Variance and Realism:Essays in Metaontology,’ Notre
Dame Philosophical Reviews, see
2013. ‘Carnap’s Metaontology,’ Noûs 47.2, 229–249.
2014. ‘Rayo’s Metametaphysics,’ Inquiry: An Interdisciplinary Journal of Philosophy
57.4, 483–497.
Enderton, H. B. 2012. ‘Second-order and Higher-order Logic,’ in E. N. Zalta (ed.),
The Stanford Encyclopedia of Philosophy (Fall 2012 edn), see http://plato.stan-
Fine, K. 1994a. ‘Essence and Modality,’ Philosophical Perspectives 8, 1–16.
1994b. ‘OntologicalDependence,’ Proceedings of the Aristotelian Society 95, 269–290.
1995. ‘Logic of Essence,’ Journal of Philosophical Logic 24, 241–273.
2001. ‘The Question of Realism,’ Philosophers Imprint 1, 1–30.
2009. ‘The Question of Ontology,’ in D. Chalmers, D. Manley, and R.
Wasserman (eds.), Metametaphysics. Oxford University Press, pp. 157–177.
2010. ‘Towards a Theory of Part,’ Journal of Philosophy 107.11, 559–589.
2012. ‘Guide to Ground,’ in F. Correia and B. Schnieder (eds.), Metaphysical
Grounding. Cambridge University Press, pp. 37–80.
Bibliography 247

2012. ‘What Is Metaphysics?’, in T. E. Tahko (ed.), Contemporary Aristotelian

Metaphysics. Cambridge University Press, pp. 8–25.
Fiocco, M. O. 2007. ‘Conceivability, Imagination, and Modal Knowledge,’
Philosophy and Phenomenological Research, 74.2, 364–380.
Fischer, R. W. and F. Leon (eds.). 2015. Modal Epistemology After Rationalism,

Synthese Library. Dordrecht: Springer (forthcoming).

French, S. 2011. ‘Metaphysical Underdetermination: Why Worry?’, Synthese
180, 205–221.
2014. The Structure of the World:Metaphysics and Representation. Oxford University
Gendler, T. S. and J. Hawthorne (eds.). 2002. Conceivability and Possibility. Oxford
University Press.
Georgi, H. 1989. ‘Effective Quantum Field Theories,’ in P. Davies (ed.), The New
Physics. Cambridge University Press, pp. 446–457.
Gibbard, A. 1975. ‘Contingent Identity,’Journal of Philosophical Logic 4, 187–221.
Hackett, J. 2007. ‘Locality and Translations in Braided Ribbon Networks,’
Classical and Quantum Gravity 24.23, 5757–5766.
Hale, B. 2013. Necessary Beings: An Essay on Ontology, Modality, and the Relations
Between Them. Oxford University Press.
Haug, M. C. (ed.). 2014. Philosophical Methodology: The Armchair or the Laboratory?
Abingdon: Routledge.
Hawley, K. 2006. ‘Science as a Guide to Metaphysics?’, Synthese 149.3, 451–470.
Hawthorne, J. 2008. ‘Three-dimensionalism vs. Four-Dimensionalism,’ in T.
Sider, J. Hawthorne and D. W. Zimmerman (eds.), Contemporary Debates in
Metaphysics. Oxford: Blackwell Publishing, pp. 263–282.
Heil, J. 2003. From an Ontological Point of View. Oxford: Clarendon Press.
2012. The Universe As We Find It. Oxford University Press.
Hirsch, E. 2002. ‘Quantifier Variance and Realism,’ Philosophical Issues 12, 51–73.
2005. ‘Physical-Object Ontology, Verbal Disputes, and Common Sense,’
Philosophy and Phenomenological Research 70, 67–97.
2009. ‘Ontology and Alternative Languages,’ in D. Chalmers, D. Manley, and
R. Wasserman (eds.),Metametaphysics. Oxford University Press, pp. 231–259.
Hofweber, T. 2009. ‘Ambitious, Yet Modest, Metaphysics,’ in D. Chalmers, D.
Manley, and R. Wasserman (eds.), Metametaphysics. Oxford University Press,
pp. 260–289.
Horvath, J. and T. Grundmann (eds.). 2012. Experimental Philosophy and Its Critics.
London: Routledge.
Hüttemann, A. and D. Papineau. 2005. ‘Physicalism Decomposed,’ Analysis 65,
248 Bibliography

Jackson, F. 1986. ‘What Mary Didn’t Know,’ The Journal of Philosophy 83,
1998. From Metaphysics to Ethics: A Defence of Conceptual Analysis. Oxford: Clarendon
Jenkins, C. S. 2010. ‘Concepts, Experience and Modal Knowledge,’ Philosophical
Perspectives 24, 255–279.
2010. ‘What Is Ontological Realism?’, Philosophy Compass 5/10, 880–890.
2011. ‘Is Metaphysical Dependence Irreflexive?’, The Monist 94.2, 267–276.
2014. ‘Intuition, “Intuition”, Concepts and the A Priori,’ in A. R. Booth and D.
P. Rowbottom (eds.), Intuitions. Oxford University Press, pp. 91–115.
2014. ‘Merely Verbal Disputes,’ Erkenntnis 79.1, 11–30.
Kim, J. 2010. Essays in the Metaphysics of Mind. Oxford University Press.
Koslicki, K. 2005. ‘On the Substantive Nature of Disagreements in Ontology,’
Philosophy and Phenomenological Research 71, 85–105.
2012. ‘Varieties of Ontological Dependence,’ in F. Correia and B. Schnieder
(eds.), Metaphysical Grounding. Cambridge University Press, pp. 186–213.
Kripke, S. 1980. Naming and Necessity . Cambridge, MA: Harvard University
2013. Reference and Existence. Oxford University Press.
Ladyman, J. 1998. ‘What Is Structural Realism?’, Studies in History and Philosophy
of Science Part A 29, 409–424.
2012. ‘Science, Metaphysics and Method,’ Philosophical Studies 160.1, 31–51.
Ladyman, J. and D. Ross (with D. Spurrett and J. Collier). 2007. Every Thing Must
Go. Oxford University Press.
Lange, M. 2008. ‘Could the Laws of Nature Change?’, Philosophy of Science 75.1,
Lewis, D. 1990. ‘Noneism or Allism?’, Mind 99.393, 23–31.
Lowe, E. J. 2006. The Four-Category Ontology: A Metaphysical Foundation for Natural
Science. Oxford: Clarendon Press.
2008. ‘Two Notions of Being: Entity and Essence,’ Royal Institute of Philosophy
Supplements 83.62, 23–48.
2010. ‘Ontological Dependence,’ in E. N. Zalta (ed.), The Stanford Encyclopedia of
Philosophy (Spring 2010 edn), see
2011. ‘Locke on Real Essence and Water as a Natural Kind: A Qualified
Defence,’ Aristotelian Society Supplementary Volume 85, 16–17.
2011. ‘The Rationality of Metaphysics,’ Synthese 178.1, 99–109.
2012. ‘What Is the Source of Our Knowledge of Modal Truths?’, Mind 121,
Bibliography 249

2014. ‘Essence vs. Intuition: An Unequal Contest,’ in A. R. Booth and D. P.

Rowbottom (eds.), Intuitions. Oxford University Press, pp. 256–268.
Lowe, E. J. and S. McCall. 2006. ‘3D/4D Controversy: A Storm in a Teacup,’ Noûs
40, 570–578.
Markosian, N. 2005. ‘Against Ontological Fundamentalism,’ Facta Philosophica 7,

2014. ‘Time,’ in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Spring
2014 edn), see
Marcus, R.B. 1967. ‘Essentialism in Modal Logic,’ Noûs 1.1, pp. 91–6.
Marsh, G. 2010. ‘Is the Hirsch–Sider Dispute Merely Verbal?’, Australasian Journal
of Philosophy 88.3, 459–469.
Maudlin, T. 2007. The Metaphysics Within Physics. Oxford University Press.
McDaniel, K. 2013. ‘Degrees of Being,’ Philosophers’ Imprint 13.19, 1–18.
‘Ways of Being,’ in D. Chalmers, D. Manley, and R. Wasserman (eds.), 2009.
Metametaphysics. Oxford University Press, pp. 290–319.
McKenzie, K. 2014. ‘Priority and Particle Physics: Ontic Structural Realism as a
Fundamentality Thesis,’ British Journal for the Philosophy of Science 65, 353–380.
McTaggart, J. M. E. 1993. ‘The Unreality of Time,’ Mind 17, 1908, 457–473;
reprinted in R. Le Poidevin and M. McBeath (eds.), The Philosophy of Time.
Oxford University Press, pp. 23–34.
Merricks, T. 2001. Objects and Persons. Oxford University Press.
Miller, K. 2005. ‘The Metaphysical Equivalence of Three and Four
Dimensionalism,’ Erkenntnis 62.1, 91–117.
Morganti, M. 2013.Combining Science and Metaphysics. New York: Palgrave Macmillan.
2014. ‘Metaphysical Infinitism and the Regress of Being,’ Metaphilosophy 45.2,
2015. ‘Dependence, Justification and Explanation: Must Reality Be
Well-Founded?’, Erkenntnis 60.3, 555–572.
Mulligan, K., P. Simons and B. Smith. ‘Truth-Makers,’ 1984. Philosophy and
Phenomenological Research 44.3, 287–321.
Nelson, M. ‘Existence,’ in E. N. Zalta (ed.), 2012. The Stanford Encyclopedia of
Philosophy (Winter 2012 edn), see
Ney, A. 2014. Metaphysics: An Introduction. Abingdon: Routledge.
Nolan, D. 1997. ‘Impossible Worlds: A Modest Approach,’ Notre Dame Journal of
Formal Logic 38.4, 535–572.
2015. ‘The A Posteriori Armchair,’Australasian Journal of Philosophy 93.2, 211–231.
O’Conaill, D. 2014. ‘Ontic Structural Realism and Concrete Objects,’ The
Philosophical Quarterly 64.255, 284–300.
250 Bibliography

Oppenheim, P. and H. Putnam. 1958. ‘Unity of Science as a Working Hypothesis,’

in H. Feigl et al. (eds.), Concepts, Theories, and the Mind-Body Problem, Minnesota
Studies in the Philosophy of Science Vol. II. Minneapolis: University of
Minnesota Press, pp. 3–36.
Papineau, D. 2012. Philosophical Devices: Proofs, Probabilities, Possibilities, and Sets.

Oxford University Press.

Parsons, T. 1980. Nonexistent Objects. New Haven, CO: Yale University Press.
Paul, L. A. 2004. ‘The Context of Essence,’ Australasian Journal of Philosophy 82,
2010. ‘Temporal Experience,’ Journal of Philosophy 107.7, 333–359.
2012. ‘Metaphysics as Modeling: The Handmaiden’s Tale,’Philosophical Studies
160, 1–29.
Peijnenburg, J. and D. Atkinson. 2003. ‘When Are Thought Experiments Poor
Ones?’, Journal for General Philosophy of Science 34.2, 305–322.
Plato. Phaedrus (R. Hackforth, trans.). 1972. Cambridge University Press.
Priest, G. 2005. Towards Non-Being:The Logic and Metaphysics of Intentionality. Oxford
University Press.
2008. ‘The Closing of the Mind: How the Particular Quantifier Became
Existentially Loaded Behind Our Backs,’ The Review of Symbolic Logic 1.1,
Pust, J. 2014. ‘Intuition,’ in E. N. Zalta (ed.),
The Stanford Encyclopedia of Philosophy(Fall
2014 edn), see
Putnam, H. 1979. ‘The Meaning of “Meaning,”’ reprinted in his Mind, Language
and Reality: Philosophical Papers , Vol. 2. Cambridge University Press, pp.
1990. ‘Is Water Necessarily H 2O?’, in J. Conant (ed.), Realism with a Human Face.
Cambridge, MA: Harvard University Press, pp. 54–79.
Pylkkänen, P., B. J. Hiley and I. Pättiniemi. ‘Bohm’s Approach and Individuality,’
in A. Guay and T. Pradeu (eds.) 2015. Individuals Across the Sciences. Oxford
University Press, Ch. 12.
Quine, W. V. 1969. ‘Existence and Quantification,’ in his Ontological Relativity and
Other Essays New York: Columbia University Press.
1980. ‘On What There Is,’ The Review of Metaphysics 2, 1948, 21–38; reprinted
in his From a Logical Point of View. Cambridge, MA: Harvard University Press,
pp. 1–19.
1986. Philosophy of Logic, (Second edn). Cambridge, MA: Harvard University
Raven, M. 2013. ‘Is Ground a Strict Partial Order?’, American Philosophical Quarterly
50.2, 191–199.
Rayo, A. 2007. ‘Ontological Commitment,’ Philosophy Compass 2/3, 2013. 428–444.
Bibliography 251

2013. Construction of Logical Space. Oxford University Press.

Reicher, M. ‘Nonexistent Objects,’ in E. N. Zalta (ed.), The Stanford Encyclopedia
of Philosophy (Summer 2014 edn), see
Ritchie, J. 2008. Understanding Naturalism. Stocksfield: Acumen.

Roca-Royes, S. 2011. ‘Conceivability and De Re Modal Knowledge,’ Noûs

45.1, 22–49.
Rodriguez-Pereyra, G. 2005. ‘Why Truthmakers?’, in H. Beebee and J. Dodd
(eds.), Truthmakers: The Contemporary Debate. Oxford University Press, 2011,
pp. 17–31.
‘Grounding Is Not a Strict Order,’ Journal of the American Philosophical Association
Rose, D. and D. Danks. 2012. ‘Causation: Empirical Trends and Future Directions,’
Philosophy Compass 7.9, 643–653.
Rosen, G. 2010. ‘Metaphysical Dependence: Grounding and Reduction,’ in B.
Hale and A. Hoffman (eds.), Modality: Metaphysics, Logic, and Epistemology.
Oxford University Press, pp. 109–135.
Routley, R. ‘On What There Is Not,’ 1982. Philosophy and Phenomenological Research
43, 151–178.
Rowbottom, D. P. 2014. ‘Intuitions in Science: Thought Experiments as
Argument Pumps,’ in A. R. Booth and D. P. Rowbottom (eds.), Intuitions.
Oxford University Press, pp. 119–134.
Rueger, A. and P. McGivern. 2010. ‘Hierarchies and Levels of Reality,’ Synthese
176, 379–397.
Russell, B. 1905. ‘On Denoting,’ Mind 14, 479–493.
Ryckman, T. A. ‘Early Philosophical Interpretations of General Relativity,’ in
E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Spring 2014 edn), see
Saunders, S., J. Barrett, A. Kent, and D. Wallace (eds.). 2010. Many Worlds? Everett,
Quantum Theory, and Reality. Oxford University Press.
Schaffer, J. 2003. ‘Is There a Fundamental Level?’, Noûs 37, 498–517.
2010. ‘Monism: The Priority of the Whole,’ Philosophical Review 119, 31–76.
‘On What Grounds What,’ in D. Chalmers, D. Manley, and R. Wasserman
(eds.), 2009. Metametaphysics. Oxford University Press, pp. 347–383.
2012. ‘Grounding, Transitivity, and Contrastivity,’ in F. Correia and
B. Schnieder,Metaphysical Grounding. Cambridge University Press, pp. 122–138.
Schnieder, B. 2006. ‘Truth-Making Without Truth-Makers,’ Synthese 152.1,
Schrödinger, E. 1935. ‘Discussion of Probability Relations Between Separated
Systems,’ Proceedings of the Cambridge Philosophical Society 31, 555–563.
252 Bibliography

Sidelle, A. 2002. ‘On the Metaphysical Contingency of Laws of Nature,’ in T.

S. Gendler and J. Hawthorne (eds.), Conceivability and Possibility. Oxford
University Press, pp. 309–336.
2009. ‘Conventionalism and the Contingency of Conventions,’ Noûs 43.2,

2010. ‘Modality and Objects,’ The Philosophical Quarterly 60.238, 109–125.

Sider, T. Ontological Realism,’ in D. Chalmers, D. Manley, and R. Wasserman
(eds.), 2009. Metametaphysics. Oxford University Press, pp. 384–423.
2011. Writing the Book of the World. Oxford University Press.
2014. ‘Hirsch’s Attack on Ontologese,’ Noûs 48.3, 565–572.
Simons, P. 1987. Parts: A Study in Ontology. Oxford: Clarendon Press.
Smith, B. and A. C. Varzi. 1997. ‘Fiat and Bona Fide Boundaries,’ Philosophy and
Phenomenological Research 60.2, 401–420.
Smith, B. and K. Mulligan. 1983. ‘Framework for Formal Ontology,’Topoi 3, 73–85.
Smolin, L. 2006. The Trouble with Physics. London: Penguin Books.
Sorensen, R. 1992. Thought Experiments. Oxford University Press.
Sosa, E. 2014. ‘Intuitions: Their Nature and Probative Value,’ in A. R. Booth and
D. P. Rowbottom (eds.), Intuitions. Oxford University Press, pp. 36–49.
Stoljar, D. ‘Physicalism,’ in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy
(Spring 2015 edn, see
Su, G. and M. Suzuki. 1999. ‘Towards Bose–Einstein Condensation of Electron
Pairs: Role of Schwinger Bosons,’ International Journal of Modern Physics B 13.8,
Szabó, Z. G. 2003. ‘Believing in Things,’ Philosophy and Phenomenological Research
66, 584–611.
Tahko, T. E. 2008. ‘A New Definition ofA Priori Knowledge: In Search of a Modal
Basis,’ Metaphysica 9.2, 57–68.
2011. ‘ A Priori and A Posteriori: A Bootstrapping Relationship,’ Metaphysica 12.2,
2012. ‘Boundaries in Reality,’ Ratio 25.4, 405–424.
(ed.). 2012. Contemporary Aristotelian Metaphysics. Cambridge University Press.
2012. ‘Counterfactuals and Modal Epistemology,’ Grazer Philosophische Studien
86, 93–115.
2012. ‘In Defence of Aristotelian Metaphysics,’ in T. E. Tahko (ed.), Contemporary
Aristotelian Metaphysics. Cambridge University Press, pp. 26–43.
2013. ‘Truth-Grounding and Transitivity,’ Thought: A Journal of Philosophy 2.4,
2014. ‘Boring Infinite Descent,’ Metaphilosophy 45.2, 257–269.
2015. ‘Natural Kind Essentialism Revisited,’ Mind 124.495, 2015, 795–822.
Bibliography 253

2015. ‘The Modal Status of Laws: In Defence of a Hybrid View,’ The Philosophical
Quarterly, (online) (forthcoming).
‘Empirically-Informed Modal Rationalism,’ in R. W. Fischer and F. Leon (eds.),
Modal Epistemology After Rationalism, Synthese Library. Dordrecht: Springer

Tahko, T. E. and E. J. Lowe. ‘Ontological Dependence,’ in E. N. Zalta (ed.), The

Stanford Encyclopedia of Philosophy (Spring 2015 edn), see http://plato.stan-
Tahko, T. E. and D. O’Conaill. 2015. ‘Minimal Truthmakers,’ Pacific Philosophical
Quarterly, (online) (forthcoming).
Tallant, J. 2013. ‘Intuitions in Physics,’ Synthese 190, 2959–2980.
Thomas, C. 2008. ‘Speaking of Something: Plato’s Sophist and Plato’s Beard,’
Canadian Journal of Philosophy 38.4, 631–667.
Thomasson, A. 2007. Ordinary Objects. Oxford University Press.
2009. ‘Answerable and Unanswerable Questions,’ in D. Chalmers, D. Manley,
and R. Wasserman (eds.), Metametaphysics. Oxford University Press, 444–471.
2009. ‘The Easy Approach to Ontology,’ Axiomathes 19, 1–15.
2015. Ontology Made Easy. Oxford University Press.
Trogdon, K. 2009. ‘Monism and Intrinsicality,’ Australasian Journal of Philosophy
87, 127–148.
2013. ‘Grounding: Necessary or Contingent?’, Pacific Philosophical Quarterly 94,
2013. ‘An Introductionto Grounding,’ in M. Hoeltje,B. Schnieder, andA. Steinberg
(eds.), Varieties of Dependence. Munich:Philosophia Verlag, pp. 97–122.
Vaidya, A. 2010. ‘Understanding and Essence,’ Philosophia 38, 811–833.
2012. ‘Intuition and Inquiry,’ Essays in Philosophy 13.1, Article 16.
Van Brakel, J. 2010. ‘Chemistry and Physics: No Need for Metaphysical Glue,’
Foundations of Chemistry 12.2, 123–136.
Van Fraassen, B. C. 2002. The Empirical Stance . New Haven: Yale University
Van Inwagen, P. 1990. Material Beings. Ithaca, NY: Cornell University Press.
1998. ‘Meta-ontology,’ Erkenntnis 48, 233–250.
2009. ‘Being, Existence, and Ontological Commitment,’ in D. Chalmers, D.
Manley, and R. Wasserman (eds.), Metametaphysics. Oxford University Press,
pp. 472–506.
Varzi, A. C. 2011. ‘Boundaries, Conventions, and Realism,’ in J. K. Campbell, M.
O’Rourke, and M. H. Slater (eds.), Carving Nature at Its Joints: Natural Kinds in
Metaphysics and Science. Cambridge, MA: MIT Press, pp. 129–153.
Von Solodkoff, T. and R. Woodward. 2013. ‘Noneism, Ontology, and
Fundamentality,’ Philosophy and Phenomenological Research 87.3, 558–583.
254 Bibliography

Williamson, T. 2007. The Philosophy of Philosophy. Oxford: Blackwell Publishing.

2013. ‘How Deep is the Distinction Between A Priori and A Posteriori
Knowledge?’ in A. Casullo and J. C. Thurow (eds.), The A Priori in Philosophy.
Oxford University Press, pp. 291–312.
Wilson, J. 2014. ‘No Work for a Theory of Grounding,’ Inquiry 57.5–6, 1–45.

Woodward, R. 2013. ‘Towards Being,’ Philosophy and Phenomenological Research

86.1, 183–193.
Yablo, S. 1993. ‘Is Conceivability a Guide to Possibility?’, Philosophy and
Phenomenological Research, 53, 1–42.
2009. ‘Must Existence-Questions have Answers?’, in D. Chalmers, D. Manley,
and R. Wasserman (eds.), Metametaphysics. Oxford University Press,
pp. 507–525.

a posteriori necessity, 80, 154, 164 constitutive facts, 161–63, 229, 230, 236
allism, 22, 236. See also noneism conventional, conventionalism,9, 63, 65,
analyticity, 27, 73, 80, 113, 155, 183, 191 76–83, 86, 89, 90, 169, 233, 236, 238,
application conditions, 74–76, 236 240
apriority, 10, 11, 80, 146, 151, 152–55, Correia, Fabrice, 108
159, 164, 168–76, 178, 180, 181, 190, counterfactuals, 81, 82, 160, 161, 162, 167,
193, 203, 206, 217, 231, 232 229, 230, 231, 236
Aristotle, Aristot elian, 3, 4, 9, 46, 57, Crane, Tim, 23, 25–27, 44, 46, 47, 62

64, 67, 93, 100, 101, 103, 127, 128,

163, 195, 196, 206, 207, 208, 209, definite descriptions, 19, 49
240, 241 Dennett, Daniel, 177
armchair metaphysics, 152, 169, 170, 171, dependence. See ontological dependence
193, 194, 206 Dorr, Cian,32, 33, 34, 35, 220
Armstrong, D. M., 137
atomism, 120, 125–30, 133, 134, 136, 139, 'easy' ontology,65, 66, 73, 74, 75, 91
142, 236, 238 Einstein, Albert, 88, 178, 196
Eklund, Matti, 29, 37
Bealer, George, 156, 178, 190–93 emergence, emergent, 143, 146, 208, 237
Bennett, Karen, 68–71, 72, 77, 89, 135 endurantism and perdurantism, 53–55,
Bohm, David,129, 141 187, 237. See also three- and four-
Cameron, Ross, 130–33 epistemicism, 69, 70, 71, 72, 77, 83
Carnap, Carnapian, 8, 13–15, 20, 27, 28, essence, 82, 95, 98, 100–4, 115, 152, 156,
29, 31, 32, 33, 35, 37, 38, 41, 49, 50, 157, 162–66, 169, 173, 174, 225, 232,
51, 53, 64, 75, 76, 203, 211, 237, 238 233, 234, 236, 237, 239, 240
causation, 9, 56, 93, 94, 108, 111, 113, essentialism, 100, 156, 235, 237
114, 144, 185, 186, 187, 218, 229, 230, existence, 6, 8, 10, 13, 15, 16–29, 32,
235, 238 35, 37, 39, 40, 41, 42, 44, 46, 47,
Chalmers, David, 5, 14, 32, 68, 73, 156, 51, 57–62, 69, 73, 74, 75, 85, 91,
158, 184 93, 95–100, 102, 104, 110, 112, 120,
chemical substance, 125, 174, 180, 202, 131, 136, 144, 166, 210, 221, 237,
233 240
conceptual analysis, 169, 182, 183, 200, external questions. See internal and
206, 217, 236, 237 external questions

256 Index

factivity, 29, 35 Ladyman, James, 11, 123, 125–27, 128,

fallibilism, 85, 167, 173, 182, 191, 192, 129, 132, 133, 137, 140, 141, 142, 144,
215, 218, 222, 237 193, 194, 213, 217–23, 228
Fine, Kit, 8, 40, 57–63, 67, 71, 73, 98, 101, language pluralism, 8, 13, 15, 29, 35–39,
102, 103, 104, 105, 106, 107, 108, 113, 53, 76, 79, 238
115, 116, 118, 139, 156, 162, 163, 165, levels (of reality), 9, 83, 107, 120–50, 152,
225 221, 238
fundamental physics, 7, 86, 124, 129, 143, Lewis, David, 21, 22, 23, 240
186, 205, 208, 219, 222 logical empiricism. See logical positivism
fundamentality, 9, 11, 34, 41, 43, 62, 79, logical positivism, 14, 155, 203, 204, 205,
83, 85, 86, 91, 105, 110, 115, 119, 219
120–51, 204, 208, 218, 221, 223, 226, Lowe, E. J., 5, 98, 101, 102, 104, 105, 115,
229, 238, 239, 241, 242 156, 163, 164, 165, 166, 169, 170, 172,
173, 216, 217, 218, 225, 231, 232, 233,
general relativity, 132, 215 234, 235
ground, grounding, 9, 62, 91, 93, 94,
104–19, 127, 130, 134, 135, 138, 140, Markosian, Ned, 130
156, 165, 227, 231, 238, 242 Maudlin, Tim, 204
gunk, 128, 130, 132, 133, 134, 138, 238 McDaniel, Kris, 127
McTaggart, John, 20
Hawley, Katherine, 220, 221 Meinong, Alexius, 8, 15, 20–27, 39, 46,
Hawthorne, John, 131, 132 240
Heil, John, 147 Meinongianism, 20–27, 39, 41
Hiley, Basil J., 130 mentality, mental states, 93, 113, 114,
Hirsch, Eli, 8, 9, 14, 36, 40, 50–57, 62, 65, 116, 153, 154, 198
66, 68, 69, 71, 72, 73, 78 mereological atomism. See atomism
Hofweber, Thomas,30–32, 36 mereology, 9, 33, 55, 120, 238, See also
identity (criteria of), 26, 77, 78, 100, 101, metaontology, 1, 3–6, 7, 8, 11, 15, 16, 17,
165, 166, 173, 175, 222 18, 24, 25, 35, 38, 39, 40, 43, 49, 50,
infinite regress. See regress 53, 54, 57, 59, 60, 62, 63, 70
infinitism, 10, 122, 133, 140, 147, 148, 239 metaphysical infinitism. See infinitism
internal and external questions, 8, 15, microstructure, 81, 115, 173–76, 180, 183,

27–35, 37, 76, 237, 238 202

intuitions, 10, 79, 112, 131, 132, 148, 156, Miller, Kristie, 54
172, 177–94, 198, 199, 202, 203, 210, modal empiricism,10, 152, 166, 167–72,
217, 221, 232, 237, 239, 240. See also 239
rational intuition modal rationalism, 10, 152, 155–63, 166,
167, 168, 172, 232, 239
Jackson, Frank, 200, 201 modality
Jenkins, Carrie, 68, 167–70, 181 conceptual, 157, 191, 236
joint-carving, 50, 57, 66, 68, 77, 79, 83–90, metaphysical, 95, 97, 115, 146, 158, 165,
138, 140, 205, 238 174, 175, 176, 180, 210, 224, 230, 236,
Kim, Jaegwon, 132 physical, 144, 239
Koslicki, Kathrin, 102, 103, 104 monism, 128, 133, 137, 208, 239. See also
Kripke, Saul, 163, 173 pluralism
Index 257

natural kinds, 89, 166, 169, 172, 173, 174, principle of charity, 52, 240
205, 232, 237, 239 principle of independence, 21, 23, 25, 240
natural necessity. See modality:physical Putnam, Hilary, 81, 124, 125, 173, 177,
naturalism, 85, 191, 207, 212, 225, 239 179, 180, 183, 200
necessity. See modality
nihilism (about composition), 34, 35, 55, quantification, quantifiers, 4, 6, 8, 14,
187, 239, See also universalism (about 16–18, 19, 21, 23, 26, 27, 30, 34, 35,
composition) 36, 39–52, 56, 57, 58, 59, 60, 61, 62,
Nolan, Daniel, 169–72 66, 68, 69, 85, 228
nominalism, 17, 53 quantifier variance, 8, 9, 36, 40, 42, 49–57,
nomological necessity.See modality: 66, 68, 69, 73, 75, 77, 240
physical quantum field theory, 132, 148
noneism, 22, 23, 24, 25, 39, 236, See quantum mechanics, 28, 125, 129, 132,
also allism 148, 197, 204, 214, 221, 222
Quine, Quinean, 4, 5, 8, 13–27, 30, 35, 38,
ontic structural realism. See structuralism 39, 40, 41, 42, 44, 45, 46, 47, 48, 49,
Ontologese, 32, 49, 56, 239 50, 57, 59, 64, 138, 155, 191, 192, 203,
ontological anti-realism, 9, 64, 72, 76 216, 217, 228
ontological commitment, 4, 6, 8, 14, 16,
18, 19, 23, 39–49, 58, 63, 64, 227 rational intuition, 11, 190–94, 240, See also
ontological deflationism, 9, 14, 36, 62, 64, intuitions
65, 66, 67, 71–76, 78, 79, 84, 85, 90, Rayo, Augustín, 43, 44, 45, 47, 51
91, 92, 236 real (metaphysically robust), 58, 60, 61,
ontological dependence 233
essential, 95, 98, 100, 101–4, 165, 237 real definition, 101, 104, 163, 164, 165,
generic existential, 95, 96, 97, 135, 237, 240
146 reduction, reductionism, 9, 13, 52, 94,
identity-dependence, 95, 98–101, 104, 102, 113, 114–16, 124–26, 152, 167,
223, 238 168, 176, 188, 189, 223, 225, 237, 240
rigid existential, 96, 97, 102, 110, 136 reference magnetism, 86, 87
ontological realism, 2, 9, 50, 64–73, 77, regress, 131, 139, 148, 149, 163, 214, 238
83, 85, 89, 240, See also ontological relativism, 29, 72, 241
anti-realism; ontological deflationism Ross, Don, 11, 123, 125–27, 128, 129, 132,

133, 137, 140, 141, 142, 144, 193, 194,

paraphrase, 19, 49 213, 217–23
Parsons, Terence,24 Routley, Richard, 24, 25
Paul, L.A., 225–31, 234, 235 Rowbottom, Darrell, 178
physical constants, 136, 144–47 Russell, Bertrand, 19, 20, 24, 45, 46
Plato, Platonism, 8, 15, 18, 19, 20, 36, 53,
59, 84 Schaffer, Jonathan,62, 114, 128, 133, 135,
pluralism (as opposend to monism), 105, 138, 140, 141, 208
128, 133, 137, See also monism Schrödinger, Erwin, 129
possibility. See modality Sidelle, Alan, 80–83
possible worlds, 23, 25, 116, 157, 180, 224, Sider, Ted,8, 9, 33, 34, 40, 49, 50, 54,
230, 236, 240 56, 57, 62, 65, 66, 67, 68, 69, 71, 73,
powers, 56 83–90, 138
Priest, Graham, 21, 24, 25, 45–47 Simons, Peter, 99
258 Index

Smith, Barry, 78 Twin Earth, 81, 177, 179, 180, 183, 184,
Smolin, Lee, 214 192, 200, 201
Sorensen, Roy,199, 200, 201
Sosa, Ernest, 183, 184, 193 universalism (about composition), 34,
special composition question, 33, 34, 35, 55, 56, 75, 187, 241. See also nihilism
55, 241 (about composition)
structuralism, 79, 223, 241 universals, 17, 18, 19, 20, 22, 32, 208, 215,
structure, 77, 79, 83, 84, 119, 120, 121, 233, 236
124, 127, 131, 136, 137, 138, 139, 140,
146, 147, 185, 204, 205, 206, 217, 223, vague, vagueness, 4, 5, 69, 153, 172, 181,
227, 228 202, 207, 241
substance, 95, 99, 128, 215, 227, 239, 241 Vaidya, Anand, 156, 186, 187
supervenience, 113, 116, 132 Van Fraassen, Bas, 211–17, 218
Szabó, Zoltán Gendler, 47, 48 Van Inwagen, Peter, 4–6, 15, 16, 33, 42, 50,
55, 70, 170, 241
theoretical virtues, 170, 171, 228, 230, Varzi, Achille C., 78–80
231, 235, 241 verbal disputes, 50, 51–57, 62, 69, 70, 71,
Thomas, Christie, 18, 19 72, 75, 240
Thomasson, Amie, 71, 72, 73–76, 78, 91
thought experiments, 10, 177–81, 185, well-foundedness, 123, 133–35, 136, 150,
186, 192, 194–202, 229, 232 242
three- and four-dimensionalism, 53, 55, Williamson, Timothy, 155, 160–63, 167,
66, 241. See also endurantism and 168, 229, 230, 231
perdurantism Woodward, Richard, 22
Trogdon, Kelly, 118
truthmaking, 9, 94, 116–19, 137, 139, 241 Yablo, Stephen, 70, 158