TRUTHMAKER

ONTOLOGY
Mark Fiddaman
Linacre College, Oxford
Department of Philosophy

A dissertation submitted for the degree of DPhil in Philosophy

Trinity Term, 2019
Abstract
Truthmaker Ontology is the metaontological thesis that ontology should be framed as

a search for truthmakers: the things in the world in virtue of which true representations

are true. This dissertation develops Truthmaker Ontology in new ways, arguing that

it offers an attractive mode of thinking about the subject matter and method of

ontological inquiry. I begin by arguing that, if ontology is to live up to its aspirations

to be a serious enterprise, we must abandon the orthodox view that ontology concerns

what there is. The remaining chapters develop Truthmaker Ontology as an alternative.

First, I clarify the two central notions on which my version of the approach rests: the

truthmaking relation between truths and their truthmakers, and the grounding relation

between truths and other truths. I then use these two notions to develop a heterodox

view of the subject matter of ontology, whereby ontological status comes apart from

what it’s true to say there is. Briefly, what has ontological status is just what makes a

contribution as a fundamental truthmaker. Since not everything that can truly be said

to exist is a fundamental truthmaker, the question of whether ‘There are Fs’ is true

doesn’t settle the ontological question about Fs. Next I turn to questions of method,

starting with the methodologically important notion of ontological commitment: the

ontological commitments of a theory, I claim, are the entities its holders rationally

ought to admit as truthmakers. I then discuss the implications of Truthmaker

Ontology for the measure of ontological parsimony, which I compare favourably to

the recent view that parsimony only measures a theory’s requirements on what is

ontologically basic. I close by consolidating the case for Truthmaker Ontology,

arguing that it avoids the pitfalls of orthodoxy and has attractive consequences for

some perennial ontological debates.

2
Acknowledgements
I am deeply grateful to Gonzalo Rodriguez-Pereyra for his generous and patient

supervision. For their encouragement and advice, philosophical and otherwise, I

thank Carine Carson, Marina Frasca-Spada, James Warren and, especially, Kathryn

Wilkinson. Thanks also to my examiners, Alexander Kaiserman and Ross Cameron,

for their very helpful comments. For brightening my days while I completed my

thesis, I give love and thanks to Neil Burridge, Steven Fiddaman, David Foster,

Alastair Leitch, Asa Madsen, Lisa Madsen, James Moran, Simon Thomas and,

especially, Talitha Slabbert. No words could express how grateful I am to Freyja

Madsen— for everything you do and are. Finally, I thank my wonderful parents,

Elaine and Paul. I dedicate this work to them.

3
Contents

INTRODUCTION 5

1 CAN ORTHODOX ONTOLOGY BE SERIOUS? 13

2 TRUTHMAKING AND GROUNDING 51

3 ON WHAT THERE REALLY IS 91

4 FUNDAMENTAL TRUTHMAKERS 130

5 TRUTHMAKER COMMITMENTS AND RATIONAL

NORMS 165

6 COUNTING COSTS 198

7 THE CASE FOR TRUTHMAKER ONTOLOGY 230

BIBLIOGRAPHY 259

4
Introduction
What ultimately makes up the world? Does the world consist of nothing more than

atoms in the void, or does it also host complex objects—tables, organisms, planets—

made up of those atoms? Is the world wholly material, or does it host abstract objects,

such as numbers? Are moral values and aesthetic properties part of the fabric of

reality, or are they merely projected onto it by us? Questions like these—ontological

questions, questions about the world's basic make-up— are among the most

interesting and important in philosophy. But this dissertation does not aim to settle

any particular ontological questions. My topic is metaontology: what exactly are

ontological questions and by what methods might they be answered?

'Ontology' derives from the Greek ὄντος: being, that which is. Quine had

etymology on his side, then, when he famously declared that the 'ontological problem'

can be put in three monosyllables: 'What is there?' (1948: 21). For Quine and his

many followers, ontological questions are quantificational questions: ‘Are there any

numbers?’, ‘Are there any complex objects?’, and so on; to find the answers, we must

determine the quantificational consequences of the sentences we ought to accept as

true. This Quine-inspired approach grew to become so dominant, in the words of one

commentator, ‘as to have become almost invisible as a methodological choice’

(Thomasson 2015: 3). But a growing number of philosophers have come to see that

it is a choice, and perhaps the wrong one. Consequently, there have recently emerged

several heterodox views of the nature of the ontologist’s task. It has been argued, for

instance, that ontology concerns what there fundamentally is, what there really is,

5
what grounds what, what there is in the most metaphysically natural sense of 'There

is', and the list goes on.1

So it's an exciting time for metaontology: the old order is increasingly being called

into question, as philosophers explore new ways to frame the task of discovering what

makes up the world. The question of this dissertation is therefore a timely and

important one. However, the diminishing consensus about the task of ontology

presents a challenge. We face a 'meta-metaontological' question: how should we

choose between rival accounts of the nature of ontological questions? Here, I’m going

to make an assumption from the outset: we should aim for an account of ontological

questions that vindicates what I shall call the serious conception of ontology—

‘Serious Ontology’, for short.2

Serious Ontology is a broad, somewhat primitive perspective on the purpose and

character of ontological inquiry. It says, in a nutshell, that ontology aims to answer

deep and substantive questions about what is real. By ‘real’ I mean, crudely, what is

out there in the world as it is in itself, as opposed to the world as we represent it in

thought, language and experience. By a ‘deep’ question, I mean one whose answer is

sensitive to objective facts about reality, as opposed to the language and concepts we

happen to employ. And by a ‘substantive’ question, I mean one with a determinate

and non-trivial answer, which can be discovered only through hard theoretical work.

This is intended to be as neutral as possible between more specific metaontological

positions. Perhaps deep and substantive questions about what is real are equivalent to

questions about what there is, what is fundamental, what (in some quasi-technical

sense) really exists, or something else. However, I take Serious Ontology to

1
See e.g. Cameron (2008a, 2010a), Dorr (2005), Horgan and Potrč (2008), Fine (2001, 2009),
Schaffer (2009), Sider (2011), Williams (2010, 2012).
2
Thomasson (2009) uses this term in a similar way.
6
capture— albeit roughly— what most of us think we’re up to when we’re grappling

with ontological questions and why we take such questions to matter. We know that

we talk and think about complex objects, minds, moral values and all the rest; but we

turn to ontology to tell us if such things really are part of the world itself, or mere

figments of our thought and language. If the questions ontologists are setting

themselves turn out to be trivial or shallow, many of us would feel that the discipline

is failing. So to refine my question, what I want to know is this: is there an account

of the task and method of ontology that would vindicate the serious conception?

This is a pressing matter, since not everyone believes that Serious Ontology is . A

particularly virulent strain of criticism traces back to Carnap (1956), who argued that

ontological questions are either meaningless or are trivially settled by the rules of the

linguistic framework in which they are posed. In a similar spirit, philosophers such

as Hirsch (2002, 2009, 2011) and Thomasson (2007, 2009, 2015) have argued that

Serious Ontologists are deeply mistaken about the character of ontological inquiry.

Hirsch contends that ontological debates are typically more sensitive to our linguistic

practices than to what is real, with the disputants often simply talking past each other;

Thomasson maintains that ontological questions are easily settled via conceptual

analysis and ordinary empirical methods, with no need for any heavy-duty

metaphysics.

I won’t directly engage with these neo-Carnapian sceptics. Indeed, there is a point

up to which I agree with them. If ontology is a question of what there is— as the

orthodox view dictates— then it is not a serious enterprise: quantificational questions

are not equivalent to deep and substantive questions about what is real. I give

arguments for this claim in Chapter 1. For one thing, the orthodox view struggles to

account for prima facie cases of unloaded quantification, where true-seeming

sentences quantify over unreal-seeming entities. For another, questions about what

7
there is and questions about what is real differ significantly in character: the former

are easily answered whereas the latter are not. For a third, the answers to

quantificational questions are disconcertingly sensitive to linguistic convention: even

if Fs are unreal, we might be speaking a metasemantically tolerant language in which

sentences quantifying over Fs are true. I conclude that ontology is not a serious

enterprise, if construed as a question of what there is.

You might think this spells doom for Serious Ontology. That would be to assume,

however, that there is nothing ontological questions could be other than questions

about what there is. I reject that assumption. If questions about what there is turn out

to be shallow or trivial, we can and should seek an alternative account of the nature

of ontological questions— one that fits better with the Serious Ontologist’s aims.

According to my preferred alternative, ontological questions are questions about

truthmakers.

Briefly, a truthmaker is a portion of reality in virtue of which some true

representation is true— on which its truth depends and which thus accounts for its

being true. Truthmaker Ontology is the metaontological view that ontology should be

framed as a search for truthmakers: in broad strokes, we are to decide what is real by

finding the theory of truthmakers that best accounts for the truths we accept. This is

by no means an original approach.3 But I believe that extant versions of Truthmaker

Ontology can be improved or expanded upon in certain key respects. My aim is to

take the project forward, clarifying and developing Truthmaker Ontology with a view

to showing that it offers an attractive mode of thinking about the nature and method

of ontology.

3
Philosophers sympathetic to some version of Truthmaker Ontology include Armstrong
(2004), Baron (2013a), Cameron (2008a, 2010a), Dyke (2007), Heil (2003, 2012), Melia
(2005), Newstead and Franklin (2012), Rettler (2016) and Tallant (2011).
8
 An important initial task is to elucidate the notion of truthmaking: the relation that

holds between true propositions and the portions of reality in virtue of which they are

true— their truthmakers. But truthmaking is not the only ‘in virtue of’ relation I’ll be

appealing to. As will emerge at several points in the dissertation, I think Truthmaker

Ontology benefits from incorporating a distinct relation of (truth-)grounding.

Whereas truthmaking holds cross-categorially between true representations and

portions of extra-representational reality, grounding is an inter-representational

relation: it takes us from a truth P to one or more further truths in virtue of which P

holds. In Chapter 2, I introduce and motivate truthmaking and grounding, explain

what distinguishes and relates them, and argue for some ‘connecting principles’

governing their interplay.

I turn to Truthmaker Ontology in earnest in Chapter 3. There, following the likes

of Cameron (2008a, 2010b) and Rettler (2016), I argue that Truthmaker Ontology

supports a coherent distinction between what has ontological status and what can truly

be said to exist. I do so by appealing to two key theses. The first is the Truthmaker

Criterion of Ontological Status: what counts as real, in the sense of serious

ontological interest, are all and only those things that make a truthmaking

contribution. (As I’ll explain, this is not the same as a truthmaker criterion of

ontological commitment, which some Truthmaker Ontologists place at the centre of

their accounts.) The second is the Divergence Thesis: what makes a statement true

can differ from what is said or implied to exist by that statement. Given these two

theses, a statement proclaiming the existence of Fs might be made true even if Fs

make no truthmaking contribution and therefore lack ontological status. What has

ontological status thus diverges from what can truly be said to exist: following

Cameron (2008a, 2010a), we can articulate this as a distinction between what really

exists (i.e. what makes a truthmaking contribution) and what merely exists (i.e. what

9
can truly be said to exist but makes no truthmaking contribution). Thus, for the

Truthmaker Ontologist, ontology is a question of what really exists, which might be

a proper subset of what it’s true to say exists.

Because it distinguishes ontological status from existential truth, Truthmaker

Ontology has some prima facie advantages over the orthodox framework. Notably, it

allows us to coherently exclude certain entities from our ontology without rejecting

or revising statements implying their existence. This is an attractive prospect for those

concerned that, under the orthodox regime, ‘serious ontological questions are being

decided by linguistic facts’ (Cameron 2008a: 5): ontological debates needn’t turn on

whether certain sentences are true at face-value or are amenable to paraphrase. It will

also appeal to those drawn to a sparse ontology, free of such bugbears as complex

objects and abstracta, but who are reluctant to deny ordinary truisms about what there

is. Last but not least, Truthmaker Ontology gives us a response to the critics of

Serious Ontology: we can concede that questions about what there is are often

shallow or trivial, while claiming that questions about what there really is— i.e. what

makes a truthmaking contribution— are deep and substantive.

But while this is all well and good, it would be useful to have a clearer idea of

where to draw the line between what makes a truthmaking contribution, and what can

truly be said to exist but makes none. Thus, in Chapter 4, I explore a way to refine

the notion of what makes a truthmaking contribution, in the course of propounding a

theory of fundamentality. I first distinguish between fundamental and non-

fundamental truths in terms of the grounding relation. I then introduce the notion of

a fundamental truthmaker— roughly, part of a collection of entities that provides a

minimally complete truthmaking base for the fundamental truths. I argue, inter alia,

that the fundamental truthmakers are all it takes to provide a minimally complete

truthmaking base for all truths. This leads to a more refined picture of what it takes

10
to answer the ontological question, as Truthmaker Ontologists construe it: the

fundamental truthmakers are all we must take to make a truthmaking contribution,

hence to really exist.

The next two chapters concern questions of ontological method: specifically, how

should Truthmaker Ontologists understand the methodologically important notions

of ontological commitment and ontological parsimony? In Chapter 5, I propose a

novel Normative Truthmaker View of Ontological Commitment. Briefly, the

Normative Truthmaker View says that the ontological commitments of a theory are

the entities one rationally ought to admit as truthmakers, given that one accepts the

theory. I argue that the Normative Truthmaker View avoids problems that afflict

some other truthmaker views, such as the view that theories are ontologically

committed to the entities that are entailed to be among their truthmakers. I also sketch

a methodological procedure, based on the Normative Truthmaker View, for deciding

what ontology to accept.

I discuss ontological parsimony in Chapter 6. According to the Truthmaker

Ontologist, not everything that exists according to a theory counts against it by the

measure of parsimony: Ockham’s Razor only applies to our theories’ requirements

on what really exists, i.e. on what makes a truthmaking contribution. This seems like

good news for those concerned to minimise their ontological costs. However, the

Truthmaker Ontologist’s position resembles one that has come under recent criticism,

according to which parsimony considerations only apply to a theory’s requirements

on what is ontologically basic.4 But I argue that, due to salient metaontological

differences, this proposal faces problems that the Truthmaker Ontologist can avoid.

4
See e.g. Schaffer (2015), Bennett (2017).
11
 At this point, having clarified its conceptual and methodological foundations, I

hope to have shown that Truthmaker Ontology is a viable metaontological

framework. Chapter 7 consolidates the case for the claim that it is also an attractive

one, focusing on two broad benefits. First, Truthmaker Ontology keeps ontology

serious, avoiding the problems raised for the orthodox view in Chapter 1. Second,

Truthmaker Ontology has attractive implications for first-order ontological debate,

stemming primarily from its ability to reconcile the benefits of sparse ontological

views with a non-revisionary appraisal of ordinary talk.

If ontology is to be a serious enterprise, we need an alternative to the post-Quinean

orthodoxy. I shall conclude that we have one: the task of discovering what ultimately

makes up the world can successfully be framed as a search for truthmakers.

12
1 Can Orthodox Ontology Be Serious?
Serious Ontology is the broad metaontological view that ontology concerns deep and

substantive questions about what is real. Orthodox Ontology is the more specific view

that ontological questions are quantificational questions: questions about what there

is. The two views are not obviously inconsistent: perhaps questions about what there

is are equivalent to deep, substantive questions about what is real. But I will argue

otherwise. The two sorts of questions are not equivalent, and so Serious Ontology and

Orthodox Ontology are in tension. Serious Ontologists should therefore part ways

with orthodoxy.

In §1.1, I will outline Serious Ontology and Orthodox Ontology in more detail. In

the rest of the chapter, I present three challenges to the conjunction of the two views,

a position I’ll call (SOX), which says that deep and substantive questions about what

is real are equivalent to questions about what there is. First, (SOX) struggles to

account for prima facie cases of ‘unloaded’ quantification: true sentences that

quantify over unreal entities (§1.2). Second, (SOX) inconsistently implies that

ontological questions are both very hard and very easy to answer (§1.3). Third, (SOX)

faces problems arising from the possibility that we are speaking a metasemantically

tolerant language, in which what can truly be said to exist does not perspicuously

reflect what is real (§1.4). The failure of (SOX) means that if ontology is a question

of what there is, it can't be a serious enterprise. Hence, Serious Ontologists should

part ways with orthodoxy.

13
1.1 Preliminaries

In this section, I will further clarify Serious Ontology, Orthodox Ontology and the

consequences of conjoining them.

1.1.1 Serious Ontology

Metaontologists are increasingly divided over what exactly it is to ask an ontological

question. For many, ontological questions are questions about what there is, or what

exists. For some, they are questions about what is fundamental, what really exists,

what exists in a maximally joint-carving sense of 'exists'; and the list goes on.5

Abstracting away from these differences, however, there is a broad contemporary

consensus about the general purpose and character of ontological inquiry. I call this

crude metaontological outlook Serious Ontology. The following passage gives a sense

of what I have in mind:

Most contemporary metaphysicians think of themselves as

concerned, not primarily with the representations of language and
thoughts, but with the reality that is represented. In the case of
ontology, there are deep and non-trivial—but still tractable—
questions about numbers, sums, events, and regions of space, as well
as about ordinary objects like turkeys and teacups. (Manley 2009:
3).

Let’s unpack this. Serious Ontologists are, in the first instance, metaphysical realists:

they believe in a 'ready-made world’6, populated by entities that exist independently

of what we think and say. The way we represent the world is one thing, says the

Serious Ontologist; the world that is out there, underlying our representations, is

5
See fn.1.
6
Goodman (1978).
14
another. For the Serious Ontologist, the purpose of ontological inquiry is to discover

what sorts of things ultimately make up the extra-representational world.

To fix terms, I will refer to the extra-representational world envisaged by Serious

Ontologists as reality and to the entities it hosts as real. The aim of Serious Ontology,

then, is to discover what is real. I offer no precise analysis of 'real'. Nor do I make

any assumptions about the extension of 'real' that would preclude any specific

metaontological view from counting as serious: what is real might be equivalent to

what there is, or to what exists. Or it might be equivalent to what is fundamental,

what (in some quasi-technical sense) really exists, or something else. I do assume,

however, that we have some intuitive grasp on the relevant sense of 'real'. Consider

the following statements:7

We think and talk about numbers. But the question remains as to

whether numbers are real.

There appears to be a table in front of me. But the question remains as

to whether there is really a table, or just a table-shaped scattering of

particles.

I think these statements are intelligible, and bring out something close to the sense of

'real' that I take to be at issue for Serious Ontologists. Crudely, what is real is what’s

out there in the world in itself, as opposed to the world as we represent it in language,

thought and experience. Real entities are what populate the 'ready-made world'

envisaged by metaphysical realists. Admittedly, this is not optimally precise; but I

take it that we have enough grip on the relevant sense of ‘real’ to work with.

7
Cf. Raven (2009: Ch. 1).
15
 Alongside this picture of ontology’s subject matter, Serious Ontologists hold

certain views about its character. First, they see ontological questions as deep: the

answers are taken to depend on objective facts about what reality is like, not on how

we happen to describe, conceptualise or experience reality. Second, they see

ontological questions as substantive: the questions have determinate but non-trivial

answers, which can only be discovered through difficult philosophical theorising.

Relatedly, some would add that serious ontological questions are epistemically

metaphysical (Sider 2011: 137). Conceptual or linguistic analysis alone cannot tell us

what is real, nor can straightforward empirical methods; only 'heavy-duty'

metaphysical theorising is up to the task. The 'preferred methodology' (Manley 2009:

3), involves a quasi-scientific comparison of the virtues of rival theories of what is

real. Sider sums this up as follows (2009: 385):

[Serious ontologists] treat competing positions as tentative

hypotheses about the world, and assess them with a loose battery of
criteria for theory choice. Match with ordinary usage and belief
sometimes plays a role in this assessment, but typically not a
dominant one. Theoretical insight, considerations of simplicity,
integration with other domains (for instance science, logic, and
philosophy of language), and so on, play important roles. (Sider
2009: 385).

Most contemporary ontologists aspire to be Serious Ontologists, and rightly so. As I

said in the introduction, I believe that Serious Ontology captures, albeit crudely, what

most of us take ontological questions to be and why they matter: we look to ontology

to tell us about the world, as opposed to how we speak or think about the world. Thus,

as we seek a more refined account of the nature of ontological questions, I assume

that our aim is to vindicate the serious conception. For example, on the orthodox view

to be outlined shortly, ontological questions are questions about what there is. If, as

I’ll claim, questions about what there is aren’t equivalent to deep and substantive
16
questions about what is real, then we should look for alternatives to the orthodox

view.

1.1.2 Orthodox Ontology

Orthodox Ontology has its origins in Quine (1948), who declared that the 'ontological

problem' can be put in three monosyllables: 'What is there?'. Thus, on the orthodox

picture, ontology concerns questions like 'Are there any numbers?', 'Are there any

complex objects?', ‘Are there any moral values?’, and so on. The Orthodox

Ontologist's ultimate goal is to provide a complete inventory of all the kinds of things

there are.

Three main theses underpin the Orthodox Ontologist’s notion of what there is (cf.

van Inwagen 1998, 2009). First:

(i) To be is the same as to exist.

Thus, contrary to neo-Meinongianism, Orthodox Ontologists deny that there is any

difference between asking whether there are Fs and asking whether Fs exist (cf.

Routley 1980, Priest 2005).8 Second:

(ii) For any Fs whatsoever, there is exactly one sense in which

there can be, or exist, Fs.

8
I won’t question this assumption and will, throughout the dissertation, treat ‘Fs exist’ and
‘There are Fs’ as equivalent.
17
Thus, contrary to what some call ontological pluralism, Orthodox Ontologists deny

that there are different 'ways' or 'modes' of being associated with entities of different

kinds (cf. McDaniel 2009, Turner 2010). Finally:

(i) The single sense of being/existence is captured by the

objectual existential quantifier, '$', of first-order logic.

Roughly, the sentence '$xFx' is true on the objectual reading just in case at least one

object in the associated domain satisfies the open formula 'Fx'. Orthodox Ontologists

take this to perspicuously capture the claim that there is at least one thing that is an

F, or more succinctly, that there are Fs. Thus, the Orthodox Ontologist's question

about Fs— whether there are, or exist, any Fs— is equivalent to the question of

whether '$xFx' is true. Orthodox Ontologists hold, in a slogan, that ontological

questions are quantificational questions (cf. Fine 2009: 158).

The orthodox method of ontology involves deciding which quantified sentences

we ought to accept as true. The procedure is roughly as follows. We first render the

sentences we want to affirm into a form that makes their quantificational implications

explicit. If the regimented sentences includes or implies the sentence '$xFx', then we

are committed to accepting Fs into our ontology— ontologically committed to Fs, as

the standard jargon has it.9 If we want to avoid this commitment, we must either

retract the original sentence or find a non-committal paraphrase: roughly, a sentence

that (in some sense) comes to the same thing as the original but does not quantify

over Fs.

This methodology is employed in different ways. Sometimes, the claim is that we

must accept certain entities into our ontology because quantification over them is

9
More on ontological commitment in Chapter 5.
18
unavoidable. For example, van Inwagen (2004) argues that we must accept

properties into our ontology, since the apt rendering of true sentences like ‘Spiders

share some of the anatomical features of insects’ quantifies over them (and no

‘nominalistically acceptable’ paraphrase is available). At other times, the claim is

that we should accept certain entities into our ontology, since quantifying over them

somehow makes our theories go better. For example, Davidson (1967) defends an

ontology of events, on the grounds that quantifying over them is the best way to

account for certain inferences involving action sentences; Lewis (1986) defends an

ontology of possible worlds on the grounds that quantifying over them brings a

range of theoretical benefits, including a neat semantic treatment of modals

(‘Possibly P’ becomes ‘$x(x is a possible world & P is true at x’). And sometimes,

the claim is that we shouldn’t accept certain entities into our ontology because

quantifying over them somehow impairs our theories. Perhaps the entities in

question are somehow problematic. Or perhaps we simply don’t need to quantify

over them and can simplify our theories by not doing so; mathematical nominalists

such as Field (1980), for example, argue against an ontology of numbers, on the

grounds that our scientific theories can be adequately paraphrased in a way that

eschews quantification over them.

1.1.3 (SOX)

The question before us is whether Orthodox Ontology can be serious. I can now

clarify what I mean by this: if ontological questions are treated as quantificational

questions, is this consistent with the aims and assumptions of Serious Ontology? That

19
is, by answering the question 'Are there Fs?', are we thereby answering the deep and

substantive question of whether Fs are real?

To fix ideas, I will focus on a position I'll call (SOX), which results from

conjoining Serious and Orthodox ontology. According to (SOX), existential

quantification is ontologically loaded in the following sense: '$xFx' (or, equivalently,

'There are Fs' or 'Fs exist') is true just in case at least one F is real in the sense of

serious ontological interest.10 So, for example, to settle the truth-value of 'There are

numbers' is to answer the deep, substantive question of whether numbers are part of

extra-representational reality. To decide whether Fs are real, we must determine

whether '$xFx' is among the implications of the (aptly regimented) sentences we

ought to accept as true.

Perhaps no-one has explicitly defended (SOX) as stated. But the position is, I think,

widely assumed. Orthodox Ontologists, such as van Inwagen and Lewis, seem to see

themselves as engaged with substantive questions about the contents of reality, rather

than shallow or trivial questions about our language or concepts. Or consider again

the debate over mathematical platonism, which Linnebo (2018) describes as 'the

metaphysical view that there are abstract mathematical objects whose existence is

independent of us and our language, thought, and practices'. This sounds like the sort

of claim a Serious Ontologist might make. Yet the recent debate between platonists

and nominalists is steeped in orthodox assumptions, having turned to a large extent

on whether quantification over mathematical objects is avoidable.

10
I’ll explain below why I take (SOX) to be committed to both natural and formal language
quantification being ontologically loaded.
20
 In what follows, I will present what I take to be three of the main challenges to

(SOX).11 If successful, the challenges will show that Orthodox Ontology and Serious

Ontology are not co-tenable: questions about what there is are not equivalent to deep

and substantive questions about what is real.

1.2 Unloaded Quantification and Paraphrase

The first challenge to (SOX) goes roughly as follows. Prima facie, it is often the case

that 'There are Fs' is true but Fs are not real. If quantification is ontologically loaded,

as (SOX) maintains, there is a conflict of intuitions here: either 'There are Fs' is true

and Fs are real, or 'There are Fs' is false and Fs are unreal. Friends of (SOX) should

be prepared to explain, or explain away, this apparent tension. The classic response

is to claim that the offending sentences can be shown to be ontologically innocent via

the method of paraphrase. But I will argue that this response is problematic.

1.2.1 Quantification in Natural and Formal Language

First, a preliminary is in order. Orthodox Ontologists sometimes acknowledge that

ordinary language is an unreliable guide to ontology. For example, Quine says that

'[m]any of our casual remarks in the "there are" form would want dusting up when

our thoughts turn seriously ontological' (1969: 100). Indeed, Quine sometimes

suggests that ontological questions only properly arise when we leave ordinary

English behind and regiment our discourse in terms of the $ of first-order logic (see

e.g. Quine 1969: 106). Let me begin, then, by saying why I think (SOX) is committed

11
I do not claim that any of the three challenges are original: my aim in what follows is to
clarify and bolster, not to break new ground.
21
to quantification in natural language— not just first-order logic— being ontologically

loaded.12

When discussing the ontological significance of $, Orthodox Ontologists tend to

have one of two accounts of quantification in mind. On the natural language account,

$ is explained as a formal abbreviation for certain natural language expressions.

Quine, for example, says that '[t]he artificial notation "$x" of existential

quantification is explained merely as a symbolic rendering of the words "there is

something x such that"' (1992: 26-27).13 Similarly, van Inwagen (1998, 2009)

explicates '$xFx' as an abbreviation of the semi-formal 'It is true of at least one thing

that it is such that it is F', which, he thinks, is just a prolix way of saying 'There is an

F'. Thus, according to van Inwagen, 'the meaning of the quantifiers is given by the

phrases of English—or of some other natural language— that they abbreviate' (1998:

240).

Drawing on the natural language account, one might argue that $ is ontologically

loaded because it inherits this feature from the natural language expressions it

abbreviates:

(i) '$xFx' straightforwardly abbreviates 'There is an F'.

(ii) 'There is an F' is true iff at least one F is real.

(iii) Therefore, '$xFx' is true iff at least one F is real.

12
What follows draws heavily on Azzouni (2004: Ch. 3).
13
See also Quine (1960: 241, 1969: 94, 1972: 111-112).
22
Clearly this argument, in its second premise, assumes that the natural language 'There

is' is ontologically loaded.

On the model-theoretic account, the semantics of $ are specified via a model: an

ordered pair consisting of a domain (a non-empty set or plurality of objects) and an

interpretation function mapping names to individuals and predicates to sets of tuples

from the domain. Intuitively, the semantic role of the quantifier is to impose

conditions on the quantity of domain-members that fall under the extension of a given

predicate. Where D is the domain (of some model of some language) the semantic

condition for $ is this:

‘$xfx’ is true iff {x : x Î D Ù fx} ≠ Æ

That is, '$xfx' is true (in the model) iff there is at least one member of D in the

extension of f.

Drawing on the model-theoretic account, one might argue that $ is ontologically

loaded in virtue of its model-theoretic semantics, together with a certain metaphysical

assumption about domains:

(i) '$xFx' is true iff there is at least one member of the domain

in the extension of F.

(ii) A domain is a collection of real objects.

(iii) Therefore, ‘$xFx’ is true iff at least one real object is in the

extension of F, i.e., iff at least one F is real.

Less obviously than the first, this argument also assumes that the natural language

'There is' is ontologically loaded. For as several have observed, nothing in the model-

theoretic semantics per se ensures that $ ranges over a domain of real objects; this

23
only follows on the further assumption that the metalanguage 'There is' used to state

the semantic condition for $ ('$xFx' is true iff there is at least one member of the

domain in the extension of F) is itself ontologically loaded (Azzouni 2004: 54-55,

Brogaard 2008: 38-40, Crane 2013: 37). As Azzouni explains (2004: 55):

Objectual quantifiers have 'objects' to range over only relative to a

body of claims in a metalanguage that itself gains access to those
'objects', if at all, via what its own quantifiers range over. And if
those (metalanguage) quantifiers do not [range over a collection of
real objects], then neither do the objectual quantifiers that the
metalanguage quantifiers help provide objectual semantics for.14

In other words, the model-theoretic semantics for $ licenses the loaded interpretation

only on the assumption that the natural language quantifiers used to specify the

domain is itself given a loaded interpretation. So again, the loadedness of $ stands or

falls with the loadedness of 'There is'.

1.2.2 The Challenge and the Paraphrase Response

So, is 'There is' ontologically loaded? If so, the truth of sentences of the form 'There

are Fs' should suffice for the reality of Fs. But there is reason to doubt that this is the

case: there are abundant examples of true-seeming English sentences that quantify

over prima facie unreal entities. Here are just a few:

(1) There is a fictional space explorer called Picard.

(2) There is a monster I dream about every night.

(3) There is a staff shortage in the NHS.15

14
Here and throughout the dissertation, all quoted italics are in the original unless otherwise
stated.
15
Similar examples are sometimes used to support the claim that ‘There are Fs’ doesn’t imply
the existence of Fs (see e.g. Azzouni 2010, Priest 2010). But I’m happy to grant the orthodox
24
Intuitively, these sentences are true— at least, in readily conceivable circumstances.

But, intuitively, extra-representational reality does not host such things as fictional

space explorers, dream monsters or staff shortages. Given (SOX), these intuitions

conflict. For if 'There is' is ontologically loaded, it seems that either (1)-(3) must be

false or the entities they quantify over must be real.16 Friends of (SOX) should be

prepared to explain, or explain away, this apparent tension.

The classic response to this challenge is the paraphrase gambit. The guiding

thought is that we can disavow the apparent ontological commitments of a quantified

sentence by finding a paraphrase: a sentence that in some sense ‘comes to the same

thing’ as the original but lacks the offending quantification. This is supposed to let us

affirm the original sentence as true without having to accept its prima facie

ontological implications.17 For example, consider the following sentence:

(Hole) There is a hole in my sock.

assumption that ‘There are Fs’ is equivalent to ‘Fs exist’. My question is whether claims of
either sort imply that Fs are real, in the sense of serious ontological interest.
16
Some innocuous-sounding sentences are incoherent on a loaded reading of the quantifiers,
e.g. ‘There are things I wish were real but aren’t, like spaceships capable of warp travel’; ‘The
planet Qo’noS is not real, so at least one thing is not real’.
17
Actually, not every version of the paraphrase gambit takes paraphrase to be truth-preserving.
On what some call the revisionary (as opposed to reconciling) model of paraphrase, the goal
of paraphrase is to replace a false sentence with a true sentence ‘in the same neighbourhood’
(Keller 2015: 90; cf. von Solodkoff 2014). Thus, if S* is a revisionary paraphrase of S, then S
is false (albeit in some sense ‘good enough’ to utter in circumstances where S* is true). I
claim, however, that our intuition about sentences like (1)-(3) is that they’re true, not just
‘nearly as good as true’ (Merricks 2001: 171-185). If I’m right, the revisionary paraphrase
gambit is not a response to the above challenge so much as a concession.

25
On a loaded reading of 'There is', (Hole) seems to imply the reality of holes. The

claim, however, is that we can avoid this apparent commitment by offering a

paraphrase like the following (cf. Lewis and Lewis 1970):

(Hole*) My sock is perforated.

Since (Hole) has a paraphrase that doesn’t quantify over holes, we can (supposedly)

affirm it as true it without committing to the reality of holes. For we have shown that

the ‘seeming reference’ to holes was an ‘avoidable manner of speaking’ (Quine 1948:

32). Accordingly, friends of (SOX) might argue that sentences like (1)-(3) do not

commit us to the reality of fictional space explorers, dream monsters, etc., since they

can be paraphrased in terms of sentences that lack the offending quantification.

But what exactly does paraphrase involve, such that paraphrasing a sentence lets

us disavow its apparent ontological implications? One might think that the relation

between (Hole) and its paraphrase (to stick with our example) is simply that the

sentences have the same meaning and/or truth-conditions. But a well-known problem

is that, since synonymy and truth-conditional equivalence are symmetric, this would

make (Hole*) a paraphrase of (Hole) and vice versa (Alston 1958). So what entitles

us to treat (Hole*) as an ‘ontologically innocent’ paraphrase of (Hole) when we could

just as well treat (Hole) as an ‘ontologically guilty’ paraphrase of (Hole*)? The onus

is on the paraphraser to give us a relevant asymmetry between (Hole) and its

paraphrase that explains why we only need to take (Hole*)’s apparent ontological

implications seriously.18

The paraphraser’s most natural option is to claim that (Hole) is only loosely true,

whereas (Hole*) is strictly true. The thought is that (Hole) has a misleading surface

18
See von Solodkoff (2014: 573), from whom I borrow the phrase ‘relevant asymmetry’.
26
form: it seems to imply that there are holes, whereas really there aren’t any (von

Solodkoff 2014: 574). But in fact, (Hole) is just a non-perspicuous way of describing

the same metaphysical situation that (Hole*) describes in more perspicuous terms.

Both sentences are true, since they both express the true proposition that a certain

sock is perforated. But while it is perfectly correct to assert (Hole) in the ‘ordinary

business of life’ (van Inwagen 1990) it has an ontologically misleading surface form,

and we should restrict ourselves to its strictly true paraphrase when ontological clarity

is at stake.

The paraphraser now has a relevant asymmetry between (Hole) and (Hole*). As

Keller puts it (2018: 371), ‘while the paraphrase of relation is symmetrical, the

perspicuous paraphrase of relation is not’ (2018: 371). Thus, in response to the

challenge above, friends of (SOX) might claim that sentences like (1)-(3) are true

even though the entities they quantify over are unreal. For the sentences are only

loosely true: they have surface forms that mislead with respect to their real,

underlying ontological implications. To bring these implications to light, we must

offer strictly true, ontologically perspicuous paraphrases.

There are several reasons to dislike this paraphrase gambit. For one thing, as

several have argued, they are unmotivated and ad hoc.19 There just isn’t any

compelling evidence to suggest that we speak loosely when uttering sentences like

(Hole), or ‘There is a shortage of staff in the NHS’. Indeed, there is reason to suspect

the contrary. As Korman points out (2009: 247), speakers are usually prepared to

disavow the strict content of sentences put forward as loosely true. Consider the

reaction you'd expect if you asked someone if they really meant to say that the sun

has moved below the horizon. (‘Of course not, don’t be pedantic.’) Compare the

19
See e.g. Burgess and Rosen (2005), Fine (2009), Korman (2009), Kripke (1982).
27
reaction you'd expect if you asked someone if they really meant to say that there's a

hole in their sock, or a staff shortage in the NHS. (‘Well… yes!’). The burden is on

the paraphraser to persuade us that a loose reading is warranted in these cases, despite

the apparent lack of independent motivation.

Another concern is that the paraphrase gambit threatens to introduce a circularity

into (SOX)’s ontological method. The assumption underlying the paraphrase gambit

is that ‘There are Fs’ commits us to the reality of Fs, unless the sentence is put

forward only as loosely true, in which case we need an ontologically perspicuous

paraphrase to show us our genuine commitments. But the problem, as Varzi puts it,

is that ‘the very issue of which sentences must be paraphrased— let alone how they

ought to be paraphrased— can only be addressed against the background of one’s

own ontological inclinations’ (2007: 277). Consider (Hole) again. The paraphraser

claims that (Hole) should be given loose reading because it ostensibly implies that

there are holes, whereas really there aren’t any. (Hole*), in contrast, is supposed to

perspicuously report the ontological facts: that a certain sock is perforated. But

suppose there really are holes. Then (Hole) is at least as ontologically perspicuous as

(Hole*), and we can rest content with the strict reading of (Hole). Circularity

threatens: we are supposed to be able to decide whether holes are real by determining

whether quantified sentences like (Hole) are strictly true or whether they require

paraphrase; but to decide whether (Hole) should be read as strictly true or whether it

requires paraphrase, it seems we must first decide whether holes are real.

Perhaps most seriously, the paraphrase gambit doesn't answer the challenge it's

supposed to resolve. To recap: cases like (1)-(3) present prima facie counterexamples

to the assumption that 'There is' is ontologically loaded. The paraphraser tells us that

we needn't take the apparent ontological implications of such sentences seriously,

since they are only loosely true: their real ontological implications are rendered

28
explicit by their strictly true paraphrases. But far from answering the challenge, the

paraphrase gambit amounts to conceding that 'There is' is not systematically

ontologically loaded: the truth of 'There are Fs' does not always suffice for the reality

of Fs. This points the way to a dilemma. If friends of (SOX) want to maintain that

'There is' is always ontologically loaded, they can't deal with sentences like (1)-(3)

by appealing to the paraphrase gambit: they must either say that the sentences are

false or that the entities they quantify over are real, thus biting one of two bullets. On

the other hand, if they do appeal to the paraphrase gambit, they thereby concede that

'There is' isn't systematically ontologically loaded. And this undercuts the motivation

for the loaded reading of ∃. For as we saw above, the two main arguments for the

loaded reading of ∃ assume a loaded reading of 'There is'. If 'There is' is only

sporadically ontologically loaded, then the most these arguments can show is that ∃

is only sporadically ontologically loaded as well. If that's the case, it seems that we

can't reliably determine whether Fs are real by deciding whether '∃xFx' is true.

1.2.3 Loaded vs. Unloaded Quantification

Rather than appeal to paraphrase, friends of (SOX) might argue that there is a

principled distinction between loaded and unloaded uses of 'There is'. In this vein,

Hofweber (2005) argues that there is an unloaded inferential role reading of the

quantifier, on which 'the inference from "F(t)" to "F(Something)" is always and

trivially valid, regardless of whether 't' refers to a real object (2005: 274). For

example, the inferential role reading lets us trivially and validly move from 'Picard is

a fictional space explorer' to 'There is a fictional space explorer', without thereby

presupposing that Picard is real. Hofweber contrasts this with the domain conditions

reading, on which the quantifier imposes a condition on a domain of worldly objects

and is therefore (he thinks) ontologically loaded. The example given is the sentence

29
'Something fell on my head', which Hofweber says is true 'if there exists an object out

there in reality that is such that it fell on my head' (2005: 271).

Following Hofweber, friends of (SOX) might argue that sentences like (1)-(3) can

be true even if the entities they quantify over are unreal. For perhaps the unloaded

inferential role reading of the quantifier is in play. Thus, one might argue that 'There

is a fictional space explorer called Picard' is true, not because there really is a fictional

space explorer, but because the sentence trivially and validly follows from 'Picard is

a fictional space explorer'; what matters for serious ontological purposes, however, is

whether 'There is a fictional space explorer called Picard' is true on the loaded domain

conditions reading, which is another matter.

The idea that ordinary language quantifiers can be ambiguous between the

inferential role and domain conditions readings has some odd implications. It implies,

for example, that 'There are Fs’ may be true on the inferential role reading of the

quantifier, but false on the domain conditions reading. In that case, we should be able

to hear 'There are Fs but there are no Fs’ as a coherent sentence, which we seemingly

cannot. But in any case, the Hofweber-style response just seems to shift the bump

under the carpet. As I said before, the fact that a quantifier is associated with a domain

does not guarantee that it is ontologically loaded— this only follows on the further

assumption that the associated domain is a collection of real objects, which the model-

theoretic semantics for $ does not ensure by itself. So we're still left without a firm

reason to think the quantifier is loaded on the domain conditions reading. Moreover,

I doubt that there is a principled way for friends of (SOX) to distinguish unloaded

inferential role quantifier-uses from loaded domain conditions uses. Why shouldn't

we think that the (allegedly) loaded domain conditions reading is at work in 'There is

a fictional space explorer called Picard'? For all I can tell, the only apparent

justification is that we have a prior intuition that fictional space explorers aren't real.
30
But it strikes me as ad hoc to insist that the inferential role reading must be in play

just because we doubt the reality of what's being quantified over.

1.3 Easy Ontology

The second challenge to (SOX) is that it misconstrues the character of ontological

questions, as Serious Ontologists construe them. The point is familiar: for just about

any philosophically controversial Fs, the claim that there are Fs seems to follow

trivially from some uncontroversial commonsense truism. For example:

(1) There are prime numbers greater than 10; so there are numbers.

(2) There is a table in front of me; so there are tables.

(3) There are properties domestic cats share with lions; so there are

properties.

These arguments look sound. The premises seem obviously true and the inferences

obviously valid. Similar arguments can be marshalled for a host of claims about what

there is. So it seems that many questions about what there is are easy. The answer is

often a trivial 'Yes', and can be found without doing any heavy-duty metaphysics: we

need only notice what obviously follows from the truisms of other disciplines or of

plain common sense. But if (SOX) is true, arguments like the above threaten to show

that questions about what is real are easy to answer. Serious Ontologists should

demur. By their lights, questions about what is real are substantive, deep and

epistemically metaphysical: we surely can't discover what it out there in extra-

representational reality just by inspecting plain, non-philosophical English truisms.

But according to (SOX), the serious ontological question of what is real is equivalent

31
to the question of what there is. This gives rise to an inconsistent triad (cf. Cameron

forthcoming(a): 9):

(i) Questions about what is real are equivalent to questions about

what there is.

(ii) Questions about what is real are not easy.

(iii) Questions about what there is are easy.

One of these claims must be rejected. But to reject (i) would be to abandon (SOX)'s

central claim about the nature of serious ontological questions. To reject (ii) would

be to abandon the serious conception of the character of ontological questions.

Friends of (SOX) therefore have no choice but to reject (iii). But (iii) is not the claim

to give up. Rather, I think we should reject (i) and (SOX) along with it.20

1.3.1 Why Existence Questions Are Easy

Apart from the prima facie soundness of arguments like (1)-(3), how else might we

support the claim that existence questions have easy answers? A prominent strategy

is to argue that existence claims can be easily established on Moorean grounds (cf.

Fine 2001, 2009; Schaffer 2009). Roughly, the thought is that the sort of

commonsense truisms that occur as premises in arguments like (1)-(3) are Moorean

facts: they have a stronger claim to our acceptance than any revisionary philosophical

20
For versions of this challenge, see Cameron (2010b, 2019), Fine (2001, 2009), Schaffer
(2009), Williams (2012). Some (e.g. Thomasson 2015) take the easiness of existence
questions to entail the easiness of ontological questions, hence that ontology is not a serious
enterprise. But I reject the assumption that there’s nothing ontological questions could be
other than existence questions.
32
argument could muster. Thus, consider Schaffer's oft-cited 'proof' of the existence of

numbers (2009: 357):

1. There are prime numbers.

2. Therefore there are numbers.

1 is a mathematical truism. It commands Moorean certainty, as

being more credible than any philosopher's argument to the contrary.
Any metaphysician who would deny it has ipso facto produced a
reductio for her premises. And 2 follows immediately, by a standard
adjective-drop inference. Thus numbers exist. End of story.

Schaffer’s thought is that the existence of numbers is just not up for serious

philosophical debate: no philosopher's argument could ever cast it into serious doubt.

Nor do we need to do any heavy-duty metaphysics to establish the existence of

numbers: we need only notice what trivially follows from plain mathematical truisms.

More generally, Schaffer thinks, the answers to existence questions are typically right

under our noses: they are easy to find because they trivially follow from claims that

no-one could reasonably deny.

But what is it about so-called Moorean facts that makes them to immune to revision

on philosophical grounds? Why should we accept that common sense has the final

word on what there is? Those who argue for the easiness of existence questions on

Moorean grounds don't always answer these questions.21 But not everyone is

convinced that the Moorean gambit is justified (see, e.g., Daly and Liggins 2010,

2014). The thought seems to be that commonsense truisms are somehow inherently

better justified than any philosophical claim to the contrary. But what gives them this

status? According to Daly and Liggins, the Moorean's most natural option is to appeal

21
Fine (2001, 2009) and Schaffer (2009), for example, both endorse a Moorean stance towards
existence questions, but offer little by way of clarification or support. See Daly and Liggins
(2014) for criticism.
33
to a principle of epistemic conservatism, according to which 'a person is to some

degree justified in retaining a given belief just because that person has that belief'

(Daly and Liggins 2010: 223).22 But as they point out, whatever justification this

might be is surely highly defeasible: if I form a belief without particularly strong

evidence, for example, I should revise it if stronger evidence to the contrary emerges.

Perhaps claims like 'There are prime numbers' and 'There is a table in front of me' are

based on good evidence and are accordingly well justified. But unless there's some

reason to think that the justification for these claims is indefeasible—on philosophical

grounds, at least— a revisionist philosopher might claim to have found arguments

strong enough to undermine them. For the Moorean to bluntly insist that no such

argument will ever be produced just seems chauvinistic.

Let me explain, then, what I think the Moorean gambit involves.23 Consider a

simplistic argument for the claim that there are no mathematical objects:

(P1) If there are mathematical objects, they are abstract.

(P2) If something it abstract, it lacks causal powers.

(P3) There are no things that lack causal powers.

(C) Therefore, there are no mathematical objects.

The argument is clearly valid and (let's allow) the premises are individually plausible.

Despite this, the Moorean wants to persuade us to reject the conclusion. How? The

Moorean might start by pointing out that the following is a corollary of (C) (Lycan

2001, 2005):

22
Cf. Daly and Liggins (2014: 462).
23
My view on this is inspired by Kelly (2005, 2008), Lycan (2001, 2005) and, especially,
Williams (2012).
34
 (C*) ¬(There are prime numbers greater than 10).

We're now confronted with a choice: accept (C*) along with the argument's premises

or reject a premise along with (C*). The Moorean says we should take the second

option, even if we can't say exactly what's wrong with the rejected premise. Why

should we do that? Because the claim that there are prime numbers greater than 10 is

somehow inherently better justified than any of the premises? Not necessarily. The

Moorean might go on to point out that (C*) is not in fact the only relevant implication

of (C), for (C) implies the falsehood of every statement that implies the existence of

a mathematical object. So the relevant corollary is actually more like this:

(C**) ¬(There are prime numbers greater than 10, the number

of my hands is two, there are 242 miles between Oxford

and Durham, the empty set is a subset of every set, the

ratio of atoms in water is 2:1, the atomic number of

caesium is 55, …).

And so on for every belief anyone's ever held about numbers, sets, functions, vectors,

trajectories, lengths, points, temperatures, and all the rest. Again, we face a choice:

accept (P1)-(P3) along with the negation of a vast number of our current beliefs, or

retain those beliefs and reject one of the premises. So, matters of relative justification

aside, the question the Moorean might press is: 'under what circumstances is it

rational for me to change my beliefs to this drastic extent?' (Williams 2012: 167). In

particular, when confronted with an argument like the above, what is the more rational

option: revise all our current beliefs about mathematical objects in one fell swoop, or

withhold belief in the conjunction of the argument's premises? The Moorean says it's

the latter. Why might this be? Why is it irrational to give up a vast number of current

35
beliefs when confronted with a seemingly compelling argument that entails their

falsehood? Here are two arguments the Moorean might give:24

(i) (P1)-(P3) are general philosophical principles, whereas the

claims negated in (C**) are judgements about particular cases. And it

seems irrational to give significantly more credence to general

principles than to case judgements when the two conflict. A single case

judgement is sometimes enough to undermine a general philosophical

principle: when we first encounter a Gettier case, for example, most of

us give up the general principle that knowledge is justified true belief,

rather than our judgement that the subject in the case doesn't count as

knowing. So if a general philosophical principle conflicts with a vast

number of case judgements, it is plausibly more rational to give up the

general principle than the case judgements. Hence, it is plausibly more

rational to give up one of the general principles encoded in (P1)-(P3)

than the vast number of case judgements negated in (C**).

(ii) According to the total evidence requirement, it is irrational for

us to bracket portions of our total evidence when deciding what to

believe. Suppose a detective has some evidence indicating that Colonel

Mustard murdered Dr. Black and other evidence implicating Professor

Plum (Kelly 2008: 64). Suppose that the evidence implicating Plum

would, on its own, make it rational for the detective to conclude that

Plum did it. The total evidence requirement tells us that it still wouldn't

be rational for the detective to accuse Plum without considering all the

24
These arguments are drawn from Kelly (2005) and Kelly (2008) respectively. I'm
compressing a lot of the detail.
36
 evidence, including that which implicates Mustard. (Perhaps Mustard

and Plum were co-conspirators, or Mustard did it after all.) Now, (P1)-

(P3) jointly entail the falsehood of a vast number of claims that we are

plausibly entitled to count among our total evidence: 'statements about

the positions of pointers on readers, statements about how experiments

were conducted, and […] numerical records of experimental data'

(Williams 2012: 167). Coming to believe (C**) on the basis of the

above argument seems to require us to bracket this evidence, contrary

to the total evidence requirement.

There are grounds for thinking, then, that when a philosophical argument purports to

undermine large swathes of our current beliefs, it is more rational to doubt the

cogency of the argument than to abandon the beliefs. Plausibly, this applies generally

to arguments against the existence of mathematical objects, complex objects,

properties, and more besides. This provides some justification for the claim that

questions about whether such entities exist have easy answers. It is highly unlikely

that a philosophical argument could ever rationally compel us to stop believing in

their existence; so when a philosopher raises the question of whether they exist, it is

a sensible policy to answer with a trivial 'Yes'.

Another motivation for the claim that existence questions are easy is broadly

linguistic. Thomasson (2010, 2014) argues that the question of 'Do Fs exist' reduces

to the question of whether the application conditions associated with the sortal term

'F' are satisfied— these being 'basic rules of use that are among those that are

meaning-constituting for the term' (Thomasson 2014: 89). Such questions are easy to

answer, Thomasson thinks, in that they can be settled without engaging in any heavy-

duty, distinctly metaphysical theorising: linguistic analysis reveals the application

37
conditions of a sortal term, and straightforward conceptual or empirical methods

reveal whether the conditions are fulfilled, hence whether the entities falling under

that sortal exist. For example, do knives exist? Well, as anyone competent with the

term 'knife' knows (at least implicitly), a knife is an object falling within a certain size

range, which consists of a handle and a blade and is typically used for cutting. Are

these conditions satisfied? Yes: just look in your kitchen drawer. Thus, knives exist.

And we didn't need to use any heavy-duty, distinctively metaphysical methods to

figure this out: just plain old conceptual analysis and ordinary observation.

Let’s now return to our inconsistent triad:

(i) Questions about what is real are equivalent to questions about

what there is.

(ii) Questions about what is real are not easy.

(iii) Questions about what there is are easy.

As I’ve said, the only option for friends of (SOX) is to deny (iii). While I don’t claim

to have established (iii) conclusively, I think the above considerations are enough to

shift the burden of proof onto (SOX). There is reason to think it very unlikely that a

philosophical argument could ever rationally compel us to abandon our beliefs in the

existence of complex objects, abstracta, and so on; such beliefs therefore seem too

epistemically secure to be worth seriously troubling about on philosophical grounds.

Thomasson’s linguistic account supports this verdict by providing a further

explanation as to why these beliefs seem so secure: all it takes for a belief in the

existence of Fs to be true is for the associated application conditions to be satisfied,

which can readily be confirmed by straightforward conceptual and empirical

methods. Until this burden of proof is shifted, it’s provisionally reasonable for Serious

Ontologists to accept (iii); and they can’t, qua Serious Ontologists, give up (ii). I
38
submit, then, that the Serious Ontologist’s best response to the inconsistent triad is to

give up (i) and (SOX) along with it.

1.4 Metasemantic Tolerance

Further problems for (SOX) arise from the possibility that we are speaking a

metasemantically tolerant language; that is, even if Fs are unreal, our linguistic

conventions might be such that ‘There are Fs’ is true on the correct semantic

interpretation of our language.

1.4.1 Assumptions

I will assume a broadly interpretationist view of metasemantics.25 Here is a brief

sketch of the basic idea. Suppose that Ida is an ideally rational being, who has been

observing a community of speakers throughout its history. Ida is omniscient with

respect to N: the set of all non-semantic facts about the world. N includes all the facts

about the speakers' worldly surroundings, as well as the worldly conditions in which

they are disposed to assent to and dissent from certain utterances. Ida knows, for

example, that the speakers are typically disposed to assent to 'The cat is on the mat'

when the world is such that f and to dissent from it otherwise. Ida's task is to

determine S: the set of all semantic facts about the speakers' language— the truth-

conditions of its sentences, the referents of its terms, and so on. According to

interpretationists, Ida will be able to determine S based on her knowledge of N,

together with certain constraints on successful interpretation. In general, the semantic

facts about some language are those postulated by the best overall interpretation of

25
See e.g. Lewis (1974), Davidson (1974) and, for a useful overview, Williams (2007).
39
that language: the assignment of semantic properties that a being in Ida's position

would settle upon.

A central constraint on successful interpretation is the principle of charity, which

I understand as follows:

(Charity) With respect to some language L, if candidate

interpretation I1 makes more sentences come out true on

typical occasions of assent than candidate interpretation

I2 then, all else equal, I1 is a better interpretation of L

than I2.

A simple way to motivate Charity is to note that competent interpreters would in fact

adhere to it in conceivable real-world scenarios. Suppose you encounter an alien

community who speak much like we do, except they invariably say 'Tree' in

circumstances where we would say 'Cat'. Thus, in the presence of cats, the aliens say

things like 'The tree is on the mat', 'The tree has caught a mouse', and so on. The

aliens' word 'Tree' might mean tree. In that case, the aliens persistently use 'Tree' to

express false (and bizarre) beliefs. But unless you have reason to think that the aliens

are deluded or deceitful, you would be naturally inclined to interpret 'Tree' to mean

cat in the aliens' language. That is, you would prefer the more charitable

interpretation, on which the truth-conditions of the aliens' sentences are more often

satisfied on typical occasions of assent. Just as competent real-world interpreters

would adhere to Charity, so, plausibly, would an interpreter in the idealised position

of Ida. Thus, all else equal, the best overall interpretation of a language will be one

that maximises the truth of assented sentences.

A final assumption before turning to the challenge to (SOX). Some philosophers

distinguish between linguistic and metaphysical truth-conditions (cf. Leslie 2008,

40
Keller 2015, Sider 2011, Williams 2012). Linguistic truth-conditions, which are

mainly of interest to natural language semanticists, are the sort of conditionals

competent speakers implicitly grasp when they understand the meaning of a sentence.

Metaphysical truth-conditions are the metaphysically interesting conditions that

reality must meet in order for a given sentence to be true. To illustrate, consider the

French sentence 'Il y a des nombres premiers'. You might grasp the linguistic meaning

of this sentence by virtue of knowing that it is true iff there are prime numbers. Yet

you might still be at least partially ignorant of the metaphysically interesting

condition that reality must meet for this sentence to be true, for example, whether it

must contain mind-independent mathematical abstracta.26 My assumption is that, if

these two kinds of truth-conditions do come apart, interpretationist principles govern

the correct assignment of metaphysical truth-conditions, and not just linguistic ones:

the metaphysical truth-conditions of sentences are those that an ideally rational

interpreter, omniscient of all the non-semantic facts about reality, would assign.

Accordingly, the correct assignment of metaphysical truth-conditions is constrained

by Charity. Suppose Ida knows that abstract objects are unreal. But the speakers she's

observing routinely utter sentences like 'There are prime numbers' and 'The empty set

is a subset of every set'. When Ida comes to interpret these speakers, she might assign

metaphysical truth-conditions to these sentences that require the reality of abstract

objects. But on this interpretation, the speakers persistently utter falsehoods. So I take

it that, all else equal, Ida would prefer to assign metaphysical truth-conditions that

don't require the reality of abstracta (whatever these might be).27

26
I adapt this example from Keller (2015: 117).
27
Cf. Cameron (2010b: 12-13), Williams (2010: 130).
41
1.4.2 Two Problems for (SOX)

Consider the following scenarios:

(Nihilism) In reality, no collection of objects ever

composes a further object. Thus, reality hosts no

complex objects such as tables or rocks; only

simple particles arranged thus-and-so.

Conventions of English use are such that

speakers systematically assent to (e.g.) 'There is

a table' iff, in reality, there are some simples

arranged table-wise.28

(Universalism) In reality, every collection of objects composes

a further object. Whenever reality hosts a

collection of simples arranged table-wise, it also

hosts a complex object—a table— composed of

those simples. Conventions of English use are

such that speakers systematically assent to (e.g.)

'There is a table' iff, in reality, there is some

complex object that is a table.

I take it that both these scenarios are epistemically possible, and that both are

consistent with our actual experience and linguistic behaviour. Consider, then, two

28
The ‘simples arranged X-wise’ locution, which I’ll employ fairly often, originates with van
Inwagen (1990). I regard this locution as a convenient shorthand for what might be a far more
complicated description of the spatiotemporal locations of, and spatial relations between, the
simple entities.
42
candidate interpretations of English. On the demanding interpretation, 'There is a

table' expresses a truth iff reality hosts a complex table (and so on mutatis mutandis

for similar sentences). On the tolerant interpretation, 'There is a table' expresses a

truth iff reality hosts some simples arranged table-wise (and so on mutatis mutandis

for similar sentences).

Suppose Universalism obtains. Then the demanding interpretation is charitable

enough: reality is such that the truth-conditions of sentences like 'There is a table' are

satisfied on typical occasions of utterance. And since the demanding interpretation

involves a somewhat simpler assignment of truth-conditions, it will plausibly prevail

over the tolerant interpretation (cf. Williams 2010). But suppose Nihilism obtains.

Then the demanding interpretation is highly uncharitable: the truth-conditions of

sentences like 'There is a table' are never met, and we speak falsely most of the time.

So given Nihilism, there would be strong pressure on an interpreter in Ida's position

to select the tolerant interpretation. If there are no countervailing pressures, the

tolerant interpretation will be the best overall. Thus, even if tables are unreal, it might

be the case that 'There is a table' is a true sentence of English.

Metasemantic tolerance poses at least two potential problems for (SOX).

According to (SOX)'s prescribed ontological method, we can determine whether Fs

are real by deciding whether 'There are Fs' must be accepted as true. But suppose that

Nihilism obtains, and the tolerant interpretation prevails. In that case, (SOX)'s

ontological method is vitiated: it would be an error to infer the reality of Fs from the

truth of 'There are Fs'. Moreover, it looks like the only way to find out whether

(SOX)'s ontological method is viable would be to discover what ontology is like via

some other method. Whether the demanding or the tolerant interpretation prevails

depends, to a large extent, on whether complex objects are real. So it seems we must

decide whether complex objects are real in order to decide whether true quantification

43
over them suffices for their reality, hence whether (SOX)'s ontological method is

viable. Clearly, in these circumstances, we couldn't decide whether complex objects

are real by deciding whether sentences quantifying over them must be accepted as

true. It therefore seems that we need an independent way to determine our ontology

in order to decide whether (SOX)'s method for determining our ontology is viable.

The second problem is based on a line of argument from Hirsch (2011).29 Suppose

some English-speaking Orthodox Ontologists are arguing about the existence of

tables. Both sides agree that there are simple particles arranged table-wise. But the

Tablists affirm, while the Anti-Tablists deny, that these particles ever compose a

further complex table. Thus, the Tablists and the Anti-Tablists disagree about the

truth-value of the following English sentence:

(T) Tables exist.

The ontologists think their debate concerns a serious ontological question, whose

answer is sensitive to what reality is like. But are they? Consider four potential

situations:

(UD) Universalism obtains and the demanding interpretation

prevails.

(UT) Universalism obtains and the tolerant interpretation prevails

(ND) Nihilism obtains and the demanding interpretation prevails.

(NT) Nihilism obtains and the tolerant interpretation prevails.

29
Hirsch thinks, however, that if the question of whether there are Fs is shallow, the
ontological question about Fs is therefore shallow. Again, I reject the assumption that
ontological questions must be construed as questions about what there is.
44
Given (UD), the Tablists are right: (T) is true in English iff tables are real, and they

are. Given (UT), the Tablists are right again: (T) is true in English iff simples arranged

table-wise are real, and they are. Given (NT), the Tablists are right yet again: the

tolerant interpretation ensures that (T) is a true sentence of English, even though

tables are unreal. The only situation in which the Anti-Tablists are right is (ND): (T)

is true in English iff tables are real and, since they're not, (T) is false. So the answer

to whether (T) is true is not sensitive to what reality is like— the truth of (T) is

consistent with tables being real or unreal. Whether (T) is true does not hinge on the

reality of tables, then, but on whether conventions of English use are such as to

warrant the demanding or the tolerant interpretation. So, given (SOX), the ontological

question about tables hinges on facts about conventions of language-use. Hence, the

ontological question is not a deep one about the contents of reality, but a shallow one

about the workings of language.

Another possibility is that the Tablists and the Anti-Tablists should not be

interpreted as speaking the same language. Perhaps the Anti-Tablists deny (T) so

frequently and fervently that they should be interpreted as speaking a version of

English in which (T) is false. And perhaps the Tablists affirm (T) so frequently and

fervently that they should be interpreted as speaking a version of English in which

(T) is true. If so, it seems that the debate over (T) is merely verbal (cf. Hirsch 2009):

the Tablists and the Anti-Tablists are talking past each other, both sides speaking truly

in their own idiolects. Again, the debate over (T) turns out not to be ontologically

serious.

45
1.4.3 Eligibility

A potential response to the above concerns is to insist that Charity is not the only

constraint on successful interpretation; there is also an eligibility constraint:

(Eligibility) With respect to some language L, if candidate

interpretation I1 assigns relatively more natural

meanings than candidate interpretation I2 then, all else

equal, I1 is a better interpretation of L than I2.

The notion of naturalness is familiar from Lewis (1983, 1984). For Lewis, some

properties are more natural than others in that (inter alia) they correspond to classes

of objectively similar objects. Thus, electron is a more natural property than electron-

or-cow, since the electrons are in reality of a kind, whereas the electron-or-cows are

a gerrymandered miscellany. Naturalness is reference magnetic, Lewis thinks, in that

more natural properties are intrinsically more eligible than less natural properties to

serve as the semantic values of predicates. Interpretationists can appeal to this

eligibility constraint in cases where Charity alone leaves predicate-reference

indeterminate. Perhaps Charity alone won't tell us whether some speakers are using

'Red' to mean red or red and inside the Milky Way or blue and outside the Milky Way.

Eligibility, however, pressures interpreters to assign the more natural candidate

meaning.

Sider (2009, 2011) argues that Eligibility constrains the assignment of meanings

to all parts of language, not just predicates. In particular, Sider holds that there is a

perfectly natural candidate meaning for the existential quantifier, which carves closer

to the joints of reality's objective quantificational structure better than any other. This

forms the basis of Sider's well-known response to Hirsch's doctrine of quantifier

variantism: the view that there are multiple candidate meanings for the existential
46
quantifier (and related expressions), none of which are uniquely best suited for

describing reality. Recall the debate between the Tablists and the Anti-Tablists. The

two sides, I said, might be interpreted as speaking different variants of English, such

that 'Tables exist' is true in one and false in the other. For Hirsch, this is due to a

difference in the meaning of 'exist': in the Tablist language, the meaning of 'exist' is

such that 'Tables exist' is true and in the Anti-Tablist language, the meaning of 'exist'

is such that 'Tables exist' is false. The quantifier variantist says that neither of these

possible languages is metaphysically privileged over the other: we could, in principle,

describe reality equally well using a language where 'Tables exist' is true as one in

which it's false. So in this and many other debates about what exists, nothing

substantive is at issue beyond the language being spoken. Sider's response is to accept

the multiplicity of candidate quantifier meanings while insisting that one of them—

the perfectly natural quantifier— is uniquely metaphysically privileged. So even if

the truth-value of 'Tables exist' varies depending on which meaning of 'exist' is in

play, there is still a metaphysically substantive question to answer about tables: can

tables truly be said to exist, in the joint-carving sense of 'exist'?

Might friends of (SOX) appeal to Eligibility to avoid the concerns outlined above?

I think not. I accept that Eligibility constrains interpretation as well as Charity. But

on the interpretationist framework I'm assuming, the semantic facts about some

language are those postulated by the best overall interpretation of that language. And

the best overall interpretation of English might be relatively ineligible if this makes

for a significantly better fit with use. Suppose again that Nihilism actually obtains.

Then, in the joint-carving sense of 'exist', only simple objects exist. But given that

English speakers routinely quantify over non-simples, interpreting the English 'exists'

to have the joint-carving meaning would be grossly uncharitable. So the possibility

remains that the best overall interpretation of English is a tolerant one on which

47
Charity trumps Eligibility, and 'Fs exist' doesn't suffice for the reality of Fs. As

Cameron says (2010b: 17), '[i]t may be the case that in English 'exists' means the most

natural of its candidate meanings […]. But we've got no guarantee that that's the case.

It will only be the case if our usage of 'exists' doesn't deviate too much from how it

should if we were only concerned to describe the quantificational structure of the

world'. Even when we factor in Eligibility, then, the possibility remains that we are

speaking a metasemantically tolerant language. And this is all that's needed to make

trouble for (SOX). It might be that 'Fs exist' doesn't suffice for the reality of Fs; to

determine whether this is the case, we'd need some capacity for determining whether

the tolerant interpretation of English prevails over more eligible alternatives, which

presupposes a capacity for determining whether Fs is real that is independent of our

ability to determine whether sentences quantifying over them are true. And the debate

between the Tablists and the Anti-Tablists, for example, might still turn out to be

shallow. If both sides are speaking the same version of English, the Tablists might

still be right about the truth-value of 'Tables exist', even if tables are unreal; if the two

sides are speaking different versions of English, they are still talking past each other.

Sider has a 'plan B' which we can implement if the English 'There is' turns out to

have an unnatural candidate meaning (2011: 74). Ontologists, he says, can simply

stipulate that their quantificational claims are to be understood in the perfectly

natural sense of the quantifier (Sider 2009: 412):

We hereby stipulate that '$' is to express an austere relative of the

ordinary English notion of existence. We hereby stipulate that
although the meaning of '$' is to obey the core inferential role of
English quantifiers, ordinary, casual use of disputed sentences
involving 'there exists' (such as 'Tables exist') are not to affect at all
what we mean by '$'. We hereby stipulate that if there is a highly
natural meaning that satisfies these constraints, then that is what we

48
 mean by '$'. Perhaps the resulting '$' has no synonym in English.
Fine— we hereby dub our new language Ontologese.

To make this move, however, is to leave Orthodox Ontology behind. The thought is

that ontology is not a question of what there is per se, but in the perfectly natural sense

of ‘There is’. But since it might be that the English 'There is' has an unnatural meaning,

it would be a mistake to treat ontological questions as questions about what there is

(see Cameron 2010b). As Cameron puts it, if we want to 'carve reality at its joints we

can't do so by asking what exists, because finding out what exists, given that 'exists'

means one of the unnatural candidate meanings, is not to find out about the

quantificational structure of reality' (2010b: 16). It would be a mistake, for example,

to try to settle the ontological question about numbers, tables or properties by asking

whether English sentences quantifying over them are strictly and literally true, or

whether they can be paraphrased one way or another. Ontologists would have to

distinguish between what can truly be said to exist in English from what can truly be

said to exist in the ontologically relevant, joint-carving sense. We might do this by

specifying that they're attempting to speak in Ontologese, or by introducing a technical

term such as really exists, which is stipulated to take the perfectly natural meaning of

the quantifier (Cameron 2010b: 17; see Ch.3). Either way, we would be venturing into

heterodox metaontological territory.

Conclusion

According to Serious Ontology, ontological questions are deep and substantive

questions about what is real. According to Orthodox Ontology, ontological questions

are questions about what there is. Combining the two yields (SOX): what is real, in

the sense of serious ontological interest, is equivalent to what there is. In this chapter,

49
I have presented and supported three prominent challenges to (SOX): (SOX)

struggles to deal with prima facie counterexamples to the claim that quantification is

ontologically loaded, with the apparent ease of settling existence questions compared

to questions about what is real, and with the possibility that we are speaking a

metasemantically tolerant language. I think these challenges add up to a strong case

for rejecting (SOX): questions about what there is are not equivalent to deep and

substantive questions about what is real. Now, some might claim that there’s nothing

ontological questions could be other than questions about what there is. In that case,

the failure of (SOX) entails the failure of Serious Ontology. But my view is that

Serious Ontologists can and should seek an alternative account of ontological

questions. On my preferred alternative, Truthmaker Ontology, ontological questions

are questions about truthmakers.

50
2 Truthmaking and Grounding
Before turning to Truthmaker Ontology proper, some groundwork is in order. The

framework I'll develop involves two key relations. One is the truthmaking relation,

which relates true propositions to the portions of reality in virtue of which they are

true— their truthmakers. The other is a (truth-)grounding relation, which relates

truths to further truths in virtue of which they hold. My aim here is to clarify my

understanding of these two relations and the connections between them.

I begin, in §2.1, with the general notion of an in-virtue-of relation: a

hyperintensional relation of non-causal metaphysical dependence. I introduce and

motivate truthmaking and grounding, both of which are relations of this kind. Some

philosophers believe in just one in-virtue-of relation, usually taken to be some form

of grounding. However, as I argue in §2.2, truthmaking can neither be reduced to or

eliminated in favour of any of the mainstream notions of grounding. Truthmaking

and grounding (specifically truth-grounding) should therefore be treated as distinct.

In §2.3, I set out some assumptions concerning the key logical and metaphysical

principles governing truthmaking and grounding. In §2.4, I motivate some

'connecting principles' governing the interaction between truthmaking and grounding,

which will play important roles in the chapters to follow.

2.1 Truthmaking, Grounding and 'In Virtue Of'

In this section, I will motivate my view that there are at least two hyperintensional

relations of non-causal metaphysical dependence (or in-virtue-of relations):

truthmaking and (truth-)grounding.

51
2.1.1 In-Virtue-Of Relations

The phrase in virtue of is useful for expressing a certain kind of philosophically

interesting claim:

(1) Complex objects exist in virtue of their simple parts.

(2) S is in pain in virtue of the fact that S’s C-fibres are firing.

(3) {Socrates} exists in virtue of Socrates.

(4) Torture is morally wrong in virtue of the fact that it decreases

overall utility.

(5) Something is a person in virtue of being self-conscious.

(6) Guernica is an artwork in virtue of being an object designed

to elicit an aesthetic response.

(7) ⟨Grass is green⟩ is true in virtue of the fact that grass is

green.30

These claims all purport to explain something: each would be a felicitous answer to

a question of the form ‘Why P?’ or ‘What makes it the case that P?’. The explanations

may or may not be correct, but they are prima facie intelligible and clearly relevant

to a range of central philosophical issues. ‘In virtue of’ has similar uses outside of

philosophy too. For example:

(8) A person has certain legal rights in virtue of being a citizen.

(9) Bitumen is viscous in virtue of having molecular structure M.

30
⟨P⟩ = the proposition that P. I will usually omit the angle brackets, unless clarity is at stake.
52
These claims are also prima facie intelligible and appear to be of a kind with those

above. Thus, the notion of one thing holding in virtue of another is not only

philosophically important, but also seems to figure in our ordinary thinking.

There is more than a superficial resemblance between these 'in virtue of' claims.

First, each claim posits the dependence of a certain phenomenon on another; for

example, (1) is naturally taken to imply that the existence of complex objects depends

on, and can thus be explained in terms of, the existence of their simple parts. Second,

the dependence posited in each case is of a non-causal, metaphysical sort; for

example (2) doesn't tell us that S's pain is an effect of which S's C-fibres firing is the

cause. We are rather dealing with a synchronic sort of dependence, whereby a certain

phenomenon arises from something metaphysically prior, or more fundamental.

Third, the dependence in question is hyperintensional, in the broad sense that it

involves more than mere modal covariation.31 For example, (3) is not equivalent to

the claim that, necessarily, {Socrates} exists if Socrates does.32 For one thing, it’s

presumably also necessary that Socrates exists if {Socrates} does; but intuitively,

{Socrates} exists in virtue of Socrates and not vice versa. For another, it’s presumably

necessary that the number 7 exists if Socrates does; but intuitively, the number 7

doesn't exist in virtue of Socrates.33

Like many others, I believe that claims like (1)-(9) are (if true) underwritten by a

certain kind of distinctively metaphysical relation. One way to motivate this is to

appeal to explanatory realism: the view that correct explanations track objective, non-

31
This use of ‘hyperintensional’ is common in the grounding/truthmaking literature, though
Duncan et. al. (2017) point out that it is somewhat idiosyncratic.
32
See Fine (1994).
33
I’m not suggesting that all ordinary uses of ‘in virtue of’ imply non-causal metaphysical
dependence. Nor am I suggesting that ‘in virtue of’ is the only means of conveying a non-
causal metaphysical dependence claim: ‘because’, ‘depends on’, ‘due to’, etc. can be used to
similar effect.
53
explanatory relations (cf. Audi 2012b, Rodriguez-Pereyra 2005). As Ruben puts it,

'[e]xplanations work, when they do, only in virtue of underlying determinative or

dependency structural relations in the world' (Ruben 1990: 210; cf. Kim 2010). If

explanatory realism is true, and if some 'in virtue of' explanations of the above sort

are correct, then there is plausibly at least one non-causal, hyperintensional

metaphysical dependence relation underlying their correctness. Call a relation with

these features an in-virtue-of relation.

How many in-virtue-of relations are there? Several philosophers think there is just

one, generally known as grounding. I, however, think there are at least two in-virtue-

of relations. As well as a grounding relation, we also need the distinct relation of

truthmaking.

2.1.2 Truthmaking Introduced

Truthmaker theory is founded on the intuition that what is true depends on what is

real:34

If a certain proposition is true, then it owes its truth to something

else; its truth is not a primitive, brute, ultimate fact. The truth of a
proposition […] depends on what reality, and in particular its subject
matter, is like. What reality is like is anterior to the truth of the
proposition, it gives rise to the truth of the proposition and thereby
accounts for it. (Rodriguez-Pereyra 2005: 21).

Those with broadly realist sympathies will be inclined to agree. We don't want to say

that truth 'floats free' from reality, neither depending on the other. As Bennett wryly

puts it, this is 'the kind of thought that leads to berets, and a job in a bad Comp Lit

department' (2001: 187). Nor do we want to adopt a 'radical semantic idealism', on

34
For more detailed overviews of truthmaker theory, see MacBride (2013), Rami (2009) and
Rodriguez-Pereyra (2006).
54
which the nature of reality depends on what representations of it are true (Rodriguez-

Pereyra 2005: 22). The most palatable alternative is to agree with the truthmaker

theorist: whether a representation of reality is true depends on what extra-

representational reality is like, and not vice versa.

More controversially, truthmaker theorists understand the dependence of truth on

reality in terms of a certain relation: truthmaking. The truthmaking relation holds

between true representations (truthbearers) and the portions of reality on which their

truth depends (truthmakers). I assume that the primary truthbearers are propositions

(though a proposition's truthmakers may also be said to make true the sentence it

expresses). And I assume that truthmakers can, in principle, belong to any ontological

category: particulars, properties, relations, events, facts, states of affairs, tropes, and

other entities might all conceivably play the truthmaking role. Truthmaking is thus a

cross-categorial relation, in that it can hold between true propositions and non-

propositions (Armstrong 2004: 5, Rodriguez-Pereyra 2006: 188). This crucial feature

is what allows the truthmaking relation to serve as a metaphysically robust conduit

between representational truth and extra-representational reality.

So much for the relata; what about the relation itself? Truthmaking is, I've said, a

relation of dependence: the dependence of the truth of true propositions on worldly

things. And this seems clearly to be a non-causal sort of dependence, whereby the

truth of true representations is determined synchronically by something

metaphysically prior, i.e. the reality of certain entities. Moreover, the dependence at

issue resists analysis in purely modal terms. The early necessitation account of

truthmaking ('x makes P true iff x exists and it is metaphysically necessary that P is

true if x exists'35) has rightly fallen out of favour, since it fails to ensure the

35
Cf. Fox (1987), Bigelow (1988).
55
explanatory relevance of truthmakers to what they make true: your coffee mug (or

any other contingent existent) would count as a truthmaker for ⟨2 + 2 = 4⟩ (or any

other necessary truth) (cf. Restall 1996). Truthmaking is therefore a hyperintensional

relation of non-causal metaphysical dependence: an in-virtue-of relation. Indeed, 'in

virtue of' is a favourite expression of truthmaker theorists: Armstrong, for instance,

says that '[t]he idea of a truthmaker for a particular truth […] is just some existent,

some portion of reality, in virtue of which that truth is true' (2004: 5; cf. Rodriguez-

Pereyra 2005).

Why believe in the truthmaking relation? One motivation, which I adapt from

Rodriguez-Pereyra (2005), centres on the compelling intuition that what is true

depends on what is real. It is natural to construe this dependence in relational terms,

especially if we think that (i) truth depends on reality in such a way that what is true

can be explained in terms of what is real, and (ii) correct explanations are

underwritten by objective determinative relations, as per explanatory realism (cf.

Rodriguez-Pereyra 2005: 29; see above). Now, as Rodriguez-Pereyra says, relations

link entities: '[s]urely if a relation is instantiated, if it links anything to anything, then

there are some things that it links, and so it links entities' (2005: 26). It is therefore

natural to think of the dependence of truth on reality as a relation between certain

entities— specifically, truthbearing entities and the entities in the world on which

their truth depends. It is natural, in other words, to believe in the truthmaking

relation.36

36
The most frequent reply to this argument is to claim that we don’t need a substantive relation
between truthbearers and truthmakers to account for the dependence of what is true on what
is real; we need only appeal to the asymmetric schema ‘P is true because P (and not vice
versa)’, or to the semantic notions of reference and satisfaction. See Dodd (2002, 2007),
Hornsby (2005) and MacBride (2014) for arguments along these lines. I refer the reader to
Asay and Baron (2020) for a reply to such ‘deflationist’ critics of truthmaker theory.
56
 Another way to motivate truthmaking is to claim that it is extrinsically justified by

its theoretical utility. There are different ideas about what theoretical roles

truthmaking might play; but my own view is that truthmaking is useful primarily as

a tool for regimenting ontological inquiry. Here's the rough idea. Ontology aims to

discover what is real. A truthmaker for a given truth is a portion of reality on which

that truth depends, and which accounts for its truth. So one way to decide our

ontology is to find the best theory of truthmakers for the truths we accept: this is the

basic idea underlying the approach which will be our focus in later chapters:

Truthmaker Ontology. Whether Truthmaker Ontology is attractive remains to be

seen; but if I'm right that it is, this is a motivation to believe in the truthmaking relation

on which the approach rests.

2.1.3 Grounding Introduced

Grounding is typically characterised as a hyperintensional relation of non-causal

metaphysical dependence, which holds between something more fundamental (the

ground) and something less fundamental (the grounded). Grounding, then, is an in-

virtue-of relation.37 Further to the features mentioned, grounding is generally thought

to be a strict partial order. That is, it is irreflexive (nothing grounds itself), asymmetric

(nothing both grounds and is grounded by something else), and transitive (if one thing

grounds another, it grounds whatever the grounded thing grounds).

The relata of grounding is a controversial issue, dividing grounding theorists into

four main camps. Fact-grounders take grounding to relate facts, construed as worldly

states of affairs consisting of objects instantiating properties and standing in relations

(see e.g. Audi 2012a, 2012b; Rosen 2010). For example, perhaps the fact that bitumen

37
For more detailed overviews of grounding theory, see the introduction to Correia and
Schnieder (2012) and Bliss and Trogdon (2014).
57
has molecular structure M fact-grounds the fact that bitumen is viscous. Truth-

grounders take grounding to relate true propositions (see e.g. Cameron 2018, Fine

2001, Rosen 2010). For example, perhaps the true proposition that S's C-fibres are

firing truth-grounds the true proposition that S is in pain. Entity-grounders hold that

grounding can relate entities of arbitrary ontological category (see Schaffer 2009).

For example, perhaps Socrates entity-grounds {Socrates}. Operator-grounders do

not, strictly speaking, take grounding to be a relation at all. They rather hold that

grounding claims should be expressed using a 'because' operator connecting

sentences to other sentences, as in 'Torture is morally wrong because it decreases

overall utility' (see e.g. Correia 2011; Fine 2001, 2012; Schnieder 2010).

Why believe in grounding? A common motivational strategy is to point to

explanatory claims like (1)-(9), claiming that grounding is the common factor

underlying the correctness of such claims and their shared features: thus, if (1) is

correct, this is because the existence of complex objects is grounded in that of their

simple parts; if (9) is correct, this is because bitumen's viscosity is grounded in its

having molecular structure M; and so on for the rest of the claims on the list.

Accordingly, grounding is often vaunted as a highly pervasive and important

phenomenon: Schaffer (2009: 364) calls it 'the primitive structuring conception of

metaphysics', while Fine (2012: 40) says that grounding 'stands to philosophy as

cause stands to science', and is the primary notion through which philosophical

questions about 'what accounts for what' are to be pursued.

I agree with this motivation up to a point. I believe that grounding (specifically

truth-grounding) is useful for capturing an important class of cases of non-causal

dependence (specifically, cases involving truths holding in virtue of other truths). As

I'll argue below, however, none of the mainstream notions of grounding can capture

all interesting cases of non-causal dependence: when it comes to the non-causal

58
dependence of what is true on what is real, we need the distinct relation of

truthmaking.

Another common motivation for grounding is that it makes sense of a certain

intuitive picture, on which the world is structured as a layered hierarchy of

dependence and determination: for example, economic phenomena plausibly depend

on psychological phenomena, which plausibly depend on biological phenomena, and

so on down until we reach a terminus in the microphysical, on which all the rest

ultimately depends. The dependence relation at issue seems to be non-causal,

explanatory, and to have the features of a strict partial order. (For example, if the

psychological depends on the biological, the biological doesn't depend on the

psychological; if the psychological depends on the biological, and the biological on

the chemical, then the psychological depends on the chemical.) It is thus tempting to

think that the relation between the 'layers' in the hierarchy is none other than

grounding.38

Note, however, that different views of grounding lend themselves to different ways

of thinking about the nature of this layered hierarchy. For some grounding theorists,

usually entity-grounders and fact-grounders, grounding imposes structure on reality.

For example, Schaffer characterises entity-grounding as 'the metaphysical notion on

which one entity depends on another for its nature and existence' (Schaffer 2010b:

345), and takes its relata to be 'full-blown, "heavyweight" entries on the roster of

entities' (2009: 360). Similarly, with respect to fact-grounding, Audi insists that

'grounded facts and ungrounded facts are equally real, and grounded facts are an

"addition of being" over and above the facts in which they are grounded' (2012b: 101-

102). On such views, grounding generates an ontological hierarchy, consisting of

38
More on this in §4.1.
59
more and less fundamental levels of reality.39 But for other grounding theorists,

specifically operator-grounders and truth-grounders, grounding imposes structure on

representations of reality: operator-grounding moves from true sentence to true

sentence, truth-grounding from true proposition to true proposition, both in such a

way that '[i]f the truth that P is grounded in other truths, then they account for its

truth; P's being the case holds in virtue of the other truths being the case' (Fine 2001:

15). Grounding, thus construed, generates an alethic hierarchy consisting of more and

less fundamental levels of truth— levels which may or may correspond to sui generis

levels of reality.

I prefer the second of these two pictures, specifically as construed in terms of the

truth-grounding relation. I will later argue that truth-grounding, in conjunction with

truthmaking, supports an attractive picture on which an abundant hierarchy of

fundamental and non-fundamental truths is founded on a sparse ontology of

fundamental truthmakers (see Ch.4). In contrast, the idea that grounding generates a

hierarchy of ontological levels incurs theoretical costs, which proponents have failed

in their attempts to mitigate (see Ch.6). The layered hierarchy, in my view, should be

construed as an alethic hierarchy, and truth-grounding is the tool we need to

underwrite it. We will later encounter other ways in which truth-grounding does

useful work alongside truthmaking in the context of Truthmaker Ontology.40 If the

overall approach turns out to be attractive, as I’ll maintain, this is a further motivation

to believe in truth-grounding.

39
This conception of grounding pairs naturally with what I’ll later call inflationary Stratified
Ontology. See §3.1.2.
40
See especially §3.2, §4.2 and §5.4.
60
2.2 Can There Be Only One?

I have given some reasons to believe in two in-virtue-of relations: truthmaking and

grounding (specifically truth-grounding). But grounding, in one form or another, is

widely thought to be the one and only in-virtue-of relation. If we grant this

assumption, we have two options: (i) assimilate truthmaking into grounding (see

§2.2.1) or (ii) eliminate truthmaking in favour of grounding (see §2.2.2). But I will

argue that, as fair as the mainstream conceptions of grounding go, neither option is

appealing.

2.2.1 Assimilation

If we take grounding to be the sole in-virtue-of relation, it doesn’t immediately follow

that there’s no such in-virtue-of relation as truthmaking. For perhaps truthmaking can

be assimilated into grounding; that is, perhaps truthmaking just is a special case of

the single, all-encompassing grounding relation.

Following Griffith (2014), I will consider four potential reductive definitions of

truthmaking in terms of grounding (corresponding to the four mainstream notions of

grounding distinguished above): these are statements of the form ‘x makes P true =df.

…’, with an instance of the relevant notion of grounding serving as definiens. At a

minimum, such a definition should provide an informative exceptionless necessary

and sufficient condition for truthmaking in terms of grounding. But as Griffith notes

(2014: 200), the definitions can also be given a stronger ‘metaphysical’ reading, as

telling us that truthmaking just is the form of grounding specified in the definiens.

This stronger reading is what interests us here. I hold that truthmaking is distinct from

grounding and must therefore deny that truthmaking just is a special case of some all-

61
encompassing grounding relation. But I need not (and in fact do not)41 deny that there

might be an exceptionless correlation between truthmaking and grounding, i.e., that

the truthmaking relation obtains just in case some grounding relation also obtains—

that is consistent with my view that the relations are distinct.42 With this in mind, I

will now consider the four definitions, briefly explaining why I reject each of them

on the strong reading— that is, as attempts to reductively identify instances of

truthmaking with instances of grounding.

Let’s start with my preferred view of grounding: the truth-grounding view, on

which grounding is a relation between true propositions. The most natural definition

of truthmaking in terms of truth-grounding is as follows:43

(TMTG) x makes P true =df. ⟨x exists⟩ grounds ⟨P is true⟩.

On the relevant reading, (TMTG) tells us that the truthmaking relation is just the

special case of the truth-grounding relation that holds between true propositions of

the form ⟨x exists⟩ and ⟨P is true⟩. This violates what I take to be an important feature

of truthmaking. The central intuition of truthmaker theory is that true representations

depend for their truth on extra-representational reality— this motivates the idea that

truthmaking is a cross-categorial relation, capable of holding between propositions

41
See §2.4.
42
This clarification is necessary, since reductively defining relation R1 in terms of relation R2
doesn’t always mean reductively identifying R1 with R2. For instance, one might reductively
define causation in terms of an entailment relation between certain propositions as follows: c
causes e =df. (⟨c occurs⟩ & ⟨The laws of nature are L1…Ln⟩) → ⟨e occurs⟩. We needn’t read
this definition as telling us that causation just is an entailment relation, or that its relata are
propositions rather than events. Rather, the definition aims to elucidate the notion of causation
by positing an informative exceptionless correlation between instances of causation and
instances of another relation of which we have some prior understanding, i.e. the entailment
relation. On the strong reading of the definitions to follow, however, the aim is to reductively
identify instances of truthmaking with instance of grounding. Thanks to Ross Cameron and
Alexander Kaiserman for pressing me to clarify this point.
43
Cf. Griffith (2014: 202), Cameron (2018: 334).
62
and non-propositions. Now, it might be the case that the truthmaking relation holds

between x and P just in case the truth-grounding relation holds between ⟨x exists⟩ and

⟨P is true⟩ (cf. Griffith 2014: 202). In fact, I think it is the case (see §2.4). But the

claim that truthmaking just is a truth-grounding relation renders truthmaking

incapable of relating representational and non-representational entities— a feature I

regard as crucial.

According to the operator-grounding view, grounding claims are properly

regimented using a ‘because’ operator connecting sentences to other sentences. One

might define truthmaking in terms of operator-grounding as follows:44

(TMOG) x makes P true =df. P is true because x exists.

There are two concerns with (TMOG). First, operator-grounding is a sentence-

connector, always moving from one sentence to another via the ‘because’ operator;

according to (TMOG), truthmaking is just the special case of operator-grounding

where the connected sentences are of the form ‘P is true’ and ‘x exists’. Again, this

seems to undermine the ability of truthmaking to provide a cross-categorial link

between the representational and the non-representational.45 Moreover, operator-

grounding is construed as a non-relational sentence-connector: it supposedly lets us

make grounding claims without assuming that there is a grounding relation, or that

‘a ground is some fact or entity in the world’ (Fine 2001: 16; cf. Correia 2010: 254).

But in my view, truthmaking is a relation between true representations and worldly

entities: I think that the dependence of truth on reality is best construed in relational

terms, and that we incur ontological commitments to the entities we designate as

truthmakers. (TMOG) seems a poor fit with this conception of truthmaking, since it

44
Cf. Correia (2011), Griffith (2014: 200-201), Schnieder (2006).
45
Cf. Asay (2017: 457-458).
63
is consistent with there being no truthmaking relation at all (cf. Griffith 2014: 201;

Liggins 2012: 267-268).

Fact-grounders take grounding to relate worldly facts, consisting of objects

instantiating properties and standing in relations. Truthmaking might be defined in

terms of fact-grounding as follows:46

(TMFG) x makes P true =df. [x exists] grounds [P is true].47

According to (TMFG), truthmaking is just the special case of fact-grounding that

holds between facts of the form [x exists] and [P is true]. Again, this conflicts with

the cross-categoriality of truthmaking, which is here construed as a relation between

non-representational entities, i.e. worldly facts. (TMFG) also comes with some

contentious ontological baggage. Since the relata of fact-grounding are construed as

‘heavyweight’ denizens of reality (Audi 2012b: 101-102), (TMFG) commits

truthmaker theorists from the outset to the reality of facts. It also commits us to the

reality of the property of existence, which is a constituent of facts of the form [x

exists]— but many, including some fact-grounders, doubt that existence is a real

property (Audi 2012b: 103). According to Griffith, these controversial features are

not decisive against (TMFG), because ‘if they are objectionable, it is for general

metaphysical reasons and not because they are obviously incompatible with

truthmaking’ (2014: 203-204). But I think they are in tension with what I see as the

main theoretical role of truthmaking: to guide ontological decision-making. I hold

that we should decide what is real by identifying the best candidate truthmakers for

the truths we accept. If facts, or the property of existence, turn out to be among the

best candidate truthmakers, we might decide on that basis to accept them as real. But

46
Griffith (2014: 202-203)
47
[P] = the fact that P.
64
(TMFG) makes the decision for us, foisting these contentious ontological choices on

us from the get-go. For my purposes, I would prefer a more ontologically neutral view

of truthmaking.

On the entity-grounding view, grounding can relate entities of arbitrary ontological

category. One could define truthmaking in terms of entity-grounding as follows:

(TMEG) x makes P true =df. x grounds [P is true].

According to (TMEG), truthmaking is just the special case of the entity-grounding

relation where the grounded entity is a ‘truth-fact’. Since truth-facts might be entity-

grounded by non-truth-facts, (TMEG) is consistent with the cross-categoriality of

truthmaking (though we still have a fact where the truthmaker theorist would

normally expect a true proposition). My concern, however, is that (TMEG)

misconstrues the nature of the questions truthmaker theory seeks to answer. Entity-

grounding is characterised as something like an ontological dependence relation: it is

‘the metaphysical notion on which one entity depends on another for its nature and

existence’ (Schaffer 2010b: 345). Thus, consider the question, ‘What makes ⟨Socrates

is wise⟩ true?’. Given (TMEG), this roughly amounts to the entity-grounding

question, ‘What does the fact, [⟨Socrates is wise⟩ is true], depend on for its nature and

existence?’. If you think that facts depend for their nature and existence on their

constituents, for example, your answer to the entity-grounding question might be

something like: ‘The proposition ⟨Socrates is wise⟩ and the property being true’. But

that’s not the sort of answer truthmaker theorists want when they ask what makes

⟨Socrates is wise⟩ true. (They expect an answer more like, ‘The state of affairs of

Socrates instantiating the universal wisdom’, or ‘Socrates’s wisdom-trope’.)

Generally, questions about the entity-grounds of truth-facts are not the same as the

questions that animate truthmaker theorists: the former concern the ontological basis
65
for the existence and nature of certain entities, while the latter concern the ontological

basis for the truth of true representations. (TMEG) leaves us unable to distinguish the

two.

I have now considered the four main candidate reductive definitions of

truthmaking in terms of grounding. I have not argued against the weaker reading of

the definitions, on which they posit an informative exceptionless correlation between

instances of truthmaking and instances of grounding. But I have argued that the

definitions fail when construed as attempts at reductive identification; that is, as

telling us that truthmaking just is a special case of grounding. And this is all I need

to support my claim that, as far as the mainstream views of grounding are concerned,

the truthmaking relation cannot be assimilated into an all-encompassing grounding

relation.

2.2.2 Elimination

If grounding is the sole in-virtue-of relation, and truthmaking can't be assimilated into

grounding, our other alternative is to eliminate truthmaking in favour of grounding.

Fine (2012) makes a spirited case for this option. He describes truthmaking and

grounding as 'close cousins', both of which are 'bound up with the general

phenomenon of what accounts for what' (2012: 43). But Fine thinks that truthmaking

is a poor guide to this general phenomenon and that we should work with grounding

instead.

Fine’s direct objections to truthmaker theory target a strawman: the necessitation

account, according to which x makes P true if the existence of x necessitates the truth

of P. Fine brings up the well-worn complaint that, on this account, necessary truths

are made true by any contingent existent. He also argues that, on certain 'quite

plausible' metaphysical views, the project of finding truthmakers is trivial: you might

66
think that the world is necessarily the way it is, for example, in which case you could

trivially cite the world as a truthmaker for any truth you like (Fine 2012: 45-46). But

the necessitation account has been out of favour since well before Fine's critique.

These objections have no force against those of us who think of truthmaking as a

hyperintensional in-virtue-of relation, and who therefore deny that being a truth-

necessitator suffices for being a truthmaker.

In the more interesting parts of his critique, Fine aims to show that truthmaking is

inferior to grounding as a means of investigating 'the general phenomenon of what

accounts for what'. One of his claims is that truthmaker theorists have an 'unduly

restricted conception of what is grounded': the truth of true representations is not the

only phenomenon that needs accounting for, says Fine, since 'whenever we consider

the question of what makes the representation that P true, there will also arise the

question of what, if anything, makes it the case that P' (2012: 43). Fine also points

out that truthmaking has the wrong structural features to underwrite the 'layered

hierarchy' picture (see §2.1.3). The grounding theorist can, for example, 'ground the

normative in the natural […], then the natural in the physical, and then the physical

in the microphysical, thereby establishing that the normative was grounded in the

microphysical' (Fine 2012: 44). Truthmaking can't be 'chained' in this way, however,

since it generally moves from truthbearers to non-truthbearers, which cannot

themselves be made true.

These objections trade on an unwarranted assumption: that truthmaker theory aims

to provide a comprehensive account of the 'general phenomenon of what accounts for

what' (cf. Asay 2017: 446, Rodriguez-Pereyra 2015: 519). On the contrary,

truthmaker theory is concerned with the more specific phenomenon of how what is

real accounts for what is true. So if we need something other than truthmaking to

account for other kinds of non-causal dependence, or to vindicate the 'layered

67
hierarchy' picture, this doesn't vitiate truthmaker theory; Fine neglects the possibility

that truthmaking and grounding might work in tandem, playing different roles in our

overall account of non-causal 'accounting for'.

What would be troubling, however, is if grounding could do the work of

truthmaking but not vice versa, rendering truthmaking superfluous. But I don't think

it can. As I've previously indicated, mainstream grounding theories are geared

towards either: (i) the dependence of what is true on what else is true, or (ii) the

dependence of what is real on what else is real (see §2.1.3). Entity-grounding and

fact-grounding move from one 'full-blown, heavyweight' denizen of reality (a fact or

other entity) to another, but never from truth to reality. If we're told, for example, that

a certain x entity-grounds a certain y, we might have learned something about what

else we must take to be real, to account for the reality of y. But there is no scope for

addressing questions about what we must take to be real, given what we take to be

true. In contrast, operator-grounding and truth-grounding only move between true

representations (sentences or propositions). Neither can do the work of truthmaking,

since neither can move from true representation to extra-representational reality.

Asay puts it as follows (2017: 457):

Suppose we are offered a completed exhaustive chart of all the

infinitely many truths and how they stand in various grounding
relations to one another. What we would have is a comprehensive
'book of the world' that showed how all the truths about the world
related to one another vis à vis what is fundamental and derivative.
What we wouldn't have is an ontology. […] Grounding theory
always moves from what is true to what is true. Hence, what [it]
ultimately offers is an alethic structure into which truths are placed.
To move from those truths to the ontology underlying them,
however, is to engage in truthmaker theory.

68
If we're told, for example, that ⟨a is F⟩ is grounded by ⟨b is G⟩, we might have learned

something about what else we must take to be true, to account for the truth of ⟨a is

F⟩. But what must we take to be real? a? b? Both? Neither? A state of affairs? A

trope? A resemblance fact? The bare grounding claim doesn't tell us. In contrast, this

is just the sort of question truthmaker theory aims to address: what should we take to

be real to account for what we take to be true?

2.2.3 Pluralism

I conclude from the above that grounding is not the only in-virtue-of relation. If it

were, we should be able to reductively identify truthmaking with, or eliminate it in

favour of grounding; but we can't do either. However, I agree with Fine that

truthmaking can't give us everything we want from a comprehensive account of non-

causal dependence. For one thing, truthmaking questions are not the only interesting

questions of non-causal dependence: the question of what in reality makes the

representation that P true is one thing, the question of what (if anything) makes P the

case is another (Fine 2012: 43). For another, truthmaking lacks the features needed to

generate a hierarchical ordering of non-causal dependence, whereby (for example)

the biological depends on the chemical and the chemical on the microphysical, and

hence the biological depends on the microphysical. There is therefore pressure to

adopt a pluralist view, on which there are at least two distinct, but importantly unified,

in-virtue-of relations. As well as truthmaking, the other relation I'm going to work

with is truth-grounding: the relation that holds between true propositions, such that

'[i]f the truth that P is grounded in other truths, then they account for its truth; P's

being the case holds in virtue of the other truths being the case' (Fine 2001: 15).

Although distinct, truthmaking and truth-grounding are indeed 'close cousins' (Fine

2012: 43). What should we say about the connection between them? One option, to

69
which I'm quite partial, is to say that truthmaking and truth-grounding are both

definable in terms of a more fundamental piece of ideology: in virtue of. Here's a

suggestion. Following Rodriguez-Pereyra (2015: 519), consider the generic 𝜑 is F in

virtue of ψ relation. We will take 'in virtue of' as primitive. But we'll hold it fixed that

any specification of 𝜑 is F in virtue of ψ is non-causal, hyperintensional, and implies

the metaphysical dependence of 𝜑's being F on ψ, in such a way that ψ can be cited

as the explanation of why 𝜑 is F. In-virtue-of relations can now be distinguished

according to the feature that goes in for F and the sort of relata that go in for 𝜑 and

ψ.48 In the case of truthmaking and truth-grounding, the relevant feature is truth. In

the case of truth-grounding, the relata on both sides are true propositions, while in the

case of truthmaking, the first relatum is a true proposition and the second is an entity

of arbitrary ontological category. Thus, where x is an entity and P and Q are true

propositions, truthmaking and truth-grounding might be defined as follows:

(Truthmaking) x makes P true iffdf. P is true in virtue of x.49

(Truth-Grounding) P truth-grounds Q iffdf. Q is true in virtue of P.

Truthmaking and truth-grounding are 'close cousins', then, because they are both

specifications of the generic 𝜑 is F in virtue of ψ relation, and are therefore non-

causal, hyperintensional metaphysical dependence relations. They differ, however, in

48
For Rodriguez-Pereyra (2015: 519) many specifications of the generic being F (non-
causally) in virtue of relation correspond to distinct in-virtue-of relations. So as well as highly
encompassing relations like being true in virtue of and existing in virtue of, there are also more
specific relations like being blue in virtue of and being right in virtue of. I’m not convinced
we need to distinguish this many sui generis in-virtue-of relations and prefer to restrict
attention to highly general features like truth and existence.
49
Cf. Rodriguez-Pereyra (2002, 2005).
70
that truth-grounding always moves between true propositions, whereas truthmaking

moves cross-categorially from a true proposition to a portion of reality. We can thus

allow the two notions to play their respective roles without attempting to define one

in terms of the other: truth-grounding pertains to the non-causal dependence of what

is true on what is true, whereas truthmaking pertains to the non-causal dependence of

what is true on what is real. Their relationship is, as Asay puts it (2017: 459),

one of cooperation. Grounding theory provides us with the 'alethic

structure' of the world, revealing how true propositions stand to one
another vis à vis fundamentality. Truthmaker theory provides the
ontological story to supplement the alethic story: truthmaker theory
tells us what [is real], whereas grounding theory tells us which
subject matters are most fundamental. (Asay 2017: 459).

Perhaps other in-virtue-of relations could be defined in a similar manner. One might

think, for example, that we need something like an entity-grounding relation,

pertaining to the non-causal dependence of what is real on what is real: x entity-

grounds y iffdf. y exists and has its features in virtue of x, or something along those

lines (cf. Schaffer 2010b: 345). I will leave this open, however. Henceforth, I will

restrict attention to truthmaking and truth-grounding.

Is my working primitive, ‘in virtue of’, clear enough to be of use? I believe so. As

stated above, the notion of one thing holding in virtue of another seems prima facie

intelligible and figures in both our philosophical and non-philosophical thinking.

Claims like (1)-(9) provide an initial grasp on the sense of ‘in virtue of’ at work in

the definitions above. Moreover, we can further refine our understanding of the notion

by specifying the logical and other features of truthmaking and truth-grounding, as

I’ll do in the next section.

71
2.3 Truthmaking and Grounding Principles

I will now set out some key principles governing truthmaking and truth-grounding

(which, henceforth, I’ll simply call ‘grounding’). For reasons of space, I’ll keep this

section fairly brisk: I will mostly limit attention to claims I’ll rely on in later chapters,

or which help to clarify the truthmaking and grounding relations.

2.3.1 Truthmaking Principles

I take the logical form of truthmaking claims to be as follows:

X⊨P

'X' is a list of one or more entities, 'P' is a true proposition, and '⊨' symbolises the

truthmaking relation. 'a ⊨ P' can be read as 'a makes P true' or 'P is true in virtue of

a'. The first argument place is plural because several entities might jointly comprise

a truthmaker for a truth: for example, perhaps ⟨Socrates and Plato exist⟩ is made true

by Socrates and Plato together. (But for stylistic reasons, I will often restrict attention

to one-one cases of truthmaking, as in ‘x ⊨ P’, where x is a single entity.)

Truthmaking is many-many. One entity can be a truthmaker for several different

truths. For example, perhaps Socrates makes true both ⟨Socrates exists⟩ and ⟨A

philosopher exists⟩. And one truth can have several different truthmakers. For

example, perhaps Socrates and Plato are both truthmakers for ⟨A philosopher exists⟩.

It is useful to distinguish between full and partial truthmakers. Suppose Socrates

and Plato together make true ⟨Socrates and Plato exist⟩. Then Socrates and Plato

jointly comprise the full truthmaker for ⟨Socrates and Plato exist⟩, whereas Socrates

is only a partial truthmaker. We may define partial truthmaking in terms of full

truthmaking as follows:

72
 (Partial Truthmaking) x is a partial truthmaker for P iffdf.

∃X(X ⊨ P & x ∈ X).

Note that, on this definition, any full truthmaker for P is also a partial truthmaker for

P but not vice versa. 'Truthmaking' shall refer to full truthmaking unless otherwise

specified.

Truthmaking is an explanation-backing relation: if x makes P true, this supports

the correctness of the explanatory claim that P is true because x exists. Accordingly,

truthmaking is hyperintensional in both argument places:

(HyperintensionalityTM) (i) If x ⊨ P and y exists at every

possible world where x exists, it does not follow that y ⊨ P; (ii) If x

⊨ P and Q is true at every possible world where P is true, it does not

follow that x ⊨ Q.

It is also non-monotonic:

(Non-MonotonicityTM) If x ⊨ P, it does not follow that ∀y(x,

y ⊨ P).

Thus, if Socrates is a truthmaker for ⟨Socrates exists⟩, it doesn't follow that Socrates

and your coffee mug comprise a truthmaker for ⟨Socrates exists⟩.

It is initially plausible to suppose that truthmaking is irreflexive and asymmetric.

But some argue that there are recalcitrant cases involving propositions serving as

truthmakers for themselves and other propositions (see e.g. David 2005, Rodriguez

Pereyra 2015). For example, one might think that ⟨This proposition exists⟩ is made

true by the entity whose existence it affirms, i.e. itself, presenting a reflexive case of

73
truthmaking. And one might think that ⟨A proposition exists⟩ makes true ⟨A

proposition about a proposition exists⟩ and vice versa, presenting an asymmetric case

of truthmaking. In a similar way, we might also generate both transitive and non-

transitive cases of truthmaking. A transitive case might be: (i) ⟨Socrates exists⟩ ⊨ ⟨A

proposition about Socrates exists⟩; (ii) ⟨A proposition about Socrates exists⟩ ⊨ ⟨A

proposition exists⟩; (iii) ⟨Socrates exists⟩ ⊨ ⟨A proposition exists⟩. A non-transitive

case might be: (i) Socrates ⊨ ⟨Socrates exists⟩; (ii) ⟨Socrates exists⟩ ⊨ ⟨A proposition

exists⟩; (iii) ¬(Socrates ⊨ ⟨A proposition exists⟩) (Griffith 2014: 205, Rodriguez-

Pereyra 2015: 523).

A possible response to these recalcitrant cases is to deny that propositions are fit

to be truthmakers. For example, one might argue that only fundamental entities can

serve as truthmakers and that propositions aren't fundamental entities (cf. Cameron

2008a, Schaffer 2010b, Rettler 2016). But perhaps it is best not to rule out

propositions as truthmakers a priori, in which case we should say that truthmaking is

merely non-reflexive, non-symmetric and non-transitive. One might worry that

conceding the asymmetry and irreflexivity of truthmaking isn't an option for

truthmaker theorists, given their guiding intuition that truth depends on reality and

not vice versa. But in cases like the above— assuming they are genuine cases of

truthmaking— we still have the truth of a true representation depending on the reality

of a certain entity; it just happens that the entity is also a true representation.

Here, briefly, are some further assumptions about the logic of truthmaking. The

following principles concerning disjunction and conjunction all appear plausible:

(v-IntroTM) If X ⊨ P or X ⊨ Q, then X ⊨ ⟨P v Q⟩.

(v-ElimTM) If X ⊨ ⟨P v Q⟩, then X ⊨ P or X ⊨ Q.

74
 (∧-IntroTM) If X ⊨ P and X ⊨ Q, then X ⊨ ⟨P ∧ Q⟩.

However, I join Rodriguez-Pereyra (2006) in rejecting the following:

(∧-ElimTM) If X ⊨ ⟨P ∧ Q⟩, then X ⊨ P and X ⊨ Q.

You might think, for example, that

Socrates, Plato ⊨ ⟨Socrates exists and Plato exists⟩.

But you shouldn't then infer that

Socrates, Plato ⊨ ⟨Socrates exists⟩,

since Plato is irrelevant to the truth of ⟨Socrates exists⟩, and thus shouldn't be

considered part of the truthmaker; ⟨Socrates exists⟩ might be true in virtue of Socrates

but isn’t true in virtue of Socrates and Plato.

What about existential truths? I'm happy to go along with the general consensus

that the truthmakers for existential generalisations are the truthmakers for their true

instances. The following '∃-introduction' rule also has some plausibility (cf. Correia

2011: §3):

(∃-IntroductionTM) If X ⊨ ⟨a is F⟩, then X ⊨ ⟨∃xFx⟩.

One popular claim I reject, however, is that existence claims are always made true by

their subjects (see e.g. Armstrong 2003; Mulligan et. al. 1984; Rodriguez-Pereyra

2002, 2006b). For instance, I deny that a is always a truthmaker for ⟨a exists⟩ and

75
then Fs are always among the truthmakers for ⟨Fs exist⟩. I will return to this in

Chapter 3.

Finally, consider the following truth-elimination and truth-introduction principles

(cf. Correia 2011):

(Truth-ElimTM) If x ⊨ ⟨P is true⟩ then x ⊨ P.

(Truth-IntroTM) If x ⊨ P then x ⊨ ⟨P is true⟩.

I can't see any obvious reason to deny either of these principles.50

Let's now turn to two somewhat contentious 'metaphysical' principles of

truthmaking: Truthmaker Maximalism and Truthmaker Necessitarianism. According

to Truthmaker Maximalism, every truth has a truthmaker:

(Truthmaker Maximalism) ∀P(P → x(x ⊨ P)

Truthmaker Maximalism is controversial because certain truths— paradigmatically

negative existentials— seem to lack truthmakers. Intuitively, ⟨Dragons do not exist⟩

is not true because the world contains certain entities but precisely because it lacks

certain entities (cf. Melia 2005: 69). And what could make it true that dragons do not

exist? Suggestions include an absence of dragons (Martin 1996), the totality fact that

the non-dragons exhaust the existents (Armstrong 2004), and the world itself

(Cameron 2008e, Schaffer 2010). If you’re unhappy with these options, you might

prefer to deny that negative existentials need truthmakers, thus abandoning

Maximalism. But this strikes some as ad hoc: why should we allow exceptions to the

general rule that truths depend on worldly entities?51 Indeed, it is sometimes assumed

50
Officially, however, I only need (Truth-ElimTM), which makes an appearance in §4.3.
51
Cf. Cameron (2008c: 411), Dodd (2007: 394), Saenz (2014: 84).
76
that to be a truthmaker theorist just is to be a Truthmaker Maximalist; according to

Milne, for example, ‘[t]ruthmaker theory maintains that for every truth there is

something, some thing, some entity that makes it true’ (2005: 221; cf. Dodd 2007,

Merricks 2007). If that’s so, the failure of Truthmaker Maximalism entails the failure

of truthmaker theory itself.

Must truthmaker theorists be Truthmaker Maximalists? It depends on what we

want truthmaker theory to do. For example, some see truthmaker theory as an account

of what it is for a truth to be true, in the mould of the correspondence theory of truth:

to be true is to correspond to reality, and to correspond to reality is to have a

truthmaker (cf. Armstrong 1997: 14). This makes Truthmaker Maximalism

compulsory: if some truths lack truthmakers, then having a truthmaker can’t be part

of what it is for a truth to be true. But in contrast, Truthmaker Maximalism is not

compulsory for those of us who see truthmaker theory as a guide to ontology. By our

lights, the way to determine what is real is to find the best theory of truthmakers for

the truths we accept. This project is not vitiated if there turn out to be truthmakerless

truths. Such truths, as Cameron puts it, are just ones ‘whose truth makes demands on

the world other than ontological demands: i.e., a demand other than that such-and-

such exists’ (2010b: 251-252). Even if Truthmaker Maximalism fails, then,

truthmaker theory can still serve its purpose: we can still answer the ontological

question by specifying truthmakers for all the truths that need them.

That said, I’m drawn to what Asay (2011: 58) calls methodological maximalism:

we should proceed on the assumption that any given truth has a truthmaker unless we

find compelling reason to think otherwise, in which case we should explain what

makes that truth an exception. The motivation for this is that we stand to gain certain

77
theoretical benefits if Truthmaker Maximalism proves correct. One benefit is that

Maximalism affords a simple, uniform account of the relation between what is true

and what is real: every truth is made true by some portion of reality (cf. Griffith 2013:

§1.3). Another, according to Cameron (2018), is that Truthmaker Maximalism

minimises the kinds of facts that must be taken as explanatorily brute: all the facts

about what is true ultimately depends on facts concerning the existence of entities.

I’m therefore inclined to adopt Truthmaker Maximalism pro tanto, on the grounds

that it is ‘beneficial if true’ (Cameron 2018: 350). But since Truthmaker Ontology—

the view that ontology should be framed as a search for truthmakers— is compatible

with the failure of Truthmaker Maximalism, I can afford to remain officially neutral

on the issue.

According to Truthmaker Necessitarianism, truthmakers metaphysically

necessitate what they make true:

(Truthmaker Necessitarianism) ∀x∀P(x ⊨ P ⟶ □(x exists

⟶ P is true)).

The best-known argument for Truthmaker Necessitarianism is due to Armstrong

(1997: 116) and runs as follows. Suppose it is claimed that x is the truthmaker for P.

But suppose there is some possible world, w, where x exists and P is false. Then, says

Armstrong, ‘we will surely think that the alleged truthmaker was insufficient by itself

and requires to be supplemented in some way’ (1997: 116). That is, some y must

actually exist that doesn’t exist at w, such that x and y jointly comprise the actual

truthmaker for P. Armstrong concludes that x can’t be the full truthmaker for P if x’s

existence doesn’t necessitate P’s truth. But Armstrong seems to be begging the

question: if I don’t already think that truthmakers must necessitate their truths, why

78
should I take the possibility that x exists and P is false to imply that x is not an

adequate truthmaker for P?52

Perhaps we can fix Armstrong’s argument by finding an independent reason to

think that x can’t be an adequate truthmaker for P unless x’s existence necessitates

P’s truth. Here’s a suggestion: x is an adequate truthmaker for P only if x’s existence

is an adequate explanation for the truth of P. And for reasons independent of

Truthmaker Necessitarianism, x’s existence is not an adequate explanation for the

truth of P unless it necessitates the truth of P.

In a different context, deRosset (2010) argues that a mark of adequate explanations

is that they lack confounding cases. Roughly, where E is an explanatory claim of the

form ‘𝜑 because ψ', a confounding case for E is a situation where ψ is satisfied but 𝜑

is not. For example, suppose E says that bitumen is viscous because it has molecular

structure M. And suppose some substance, X, has molecular structure M but isn’t

viscous. This presents a confounding case for E, which shows E to be at best

incomplete (deRosset 2010: 80): E either needs to be supplemented or is off the mark

altogether.

As stated above, truthmaking is an explanation-backing relation: if x makes P true,

this implies the correctness of the following explanatory claim:

(E*) P is true because x exists.

Now suppose that, at possible world w, x exists but P is false. This seems to present

a confounding case for E*: x’s existence at a given world is an at-best-incomplete

explanation for the truth of P at that world. So in particular, x’s existence at the actual

world is an at-best-incomplete explanation for the truth of P, in which case x isn’t an

52
Cameron (2018: 339-340), MacBride (2013: §1.2).
79
adequate truthmaker for P. Generally, x is not an adequate truthmaker for P unless

there is no possible world at which x exists and P is false— as per Truthmaker

Necessitarianism.

You might point out that there is an important difference between E and E*. E is

inadequate because it has an actual confounding case: substance X has molecular

structure M but is not viscous. But I claimed that E* is inadequate because it has a

possible confounding case: there is a possible world at which x exists but P is false.

So the above argument assumes that fully adequate truthmaker-explanations must

lack possible confounding cases, not merely actual ones. But this is a principled

assumption. When checking an explanatory claim for confounding cases, we hold

fixed the putative explanans and vary the circumstances to see if the explanandum is

still satisfied. But the circumstances we need to vary to find a relevant confounding

case depend on the nature of the explanatory proposal. We can check E for

confounding cases by looking for actual cases where a substance has molecular

structure M but isn’t viscous, since the same possible world can host two substances

that share one property but fail to share another. But we can’t check E* for

confounding cases by looking for actual cases where x exists but P is false, since P

can’t be true and false at the same world. So the requirement that fully adequate

truthmaker-explanations must lack possible confounding cases seems principled,

given what such explanations purport to explain; that is, how reality accounts for the

truth of a given proposition.

2.3.2 Grounding Principles

I take the logical form of grounding claims to be as follows:

Γ⇒P

80
'Γ' is a list of one or more true propositions, 'P' is a single true proposition and '⇒'

symbolises the grounding relation: 'P ⇒ Q' can be read as 'P grounds Q' or 'Q is true

in virtue of P'. As with truthmaking, the first argument place is plural because the

ground for some truth might consist of several different truths.

If P is all it takes to ground Q, then P fully grounds Q. But if, for example, ⟨a

exists⟩ and ⟨b exists⟩ jointly comprise the full ground for ⟨a and b exists⟩, then ⟨a

exists⟩ is a partial ground for ⟨a and b exists⟩. Partial grounding may be defined in

terms of full grounding as follows:

(Partial Grounding) P is a partial ground for Q iffdf. ∃Γ(Γ

⇒ Q & P ∈ Γ).

Like truthmaking, I take grounding to be an explanation-backing relation: if P

grounds Q, this implies the correctness of the explanatory claim that P is the case

because Q is the case. Accordingly, grounding is hyperintensional (in both argument

places) and monotonic:

(HyperintensionalityG) (i) If P ⇒ Q and R is true in every

possible world where P is true, it does not follow that R ⇒ Q; (ii) If

P ⇒ Q and R is true in every possible world where Q is true, it does

not follow that P ⇒ Q.

(Non-MonotonicityG) If P ⇒ Q, it does not follow that ∀R(P,

R ⇒ Q).

81
As is standard, I take grounding to be irreflexive, asymmetric and transitive:53

(IrreflexivityG) ∀P¬(P ⇒ P).

(AsymmetryG) ∀P∀Q¬(P ⇒ Q & Q ⇒ P).

(TransitivityG) ∀P∀Q∀R(If P ⇒ Q & Q ⇒ R then P

⇒ R).

And here are some further assumptions concerning the logic of grounding. I take all

of the following to be plausible:

(v-IntroG) If Γ ⇒ P or Γ ⇒ Q, then Γ ⇒ ⟨P v Q⟩.

(v-ElimG) If Γ ⇒ ⟨P v Q⟩, then Γ ⇒ P or Γ ⇒ Q.

(∧-IntroG) If Γ ⇒ P & Γ ⇒ Q, then Γ ⇒ ⟨P ∧ Q⟩.

But like (∧-EliminationTM), the following is questionable:

(∧-ElimG) If Γ ⇒ ⟨P ∧ Q⟩, then Γ ⇒ P & Γ ⇒ Q.

You might think, for example, that

⟨Socrates exists⟩, ⟨Plato exists⟩ ⇒ ⟨Socrates exists and Plato exists⟩.

But you shouldn't then infer that

53
This is not uncontroversial. For instance, Jenkins (2011) offers a counterexample to the
IrreflexivityG (which, if successful, would also counterexemplify AsymmetryG), while
Schaffer (2012) offers counterexamples to TransitivityG. I will not discuss these here,
however, as I believe adequate replies are available elsewhere: see Litland (2013) and Raven
(2013).
82
 ⟨Socrates exists⟩, ⟨Plato exists⟩ ⇒ ⟨Socrates exists⟩.

The truth that Plato exists is irrelevant to the truth that ⟨Socrates exists⟩, and so

shouldn't be considered part of the ground.

The following existential introduction principle seems plausible:

(∃-IntroG) If Γ ⇒ ⟨a is F⟩, then Γ ⇒ ⟨∃xFx⟩.

Nor can I see any obvious reason to deny the following truth-introduction and

elimination principles (cf. Correia 2011: §4):

(Truth-IntroG) If P ⇒ Q, then ⟨P is true⟩ ⇒ ⟨Q is true⟩.

(Truth-ElimG) If ⟨P is true⟩ ⇒ ⟨Q is true⟩, then P ⇒ Q.54

2.4 Connecting Principles

I will now discuss three 'connecting principles' governing the interplay between

truthmaking and grounding.

2.4.1 Interaction

The principle of Interaction is as follows:

(Interaction) x ⊨ P iff ⟨x exists⟩ Þ ⟨P is true⟩

Interaction should not be taken as an attempt to reductively identify truthmaking with

grounding. It is just an exceptionless biconditional, telling us that whenever the

truthmaking relation holds between x and P, the distinct grounding relation holds

54
Officially, however, I only need (Truth-IntroG), which features in an argument in the next
section.
83
between the propositions ⟨x exists⟩ and ⟨P is true⟩. This seems plausible. I can't see

how one could intelligibly affirm that P is true in virtue of x while denying that ⟨P is

true⟩ is true in virtue of ⟨x exists⟩, or vice versa. A potential explanation for this is

that for x to make P true just is for ⟨x exists⟩ to ground ⟨P is true⟩, but I have argued

against this: the truthmaking relation and the grounding relation are distinct. The

nearest non-reductive alternative to Interaction is as follows:

x ⊨ P iff ⟨x exists⟩ Þ P.

But this entails that if a ⊨ ⟨a exists⟩ then ⟨a exists⟩ Þ ⟨a exists⟩, violating

(AsymmetryG).

2.4.2 Quasi-Asymmetry

Grounding is asymmetric, whereas truthmaking is non-symmetric. But there is a

plausible 'asymmetry-like' connecting principle, which I'll call Quasi-Asymmetry:

(Quasi-Asymmetry) "x"P((x ⊨ P) ® ¬ (P Þ ⟨x exists⟩))

Informally, Quasi-Asymmetry tells us that a truth cannot explain the existence of its

own truthmaker and that a truthmaker can't make true the explanation of its own

existence. This principle is analogous to (AsymmetryG) and is plausible for similar

reasons. To see this, consider some violations of Quasi-Asymmetry. First:

(Socrates ⊨ ⟨Socrates exists⟩) & (⟨Socrates exists⟩ Þ ⟨Socrates

exists⟩)

The second conjunct is false, since it violates (AsymmetryG). Second:

84
 (Socrates, Plato ⊨ ⟨Socrates and Plato exist⟩) & (⟨Socrates and Plato

exist⟩ Þ ⟨Socrates exists⟩, ⟨Plato exists⟩)

This time, the second conjunct seems false because conjunctions are grounded in their

conjuncts, not the other way around. Third:

(⟨Socrates exists⟩ Þ ⟨{Socrates} exists⟩) & ({Socrates} ⊨ ⟨Socrates

exists⟩)

Again, the second conjunct seems false: Socrates might make it true that {Socrates}

exists, but {Socrates} does not plausibly make it true that Socrates exists. Finally:

(⟨Simples arranged Socrates-wise exist⟩ Þ ⟨Socrates exists⟩) &

(Socrates ⊨ ⟨Simples arranged Socrates-wise exist⟩)

Again, the second conjunct seems dubious: surely, if anything, simples arranged

Socrates-wise make it true that Socrates exists, not the other way around.

So Quasi-Asymmetry is initially plausible. However, there is a specific sort of case

where violations of Quasi-Asymmetry are not clearly false. These arise when a

truthmaking claim designates a truthmaker using a logically complex singular term,

such as 'the fact that a is F' or 'a's F-ness'. Consider the following:

([Socrates is wise] ⊨ ⟨Socrates is wise⟩) & (⟨Socrates is wise⟩ Þ

⟨[Socrates is wise] exists⟩)

(Socrates' wisdom ⊨ ⟨Socrates is wise⟩) & ⟨Socrates is wise⟩ Þ

⟨Socrates's wisdom exists⟩).

85
Some truthmaker theorists find claims like those in the first conjuncts plausible. And

it is not obvious that the second conjuncts are false (cf. Schnieder 2006, Correia and

Schnieder 2012). Couldn't it be that Socrates's wisdom, or the fact that Socrates is

wise, exist because Socrates is wise?

I suspect, however, that these grounding claims are bogus. Note first that the claims

don't seem particularly explanatory. 'Socrates's wisdom exists' and 'The fact that

Socrates is wise exists' just seem like oddly laboured ways of saying that Socrates is

wise. And we might expect that what grounds truths about the existence of facts and

tropes should tell us something informative about the nature or constitution of facts

and tropes. For example, on an Armstrongian metaphysics of facts, a more

informative grounding explanation for the existence of [Socrates exists] might cite

the existence of the fact's non-mereological constituents, the particular Socrates

standing in the instantiation relation to the universal wisdom (Armstrong 1997).

Furthermore, Liggins (2012) offers a plausible account of why it might seem

reasonable to say things like 'Socrates's wisdom exists because Socrates is wise' even

if the 'because' doesn't indicate a genuine grounding relation. Consider the following

(Liggins 2012: 262):

Someone has taken the diamonds, because they're not where I left them.

The 'because' doesn't indicate the presence of a grounding relation here, but rather

functions as an inference-marker. The sentence is a truncated statement of the

following inference:

If the diamonds are not where I left them, somebody has taken them.

They are not where I left them.

Someone has taken them.

86
Similarly, 'Socrates' wisdom exists because Socrates is wise' is plausibly a truncated

statement of the following inference:

If Socrates is wise, then Socrates' wisdom exists.

Socrates is wise.

So Socrates' wisdom exists.

This would explain why claims like 'Socrates' wisdom exists because Socrates is wise'

seem true, but not particularly explanatory: they truncate formally valid inferences,

but do not involve genuine grounding claims.

2.4.3 The Grounding Principle

Perhaps the most interesting connecting principle is the Grounding Principle:

(Grounding Principle) If x ⊨ P and P Þ Q then x ⊨ Q.55

The Grounding Principle is an analogue of Armstrong's Entailment Principle

(2004:10):

(Entailment Principle) If x ⊨ P and P ® Q then x ⊨ Q.

The Entailment Principle has some initial plausibility. Its obvious appeal is that it

promises a significant ontological economy: given a truthmaker for P, there is no need

to posit additional 'bespoke' truthmakers for whatever P entails. Unfortunately, the

Entailment Principle faces some well-known difficulties. For one thing, since

conjunctions entail their conjuncts, the Entailment Principle implies the false (∧-

55
Correia (2011) suggests that a similar principle can be motivated if truthmaking is defined
in terms of operator-grounding.
87
ElimTM) principle (Rodriguez-Pereyra 2006; see above). It also implies that any

truthmaker for any true P is a truthmaker for all necessary truths, which are trivially

entailed by everything. Even worse, in conjunction with (v-ElimTM), the Entailment

Principle implies that every truthmaker is a truthmaker for every truth simpliciter

(Restall 1996, Rodriguez-Pereyra 2006: 963). Let Q be an arbitrary truth. Suppose x

makes true some other truth, P. By the Entailment Principle, x makes true ⟨Q v ¬Q⟩

(which is trivially entailed by everything). So by (v-ElimTM), x makes true Q or x

makes true ¬Q. Since Q is true, ¬Q is false and therefore has no truthmaker. So x

makes true Q. Thus, any truthmaker for any P is also a truthmaker for any true Q.

The Grounding Principle is at least as initially plausible and appealing as the

Entailment Principle: if x is 'ontology enough' to make P true, and P grounds Q, x is

plausibly ontology enough to make Q true as well, thus obviating the need to posit a

bespoke truthmaker for Q. But the Grounding Principle avoids the problems that

afflict the Entailment Principle. Conjunctions entail, but do not ground, their

conjuncts; so unlike the Entailment Principle, the Grounding Principle does not imply

(∧-ElimTM). Every truth entails, but does not ground, every necessary truth. So unlike

the Entailment Principle, the Grounding Principle does not imply that every

truthmaker is a truthmaker for every necessary truth. And for the same reason, the

Grounding Principle does not combine with (v-ElimTM) to imply that every

truthmaker is a truthmaker for every truth simpliciter; it is not the case that any P

grounds ⟨Q v ¬Q⟩ for arbitrary Q.

Furthermore, the Grounding Principle can be derived from three of the principles

endorsed above:

(TransitivityG) ∀P∀Q∀R(If P ⇒ Q & Q ⇒ R then P ⇒ R).

(Truth-IntroG) If P Þ Q then ⟨P is true⟩ Þ ⟨Q is true⟩

88
 (Interaction) x ⊨ P iff ⟨x exists⟩ Þ ⟨P is true⟩.

The argument is as follows:

(1) x⊨P&PÞQ [Assumption]

(2) ⟨x exists⟩ Þ ⟨P is true⟩ [Interaction, (1)]

(3) ⟨P is true⟩ Þ ⟨Q is true⟩ [Truth-IntroG, (1)]

(4) ⟨x exists⟩ Þ ⟨Q is true⟩ [TransitivityG, (2), (3)]

(5) x⊨Q [Interaction, (4)]

(6) ((x ⊨ P & P Þ Q) ® x ⊨ Q [®-Introduction, (1), (4)]

Given the Grounding Principle, we can distinguish two ways in which a truthmaker

can make a given truth true: it can do so indirectly via one or more intermediate

grounding relations, or directly without any such intermediary. More precisely:

(Direct Truthmaking) x directly makes P true iffdf. x ⊨ P &

¬$Q(x ⊨ Q & Q Þ P).

(Indirect Truthmaking) x indirectly makes P true iffdf. x ⊨ P &

$Q(x ⊨ Q & Q Þ P).

89
This distinction will come in handy later on (see especially §3.3 and §4.3).56

Conclusion

Truthmaker theory investigates the dependence of what is true on what is real, which

it construes in terms of the truthmaking relation. Truthmaker Ontology uses the idea

of truthmaking as the basis for a distinctive approach to ontological inquiry. But as

will later emerge, I think that Truthmaker Ontology benefits in several ways if it

supplements truthmaking with a distinct relation of (truth-)grounding. Thus, one of

my aims in this chapter was to clarify my understanding of truthmaking, grounding

and the connections between them. The other was to provide some independent

motivation for seeing two different relations here, where some would see only one.

But the real proof will be in the testing: as I aim to show in the following chapters,

truthmaking and grounding do useful work in tandem as the linchpins of an attractive

account of the nature and method of ontology. If so, this is reason to think that both

relations deserve a place in the metaphysician's toolkit.

56
Cameron (2018: 337) draws a comparable distinction between direct and indirect truth-
grounding: P directly grounds Q iff P grounds Q and there is no R such that P grounds Q and
Q grounds R. He notes that this definition rules out cases where P stands in an unmediated
grounding relation to Q and also grounds Q by grounding some R that grounds Q (2018: 337,
n.8). If we want to allow for such cases, Cameron says, we should only talk about whether a
particular instance of grounding is direct or indirect: ‘a particular instance of <p> grounding
<q> will be direct iff that instance of grounding isn’t mediated via some <r> which grounds
<q> and is grounded by <p>’ (ibid.). Similarly, we may want to allow for cases where x makes
Q true directly and also does so by making true some P that grounds Q. If so, we should only
talk about whether a particular instance of truthmaking is direct or indirect. I’ll gloss over this
nicety, however.

90
3 On What There Really Is
I now turn in earnest to Truthmaker Ontology. I argued earlier that if ontology is to

be a serious enterprise, ontological questions must be distinguished from questions

about what there is. My aim here is to show that Truthmaker Ontology offers a viable

alternative, on which ontological status comes apart from existential truth.

Specifically, I shall join Cameron (2008a) in claiming that Truthmaker Ontology

supports a coherent distinction between what merely exists and what really exists,

where only the latter is of ontological concern (cf. Fine 2001). This distinction, as I

understand it, arises from two theses. According to the Truthmaker Criterion of

Ontological Status, what counts as real, in the sense of serious ontological interest,

are all and only those things that make a truthmaking contribution. According to the

Divergence Thesis, the truthmakers for a claim can differ from what is said or implied

to exist by that claim. Together, these theses allow that ⟨There are Fs⟩ might be made

true even if Fs make no truthmaking contribution, and therefore lack ontological

status. We should therefore distinguish between what really exists (i.e. what makes a

truthmaking contribution) and what merely exists (i.e. what can truly be said to exist

but makes no truthmaking contribution). Thus, for Truthmaker Ontologists, ontology

is not a question of what there is, but of what there really is.

In §3.1, I provide some background and motivation for the idea that truthmaking

lets us distinguish ontological status from existential truth. I argue (pace Cameron

and others) that it's a mistake to base this distinction on a truthmaker criterion of

ontological commitment. Instead, the Truthmaker Ontologist should appeal to a

criterion of ontological status, which I discuss in §3.2. In §3.3, I defend the

Divergence Thesis, offering an account of it based on the Grounding Principle (see

91
§2.4). In §3.4, I articulate the distinction between mere and real existence and defend

its coherence.

3.1 Truthmaker Ontology and Stratified Ontology

I will start by introducing Truthmaker Ontology and placing it in the broader

context of what I call Stratified Ontology.

3.1.1 Truthmaker Ontology

Ontology aims to provide a complete inventory of the contents of reality. A

truthmaker is a portion of reality in virtue of which some true proposition is true. So,

prima facie, truthmaker theory offers a way of regimenting ontological inquiry: we

can decide what is real by finding the theory of truthmakers that best accounts for the

truths we accept. This is the basic idea underlying Truthmaker Ontology.

In one form or another, Truthmaker Ontology has been around for a while.57 In

recent years, the approach has enjoyed a modest resurgence of interest, as several

have come to see it as an attractive alternative to Orthodox Ontology.58 But there are

different views about where the two approaches diverge. For Armstrong (2004), the

main advantage of Truthmaker Ontology is that it does ontological justice to the

predicate. Armstrong thinks that, by taking bound variables as the sole vehicles of

ontological commitment, Orthodox Ontologists illicitly 'stack the ontological deck'

in favour of nominalism. In contrast, 'when we look to truthmakers for truths, subject

and predicate start out as equals, and we can consider the ontological implications of

both in an unbiased way' (Armstrong 2004: 23-24). But Armstrong never explicitly

57
Historical antecedents of Truthmaker Ontology have been detected in the work of Aristotle
(Fox 1987), Leibniz (Hillman 2009), and several early modern Scholastics (Embry 2015).
58
See inter alia Armstrong (2004), Cameron (2008a, 2010a), Dyke (2007), Heil (2003), Melia
(2005), Rettler (2016).
92
questions the basic orthodox tenet that ontology concerns what there is (cf. Lewis

1992: 203). He seems to see Truthmaker Ontology as a more property-friendly way

of deciding what there is.

For others, the appeal of Truthmaker Ontology lies precisely in its ability to

separate ontological questions from questions about what there is. A key premise

here is what I call the Divergence Thesis, according to which the truthmakers for

existence claims need not involve the entities being claimed to exist.59 The thought is

roughly that, given the Divergence Thesis, the truth of the sentence 'There are Fs'

doesn't settle the ontological question concerning Fs, since it might not take an

ontology containing Fs to make this sentence true. Thus, as Melia puts it (2005: 77):

[The] truthmaking project gives us a way of accounting for

[quantified sentences] without dismissing them as false or
meaningless, without having to find a paraphrase, and without
having to accept the entities that the sentence apparently quantifies
over and refers to into our ontology.

It's an attractive prospect. If Melia is right, the ontological question about Fs is not

bound up with the question of whether sentences quantifying over them are true.

Excluding Fs from our ontology does not compel us to deny the truth of such

sentences, or advance revisionary claims about their meaning or underlying logical

form. This is an attractive prospect for those concerned that, under the orthodox

regime, 'serious ontological questions are being decided by linguistic facts' (Cameron

2008a: 5).60 It also opens the appealing possibility of endorsing a sparse ontology,

free of such bugbears as complex objects and abstracta, without rejecting

commonsense truisms about what there is.61 Last but not least, it suggests a line of

59
See Cameron (2008a: 4), Dyke (2007: 4), Heil (2003: 55), Melia (2005: 77).
60
Cf. Dyke (2007: 1-13), Heil (2003: 5-7), Rettler (2016: 1414).
61
Cf. Cameron (2010a: 249-250), Rettler (2016: 1413)
93
reply to the critics of Serious Ontology. It might be conceded that questions about the

truth of existence claims are shallow or trivial; but the Truthmaker Ontologist can

insist that deep and substantive questions remain about what, if anything, makes such

claims true.62

But what does Melia mean by 'ontology' in the above passage? If 'our ontology' is

what we think there is, it seems flatly inconsistent to claim that 'There are Fs' is

literally true while denying that Fs are part of our ontology. Not that we should equate

'ontology' with 'what there is'— I've argued to the contrary. But to ensure consistency,

it seems that Truthmaker Ontologists must articulate a non-standard notion of

'ontology'. Unfortunately, not all of them are clear about what this might be.

Cameron (2008a, 2010a) and Rettler (2016) are notable exceptions. Their approach

is to construe Truthmaker Ontology as a version of what I'll call Stratified Ontology.

3.1.2 Stratified Ontology

Fine (2001) influentially distinguishes between what exists and what really exists.63

Thus, it might be that tables exist but do not really exist; perhaps all that really exist

are simple particles arranged thus-and-so. The distinction is designed to drive a wedge

between existential and ontological questions, so that the question of whether

something exists needn't be taken to settle the ontologically interesting question of

whether it really exists. There has since emerged a minor industry of metaontological

views based on functionally similar distinctions, sometimes articulated as a

62
I’ll expand on these advantages in Chapter 7.
63
Actually, this is a specific instance of Fine's more general distinction between what is
merely and what is really the case.

94
distinction between what fundamentally exists and what derivatively exists: these are

the views I group under the broad heading of 'Stratified Ontology'.64

It is easy to see why Stratified Ontology appeals to serious ontologists disillusioned

with the orthodox framework. Stratified Ontologists can agree with the Moorean that

common sense and science have the final say over what exists. And they can concede

to the neo-Carnapian that debates over what exists are shallow or trivial. They can

insist, however, that there are still deep, substantive, philosophical questions about

what really or fundamentally exists, and that these questions are the proper focus of

ontology. But while Stratified Ontologists tend to have broadly similar motivations,

they have very different views about the nature of the distinction between what exists

and what really, or fundamentally, exists. Following von Solodkoff and Woodward

(2013), we can roughly divide Stratified Ontologists into two camps: inflationary and

deflationary.

On the inflationary picture, reality is stratified into layers. The 'bottom' layer

consists of fundamental (ungrounded, ontologically independent) entities, while the

'upper' layers consist of non-fundamental (grounded, ontologically dependent)

entities. Non-fundamental entities are seen as 'heavyweight' additions to ontology, no

less real than their fundamental grounds65. Thus, since they take both fundamental

and non-fundamental existents to have full-blown ontological status, inflationists

accept the orthodox view that ontology comprises everything that exists. They deny,

however, that merely listing the existents suffices for answering the ontological

question: we must also say where all the existents fit into the 'great chain of being'

(Schaffer 2009: 367). Inflationists hold that most existence questions are easily

64
For views in the vicinity, see Bennett (2017), Cameron (2008a, 2010a, 2010b), Dorr (2005),
Fine (2001, 2009), Horgan and Potrč (2008), Schaffer (2009), Sider (2011), Williams (2010,
2012).
65
See Audi (2012b: 101-102), Bennett (2017: Ch.8) and Schaffer (2009: 360, 2017: 2456).
95
settled by common sense, and that we should therefore take a permissive attitude to

what exists (see esp. Schaffer 2009). But given that Fs exist, the interesting question

is whether Fs are fundamental or, if not, how they are grounded in the fundamental.

Deflationists do not see reality as stratified so much as our representations of it.

The basic idea is that it might be true to represent the existence of Fs in thought and

language, even if reality hosts no Fs. Thus, the sentence 'Tables exist' might be

literally true, even if reality hosts no tables; perhaps all it takes for this sentence to be

true is for reality to host simples arranged thus-and-so. If so, then tables exist— it is

literally true to say so— yet lack ontological status. The deflationist articulates this

by saying that tables exist but don't fundamentally, or really, exist. But the claim here

is not that tables occupy a certain stratum in a 'great chain of being' comprising

different layers of reality. The claim is that tables are not part of reality at all, but we

can still truly represent our table-less reality using sentences like 'Tables exist'. Thus,

whereas inflationists take the fundamental existents to be a proper subset of what has

ontological status, deflationists take the fundamental, or real, existents to exhaust

what has ontological status.

So although inflationists and deflationists make similar-sounding claims, they have

very different ideas about what Stratified Ontology involves. They also face different

explanatory burdens. Inflationists must explain what allows us to be so permissive

about what exists non-fundamentally, given that non-fundamental existents are

heavyweight ontological additions— I will argue in Chapter 6 that inflationists have

failed to discharge this burden. Deflationists, on the other hand, must offer an account

of how an F-less ontology could support the literal—albeit non-fundamental— truth

of representations concerning the existence of Fs. The main appeal of Truthmaker

Ontology, as I see it, is that it offers such an account, making sense of the deflationary

distinction between mere and real existence. Here I follow in the footsteps of

96
Cameron (2008a). As I'll now explain, however, I don't quite agree with Cameron

about how Truthmaker Ontologists should draw this distinction.

3.1.3 Cameron on Truthmaker Commitments and Real Existence

Cameron's (2008a) account centres on a truthmaker criterion of ontological

commitment, which he proposes as a replacement for Quine's quantifier criterion: 'the

ontological commitments of a theory are just those things that must exist to make true

the sentences of that theory' (Cameron 2008a: 4). Alongside this, Cameron endorses

the Divergence Thesis, which allows that '⟨x exists⟩ might be made true by something

other than x, and hence that "a exists" might be true according to some theory without

a being an ontological commitment of that theory' (Cameron 2008a: 4). Thus, if ⟨The

sum of a and b exists⟩ might be made true by just a and b, then one can affirm that

the sum exists without incurring an ontological commitment to sums (Cameron

2008a: 5). Generalising, Cameron's truthmaker criterion of ontological commitment

makes it consistent to affirm that something exists without accepting it into your

ontology. This prompts Cameron to draw the following distinction (2008a: 6):

Let us say that a really (or, equivalently, fundamentally) exists iff

we are ontologically committed to a, and that a exists, but doesn't
really exist (or, equivalently, that a exists derivatively), iff ⟨a exists⟩
is true but is made true by something other than a.

By Cameron's lights, ontology is exhausted by what really exists, i.e., what 'does

some truthmaking' (2008a: 7). The distinction between mere and real existence is a

device that lets us separate true existential representations from what, out there in

reality, accounts for their truth. Cameron thinks ontologists should only be concerned

with the latter— that is, with what really exists. This seems to offer a coherent sense

in which ontology can come apart from what can truly be said to exist, and a cogent

97
version of deflationary Stratified Ontology. Where I disagree with Cameron is that I

don't think the Truthmaker Ontologist should draw the distinction between mere and

real existence in terms of ontological commitment.

On my understanding, the ontological commitments of a theory are the

requirements that the theory's truth imposes on ontology; roughly, what ontology

must include if the theory is to be true. Thus, if ontology consists of what there is,

ontological commitments are requirements on what there is; if ontology consists of

what is fundamental, ontological commitments are requirements on what is

fundamental, and so on. A criterion of ontological commitment is a principle by

which to determine a theory's requirements on ontology. And in my view, we're in no

position to decide whether to accept a given criterion of ontological commitment

before we've decided on the scope of ontology— that which requirements on ontology

are requirements on. Cameron, however, argues from the truthmaker criterion of

ontological commitment to a claim about the scope of ontology. His argument is

roughly this (Cameron 2008a: 6-7):

(1) The ontological commitments of a theory are the things needed

to make that theory true.

(2) If a theory affirms 'There are Fs', it does not follow that Fs are

needed to make that theory true.

(3) So it is consistent to affirm that there are Fs while denying that

Fs are among the ontology one accepts.

(4) So ontology is not a question of what there is. (It only concerns

the subset of what there is that does some truthmaking, or what

there really is.)

98
This is a bit like a Quinean arguing as follows:

(1) The ontological commitments of a theory are the things that

theory's quantifiers must range over if the theory is to be true.

(2) If a theory affirms 'There are Fs', that theory's quantifiers must

range over Fs.

(3) So it is inconsistent to affirm that there are Fs while denying

that Fs are among the ontology one accepts.

(4) So ontology is a question of what there is.

This argument seems odd. The reason for this, I think, is that we’re in no position to

evaluate the first premise prior to accepting the conclusion. Suppose we assume that

ontology consists of what there is, and hence that ontological commitments are

requirements on what there is. We’re then in a position to judge whether the quantifier

criterion captures, or fails to capture, the ontological requirements of theories. We

might decide, for example, that the quantifier criterion neglects certain requirements

on ontology, construed as requirements on what there is. (Example: the sentence

‘Everything the Bible says is true’ does not quantify over God, but its truth plausibly

requires that there is a God.66) But if we haven't already assumed that ontology is a

question of what there is, it's not clear how we could evaluate the quantifier criterion:

how could we decide whether the criterion captures a theory's ontological

requirements without some idea of what such requirements are meant to be on? It

66
This example is adapted from Peacock (2011: 81). See §5.1.1.

99
would be odd for a Quinean to argue as above, then, because the acceptability of the

first premise depends on the truth of the conclusion.

It is similarly inappropriate to argue from a truthmaker criterion of ontological

commitment to the claim that ontology comes apart from what there is. The

truthmaker criterion is adequate insofar as it captures the ontological requirements of

theories. Perhaps it does that, if ontology comes apart from what there is. But if

ontology doesn't come apart from what there is, the truthmaker criterion will be

inadequate, since it doesn't count every requirement on what there is as an ontological

commitment. So it seems to me that we're not in a position to accept Cameron's

truthmaker criterion unless we've already accepted his conclusion about the scope of

ontology. The claim that ontology comes apart from what there is must be seen as a

precondition for, not a consequence of, adopting truthmaker criterion of ontological

commitment.

Like Cameron and others, I think the main appeal of Truthmaker Ontology is that

it can support a coherent distinction between what is in the ontologist’s purview and

what can truly be said to exist. But ontological commitment is a red herring here. I

suggest that Truthmaker Ontologists should start with the claim that all and only the

truthmakers are real, in the sense of serious ontological interest. That is, they should

start with what I’ll call a truthmaker criterion of ontological status (see below).

Combined with the Divergence Thesis, this will give us the desired distinction

between what has ontological status and what can truly be said to exist.

No doubt my disagreement with Cameron is partly terminological. As Cameron

points out (2010a: 250), 'ontological commitment' is a technical term, and we are free

to define it in whatever way proves most useful, even if our definition is at odds with

the term's past associations. Perhaps Cameron wants to use 'criterion of ontological

100
commitment' in the way I would use 'criterion of ontological status'.67 But I think this

muddies the water. The question of the scope of ontology and the question of what

theories require of ontology are distinct, and matter for different reasons: the former

pertains to what ontological inquiry is all about in the first place, the latter to how we

decide what ontology to accept based on what theories we accept as true, and how we

assess theories based on their ‘ontological pre-conditions’ (Peacock 2011).68

3.2 The Truthmaker Criterion of Ontological Status

Serious Ontology aims to discover what is real. I've assumed that we have some

intuitive grasp of what this means: what is real is what is out there in the ready-made

world envisaged by metaphysical realism, as opposed to the world as we represent it

in language, thought and experience (see §1.1.1). But I haven’t tried to give any

precise analysis of 'real'— perhaps no such analysis is available. Even so, we can still

try to clarify our task by finding an informative necessary and sufficient condition on

what counts as real in the sense of serious ontological interest. Call such a condition

a criterion of ontological status.

I take the central tenet of Truthmaker Ontology to be the Truthmaker Criterion of

Ontological Status, or (TOS):

(TOS) x is real iff x makes a truthmaking contribution.

67
Some of Cameron’s remarks suggest a conflation between ontological commitment and
ontological status, for example: 'a really (or fundamentally) exists iff we are ontologically
committed to a' (2008a: 6); 'the ontological commitments of a theory have real being' (2010a:
250). (Cf. Rettler (2016: 1412): 'we are ontologically committed to the fundamental and all
and only the truthmakers are fundamental'.) Taken at face-value, these remarks suggest that
the truthmakers 'we are ontologically committed to' in fact really/fundamentally exist, as
opposed to their real/fundamental existence being a pre-condition of the truth of the theories
we accept. Cameron (2019) has recently clarified his position, however. See §5.2.3 for
discussion.
68
More on the methodological importance of the notion of ontological commitment in §5.1.1.
101
(TOS) is deliberately broad in two respects. First, making a 'truthmaking contribution'

needn't mean being a full truthmaker for some truth, but also applies to partial

truthmakers. Some may also wish to include components of truthmakers, such as a

mereological part of a complex truthmaking object, or a non-mereological constituent

of a truthmaking state of affairs. Second, there are different ways we might refine the

notion of making a truthmaking contribution. For example, some may wish to restrict

(TOS) to what makes an indispensable or a minimal truthmaking contribution.69 I will

discuss my preferred way of refining (TOS) in the next chapter, where I suggest that

the criterion should be restricted to what makes a fundamental truthmaking

contribution. For present purposes, however, we can set this matter to one side.

(TOS) is not a criterion of ontological commitment. It doesn't pertain to the

requirements theories impose on ontology, but to that which requirements on

ontology are requirements on.70 In other words, (TOS) is a claim about the scope of

ontology. It tells us that a complete and correct answer to the ontological question

would take the form of a list of everything that makes a truthmaking contribution;

whatever doesn't make a truthmaking contribution lacks ontological status and is

outside the ontologist's purview. Making a truthmaking contribution is what marks

the boundary between what is real, in the sense of serious ontological interest, and

what is not.

69
For discussion of minimal truthmakers, see Armstrong (2004: 19-20), O’Conaill and Tahko
(2016). MacBride (2013: §2.1.4.2) takes Armstrong to hold that ‘to be a genuine addition to
being is to be a net (indispensable) contributor to the schedule of truthmakers’.
70
Given (TOS), we will need some sort of truthmaker criterion of ontological commitment.
For if all and only the truthmakers have ontological status, then requirements on ontology are
requirements on truthmakers. So if we accept (TOS), the question is not whether we need a
truthmaker criterion of ontological commitment, but what kind. As we’ll see in Chapter 5,
spelling out the relevant notion of requirement is not a straightforward matter.
102
 Without attempting an analysis, is there more we can say about this notion of

being ‘real in the sense of serious ontological interest’, which (TOS) is supposed to

elucidate? I think the Truthmaker Ontologist has a couple of options here. The first

is broadly Finean.71 This involves claiming that ‘real’ is an unavoidably primitive

notion, albeit one on which we can get a decent intuitive and working grasp (Fine

2009: 175-176). For example, I suggested in §1.1 that the ontologically relevant sense

of ‘real’ is brought out by claims such as: ‘We think and talk about numbers, but the

question remains as to whether numbers are real’. Fine points to claims such as:

‘Democritus thought that there was nothing more to the world than atoms in the void’

(2009: 175); that is, Democritus thought that nothing but atoms in the void were real.

But perhaps we ultimately cannot 'define the concept of reality in essentially different

terms' (Fine 2009: 175). (TOS) might offer its services here, not as an analysis of

'real', but as an informative necessary and sufficient condition on what we should

count as real in the targeted sense. Construing the real as what makes a truthmaking

contribution might guide us towards a clearer understanding of how to approach the

question of what is real, even if 'real' itself resists analysis.72

A second option, which I’ll look at in more detail, is broadly Siderean.73 This

involves claiming that there is an elite candidate meaning for the existential quantifier

and that what is real, in the ontologically relevant sense, is what can truly be said to

exist using that quantifier. Against this background, (TOS) might be interpreted as

the claim that what can truly be said to exist using the elite quantifier are just those

things that make a truthmaking contribution (cf. Cameron 2010a, 2010b; Rettler

71
Cf. Fine (2001, 2009).
72
‘This search for a criterion for the real must be understood as a criterion for us to count
something as real; it will be a principle to apply in determining whether to accord that status,
given our current stage of epistemic development. There need not be, and probably cannot be,
any criterial mark of the real itself: the real is what is, period’. (Campbell 1994: 28).
73
Cf. Sider (2009, 2011).
103
2016). But what is it for there to be an elite ‘quantifier meaning’? And why think that

what can truly be said to exist using the elite quantifier are just the things that make

a truthmaking contribution?

For Sider, the elite meaning for the quantifier is the most eligible of its candidate

meanings, in the sense at issue in the Eligibility constraint on adequate interpretation

(see §1.4):

(Eligibility) With respect to some language L, if candidate

interpretation I1 assigns relatively more natural

meanings than candidate interpretation I2 then, all else

equal, I1 is a better interpretation of L than I2.

For Lewis (1983, 1984), Eligibility primarily constrains interpreters to assign more

natural, ‘joint-carving’ extensions to predicates. But Sider, recall, argues that it

extends to other parts of language, including logical terms like conjunction, negation

and quantification. He holds that reality has a distinguished logical structure—

including a quantificational structure— that exerts a reference-magnetic force on

such terms. The elite meaning for the quantifier, for Sider, is that which carves

reality’s objective quantificational structure perfectly at the joints. By his lights, the

question of serious ontological interest is 'What exists, in the perfectly natural sense

of "exists"?'.

I confess to finding the notion of ‘worldly’ logical structure a bit mysterious.74 I

see how certain meanings for names and predicates might be especially natural, in

virtue of the way reality is objectively 'carved' into objects and those objects into

property-classes. But I'm less sold on the idea that there is some kind of 'worldly

74
For similar concerns, see Contessa (2013), Hirsch (2013), Goldwater (2014), Schaffer (MS).
104
preference' for logical terms to have certain meanings.75 What is there, in reality, for

terms like 'And', 'Or' and 'There is' to latch onto? This is admittedly a blank stare

rather than an objection. Still, I prefer a different account of what elite quantification

involves.

On my preferred view, the claim that there is an elite 'meaning' for the existential

quantifier simply amounts to the claim that there is an elite domain of quantification

(cf. Brons 2014, Brouwer 2012, Schaffer MS). By way of illustration, consider

quantifier variantism (Hirsch 2011). On this view, there are multiple, unrestricted

'logically $-like' quantifier-meanings, such that '$xFx' can have different truth-values

depending on which meaning of '$' is in play. Moreover, since none of these

quantifier-meanings is metaphysically elite, debates about whether Fs exist are often

shallow; for in many cases, we could describe reality equally well using a language

in which 'Fs exist' is true or one in which it's false. (Sider's well-known response to

quantifier variantism is to accept the multiplicity of quantifier-meanings while

insisting that one of them—the perfectly natural quantifier— is metaphysically elite.)

But what is it for there to be multiple 'quantifier-meanings', in the sense at issue? I

assume a model-theoretic definition of '$' whereby, in a model with domain D,

‘$xfx’ is true iff {x : fx Ù x Î D} ≠ Æ

That is, '$xfx' is true iff f has a non-empty extension in D. Now let 'Universalese' be

a language whose quantifier is $U and let 'Nihilese' be a language whose quantifier is

$N. $U and $N are both 'logically $-like', obeying the standard introduction and

elimination rules for $. Yet $U and $N somehow differ in 'meaning' such that

'$U(Table(x)' is true but '$Nx(Table(x)' is false (with no shift in the meaning of 'Table').

75
So is Schaffer (MS), from whom I borrow this phrase.
105
What explains this difference? On the above definition of '$', it seems the only

possible explanation is that $U and $N have different domains, such that 'Table' has a

non-empty extension in the domain of $U and an empty extension in the domain of

$N. Given that $U and $N are logically $-like, there just doesn't seem to be anything

else that could differ. The quantifier variantist doesn't want to say that the difference

in quantifier-meanings results from domain restriction. So it looks like the difference

between $U and $N must be a difference between two distinct 'absolute' domains of

quantification.76

So I'm inclined to think that the claim that there are multiple candidate 'quantifier-

meanings' really boils down to the claim that there are multiple candidate domains.

76
You might argue that this is uncharitably hostile to quantifier variantism. (see Schaffer MS:
14). Consider the following argument (Sider 2009: 393-394; cf. Hawthorne 2006, Eklund
2009b). Suppose Quentin is a quantifier variantist and a native speaker of Nihilese. As a
speaker of Nihilese, Quentin wants to say:

(1) Tables do not exist.

But as a quantifier variantist, Quentin wants to say:

(2) 'Tables exist' is a true sentence of Universalese.

But it doesn't look like he can say both. For (2) to be true in Nihilese, the following would
also have to be true in Nihilese:

(3) 'Table' has a non-empty extension in the domain of $U.

But then it would have to be true, in Nihilese, that there is a domain of $U and that some
member of that domain falls under the extension of 'Table'. Thus, 'Tables exist' would have to
be true in Nihilese, contradicting (1)
But I don't see why Quentin should have to express his quantifier variantist claim in Nihilese.
As Schaffer points out (MS: 14), 'quantifier variantism involves semantic claims about these
object languages [i.e. Universalese and Nihilese], and in orthodox model-theoretic semantics
we are careful to couch semantic claims about object languages in the metalanguage'. So
Quentin should express his nihilism in Nihilese, but his quantifier variantism in a
metalanguage with a more encompassing domain, in which he can consistently say:

106
The claim that there is no elite 'quantifier-meaning' amounts to the claim that there is

no metaphysically elite choice of domain. And the claim that there is an elite

'quantifier-meaning' amounts to the claim that there is a metaphysically elite domain.

Schaffer (MS: 11) calls the latter claim Domain Realism.

Whether you’re amenable to Domain Realism depends, basically, on the strength

of your realist convictions. If you think that reality comes 'pre-carved into objects'

(Cameron 2010b: 17), you'll be inclined to think that there is an elite domain

containing just those objects. This domain will make for 'loaded' quantification, in

that Fs can truly be said to exist, using a quantifier with the elite domain, just in case

Fs really are out there among the pre-carved objects. Beyond this, however, there are

different constraints Domain Realists might appeal to in order to characterise the elite

domain.

As I said back in §1.2, I don't think that existential quantification is inherently

loaded in virtue of its semantics. The loadedness or otherwise of a quantifier is fixed

by how the domain is specified in the metalanguage, and not all specifiable domains

make for loaded quantification. There is no logical obstacle to defining a domain D

such that D =df. {x : x is imaginary}, but this presumably won't make for loaded

quantification. Nor, in my view, will the maximally encompassing domain,

containing absolutely everything that can be quantified over. As I see it, this will just

be the superset of all specifiable domains, including the domain of imaginary objects,

the domain of impossible objects, and so on.

(4) 'Table' has a non-empty extension in the domain of $U and an empty extension in the
domain of $N.

See Schaffer (MS) for further discussion.

107
 But there are more discerning constraints we might use to characterise the elite

domain. For example, we might think that the elite domain will comprise just those

things that are indispensable to our best overall theory of the world (cf. Sider 2011).

This does not tell us in advance what the specific contents of the elite domain will be

(e.g. whether it will contain only spacetime points and sets). It is more like a

methodological recommendation: finding out what is indispensable to our best theory

is the way to determine the contents of the elite domain, hence the range of the elite

quantifier.

In this spirit, we might think of (TOS) as the claim that the metaphysically elite

domain comprises just those things that make a truthmaking contribution. Again, this

does not tell us whether the elite domain contains particular entities, like spacetime

points or sets. But it offers relatively tractable guidance as to how to determine the

contents of the elite domain—hence how to answer the ontological question. The

suggestion is that what is real, in the sense of ontological interest, is what can truly

be said to exist using the metaphysically elite quantifier, which has that status in

virtue of ranging over the metaphysically elite domain; and we should think of the

elite domain as containing just those things that make a truthmaking contribution.

This is one way to make sense of (TOS)'s claim that x is real iff x makes a truthmaking

contribution.

But why should we accept (TOS)? Why should we think that what is real in the

sense of ontological interest (or what makes up the metaphysically elite domain) is

just what makes a truthmaking contribution? I doubt that this is the sort of claim that

could be established via deductive argument. It is a posit— one to be accepted or

rejected based on whether it makes for an attractive picture of ontological inquiry.

For example, we take the aim of ontology to be to discover what makes a truthmaking

contribution, does this tally with the idea that ontological questions are substantive

108
and deep, rather than trivial or shallow? Is there a viable methodology for settling

ontological questions, construed as questions about truthmakers? What implications

does (TOS) have for first-order ontological debate? We won't be in a position to

answer these questions until we have a fuller picture of Truthmaker Ontology, which

will unfold over the chapters to follow.

(TOS) should seem at least initially plausible, however— at least to those

sympathetic to the notion of truthmaking. Ex hypothesi, a truthmaker is a part of extra-

representational reality—something out there in the world— that accounts for the

truth of some truthbearer. Given this, it is natural to think that if x makes a

truthmaking contribution then x is real. Conversely, if x is real, it is natural to think

that x makes a truthmaking contribution. For if x is real, there will presumably be

truths that wouldn't otherwise hold, and for which x is responsible— the truth that x

exists, for one.77 So, intuitively, whatever makes a truthmaking contribution is real

and vice versa.

Furthermore, it is initially plausible to think that if (TOS) is true, ontological

questions have the character ascribed to them by the serious conception of ontology.

The founding intuition of truthmaker theory is the idea that representational truth

depends, cross-categorially, on extra-representational reality. So the search for

truthmakers is, by design, an attempt to get beyond our representations of the world

to the reality that underlies them. This resonates, prima facie, with the aims of Serious

Ontology. Serious Ontologists, as Williams puts it (2010: 107), 'are not just interested

in what the true sentences or propositions are: we are interested in the way reality is,

in the objects and properties and their arrangements that support the truth of the

propositions. Truthmaker theory is, basically, just a way of trying to understand the

77
I take it that if x is real then ⟨x exists⟩ is true, but I will deny the converse.
109
relevant notion of 'support'. And as will shortly emerge, when (TOS) is combined

with the Divergence Thesis, we have a coherent way to distinguish what has

ontological status from what can truly be said to exist— a distinction which is crucial,

I have argued, if ontology is to be a serious enterprise.

I argued above that Truthmaker Ontology should be founded on a truthmaker

criterion of ontological status, (TOS), rather than a criterion of ontological

commitment. I have now sketched a way to clarify (TOS) and have given some

reasons to think it initially plausible. I end this section, however, on a promissory

note. (TOS) is an axiom, which is ultimately to be assessed based on whether the

approach based upon it offers a useful way to think about the subject matter and

method of ontology— this has yet to be established.

3.3 The Divergence Thesis

My overall aim in this chapter is to explain how Truthmaker Ontology supports a

coherent what is real, in the sense of ontological interest, and what can truly be said

to exist. (TOS) doesn't let us do this by itself. For perhaps everything that can truly

be said to exist must be taken to make a truthmaking contribution. To get the desired

distinction, we have to combine (TOS) with the Divergence Thesis.

Roughly, the Divergence Thesis says that that the truthmakers for a claim need not

involve the entities said or implied to exist by that claim. Truthmaker Ontologists

express the Divergence Thesis in different ways, for example: 'Truthmakers may or

may not involve the existence of entities apparently referred to by terms in the

sentence' (Dyke 2007: 4); 'If "a is F" is true, this need not be because there is some

entity corresponding to a, some property corresponding to F, and the entity, a,

possesses the property F' (Heil 2003: 55); and 'the truthmakers for certain sentences

[…] do not involve the entities which the sentence apparently refers to or quantifies
110
over' (Melia 2005: 77). I will focus on the Divergence Thesis as it applies to

existential propositions (cf. Cameron 2008a: 5):

(Divergence Thesis) If ⟨x exists⟩ is made true, it does not

follow that x is a truthmaker for ⟨x

exists⟩.

(Likewise, if a general existence proposition of the form ⟨Fs exist⟩ is made true, it

does not follow that Fs are among the truthmakers. The same goes, mutatis mutandis,

for propositions of the form ⟨There is an x⟩ and ⟨There are Fs⟩. Since I take the

truthmakers for a sentence to be the truthmakers for the proposition that sentence

expresses, we can also say that x need not be a truthmaker for 'x exists', Fs need not

be among the truthmakers for 'There are Fs', and so on.)

The Divergence Thesis is controversial. While it is generally agreed that truths can

have several truthmakers, truthmaker theorists often assume that ⟨x exists⟩ always has

x as at least one of its truthmakers (e.g. Armstrong 2003, Rodriguez-Pereyra 2002).

The Divergence Thesis implies that they are wrong: x need not make any truthmaking

contribution to ⟨x exists⟩. Friends of the Divergence Thesis should explain why these

other truthmaker theorists are mistaken. Moreover, the Divergence Thesis faces at

least two potential objections:

Aboutness. Truthmakers are supposed to explain the truth of

truthbearers. And some argue that an adequate truthmaker-explanation

for P must appeal, at least in part, to what P is about. Call this the

Aboutness Constraint (cf. Baron 2013b, Baron et. al 2019, Merricks

2007, Schipper 2019). But the Divergence Thesis permits violations of

the Aboutness Constraint; it implies, for example, that the truthmaker-

111
 explanation for ⟨x exists⟩ need not appeal to x at all. Thus, one might

worry that the Divergence Thesis permits inadequate truthmaker-

explanations.

Identification. Truthmaker Ontologists tell us to decide our ontology

by asking what it takes to make true the truths we accept. Given the

Divergence Thesis, existence claims may or may not be made true by

what is claimed to exist. But how are we supposed to tell whether, for

example, it takes Xs to make true ⟨Xs exist⟩, or if something else would

do? Without some guidance as to when and why the alleged divergence

occurs, it's not clear how we're supposed to decide what ontology we

need to make true the existential truths we accept. Thus, one might

worry that the Divergence Thesis leaves Truthmaker Ontologists

without a workable ontological method (cf. Hale and Wright 2009:

186, Thomasson MS).

But despite its controversial status, proponents of the Divergence Thesis rarely argue

for it in detail. In what follows, I will offer an explanation of why the Divergence

Thesis holds and argue that the account meets the objections above.

3.3.1 Simple and Substantial Truthmaking Claims

Schulte (2011a, 2011b) distinguishes between simple and substantial truthmaking

claims. In a simple truthmaking claim, the terms used to express the truthbearer

reoccur in the description of the truthmaker, as in:

(Sim) Socrates makes true ⟨Socrates exists⟩.

112
(Sim) seems fairly uninformative. Consider the corresponding explanatory claim:

'⟨Socrates exists⟩ is true in virtue of Socrates'. As Schulte points out (2011b: 418), the

predicate 'is true' is all that prevents this claim being outright trivial: 'Socrates exists

in virtue of Socrates' is not a genuine explanation at all.

In a substantial truthmaking claim, the terms used to express the truthbearer differ

from those used to describe to truthmaker, as in:

(Sub) Simples arranged Socrates-wise make true ⟨Socrates exists⟩.

(Sub) seems more informative than (Sim). The corresponding explanatory claim,

'⟨Socrates exists⟩ is true in virtue of simples arranged Socrates-wise', remains

explanatory when the truth-predicate is eliminated.

Since truthmaker theorists sometimes appeal to substantial truthmaking claims, the

difference between substantial and simple truthmaking claims is of independent

interest. My interest, however, is in how an account of substantial truthmaking claims

might shed light on the Divergence Thesis. For it might be the case that whenever a

truth has a 'substantial' truthmaker it also has a 'simple' truthmaker. If (Sub) is true,

for example, perhaps (Sim) must also be true. However, if we can explain why

existence claims can have substantial truthmakers without the need for bespoke

simple truthmakers, we will have explained why the Divergence Thesis holds. We

could claim, for example, that ⟨Socrates exists⟩ is made true by simples arranged

Socrates-wise but not by Socrates. Is there an account of substantial truthmaking

claims that allows this?

Schulte's interesting proposal is that substantial truthmaking claims involve two

distinct explanatory components. The first is a truthmaking claim connecting a truth

to a simple truthmaker. The second is a reductive explanation of the existence of this

113
truthmaker in terms of the existence of something else (Schulte 2011b: 419). For

example, on Schulte's view, (Sub) is really a truncated version of the following:

(SubRE) (i) Socrates makes true ⟨Socrates

exists⟩.

(ii) The existence of simples

arranged Socrates-wise reductively

explains the existence of Socrates.

Schulte thus seems to be appealing to the following Reduction Principle:

(Reduction Principle) If y ⊨ P, and the existence of x

reductively explains the existence of y,

then x ⊨ P.

Given the Reduction Principle, (SubRE) implies (Sub). By Schulte's lights, it is the

presence of the additional reductive explanation that makes (Sub) more informative

than (Sim).

I like the shape of this proposal, but it doesn't give friends of the Divergence Thesis

what they want. Schulte is not clear whether he thinks that reductive explanation

yields identity (i.e., whether ‘the existence of x reductively explains the existence of

y’ entails that x = y). But suppose, first, that it doesn't. Then substantial truthmaking

claims require us to posit both a substantial truthmaker and a distinct simple

truthmaker. For example, by (SubRE), we'd need Socrates as a bespoke simple

truthmaker for ⟨Socrates exists⟩ as well as the simples explicitly cited in (Sub). This

is what the Divergence Thesis is supposed to avoid. Suppose, on the other hand, that

114
reductive explanation does yield identity. Then substantial truthmaking claims will

commit us to some contentious metaphysics. For example, the claim that simples

arranged Socrates-wise make true ⟨Socrates exists⟩ will commit us to composition-

as-identity; the claim that Socrates's being in neural state N makes true ⟨Socrates is

in pain⟩ will commit us to an identity theory of mind. My Truthmaker Ontologist

would rather avoid this sort of baggage.

Another potential account, with a similar shape to Schulte's, is based on

Armstrong's (2004: 10) Entailment Principle:

(Entailment Principle) If x ⊨ P and P ® Q then x ⊨ Q.

Given the Entailment Principle, we might take (Sub) to be a truncated version of the

following:

(SubEP) (i) Simples arranged Socrates-wise

make true ⟨Simples arranged Socrates-

wise exist⟩.

(ii) ⟨Simples arranged Socrates-wise

exist⟩ entails ⟨Socrates exists⟩.

By the Entailment Principle, (SubEP) implies that simples arranged Socrates-wise

make true ⟨Socrates exists⟩. But this time, there is no need to posit Socrates as a

bespoke simple truthmaker: the simples by themselves do all the truthmaking needed,

and so are 'ontology enough' to account for the truth of ⟨Socrates exists⟩. However,

this account has a major disadvantage: it fails to ensure the explanatory relevance of

substantial truthmakers to what they make true (see §3.4). For example, if simples

115
arranged Socrates-wise make true ⟨Socrates exists⟩ they will also make true

⟨Everything is self-identical⟩.

So the entailment account gives us too little explanatory relevance, while the

reductive explanation account gives us too much ontology (or commits us to some

contentious metaphysics). But a third account, with the same general shape, gives us

the best of both worlds. This account is based on the Grounding Principle, and the

associated distinction between direct and indirect truthmaking (see §2.4):

(Grounding Principle) If x ⊨ P and P Þ Q then x ⊨ Q.

(Direct Truthmaking) x directly makes P true iffdf. x ⊨ P &

¬$Q(x ⊨ Q & Q Þ P).

(Indirect Truthmaking) x indirectly makes P true iffdf. x ⊨ P &

$Q(x ⊨ Q & Q Þ P).

Given the Grounding Principle, we might take (Sub) to be a truncated version of the

following:

(SubGP) (i) Simples arranged Socrates-wise

make true ⟨Simples arranged Socrates-

wise exist⟩.

(ii) ⟨Simples arranged Socrates-wise

exist⟩ grounds ⟨Socrates exists⟩.

By the Grounding Principle, (SubGP) implies that simples arranged Socrates-wise

indirectly make true ⟨Socrates exists⟩. In general, I submit that substantial

116
truthmaking is indirect truthmaking: truths are connected to their substantial

truthmakers via intermediate grounding relations. In contrast, simple truthmaking is

direct truthmaking: truths are connected to their simple truthmakers directly, without

any intermediate grounding relations. So the reason substantial truthmaking claims

seem more informative than simple truthmaking claims is that the former involve an

extra explanatory component, i.e., one or more intermediate grounding relations. This

has the same advantage as the entailment account. We can say, for example, that

simples arranged Socrates-wise are all it takes to make ⟨Socrates exists⟩ true, without

the need for Socrates as a bespoke simple truthmaker. But it also ensures the

explanatory relevance of substantial truthmakers to what they make true. For

example, if simples arranged Socrates-wise make true ⟨Socrates exists⟩, it does not

follow that they make true ⟨Everything is self-identical⟩, which is entailed, but not

grounded, by ⟨Socrates exists⟩.

Let's now apply this account of substantial truthmaking to the Divergence Thesis:

(Divergence Thesis) If ⟨x exists⟩ is made true, it does not

follow that x is a truthmaker for ⟨x

exists⟩.

First, consider the following claim: if x indirectly makes true ⟨y exists⟩, then x ≠ y.

This follows from Quasi-Asymmetry, which I have previously defended (§3.4):

(Quasi-Asymmetry) "x"P((x ⊨ P) ® ¬(P Þ ⟨x exists⟩))

Suppose ⟨a exists⟩ is indirectly made true by some x. Then there is some P such that

x ⊨ P & P Þ ⟨a exists⟩. Now suppose that x = a. Then a ⊨ P & P Þ ⟨a exists⟩. But,

117
by Quasi-Asymmetry, ¬(a ⊨ P & P Þ ⟨a exists⟩). Contradiction. So any indirect

truthmaker for ⟨x exists⟩ must be something other than x.

(In contrast, I think that if x directly makes true ⟨y exists⟩, then x = y. I have no

direct argument for this claim, but I find it very plausible. If x makes true ⟨y exists⟩

and x ≠ y, it is natural to think that there must be some grounding connection between

the truths ⟨x exists⟩ and ⟨y exists⟩. Otherwise, it just seems to be a brute fact that ⟨y

exists⟩ is true in virtue of something that bears no immediate connection to its

content.)

Now, suppose that ⟨x exists⟩ is indirectly made true by some y. By the previous

argument, x ≠ y. Thus, ⟨x exists⟩ is made true by something other than x. Now since

a truth can have several truthmakers, it might also be the case that ⟨x exists⟩ is directly

made true by x. But since y itself is a sufficient truthmaker for ⟨x exists⟩, ⟨x exists⟩

might be made true without help of x. And that's all the Divergence Thesis requires:

if ⟨x exists⟩ is made true, it does not follow that x is a truthmaker for ⟨x exists⟩,

because it might be that ⟨x exists⟩ is indirectly made true but is not directly made true.

In sum, the Grounding Principle offers a natural and attractive explanation as to how

existence claims need not be made true by what is claimed to exist. (And the account

can be generalised beyond existence claims. For example, the predication ⟨Socrates

is in pain⟩ might be indirectly made true as follows: (i) [Socrates is in neural state N]

⊨ ⟨Socrates is in neural state N⟩; (ii) ⟨Socrates is in neural state N⟩ Þ ⟨Socrates is in

pain⟩. If so, then the fact that Socrates is in neural state N might be all it takes to make

true ⟨Socrates is in pain⟩, without the need to posit the fact that Socrates is in pain as

a bespoke simple truthmaker.)

118
3.3.2 Replies to Objections

According to the Aboutness Objection, the Divergence Thesis permits violations of

the Aboutness Constraint, which says that an adequate truthmaker-explanation for P

must at least partly appeal to what P is about. To see the Aboutness Constraint in

action, consider presentism: the view that everything that exists exists presently.

Presentism is thought to face the challenge of finding presently existing truthmakers

for truths about the past, such as ⟨Shakespeare existed⟩. One strategy is to posit tensed

facts, such as the fact that the world is such that it once contained Shakespeare

(Bigelow 1996). But some object to this on the grounds that it violates the Aboutness

Constraint (Baron 2013b, Merricks 2007). Intuitively, ⟨Shakespeare existed⟩ is about

Shakespeare, rather than the world's instantiating a certain tensed property. Hence,

the objectors argue, the presentist's truthmaker-explanation doesn't pass muster.

But by my lights, there is no obstacle (in principle) to the presentist's truthmaker-

explanation being correct. For perhaps ⟨Shakespeare existed⟩ is not directly made true

by anything involving Shakespeare himself, but is indirectly made true as follows:

(i) [The world is such that it once contained Shakespeare] ⊨ ⟨The

world is such that it once contained Shakespeare⟩.

(ii) ⟨The world is such that it once contained Shakespeare⟩ Þ

⟨Shakespeare existed⟩.

That is, perhaps [The world is such that it once contained Shakespeare] indirectly

makes true ⟨Shakespeare existed⟩, and so the truthmaker-explanation for

⟨Shakespeare existed⟩ might not appeal to Shakespeare. So, given the Aboutness

119
Constraint, it looks like the Divergence Thesis permits explanatorily inadequate

truthmaker-explanations.

I accept that the Divergence Thesis permits violations of the Aboutness Constraint.

But I reply that, on my account of how these violations occur, retaining the Aboutness

Constraint is unmotivated. The main motivation for the Aboutness Constraint is that

we can't otherwise ensure the explanatory relevance of truthmakers to what they make

true. Thus, Merricks (2007: 22-34) and Schipper (2019) claim that we need the

Aboutness Constraint to avoid the awkward consequences of the view that

truthmaking is nothing more than necessitation, e.g. that your coffee mug is a

truthmaker for ⟨2 + 2 = 4⟩. Baron (2013b: 550) argues that rejecting the Aboutness

Constraint leads to a kind of 'permanent epistemic chaos'. For example, suppose I

know some true P about the mating habits of pandas. If the Aboutness Constraint is

false, Baron claims, then for all I know P might be true in virtue of something

completely irrelevant, such as your coffee mug.

But it doesn't follow from the falsehood of the Aboutness Constraint that

truthmaker might be explanatorily irrelevant to what they make true. Explanatory

relevance is ensured as long as the truthmakers are appropriately connected to what

their truths are about (cf. Baron 2013b: 555). And by my lights, violations of the

Aboutness Constraint occur in cases of indirect truthmaking, where truths are

connected to truthmakers via one or more intermediate grounding relations; and truths

must be explanatorily relevant to what they ground. Perhaps P, the truth about the

mating habits of pandas, is grounded by certain physiological truths, which are

themselves grounded by certain chemical truths, and so on until we ultimately reach

certain truths about the existence and features of subatomic particles. Then we might

think that P is indirectly made true by facts involving the existence and features of

these subatomic particles. This violates the Aboutness Constraint, since P is not about

120
subatomic particles. But intuitively, our indirect truthmakers are relevant to the truth

of P, since they are appropriately connected to P via a series of explanatorily relevant

grounding relations. In contrast, your coffee mug will not be an indirect truthmaker

for P, or for ⟨2 + 2 = 4⟩: any truths your coffee mug directly makes true (⟨Your coffee

mug exists⟩, perhaps) are not plausibly connected by grounding to these truths. So

although my account permits violations of the Aboutness Constraint, it still ensures

the explanatory relevance of truthmakers to what they make true.

According to the Identification objection, the Divergence Thesis threatens to leave

Truthmaker Ontologists without a workable ontological method. This is because there

is no clear procedure for telling whether an existence claim must be made true by

what it says there is, or whether something else would suffice. Must we admit tables

into our ontology, for example? According to Truthmaker Ontologists, this question

largely hinges on whether sentences like 'Tables exist' express propositions that

require tables as truthmakers; if simples arranged table-wise are all it takes to make

these sentences true then we can get by without tables in our ontology. But given the

Divergence Thesis, we can't tell whether this is so just by examining the table-

involving truths we accept. So how are we supposed to tell whether we need tables

as truthmakers, hence whether we need them in our ontology?

Talk of what's needed to make sentences true takes us into the arena of ontological

commitment. Given that ontology consists of all and only the truthmakers, the

ontological commitments of a theory are its requirements on truthmakers. The

objection is asking, effectively, how are we supposed to tell which truthmakers a

theory requires, given that existence claims may or may not be made true by what is

claimed to exist. I won't fully address this challenge until Chapter 5, where I go into

detail about how Truthmaker Ontologists should understand and identify the

ontological commitments of theories. Here, I'll just say something about Thomasson's

121
demand that Truthmaker Ontologists provide what she calls a negative

methodological principle (MS: 4-5):

[E]ven if truthmaker theorists don't give us a recipe for determining

what we are committed to, they must at least provide a criterion that
will tell us when we aren't committed to entities of a kind apparently
referred to in a given sentence or theory. Under what conditions can
we say that an entity isn't needed as a truthmaker and so deny the
apparent commitment?

Here's my suggestion. Recall that, given Quasi-Asymmetry, any indirect truthmaker

for ⟨a exists⟩ must be something other than a. And ⟨a exists⟩ will have an indirect

truthmaker as long as it is grounded by some P such that P has a truthmaker (the latter

clause being redundant, of course, if Truthmaker Maximalism is assumed). So the

question of whether a is needed as a truthmaker for ⟨a exists⟩ reduces to the question

of whether ⟨a exists⟩ is grounded by some truth (with a truthmaker). If it is, then

whatever makes true the truth that grounds ⟨a exists⟩ will suffice to make true ⟨a

exists⟩. It doesn't follow that a does not make true ⟨a exists⟩. But it does follow that

a need not make true ⟨a exists⟩, since the indirect truthmaker would suffice by itself.

Thus, the negative methodological principle I'd suggest is as follows: if ⟨a exists⟩ is

plausibly grounded by some other truth(s), then a plausibly needn't be among the

truthmakers for ⟨a exists⟩. Of course, developing a positive account what is required

to make a given claim true is another matter— I'll postpone this until Chapter 5.

122
3.4 Mere and Real Existence

I will now clarify and defend the distinction between what merely and what really

exists.

3.4.1 The Distinction

By the Divergence Thesis, x need not be among the truthmakers for ⟨x exists⟩. Hence,

⟨x exists⟩ might be made true even if x doesn't do any truthmaking. By (TOS), x counts

as real— as having ontological status— just in case x makes a truthmaking

contribution. So ⟨x exists⟩ might be made true (hence true simpliciter) even if x lacks

ontological status. Thus, existential truth and ontological status come apart: what is

real, in the sense of ontological interest, might be a proper subset of what can truly

be said to exist. Following Cameron (2008a), we might articulate this by introducing

a pair of technical terms: merely exists and really exists:

(Real Existence) x really exists iffdf. x makes a truthmaking

contribution.

(Mere Existence) x merely exists iffdf. ⟨x exists⟩ is true but x does

not make a truthmaking contribution.

The view, then, is that ontology does not concern what exists but the subset thereof

that really exists, i.e. makes a truthmaking contribution.

It's crucial to be clear about the nature of this distinction. Earlier, I distinguished

between inflationary and deflationary forms of Stratified Ontology— the broad

family of metaontological views that some sort of distinction between what exists and

what really, or fundamentally, exists. The distinction above is intended in a

deflationary spirit. The thought is not that reality divides into the 'elite' entities that

123
make a truthmaking contribution and the 'non-elite' entities that don't. The thought is

that, if ⟨Fs exist⟩ is true, but Fs don't do any truthmaking, then Fs aren't part of reality

at all— out there in the world, there are really no such things. Yet representations

concerning the existence of Fs can still be made true by an F-less reality. Given this,

ontologists need an expressive device to distinguish what can truly be said to exist

from what has ontological status (cf. Williams 2010): that is the function of the

technical term really exists. To quote Cameron (2008a: 7):

Do not think of the distinction as dividing the entities in the world

into the privileged real existents and the impoverished unreal
existents. All there is in the world are the real existents: once you've
said what there really is, you have answered the ontological
question. The distinction between mere existence and real existence
is just a way of talking.

3.4.2 Is the Distinction Coherent?

If I'm labouring the point, it's because I'm anxious to avoid the sort of

misunderstanding on display in a recent paper by Schipper (2019). Schipper

distinguishes between deflationary and inflationary forms of Truthmaker Ontology.

His deflationary Truthmaker Ontologist thinks that 'reality is exhausted by

fundamental reality', which consists of all and only the truthmakers (2019: §3, my

emphasis). His inflationary Truthmaker Ontologists 'takes reality not to be exhausted

by fundamental reality (rather, it consists of both fundamental and non-fundamental

reality)', but insists that only fundamental entities do any truthmaking (ibid.). My

view resembles that of Schipper's deflationary Truthmaker Ontologist (though I

prefer to talk in terms of real existence, rather than fundamentality). But according to

Schipper, both views face serious difficulties. Briefly, his objection to the inflationary

view is that there is no principled basis on which to rule out real, albeit non-

fundamental, entities as truthmakers. His objection to the deflationary view is that it

124
falls into contradiction. For according to Schipper, deflationary Truthmaker

Ontologists are 'nihilists about what is not fundamental', endorsing the following

schema (2019: §4):

(S) ¬(In fundamental reality, f) ® ¬f.

Yet Schipper's deflationist also wants to say that (for example) 'Tables exist' is true,

albeit made true by a table-less 'fundamental reality'. But if 'Tables exist' is true, then

tables exist. And if ¬(In fundamental reality, tables exist), it follows by (S) that

¬(Tables exist). Hence, Schipper concludes, deflationary Truthmaker Ontology is

incoherent.

Schipper's assumption that deflationary Truthmaker Ontologists are committed to

(S) rests on a general confusion about the distinction between inflationary and

deflationary forms of Stratified Ontology. 'Inflationists', he says, 'think that reality is

layered or has levels; everything that is part of fundamental reality is at one level, but

there is at least one other level […] containing everything non-fundamental' (2019:

§3). So far so good. But he continues (op. cit.):

By contrast, according to the deflationist, reality is not layered and

consists only of fundamental reality […]. [E]verything which
deflationists might call 'derivative' is not, as the inflationist would
construe it, 'merely existing', but just not existing at all.

Schipper is assuming here that whatever exists must be part of reality. Given this, the

deflationist's claim that 'reality […] consists only of fundamental reality' entails the

non-existence of the non-fundamental. This would explain why Schipper thinks

deflationary Truthmaker Ontologists are committed to (S), leading to self-

contradiction. But to assume that whatever exists (or perhaps better: can truly be said

to exist) must be part of reality just begs the deflationist's question. The whole crux
125
of deflationary Stratified Ontology is that an F-less reality can support the literal truth

of representations like 'Fs exist'; hence the need to distinguish what can truly be said

to exist from what really (or fundamentally) exists. Truthmaker Ontology, as I’ve

presented it, offers a specific account of how this distinction works: given that reality

is exhausted by the things that make a truthmaking contribution, and ⟨Fs exist⟩ can

be made true indirectly by non-Fs, it is perfectly consistent to think that Fs can truly

be said to exist yet aren't part of reality. By lumbering deflationary Truthmaker

Ontologists with schema (S), Schipper ignores all this, simply assuming that what is

real and what can truly be said to exist can't be intelligibly distinguished.

But perhaps there are independent grounds for denying that the distinction is

intelligible. One might argue that the truth of 'P' implies that reality is such that P. So

it can't be true that x exists unless reality is such that x exists, which seems equivalent

to x's being real. Given this, the claim that x might truly be said to exist despite being

unreal sounds incoherent (cf. Eklund 2009a: 316).

But there is reason to doubt that true representations mirror reality as closely as

this objection assumes. Consider Sider's (2013a) story of Nihilo, a god who creates a

world made up solely of simple particles:

On the first day Nihilo creates some particles and arranges them in
beautiful but lifeless patterns. He becomes lonely, so on the second
day he creates some minions (or rather, particles arranged minion-
wise). On the third day he tries to teach his minions so speak. But
this goes badly. The dim-witted minions struggle to understand
Nihilo's talk of subatomic particles and their physical states. So on
the fourth day he teaches them an easier way to speak. Whenever an
electron is bonded (in a certain way) to a proton, he teaches them to
say 'there is a hydrogen atom'; whenever some subatomic particles
are arranged chair-wise he teaches them to say 'there is a chair', and
so on. (Pretend that electrons and protons have no proper parts.)
(Sider 2013a: 248-249).

126
Intuitively, chairs and other complex objects are unreal in the scenario described: all

Nihilo created when he created the world were simple particles. But when the minions

utter sentences like 'There is a chair', I think they plausibly speak the truth. For it is

reasonable to suppose that the minions are best interpreted as speaking a

metasemantically tolerant language, in which 'There is a chair' expresses a truth just

in case reality hosts some simples arranged chair-wise (see §1.4). It thus seems at

least coherent to think that chairs can truly be said to exist in the minions' language

despite being unreal. Indeed, it is conceivable that conventions of English use are

such that 'There is a chair' might be true in our language, even if we inhabit a chair-

less reality. If this is a coherent possibility, as I think it is, then there is an intelligible

sense in which what is real can come apart from what can truly be said to exist.

Alternatively, consider sentences like 'There is a fictional space explorer called

Picard' and 'There is a shortage of staff in the NHS' (see §1.2). I submit that the three

following claims are highly plausible: (i) These sentences are both true; (ii) When

speakers typically utter these sentences, they are not speaking non-strictly, or non-

literally; (iii) Reality does not host such things as fictional space explorers or

shortages or staff. If the strict and literal truth of 'There are Fs' implies the reality of

Fs, one of these plausible assumptions must be given up. Isn't it at least somewhat

reasonable to conclude that, instead of giving up (i)-(iii), we should instead revise our

assumptions about the relation between reality and existential truth?

Here's Williams (2012: 171) with a nice description of the sort of alternative

picture deflationary Stratified Ontology allows for:

Mind-independent reality is thus-and-so (perhaps an ontology of

concreta spread through space-time, characterised by the
instantiation of various natural properties). The relation between this
reality and representations of it—including the sentences of natural
language and the language of thought— is complex and demands

127
 analysis. When the dust settles, some of these representations will
turn out to be true. But—for example— all that may be required of
reality for the representation 'there are tables' to be true, is that
certain simple particles stand in certain arrangements.

The Truthmaker Ontologist adds that the relevant connection between reality and true

representations of it is the truthmaking relation A table-less ontology might support

the truth of 'There are tables' because the proposition expressed might be indirectly

made true by simples arranged thus-and-so. If so, then the simples have genuine

ontological status and are thus said to really exist. Tables can truly be said to exist

but have no ontological status, and therefore merely exist. We could tell a similar

story about sentences like 'There is a fictional space explorer called Picard' or 'There

is a chair', as spoken by Nihilo's minions. Such sentences are true; but perhaps reality

doesn't need to host fictional space explorers or chairs in order to adequately account

for their truth. I see no good reason to think this picture incoherent or unintelligible.

Conclusion

If ontology is to be a serious enterprise, we need a way to coherently distinguish

ontological questions from questions about what there is. I have argued that

Truthmaker Ontology offers one. Since all and only the truthmakers have ontological

status, and since Fs need not be among truthmakers for ⟨There are Fs⟩, an F-less

ontology can support the truth of ⟨There are Fs⟩. Truthmaker Ontology thus emerges

as a deflationary form of Stratified Ontology, on which ontological status comes apart

from what can truly be said to exist. We can articulate this by distinguishing what

really exists (i.e. what makes a truthmaking contribution) and what merely exists (i.e.

what can truly be said to exist but makes no truthmaking contribution). The claim,

128
then, is that ontology only concerns what really exists; and we've already begun to

see why this might appeal to Serious Ontologists.

129
4 Fundamental Truthmakers
We now have a general idea of what Truthmaker Ontology is and how it diverges

from Orthodox Ontology. The key claim is that the task of ontology is to determine

what makes a truthmaking contribution— what really exists— which might be a

proper subset of what can truly be said to exist. Here I suggest a way to refine this

proposal. I will do so in the course of propounding a theory of fundamentality: an

account of what is fundamental and how the fundamental relates to the non-

fundamental. Briefly, on the theory proposed, a collection of fundamental

truthmakers provides the ontological foundation for an abundant, layered hierarchy

of fundamental and non-fundamental truths. I claim that this theory offers an

independently attractive way to think about the contentious notion of fundamentality.

But moreover, it guides Truthmaker Ontologists towards a clearer understanding of

what it takes to answer the ontological question. For the fundamental truthmakers, I'll

suggest, are the only things we need to take to make a truthmaking contribution; hence

they are all we need to admit to our ontology.

In §4.1, I outline some general aims and desiderata of a theory of fundamentality.

I also consider some of the benefits and drawbacks of accounting for fundamentality

in terms of truthmaking and/or grounding (broadly construed). I then set about

developing a theory of fundamentality, based on the distinct relations of truthmaking

and (truth-)grounding. First, in §4.2, I use grounding to define what I take to be the

core properties of fundamental truth—basicness and integrality— and to elucidate

the interaction between them. Then, in §4.3, I introduce the notion of a fundamental

truthmaker: roughly, a member of a collection of entities that provides a non-

redundant truthmaking base for the fundamental truths. I argue, inter alia, that the

fundamental truthmakers can provide a non-redundant truthmaking base for all truths

130
and are therefore all we need to take to make a truthmaking contribution. In §4.4, I

consolidate the foregoing results into a theory of fundamentality and explain its

relevance to Truthmaker Ontology.

4.1 Fundamentality

In broad terms, a theory of fundamentality is an account of (i) what is fundamental

and (ii) how the fundamental relates to the non-fundamental.78 Not all theories of

fundamentality, however, target quite the same phenomenon.

Some theories are primarily concerned with ontological fundamentality. Such

theories treat ‘fundamental’ and ‘non-fundamental’ as predicates of non-

representational, ‘worldly’ entities (e.g. quarks, cities, sets). Roughly speaking,

theories of ontological fundamentality aim to capture the idea that there are certain

ontologically basic 'building blocks' on which the rest of reality ultimately depends.

Schaffer (2009), for example, distinguishes between fundamental and non-

fundamental entities in terms of the entity-grounding relation (see §2.1): x is a non-

fundamental entity iff something entity-grounds x, and a fundamental entity iff

nothing does.

Other theories are primarily concerned with alethic fundamentality. On such

theories, 'fundamental' and 'nonfundamental' are seen as predicates of

representational entities (e.g. true sentences, true propositions, sub-sentential terms).

Roughly speaking, such theories aim to capture the idea that there are certain

explanatorily basic or metaphysically perspicuous truths, in virtue of which all others

hold. Some distinguish between fundamental and non-fundamental truths in terms of

truth-grounding: non-fundamental truths are grounded by fundamental truths, which

78
See Fisher (2016: 449), Sider (2011: 105, 2013b: 760).
131
are not grounded by any further truths (cf. Cameron 2018). For Sider (2011),

fundamental truths are those formulated in fundamental ('joint-carving') terms and

relate to non-fundamental truths via what he calls 'metaphysical semantics'.

I take there to be two core concepts associated with the notion of fundamentality

(whether ontological or alethic).79 The first is basicness. Intuitively, the fundamental

is in some sense the 'ground floor': it does not depend on or derive from anything else.

Basicness is perhaps the most natural way to think about fundamentality, underlying

the widespread idea that fundamental entities/truths are those that don't exist/hold in

virtue of any other entities/truths (cf. Bennett 2017: Ch.8). The second core concept

is integrality. Intuitively, fixing what is fundamental is 'all God would have to do' in

order to fix everything. There are two aspects of the concept of integrality. One is

completeness: the fundamental must in some sense be responsible for everything else

(cf. Schaffer 2010: 39, Sider 2011: 105). The other is non-redundancy: not only does

the fundamental account for everything else, but nothing less than the fundamental

suffices to do so. If basicness is the most natural way to characterise fundamentality,

integrality is perhaps the most philosophically interesting. It underlies one of the main

reasons for regarding fundamentality as metaphysically important: to give an account

of what is fundamental, qua integral, is to give a minimally complete account of

everything (cf. Sider 2011: 105, Tahko 2018: §1.3).

Another reason for interest in fundamentality comes from the previously

mentioned idea that the world is structured as a layered hierarchy, whereby (e.g.) the

economic depends on the psychological, the psychological on the biological, and so

on (see §2.1.3). It is natural to think of this layered hierarchy as corresponding to a

notion of relative fundamentality: biological phenomena are more fundamental than

79
See Bennett (2017: Ch.8) and Tahko (2018).
132
psychological phenomena but less fundamental than chemical phenomena, for

example. And perhaps the hierarchy terminates in a set of absolutely fundamental

phenomena, which do not depend on anything else but on which everything else

ultimately depends. I will assume, in what follows, that as well as elucidating the

notions of basicness and integrality, a theory of fundamentality should also make

sense of this layered picture.

4.1.1 Grounding, Truthmaking and Fundamentality

The intuitive appeal of the layered conception perhaps explains why several find it

congenial to characterise fundamentality in terms of some form of grounding

relation.80 Grounding is generally construed as a relation of non-causal dependence,

closely related to explanation. It thus seems well-suited to capture the sense in which

the less fundamental depends on the more fundamental. And, like grounding, relative

fundamentality seems to be a strict partial order. If chemistry is more fundamental

than biology, biology is not also more fundamental than chemistry; so more

fundamental than is plausibly asymmetric (hence irreflexive). If physics is more

fundamental than chemistry, and chemistry more fundamental than biology, then

physics is more fundamental than biology; so more fundamental than is plausibly

transitive. Prima facie, then, grounding has the right features to make sense of the

notion of relative fundamentality corresponding to the layered hierarchy.81 But as I

mentioned in §2.1.3, there are different ways to think about the hierarchical picture;

80
See e.g. Cameron (2018), Rosen (2010), Schaffer (2009). Note that I’m using ‘grounding’
broadly here and am not specifically referring to truth-grounding.
81
But see Rabin (2018).
133
in particular, there is a question as to whether to think of it as an alethic or an

ontological hierarchy.

Recall the inflationary conception of Stratified Ontology, on which reality divides

into layers of fundamental and non-fundamental entities (§3.1.2). Inflationists tend to

think of hierarchical fundamentality as an ontological hierarchy, structured by a

grounding relation whose relata are construed as elements of extra-representational

reality. They might appeal to entity-grounding to distinguish between fundamental

(ungrounded) and non-fundamental (grounded) entities, both construed as 'full-blown

"heavyweight"' denizens of reality (Schaffer 2009: 360). Or they might appeal to fact-

grounding to distinguish fundamental from non-fundamental facts, insisting that

'grounded facts and ungrounded facts are equally real, and grounded facts are an

"addition of being" over and above the facts in which they are grounded' (Audi 2012b:

101-102). The hierarchy, with economics and psychology towards the 'top' and

physics towards the 'bottom', is thus supported by an abundant ontology of more and

less fundamental 'levels of being'.

You may have guessed that I'm not a fan of this inflationary picture. One reason

for this has to do with parsimony. Ockham's Razor tells us to avoid an abundant

ontology if we can get by with a sparser one. And given that the fundamental is

supposed to somehow explain everything else, there is a concern that the inflationist's

abundant non-fundamental ontology is redundant. Some hope to avert this worry by

arguing that, although grounded and ungrounded entities are equally real, Ockham's

Razor only cares about the ungrounded (Schaffer 2015, Bennett 2017). But as I'll

argue in a later chapter, I don't think this works (see Ch.6).

One might instead think of the hierarchy as an alethic hierarchy. One might, for

instance, distinguish between fundamental and non-fundamental truths in terms of a

grounding relation between true propositions (cf. Cameron 2018). This could provide

134
for an abundant, structured hierarchy of fundamental and non-fundamental truths—

the less fundamental holding in virtue of the more fundamental— without the need

for a correspondingly abundant ontological hierarchy. Perhaps, for example, the

truths of psychology are derivative and dependent on the truths of biology, even if

reality hosts no sui generis psychological or biological entities. But the question

arises: how does this alethic hierarchy relate to extra-representational reality? As a

truthmaker theorist, I think the alethic needs some foundation in the ontological—

albeit, perhaps, in a sparser ontology then we get on the inflationary entity-grounding

picture. Moreover, I think this foundation must involve a cross-categorial relation of

non-causal dependence, connecting representational truth to extra-representational

reality (see Ch.2). A grounding relation that only obtains between true representations

doesn't seem to allow for this on its own.

Perhaps, instead of grounding, fundamentality could be characterised in terms of

truthmaking. An interesting possibility here is that, because truthmaking is cross-

categorial, a truthmaker theory of fundamentality could speak to both alethic and

ontological fundamentality. Fisher (2016: 449) outlines the basic idea as follows:

A truthmaker theory of fundamentality […] posits a relation of

truthmaking in order to specify which truths are fundamental, how
they are related to the world, and how they are related to
nonfundamental truths. On the truthmaking approach, entity-
fundamentality and truth-fundamentality are combined to give a
theory about the world's structure in terms of a set of fundamental
truthmakers and a fundamental description that is made up of a set
of fundamental truths.

Here's a rough picture of how this might go. We have a sparse set of fundamental

entities— say, microphysical entities—serving as the ultimate truthmakers for all the

truths. Fundamental truths might be defined in relation to these entities; perhaps they

are simply truths that proclaim the existence of the microphysical truthmakers. Non-
135
fundamental truths stand in some less immediate relation to the fundamental

truthmakers but are still made true by the same sparse microphysical ontology. So (as

per the truth-grounding approach) there might be psychological and biological truths

without the need to admit sui generis psychological or biological entities into our

ontology (as per the entity-grounding approach). But we also have a story to tell about

how these truths depend on ontology, in terms of their relation to the fundamental

truthmakers. As deRosset puts it, '[t]ruthmaking has the right features to explain the

rich panoply of truths while keeping the ontology […] sparse and flat' (2017: 543).

But as deRosset goes on to point out (2017: 544-545), truthmaking seems to have

the wrong features to underwrite the hierarchical picture. On the truth-grounding

approach, truths about the microphysical might provide the ultimate grounds for all

non-fundamental truths; but there will also be grounding relations among the non-

fundamental truths, the less fundamental holding in virtue of the more fundamental.

On the truthmaker approach, microphysical entities might be the ultimate truthmakers

for all truths. But since truthmaking is cross-categorial, there will be no truthmaking

relations between, for example, the truths of biology and the truths of psychology. So

while the notion of truthmaking may shed light on how fundamental and non-

fundamental truths relate to fundamental ontology, it 'lacks the right formal and

material features to capture the rich layered structure of theories' (deRosset 2017:

545).

Let’s take stock. A theory of fundamentality aims to say (i) what is fundamental

and (ii) how the fundamental relates to the non-fundamental. A theory of

fundamentality might primarily concern alethic or ontological fundamentality or, in

the case of truthmaker theories, elements of both. In either case, we would like a

theory of fundamentality that makes sense of the two core concepts of basicness and

integrality. We would also like one that makes sense of the compelling hierarchical

136
picture. Grounding-based approaches seem particularly well-suited to the latter. But

'worldly' conceptions of grounding, on which grounding is a relation between 'full-

blown, heavyweight' denizens of reality, give us too much ontology for my liking.

Representational conceptions, on the other hand, give us too little and do not speak

to the truthmaker theorist's guiding intuition that the alethic depends cross-

categorially on the ontological. Truthmaker approaches, meanwhile, have the

potential to give us the best of both worlds, but seem ill-equipped to make sense of

the layered conception. To quote deRosset again (2017: 545): 'We would like some

way to stitch together ground and truthmaking so that the result has the formal and

material features of ground, but relies only on a sparse and flat ontology'.

4.1.2 Hybrid Theories

Perhaps the way forward is to adopt some sort of hybrid truthmaker theory of

fundamentality, appealing to other notions alongside truthmaking. Cameron (2018)

and Fisher (2016) might be read as endorsing theories of this sort.

Cameron (2018) outlines a theory of fundamentality in the course of an attempt to

assimilate truthmaking into truth-grounding. Let a fundamental truth be a true

proposition that is not grounded by (does not hold in virtue of) any further truth. Let

a pure existence claim be a proposition whose sole content is that some x or Xs

exist(s). Cameron uses these notions to define truthmaking as follows (2018: 338):

[A] proposition is made true by some things, the Xs, if and only if it
is the fundamentally true pure existence claim that the Xs exist or it
is true in virtue of the fundamentally true pure existence claim that
the Xs exist.82

82
I have substituted 'fundamentally' for Cameron's 'brutely'.
137
Essentially, then, Cameron holds a grounding theory of alethic fundamentality. On

his view, fundamental and non-fundamental truths are connected by the (irreflexive,

asymmetric and transitive) grounding relation whose relata are true propositions; the

fundamental truths are those which are not grounded by further truths. Truthmaker

theory, as Cameron sees it, is just a special version of this grounding theory on which

the only fundamental truths are existential truths, which serve as the ultimate grounds

for all other truths. This is how Cameron understands 'the characteristic claim of

truthmaker theory: the claim that every truth is made true by some thing(s)' (2018:

338).

Cameron's account gives us an abundant, structured alethic hierarchy without the

need for a correspondingly abundant ontological hierarchy. It also sounds like

Cameron has the resources to relate this alethic hierarchy to a sparse ontology of

fundamental truthmakers. But because Cameron assimilates truthmaking into truth-

grounding, his theory jettisons the truthmaker theorist's usual contention that this is a

cross-categorial relation.83 Suppose that, out in reality, there is a certain electron e.

And suppose that ⟨e exists⟩ is a fundamental truth. It follows from Cameron's

definition of truthmaking that ⟨e exists⟩ is made true by e. But the relation between e

and ⟨e exists⟩ apparently can't be a cross-categorial truthmaking relation. The

grounding relation in terms of which Cameron defines truthmaking only obtains

between true propositions, and e is not a true proposition but an electron. So if there

is a cross-categorial relation between e and ⟨e exists⟩, truthmaking isn't it.

83
Cf. Asay (2017: 459-460). I am assuming here that Cameron’s reductive definition of
truthmaking is an attempt to reductively identify the truthmaking relation with certain
instances of the truth-grounding relation (see §2.2.1). In the context, this seems to be what he
intends. But perhaps Cameron has a weaker claim about the connection between truthmaking
and truth-grounding in mind, in which case the above concern about cross-categoriality may
not apply.
138
 Fisher (2016) offers a truthmaker theory of fundamentality that preserves the cross-

categoriality of truthmaking. His key move is to deny Truthmaker Maximalism. This

allows him to define the fundamental truths as those with truthmakers, which (given

non-maximalism) are a proper subset of all the truths (Fisher 2016: 460). To account

for the relation between fundamental and non-fundamental truths, Fisher appeals to

Sider's (2011) notion of metaphysical semantics. Briefly, the thought is that non-

fundamental truths relate to fundamental truths via conditionals of the form 'S is true

in L iff f', where L is a 'fundamental language' whose true sentences are just those

with truthmakers (Fisher 2016: 461-462). Thus, on Fisher's theory, non-fundamental

truths relate to fundamental truths via metaphysical semantics, and the fundamental

truths are all and only the truths with truthmakers.

I find all this rather ad hoc. Fisher's only discernible motivation for rejecting

Truthmaker Maximalism is that it lets him stipulate that the fundamental truths are

just the ones with truthmakers. (And his only discernible motivation for wheeling in

metaphysical semantics is that this stipulation prevents him from characterising non-

fundamental truths in terms of their relations to truthmakers.) But there is no obvious

independent reason to think that having a truthmaker is what makes a truth

fundamental. Consider an intuitively non-fundamental truth: ⟨The London Symphony

Orchestra exists⟩. What would Fisher say to someone who insisted that this truth has

a truthmaker, and so should count as fundamental by his lights? I can't think of much

he could say without relying on prior ideas about what makes a truth fundamental,

which is what the notion of having a truthmaker is supposed to illuminate. It would

be no good to say, for instance, that because ⟨The London Symphony Orchestra

exists⟩ is plausibly non-basic, or non-integral, we shouldn't take it to be a fundamental

truth, i.e. one with a truthmaker. This would be to appeal to prior ideas about what

makes a truth fundamental in order to decide which truths have truthmakers, and
139
hence which ones are (for Fisher) fundamental. (Cameron, in contrast, could say that

⟨The London Symphony Orchestra exists⟩ is plausibly non-fundamental because it is

plausibly grounded by other truths, and so isn't basic.)

4.1.3 Next Steps

In Chapter 2, I gave my view that truthmaking and (truth-)grounding are distinct

relations, though both are definable in terms of the primitive in virtue of. Over the

next two sections, I will offer a theory of fundamentality which incorporates both

truthmaking and grounding, construed as distinct relations— I'll call it (TGF) ('T' for

truthmaking, 'G' for grounding, 'F' for fundamentality).

In outline, my approach will be as follows. I will start by using the grounding

relation to define the core properties of fundamental truth. I will then discuss the

connection between fundamental truths and truthmakers. It will emerge that whatever

(minimally) suffices to make true the fundamental truths (minimally) suffices to make

true all the truths. These are the fundamental truthmakers. On the resulting picture,

the fundamental truthmakers provide a sparse ontological foundation for an abundant,

structured hierarchy of fundamental and non-fundamental truths— an alethic

rainforest in an ontological desert. (TGF) offers what I think is an independently

attractive way to think about fundamentality, and of how reality and our

representations of it are structured. But I'll go on to claim that (TGF) also offers

Truthmaker Ontologists a more refined conception of what makes a truthmaking

contribution, and therefore has ontological status.

4.2 Grounding and Fundamental Truth

The two core concepts associated with fundamentality are basicness and integrality.

In this section, I will use the grounding relation to define basicness and integrality

140
with respect to truth. I will then argue that any basic truth is an integral truth and vice

versa. The set of fundamental truths, as I will understand them, is the set of basic and

integral truths.

4.2.1 Basic Truths

Intuitively, the basic truths are those at the 'ground floor' of the alethic hierarchy; they

are explanatorily basic and do not derive from or depend on any further truths. I will

define basic truth, in a familiar way, in terms of ungroundedness:

(Basic Truth) P is a basic truth iffdf. P is true and ¬$Q(Q Þ

P).

That is, P is a basic truth just in case there are no other truths in virtue of which P is

true. As far as the non-causal explanatory dependence of truth on truth goes, P is at

the ground floor.

Are there any basic truths? I'm going to assume so. Moreover, I'm going to assume

that every truth is either basic or is grounded in some basic truth(s), as per

Foundations:84

(Foundations) "P(P is a basic truth ∨ $Q(Q is a basic truth &

Q Þ P)).85

I find Foundations intuitively compelling. As Fine says (2010: 105), 'given a truth

that stands in need of explanation, one naturally supposes that it should have a

84
Cameron (2018: 338) endorses a similar principle under the name 'Universal Grounding'.
The quantifiers in Foundations range over true propositions.
85
For simplicity, I’m ignoring the fact that P might be jointly grounded by several basic
truths— I take it that (Foundations) could be amended to allow for this.
141
"completely satisfactory" explanation, one that does not involve cycles and terminates

in truths that do not stand in need of explanation'. But there are those who endorse

infinitism, on which some grounding chains never terminate in basic truths, and

coherentism, on which some grounding chains involve cycles.86 I lack the space to

rebut these views in detail. Briefly, however, I think the best reason to assume

Foundations is that it is theoretically beneficial to do so— here I follow Cameron

(2008d). 'If we seek to explain some phenomena', writes Cameron (2008d: 12), 'then,

other things being equal, it is better to give the same explanation of each phenomenon

than to give separate explanations of each phenomenon'. But if some grounding

chains never terminate or involve cycles, 'there is no explanation of everything that

needs explaining. That is, it is true for every [non-basic truth P] that [the truth of P]

is explained by [some prior truth], but there is no collection of [truths] that explains

the [truth] of every [non-basic P]' (ibid.).87 Assuming Foundations, however, all truths

that require grounding explanations are ultimately accounted for in terms of a set of

truths that require no grounding-explanations; we can thus give a common, unified

explanation of every non-basic truth. This is a theoretical benefit, giving us some

motivation for assuming Foundations instead of the alternatives.88

86
For recent discussion, see the relevant papers in Bliss and Priest (2018).
87
Cameron’s focus is on ontological fundamentality, but I think the point carries over to
alethic fundamentality.
88
As Cameron points out, this line of argument can’t establish that Foundations is necessarily
true: its status is rather ‘like that enjoyed by Ockham’s razor: we should accept it because if
it is true the theories we arrive at give a better explanation of the phenomena to be explained,
and hence are more likely to be true. But such principles of theory-choice do not appear
necessary; it is not as if the world is necessarily such that the simplest explanation is the right
one—we just hope that our world is like this’ (2008d: 13). In keeping with this, I won’t assume
that Foundations holds necessarily; only that we should think it holds at our world.
142
4.2.2 Integral Truths

Intuitively, an integral truth is a member of an elite collection of truths that forms a

complete and non-redundant grounding base for all other truths. To make this more

precise, I will define the notion of minimal completeness with respect to grounding,

or 'minimal G-completeness':

(MCG) (i) A set of true propositions T is minimally G-

complete with respect to a set of true propositions T*

iffdf. T contains a full ground for every member of T*

and no proper subset of T contains a full ground for

every member of T*.

(ii) A set of true propositions T is minimally G-

complete simpliciter, iffdf. T is minimally G-complete

with respect to all other truths.

(T contains a full ground for P just in case one of T's members fully grounds P or

several of T's members jointly comprise a full ground for P. 'Minimally G-complete'

shall mean 'minimally G-complete simpliciter', unless otherwise specified.) I now

define 'integral truth' in terms of minimal G-completeness as follows:

(Integral Truth) P is an integral truth iffdf. P Î T, where T is the

unique minimally G-complete set of truths.89

By (MCG), if T is minimally G-complete, T contains a full ground for every other

truth. The integral truths thus provide a complete grounding base for the truths about

89
More on uniqueness below.
143
the world. And, if P is a member of T, then there is some true Q such that (i) T contains

a full ground for Q but (ii) the proper subset of T consisting of all its members apart

from P does not contain a full ground for Q. The integral truths thus provide a non-

redundant grounding base for all the truths about the world. (In future, I will write

'T\P' for 'the proper subset of T consisting of all its members apart from P'. Note also

that P itself need not be a full ground for any Q; it might only be part of a full ground

for some Q that involves other members of T.)

4.2.3 The Parity Thesis

The relation between the basic and integral truths is captured by the Parity Thesis:

(Parity Thesis) "P(P is a basic truth iff P is an integral truth).

Let’s start by assuming that there is exactly one minimally G-complete set of truths,

T. We can show that, given this assumption, any basic truth is an integral truth and

vice versa.90

If P is a basic truth, then P is an integral truth. Assume, for reductio, that some

true P is basic but isn’t integral. Since T is minimally G-complete, it follows from

(MCG) that either P Î T or T contains a full ground for P. Since P is not integral, P

Ï T. Hence, T contains a full ground for P. So, contrary to our assumption, P is not

basic. Any basic truth is therefore an integral truth.

90
The following argument is inspired by Bennett (2017: §5.4) (who characterises
fundamentality in terms of what she calls ‘building’). A minor difference is that Bennett is
concerned with ontological fundamentality, not alethic fundamentality. Another is that the
claim she defends is conditional: if building is transitive, then the ‘unbuilt’ entities coincide
with the entities that completely and non-redundantly ‘build’ all others. I hold that grounding
is transitive, and that the ungrounded truths therefore coincide with the truths that completely
and non-redundantly ground all others.
144
 If P is an integral truth, then P is a basic truth. Assume, for reductio, that some

true Q is integral but isn’t basic. Since T is minimally G-complete, every grounded

truth is fully grounded by some member(s) of T. Since Q is non-basic, it’s grounded.

Thus, T contains a full ground for Q; call it P. Since Q is integral, Q Î T. By (MCG),

there is some true R such that T contains a full ground for R, but T\Q doesn’t contain

a full ground for R. There are two possibilities here.

First, it might be that Q itself fully grounds R. Then, (by TransitivityG), since P

grounds Q and Q grounds R, P grounds R. Therefore, T\Q does contain a full ground

for R, namely P. So, contrary to our assumption, Q is not integral.

Second, it might be that Q is a part of a full ground for R that involves other

members of T. Let Q* be another member of T such that Q and Q* jointly comprise

a full ground for R (though neither does so individually). Thus, we have:

(i) PÞ Q

(ii) Q, Q* Þ R.

I take it to follow that:

(iii) P, Q* Þ R.

The general principle I’m appealing to is this: if (i) P is a ground for Q and (ii) Q, Γ

is a ground for R, then (iii) P, Γ is also a ground for R.91 I won’t argue for this in

detail. But for a plausible instance, assume the following:

(i*) ⟨Simples arranged Socrates-wise exist⟩ Þ ⟨Socrates exists⟩

(ii*) ⟨Socrates exists⟩, ⟨Plato exists⟩ Þ ⟨Socrates and Plato exist⟩.

91
‘Γ’ here is a collection of one or more truths.
145
Given (i*) and (ii*), it plausibly follows that:

(iii*) ⟨Simples arranged Socrates-wise exist⟩, ⟨Plato exists⟩ Þ

⟨Socrates and Plato exist⟩.

Resuming the argument, we again have that T\Q contains a full ground for R, namely

{P, Q*}. So again, contrary to our assumption, Q is not integral. Any integral truth

must therefore be a basic truth.

I’ve shown that the Parity Thesis is true, assuming that there is exactly one

minimally G-complete set of truths. But is there? An argument adapted from Bennett

(2017: 117-118) shows that the answer is yes, as long as Foundations is assumed:

(Foundations) "P(P is a basic truth ∨ $Q(Q is a basic truth &

Q Þ P)).

Given Foundations, there is a non-empty set— call it B— containing all and only the

basic truths. Moreover, any true P is either (i) a member of B or (ii) is grounded by

some member(s) of B. B therefore contains a full ground for every other truth. Does

any proper subset of B contain a full ground for every other truth? No. For let Q be

an arbitrary member of B. B\Q doesn’t contain a full ground for every other truth

because it doesn’t contain a full ground for Q, which is ungrounded. So B is minimally

G-complete. Is B the only minimally G-complete set of truths? Yes. For the argument

above shows that, if there’s a minimally G-complete set of truths, no non-basic truth

can belong to that set. So, since B contains all and only the basic truths, no non-

member of B can belong to any minimally G-complete set. And by definition of

‘minimal G-completeness’, no proper subset of B is minimally G-complete. Thus,

given Foundations, there is exactly one set of integral truths: the set of all and only

146
the basic truths.

I take basicness and integrality to be the two core properties of fundamental truths,

and I have shown how to construe these properties in terms of grounding. I have also

argued that basicness and integrality coincide, given Foundations (which I assume).

Hence, I endorse the following:

(Fundamental Truth) (i) P is a fundamental truth iff P is a

basic truth iff P is an integral truth.

(ii) P is a non-fundamental truth iff

P is true and not fundamental.

We now have the necessary components of a theory of alethic fundamentality:

fundamental and non-fundamental truths are connected by the grounding relation, and

the fundamental truths are those which are basic (i.e. ungrounded) and integral (i.e.

minimally complete with respect to grounding). As a non-causal explanatory

dependence relation, grounding seems well-suited for capturing the sense in which

less fundamental truths derive from more fundamental truths. And as a strict partial

order, grounding has the right features to capture the hierarchical conception of

relative alethic fundamentality.92

4.3 Fundamental Truthmakers

The truths about the world are hierarchically structured by the grounding relation,

with a collection of basic and integral truths at the ground floor. This alethic hierarchy

92
I have not attempted to explicitly define relative fundamentality, however. See Bennett
(2017: Ch.6) for a useful discussion of some of the issues involved in this.
147
now needs an ontological foundation. I will argue that the only foundation it needs is

a (potentially sparse) set of fundamental truthmakers.

4.3.1 Definitions

I define minimal completeness with respect to truthmaking, or 'minimal TM-

completeness', as follows:

(MCTM) (i) A set of entities E is minimally TM-complete

with respect to a set of true propositions T iffdf. E

contains a full truthmaker for every member of T

and no proper subset of E contains a full truthmaker

for every member of T.

(ii) A set of entities E is minimally TM-complete

simpliciter iffdf. E is minimally TM-complete with

respect to the set of all true propositions.

(E contains a full truthmaker for P just in case one of E's members fully makes P

true or several of E's members jointly comprise a full truthmaker for P.) If E is

minimally TM-complete with respect to a set of truths T, then E contains a full

truthmaker for every truth in T. E thus provides a complete truthmaking base for the

truths in T. And for any x Î E, there is some P Î T such that (i) E contains a full

truthmaker for P, but (ii) the proper subset of E containing all of E's members apart

from a (written 'E\a') does not contain a full truthmaker for P. E thus provides a non-

redundant truthmaking base for the truths in T.

Note that a might be a member of E without itself being a full truthmaker for any

truth in T. It might be a part of a full truthmaker for some truth in T, which involves

148
other members of E. Note also that E's being minimally TM-complete with respect to

T does not mean that E contains a minimal truthmaker for every truth in T (where x

is a minimal truthmaker for P just in case x makes P true and no proper part of x does

so; see Armstrong 2004: 19-20). Let T be a set of truths concerning the existence of

electrons and suppose that ⟨Infinitely many electrons exist⟩ is a member of T. Let E

be a set of entities containing an infinite number of electrons, which together

comprise a full truthmaker for ⟨Infinitely many electrons exist⟩. Subtract one of these

electrons (or every second electron, every third electron, …), and E still contains a

full truthmaker for ⟨Infinitely many electrons exist⟩. So E doesn't contain a minimal

truthmaker for this truth. But E might still be minimally TM-complete with respect

to T as long as subtracting the electron(s) leaves some truth in T without a full

truthmaker in E. For example, perhaps subtracting electron e leaves a full truthmaker

for ⟨Infinitely many electrons exist⟩, but not for ⟨e exists⟩. E's job, as it were, is to

contain the minimum number of entities needed to make true all the truths in T, but

not necessarily each individual truth in T.

I now define fundamental truthmaker as follows:

(Fundamental Truthmaker) a is a fundamental truthmaker iffdf. a Î

E, where E is the unique set of entities

that is minimally TM-complete with

respect to the set of fundamental truths.93

93
The term is perhaps a bit misleading: for reasons mentioned above, an entity could in
principle be a ‘fundamental truthmaker’ by virtue of being a partial truthmaker.
149
That is, the fundamental truthmakers jointly comprise a complete and non-redundant

truthmaking base for the fundamental truths, i.e. the unique set containing all and only

the basic and integral truths.

What justifies the assumption that there is a set of fundamental truthmakers? As

with Foundations, I think the most compelling rationale is methodological. I will

shortly argue that, if there is a set of fundamental truthmakers, they are not only

minimally TM-complete with respect to the fundamental truths but with respect to all

truths. Consequently, an inventory of fundamental truthmakers would amount to an

optimally comprehensive, parsimonious and unified account of what ontology is

needed to account for the truths about the world: comprehensive because the

fundamental truthmakers would account for all the truths; parsimonious because the

fundamental truthmakers would non-redundantly account for all the truths; and

unified because the fundamental truthmakers would provide the sparsest possible

ontological foundation for the broadest possible array of truths. There is thus a pro

tanto case for the assumption that there is a set of fundamental truthmakers,

resembling the case for Foundations outlined above: in the words of Cameron (2008d:

13), ‘we should accept it because if it is true the theories we arrive at give a better

explanation of the phenomena to be explained, and hence are more likely to be true’.

I will now argue for three claims about fundamental truthmakers, and their relation

to fundamental and non-fundamental truths.

4.3.2 Thrift

The first and perhaps most significant claim is what I'll call Thrift:

(Thrift) The fundamental truthmakers are minimally TM-

complete simpliciter.

150
In other words, given a set of entities that’s minimally TM-complete with respect to

the fundamental truths, that set is minimally TM-complete with respect to all truths.

The argument for Thrift assumes the Grounding Principle (§2.4.3):

(Grounding Principle) If x ⊨ P and P Þ Q then x ⊨ Q.

Let T be the set of fundamental truths, let E be minimally TM-complete with respect

to T, and let Q be an arbitrary true proposition. Since T is minimally G-complete,

either Q Î T or T contains a full ground for Q. Suppose first that Q Î T. Then, since

E is minimally TM-complete with respect to T, E contains a full truthmaker for Q.

Suppose next that Q ∉ T and so T contains a full ground for Q; call it P. Since P Î T,

E contains a full truthmaker for P; call it a. So a ⊨ P and P Þ Q. By the Grounding

Principle, a ⊨ P. So whether or not Q Î T, E contains a full truthmaker for Q. Since

Q was arbitrary, we can infer that E contains a full truthmaker for every truth.

Moreover, no proper subset of E contains a full truthmaker for every truth since, ex

hypothesi, no proper subset of E contains a full truthmaker for every truth in T.

Therefore, if E is minimally TM-complete with respect to the set of fundamental

truths, E is minimally TM-complete simpliciter.

Given Thrift, the fundamental truthmakers provide a complete and non-redundant

truthmaking base for all truths. There is thus a natural sense in which the fundamental

truthmakers are integral: they are 'all God would have to create' in order to make the

true overall theory of the world true. Consequently, Thrift shows us how an abundant

structured hierarchy of fundamental and non-fundamental truths can be supported by

a sparse ontological foundation.

151
4.3.3 Matching

The second claim, Matching, concerns the status of truths about the existence of

fundamental truthmakers:

(Matching) If x is a fundamental truthmaker, then ⟨x

exists⟩ is a fundamental truth.

The argument for Matching assumes the Grounding Principle, along with three other

principles from Chapter 2:

(Interaction) x ⊨ P iff ⟨x exists⟩ Þ ⟨P is true⟩

(Quasi-Asymmetry) "x"P((x ⊨ P) ® ¬(P Þ ⟨x exists⟩))

(Truth-ElimTM) If x ⊨ ⟨P is true⟩ then x ⊨ P.

I’ll proceed in two stages. First, we need the following Lemma:

(Lemma) If x is a fundamental truthmaker, then x

makes true ⟨x exists⟩ and no

fundamental truthmaker other than x

makes true ⟨x exists⟩.

Let E be the set of fundamental truthmakers and suppose that a Î E. Since a exists,

⟨a exists⟩ is true. Since E is minimally TM-complete (simpliciter), E contains a full

truthmaker for ⟨a exists⟩; suppose it's x. Thus:

(1) x ⊨ ⟨a exists⟩.

152
Since a Î E, there is some true P such that E contains a full truthmaker for P but E\a

does not contain a full truthmaker for P. For simplicity, we'll assume that a itself is a

full truthmaker for P (as opposed to a part of a full truthmaker for P). Thus:

(2) a ⊨ P.

It follows via Interaction that:

(3) ⟨a exists⟩ Þ P is true.

From (1), (3) and the Grounding Principle, it follows that:

(4) x ⊨ ⟨P is true⟩.

Hence, by Truth-ElimTM:

(5) x ⊨ P.

Now, x and a are both members of E. And either x = a or x ≠ a. If x ≠ a, then E\a

contains a full truthmaker for P, namely x. But as we've seen, our assumptions entail

that E\a does not contain a full truthmaker for P. Therefore, x = a. So by substitution

into (1):

(6) a ⊨ ⟨a exists⟩.

Thus, if a is a fundamental truthmaker, then any fundamental truthmaker that makes

true ⟨a exists⟩ is identical to a.

Having established Lemma, the rest of the argument for Matching is as follows.

Again, let E be minimally TM-complete, and let F be the set of fundamental truths

(i.e., P Î F iff P is a basic truth iff P is an integral truth). And assume, for reductio,

153
that there is some a such that a is a fundamental truthmaker but ⟨a exists⟩ is a non-

fundamental truth (i.e., a Î E but ⟨a exists⟩ is true but not in F). Since F is minimally

G-complete, F contains a full ground for ⟨a exists⟩; call it P. Since E is minimally

TM-complete, E contains a full truthmaker for P; call it x. Thus:

(7) x ⊨ P and P Þ ⟨a exists⟩.

By the Grounding Principle:

(8) x ⊨ ⟨a exists⟩.

Since a Î E, it follows from Lemma that:

(9) "x((x Î E & x ⊨ ⟨a exists⟩) ® x = a).

So, from (8) and (9):

(10) x = a.

And from (7) and (10):

(11) a ⊨ P and P Þ ⟨a exists⟩.

But by Quasi-Asymmetry:

(12) ¬(a ⊨ P and P Þ ⟨a exists⟩)

154
Contradiction. So our assumption that a is a fundamental truthmaker but ⟨a exists⟩ is

not a fundamental truth is false. Therefore, if a is a fundamental truthmaker, ⟨a exists⟩

is a fundamental truth.

Given Thrift, there’s a natural sense in which the existence of fundamental

truthmakers is integral. Given Matching, there’s also a natural sense in which the

existence of fundamental truthmakers is basic: if x is a fundamental truthmaker, then

the truth that x exists does not hold in virtue of any further truths.

4.3.4 FT-Truth

The third claim, FT-Truth, concerns the nature of the truthmaking relations between

fundamental truthmakers, and fundamental and non-fundamental truths respectively.

First, recall the distinction between direct and indirect truthmaking (§2.3):

(Direct Truthmaking) x directly makes P true iffdf. x ⊨ P &

¬$Q(x ⊨ Q & Q Þ P).

(Indirect Truthmaking) x indirectly makes P true iffdf. x ⊨ P &

$Q(x ⊨ Q & Q Þ P).

FT-Truth can be stated as follows:

(FT-Truth) (i) P is a fundamental truth iff P is

directly made true by some fundamental

truthmaker(s) and P is not indirectly

made true by any fundamental

truthmaker(s).

155
 (ii) P is a non-fundamental truth iff

P is indirectly made true by some

fundamental truthmaker(s).

Start with the second clause. Let E be the set of fundamental truthmakers. Since E is

minimally TM-complete (simpliciter), every true P has a full truthmaker among the

members of E. Since every instance of truthmaking is either direct or indirect, every

true P is made true directly or indirectly by some member(s) of E. If P is a non-

fundamental truth, then P is grounded by some fundamental truth(s). In that case, P

is indirectly made true by whatever members of E make true its fundamental grounds.

And if P is indirectly made true by some members of E then P is grounded, since only

grounded truths can be indirectly made true, and is therefore non-fundamental.

Hence, P is a non-fundamental truth if and only if P is indirectly made true by some

fundamental truthmaker(s).

Now for the first clause. Again, let E be the set of fundamental truthmakers. Thus,

every true P is made true directly or indirectly by some member(s) of E. If P is a

fundamental truth, then P is ungrounded. Since ungrounded truths can't be indirectly

made true, P is directly made true by some member(s) of E. However, if P is directly

made true by some member(s) of E, it doesn't immediately follow that P is a

fundamental truth. One truth can have several truthmakers and, in principle, P might

be made true both directly by some member(s) of E and indirectly by others. But as

we've seen, P is made true indirectly by some member(s) of E if and only if P is a

non-fundamental truth. So if every instance of truthmaking between P and the

members of E is direct, then P is a fundamental truth. Thus, P is a fundamental truth

if and only if P is directly made true by some fundamental truthmaker(s) and P is not

indirectly made true by any fundamental truthmaker(s).

156
4.4 A Theory of Fundamentality

I will now outline and motivate the theory of fundamentality based on the foregoing

results: (TGF).

4.4.1 (TGF)

A theory of fundamentality aims to say (i) what is fundamental and (ii) how the

fundamental relates to the non-fundamental. 'Fundamental' might be treated either as

a predicate of non-representational 'worldly' entities (ontological fundamentality) or

of true representations (alethic fundamentality). In either case, there are two core

ideas a theory of fundamentality should aim to capture: the idea that the fundamental

entities/truths are in some sense basic and the idea that they are in some sense

integral. A further desideratum of a theory of fundamentality is that it should make

sense of the idea that the 'facts' about the world fall into a hierarchical structure of

explanatory dependence, wherein the less fundamental derives from and depends on

the more fundamental.

(TGF) brings alethic and ontological fundamentality together to yield a picture of

how the true representations of the world are structured and how they relate to extra-

representational reality. Fundamental and non-fundamental truths are related to each

other by the grounding relation. The fundamental truths are basic in that they are

ungrounded, and integral in that they form a complete and non-redundant grounding

base for all other truths. The fundamental entities are the fundamental truthmakers.

The fundamental truthmakers are integral in that they provide a complete and non-

redundant truthmaking base for all the truths. They are basic in the sense that truths

proclaiming the existence of fundamental truthmakers are fundamental, hence

ungrounded, truths. As well as the grounding relations they bear to each other,

157
fundamental and non-fundamental truths can also be characterised in terms of their

relations to the fundamental truthmakers: non-fundamental truths are made true

indirectly by the fundamental truthmakers, while fundamental truths are (only) made

true directly by the fundamental truthmakers.

(TGF) thus makes sense of the idea that the fundamental entities and truths are

integral and basic. It also offers a prima facie attractive way to think about the

hierarchical picture. Since grounding is a relation of non-causal explanatory

dependence, it is well-suited for capturing the sense in which less fundamental truths

depend on and derive from more fundamental truths. Since grounding is asymmetric,

irreflexive and transitive, it is well-suited for capturing the structure of the alethic

hierarchy, which is plausibly a strict partial order. In this respect, (TGF) contrasts

favourably with the 'pure' truthmaking approach, which gives no account of how more

and less fundamental truths relate to each other. Unlike the inflationary entity-

grounding approach, there is no need to think that the alethic hierarchy corresponds

to an abundant ontological hierarchy: on (TGF), the true representations of the world

are stratified, but the ontology is sparse and flat. Unlike the 'pure' truth- grounding

approach, (TGF) tells us not just how fundamental and non-fundamental truths relate

to each other but also how they relate, via the cross-categorial truthmaking relation,

to extra-representational reality.

Unlike Cameron's (2018) truthmaker theory of fundamentality, (TGF) does not

involve any attempt to assimilate truthmaking into truth-grounding, allowing us to

retain the cross-categoriality of truthmaking. Unlike Fisher's (2016) theory, (TGF)

characterises fundamentality in a non-ad hoc and independently plausible way. For

example, why is ⟨The London Symphony Orchestra exists⟩ plausibly a non-

fundamental truth? The only answer available on Fisher's account is: 'Because it

doesn't have a truthmaker'. On (TGF), in contrast, we could explain why this truth is

158
non-fundamental by pointing out that it is plausibly non-integral and non-basic. Nor

does (TGF) compel us to abandon Truthmaker Maximalism merely for the sake of

having a coherent truthmaker theory of fundamentality. Indeed, (TGF) makes

Truthmaker Maximalism easier to satisfy; for given Thrift, we can provide

truthmakers for all the truths by providing truthmakers for all the fundamental truths.

(TGF) involves two distinct notions— grounding and truthmaking— where other

theories of fundamentality only involve one. One might worry that this is

ideologically unparsimonious. In fact, this is the concern that prompts Cameron

(2018: 335) to identify truthmaking with truth-grounding, rather than treating them

as distinct: ‘It would be uneconomical if we had to have two primitive notions here,

so we should try and define one of makes true and [grounds] in terms of the other’. I

would first point out, however, that I don’t regard truthmaking and grounding as

distinct primitive notions. Rather, I think they’re both definable in terms of a single

primitive: in virtue of (see §2.1). So it’s not immediately clear that (TGF) is less

ideologically parsimonious than a theory of fundamentality where either truthmaking

or grounding is treated as the sole primitive. But even if I were committed to treating

truthmaking and grounding as distinct primitives, it doesn’t immediately follow that

(TGF) would be ideologically uneconomical. Ideological economy is not just a

measure of the number of primitive concepts a theory invokes, but of how useful

those concepts are. And as I’ve said, there are several benefits to building distinct

truthmaking and grounding relations into our theory of fundamentality.

4.4.2 Sider’s Objection

Sider (2011) objects that truthmaker theories of fundamentality result in an

explanatorily impoverished conception of fundamental truth. According to

truthmaker theories, as Sider construes them, the only fundamental truths are bare

159
existence claims of the form 'x exists', where x is a truthmaker (cf. Cameron 2018).

But Sider argues that a list of bare existence claims can't do what we want

fundamental truths to do, namely, to provide satisfactory ultimate explanations for

everything else. Ultimate explanations that always terminate in a list of entities can't

do this, Sider thinks, since they 'cannot be systematised with detailed general laws,

patterns or mechanisms' (2011: 160):

Think of the point this way. Suppose God hands you a collection of
entities: Alexander, Buffy, Cordelia, Dawn, …, and asks you to
work out the rest. Are there helium molecules? Are there cities?
Why did Buffy the Vampire Slayer end after season seven? You
wouldn't have any idea how to respond.

Suppose Sider is right: truthmaker theories of fundamentality imply that the only

fundamental truths are bare existence claims about truthmakers. It's not clear to me

that the fundamental truths, so construed, would be explanatorily inadequate.

Explanation is intensional, and the extent to which we find a claim explanatory

depends on how the explanantia are described. Consider the following causal

explanatory claim:

The most reported news event of May 1915 caused the US to enter the

First World War.

Whether you hear this as an adequate explanation depends on whether you know that

'the most reported news event of May 1915' refers to the sinking of the Lusitania.94

Something similar applies to truthmaker explanations.95 Consider the following:

⟨Helium atom h exists⟩ is true because Buffy exists.

94
Cf. Davidson (2001: Chs. 7&11) on the distinction between causation and explanation.
95
The following point is due to Cameron (2015: 122-123).
160
This will sound unexplanatory if we don't know what 'Buffy' refers to. It will seem

more explanatory, however, if we know that 'Buffy' refers to (say) the fact that

electrons e1 and e2 are orbiting nucleus n. It's not obvious, then, that fundamental

explanations terminating in the citation of entities would be explanatorily

unsatisfactory.

Sider thinks this is cheating (2011: 157-159). The truthmaker theory of

fundamentality he targets is supposed to give us fundamental truths involving only

the notion of existence. But in the example above, the truthmaker theorist has used a

'canonical name' involving predication ('orbiting') to describe the truthmaker for

⟨Helium atom h exists⟩ (Sider 2011: 157). For Sider, this amounts to 'smuggling in'

non-existential ideology into the fundamental truths. But this rejoinder rests on

Sider's own theory of fundamentality, on which 'fundamental' is primarily a predicate

of the concepts picked out by sub-sentential terms (cf. Cameron 2015: 122-123). By

Sider's lights, there are fundamental predicational truths if there are fundamental

truths formulated using predicates. This is not something the truthmaker theorist need

accept.

But in any case, (TGF) does not commit us to taking all fundamental truths to be

bare existence claims about truthmakers. Perhaps ⟨a is F⟩ is a fundamental truth, for

example. Given FT-Truth, this would mean that ⟨a is F⟩ is made true directly by some

fundamental truthmaker(s). Suppose the truthmaker is a trope: a's F-ness. Given

Matching, ⟨a's F-ness exists⟩ must also be a fundamental truth, made true directly by

some fundamental truthmaker(s). Since existence claims are directly made true (if at

all) by what they claim to exist (see §3.3), ⟨a's F-ness exists⟩ is also made true by a's

F-ness. But this doesn't imply that ⟨a is F⟩ can't be a fundamental truth; it rather

implies that ⟨a is F⟩ and ⟨a's F-ness exists⟩ are both fundamental truths, made true

161
directly by a's F-ness. So if there is a worry about taking the only fundamental truths

to be bare existence claims, I am happy to allow that there might be non-existential

fundamental truths— as long as they can be construed to have direct fundamental

truthmakers, and as long as the truths proclaiming the existence of these truthmakers

can be construed as fundamental.

4.4.3 Refining the Truthmaker Criterion of Ontological Status

In the last chapter, I argued that Truthmaker Ontology supports a coherent distinction

between what merely and what really exists. Since all and only the things that make

a truthmaking contribution have ontological status, and since Fs need not be among

the truthmakers for ⟨Fs exist⟩, it is possible for an F-less ontology to support the truth

of ⟨Fs exist⟩. What can truly be said to exist thus comes apart from what has

ontological status, and we can articulate this by distinguishing what really exists (i.e.

what makes a truthmaking contribution) from what merely exists (i.e. can truly be

said to exist but makes no truthmaking contribution). Truthmaker Ontology, thus

construed, is a deflationary form of Stratified Ontology (see §3.1.2).

Some Stratified Ontologists prefer to distinguish between the fundamental and the

non-fundamental, instead of what really and what merely exists. This is particular

germane on the inflationary approach, which takes everything that exists to have

heavyweight ontological status: the inflationist wants to say that reality divides into

fundamental and non-fundamental entities, but not that non-fundamental entities are

unreal. But deflationists also go in for fundamentality-talk, sometimes treating

'fundamentally exists' and 'really exists' as synonyms (e.g. Cameron 2008a, Williams

2010). Here is Cameron, for example, with a deflationary take on the

fundamental/non-fundamental distinction (2008c: 303):

162
 When I say that statues are not fundamental I do not mean to say that
they have being but are kind of less important than the things that
both have being and are fundamental. I mean to deny that statues are
elements of our ontology. […] When I say that it is true that statues
exist but that statues do not fundamentally exist, I mean that there
are no such things as statues, but nevertheless the English sentence
'statues exist' is true.

I, however, don't regard 'really exists' and 'fundamentally exists' as synonyms. The

role of 'really exists' is to let us express the distinction between what has ontological

status and what can truly be said to exist: if ⟨x exists⟩ is made true but x makes no

truthmaking contribution, we can express this by saying that x exists but doesn't really

exist. But as I see it, the role of fundamentality is to refine our understanding of what

makes a truthmaking contribution, and hence what really exists. Recall the truthmaker

criterion of ontological status:

(TOS) x is real iff x makes a truthmaking contribution.

(TOS) tells us that a complete and correct answer to the ontological question would

take the form of a list of all and only the things that make a truthmaking contribution.

I now want to suggest a way to refine that claim: what makes a truthmaking

contribution are just the fundamental truthmakers, i.e., the things that comprise a

complete and non-redundant truthmaking base for the fundamental truths. So what is

real, in the sense of ontological interest, are the fundamental truthmakers.

The case for this claim is intuitively compelling. Given Thrift, the fundamental

truthmakers provide a complete and non-redundant truthmaking base for all the

truths. Thus, if a is a fundamental truthmaker, then a is indispensable to a complete

and correct account of what reality has to be like to account for the overall true theory

of the world; no account of what reality determines what is true could be complete

and correct unless it counted a among the real entities. On the other hand, if ⟨a exists⟩
163
is true but a is not a fundamental truthmaker, then an account of how reality

determines what is true could be complete and correct without counting a among the

real entities. Since all the requisite truthmaking work can be done without help of a,

there is no need to take a to make a truthmaking contribution. There is thus an

intuitive sense in which a should not be counted as real.

Furthermore, there is a methodological case for taking fundamental truthmakers to

be the target of ontological inquiry. As I said above, a complete inventory of

fundamental truthmakers would amount to a maximally comprehensive,

parsimonious and unified account of what ontology must be like in order to account

for the truths about the world. In at least these respects, an optimally virtuous account

of what is real coincides with what is real according to the revised (TOS).

Conclusion

(TGF) offers an attractive picture of the alethic and ontological structure of the world:

we have an abundant, layered hierarchy of fundamental and non-fundamental truths,

with a (perhaps sparse) ontological foundation consisting of fundamental

truthmakers. But beyond its independent merits, (TGF) provides Truthmaker

Ontologists with a more refined view of what it takes to discover what really exists.

The fundamental truthmakers are all we have to take to make a truthmaking

contribution— they are 'all God would have to create' to make the true description of

the world true. Thus, to decide on a set of fundamental truthmakers is to answer the

ontological question.

164
5 Truthmaker Commitments and
Rational Norms

Over the last two chapters, I have articulated and refined a heterodox view of the

scope of ontology; that is, a view of what counts as having ontological status and is

therefore the focus of ontological inquiry. The next two chapters focus on issues

concerning ontological method. Specifically, I aim to clarify how Truthmaker

Ontologists should understand the methodologically important notions of ontological

commitment and ontological parsimony. I begin, in this chapter, with ontological

commitment.

The ontological commitments of a theory are the requirements that the theory's

truth imposes on ontology— roughly, the things that must have ontological status if

the theory is true. Since I take all and only the truthmakers to have ontological status,

I accordingly think of ontological commitments as requirements on truthmakers.96 I

therefore hold what some call a truthmaker view of ontological commitment.97 As

we'll see, however, not all truthmaker views of ontological commitment are equal.

For even if we grant that ontological commitments are requirements on truthmakers,

difficult questions remain as to how to construe the relevant notion of requirement. I

shall therefore advance a novel view I call the Normative Truthmaker View of

ontological commitment. Briefly, the Normative Truthmaker View says that a

theory's ontological commitments are the entities one rationally ought to admit as

truthmakers, given that one accepts that theory.

96
Or more precisely, on what makes a truthmaking contribution. I’ll be a little fast and loose
about this distinction in what follows, where clarity is not at issue.
97
Cf. Rettler (2016).
165
 In §5.1, I discuss some preliminaries concerning the notion of ontological

commitment and its role within Truthmaker Ontology. In §5.2, I consider some

previous efforts to account for ontological commitment in terms of truthmaking,

highlighting problems where they arise. In §5.3, I present and motivate the Normative

Truthmaker View, explaining how it avoids the problems that afflict other accounts.

Finally, in §5.4, I sketch a methodological procedure involving the Normative

Truthmaker View for deciding on the ontology we ought to accept.

5.1 Ontological Commitment and Truthmaker Ontology

I begin with some preliminaries concerning the nature and import of the notion of

ontological commitment. I will then briefly comment on the role of the notion of

ontological commitment in the context of Truthmaker Ontology.

5.1.1 Preliminaries

In broad and general terms, the ontological commitments of a theory98 are the

requirements that the theory's truth imposes on ontology. If theory T is ontologically

committed to Fs, then Fs are required to be among the contents of ontology if T is to

be true. Accepting T thus commits you to an ontology containing Fs; conversely,

excluding Fs from your ontology commits you to rejecting or revising T.

This is a neutral characterisation of ontological commitment, insofar as it doesn't

assume any particular view of the scope of ontology. If ontology consists of what

there is, ontological commitments are requirements on what there is; if ontology

consists of what is fundamental, then ontological commitments are requirements on

98
A theory, in this context, is just a set of one or more interpreted sentences. It is sometimes
convenient to speak of persons having certain ontological commitments, by dint of the
theories they accept, or the sentences they affirm.
166
what is fundamental, and so on. So as I see it, one's understanding of ontological

commitment depends on one's underlying metaontological outlook: to understand

requirements on ontology, we must first decide what it is that such requirements are

on (see §3.1.3).

That said, you don't need a settled view of the scope of ontology to see why it is

methodologically important to be able to determine the ontological requirements of a

theory. If you know that a highly credible theory is ontologically committed to Fs,

then you have reason to accept Fs into your ontology; you might accept an ontology

of numbers, for instance, because you think that our best scientific theories are

ontologically committed to them. Conversely, if you know that a theory is

ontologically committed to something dubious, you have reason to reject or revise

the theory; you might reject Cartesian dualism, for instance, because you think it

requires a naturalistically unacceptable ontology of immaterial substances. Thirdly,

if you know that one theory has fewer ontological commitments than a rival, this

might lead you to prefer the more parsimonious theory. For instance, if a theory

committed to simples and composites is explanatorily on a par with a theory

committed only to simples, the latter is more choice-worthy for making fewer

demands on the world.99 In at least these ways, ontological commitment is a useful

tool of theory choice and ontic decision-making. Understanding it should be high on

the metaontologist's agenda, whatever the scope of ontology might be.

The main task for a theory of ontological commitment is to clarify the notion of a

requirement on ontology. (Clarifying the scope of ontology is, as I’ve said, a distinct

and prior task.) This task breaks down into two questions. The substantive question

99
More on ontological parsimony in the next chapter.
167
is: what is it for a theory's truth to impose requirements on ontology? The procedural

question is: how do we tell what a theory's requirements on ontology are?

Consider Quine, whose background metaontological assumption is that ontology

consists of what there is (1948: 21). On this assumption, the substantive question of

ontological commitment is: what is it for a theory's truth to impose requirements on

what there is? And Quine's answer seems to understand requirements on what there

in terms of entailment (most likely logical entailment):

To show that a theory assumes a given object, or objects of a given

class, we have to show that the theory would be false if that object
did not exist, or if that class were empty; hence that the theory
requires that object, or members of that class, in order to be true.
How are such requirements revealed? (Quine 1969: 93).

With the last sentence ('How are such requirements revealed?'), Quine shifts focus to

the procedural question of ontological commitment. His answer to this is well-

known:

A theory is committed to those and only those entities to which the

bound variables of the theory must be capable of referring in order
for the affirmations made in the theory be true. (Quine 1948: 33).

Quine isn't telling us here that what it is for a theory to require certain entities is for

it to have them in the range of its bound variables. He's telling us that looking at a

theory's bound variables is the only reliable way to identify its requirements on (i.e.

its logical entailments concerning) what there is.

Not every Orthodox Ontologist accepts both the substantive and procedural

components of Quine's account. Some, for instance, would agree that for a theory to

require that there are Fs is for it to logically entail that there are Fs, yet deny that only

first-order quantification reliably conveys such requirements; they might think

168
second-order quantification also does so, for instance. Others would reject Quine's

view of what ontological requirement consists in; for instance, instead of logical

entailment, one might think of requirements on what there is in terms of metaphysical

entailment (Jackson 1989, Peacock 2011). They might agree that a theory's including

the sentence '$xFx' suffices for its being committed to Fs, though for a different

underlying reason: the theory isn't ontologically committed because it logically

entails that there are Fs, but because it's false at every possible world where there are

no Fs.

This brings us to the distinction between explicit and implicit ontological

commitment. Following Krämer (2014: 2149), an orthodox ontologist might draw

this distinction as follows:

T is explicitly ontologically committed to Fs iffdf. T includes some

sentence that means there are Fs. T is implicitly ontologically
committed to Fs iffdf. T is ontologically committed to Fs & T is not
explicitly ontologically committed to Fs.100

Our answer to the substantive question of what ontological requirement consists in

will shape our understanding of implicit commitment. Suppose the orthodox

ontologist sees ontological commitments as what theories logically entail that there

is. Then it would seem that the sentence 'Everything the Bible says is true' isn't even

implicitly ontologically committed to God (cf. Peacock 2011: 81). Yet intuitively

there must be a God for everything the Bible says to be true. Such concerns about

implicit commitment motivate some Orthodox Ontologists to prefer views based on

metaphysical, rather than logical, entailment.

100
More neutrally, we might say that T is explicitly ontologically committed to Fs iff T
includes some sentence that means Fs ___, where what fills the blank depends on what we
take the scope of ontology to be. T would be implicitly committed to Fs iff T doesn't include
a sentence meaning that Fs ___, but still requires that Fs ___.
169
5.1.2 The Truthmaker Criterion

This is all by way of introduction— I won't discuss here the relative merits of different

accounts of requirements on what there is. For, to repeat, I don't think that ontology

consists of what there is; I think it consists of the subset thereof that makes a

truthmaking contribution, or what there really is. So by my lights, ontological

commitments are requirements on truthmakers. Hence, the substantive question I'm

interested in is: what is it for a theory to require certain truthmakers? And the

procedural question I'm interested in is: how do we tell what truthmakers a theory

requires?

My position commits me to what some call a truthmaker criterion (or ‘view’101)

of ontological commitment. Cameron sums up the basic idea as follows (2008a: 4):

What are the ontological commitments of a theory? For Quine, it is

those things that must be said to lie within the domain of the
quantifiers if the sentences of the theory are to be true. I am a
truthmaker theorist: I hold that the ontological commitments of a
theory are just those things that must exist to make true the sentences
of that theory.

While several philosophers sympathise with some form of truthmaker criterion, only

two offer a detailed defence: Cameron and Rettler.102 But on their accounts, the

truthmaker criterion of ontological commitment is what supports the distinction

between what has ontological status and what can truly be said to exist. As I’ve said

previously (§3.1.3), I think this is dialectically back to front: it’s inappropriate to

argue from a claim about which view of ontological commitment is correct to a claim

101
Rettler (2016).
102
See Cameron (2008a, 2010a, forthcoming(a)), Rettler (2016). Other sympathisers include
Armstrong (2004) and Heil (2003), but neither offer many details.
170
about the scope of ontology. It is rather the truthmaker criterion of ontological status

that, in tandem with the Divergence Thesis, supports the claim about the scope of

ontology Cameron and Rettler want to defend. Formulating a truthmaker criterion of

ontological commitment is a secondary, methodological concern, albeit an important

one: given that ontology consists of truthmakers, we'd like to know how to understand

and identify the requirements theories impose on what truthmakers there are. And this

is by no means a fait accompli: hard questions arise about how to understand the

notion of requirement. For example, should we construe requirements on truthmakers

in terms of entailment? If so, what kind of entailment? If not, what? How do a theory's

truthmaker-requirements relate to the language it employs? How might they be

identified? In what follows, I’ll attempt to shed light on some of these questions.

5.2 In Search of a Truthmaker Criterion

First, however, I will look at some extant truthmaker views of ontological

commitment.

5.2.1 The General Truthmaker View

Rettler (2016) distinguishes between two truthmaker views of ontological

commitment. On the 'Specific Truthmaker View', which Rettler rejects, affirming a

sentence ontologically commits you to whatever in fact makes that sentence true. On

the 'General Truthmaker View', which Rettler endorses, affirming a sentence only

ontologically commits you to there being some truthmaker(s) or other for the

sentence. Here's his illustration (Rettler 2016: 1419):

[B]y saying 'tables exist', we're not ontologically committed to the

truthmakers; we're just ontologically committed to there being
truthmakers. Namely, we're committed to the sentence being true
and having a truthmaker (or several).
171
The Specific Truthmaker View seems an obvious non-starter. It entails that only true

theories have ontological commitments, and that any ontological commitment is to

something that really exists. This makes ontological commitment useless as a tool of

theory choice and ontological decision-making: we couldn't evaluate a theory's

ontological commitments prior to accepting the theory as true and its commitments

as real. Rettler is therefore quite right to reject the Specific Truthmaker View.

But Rettler's General Truthmaker View is hardly more useful. All the General

Truthmaker View tells us is that affirming a sentence incurs some sort of ontological

commitment to truthmakers— we know not what. This doesn't seem to leave much

scope for assessing and comparing the ontological commitments of different theories,

or for deciding what to admit to our ontology based on what we accept as true.

Anticipating this objection, Rettler replies that we never should have expected to

be able to determine ontological commitments just by looking at what sentences

say— this is Orthodox Ontology's mistake (2016: 1423). I agree. But just because we

can't read off ontological commitments directly from sentences doesn't mean there's

nothing more informative to say about what they commit us to. The truth of 'Tables

exist' might not require an ontology containing tables (or so Truthmaker Ontologists

think); but one would hope to be able to say more about what it does require than just

'some truthmakers'.

In any case, I'm not sure that the General Truthmaker View amounts to a view of

ontological commitment at all. It sounds more like a roundabout way of affirming

Truthmaker Maximalism: every true sentence must be made true by some bit of

ontology, so affirming a sentence commits you to there being some bit of ontology

that makes it true. But couldn’t you accept this without thinking that truthmaking has

anything to do with ontological commitment as such? Consider the view that a

172
sentence is ontologically committed to the things whose existence it entails. Given

Truthmaker Maximalism, every sentence entails that there is something (or other)

that makes that sentence true. Combining the two, it seems to follow that every

sentence is ontologically committed to there being something or other that makes that

sentence true. This sounds like Rettler's General Truthmaker View. But it’s a stretch

to call this a truthmaker view of ontological commitment: it's just a standard

entailment-based view combined with Truthmaker Maximalism.

5.2.2 The Metaphysical Entailment View

It’s natural to think of a theory's requirements on ontology in terms of its entailments

concerning ontology. Indeed, as we saw earlier, those who construe ontological

commitments as requirements on what there is often cash this out in terms of some

kind of entailment. So perhaps we should think of a theory's requirements on

truthmakers in terms of its entailments concerning truthmakers. This seems to fit with

certain ways of articulating the truthmaker criterion, such as 'the ontological

commitments of a theory are just those things that must exist to make true the

sentences of that theory' (Cameron 2008a: 4, my emphasis. But see below). It also

seems to let us ascribe ontological commitments without assuming the truth of the

committed theories, or the reality of what they commit to (unlike the so-called

Specific Truthmaker View).

But what, specifically, would an entailment-based truthmaker look like? We are

unlikely to be able to capture truthmaker-commitments in terms of logical entailment.

This would make ontological commitment too bound up with logical form for most

Truthmaker Ontologists. The likeliest candidate seems to be metaphysical entailment.

Bricker (2014: §3.1), for example, construes the truthmaker criterion as follows:

173
 (TOCM) (i) T is ontologically committed to a iff,

necessarily, if T is true then a is a truthmaker for T.

(ii) T is ontologically committed to Fs iff,

necessarily, if T is true then Fs are among the

truthmakers for T.

An objection to (TOCM), due to Bricker (2014: §3), is that it seems to undergenerate

ontological commitments. Assume Truthmaker Necessitarianism, and consider the

following cases:

(C1) Let a and b be mutually contingent existents. Suppose that,

necessarily: (i) if 'a and b exists' is true, it is made true by a and b; (ii)

if 'a exists’ is true, it is made true by a. Plausibly, a doesn't necessitate,

hence cannot make true, 'a and b exists'. So (TOCM) has the implausible

consequence that 'a exists' is ontologically committed to a but 'a and b

exist' is not.

(C2) Let c be a complex object which essentially has c* as a proper

part. Suppose that, necessarily, if 'c exists' is true, it is made true by c.

Plausibly, c* doesn't necessitate, hence cannot make true, 'c exists'. So

(TOCM) has the somewhat implausible consequence that 'c exists' is

ontologically committed to c but not to its essential proper part, c*.

(C3) Let 'a is F' be an inessential predication. Suppose that,

necessarily, if 'a is F' is true, it is made true by the fact [a is F].

Plausibly, a doesn't necessitate, hence cannot make true 'a is F'. So

(TOCM) has the somewhat implausible consequence that 'a is F' is

174
 ontologically committed to the fact that a is F but not to its constituent,

a.

This objection isn't as devastating as Bricker suggests (2014: §3.4), but it does

motivate a more careful formulation of (TOCM). Let's stipulate that a is an element

of a truthmaker for P iff either (i) a is a partial truthmaker for P103, (ii) a is a

mereological part of a truthmaker for P, or (iii) a is a non-mereological constituent

of a fact, or state of affairs, that is a truthmaker for P (cf. Armstrong 1997). Now

consider the following revision to (TOCM):

(TOCM*) (i) T is ontologically committed to a iff,

necessarily, if T is true then a is a truthmaker for T or an

element of one.

(ii) T is ontologically committed to Fs iff,

necessarily, if T is true then Fs are among the

truthmakers for T or their elements.

(TOCM*) avoids the problems above. Suppose that, necessarily, if 'a and b exists' is

true, it is made true by a and b. Then, since a is required as a partial truthmaker,

(TOCM*) implies that 'a and b exists is ontologically committed to a, even though a

doesn't necessitate the truth of 'a and b exist'. Suppose that, necessarily, if 'c exists' is

true, it is made true by c, which essentially has c* as a proper part. Then, since c* is

required as a mereological part of the truthmaker, (TOCM*) implies that 'c exists' is

ontologically committed to c*, even if c* doesn't necessitate the truth of ‘c exists'.

Suppose that, necessarily, if 'a is F' is true, it is made true by [a is F]. Then, since a

103
I.e. ∃X(X ⊨ P & a Î X). See §2.3.1.
175
is required as a non-mereological constituent of the truthmaker, (TOCM*) implies

that 'a is F' is ontologically committed to a, even if a doesn't necessitate the truth of

'a is F'.

But although (TOCM*) is an improvement, there remain serious problems with

construing truthmaker-commitments in terms of metaphysical entailment. Suppose

that T is necessarily false. The following is a theorem in the weakest system of modal

logic: □¬P ® □(P ® Q). It follows that for any P, □(T ® P). So the right bijunct of

(TOCM*) is vacuously satisfied: for any Fs whatsoever, T is ontologically committed

to Fs. Thus, (TOCM*) has the strange consequence that impossible theories are

ontologically committed to everything.

The same problem arises on the following familiar view of ontological

commitment (cf. Jackson 1989, Peacock 2011):

(EOCM) T is ontologically committed to Fs iff, necessarily, if T

is true then Fs exist.

Arguably, however, the problem is even worse for (TOCM*) than it is for (EOCM).

When we plug impossible theories into (EOCM), we get vacuous satisfactions of the

right bijunct like the following: necessarily, if 'Round squares exist' is true, then

unicorns exist. This sounds odd, but we're used to such quirks of entailment. But when

we plug the same theory into (TOCM*), we get this: necessarily, if 'Round squares

exist' is true, then unicorns are among the (elements of the) things in virtue of which

it is true. To me, this sounds outright false. If (per impossibile) 'Round squares exist'

were true, the explanation for its truth might have something to do with round squares,

but certainly nothing to do with unicorns. So friends of (TOCM*) can't dismiss the

176
above problem as a technical oddity; it has results that conflict with our intuitions

about truthmaking.104

Another aspect of the problem concerns theory choice (cf. Jackson 1989, Krämer

2014, Michael 2008). Following Krämer (2014: 2150ff.), consider the following pair

of theories:

TN At least two objects exist and, necessarily, no two objects

compose a further object.

TU At least two objects exist and, necessarily, any two objects

compose a further object.

Suppose we want to use (TOCM*) to compare the ontological commitments of TN and

TU for parsimony. To do this, we must make certain assumptions about their modal

status (Krämer 2014: 2151). If both theories are necessarily false, then they are

equally and maximally unparsimonious, since both are ontologically committed to

everything. So if we're to count one of the theories more parsimonious than the other,

we need to assume that one of them is possibly true, hence that the other is necessarily

false. But it then seems that we can't count one of the theories more parsimonious

than the other without begging the question against one of them (Krämer: op. cit..).

To assume that TN is possibly true is to assume that it isn't necessary that any two

objects compose a further object, hence that TU is false; to assume that TU is possibly

true is to assume that it isn't necessary that no two objects compose a further object,

hence that TN is false.

104
There is, however, another familiar problem for (EOCM) which (TOCM*) avoids. On
(EOCM), any theory whatsoever is ontologically committed to every necessary existent. Not
so on (TOCM*). If Fs necessarily exist, and therefore exist at every world where T is true, it
doesn't follow that T is true in virtue of Fs at those worlds.
177
5.2.3 The Potential Grounds View

In earlier papers, it wasn't too clear what Cameron (2008a, 2010a) meant by the claim

that theories are ontologically committed to what must exist to make them true.105

Some took him to hold a version of the Metaphysical Entailment View (Bricker 2014:

§3.1, Brogaard 2008), others the non-starter that is the Specific Truthmaker View

(Rettler 2016). Recently, however, Cameron (2019) has clarified his view of

ontological commitment, which turns out to be more nuanced than his interpreters

realised. I'll call it the 'Potential Grounds View'.

Consider the question: what conditions must reality meet in order for the claim

'Ball is red' to be true? The answer, according to Cameron, is that there must be

adequate grounds for the claim. But there are different potential ways the world might

furnish such grounds. Following Cameron (2019: 357-358), consider three potential

ways of grounding 'Ball is red':

(1) The grounds of 'Ball is red' are (i) That Ball exists, and (ii) That

Ball is red.

(2) The grounds of 'Ball is red' are (i) That Ball exists, (ii) That

Redness exists, and (ii) That Ball instantiates Redness.

(3) The grounds of 'Ball is red' are (i) That Ball exists, (ii) That

Redness exists, and (iii) That the state of affairs of Ball's

instantiating Redness exists.

Assume that (1)-(3) exhaust the potential ways grounding 'Ball is red'. Then, the

thought goes, the truth of 'Ball is red' requires the world to furnish one of these three

105
A point he has now acknowledged (Cameron 2019: 352, fn. 43).
178
sets of potential grounds. So affirming 'Ball is red' commits us to the world being

such that either: (1) Ball exists and Ball is red, (2) Ball exists, Redness exists, and

Ball instantiates Redness, or (3) Ball exists, Redness exists, and the state of affairs of

Ball's instantiating Redness exists.

Now for Cameron, truthmaking is a special case of truth-grounding (2018: 338,

2019: 354). What it is for a to make P true, he thinks, is for P to be grounded by the

true proposition that a exists. So the range of potential grounds for a claim

encompasses its potential truthmakers; these are given by those among the claim's

potential grounds that concern the existence of entities (Cameron 2019: 354). Thus,

if (1) is true, then the truthmaker for 'Ball is red' is Ball; if (3) is true, then the

truthmakers are Ball, Redness, and the state of affairs of Ball's instantiating

Redness.106 'The truthmaker criterion of ontological commitment', says Cameron,

'tells us that we are committed to there being some or other of the potential

truthmakers for the claims we take to be true' (2019: 351-352).

For example, still assuming that (1)-(3) are exhaustive, our ontological

commitment in affirming 'Ball is red' is to the world's being such that either: (1) Ball

exists, (2) Ball and Redness exist, or (3) Ball, Redness, and the state of affairs of

Ball's instantiating Redness exist. In this case, Cameron would say that 'Ball is red' is

ontologically committed to Ball, since 'Ball exists' is on all its lists of potential

grounds (2019: 355). But now suppose there's a fourth potential way to ground 'Ball

is red':

(4) The grounds of 'Ball is red' are (i) That simples arranged Ball-

wise exist, and (ii) That these simples are arranged red-wise.

106
As Cameron explains (2019: 358), 'Ball is red' has necessitating truthmakers in the latter
case but not the former.
179
In this case, 'Ball is red' incurs a disjunctive ontological commitment: the world must

be such that either Ball exists or simples arranged Ball-wise exist (Cameron 2019:

356). Now consider a fifth potential way to ground 'Ball is red':

(5) The grounds of 'Ball is red' are (i) That it is Ball-and-red-ing.

In this case, Cameron would say that 'Ball is red' has no ontological commitments,

since there's a potential way of grounding it that doesn't involve the existence of

entities at all (2019: 357).

Cameron is clear that he's not thinking of 'potential grounds' in modal terms (2019:

355): the thought is not that, in every possible world where P is true, it is grounded

either thus-and-so, or so-and-thus, or … . Rather, Cameron seems to take 'potential

grounds' as primitive: 'we simply have to take claims— whether they be truths,

contingent falsehoods, or impossibilities— to generate, by virtue of what they say

about the world, these lists of potential grounds' (2019: 355). If we're happy with this,

Cameron's Potential Grounds View looks quite attractive. Because 'potential grounds'

needn't be actual grounds, we can assign ontological commitments without assuming

the truth of the committed theories, or the reality of what they commit to. Because

'potential grounds' is (by stipulation) a non-modal notion, we can avoid the sort of

problems that afflict the Metaphysical Entailment View. And although the Potential

Grounds View sometimes makes ontological commitments disjunctive, it offers

greater precision than Rettler's General Truthmaker View, on which the answer to

'What are the ontological commitments of a theory?' is always just 'Some

truthmakers'.

180
 The view I will outline below resembles the Potential Grounds View in some

respects.107 It too rests on a non-modal notion of a theory's potential— or as I prefer,

candidate— truthmakers. But there are at least two differences. First, as I've said

before, I don't think that truthmaking can be assimilated into grounding— my view

won’t assume otherwise.108 Second, I aim to say more about what potential, or

candidate, truthmakers for a theory are, and how these might be determined.

5.3 Towards a Normative Truthmaker View

I will now propose a new truthmaker view of ontological commitment. I call it the

Normative Truthmaker View.

5.3.1 Normative Commitments

Suppose I promise to mow your lawn every week, but rarely turn up. Am I committed

to mowing your lawn? In one sense, I'm not: a truly committed lawn-mower would

show up without fail. But in another sense, I am: I ought to mow your lawn weekly,

because I promised I would. 'Commitment' thus seems to have both descriptive and

normative senses: I have my descriptive commitments in virtue of what I actually do,

and my normative commitments in virtue of certain constraints on what I ought to do

(Dieveney 2008). I want to suggest that we think of ontological commitment in the

broadly normative sense. The basic idea is this: for T to be ontologically committed

to Fs is for T to be such that one rationally ought to admit Fs into one's ontology,

given that one accepts T.

107
The views were developed independently, however— I came across the manuscript for
Cameron’s paper after the bulk of this chapter was written.
108
See §2.2.1. Again, I’m assuming that Cameron thinks that truthmaking just is a case of
truth-grounding (cf. p.138 fn.83).
181
 It is hardly novel to think that ontological commitment is connected to rational

obligation. Many would agree that we rationally ought to endorse the ontological

commitments of the theories we accept. But we tend to think that this obligation

arises from prior facts about ontological commitment. You might think that T's

ontological commitment to Fs consists in its logically entailing that there are Fs, and

that you rationally ought to believe what logically follows from your beliefs. In that

case, it's natural to say that believing T rationally obliges you to admit Fs to your

ontology because T is ontologically committed to Fs. What I'm suggesting, however,

is that if T is ontologically committed to Fs, this is because believing T rationally

obliges you to admit Fs to your ontology. Imposing such rational obligations is what

ontological commitment consists in. (You might think that entailment is a way for

theories to create such obligations, but it needn't be the only way.)

My view is that ontology consists of all and only the things that make a

truthmaking contribution. Hence, I take a theory's ontological commitments to be the

things one rationally ought to take to make a truthmaking contribution, given that one

accepts the theory. This is the substantive claim underlying what I call the Normative

Truthmaker View of ontological commitment. The pressing question in the

development of such a view is what truthmaking-contributors a theory rationally

obliges its holders to posit. I will now sketch an answer to this question.

5.3.2 The Normative Truthmaker View Outlined

I take it that one way for a theory to rationally oblige its holders to admit certain

entities as truthmakers is to explicitly designate those entities as truthmakers. This

suggests the following truthmaker criterion of explicit ontological commitment (see

§5.1.1):

182
 (TOCEX) T is explicitly ontologically committed to Fs iffdf. T

contains a sentence designating Fs as truthmakers, or as elements of

truthmakers.

I don’t claim that there’s some 'canonical language' that's uniquely best for

designating entities as truthmakers. All that matters for my purposes is that there is

some language we can use to designate truthmakers clearly enough for our liking. But

to fix ideas, I'll assume that we can explicitly designate truthmakers using sentences

like '$x(x = a & x ⊨ P)' and '$x(Fx & x ⊨ P)'. These sentences are explicitly

ontologically committed to a and Fs respectively, because they designate them as

truthmakers. The sentence '$X(X ⊨ P & b Î X)' is explicitly ontologically committed

to b (along with the rest of X), because it designates b as a partial truthmaker. And

recall that, on my view, truthmaking can be either direct or indirect (see §2.4): x

directly makes P true iff x ⊨ P & ¬$Q(x ⊨ Q & Q Þ P)'; x indirectly makes P true

iff x ⊨ P & $Q(x ⊨ Q & Q Þ P)'. So, for instance, '$x(x = c & x ⊨ P & P Þ Q)' is

explicitly ontologically committed to c, which it designates as a truthmaker for P and

as an indirect truthmaker for Q. I take it that, if a theory contains sentences like these,

then its holders are rationally obliged to admit the truthmakers designated— that’s

what the ontological commitment consists in.

Of course, most theories don't explicitly designate truthmakers. So most theories

lack explicit ontological commitments. But we can render any theory explicitly

ontologically committal by means of what I call a TM-extension. A TM-extension

enriches a theory with sentences designating truthmakers for the propositions the

theory expresses. It also fully specifies all intermediate grounding claims when the

designated truthmakers are indirect. To ensure the latter, I’ll stipulate that '⊨' always

183
indicates direct truthmaking and 'Þ' direct grounding109 when they occur in a TM-

extension. This means that a TM-extension must articulate any indirect truthmaking

claims by starting with a direct truthmaking claim and progressing via a chain of

direct grounding claims to the part of the theory that is indirectly made true.

To illustrate, consider a two-sentence theory:

T {'P', 'Q'}

One way to construct a TM-extension for T is simply to designate direct truthmakers

for P and Q. For example:

T+ 1 {'P', 'Q', '$x(x = a & x ⊨ P)', '$x$y(x = b & y = c & x, y ⊨ Q)'}

But another way is to find direct grounds for P and Q, direct grounds for those

grounds, and so on, until we reach indirect truthmakers for P and Q. We might end

up with something like this:

T+ 2 {'P', 'Q', '$x(x = a & $R(x ⊨ R & R Þ P))', '$y(y = b &

$S1$S2((y ⊨ S1) & (S1 Þ S2) & (S2 Þ Q)))'.110

Now, as it stands, TM-extensions don't tell us much about what theory-holders

rationally ought to admit as truthmakers. The only constraints on them are syntactic;

there are no restrictions on what truthmakers or grounds a TM-extension can assign.

109
An instance of grounding between P and R is direct iff P grounds R and there is no Q such
that P grounds Q and Q grounds R.
110
A TM-extension might also posit direct truthmakers for some parts of the theory and
indirect truthmakers for others.
184
A TM-extension might say that the number 7 makes true ⟨Socrates exists⟩, for

instance, or that ⟨Snow is white⟩ grounds ⟨Grass is green⟩.

The way forward is to find constraints on what counts as an admissible TM-

extension of a theory. The aim is to ensure that the truthmakers posited by an

admissible TM-extension are rationally acceptable candidates for making the

extended theory true— rationally acceptable, that is, by that theory's lights. We're not

looking for constraints on what truthmakers it is rational to admit simpliciter, but on

what truthmakers it is rational to admit given that one accepts a certain theory. For

instance, it may be irrational from a theory-external perspective to admit ghosts as

truthmakers, but not from the perspective of a theory about ghosts. With this in mind,

I will now suggest some constraints on admissible TM-extensions that are primarily

based on (i) what a theory says, and (ii) what follows from the nature of truthmaking

and grounding (as opposed to prior assumptions about what truthmakers or grounds

are independently plausible).

(1) Admissible TM-extensions are subject to the logical and

structural principles governing truthmaking and grounding. For example, a

TM-extension will be inadmissible if it involves symmetric or reflexive

cases of grounding or truthmaking111. They are also subject to principles

governing the interplay between truthmaking and grounding (see §2.4). For

example, given Quasi-Asymmetry, an admissible TM-extension cannot

verify both 'a ⊨ P' and 'P Þ ⟨a exists⟩'. Given the Grounding Principle, an

admissible TM-extension cannot verify 'a ⊨ P' and 'P Þ Q' while falsifying

'a ⊨ Q'. Admissible TM-extensions are also subject to constraints on what

111
Save, perhaps, for certain exceptional cases involving propositions serving as truthmakers
for truths about propositions. See §2.3.1.
185
makes true or grounds truths of certain logical forms. For example, an

admissible TM-extension will ground conjunctions in their conjuncts and

existential generalisations in their instances, and not vice versa.

(2) Any claims of the form 'x ⊨ P' in a TM-extension indicate direct

truthmaking, which I take to be subject to a version of the Aboutness

Constraint (see §3.3.1). That is, where direct truthmaking is concerned, the

terms that occur in the truthbearer also occur in the description of the

truthmaker. Thus, a could directly make true ⟨a exists⟩, but b couldn't. The

fact that a is F, or a's F-ness trope, could directly make true ⟨a is F⟩, but the

fact that b is G or c's H-ness trope couldn't. So, for example, a TM-extension

will be inadmissible if it posits the number 7 as the direct truthmaker for

⟨Socrates exists⟩. Where indirect truthmaking is concerned, the terms that

occur in the truthbearer need not occur in the description of the truthmaker

(see §3.3). But in such cases, a TM-extension must display how the subject

matter of the indirectly made true proposition is connected, via a chain of

direct grounding claims, to its indirect truthmaker.

(3) Admissible TM-extensions can only feature truthmaking and

grounding claims that meet a basic threshold of intuitive explanatory

adequacy. If a makes P true, this implies the correctness of the explanatory

proposal, 'P is true because a exists'; and if P grounds Q, this implies the

correctness of the explanatory proposal, 'Q because P' Q’ (see §2.3). My

crude suggestion is that a TM-extension is inadmissible if it features a

truthmaking or grounding claim such that the corresponding explanatory

proposal, 'f because y', is inherently inapt. Very roughly, an inherently inapt

186
explanatory proposal is one where the content of the putative explanans

precludes it from having any explanatory bearing on the explanandum. For

example, I take it that 'Grass is green because snow is white' is an inherently

inapt explanatory proposal, and so a TM-extension will be inadmissible if it

features the grounding claim '⟨Snow is white⟩ Þ ⟨Grass is green⟩'.

Clearly, the notion of an explanatory proposal being inherently inapt

needs fleshing out. Here, I’ll limit myself to sketching a rough and ready test

(inspired by Joyce 2001: 3). Imagine yourself in the position of a radical

interpreter, observing an alien community whose language is

homophonically indistinguishable from English. You are satisfied that the

alien’s word ‘because’ has the same meaning as its homophonic English

counterpart. You notice, however, that the aliens are disposed to sincerely

assert the sentence 'f because y'. If, qua charitable interpreter, you would

take this as evidence that either ‘f’ or ‘y' cannot have the same meaning as

its homophonic English counterpart, this indicates that the English claim 'f

because y' is inherently inapt. For example, if the aliens sincerely asserted

‘Grass is green because snow is white’ and meant what we mean by it, this

would be a sign of conceptual incompetence— given that ‘because’ has its

English meaning, a charitable interpreter would be inclined to judge that at

least one of the alien’s sentences, ‘Grass is green’ and ‘Snow is white’, does

not have its English meaning. This indicates that the English claim ‘Grass is

green because snow is white’ is inherently inapt, and so the grounding claim,

'⟨Snow is white⟩ Þ ⟨Grass is green⟩' can’t feature on an admissible TM-

extension.

187
 However, consider the following explanatory proposals: (i) Ghosts exist

because disembodied souls exist; (ii) It is true that disembodied souls exist

because disembodied souls exist. Dubious as these claims might be, I submit

that the above test would not indicate that they are inherently inapt. So an

admissible TM-extension of 'Ghosts exist' might include the claims: (i)

⟨Disembodied souls exist⟩ Þ ⟨Ghosts exist⟩, and (ii) Disembodied souls ⊨

⟨Disembodied souls exist⟩. This is as it should be. The truthmakers and

grounds featured on an admissible TM-extension should be rationally

acceptable given what the theory says but need not be so from a theory-

external perspective.

(4) Admissible TM-extensions should consistently designate entities

of a particular ontological category as truthmakers for a particular kind of

truth. To illustrate, suppose that T consists of two atomic predications: 'a is

F' and 'b is G'. An admissible TM-extension of T might designate the facts

[a is F] and [b is G], or the tropes a's F-ness and b's G-ness as direct

truthmakers for these claims. But a TM-extension of T will be inadmissible

if it designates the fact [a is F] as the direct truthmaker for ⟨a is F⟩ and b's

G-ness trope as the direct truthmaker for ⟨b is G⟩. For plausibly, whatever

entities serve as truthmakers for atomic predications, they are uniform across

all atomic predications.

I don't claim that these constraints are exhaustive, and some of the above will need

ironing out on another occasion. But I've tried to indicate that, based on what a theory

says and what follows from the nature of truthmaking and grounding, there are

plausible constraints on what theory-holders might rationally admit as truthmakers.

188
This puts us in a position to see how TM-extensions bear on the truthmakers theory-

holders rationally ought to admit. If T has an admissible TM-extension T+, then the

truthmakers designated by T+ are rationally acceptable candidate truthmakers for T.

Thus, suppose T has exactly three admissible TM-extensions: T+1 designates the Xs

as truthmakers for T, T+2 designates the Ys, and T+3 designates the Zs. Then the things

you rationally ought to admit as truthmakers if you accept T are either the Xs or the

Ys or the Zs. Now suppose that every admissible TM-extension of T designates Fs as

truthmakers. Then someone who accepts T rationally ought to admit Fs as

truthmakers. So, given that a theory's ontological commitments are the things its

holders rationally ought to admit as truthmakers, this suggests the following view of

ontological commitment:

(TOCN) (i) For any theory T, with admissible TM-

extensions T+1, T+2, …, T+n, T is ontologically

committed to there being either the things to which T+1

is explicitly ontologically committed, or the things to

which T+2 is explicitly ontologically committed, …, or

the things to which T+n is explicitly ontologically

committed.

(ii) T is ontologically committed to a/Fs iff every

admissible TM-extension of T is explicitly

ontologically committed to a/Fs.

As I've said, the Normative Truthmaker View has some aspects in common with

Cameron's Potential Grounds View. To borrow Cameron’s toy example, let B be the

single-sentence theory, 'Ball is red'. Suppose B has exactly two admissible TM-

189
extensions, which incur the following explicit commitments (as measured by

(TOCEX)):

B+1 Ball; Redness; The state of affairs of Ball's instantiating

Redness.

B+2 Ball; Ball's redness trope.

In this case, like Cameron, I say that B is ontologically committed to Ball: if every

admissible TM-extension of B designates Ball as a truthmaker (or an element of one),

then you rationally ought to admit Ball as a truthmaker (or an element of one), given

that you accept B. Now suppose there is a third admissible TM-extension of B, whose

explicit commitments are as follows:

B+3 Simples b1, …, bn; The fact that b1, …, bn are arranged

Ball-wise and Redness-wise.

In this case, like Cameron, I say that the ontological commitment of B is to there

being either Ball or b1, …, bn: given that you accept B, you rationally ought to think

that either Ball or b1, …, bn make a truthmaking contribution to B. Thus, the

Normative Truthmaker View might be seen as a cousin of the Potential Grounds

View— albeit one that doesn't define truthmaking in terms of grounding, and which

attempts to shed more light on what a (non-modal) notion of potential/candidate

truthmakers might be.

5.3.3 In Favour of the Normative Truthmaker View

I will now outline some of the benefits of the Normative Truthmaker View and reply

190
to recurrent objection to truthmaker views of ontological commitment.

First, as a reminder, the notion of ontological commitment plays a different role in

my overall account than it does for other Truthmaker Ontologists, and the motivations

for the Normative Truthmaker View are accordingly different. By my lights, the

benefits arising from the ability to distinguish ontology proper from what can truly

be said to exist are not, properly speaking, benefits of a truthmaker criterion of

ontological commitment. They rather arise from the truthmaker criterion of

ontological status, in conjunction with the Divergence Thesis. Given the assumptions

I began with in this chapter— that ontological commitments are requirements on

ontology, and that ontology consists of the truthmakers— the question is not whether

to adopt a truthmaker view of ontological commitment but which one. Accordingly,

the motivations for the Normative Truthmaker View are primarily reasons for

preferring it over other candidate truthmaker views.

Unlike what Rettler calls the Specific Truthmaker View, the Normative

Truthmaker View lets us ascribe ontological commitments without assuming the truth

of the committed theories or the reality of what they commit to. That T rationally

obliges its holders to admit Fs as truthmakers does not imply that T is true or that Fs

actually do any truthmaking. This is in keeping with the idea that ontological

commitment is an important tool of theory choice and ontic decision-making. If T is

ontologically committed to Fs, i.e. rationally obliges its holders to admit Fs as

truthmakers, then if you have reasons to accept T, you have reason to admit Fs to your

ontology. But conversely, if you have reason to believe that Fs are dubious, you have

reason to reject or revise T.

The Normative Truthmaker View also differs from Rettler's General Truthmaker

View. On the Normative Truthmaker View, the answer to 'What is T ontologically

committed to?' might be 'Fs', or it might be 'Fs or Gs or Hs'. Such answers are more

191
informative than the answer we get on the General Truthmaker View, which is always

just 'Some truthmakers'.

Unlike the Metaphysical Entailment View, the Normative Truthmaker View does

not imply that impossible theories are ontologically committed to everything. It might

be that 'Round squares exist' is ontologically committed to round squares, if round

squares are on every admissible TM-extension of 'Round squares exist'. But it will

not be ontologically committed to unicorns, which won't be candidate truthmakers

given the sort of constraints on admissibility suggested above.

Cameron's Potential Grounds View promises similar advantages over these other

views, and I won't argue that the Normative Truthmaker View is superior. You may

find the Normative Truthmaker View more congenial, however, if you think that

truthmaking is distinct from truth-grounding, or want more details than Cameron

offers on what the notion of ‘potential’ truthmakers amounts to.

A recurrent objection to truthmaker views of ontological commitment is that they

presuppose the standard Quinean quantifier criterion (Asay 2011, Mantegani 2014,

Schaffer 2008). The thought is that, in order to designate certain entities as

truthmakers, we'll have to use sentences like 'There is an F, such that F is a

truthmaker'. Indeed, I have explicitly said that we can incur explicit ontological

commitments to truthmakers by using quantified sentences such as '$x(x = a & x ⊨

P)'. Because of this, the objectors would claim, I'm unwittingly assuming the familiar

quantifier view: '$' is still doing the work. As Schaffer puts it (2008: 16), 'truthmaker

commitments are parasitic upon quantifier commitments, and so the truthmaker view

cannot possibly replace the quantifier view'.

This objection rests on a confusion about what a 'view' of ontological commitment

involves. Suppose Moe thinks that theories are ontologically committed to what they

metaphysically entail there is, Larry thinks that theories are committed to what they
192
logically entail there is, and Curly thinks that theories are committed to what they

conceptually entail there is. All three agree that if a theory includes the sentence

'$xFx', it is ontologically committed to Fs. Do Moe, Larry and Curly have the same

view of ontological commitment? No. While they agree that sentences of the form

'$xFx' convey ontological commitment, they hold substantially different views about

ontological commitment consists in.

I think that a theory's being ontologically committed to Fs consists in its rationally

obliging its holders to admit Fs as truthmakers. And I've suggested that if a theory

includes the sentence '$x(x = a & x ⊨ P)', then it is ontologically committed to a. But

this doesn't mean I'm unwittingly assuming a 'quantifier view' of ontological

commitment. The above sentence isn't ontologically committed to a because it

quantifies over a; it is ontologically committed to a because it rationally obliges

someone who affirms it to admit a as a truthmaker— the use of quantification just

makes this commitment manifest. More generally, just because two philosophers

think that ontological commitment can sometimes be conveyed by the same linguistic

means doesn't mean they have the same underlying account of ontological

commitment. I have given an account of what ontological commitment (and indeed

ontology) consists in that differs significantly from the standard Quinean approach.

The fact that certain (highly circumscribed) uses of quantification might be used to

convey ontological commitments, as I construe them, in no way undermines this.

5.4 Determining an Ontology of Truthmakers

I will now sketch a (somewhat idealised) procedure for determining an ontology of

truthmakers. In a nutshell, I suggest that our aim should be to find the best TM-

extension for the truths we have best reason to accept. This is the TM-extension that

193
is both (i) admissible, and (ii) scores higher than any other admissible TM-extension

by the measure of overall theoretical virtue. I’ll break the procedure into three stages.

(Stage One) Determine the Truths

The first stage involves deciding on an initial body of claims that we have good reason

to accept as true. I have in mind something like what Armstrong (2004: 26) calls our

'epistemic base': a collection of truths— or what we take to be truths— that serves as

our starting point in the search for truthmakers. I think this collection will include

'Moorean' commonsense truisms: 'There are tables', 'I have hands', 'Snow is white',

and so on (Armstrong 2004: 26-30, 2006). It will also include the well-established

claims of science, in a broad sense that includes natural science, social science, and

mathematics. Criteria internal to these disciplines will determine what we're to

include in our initial stock of truths.

We shouldn’t think of the truths included in the initial stock as inviolable. For one

thing, we might have to reject or revise certain claims from our initial stock, if we

find that they make unacceptable demands on ontology. For another, we may find

apparent inconsistencies among our initial stock— between certain scientific claims

and certain commonsense claims, for example— that we'd need to resolve one way

or another (see §7.3.3). Finding a plausible and consistent stock of truths, and the best

theory of truthmakers for those truths, is likely to require some reflective equilibrium.

As Armstrong puts it (2004: 26):

We take certain things to be true, and then ask what truthmakers

these truths demand. It may be that at times we will think that we
must have certain truthmakers, and as a result add to what we take
to be our stock of truths. Or we might find ourselves unwilling to
postulate certain truthmakers and therefore having to reject what we
had previously taken to be truths.

194
 (Stage Two) Determine What Grounds What

The next stage involves questions of grounding: what is the best account of the

grounding relations among the truths we accept, and of the fundamental truths in

which they are ultimately grounded? As Fine says (2012: 44), 'the attempt to

determine what grounds what naturally proceeds in stages— one first determines the

relatively immediate grounds for the truths in question, then the relatively immediate

grounds of those grounds, and so on until one reaches the ultimate grounds'. As well

as abiding by the sort of constraints on admissibility described earlier, we should seek

the most virtuous system of grounds available, as measured in terms of familiar

theory-choice criteria. Here's Fine again (2001: 22):

[A] system of grounds may be appraised, in much the same way as

any other explanatory scheme, on the basis of such considerations
as simplicity, breadth, coherence, or non-circularity. Perhaps the
most important virtue in this regard is explanatory strength, the
capacity to explain that which stands in need of explanation and
would otherwise be left unexplained.

What sort of principles might guide our choice between different candidate grounds?

One candidate is a principle of economy: all else equal, it is preferable to ground a

given truth in as few truths as possible. Another is a principle of unification: all else

equal, it is preferable to find common grounds for different kinds of truth than to posit

bespoke grounds for each. A third is a principle of explanatory completeness: all else

equal, it is preferable to minimise the number of ungrounded truths. Intuitive,

scientific and logical considerations will also plausibly play a role.

A pressing question at this stage is where to stop positing further grounds and take

certain truths as fundamental. Plausibly, given the sort of principles just suggested,

the aim is to arrive at a minimally complete set of grounds, in the sense discussed in

Chapter 4. By definition, this would be a maximally parsimonious and explanatorily

195
complete set of fundamental truths in terms of which all others are ultimately

explained. But there are, of course, difficult questions about what truths will turn out

to be fundamental. Perhaps we'll be able to ground all other truths in truths about the

microphysical, but perhaps not. Perhaps, for example, certain truths about minds or

moral properties can't be accounted for in terms of anything more basic and should

be treated as fundamental.

(Stage Three) Determine the Truthmakers

By now, we have an 'alethic hierarchy' of fundamental and non-fundamental truths,

structured by grounding relations (see §4.1). The last stage of our idealised procedure

involves designating truthmakers, thus providing the ontological foundation for our

alethic hierarchy. Again, we want the most theoretically virtuous account of

truthmakers available, guided by principles of parsimony, unification, explanatory

completeness, and so on. Plausibly, our aim should be to find the most plausible set

of fundamental truthmakers for the theory we started with (see §4.2). This would

involve finding the most plausible candidate set of truthmakers that is minimally

complete with respect to the fundamental truths, as determined at Stage Two. These

entities would then provide a complete and non-redundant truthmaking base for all

the truths in our alethic hierarchy.

At this point, we have constructed the best candidate TM-extension for our overall

theory of the world. The entities designated as truthmakers at Stage Three are the best

candidate direct truthmakers for the best candidate fundamental grounds of our initial

stock of truths (and hence indirectly make true the non-fundamental truths). I suggest

that these truthmakers exhaustively comprise the ontology we ought to accept.

As I've said, this is an idealised procedure. In practice, difficult philosophical

questions will arise at every stage, and it will take hard work to achieve reflective
196
equilibrium between our accounts of what is true, what grounds what, and what the

truthmakers are. But I hope to have given some sense of what an approach to the

question of ontology might look like from the perspective of Truthmaker Ontology,

even if the answer would be extremely hard to find. To quote Armstrong again (2004:

23):

[T]he hunt for truthmakers is as controversial and difficult as the

enterprise of metaphysics. I think that proceeding by looking for
truthmakers is an illuminating and useful regimentation of the
metaphysical enterprise, or at least the enterprise of a realist
metaphysics. But it is no easy and automatic road to the truth in such
matters.

Conclusion

It is important to be able to determine a theory's requirements on ontology— its

ontological commitments. I hold that ontology consists of what makes a truthmaking

contribution, and hence that ontological commitments are requirements on

truthmakers. It proves difficult, however, to spell out the relevant notion of

requirement. I have suggested that we think of 'requirements on truthmakers' as

rational requirements on what theory-holders ought to admit as truthmakers, given

what their theory says. I also offered a way to flesh this out, based on the technical

notion of an admissible TM-extension. The Normative Truthmaker View avoids the

major difficulties that afflict some previous efforts to account for ontological

commitment in terms of truthmaking. It also forms the basis of a (somewhat idealised)

procedure for deciding on an ontology of truthmakers: we attempt to find the best

TM-extension of the claims we accept as true. The entities designated as truthmakers

by this TM-extension comprise the ontology we ought to accept.

197
6 Counting Costs
I now turn to another methodologically important notion: ontological parsimony.

Truthmaker Ontology, I've said, is a form of Stratified Ontology (§3.1); and Stratified

Ontologists often claim that, in light of their views, we must revise our ideas about

how ontological parsimony is to be measured. Ockham's Razor— 'Do not multiply

entities without necessity'— is usually taken to imply that theories are generally more

choice-worthy if they imply the existence of fewer entities. But according to several

Stratified Ontologists, parsimony considerations only apply to claims about what

fundamentally or really exists. Hence, on their view, theories are not necessarily less

choice-worthy for multiplying entities. These 'Razor Revisers', as Baron and Tallant

(2018) call them, have come in for some recent criticism. What the critics often

ignore, however, is that Razor revision amounts to very different things depending on

whether inflationary or deflationary Stratified Ontology is in play. I'm going to argue

that the most prominent inflationary attempt at Razor revision—endorsed by Schaffer

(2009), Bennett (2017), and others—doesn't succeed. But matters are different, I'll

argue, in the deflationary context of Truthmaker Ontology. This chapter, then, has

two main aims: (i) to explain how the methodologically important notion of

ontological parsimony works in the context of Truthmaker Ontology, and (ii) to make

trouble for one of Truthmaker Ontology's major rivals.

In §6.1, I introduce ontological parsimony in general terms and clarify the

distinction between inflationary and deflationary forms of Razor revision. In §6.2, I

demotivate and criticise the most prominent inflationary attempt to revise the Razor:

Schaffer’s Laser. In §6.3, I outline a view of how Truthmaker Ontologists should

measure the ontological parsimony of theories, based on the view of ontological

198
commitment defended in the previous chapter. In §6.4, I argue that this view is better

motivated than Schaffer’s Laser and doesn’t suffer from the same problems.

6.1 Ontological Parsimony and Stratified Ontology

I’ll begin by introducing the notion of ontological parsimony in general terms. I will

then distinguish between two forms of ‘Razor revision’, based on inflationary and

deflationary versions of Stratified Ontology.

6.1.1 Ockham's Razor and Theory Choice

Ockham's Razor is the following familiar maxim:

(Ockham's Razor) Do not multiply entities without necessity.

Typically, Ockham's Razor is interpreted as a comparative principle of theory choice:

all else equal, it is rational to prefer a more ontologically parsimonious theory over a

less ontologically parsimonious rival. Ontological parsimony is usually understood in

terms of ontological commitment, along the following lines:112

(Ontological Parsimony) Theory T1 is more ontologically

parsimonious than a rival theory T2 iff T1 has fewer ontological

commitments than T2.

(If T1 is ontologically committed to fewer token entities than T2, T1 is said to be more

quantitatively ontologically parsimonious than T2. If T1 is ontologically committed

to fewer kinds of entity than T2, T1 is said to be more qualitatively ontologically

parsimonious than T2. It is widely agreed that qualitative parsimony is a theoretical

112
See e.g. Baker (2007: 196-197).
199
virtue, though there is some disagreement over the importance of quantitative

parsimony. I will skirt over this issue in what follows, but see Baker 2003, Nolan

1997 and Lewis 1973: 87 for relevant discussion.)

I assume that ontological parsimony is an epistemic virtue of theories (rather than,

say, a pragmatic or an aesthetic virtue). That is, all else equal, a more ontologically

parsimonious theory is more likely to be true than a less parsimonious rival. The exact

justification for this is a contentious issue.113 But the basic rationale is that choosing

more ontologically parsimonious theories 'diminishes your risk of error', as Russell

puts it (1919: 379). A theory's ontological commitments are the requirements its truth

imposes on ontology (see §5.1). And the more requirements a theory imposes on

ontology, the greater the risk that ontology will fail to comply, hence that the theory

will be false. We might put the point in terms of basic probability (cf. Brenner 2017:

2690, Rodriguez-Pereyra 2002: 205-207). The probability of a conjunction whose

conjuncts are mutually independent is less than the probability of either conjunct

alone. Hence, if O1 and O2 are mutually independent ontological claims, O1 is more

probable than O1&O2. Thus, all else equal, theories are more likely to be true if they

rely on fewer independent ontological claims.

6.1.2 Stratified Ontologists as 'Razor Revisers'

Our understanding of ontological parsimony depends on our understanding of

ontological commitment, which depends in turn on our understanding of what has

ontological status (see §5.1). Orthodox Ontologists think that ontology concerns what

exists, and hence that ontological commitments are requirements on what exists. So

by their lights, the claim that T1 is more ontologically parsimonious than T2 means

roughly that T1 requires the existence of fewer entities than T2. But according to a

113
See Baker (2004) for an overview.
200
number of Stratified Ontologists, we need a more discerning measure of ontological

parsimony— one that only measures a theory's requirements on what really or

fundamentally exists (see §3.1). Thus, Sider (2013: 240) says that parsimony

considerations 'should be restricted to theories about the fundamental nature of the

world'; Cameron (2010a: 262) says that 'the correct formulation of Ockham's Razor

[…] must tell us to judge theories by what entities they admit to their fundamental

ontology'; Bennett (2017: 222) says that 'nonfundamentalia do not count against the

ontological simplicity of a theory […]; Ockham's Razor does not even "see" the

nonfundamentalia'; and Schaffer (2015) argues at length that 'the Razor' should be

replaced by a more discerning maxim he calls the Laser: 'Do not multiply

fundamental entities without necessity'. Baron and Tallant (2018) dub these

philosophers 'Razor Revisers'.

It is easy to see why a Stratified Ontologist would want to be a Razor Reviser. A

common motivation for Stratified Ontology is that it lets us reap the theoretical

benefits of sparse ontological views (compositional nihilism, mathematical

nominalism, etc.) while avoiding the usually associated error theories. For Stratified

Ontology gives us the option of affirming the existence of, e.g., composite objects or

numbers while denying that they fundamentally/really, exist. If parsimony only cares

about fundamental/real existence, this would seem to yield the parsimony benefits of

standard nihilism and nominalism without the revisionary consequences.

However, pace Baron and Tallant, it's a mistake to lump all Razor Revisers into

the same category. To do so is to neglect relevant differences between inflationary

and deflationary versions of Stratified Ontology. To briefly recap: deflationists hold

that reality is exhausted by what fundamentally/really exists. If a deflationist says that

tables do not fundamentally really exist, they mean that reality hosts no tables but that

we can still truly represent our table-less reality using sentences like 'Tables exist'.

201
Inflationists hold that reality consists of both fundamental and non-fundamental

'layers'. If an inflationist says that tables are non-fundamental, they mean that tables

are full-blown—albeit dependent— denizens of reality. Razor Revisers on both sides

of the inflationary/deflationary divide might claim that parsimony considerations are

only relevant where fundamental existence is concerned. But this means different

things, depending on whether 'fundamental existence' is given an inflationary or a

deflationary reading. To appreciate the difference, consider these two familiar claims:

(OC) The ontological commitments of a theory are its requirements

on what exists.

(OP) T1 is more ontologically parsimonious than T2 iff T1 has fewer

ontological commitments than T2.

Inflationist Razor Revisers accept (OC) but reject (OP). For the distinction between

fundamental and non-fundamental existents is exhaustive and, according to the

inflationist, both kinds of existent have full-blown ontological status. So by their

lights, all requirements on what exists are requirements on ontology; hence, they

accept the standard view that ontological commitments are requirements on what

exists. But contrary to (OP), inflationist Razor Revisers hold that only a proper subset

of a theory's ontological commitments count against it by the measure of parsimony.

Thus, inflationist Razor Revisers say things like: 'Nonfundamentalia are no addition

to fundamental being, which is all parsimony cares about. But they are an addition to

being full stop' (Bennett 2017: 222); '[D]erivative entities are an "ontological free

lunch", in the sense that they are genuinely new and distinct entities [and] additional

commitments but they cost nothing' (Schaffer 2015: 647-648). As one sympathiser

puts it:

202
 The basic idea is to distinguish between the ontological
commitments of a theory and its costs, i.e. between the entities that
must exist if a theory is true and the entities that count against the
theory when we just it for ontological parsimony. For […] some
existence claims of a theory T are unproblematic in the sense that
these claims do not count as a disadvantage when we compare T
with rival theories. (Schulte 2011a: 253-254)

Deflationist Razor Revisers, in contrast, accept (OP) while rejecting (OC). On their

view, ontology consists of only what fundamentally/really exists; so contrary to (OC),

a theory's ontological commitments are just its requirements on fundamental reality.

But given this, the claim that parsimony is only relevant to claims about fundamental

existence is consistent with (OP): all a theory's ontological commitments are relevant

to the measure of parsimony, but only requirements on the fundamental count as

ontological commitments. Perhaps, for instance, the truth of 'There are tables' requires

that fundamental reality host certain simples, but no tables. Then, for the deflationist

Razor Reviser, the ontological cost of affirming 'There are tables' is just the cost of

the simples. This is not because tables are somehow a 'commitment without a cost'

(Schulte 2011a: 253), but because we were never ontologically committed to tables

in the first place: positing tables doesn't decrease ontological parsimony because a

table-free ontology can support the truth of 'There are tables' (cf. Cameron 2010a:

263-264).

In short, 'Razor revision' amounts to quite different things depending on the

metaontological context. For deflationary Stratified Ontologists, ontological

parsimony doesn't concern the non-fundamental because non-fundamental entities

lack ontological status— they don't really exist. But for inflationary Stratified

Ontologists, ontological parsimony doesn't concern the non-fundamental, even

though non-fundamental entities are genuine additions to ontology— they really do

203
exist, but Ockham's Razor is somehow blunt against them. The inflationary and

deflationary strategies therefore can and should be evaluated separately. In the next

section, I will look at the most prominent version of the inflationary strategy, arguing

that it doesn't succeed.

6.2 Schaffer's Laser

Schaffer (2009, 2015, 2017) is the clearest example of an inflationary Stratified

Ontologist (see §3.4). He believes that reality divides into fundamental (or

ungrounded) and derivative (or grounded) existents, and sees both as 'full-blown

"heavyweight"' entries on the roster of entities' (2009: 360).114 Moreover, Schaffer

takes a permissive attitude to questions of existence: the question is not whether

numbers, tables, properties, …, exist—of course they do—, but whether they are

fundamental (2009: 347). So Schafferian reality is highly abundant, in a way that

might seem to fall afoul of Ockham's Razor (Schaffer 2009: 361). But Schaffer (2015)

argues that 'the Razor' is the wrong measure of ontological parsimony, and should be

replaced by the more discerning 'Laser':

(The Laser) Do not multiply fundamental entities without necessity.

As Schaffer says (2015: 647): 'The Razor measures the ontological economy of a

theory by the entities it posits, while the Laser measures the ontological economy of

a theory by the fundamental entities it posits'.115 And since the Laser is supposed to

114
For Schaffer, grounding is entity-grounding (see §2.1). I’ll stick with this use of the term
while discussing the Laser.
115
It’s somewhat unclear what Schaffer means when he refers to the fundamental entities
‘posited’ by a theory. He sometimes seems to mean 'entities that are fundamental according
to the theory'. But at other times, he seems to mean 'entities posited by the theory that are in
fact fundamental' (or at least, whose status as fundamental posits is somehow independent of
what the theory says). I'll gloss over this in what follows.
204
replace the Razor, it follows that only fundamental posits matter for parsimony:

'[t]here is no problem with the multiplication of derivative entities' (Schaffer 2009:

36). Thus, Schaffer's Laser comes as a package with what Baron and Tallant call the

thesis of Free Derivatives (2018: 598, minor changes):

(Free Derivatives) For any two theories, T1 and T2, the

preferability of T1 over T2 (or vice versa) on parsimony grounds is
unaffected by any differences in the number of derivative entities
posited by the two theories.

Note that Schaffer doesn't tie the Laser to any particular way of drawing the

distinction between fundamental and derivative entities (Schaffer 2015: 647). He

does, however, assume an inflationary reading of the distinction, whereby the

derivative posits of theories are genuine— albeit cost-free— ontological

commitments:

By the lights of The Laser, derivative entities are an 'ontological free

lunch', in the sense that they are genuinely new and distinct entities
but they cost nothing by the measure of economy. The Laser thus
incorporates an implicit distinction between the commitments of a
theory, and the cost of such commitments. By the lights of The
Laser, derivative entities are additional commitments, but they cost
nothing. (Schaffer 2015: 647-648).

6.2.1 The Case for the Laser

Schaffer (2015) gives three arguments for replacing the Razor with the Laser, with

Bennett (2017) adding a fourth. I will briefly outline each argument and say why I

find them unpersuasive.116

116
I will keep the objections fairly brisk, as I have criticised the Laser (in collaboration)
elsewhere; see Fiddaman and Rodriguez-Pereyra (2018). For further criticisms of the Laser,
see Baron and Tallant (2018) and Turner (2016).
205
Feng and Esther

Schaffer (2015: 648) argues that the Laser fits better than the Razor with plausible

intuitions about parsimony-driven theory choice (cf. Bennett 2017: 221-222). He

illustrates with the following case:

Esther posits a fundamental theory with 100 types of fundamental

particle. Her theory is predictively excellent and is adapted by the
scientific community. Then Feng comes along and—in a moment of
genius— builds on Esther's work to discover a deeper fundamental
theory with 10 types of fundamental string, which in varying
combinations make up Esther's 100 types of particle. This is
intended to be a paradigm case of scientific progress in which a
deeper, more unified and more elegant theory ought to replace a
shallower, less unified, and less elegant theory. Feng's theory is
evidently better in every methodological respect. (Schaffer 2015:
648).

The Razor favours Esther's theory, since it posits fewer entities overall. The Laser

favours Feng's theory, since it posits fewer fundamental entities. Since Feng's theory

is 'evidently better', Schaffer claims that the Laser's verdict is correct. He concludes

that 'what should be compared for economy are Esther's 100 types of fundamental

particle with Feng's 10 types of string. That is, what should be compared for economy

are the fundamental entities' (2015: 648-649).

There are at least two problems with this argument.117 One problem is that not all

else is equal between the two theories. Feng's theory is said to be better in several

methodological respects, including depth, unification and elegance. So even if Feng's

theory is better than Esther's overall, it doesn't follow that it is better with respect to

ontological parsimony. It might be that Feng's theory is worse with respect to

117
Cf. Fiddaman and Rodriguez-Pereyra (2018: 342-344).
206
parsimony, as the Razor predicts, but makes up for it by being better in other respects.

We don't need the Laser to explain why Feng's is the better theory.

A second problem is that the argument does nothing to support the claim that the

Laser should replace the Razor. At most, it shows that theories are more choice-

worthy if they posit fewer fundamental entities. But it is fallacious to move from the

premise that theories are better off for positing fewer fundamental entities to the

conclusion that theories are no worse off for positing derivative entities. It might be

the case that, if T1 and T2 posit the same fundamental entities but T1 posits more

derivative entities, T1 is less choice-worthy for that reason. So even if Schaffer’s

argument shows that positing fewer fundamental entities is preferable, it doesn’t

support the conclusion that derivative entities are cost-free.

Analogy With Conceptual Parsimony

Schaffer (2015: 648-651) argues that the Laser is supported by an analogy between

ontological and conceptual parsimony. He writes:

Primitive concepts are to ideology what fundamental entities are to

ontology, and defined concepts are to ideology what derivative
entities are to ontology. In both the ontological and conceptual
domains, one finds a division between the basis (fundamental
entities, primitive concepts) and the superstructure (derivative
entities, defined concepts)' (Schaffer 2015: 649).

When measuring conceptual parsimony, we tend to think that only primitive

(undefined) concepts count. To use Schaffer's example (2015: 649), suppose Georg

develops a regimentation of set theory, defining 40 set-theoretic concepts from 10

primitives; Hamsa then discovers that 99 set-theoretic concepts, including all 40 of

Georg's, can be defined using a single primitive. Schaffer claims that Hamsa's extra

defined concepts are evidently no mark against her theory, illustrating the point that

207
defined concepts incur no further cost to conceptual parsimony, relative to the cost

the primitives). As he puts it, we should measure conceptual parsimony with the

'Conceptual Laser' ('Do not invoke primitive concepts without necessity'), rather than

the 'Conceptual Razor' ('Do not invoke concepts without necessity') (2015: 649). On

the strength of the analogy above, Schaffer infers that derivative entities incur no

further cost to ontological parsimony, relative to the cost of the fundamentals. That

is, we should measure ontological parsimony with the Laser rather than the Razor.

I will add just one objection to this argument to those discussed elsewhere:118

Schaffer's argument from analogy should not persuade anyone who doesn't already

think that the Laser supplants the Razor.

First, why are theories no worse off for multiplying defined concepts? I submit that

part of the reason is that multiplying defined concepts incurs no further ontological

costs. The basic rationale behind the preference for ontological parsimony is this: the

more requirements a theory imposes on reality, the greater the risk that reality won't

comply, hence that the theory is false. But multiplying defined concepts doesn't

increase the risk of falsehood in this way. Suppose we already have the concepts

female and fox and go on to define the concept vixen. Then, by our lights, claims

about female foxes and claims about vixens impose the same requirements ontology.

If vixen-hood and female fox-hood are one and the same, it surely takes no more

ontology to ensure the truth of 'There are vixens' true than it takes to ensure the truth

of 'There are female foxes’. Plausibly, then, the defined concept doesn't increase our

risk of error at least partly because it doesn’t increase the risk that reality won’t

comply with our theory’s requirements.

118
See references in fn. 57.
208
 Now we’ve seen that in Schaffer's inflationary framework, derivative entities are

'genuinely new and distinct entities'— full-blooded ontological additions to their

fundamental grounds (Schaffer 2015: 647). Given this, there is a relevant disanalogy

between multiplying derivative entities and multiplying defined concepts: the former

imposes additional requirements on ontology, but the latter does not. We might agree

that theories are no less likely to be true for multiplying defined concepts. But since

multiplying defined concepts does not impose additional requirements on ontology,

why should this persuade us that theories are no less likely to be true for multiplying

derivative entities? It will only do so, it seems to me, if we’ve already accepted that

the Laser replaces the Razor. If you’re not already convinced that multiplying

requirements on non-fundamental ontology doesn’t increase a theory’s risk of

falsehood, the analogy with conceptual parsimony shouldn’t convince you otherwise.

Bang for the Buck

Schaffer (2015: 651-653) gives another argument based on the analogy between

conceptual and ontological parsimony. He points out that having a small stock of

primitive concepts is not a theoretical virtue in itself. Rather, we value theories that

use a relatively small stock of primitives to define a relatively large number of further

concepts. Hamsa's axiomatisation of set theory, for example, is impressive because it

uses a single primitive to define 99 further concepts, not simply because it has a single

primitive (Schaffer 2015: 651). Schaffer takes this to motivate the following

principle, integrating the virtues of conceptual parsimony and fruitfulness (2015:

652):

(Conceptual Bang for the Buck) Optimally balance minimisation

of primitive concepts with maximisation of defined concepts
(especially useful ones).

209
On the strength of the analogy between defined concepts and derivative entities,

Schaffer proposes an analogous principle integrating ontological parsimony and

fruitfulness:

(Ontological Bang for the Buck) Optimally balance minimisation

of fundamental entities with maximisation of derivative entities
(especially useful ones).

Schaffer then argues that, if Ontological Bang for the Buck is the underlying principle

underlying the integration of ontological economy and fruitfulness, it underlies the

Laser and not the Razor (2015: 653):

Ontological Bang for the Buck pressures one to minimise

fundamental entities (that's the buck), and thus supports a specific
methodological injunction not to multiply such fundamental entities
where possible. It supports no methodological injunction against
derivative entities, but actually favours the generation of derivative
entities (that's the bang).

The 'especially useful ones' clause in Ontological Bang for the Buck bears a lot of

weight in this argument— why not only useful ones? (cf. Fiddaman and Rodriguez-

Pereyra 2018: 348). Schaffer says that, much as defined concepts show that a stock

of primitive concepts is fruitful, '[d]erivative entities are part of what makes a package

of fundamental entities fruitful. They show that these fundamental entities can be used

to produce something' (2015: 652). But merely producing 'something' is surely not

what makes a theory fruitful; what is produced should do some explanatory work.

Given this, one might think, the principle integrating ontological parsimony and

fruitfulness is actually this:

(Ontological Bang for the Buck*) Optimally balance minimisation

of fundamental entities with maximisation of useful derivative
entities.
210
Now, the Razor has no issue with useful derivative posits: if a derivative posit is

needed to do some explanatory work, then it isn't posited without necessity. So it

seems that one could adopt the revised Ontological Bang for the Buck* without

adopting the Laser. Schaffer's argument therefore hinges on our accepting the

unmodified principle— with its 'especially useful ones' clause— rather than the

modified version. But we can't do this unless we already think that the Laser supplants

the Razor. For we'd need to assume that it is permissible— indeed, to some extent

virtuous— to derive further entities from our fundamental posits, even if the

derivative entities aren't useful. And if we assume this, we're assuming that we can

permissibly multiply derivative entities without necessity. So Schaffer's argument,

like the last one, implicitly assumes that the Laser supplants the Razor.

Necessitation

Bennett (2017: 223) argues that derivative entities incur no additional costs relative

to their fundamental grounds, because the existence of the fundamental entities

necessitates that of the derivatives. Suppose theories T- and T+ both posit the same

fundamental entities: the Fs. But according to T+, there are also some derivative

entities, the Ds, grounded by the Fs; according to T-, there are no such Ds. By T+'s

lights, assuming that fundamental entities necessitate the existence of what they

ground, the existence of the Fs entails the existence of the Ds. And the following is a

theorem of basic probability theory:

If A ® B, then P(A) = P(A&B).

211
So by T+'s lights, Bennett argues, the probability that the Fs exist is equal to the

probability that both the Fs and the Ds exist. She concludes that 'T+ is exactly as

likely as T-. Its extra ontology does not make it less likely to be true' (2017: 223).

There's something fishy about this. Suppose you and I agree that there are bats. I

say that there are also vampires, since any bat is necessarily a disguised vampire; you

say that there are no such things as vampires. By Bennett's reasoning, my theory is

exactly as likely as yours. For by my lights, the existence of bats entails the existence

of vampires. The probability that there are both bats and vampires is thus equal to the

probability that there are only bats, and so my theory's extra posits—vampires— 'do

not tell against its simplicity in a way that makes it less likely to be true' (Bennett

2017: 223).

Here's the problem. The claim that's supposed to make vampires cost-free on my

theory— that any bat is necessarily a disguised vampire— is itself part of my theory.

You should only infer that my theory is no less likely to be true for positing vampires

if you accept what my theory entails about its own probability: if any bat is necessarily

a disguised vampire, then the probability that there are bats and vampires is equal to

the probability that there are bats. But a third party comparing our theories can't grant

that any bat is necessarily a disguised vampire without begging the question in my

favour. That would be to assume that, even if our shared claim that there are bats is

true, your claim that there are no vampires is false. Compare the case of T- and T+.

On the assumption that the existence of the Fs necessitates that of the Ds, the

probability that both the Fs and the Ds exist is indeed equal to the probability that the

Fs exist. But to make that assumption would be to rule out T-, since T- says that the

Fs exist but the Ds do not. So if we want to fairly compare T- and T+ for parsimony,

we can’t grant the necessitation claim that supposedly makes the Ds cost-free on T+.

212
Given Bennett’s argument, we can only deem derivative posits cost-free by stacking

the deck in favour of the theories that posit them.119

6.2.2 Objections to the Laser

Having rebutted the main arguments for the Laser, I will now outline two objections

to it: the Special Sciences objection and the Overgeneration objection.120

Special Sciences

Parsimony seems to constrain theory choice in the special sciences. It has been

argued, for example, that considerations of parsimony played a role in Avogadro's

theory of the molecular composition of gases (Nolan 1997) and in the nineteenth-

century debate between rival theories of biogeographical dispersal (Baker 2007).121

But disciplines like chemistry and biogeography deal with paradigmatically non-

fundamental entities, raising the concern that the Laser offers no guidance in the

special sciences. Not only does this seem revisionary, it arguably undermines a major

reason for philosophers to care about parsimony in the first place: the idea that

parsimony is a general constraint on scientific methodology, which philosophers

should strive to emulate.

A potential reply, ventured by Schaffer (2015: 661) and Bennett (2017: 229), is

that parsimony isn't really a going concern in the special sciences. The thought is that,

when chemists and biologists sound like they're appealing to parsimony, they're really

appealing to a kind of 'inference to the simplest explanation but not one involving

119
Note that Schaffer (2010b) rejects grounding necessitarianism and is therefore unlikely to
endorse this argument for the Laser.
120
This sub-section draws on Fiddaman and Rodriguez-Pereyra (2018). See there for further
details.
121
See Baron and Tallant (2018) for discussion of these cases in relation to the Laser.
213
concerns about the multiplication of entities' (Schaffer 2015: 661). Here's Bennett

(2017: 228):

The psychologist and biologist whose theory has fewer memory

systems or species might say that her theory is simpler. But what she
means is that her theory has simpler explanations, not that it has a
smaller ontology. The problem with her opponents' views […] isn't
about ontology at all. It's about complexity of explanation.

Unfortunately, Schaffer and Bennett don't argue for these speculations; nor do they

engage with any of the actual cases, like the two mentioned above, that some claim

to show the relevance of parsimony in the special sciences.

A more promising reply is to claim that multiplying non-fundamental entities, such

as chemicals or organisms, multiplies implicit costs at the fundamental level. Thus,

Schaffer says that 'derivative entities impose an indirect cost, in terms of whatever

fundamental entities serve as their grounds'; for example, the Laser might penalise a

theory for positing telekinetic powers on the grounds that this would require novel

fundamental physical forces (Schaffer 2015: 662). Similarly, Bennett points out that

'a world just like ours but also containing the Loch Ness monster has more

fundamental bits too: it has whatever composes, constitutes, or otherwise builds the

monster' (Bennett 2017: 227).

This reply might work, but we'd need to know more about how theories incur

implicit fundamental commitments via their non-fundamental posits. Neither

Schaffer nor Bennett say much about this. Schaffer's wording—'derivative entities

impose an indirect cost, in terms of whatever fundamental entities serve as their

grounds'— suggests that the implicit costs of derivative posits are those imposed by

their actual fundamental grounds. Apart from only letting us ascribe implicit costs to

true theories, this suggestion further undermines Schaffer's Feng and Esther

214
argument. For suppose Feng's theory is true, and so Esther's 100 particles are in fact

grounded in Feng's 10 strings. Then if derivative posits incur indirect costs in terms

of their actual fundamental grounds, the Laser tells us that Esther's theory was just as

parsimonious as Feng's all along, since her implicit ontological costs are just those

imposed by the strings that in fact ground her particles. Bennett's wording suggests

something like an entailment-based account of implicit commitment: positing the

Loch Ness monster incurs an implicit cost in terms of certain fundamental Fs because,

necessarily, if the Loch Ness monster exists, it is grounded by the Fs. But entailment-

based accounts of implicit commitment face familiar difficulties, which would carry

over to this analysis of implicit cost— for example, metaphysically impossible

theories will be maximally parsimonious (see Fiddaman and Rodriguez-Pereyra

2018: 353-354; cf. §5.2). Until friends of the Laser offer a convincing account of how

implicit costs are incurred, the Special Sciences objection will still be a concern.

Overgeneration

If the Laser supplants the Razor, then any difference in the number of derivative

entities posited by rival theories shouldn't affect whether either theory is preferable

on grounds of parsimony. An apparent consequence of this is that theories can't be

penalised for spuriously overgenerating derivative entities, making it hard to say

what's wrong with certain dubious-sounding theories. To use Schaffer's example,

consider doubled mereology (2015: 656-657). Doubled mereology is a non-

extensional rival to classical mereology with a 'mirroring' function: for any plurality

of simples, s1, …, sn, there are two numerically distinct composite objects, c and

mirror-c, with s1, …, sn as proper parts. Doubled mereology sounds like a bad

theory— one we should be able to criticise on parsimony grounds. But if the Laser

215
supplants the Razor, it looks like doubled mereology is just as parsimonious as

classical mereology, since both have the same fundamental simples.

Schaffer's reply is to stress that his Bang for the Buck principle tells us to maximise

derivative entities, especially useful ones, pointing out that the doubled mereologist's

mirror sums haven't been put to any use (2015: 657). But this doesn't seem to get to

the heart of what's wrong with doubled mereology. Schaffer's Bang for the Buck

principle implies that the extra mirror sums are not just harmless, but are to some

extent a virtue of doubled mereology— it would be better if mirror sums were useful,

but they still give us more bang for the same buck. Intuitively, however, doubled

mereology is a bad theory for positing these useless extra sums; it's not a good theory

that would be even better if the sums were useful. Schaffer's Bang for the Buck

principle gets this wrong. But doubled mereology could be criticised for flouting the

modified Bang for the Buck principle, which tells us to maximise derivative entities

only insofar as they're useful. As we saw, however, the modified Bang for the Buck

principle is consistent with the vanilla Razor.122

6.3 The Truthmaker Razor

For Truthmaker Ontologists, ontology doesn't consist of what can truly be said to

exist, but only the proper subset thereof that makes a truthmaking contribution— what

really exists. Accordingly, the ontological commitments of a theory are its

requirements on truthmakers. And ontological parsimony is a measure of the number

of truthmakers a theory requires. Given this, Truthmaker Ontologists can accept

122
Schaffer has a back-up reply, which is to claim that doubled mereology has other, non-
parsimony-based defects, namely, that it would require an inelegant axiomatisation (2015:
657). But see Turner (2016: 383-384) for doubts about this.
216
(something close to) the standard view that the more ontological commitments a

theory has, the less ontologically parsimonious it is:

(OP) T1 is more ontologically parsimonious than T2 iff T1 has

fewer ontological commitments than T2.

But given that ontological commitments are requirements on truthmakers, (OP) has

non-standard implications. It tells us that ontological parsimony only concerns the

truthmakers required by theories; and since not every existence claim must be made

true by what is claimed to exist, not every entity posited by a theory increases its

ontological costs. The appropriate rule for assessing ontological parsimony, then, is

what I’ll call the Truthmaker Razor:

(Truthmaker Razor) Do not multiply truthmakers without

necessity.

Now for some details. In the last chapter, I suggested that a theory's ontological

commitments (i.e., its requirements on truthmakers) are the entities one rationally

ought to admit as truthmakers, given that one accepts the theory. This is the

substantive claim underlying what I called the Normative Truthmaker View of

ontological commitment (§5.3):

(TOCN) (i) For any T, with admissible TM-extensions T+1,

T+2, …, T+n, T is ontologically committed to there being

either the things to which T+1 is explicitly ontologically

committed, or the things to which T+2 is explicitly

ontologically committed, …, or the things to which T+n

is explicitly ontologically committed.

217
 (ii) T is ontologically committed to a/Fs iff every

admissible TM-extension of T is explicitly

ontologically committed to a/Fs.

(To briefly recap: T is explicitly ontologically committed to Fs iff T contains a

sentence designating Fs as truthmakers or as elements of truthmakers. A TM-

extension is an enriched version of theory containing sentences explicitly designating

entities as truthmakers for the theory and specifying any intermediate grounding

relations where the designated truthmakers are indirect. An admissible TM-extension

is a TM-extension on which the designated truthmakers are rationally acceptable

candidates for making the extended theory true, in light of what the theory says and

certain constraints stemming from the nature of truthmaking and grounding. See §5.3

for details.)

Given that (TOCN) is our view of ontological commitment, how should we

understand ontological parsimony? I suggest the following:

(TOP) T1 is more ontologically parsimonious than T2 iff every

admissible TM-extension of T1 has fewer explicit

ontological commitments than every admissible TM-

extension of T2.123

If T+ is an admissible TM-extension of T, then the explicit ontological commitments

of T+ are candidate truthmakers for the propositions T expresses. So what (TOP) tells

123
We may wish to distinguish between quantitative and qualitative versions of (TOP): T1 is
more quantitatively ontologically parsimonious than T2 iff every admissible TM-extension of
T1 is explicitly committed to fewer individuals than every admissible TM-extension of T2;
T1 is more qualitatively ontologically parsimonious than T2 if every admissible TM-extension
of T1 is explicitly committed to fewer kinds than every admissible TM-extension of T2.
218
us is that T1 is more ontologically parsimonious than T2 iff every candidate way of

making T1 true involves fewer truthmakers than every candidate way of making T2

true.

Here's a toy illustration (cf. Cameron 2019; §5.3). Consider the following pair of

theories:

B {'Ball is red'}

BC {'Ball is red', 'Cube is blue'}

Suppose B has exactly two admissible TM-extensions, B+1 and B+2, which incur the

following sets of explicit ontological commitments:

B+1 {Ball, Redness, The state of affairs of Ball's

instantiating Redness}

B+2 {Ball, Ball's redness trope}

Suppose BC also has exactly two admissible TM-extensions, BC+1 and BC+2, which

incur the following sets of explicit ontological commitments:

BC+1 {Ball, Redness, The state of affairs of Ball's

instantiating Redness, Cube, Blueness, The state of

affairs of Cube's instantiating Blueness}

BC+2 {Ball, Ball's redness trope, Cube, Cube's blueness

trope}

219
By (TOP), B is more ontologically parsimonious than BC: every candidate way of

making B true involves fewer truthmakers than every candidate way of making BC

true.

What does this tell us about how likely B and BC are to be true? We can put it in

terms of basic probability. In a nutshell, the probability that B is made true is the

probability that either the truthmakers of B+1 or the truthmakers of B+2 exist; the

probability that BC is made true is the probability that either the truthmakers of BC+1

or the truthmakers of BC+2 exist. Now, not all the truthmakers on the lists above are

probabilistically independent. For instance, the existence of Ball's redness trope

plausible entails the existence of Ball. If so:

P(Ball exists & Ball's redness trope exists) = P(Ball's redness trope

exists)

(And the same goes, mutatis mutandis, for Cube and Cube's blueness trope.)

Likewise, the existence of the state of affairs of Cube's instantiating Blueness

plausibly entails the existence of Cube and Blueness. If so:

P(Cube exists & Blueness exists & The state of affairs of Cube's
instantiating Blueness exists) = P(The state of affairs of Cube's
instantiating Blueness exists)

(And the same goes, mutatis mutandis, for Ball, Redness, and the state of affairs of

Ball's instantiating Redness.) But I take it that the candidate truthmakers for ⟨Ball is

red⟩ are probabilistically independent of the candidate truthmakers for ⟨Cube is blue⟩,

since one might be made true without the other. For example, the existence of Ball's

220
redness trope and the existence of Cube's blueness trope are mutually independent.

On these assumptions:

R(B is made true) = P(The state of affairs of Ball's instantiating

Redness exists or Ball's redness trope exists)

And the probability that BC is made true is equal to:

R(BC is made true) = P([The state of affairs of Ball's instantiating

Redness exists & The state of affairs of Cube's instantiating

Blueness exists] or [Ball's redness trope exists & Cube's blueness

trope exists])

Now, given that A and B are mutually independent, R (A) > R (A&B). And R (A or B)

> R ([A & C] or [B & D]). Hence, all else being equal, B is more likely to be made

true than BC. More generally, if every candidate way of making theory T1 true

involves fewer mutually independent truthmakers than every candidate way of

making theory T2 true, then (all else equal) T1 is more likely to be made true than T2.

(TOP) therefore tallies with the basic rationale behind ontological parsimony given

at the outset: the more requirements a theory imposes on ontology, the less likely

ontology is to comply, and so choosing more ontologically parsimonious theories

diminishes our risk of error. The main difference is that 'ontology' is taken to consist

solely of what makes a truthmaking contribution, so the risk we're trying to diminish

is imposing demands on truthmakers that aren't met.

221
6.4 Contrast with the Laser

Like the Laser, the Truthmaker Razor allows us to say that not everything a theory

claims to exist counts against it by the measure of ontological parsimony. But the

reasons underlying this are quite different. The Laser's inflationist proponents want

to drive a wedge between the ontological commitments of a theory and the costs of

those commitments. They see requirements on the non-fundamental as full-blooded

requirements on ontology but claim that multiplying such requirements is somehow

innocuous. In contrast, the Truthmaker Razor involves no distinction between

commitments and costs: all ontological commitments are ontological costs, but not

all existence claims incur ontological commitment to what is said to exist. If a theory

affirms 'Fs exist', but Fs aren't needed to make 'Fs exist' true, then Fs don't count

towards the theory's ontological costs; the costs are just those incurred by whatever it

takes to make 'Fs exist' true. Fs are not, in this case, a 'commitment without a cost',

but aren't ontological commitments at all: the theory is no less ontologically

parsimonious for positing Fs, because ontology needn't contain any Fs for the theory

to be true.

So the Truthmaker Razor rests on an unfamiliar picture of ontology and

ontological commitment; but I've argued in previous chapters that this picture is

coherent. And once we understand this picture, I think the Truthmaker Razor offers

a better motivated form of 'Razor revision' than the inflationary Laser.

Following Turner (2016)— a critic of the Laser— consider Argle, the protagonist

of Lewis and Lewis (1970). On Turner's reading, Argle thinks that there are no holes,

but that the sentence 'There is a hole in my cheese' is true; 'That's because Argle

endorses a fancy semantics according to which, when "a hole" is attached to "there

is", it doesn't act like a quantifier but instead a predicate which means "is perforated"'

222
(Turner 2016: 378). Now, friends of the Laser say that the 'grounded' parts of theories

don't count by the measure of ontological parsimony. Turner thinks this has some

plausibility, if 'Fs ground Gs' means something like what Argle means when he

claims that 'hole-free' states account for the truth of hole-talk:

When we calculate Argle's ontology, we don't include holes. And we

do this for the obvious reason that, in some very real sense Argle
thinks there aren't holes— even though he assents to 'there is a hole
in my cheese'. Hole talk is, for Argle, just talk; and its being just talk
shows up in the way existential claims about holes are grounded in
entirely hole-free states. (Turner 2016: 381).

But this sense of 'grounding' is not what the Laser's inflationist proponents have in

mind. By their lights, the sort of 'grounding' that's supposed to mitigate the costs of

non-fundamental posits is a kind of ontological dependence relation between 'full-

blown "heavyweight"' denizens of reality (Schaffer 2009: 360). So if a friend of the

Laser tells us that holes are grounded in hole-free states, we do have to include holes

when we calculate their ontology. Their task is to persuade us that, despite being

genuine ontological additions, 'grounded' holes are irrelevant to ontological

parsimony; as we've seen, the have so far failed to do this.

But Truthmaker Ontology offers something like an Argle-style attitude towards

hole-talk. The main difference is that the Truthmaker Ontologist doesn't appeal to any

'fancy semantics'. It might be that the 'There is a hole in my cheese' is true— literally

and without qualification— even if reality is hole-free. Given this, we should exclude

holes from the Truthmaker Ontologist's ontology for the same reason Turner thinks

we should exclude them from Argle's ontology: in a very real sense, the Truthmaker

Ontologist thinks there aren't holes and that hole-talk is just talk. Because it doesn't

take an ontology of holes to make hole-talk true, we can truly represent the existence

of holes without admitting them to our ontology. There is thus a clear and natural
223
sense in which holes shouldn't count when we're assessing the Truthmaker

Ontologist's view for ontological parsimony.

Another way to bring out the difference between the Laser and the Truthmaker

Razor is to consider Schaffer's argument from the analogy with conceptual

parsimony. Schaffer is right to say that we don't make a theory worse off by defining

additional concepts from its primitives. But part of the reason for this, I said, is that

multiplying defined concepts doesn't result in additional ontological requirements: if

we use 'female fox' to define 'vixen' then we can affirm 'Vixens exist' as well as

'Female foxes' exist without imposing extra requirements on ontology. This is why

the analogy fails to support the Laser. For by the inflationist's lights, we do impose

extra requirements on ontology when we posit additional non-fundamentals. So the

analogy between defined concepts and derivative entities won't persuade us to accept

the Laser unless we're already convinced that derivative entities are cost-free.

The Truthmaker Razor, however, is supported by an analogy between multiplying

defined concepts and multiplying 'derivative entities' (understood as entities posited

by a theory but which are not among the truthmakers to which it is ontologically

committed). Suppose the Truthmaker Ontologist says that simples arranged table-

wise are all it takes to make it true that tables exist. Then what they're claiming is that

the same portion of reality, the simples, is enough to make true the sentence 'Tables

exist', as well as the sentence 'Simples arranged table-wise exist'. Whatever ontology

is enough for one sentence to be true is ontology enough for both sentences to be true.

In this respect, multiplying 'derivative entities' is analogous to multiplying defined

concepts, in terms of the effect on parsimony: we're multiplying ways of truly

representing the same portion of ontology, as it were, without multiplying the

ontology itself. So if you agree that defined concepts don't make theories worse off

224
partly because they don't create new ontological demands, you should think the same

about the Truthmaker Ontologist's 'derivative entities'.

6.3.3 The Objections Revisited

I will now argue that the Truthmaker Razor avoids the objections that afflict

Schaffer’s Laser.

According to the Special Sciences objection, the Laser offers no methodological

guidance to disciplines such as biology and chemistry, which are concerned with

paradigmatically non-fundamental entities. In reply, friends of the Laser suggest that

non-fundamental posits incur implicit commitments, hence costs, at the fundamental

level, but have yet to give a clear account of how this works.

The Truthmaker Razor is well-equipped to handle implicit commitments— the

view of ontological commitment on which it rests construes most ontological

commitments as implicit. Consider two theories of biological development. G posits

a single kind of entity: genes. GE posits two kinds of entity: genes and élan vital.

Thus, among other things, G and GE affirm the following:

G {'Genes exist'}

GE {'Genes exist', 'Elan vital exists'}

Prima facie, G is more parsimonious than GE. The Truthmaker Razor confirms this.

By (TOP), G is more parsimonious than GE if every admissible TM-extension of G

posits fewer truthmakers than every admissible TM-extension of GE. Now

presumably, ⟨Genes exist⟩ and ⟨Elan vital exists⟩ are mutually independent, in the

sense that one could be made true but not the other. Thus, every admissible TM-

extension of G will need to posit truthmakers for ⟨Genes exist⟩ and every admissible

225
TM-extension of GE will need to posit truthmakers for ⟨Genes exist⟩ and truthmakers

for ⟨Elan vital exists⟩. Hence, by (TOP), G is more ontologically parsimonious than

GE. In general, if T1 includes the claims P1, …, Pn and T2 includes the claims P1,

…, Pn & Q, for some mutually independent Q, T1 will be more ontologically

parsimonious than T2, regardless of what the theories are about. In this way, there is

no reason why the Truthmaker Razor couldn't guide theory choice in the special

sciences or elsewhere.

What about the Overgeneration Objection? To return to the original example,

consider a disagreement between a classical mereologist and a doubled mereologist.

According to the classical mereologist, there is a composite object c. According to

the doubled mereologist, there are two composite objects: c and mirror-c. Thus, the

two mereologists affirm the following:

CM {'c exists'}

DM {'c exists', 'Mirror-c exists'}

According to (TOP), CM is more ontologically parsimonious than DM only if every

admissible TM-extension of CM posits fewer truthmakers than every admissible TM-

extension of DM. But it's plausible to think that truths about complex objects might

be made true indirectly by their simple parts. And according to the doubled

mereologist, c and mirror-c both have the same simple parts—s1, …, sn— that c has

according to the classical mereologist. So perhaps there's an admissible TM-extension

of CM, CM+, on which s1, …, sn indirectly make true ⟨c exists⟩ and an admissible

TM-extension of DM, CM+, on which s1, …, sn indirectly make true both ⟨c exists⟩

and ⟨Mirror-c exists⟩. CM+ and DM+ posit exactly the same truthmakers. Hence, if

both are admissible TM-extensions of their respective theories, it is not the case that

226
every admissible TM-extension of CM posits fewer truthmakers than every

admissible TM-extension of DM, so (TOP) tells us that they're equally ontologically

parsimonious.

I think this is actually the right result. By my lights, CM and DM don't differ with

respect to their requirements on ontology. It is therefore natural to think that they

shouldn't differ with respect to ontological parsimony. (In contrast, friends of the

Laser are committed to saying that CM and DM are equally ontologically

parsimonious despite differing with respect to their ontological requirements.) But

how can we then explain why DM seems like the worse theory? I submit that it's

perfectly reasonable for the Truthmaker Ontologist to deem DM the worse theory

because of its superfluous posits— it's just that these aren't ontological posits and so

aren't relevant to ontological parsimony. Recall Schaffer's Ontological Bang for the

Buck principle:

(Ontological Bang for the Buck) Optimally balance minimisation of

fundamental entities with maximisation of derivative entities (especially

useful ones).

I argued above that Ontological Bang for the Buck presupposes the Laser rather than

supports it, because the 'especially useful ones' clause implies that it's permissible to

multiply useless derivative entities. Why shouldn't the relevant principle tell us to

maximise only useful derivative entities, which seems consistent with the standard

Razor? Truthmaker Ontologists, however, can adopt the following variant of the

principle:

(Truthmaker Bang for the Buck) Optimally balance minimisation of

truthmakers with maximisation of useful truths.

227
This seems like a sensible policy: account for as many plausible or explanatorily

useful truths as possible on the slenderest ontological basis possible. However, truths

should not be multiplied without good reason. Here are some truisms: (i) There are

cars; (ii) There are garages. Suppose a Truthmaker Ontologist claims that all it takes

to make these truisms true are simples arranged car-wise and garage-wise

respectively. Then another comes along and claims that, with only minor adjustments,

we can get more 'bang for the buck' out of these truthmakers. For they could also be

taken to make it true that there are incars (cars in a garage) and outcars (cars outside

of a garage) (Hirsch 1982: 32). The second Truthmaker Ontologist might insist that

there's no harm in positing these extra entities— after all, we're not making any extra

demands on ontology. But one wonders: what's the point?'. Even if we're not adding

to our ontology, why add the concepts incar and outcar to our ideology if we can get

along well enough without them?

In this spirit, we could criticise DM for flouting the Truthmaker Bang for the Buck,

even if it is on a par with CM with respect to ontological parsimony. The mirror-

fusions of DM may not impose any additional requirements on ontology; but they do

add to our ideology in needless and explanatorily irrelevant ways. DM is a bad theory,

not because it makes extravagant demands on ontology, but because it uses that

ontology to pointlessly generate useless truths proclaiming the existence of mirrored

fusions. Inflationist friends of the Laser can’t draw the same distinction: by their

lights, the mirrored fusions are an addition to the doubled mereologist’s ontology, yet

somehow don’t diminish its parsimony. And, as we’ve seen, the arguments for that

claim are inconclusive.

228
Conclusion

If we accept Truthmaker Ontology, we need to revise standard ideas about the

measure of ontological parsimony: not everything that exists according to a theory

increases its ontological costs. This has its benefits. It means that, for example, the

Truthmaker Ontologist can claim the parsimony benefits of compositional nihilism

without having to reject familiar truisms concerning the existence of tables, planets,

organisms, and all the rest. Inflationary Stratified Ontologists claim similar benefits

but do so in a different way. They hold that, even though derivative entities are

genuine additions to ontology, they are somehow irrelevant to the measure of

ontological parsimony. I’ve argued that the case for this claim is unconvincing.

However, I’ve argued that Truthmaker Ontology does have a viable measure of

ontological parsimony. The Truthmaker Razor can account for the relevance of

ontological parsimony in the special sciences and doesn’t permit us to spuriously

overgenerate non-fundamental entities. However, it also gives us the desired result:

not everything that exists according to a theory is an ontological cost.

229
7 The Case for Truthmaker Ontology
I began by arguing that, if ontology is to be a serious enterprise, we must part ways

with the orthodox view that ontology concerns what there is. I then set about

developing Truthmaker Ontology as an alternative, which offers a cogent heterodox

picture of the subject matter and method of ontology. In this chapter, I consolidate

the case for Truthmaker Ontology, outlining its advantages over Orthodox Ontology

and arguing that it has attractive implications for first-order ontological debate.

In §7.1, I summarise the approach I have developed, highlighting some differences

with other versions of Truthmaker Ontology. I then outline and defend the main

advantages of Truthmaker Ontology, which I group under two broad headings:

keeping ontology serious (§7.2) and reconciling modesty and sparsity (§7.3). Finally,

in §7.4, I defend Truthmaker Ontology against some remaining potential objections.

7.1 Ontology as a Search for Truthmakers

Truthmaker Ontology uses the notion of truthmaking as the basis of a metaontological

approach on which ontological questions come apart from questions about what there

is— a necessity, I’ve argued, if ontology is to be a serious enterprise. The idea that

ontology should be construed as a search for truthmakers is not new and has several

contemporary advocates. I hope, however, to have put Truthmaker Ontology on a

firmer footing by clarifying and developing its conceptual and methodological

foundations.

One of my aims was to clarify the basis of the crucial distinction between ontology

and what can truly be said to exist. Some Truthmaker Ontologists draw this

distinction by appealing to a truthmaker criterion of ontological commitment. But we

should be careful not to conflate the notion of what theories are ontologically

230
committed to—i.e. what requirements their truth imposes on ontology— with the

notion of what has ontological status in the first place. In my view, the latter notion

takes priority, and I have therefore taken the central axiom of Truthmaker Ontology

to be a truthmaker criterion of ontological status, (TOS): what counts as real, in the

sense of serious ontological interest, are all and only those things that make a

truthmaking contribution.

A second crucial claim is the Divergence Thesis, according to which the

truthmakers for a statement can differ from what is said or implied to exist by that

statement. Several Truthmaker Ontologists endorse the Divergence Thesis, but aren’t

always clear about when and why the alleged divergence occurs. I suggested that this

is one of several areas in which it is useful to have, alongside truthmaking, a distinct

relation of (truth-)grounding: given the Grounding Principle, ⟨Fs exist⟩ might be

made true by non-Fs indirectly via intermediate grounding relations (see §3.3).

(TOS) and the Divergence Thesis combine to allow ontological status to come

apart from existential truth. ⟨Fs exist⟩ might be made true even if Fs make no

truthmaking contribution and hence lack ontological status; so an F-less ontology

might support the literal truth of representations concerning the existence of Fs.

Truthmaker Ontology thus offers a coherent way to make sense of the distinction

between what merely and what really exists. The real existents are the things out there

in the world making a truthmaking contribution; the mere existents can truly be said

to exist but have no ontological status— out there in the ready-made world, there are

really no such things. So, contrary to orthodoxy, ontology is a question of what really

exists; and this might be a proper subset of what can truly be said to exist.

Some Truthmaker Ontologists use the term fundamentally exists as I use really

exists, or treat the two as synonyms (Cameron 2008a, Rettler 2016). I think it is useful

to keep these two notions separate. 'Really exists' is an expressive device used to mark
231
off what has ontological status (i.e. makes a truthmaking contribution) from what can

merely be truly said to exist. ‘Fundamental' pertains to entities and/or truths which

are in some sense basic and integral. So in principle, fundamentality may or may not

coincide with real existence. I've suggested that Truthmaker Ontologists can use the

notion of fundamentality as a constraint on what must be taken to really exist. In

Chapter 4, I proposed a theory of fundamentality on which an abundant, structured

hierarchy of fundamental and non-fundamental truths is supported by a perhaps

sparse foundation of fundamental truthmakers; the fundamental truthmakers are all

the ontology needed to make the true description of the world true, completely and

without redundancy. This leads to a more refined conception of what we must take to

make a truthmaking contribution: one could give a complete, correct and non-

redundant answer to the ontological question by specifying a minimally complete

truthmaking base for the fundamental truths.

The notion of ontological commitment is not as central in my account as it is for

other Truthmaker Ontologists, but nevertheless plays an important methodological

role. On what I called the Normative Truthmaker View, the ontological commitments

of a theory are the things one rationally ought to admit as truthmakers, given that one

accepts the theory. The idea of an admissible TM-extension brings this into sharper

focus. The truthmakers designated by an admissible TM-extension of a theory are

rationally acceptable candidates for making that theory true, given what the theory

says; a theory thus rationally obliges its holders to accept one or other of its sets of

candidate truthmakers (and to accept any truthmakers that are common to all sets).

The ontological parsimony of a theory can be understood as a measure of its

truthmaker-commitments: if every candidate way of making theory T1 true involves

fewer truthmakers than every candidate way of making T2 true, then T1 is the more

parsimonious.

232
 The idea of a best TM-extension forms the basis of a (somewhat idealised) three-

step procedure for determining the ontology we ought to accept. First, we decide what

we ought to accept as true. Second, we decide on the most rational and plausible

account of how these truths are structured with respect to grounding, and which

should be construed as fundamental. Third, we decide on the most rational and

plausible candidate truthmakers for the fundamental truths. The inventory of

truthmakers thereby determined will be our answer to the ontological question.

So although Truthmaker Ontology is not a new idea, I have developed it in new

ways. My hope is that the new developments prove useful, or at least breathe new life

into the question of what ontology might look like through the lens of truthmaker

theory. But while I have often disagreed with other Truthmaker Ontologists over

details, my basic contention is the same: ontology is not a question of what there is,

but of what makes a truthmaking contribution. The advantages I shall now claim for

Truthmaker Ontology also resemble those often claimed by other advocates. I divide

them into two broad categories: keeping ontology serious and reconciling modesty

and sparsity.

7.2 Keeping Ontology Serious

Serious Ontologists hold that ontology concerns deep and substantive questions about

what is real. Orthodox Ontologists hold that ontology concerns questions about what

there is. I began by arguing that these two views are in tension and that Serious

Ontologists therefore need a better alternative. I will revisit the three challenges raised

in the first chapter to (SOX)— the view that serious ontological questions are

equivalent to questions about what there is— and explain why they have no purchase

on Truthmaker Ontology.

233
7.1.1 Unloaded Quantification

The first challenge to (SOX) was as follows (see §1.2). (SOX) assumes that objectual

existential quantification is ontologically loaded, in the sense that '$xFx' is true iff at

least one F is real. And on two widespread accounts of $— the natural language

account and the model theoretic account— $ is ontologically loaded only if the

English 'There is' is ontologically loaded. But that's questionable: there are many true-

seeming English sentences that quantify over unreal-seeming entities ('There is a

fictional space explorer called Picard', for example). To avoid biting a bullet, (SOX)

gives us the option of claiming that the recalcitrant sentences are amenable to

paraphrase. But to take the paraphrase gambit is to concede that 'There is' is

intermittently ontologically loaded at best, thus undermining the case for the loaded

reading of $.

Truthmaker Ontology avoids this challenge because it rejects the assumption that

existential quantification is by nature ontologically loaded. The truth of 'There are Fs'

doesn't suffice for the reality of Fs because it doesn't suffice for Fs making a

truthmaking contribution. Quantification over Fs suffices for their reality only when

the Fs are being designated as truthmakers; and it is not quantification per se that's

doing the work here, but the fact that the Fs are said to play a truthmaking role (see

§5.3). Thus, Truthmaker Ontology permits true quantification over unreal entities,

preserving our intuitions about cases like 'There is a fictional space explorer called

Picard'. The way to handle such sentences is to claim that the propositions they

express are not made true by what is quantified over; for instance, ⟨There is a fictional

space explorer called Picard⟩ is plausibly made true indirectly by a reality devoid of

fictional space explorers. And there is no need to claim that the original sentence has

a misleading surface form or isn’t put forward literally true. We can say that the

234
sentence expresses the proposition it seems to express and is used to seriously assert

that proposition— it's just that reality needn't host any fictional space explorers in

order to account for that proposition's truth.

More generally, Truthmaker Ontology puts ontological questions at arm's length

from questions about the semantics and pragmatics of English sentences— questions

such as whether some sentence has a misleading surface form, or is typically put

forward with an 'attitude of frivolity' (Quine 1961: 103), or is equivalent in meaning

or functional role to some other sentence. We can leave such questions to the

experts—linguists and philosophers of language— and put our own focus on the

search for truthmakers. In contrast, Orthodox Ontology allows such linguistic

questions to drive our ontological decision-making (cf. Cameron 2008a, 2010b;

Rettler 2016). We find philosophers 'exam[ining] the meaning' of English sentences

in order to read off ontological conclusions from their quantificational structure (van

Inwagen 2004: 114). The prospects for prima facie revisionary views, like

mathematical nominalism and compositional nihilism, are thought to hinge on

whether sentences quantifying over tables or numbers can be successfully

paraphrased into ones that don't.124 (see e.g. Field 1980, Melia 1998, Uzquiano 2004,

van Inwagen 1990). Serious Ontologists, whose aim is to discover the contents of

extra-representational reality, should be sceptical about this state of play: 'serious

ontological questions are being decided by linguistic facts' (Cameron 2008a: 5).

Truthmaker Ontology discourages this practice, which is a general respect in which

it helps to keep ontology serious. As Rettler puts it, we can 'resolve ontological

questions by doing metaphysics, and not by investigating our use of language' (2016:

1414).

124
See, for example, the exchanges between Field (1980) and Melia (1998), and van Inwagen
(1990) and Uzquiano (2004).
235
 Schaffer (2008: 9) objects that the attempt to divorce ontology from language is an

unrealistic goal. After all, Truthmaker Ontologists are still in the business of

determining the ontological implications of the sentences we accept, which must be

'at least in part a function of what these sentences mean'. But I don't deny this. We

might have used the sentence 'Tables exist' so that it means what 'Chairs exist' actually

means and vice versa. If so, then the actual truthmakers for Tables exist' would have

been the truthmakers for 'Chairs exist' and vice versa (cf. Williams 2012: 178). So the

ontological implications of these sentences are a function of their meaning. But the

truthmakers in question might be simples arranged thus-and-so, rather than tables and

chairs. What I deny is that the ontological implications of a sentence are tied to their

surface form, so that 'Fs exist' can only be strictly and literally true if reality hosts Fs.

It is this assumption that drives ontologists towards an excessive preoccupation with

language, but which Truthmaker Ontology rejects.

7.1.2 Easy Ontology

The second challenge to (SOX) was as follows (see §1.3). Questions about what is

real are deep and substantive— they don't have easy answers but demand careful and

difficult philosophical scrutiny. If (SOX) is true, the same should go for questions

about what there is. But it doesn't. For one thing, beliefs in the existence of numbers,

tables, (etc.) are highly epistemically secure, insofar as it's very unlikely that a

philosophical argument could ever make it rational for us to give them up. Moreover,

Thomasson (2015) offers a cogent explanation of why such beliefs seem so secure.

All it takes for the belief that tables exist to be true is for the application conditions

associated with the term 'Table' to be satisfied. And as long as you're conceptually

competent with the term 'Table', you can tell that the relevant conditions are satisfied

through straightforward observation. Plausibly, then, questions about whether tables

236
and other commonsense posits exist just aren't up for serious philosophical debate

and should be answered with a trivial 'Yes'. If so, existence questions have a

fundamentally different character to serious ontological questions about what is real.

Truthmaker Ontology allows us to say that existence questions often are easy, but

that the corresponding ontological questions are not. Perhaps no philosopher's

argument could ever rationally compel us to stop believing that tables exist; and

perhaps all it takes to know that tables exist is to be competent with the term 'Table'

and look in your dining room. But the question of serious ontological interest is not

whether tables exist. It is whether they really exist, i.e. make a truthmaking

contribution. And we can't infer that tables make a truthmaking contribution from the

fact that 'Tables exist' is a true sentence. We can only find out by doing some

metaphysics. On the methodology I've suggested, we determine what ontology we

ought to accept by determining the best candidate set of fundamental truthmakers for

the claims we accept as true (see §5.5). To find out whether tables really exist, we

must decide whether they must number among these truthmakers— a decision that

turns on considerations of what grounds what, what makes what true, and criteria of

theoretical virtue. Thus, even if the existence of tables is a trivial matter, the

ontological question about them remains, and requires distinctly metaphysical

attention.

7.1.3 Metasemantic Tolerance

The third challenge to (SOX) was based on the possibility that we are speaking a

metasemantically tolerant language (see §1.4). Here are two epistemically possible

scenarios:

(Nihilism) Reality hosts no complex objects, only simples

arranged thus-and-so. Speakers are disposed to

237
 assent to (e.g.) 'There is a table' iff, in reality,

there are some simples arranged table-wise.

(Universalism) Reality hosts simples arranged thus-and-so as

well as complex objects composed of those

simples. Speakers are disposed to assent to (e.g.)

'There is a table' iff, in reality, there is some

complex object that is a table.

Whether Nihilism or Universalism obtains, 'There is a table' (and similar sentences)

might be true on the best overall interpretation of English. Thus, consider two groups

of Orthodox Ontologists: the Tablists say that 'There is a table' is true, while the Anti-

Tablists say it's false. The Tablists might be right, even if complex objects are unreal.

Generally, the truth of 'There are Fs' does not reliably track whether Fs are real,

depending to a large extent on conventions of linguistic use.

Truthmaker Ontologists can agree that, owing to conventions of language use,

'There is a table' might be true whether Nihilism or Universalism obtains. They deny,

however, that the ontological question about tables is equivalent to the question of

whether 'There is a table' is true. The question is whether tables make a truthmaking

contribution; and the answer to that question is appropriately sensitive to whether

Nihilism or Universalism obtains. If Universalism obtains, it is natural to think that

'There is a table' expresses a proposition that's made true by a table; if Nihilism

obtains, it is natural to think that 'There is a table' expresses a proposition that's made

true by simples arranged table-wise (cf. Cameron 2010b: 12). So if we take the

ontological question about tables to be whether they make a truthmaking

contribution, the debate is not hostage to linguistic fortune. We can concede that

questions about what can truly be said to exist hinge, to a large extent, on linguistic
238
use; but we can insist that substantive questions remain about what the truthmakers

are.

7.3 Reconciling Modesty and Sparsity

The second broad advantage of Truthmaker Ontology is that it provides attractive

new perspectives on a range of issues in first-order ontology. It does so, in a nutshell,

because it offers a cogent way to reconcile the pressure towards ontological sparsity

with the pressure towards a modest appraisal of ordinary discourse. I will briefly

compare the Truthmaker Ontologist's approach to reconciling these pressures with

some of the other main options, and then consider some specific applications. I will

then discuss two neglected potential challenges to the Truthmaker Ontologist's

reconciliation strategy: the Problem of Consistency and the Problem of Fit.

7.3.1 Reconciliation Strategies

Ontologists often find themselves caught between two powerful theoretical pressures.

On the one hand, there is pressure towards modesty: most of us are reluctant to reject

familiar truisms about numbers, tables, minds, moral values, and all the rest. On the

other hand, there is pressure towards sparsity: excluding familiar entities from our

ontology often seems the best way to avoid nagging philosophical problems and

satisfies the more general Ockhamite urge to pare our ontology down where possible.

It would be nice to have a reconciliation strategy, which allows one to 'have one's

cake and eat it: endorse the attractive package of a radically minimal metaphysics,

while endorsing the […] truth of the corpus of commonsense belief' (Williams 2010:

105).

But on orthodox ontological assumptions, this is a hard trick to pull off. If what

has ontological status is equivalent to what there is, then excluding Fs from our

239
ontology commits us to denying that there are any Fs, hence that truisms quantifying

over Fs are true at face value. The aims of modesty and sparsity thus seem directly at

odds, and it is difficult to resolve the tension in a satisfying way.

One option is to deny that Fs have ontological status (i.e. that there are Fs) while

advancing a revisionary view of what ordinary talk of Fs involves: sentences

quantifying over Fs mislead with respect to logical form, have non-obvious

meanings, are not truth-apt, are typically used figuratively, and so on. The debate

over Fs thus descends into a debate about language. If one of these revisionary

linguistic proposals succeeds, then perhaps we can have our cake and eat it; if not,

and F-sentences must be taken at face value, we must choose between modesty and

sparsity.

Another option is to grant that Fs have ontological status (i.e. that there are Fs)

while downplaying the costs of doing so by claiming that Fs are an ontological free

lunch. But what does this mean? On one view, Fs are an ontological free lunch if they

are 'not ontologically additional' to something we've already admitted to our ontology.

Armstrong, for instance, says that 'whatever supervenes is not something

ontologically additional to the subvenient […]. What supervenes is no addition of

being' (1997: 12, emphasis mine). But if there's no difference between ontological

status and existence, this presents a puzzle125. For then, if Fs and Gs exist, both have

ontological status. So either Fs are identical to Gs or they are ontologically distinct.

But supervenience does not entail identity. So it seems that Fs might supervene on

Gs and yet be ontologically distinct, making them 'something ontologically

additional' in the only discernible sense126. On another view, ontological free lunches

125
Cf. David (2005: 148), Oliver (1996: 31), Melia (2005: 74-75).
126
A similar puzzle arises for Lewis's claim that mereological fusions are ontologically
'nothing over and above' the objects fused (1991: 81-82). Unless the objects are identical to
their fusion— which Lewis denies, except in an analogical sense— the fusion is a distinct
240
are genuine ontological additions but incur no cost to ontological parsimony— this,

we've seen, is popular among inflationary Stratified Ontologists. The thought is that,

while both fundamental and non-fundamental entities have ontological status,

parsimony considerations don't apply to the non-fundamental. Hence, non-

fundamental entities are 'an "ontological free lunch", in the sense that they are

genuinely new and distinct entities but they cost nothing by the measure of economy'

(Schaffer 2015: 647; cf. Bennett 2017, Schulte 2011). I have argued, however, that

there are problems with the claim that parsimony considerations don't apply to non-

fundamental entities, as inflationists understand them.

In contrast to these proposals, Truthmaker Ontology offers a clear and cogent

reconciliation strategy (cf. Cameron 2008a: 7, 2010a: 249; Melia 2005: 74; Rettler

2016: 1413). If ⟨Fs exist⟩ is made true, then Fs exist— it is literally true to say so.

But consistently with that, Fs might make no truthmaking contribution, in which case

they lack ontological status. In that case, there is a natural sense in which Fs are an

'ontological free lunch': it is true that Fs exist, but ontology hosts no Fs. The truth of

familiar claims about Fs is thus consistent with an F-less ontology. If we want to

exclude Fs from our ontology, then, there is no need to advance revisionary linguistic

claims about what F-talk involves: we can say that true statements about Fs mean

what they seem to mean, have the logical forms they seem to have, etc., but are

nonetheless accounted for by an F-less ontology. Nor do we need to cash out the

claim that Fs are 'not ontologically additional' to some other constituents of ontology:

we can say that Fs are not ontologically additional to anything because they have no

existent. So if whatever exists has ontological status, the fusion seems to be ontologically
'over and above' the fused objects in the only sense available.
Of course, some might claim that fusions are an ontological free lunch precisely because
they are strictly identical to the objects fused (see e.g. Baxter 1998). But to quote Cameron
(2008a: 7), 'if many-one identity is the price to pay, then it really does seem that there's no
such thing as a free lunch'
241
ontological status. Nor do we need to claim that we can somehow add Fs to our

ontology without decreasing ontological parsimony: we can say that all additions to

ontology decrease ontological parsimony but not everything that can truly be said to

exist is an addition to ontology.

7.3.2 Applications

So compared to the standard alternatives, Truthmaker Ontology offers a cogent

strategy for reconciling modesty and sparsity. The general recipe is to 'show that an

ontology lacking in Xs can nonetheless make true sentences proclaiming the existence

of, or attributing features to, the Xs' (Cameron 2010a: 250). This has potentially

attractive implications for several live ontological debates.

The debate over complex objects is a good, if well-worn, case in point. Familiar

truisms about tables, organisms, planets, and so on might be made true indirectly by

simples standing in certain arrangements. If so, complex objects are an ontological

free lunch. They can truly be said to exist, but don't really exist: they make no

truthmaking contribution and therefore lack ontological status. Thus, without revising

familiar truisms about complex objects, the Truthmaker Ontologist can claim the

ontological benefits of compositional nihilism. Apart from the obvious parsimony

benefits, we can give a straightforward and systematic answer to the Special

Composition Question (van Inwagen 1990): in reality, there are no conditions under

which some objects compose a further object. We also potentially have some

attractive solutions to some thorny puzzles, such as the Statue and the Clay. This

morning Clay was shaped into Statue. Since Statue and Clay now exactly coincide,

it's natural to think they're identical. But since Statue and Clay have different

properties (e.g. Clay would survive a hammer blow, but Statue would not), Leibniz's

Law entails that they are not identical. What to say? Constitution theorists say that

242
Statue and Clay are non-identical despite exactly coinciding: Clay constitutes but is

nonetheless distinct from Statue. Perdurantists say that Statue and Clay don't exactly

coincide, since Statue is a proper temporal part of Clay. Nihilists, standardly, deny

that Statue or Clay exist: all that exists are some simples which were once arranged

Clay-wise and are now also arranged Statue-wise. Truthmaker Ontologists suggest a

different perspective (see Heil 2003, Barnes and Cameron 2007). Our task is to say

what truthmakers best account for the fact that 'Clay exists' and 'Statue exists' are both

true today, whereas only 'Clay exists' was true yesterday. An attractive option is to

say that, yesterday, some simples were arranged so as to make 'Clay exists' but not

'Statue exists' true; today, they are arranged so as to make 'Statue exists' true as well.

All that really exists, then, are simples arranged thus-and-so; there is no need to admit

distinct but exactly coinciding objects, temporal parts, or even Statue and Clay into

our ontology. Nevertheless, we can affirm what seems obviously true: Statue and

Clay exist.

Abstract objects might be given a similar treatment. Perhaps it doesn't take an

ontology of mathematical abstracta to make it true that there are numbers, for

example. Perhaps it doesn't take any ontology at all: you might agree with Cameron

(2010c) that the truths of (pure) mathematics are necessary, and hence require no

truthmakers since they hold no matter what is real. Or perhaps statements implying

the existence of numbers might be made true by non-numbers: proofs, say, or certain

concrete structures. Cameron (2008a) makes an interesting suggestion drawing on

neo-Fregeanism, according to which the existence of numbers (along with all the

truths of arithmetic) can be derived from the following abstraction principle: the

number of Fs is identical to the number of Gs just in case there is a one-one

correspondence between the Fs and the Gs. Thus, for any Fs and Gs standing in one-

one correspondence, the number of Fs is identical to the number of Gs, from which

243
it follows that there is at least one number. By orthodox lights, this is enough to settle

the ontological question about numbers in the affirmative (arguably somewhat

suspiciously). But Truthmaker Ontologists, Cameron suggests, can treat it as an

opportunity to claim numbers as an ontological free lunch (2008a: 12-13). Perhaps

⟨There is a one-one correspondence between the Fs and the Gs⟩ is made true by the

Fs, the Gs and the fact that nothing else is an F or a G. In that case, given the

abstraction principle, equinumerous collections of entities and 'That's all' facts are

ontology enough to ensure the truth that there are numbers.

The metaphysics of mind is another potential area of application. Truthmaker

Ontologists inclined towards a physicalist ontology might claim that truths about the

mental have physical truthmakers. Perhaps, for example, what makes it true that I am

in pain is the fact that I am in neural state N. On this view, we avoid the need for a

dualistic ontology containing sui generis mental substances. But we can also claim

some prima facie advantages over more familiar forms of physicalism. A reductive

physicalist might claim that being in pain just is being in neural state N. This faces

the familiar problem of multiple realisability: pain can't be identical to N, it seems,

since an octopus might also be in pain but occupy a very different neural state, N*.127

The Truthmaker Ontologist can avoid this worry, pointing out that the same kind of

mental property-ascription can have different truthmakers: perhaps what makes it true

that I am in pain is the fact that I'm in neural state N, whereas what makes it true that

the octopus is in pain is the fact that the octopus is in neural state N*. A non-reductive

physicalism might claim that mental states are irreducible to but in some sense

supervenient on physical states. This leads to concerns about causal exclusion.128 My

pain seems to cause me to swear. But physicalism is thought to entail that every event

127
See e.g. Putnam (1967).
128
See e.g. Kim (1993).
244
has a sufficient physical cause. So unless my pain is identical to something physical

after all, it seems that my pain is causally redundant with respect to my swearing. If

we reject redundant causes, we are thus drawn to the unpalatable conclusion that the

mental is causally inert129. The Truthmaker Ontologist might say something along the

following lines. To preserve both commonsense intuitions and our physicalist

scruples, we must uphold the literal truth of both these claims:

(i) ⟨My being in pain caused me to swear⟩

(ii) ⟨My being in neural state N caused me to swear⟩

But these claims might have the same truthmaker. Perhaps (ii) is made true by a fact

involving physical events related by causation. Plausibly, then, (i) is made true

indirectly be the same fact. If so, there is no objectionable causal redundancy: in

reality, there is just the one causal relation holding between physical events.

Nevertheless, it is literally true that my being in pain caused me to swear.

In a similar vein, Truthmaker Ontology may allow us to reconcile a naturalistic

ontology with a modest appraisal of moral discourse. Mackie (1977) rejects the

existence of sui generis moral properties because he thinks (inter alia) that they have

no place in a naturalistically respectable ontology; and since moral statements

presuppose the instantiation of such properties, argues Mackie, they are

systematically false. One might resist this error-theoretic conclusion by conceding

that moral properties don't exist, while claiming that moral statements aren’t truth-

apt, or aren't used to make serious assertions about moral properties. Or one might

129
'[I]f it isn't literally true that my wanting is causally responsible for my reaching, and my
itching is causally responsible for my scratching, and my believing is causally responsible for
my saying […], then practically everything I believe about anything is false and it's the end
of the world'. (Fodor 1990: 156).
245
argue that moral properties do exist but are identical to certain natural properties.

Truthmaker Ontology gives us another option. We can claim that moral discourse is

truth-apt, assertoric, etc., that ontology doesn't host anything that is identical to a

moral property, but that moral truths have non-moral truthmakers: perhaps causal

facts about the non-moral consequences of actions (e.g. the fact that a certain action

causes pain), psychological facts to the evaluative practices of agents, or some

combination of the two.130

7.3.3 The Problem of Consistency

I will now discuss two potential concerns about the Truthmaker Ontologist's

reconciliation strategy. The first concern is that we can't feasibly reconcile all the

truths we want to accept with a sparse ontology, since the truths we want to accept

are sometimes mutually inconsistent (cf. Lipman 2018).131 To bring this out, let us

return to the Statue and the Clay. The puzzle arises due to an apparent inconsistency

between some individually plausible claims:

(1) Statue and Clay exist.

(2) Statue and Clay exactly coincide.

(3) Statue and Clay have different properties (e.g. Clay would

survive a hammer blow, but Statue would not).

(4) Objects that exactly coincide are identical.

(5) Objects with different properties are not identical.

130
See Asay (2013) for an interesting account of how Blackburn's (1984, 1993) quasi-realism
might be understood in terms of truthmaking.
131
For concerns along these lines, see Lipman (2018), Korman (2015).
246
The Truthmaker Ontologist's proposal above was that we can consistently affirm (1)

while denying that Statue and Clay have ontological status; for all that must really

exist to make (1) true are simples arranged thus-and-so. But an objector might say the

following (cf. Korman 2015):

So what? I'll grant that, by your lights, Statue and Clay can truly be said

to exist, even though neither has ontological status. But (1) still can't

be true together with (2)-(5). So the puzzle you promised to solve

remains intact: we still have five individually plausible but mutually

inconsistent claims, one of which must be given up. If you ignore this

inconsistency, you can't claim to have solved the puzzle. But if you

address the inconsistency by rejecting one of the claims, your solution

is no advance on the standard alternatives which must also deny

something plausible.

The first point to make is that Truthmaker Ontology makes no promise to be able to

preserve every plausible-seeming truth. The aim is to find a theory of truthmakers that

strikes the best balance between ontological plausibility and accounting for what is

plausibly true. In pursuit of this aim, we will occasionally have to engage in some

reflective equilibrium:

We take certain things to be true, and then ask what truthmakers

these truths demand. It may be that at times we will think that we
must have certain truthmakers, and as a result add to what we take
to be our stock of truths. Or we might find ourselves unwilling to
postulate certain truthmakers and therefore having to reject what we
had previously taken to be truths. (Armstrong 2004: 26).

We can expect the occasional conflict between the truthmakers we'd like to posit and

the claims we'd like to make true. When such conflicts arise, we'll need to do some
247
cost-benefit analysis. Suppose the best or only candidate truthmakers for certain

candidate truths seem problematic in some respect— perhaps admitting these

truthmakers would make our ontology less unified, or less parsimonious than we'd

like. If the candidate truths are especially plausible (e.g. Moorean truisms or secure

results of science), then the least costly option might be to admit the candidate

truthmakers as least provisionally. But if the candidate truths are only somewhat

plausible, we might deem the cost of admitting the truthmakers too high and give up

on the candidate truths instead.

Similar cost-benefit reasoning applies when conflicts arise between candidate

truths, as with the five claims above: the task is to find the best candidate truthmakers

for the most plausible claims, then decide what to say about the rest. Looking down

the list (1) has the strongest claim to be a datum. So our priority is to find a truthmaker

for (1): say we settle on the fact that simples s1, …, sn are in spatial configuration C.

Next come (2) and (3). Plausibly, (2) is true because Statue and Clay both exactly

occupy region R. So perhaps the fact that s1, …, sn exactly occupy R is enough to

make it true that Statue and Clay exactly occupy R, hence that they exactly coincide.

(3) is true partly because Clay would survive a hammer blow and Statue would not.

Perhaps this is true in virtue of certain dispositional facts to do with the configuration

of s1, …, sn: under such-and-such physical conditions, s1, …, sn would still be

configured so as to make ⟨Clay exists⟩ true but would no longer be configured so as

to make ⟨Statue exists⟩ true.

That leaves (4) and (5). The question to ask is whether we have a better candidate

truthmaker for ⟨Statue and Clay are identical⟩ (which follows from (2) and (4)) or for

⟨Statue and Clay are distinct⟩ (which follows from (3) and (5)). Deciding one way or

the other will resolve the inconsistency, at the expense of having to give up either (4)

or (5). But contrary to the above objection, I think this way of resolving the
248
inconsistency is better than most of the standard options. We still have a sparse

ontology of simples standing in certain arrangements, with no need for distinct

coinciding objects, temporal parts, or complex objects. At the same time, we have

preserved the truth of the most plausible claims on the list: (1)-(3). Compared to these

benefits, giving up a somewhat plausible philosophical claim about identity seems a

relatively small price to pay.

An alternative is to say that it's indeterminate whether we have a truthmaker for

⟨Statue and Clay are identical⟩ or for ⟨Statue and Clay are distinct⟩. There's perhaps

some pressure to say that Statue and Clay are identical because it's true that they

exactly coincide. And there's perhaps some pressure to say that Statue and Clay are

distinct because it's true that they have different properties. This might lead us to

conclude that it's indeterminate whether Statue and Clay are identical. But on the

Truthmaker Ontologist's diagnosis, this indeterminacy only arises at the level of

linguistic representation and need not carry over into our account of what really

exists. Compare what Cameron says about indeterminate existence (2008a: 16):

Grant, for the sake of argument, that indeterminate existence is to be

avoided; this seems only to mandate that our fundamental ontology
be precise. It doesn't seem at all worrying if it's indeterminate
whether that precise fundamental ontology results in the truth of
certain propositions concerning the existence of complex objects.
After all, this wouldn't be a case of it being indeterminate whether
there is some thing in our ontology. […] It doesn't seem worrying if
the precise way things fundamentally are results in the
indeterminacy of certain claims concerning what exists derivatively.
That's just a case of indeterminacy in what makes what true— it's
not really a case of indeterminate existence.

The same plausibly goes for indeterminate identity. The priority is to ensure that our

account of what really exists does not involve cases of indeterminate identity. But

perhaps due to semantic indecision, it might still be an indeterminate matter whether

249
'Statue is identical to Clay' expresses a true proposition. This seems tolerable,

however, since the indeterminacy involved is linguistic rather than ontological; as

Cameron says elsewhere, 'indeterminacy arising from linguistic or conceptual

unsettledness is utterly unproblematic' (2010a: 263). As before, we still have an

attractive response to the puzzle, which optimises the balance between ontological

plausibility and preserving what is plausibly true.

7.2.4 The Problem of Fit

Here's another potential concern. If our aim is to preserve the truth of familiar claims

about Fs, there are plausibly limits on what truthmakers we can posit for those claims.

You might argue that, if our aim is to preserve the truth of familiar claims about Fs,

it is not enough to cite just any truthmakers for those claims. The truthmakers cited

must 'fit' with the way Fs are ordinarily conceived. Suppose we were to claim that

⟨God exists⟩ is true, since it is made true by certain facts about the human imagination,

or by the universe. Presumably this wouldn't move many atheists, who would likely

reply that these truthmakers don't fit with the theistic conception of God as the all-

powerful creator and sustainer of the universe. Or suppose we were to claim that the

truthmakers for moral statements are the subjective mental attitudes of speakers.

Presumably this wouldn't move a moral error theorist like Mackie (1977), who would

likely reply that these truthmakers don't fit with the ordinary conception of moral

values as objectively prescriptive.

In general, there seem to be constraints on what truthmakers we can appeal to if we

want to preserve truths about Fs as ordinarily conceived. And, according to Audi

(2015), the Truthmaker Ontologist's reconciliation strategy struggles to meet these

constraints. To use his example, consider the sentence 'There are cars'. The

Truthmaker Ontologist might claim that the truth of this sentence is consistent with a
250
car-less ontology, since it could be made true by simples arranged car-wise. But, Audi

argues, part of the ordinary concept of a car is that a car is a certain kind of unit (2015:

223). Hence, 'no swarm of non-cars can be a truthmaker for "There are cars". Certain

swarms may convince us that "There are cars" is a harmless fiction, but harmless is

one thing and truth is another' (Audi 2015: 224). Audi seems to think (but does not

argue) that this generalises: you can't claim to have preserved the truth of statements

about Fs, as Fs are ordinarily conceived, by citing non-Fs as truthmakers (2015: 222).

Let's start by granting the assumption that our aim is to preserve truths about Fs as

Fs are ordinarily conceived. While I agree that there are constraints on what

truthmakers we can then appeal to, I don't agree that the truthmakers must be Fs.

Suppose we want to find a truthmaker for the claim that a certain car ('Car') exists,

which fits with the way cars are ordinarily conceived. And suppose Audi is right that

cars are ordinarily conceived as units. Then it seems to me that our task is to find

suitable truthmakers for both the claim that Car exists and the claim that Car is a unit.

And it’s not clear that these truthmakers couldn't be simples arranged thus-and-so.

Suppose, for the sake of illustration, that cars are ordinarily conceived as units just

because they seem to be objects with fairly well-demarcated boundaries. Thus,

perhaps Car occupies a certain region R, and we think of Car as a single unit because

the boundaries of R mark a clear division between where Car ends and everything

else begins. In that case, perhaps the fact that certain simples are arranged Car-wise

and exactly occupy R is enough to make it true both that Car exists and that Car is a

unit. If so, perhaps this is enough to make it true that Car exists in a way that fits the

(alleged) fact that cars are ordinarily conceived as units.

More generally, if we want to ensure that the truthmakers for F-claims fit with the

way Fs are ordinarily conceived, we needn't impose a priori constraints on what those

truthmakers must be; we must rather focus on what those truthmakers would have to

251
make true if the ordinary conception of F-hood is to be vindicated. This will require

a degree of conceptual analysis. To illustrate, suppose we want to know what it would

take to make ⟨God exists⟩ true, given the way God is conceived within theistic

discourse. We might approach by identifying the core platitudes associated with God

within that discourse— perhaps 'God is all powerful', 'God is all-knowing', 'God is

the creator and sustainer of the universe', etc.— and ask what truthmakers it would

take for these platitudes to be satisfied.132 This might lead us to conclude that, say,

the universe itself isn't a suitable truthmaker for ⟨God exists⟩, because it isn't adequate

with respect to the core God-platitudes.

So on the one hand, there are constraints on what truthmakers we can appeal to if

we want to preserve truths about Fs as ordinarily understood: the truthmakers, I'm

suggesting, should be adequate with respect to the core platitudes associated with F-

hood. This does not mean, however, that we can't preserve truths about Fs as

ordinarily understood by citing non-Fs as truthmakers: as long as the relevant

platitudes are upheld, I don't see what more it would take to vindicate the ordinary

concept of F-hood.

Having said all that, it may not always be desirable to provide truthmakers for F-

claims that leave the ordinary concept of F-hood perfectly intact, since the required

truthmakers may be problematic. Suppose we want to know what it would take to

make ⟨Murder is wrong⟩ true, given the way properties like wrongness are ordinarily

conceived. When we reflect on the platitudes associated with wrongness, we find we

132
Joyce (2001: 3) suggests a rough test for identifying the core platitudes of a discourse,
which I adapted to a slightly different use in §5.3.2. To briefly illustrate, suppose we want to
decide whether some claim, ‘God is F’, expresses a core platitude about God. We can do so,
Joyce suggests, by imagining ourselves encountering some community who have a concept
God*, which seems to behave like our concept God, but who are not disposed to assent to
‘God is F’. If this would lead us not to translate ‘God*’ as ‘God’, then ‘God is F’ is indeed a
core God-platitude.
252
agree with Mackie that wrongness is conceived as an objectively prescriptive

property. Suppose this leads us to conclude that leaving the ordinary concept of

wrongness intact would force us to admit sui generis moral properties as

truthmakers— an ontological cost we're reluctant to bear. In that case, we might

decide it's better to posit truthmakers for ⟨Murder is wrong⟩ that are less costly, but

which provide a less-than-perfect fit with the ordinary concept of wrongness, being

adequate with respect to only some of the wrongness-platitudes. Again, this is a

matter to be decided by cost-benefit analysis: do the benefits of leaving the

wrongness-platitudes intact outweigh the costs of admitting problematic truthmakers

to our ontology?

7.4 Defending Truthmaker Ontology

I will now reply to some remaining objections to Truthmaker Ontology.

Objection: Truthmaker Ontology Changes the Subject

The most likely objection is that Truthmaker Ontology isn't really an approach to

ontology at all. One might insist that, by definition, ontology concerns what there is.

So the alleged advantages of Truthmaker Ontology are the result of definitional

sleight of hand: we can't change the scope of ontology by insisting that only

truthmakers have ontological status any more than we can make dogs have five legs

by counting tails as legs.

Reply

In my view, insisting that ontology can only be about what there is a bit like insisting

that economics can only be about household management: it puts etymology before

our ongoing theoretical concerns. I believe that most ontologists aspire to be Serious

253
Ontologists, addressing deep and substantive questions about the contents of reality.

We should tailor our understanding of 'ontology' to suit this aspiration, even if it

means departing from etymology or historical usage. And, I have argued, Serious

Ontology is better served if we take ontology to address questions about truthmakers

rather than about what there is. So although my view does involve a revisionary

notion of 'ontology', I submit that it is a principled revision.

Objection: Truthmaker Ontology is Esoteric

I claim that ontology is a question of what really exists. What really exists is what

makes a truthmaking contribution, and truthmaking is understood in terms of a

metaphysical primitive: in virtue of. One might object, then, that Truthmaker

Ontology is an example of what Hofweber calls esoteric metaphysics: 'one needs to

understand distinctly metaphysical terms in order for one to understand what the

questions are that metaphysics tries to answer. You have to be an insider to get in the

door' (Hofweber 2009: 267). According to Hofweber, esoteric metaphysics should be

rejected in favour of egalitarian metaphysics, whose questions 'are expressed in

ordinary everyday terms, accessible to all […]. One does not need to understand

special metaphysical terms to understand the questions that we are trying to ask'

(2009: 266).

Reply

First, it's not clear to me that the questions Truthmaker Ontology addresses can't be

ordinarily understood, even if they involve distinctly metaphysical terms. To

understand the questions, the first notion we must grasp is the primitive in virtue of.

We seem to have some ordinary understanding of this notion, as witnessed by the

intelligibility of the sort of examples I used to introduce in-virtue-of relations (§2.1).

Speaking personally, I have found that non-philosophers can easily get an initial grasp
254
on the use of 'in virtue of' as a way of talking about explanatory non-causal

dependence. Matters get more complicated, of course, as we seek to develop a more

refined philosophical account of in-virtue-of relations. But we also have some

ordinary understanding of the notion of causation, which philosophers refine in

sometimes complicated ways; but Hofweber does not see causation as an esoteric

notion (2009: 268). The next idea we need to grasp is the distinction between what is

real and what is merely represented. Again, we seem to have some ordinary

understanding of this distinction, as witnessed by the intelligibility of statements like

'We think and talk about numbers, but the question remains as to whether numbers

are real' (see §1.1). Putting the two together, we can plausibly have some ordinary

understanding of the idea of a truth holding in virtue of a portion of reality, i.e.,

truthmaking. From there, it's not too far a leap to grasp the questions Truthmaker

Ontology addresses: the term really exists is just a device for distinguishing the

portions of reality in virtue of which truths hold and the things that can truly be

represented to exist. So even if the questions of Truthmaker Ontology involve

distinctly metaphysical terms, I'm not persuaded that they are inaccessible to non-

metaphysicians.

Second, even if the questions of Truthmaker Ontology can't be ordinarily

understood, it's not immediately clear what would be wrong with this. As Raven

points out (2012: 695), 'counterfactuals, Chomskyan syntax, autism, quantum states

and economic markets are all things we do not ordinarily understand. We do not reject

them for that reason alone'. Moreover, if we insist that ontological questions must be

'expressed in ordinary, everyday terms, accessible to all' (Hofweber 2009: 266), we

risk misunderstanding the questions. 'Are there numbers?' and 'Are there tables?' may

be egalitarian questions. But as I've argued, they should not be mistaken for questions

of serious ontological interest. Perhaps serious ontological questions can only be

255
adequately posed in distinctively metaphysical terms. As long as the terms are at least

tolerably intelligible (as I've suggested they are), perhaps a degree of esotericism is

not such a bad thing.

Objection: Truthmaker Ontology Doesn't Save the Phenomena

I claim that all that really exists is what makes a truthmaking contribution, which

might be a small proper subset of what can truly be said to exist (see Ch. 4). So

perhaps there are really no such things as tables, numbers, minds, moral values, …,

even if it is true to represent their existence. But you might object that this doesn't

save the phenomena. You might push this from a Moorean angle. It's a Moorean fact,

you might insist, that tables really are out there in the world, not merely that we say

true things about them. So when Truthmaker Ontologists claim to be able to exclude

tables from their ontology without revising familiar truisms, you might worry that

we're not getting our Moorean due. Alternatively, you might argue that Truthmaker

Ontologists can't account for certain important features of the things they deem

unreal. If complex objects don't really exist, for example, how can we account for the

fact that cars, knives, rocks and electrical sockets seem to have causal powers? If

mental states don't really exist, how can we account for the qualia of pain or anger?

Reply

As to the first point, I deny that claims about what really exists have the status of

Moorean facts. Common sense just doesn't have the power to perspicuously reveal the

character of extra-representational reality; it might tell us what is true, but it doesn't

give us the full metaphysical story about why it's true. I think Moore himself might

have agreed. He insisted that, while he was in no doubt as to the truth of many

commonsense propositions, he was far from certain about how they should be

256
analysed (Moore 1959: 53). While I don't think that analysing propositions is the way

to do metaphysics, I do endorse something in a similar spirit: we can perhaps be certain

that 'There are tables' says something true, but we can't be certain about what, in

reality, makes it true.

As for the second point, it's worth noting that Truthmaker Ontology doesn't rule

out the reality of any particular entities a priori. Perhaps the best way to account for

truths about qualia is to admit sui generis mental states as truthmakers; perhaps the

best way to account for certain causal truths is to admit complex objects as

truthmakers. Truthmaker Ontology allows us to exclude certain entities from our

ontology without denying that they exist; but it doesn't force us to do so. But in any

case, I don't see why Truthmaker Ontology can't account for the important features

of entities deemed not to really exist. It seems that rocks have causal powers— they

cause glass to shatter if thrown hard enough, and so on. But one might think that rocks

would do no causal work that simples arranged rock-wise couldn't do by themselves.

Some take this to be a reason to deny the existence of rocks (cf. Merricks 2001). I say

it's a reason to consider whether truths about the existence and causal powers of rocks

might be made true just by simples arranged rock-wise. We could then truly say that

rocks exist and have certain causal powers. But—to borrow a phrase from Heil

(2012)— the 'deep story' about why rocks exist and have their causal powers would

be told in terms of simples.

Conclusion

Metaontology asks: what is the nature of ontological questions and by what methods

might they be answered? I have joined the growing number of philosophers who think

that, if ontology is to be a serious enterprise, we need an alternative to the orthodox

view that ontological questions are questions about what there is. Of the several
257
heterodox approaches that have emerged in recent years, I believe that Truthmaker

Ontology is the most promising. I have therefore sought to further the cause of

Truthmaker Ontology by clarifying its conceptual and methodological foundations.

Along the way, I have tried to shed new light on the relation between truthmaking

and grounding and their roles within Truthmaker Ontology, the distinction between

what has ontological status and what can truly be said to exist, the notion of

fundamentality and its metaontological relevance, and the way to measure the

ontological commitments and costs of theories. Truthmaker Ontology undoubtedly

requires further development. But I hope to have moved us a step closer towards

understanding how ontology might be pursued as a search for truthmakers.

258
Bibliography

Alston, W. (1958) 'Ontological Commitments'. Philosophical Studies 9: 8-17.

Armstrong, D.M. (1978) Universals and Scientific Realism. Cambridge University

Press.

Armstrong, D.M. (1997) A World of States of Affairs. Cambridge University Press.

Armstrong, D.M. (2004) Truth and Truthmakers. Cambridge University Press.

Armstrong, D.M. (2006) 'The Scope and Limits of Human Knowledge'. Australasian
Journal of Philosophy, 84: 2: 159-166.

Asay, J. (2011) Truthmaking, Truth and Realism: New Work for a Theory of
Truthmakers. Doctoral dissertation. Chapel Hill, NC: University of North Carolina.

Asay, J. (2012) ‘A Truthmaking Account of Realism and Anti-Realism’. Pacific

Philosophical Quarterly, 93: 373-394.

Asay, J. (2013) ‘Truthmaking, Metaethics, and Creeping Minimalism’. Philosophical

Studies, 163: 213-232.

Asay, J. (2016) ‘Logic and/of Truthmaking’. In D. Deng, H. Lin, and S. Yang (eds.),
Structural Analysis of Non-Classical Logics. Berlin: Springer-Verlag: 37-55.

Asay, J. (2017) ‘Run Aground: Kit Fine’s Critique of Truthmaker Theory’. The
Philosophical Quarterly, 67 (268): 443-463.

Asay, J. and Baron, S. (2020) ‘Deflating Deflationary Truthmaking’. Philosophical

Quarterly, 70 (278): 1-21.

Audi, P. (2012a) 'Grounding: Toward a Theory of the In-Virtue-Of Relation'. Journal

of Philosophy, 109: 685-711.

Audi, P. (2012b) 'A Clarification and Defense of the Notion of Grounding'. In F.

Correia and B. Schnieder (eds.), Metaphysical Grounding: Understanding the
Structure of Reality. Cambridge University Press: 101-121.

Audi, P. (2015) 'Explanation and Explication'. In C. Daly (ed.). The Palgrave

Handbook of Philosophical Methods. Palgrave: 208-230.

259
Azzouni, J. (2004) Deflating Existential Consequence: A Case for Nominalism.
Oxford University Press.

Azzouni, J. (2010) ‘Ontology and the Word ‘Exist’: Uneasy Relations’. Philosophia
Mathematica, 18(3): 74-101.

Baker, A. (2004) ‘Simplicity’. In E. Zalta (ed.) The Stanford Encyclopedia of

Philosophy.

Baker, A. (2007) ‘Occam’s Razor in Science: A Case Study from Biogeography’.

Biology and Philosophy, 22: 193-215.

Barnes, E. and Cameron, R. (2007) ‘A Critical Study of John Heil’s From an

Ontological Point of View’. Philosophy of Mind Review, 6: 22-30.

Baron, S. (2013a) 'A Truthmaker Indispensability Argument'. Synthese, 190: 2413-

2427.

Baron, S. (2013b) 'Talking About the Past'. Erkenntnis, 78: 547-560.

Baron, S., Chua, R.M., Miller, K. and Norton, J. (2019) 'Much Ado About Aboutness'.
Inquiry, DOI: 10.1080/0020174X.2019.1592705

Baron, S. and Tallant, J. (2018) ‘Do Not Revise Ockham’s Razor Without Necessity’.
Philosophy and Phenomenological Research, 96(3): 596-619.

Baxter, D. (1988) 'Many-One Identity'. Philosophical Papers, 17: 193-216.

Beebee, H. and Dodd, J. (eds.) (2005) Truthmakers: The Contemporary Debate.

Oxford University Press.

Bennett, K. (2011a) ‘By Our Bootstraps’. Philosophical Perspectives, 25,

Metaphysics: 27-41.

Bennett, K. (2011b) 'Truthmaking and Case-making'. Philosophy and

Phenomenological Research 83: 187-195.

Bennett, K. (2011b) ‘Construction Area (No Hard Hat Required)’. Philosophical

Studies, 154: 79-104.

Bennett, K. (2017) Making Things Up. Oxford University Press.

Bigelow, J. (1988) The Reality of Numbers: A Physicalist’s Philosophy of

260
Mathematics. Oxford University Press.

Bigelow, J. (1996) 'Presentism and Properties'. Philosophical Perspectives, 10: 35-

52.

Blackburn, S. (1984) Spreading the Word: Groundings in the Philosophy of

Language. Oxford: Clarendon Press.

Blackburn, S. (1993) Essays in Quasi-Realism. Oxford University Press.

Bliss, R. and Trogdon, K. (2014) ‘Metaphysical Grounding’. In E. Zalta (ed.) The

Stanford Encyclopedia of Philosophy.

Brenner, A. (2017) 'Simplicity as a Criterion of Theory Choice in Metaphysics'.

Philosophical Studies, 174: 2867-2707.

Bricker, P. (2014) 'Ontological Commitment'. In E. Zalta (ed.) The Stanford

Encyclopaedia of Philosophy.

Brogaard, B. (2008) 'Inscrutability and Ontological Commitment'. Philosophical

Studies, 141: 21-42.

Brons, L. (2013) 'What Does It Mean for Something to Exist?'. The Science of
Mind, 51: 53-74.

Brouwer, T. (2013) The Metaphysical Commitments of Logic. Doctoral dissertation,

University of Leeds.

Burgess, J. and Rosen, G. (2005) 'Nominalism Reconsidered'. In S. Shapiro (ed.), The

Oxford Handbook of Philosophy of Mathematics and Logic. Oxford University Press.

Cameron, R. (2008a) ‘Truthmakers and Ontological Commitment: Or, How to Deal

With Complex Objects and Mathematical Ontology Without Getting Into Trouble'.
Philosophical Studies, 140(1): 1-18.

Cameron, R. (2008b) ‘Truthmakers, Realism and Ontology’. Philosophy 83: 107-128.

Cameron, R. (2008c) ‘There Are No Things That Are Musical Works’. The British
Journal of Aesthetics, 48: 295-314.

Cameron, R. (2008d) ‘Turtles All the Way Down’. The Philosophical Quarterly,
58(230): 1-14.

261
Cameron, R. (2008e) 'How to Be a Truthmaker Maximalist'. Nous, 42(3): 410-421.

Cameron, R. (2010a) ‘How to Have a Radically Minimal Ontology’. Philosophical

Studies 151: 249-264.

Cameron, R. (2010b) ‘Quantification, Naturalness and Ontology’. In A. Hazlett (ed.),

New Waves in Metaphysics. Basingstoke: Palgrave-Macmillan: 8-26.

Cameron, R. (2010c) 'Necessity and Triviality'. The Australasian Journal of

Philosophy 88(3): 401-415.

Cameron, R. (2014) ‘Parts Generate the Whole, But They Are Not Identical to It’. In
A. Cotnoir and D. Baxter (eds.), Composition as Identity. Oxford University Press.

Cameron, R. (2015) The Moving Spotlight: An Essay on Time and Ontology. Oxford
University Press.

Cameron, R. (2018) ‘Truthmakers’. In M. Glanzberg (ed.), The Oxford Handbook of

Truth. Oxford University Press.

Cameron, R. (2019) 'Truthmaking, Second-order Quantification, and Ontological

Commitment'. Analytic Philosophy, 60(4): 336-360.

Cameron, R. (Forthcoming(a)). 'Truthmaking and Metametaphysics'. In R. Bliss and

J. Norton (eds.) The Routledge Handbook of Metametaphysics.

Cameron, R. (Forthcoming(b)). ‘Easy Ontology, Two-Dimensionalism, and

Truthmaking’. Oxford Studies in Metaphysics.

Campbell, K. (1994) ‘Selective Realism in the Philosophy of Physics’. The Monist

77: 27-46.

Carnap, R. (1950) 'Empiricism, Semantics and Ontology'. Revue Internationale de

Philosophie 4: 20-40.

Contessa, G. (2013) 'Reading the Book of the World'. Philosophical Studies 172:
1051-1077.

Correia, F. (2010) ‘Grounding and Truth-Functions’. Logique et Analyse, 53(211):

251-279.

Correia, F. (2011) ‘From Grounding to Truth-Making: Some Thoughts’. In A. Reboul

(ed.), Philosophical Papers Dedicated to Kevin Mulligan. (Online publication,

262
available at http://www.philosophie.ch/kevin/festschrift/Correia-paper.pdf)

Correia, F. and Schnieder, B. (eds.) (2012) Metaphysical Grounding: Understanding

the Structure of Reality. Cambridge University Press.

Crane, T. (2013) The Objects of Thought. Oxford University Press.

Daly, C. and Liggins, D. (2010) ‘In Defence of Error Theory’. Philosophical Studies,
149: 209-230.

Daly, C. and Liggins, D. (2014) ‘In Defence of Existence Questions’. The Monist,
97(4): 460-478.

David, M. (2005) ‘Armstrong on Truthmaking’. In H. Beebee and J. Dodd (eds.),

Truthmakers: The Contemporary Debate. Oxford University Press: 141-59.

Davidson, D. (1967) 'The Logical Form of Action Sentences'. In N. Rescher (ed.),

The Logic of Decision and Action. Pittsburgh: University of Pittsburgh Press.

Davidson, D. (2001) Essays on Actions and Events. Oxford University Press.

Davidson, D. (1974) 'Belief and the Basis of Meaning'. Synthese, 27: 309-323.

DeRosset, L (2010) 'Getting Priority Straight'. Philosophical Studies, 149(1): 73-97.

DeRosset, L. (2017) 'Grounding the Unreal'. Philosophy and Phenomenological

Research, 95(3): 535-563.

Dieveney, P. (2008) 'Ontological Infidelity'. Synthese, 165: 1-12.

Dodd, J. (2002) ‘Is Truth Supervenient on Being’, Proceeding of the Aristotelian

Society 102(1): 69-85.

Dodd, J. (2007) 'Negative Truths and Truthmaker Principles'. Synthese 156: 383-401.

Dorr, C. (2005) 'What We Disagree About When We Disagree About Ontology'. In

M. Kalderon (ed.), Fictionalism in Metaphysics. Oxford University Press: 234-286.

Duncan, M., Miller, K. and Norton, J. (2017) 'Is Grounding a Hyperintensional

Phenomenon?'. Analytic Philosophy, 58(4): 297-329.

Eklund, M. (2006) ‘Metaontology’. Philosophy Compass, 1(3): 317-334.

263
Eklund, M. (2009a) 'On Some Recent Criticisms of the "Linguistic" Approach to
Ontology'. Dialectica, 63(3): 313-323.

Eklund, M. (2009b) 'Carnap and Ontological Pluralism'. In D. Chalmers, D. Manley,

and R. Wasserman (eds.), Metametaphysics. Oxford University Press: 130-156.

Eklund, M. (MS) 'Existence, Existence*, and Existence in Reality'.

Embry, B. (2015) 'Truth and Truthmakers in Early Modern Scholasticism'. Journal of

the American Philosophical Association, 1(2): 196-216.

Fiddaman, M. and Rodriguez-Pereyra, G. (2018) 'The Razor and the Laser'. Analytic
Philosophy, 59(3): 341-358.

Field, H. (1980) Science Without Numbers: A Defence of Nominalism. Oxford: Basil

Blackwell.

Field, H. (1989) Realism, Mathematics and Modality. Oxford: Basil Blackwell.

Fine, K. (1994) ‘Essence and Modality’. Philosophical Perspectives, 8: 1-16.

Fine, K. (2011) ‘The Question of Realism’. Philosophers Imprint, 1(1): 1-30.

Fine, K. (2009) ‘The Question of Ontology’. In D. Chalmers, D. Manley, and R.

Wasserman (eds.), Metametaphysics. Oxford University Press: 157-177.

Fine, K. (2010) ‘Some Puzzles of Ground’. Notre Dame Journal of Formal Logic,
51(1): 97-118.

Fine, K. (2012) ‘Guide to Ground’. In F. Correia and B. Schnieder (eds.),

Metaphysical Grounding: Understanding the Structure of Reality. Cambridge
University Press: 37-80.

Finn, S. (2017) 'The Role of Existential Quantification in Scientific Realism'.

Philosophy 92: 351-367.

Fisher, A.R.J. (2016) 'Truthmaking and Fundamentality'. Pacific Philosophy

Quarterly 97(4): 448-473.

Fodor, J. (1990) A Theory of Content. Cambridge, MA: MIT Press.

Fox, J. (1987) ‘Truthmaker’. Australasian Journal of Philosophy, 65: 188-207.

264
Garner, R. (1990) ‘On the Genuine Queerness of Moral Properties and Facts’.
Australasian Journal of Philosophy, 68(2): 137-146.

Goodman, N. (1978) Ways of Worldmaking. Indianapolis: Hackett.

Goldwater, J. (2014) 'Sider's Third Realm'. Metaphysica 15: 99-112.

Griffith, A. (2013) Ways of Truthmaking: A Pluralist Theory of Truthmaking.

Doctoral dissertation. University of California, Irvine.

Griffith, A. (2014) ’Truthmaking and Grounding’. Inquiry, 57(2): 196-215.

Griffith, A. (2015) ’Towards a Pluralist Theory of Truthmaking’. Erkenntnis, 80:

1157-1173.

Hale, B. (2007) 'Neo-Fregeanism and Quantifier Variance'. Proceedings of the

Aristotelian Society 107: 375-385.

Hale, B. and Wright, C. (2009) 'The Metaontology of Abstraction'. In D. Chalmers,

D. Manley, and R. Wasserman (eds.), Metametaphysics. Oxford University Press:
178-212.

Hawley, K. (2015) 'Composition as Identity'. In A. J. Cotnoir and D.L.M Baxter (eds.)

Composition as Identity. Oxford University Press.

Hawthorne, J. (2006) ‘Plenitude, Convention, and Ontology’. Metaphysical Essays.

Oxford University Press: 53-70.

Hawthorne, J. (2009) 'Superficialism in Ontology'. In D. Chalmers, D. Manley, and

R. Wasserman (eds.), Metametaphysics. Oxford University Press: 213-230.

Heil, J. (2003) From an Ontological Point of View. Oxford: Clarendon Press.

Heil, J. (2005) 'Real Tables'. The Monist 88: 493-509.

Heil, J. (2006) 'The Legacy of Linguisticism'. The Australasian Journal of Philosophy

84: 233-244.

Heil, J. (2012) The Universe As We Find It. Oxford University Press.

Hillman, T.A. (2009) 'Substantial Simplicity in Leibniz: Form, Predication &

Truthmakers'. The Review of Metaphysics 63(1): 91-138.

265
Hirsch, E. (1982) The Concept of Identity. Oxford University Press.

Hirsch, E. (2002a) 'Quantifier Variance and Realism'. Philosophical Issues 12: 51-73.

Hirsch, E. (2002b) 'Against Revisionary Ontology'. Philosophical Topics 30(1): 103-

127.

Hirsch, E. (2005) 'Physical-Object Ontology, Verbal Disputes, and Common Sense.

Philosophy and Phenomenological Research 70: 67-97.

Hirsch, E. (2008) 'Language, Ontology, and Structure'. Noûs 42: 509-528.

Hirsch, E. (2009) 'Ontology and Alternative Languages'. In D. Chalmers, D. Manley,

and R. Wasserman (eds.), Metametaphysics. Oxford University Press: 231-259.

Hirsch, E. (2011) Quantifier Variance and Realism. Oxford University Press.

Hirsch, E. (2013) 'The Metaphysically Best Language'. Philosophy and

Phenomenological Research 87: 709-716.

Hofweber, T. (2009) 'Ambitious, Yet Modest, Metaphysics'. In D. Chalmers, D.

Manley, and R. Wasserman (eds.), Metametaphysics. Oxford University Press: 260-
289.

Hofweber, T. (2016) Ontology and the Ambitions of Metaphysics. Oxford University

Press.

Horgan, T. and Potrč, M. (2008) Austere Realism. Cambridge, MA.: MIT Press.

Hornsby, J. ‘Truth Without Truth-Making Entities’. In H. Beebee and J. Dodd (eds.),

Truthmakers: The Contemporary Debate. Oxford University Press.

Jackson, F. (1980) 'Ontological Commitment and Paraphrase'. Philosophy 55: 303-

315.

Jackson, F. (1989) 'A Puzzle About Ontological Commitment'. In J. Heil (ed.), Cause,
Mind and Reality. Dordrecht: Kluwer: 191-200.

Jenkins, C.S. (2005) 'Realism and Independence'. American Philosophical Quarterly

42(3): 199-209.

Jenkins, C.S. (2010) 'What is Ontological Realism?'. Philosophy Compass 5(10):

880-890.
266
Jenkins, C.S. (2011) 'Is Metaphysical Dependence Irreflexive?'. Monist 94(2): 267-
276.

Joyce, R. (2001) The Myth of Morality. Cambridge University Press.

Keller, J. A. (2015) 'Paraphrase, Semantics, and Ontology'. In Oxford Studies in

Metaphysics, Vol. 9. Oxford University Press.

Keller, J. A. (2017) 'Paraphrase and the Symmetry Objection'. Australasian Journal

of Philosophy. 95(2): 365-378.

Kelly, T. (2005) 'Moorean Facts and Belief Revision, or Can the Skeptic Win?'.
Philosophical Perspectives 19: 179-209.

Kelly, T. (2008) 'Common Sense as Evidence: Against Revisionary Ontology and

Skepticism'. Midwest Studies in Philosophy 32(1): 53-78.

Kelly, T. (2017) 'Paraphrase and the Symmetry Objection'. Australasian Journal of

Philosophy 95(2): 365-378.

Kim, J. (1993) Supervenience and Mind: Selected Philosophical Essays. Cambridge

University Press.

Kim, J. (2005) Physicalism, or Something Near Enough. Princeton University Press.

Kim, J. (2010) Essays in the Metaphysics of Mind. Oxford University Press.

Korman, D.Z. (2009) ‘Eliminativism and the Challenge from Folk Belief. Noûs 43(2):
242-264.

Korman, D.Z. (2015) ‘Fundamental Quantification and the Language of the Ontology
Room’. Noûs 49(2): 298-321.

Koslicki, K. (2012) ‘Varieties of Ontological Dependence’. In F. Correia and B.

Schnieder (eds.), Metaphysical Grounding: Understanding the Structure of Reality.
Cambridge University Press: 186-213.

Krämer, S. (2014) 'Implicit Commitment in Theory Choice'. Synthese 191(10): 2147-

2165.

Leslie, S. (2007) 'Generics and the Structure of the Mind'. Philosophical Perspectives
21(1): 375-403.

267
Lewis, D. (1970) ‘How to Define Theoretical Terms’. The Journal of Philosophy
67(13): 427-446.

Lewis, D. (1974) ‘Radical Interpretation’. Synthese 23: 331-344.

Lewis, D. (1975) ‘Languages and Language’. In K. Gunderson (ed.), Language, Mind

and Knowledge. Minnesota Studies in the Philosophy of Science (7). Minneapolis:
University of Minnesota Press.

Lewis, D. (1983) ‘New Work for a Theory of Universals’. Australasian Journal of

Philosophy, 61(4): 343-377.

Lewis, D. (1984) 'Putnam's Paradox'. Australasian Journal of Philosophy 62: 221-

236.

Lewis, D. (1986) On the Plurality of Worlds. Oxford: Blackwell.

Lewis, D. (1991) Parts of Classes. Oxford: Blackwell.

Lewis, D. (1992) 'Critical Notice of D.M. Armstrong, A Combinatorial Theory of

Possibility'. Australian Journal of Philosophy, 70(2): 211-224.

Lewis, D. and Lewis, S. (1970) 'Holes'. Australasian Journal of Philosophy, 48(2):

206-212.

Liggins, D. (2012) 'Truthmakers and Dependence'. In F. Correia and B. Schnieder

(eds.), Metaphysical Grounding: Understanding the Structure of Reality. Cambridge
University Press: 254-271.

Linnebo, Ø. (2018) 'Platonism in the Philosophy of Mathematics'. In E. Zalta

(ed.), The Stanford Encyclopaedia of Philosophy.

Lipman, M. A. (2018) ‘Against Fundamentality-Based Metaphysics’. Noûs 52(3):

587-610.

Litland, J. E. (2013) 'On Some Counterexamples to the Transitivity of Grounding'.

Essays in Philosophy 14(1), http://doi.org/10.7710/1526=0569.1453.

Lowe, E.J. and Rami, A. (2009) Truth and Truth-Making. Stocksfield: Acumen.

Lycan, W.G. (2001) 'Moore Against the New Skeptics'. Philosophical Studies 103:
35-53.

268
MacBride, F. (2013) ‘Truthmakers’. In E. Zalta (ed.) The Stanford Encyclopedia of
Philosophy.

MacBride, F. (2014) ‘How Truth Depends upon Being’. Analysis 74(3): 370-378.

Mackie, J. (1977) Ethics: Inventing Right and Wrong. Harmondsworth: Penguin.

Manley, D. (2009) 'A Guided Tour of Metametaphysics'. In D. Chalmers, D. Manley,

and R. Wasserman (eds.), Metametaphysics. Oxford University Press: 1-37.

Martin, C. B. (1996) ‘How It Is: Entities, Absences and Voids’. Australasian Journal
of Philosophy, 74(1): 57-65.

McDaniel, K. (2009) 'Ways of Being'. In D. Chalmers, D. Manley, and R. Wasserman

(eds.), Metametaphysics. Oxford University Press: 290-319.

Melia, J. (1998) 'Field's Programme: Some Interference'. Analysis, 58(2): 63-71.

Melia, J. (2005) ‘Truthmaking Without Truthmakers’. In H. Beebee, and J. Dodd

(eds.), Truthmakers: The Contemporary Debate. Oxford University Press: 67-84.

Merricks, T. (2001) Objects and Persons. Oxford: Clarendon Press.

Merricks, T. (2007) Truth and Ontology. Oxford University Press.

Michael, M. (2008) 'Implicit Ontological Commitment'. Philosophical Studies,

141(1): 43-61.

Milne, P. (2005) ‘Not Every Truth Has a Truthmaker’. Analysis 65(3): 221-224.

Moore, G.E. (1959) Philosophical Papers. London: Allen & Unwin.

Mulligan, K., Simons, P. and Smith, B. (1984) 'Truth-Makers'. Philosophy and

Phenomenological Research, 64: 287-321.

Newstead, A. and Franklin, J. (2012) 'On What Exists Mathematically'. In A. Bird, B.

Ellis, and H. Sankey (eds.), Properties, Powers and Structures: Issues in the
Metaphysics of Realism. New York: Routledge: 81-97.

O’Conaill, D. and Tahko, T. (2016) ‘Minimal Truthmakers’. Pacific Philosophical

Quarterly, 97(2): 228-244.

269
Oliver, A. (1996) 'The Metaphysics of Properties'. Mind, 105: 1-80.

Peacock, H. (2011) 'Two Kinds of Ontological Commitment'. The Philosophical

Quarterly, 61(242): 79-104.

Priest, G. (2005) Towards Non-Being. Oxford University Press.

Priest, G. (2008) ‘The Closing of the Mind: How the Particular Quantifier Became
Existentially Loaded Behind Our Backs’. The Review of Symbolic Logic, 1: 42-55.

Putnam, H. (1967) ‘Mental Predicates’. In W.H. Capitan & D.D. Merrill (eds.) Art,
Mind, and Religion. University of Pittsburgh Press.

Quine, W.V.O. (1948) ‘On What There Is’. The Review of Metaphysics, 2(5): 21-38.

Quine, W.V.O. (1960) Word and Object. Cambridge, Mass.: MIT Press.

Quine, W.V.O. (1969) 'Existence and Quantification'. In W.V.O Quine, Ontological

Relativity and Other Essays. New York, Columbia University Press: 91-113.

Quine, W.V.O. (1992) Pursuit of Truth. Harvard University Press.

Rabin, G. O. (2018) 'Grounding Orthodoxy and the Layered Conception'. In R. Bliss

and G. Priest (eds.), Reality and its Structure: Essays in Fundamentality. Oxford
University Press: 37-49.

Rami, A. (2009) ‘Introduction: Truth and Truth-Making’. In E.J. Lowe & A. Rami
(eds.) Truth and Truth-Making. Stocksfield: Acumen.

Raven, M. J. (2009) Ontology, From a Fundamentalist Point of View. Doctoral

dissertation, New York University.

Raven, M. J. (2012) 'In Defence of Ground'. Australasian Journal of Philosophy,

90(4): 687-701.

Raven, M.J. (2013) 'Is Ground a Strict Partial Order?'. American Philosophical
Quarterly, 50(2): 193-201.

Raven, M. J. (2015) ‘Fundamentality Without Foundations’. Philosophy and

Phenomenological Research, 93(3): 607-626.

Rayo, A. (2007) ‘Ontological Commitment'. Philosophy Compass, 2(3): 428-444.

270
Read, S. (2000) 'Truthmaker and the Disjunction Thesis'. Mind 109: 67-79.

Restall, G. (1996) ‘Truthmakers, Entailment and Necessity’. Australasian Journal of

Philosophy, 74: 331-340.

Rettler, B. (2016) 'The General Truthmaker View of Ontological Commitment'.

Philosophical Studies, 173: 1405-1425.

Rettler, B. (2017) 'Grounds and "Grounds"'. Canadian Journal of Philosophy, 47(5):

631-655.

Rodriguez-Pereyra, G. (2002) Resemblance Nominalism: A Solution to the Problem

of Universals. Oxford University Press.

Rodriguez-Pereyra, G. (2005) ‘Why Truthmakers’. In H. Beebee and J. Dodd (eds.),

Truthmakers: The Contemporary Debate. Oxford University Press: 17-31.

Rodriguez-Pereyra, G. (2006a) 'Truthmaking, Entailment, and the Conjunction

Thesis'. Mind, 115(460): 968-982.

Rodriguez-Pereyra, G. (2006b) 'Truthmakers'. Philosophy Compass 1/2: 186-200.

Rodriguez-Pereyra, G. (2015) 'Grounding is not a Strict Order'. Journal of the

American Philosophical Association, 1(3): 517-534.

Rosen, G. (2010) 'Metaphysical Dependence: Grounding and Reduction'. In B. Hale

and A. Hoffman (eds.), Modality: Metaphysics, Logic, and Epistemology. Oxford
University Press: 109-136.

Ruben, D-H. (1990) Explaining Explanation. New York: Routledge.

Russell, B. (1919) 'The Philosophy of Logical Atomism'. The Monist, 29(3): 345-380.

Saenz, N. (2014) 'The World and Truth about What is Not'. Philosophical Quarterly,
64: 82-98.

Schaffer, J. (2003) ‘Is There a Fundamental Level?’. Noûs, 37(3): 498-517.

Schaffer, J. (2007) 'From Nihilism to Monism'. Australasian Journal of Philosophy,

85(2): 175-191.

Schaffer, J. (2008a) ‘Truthmaker Commitments’. Philosophical Studies, 141: 7-19.

271
Schaffer, J. (2008b) 'Truth and Fundamentality: On Merricks's Truth and Ontology'.
Philosophical Books, 49(4): 302-316.

Schaffer, J. (2009) ‘On What Grounds What’. In D. Chalmers, M. Manley and R.

Wasserman (eds.), Metametaphysics. Oxford University Press: 347-383.

Schaffer, J. (2010a) 'Monism: The Priority of the Whole'. Philosophical Review

119(1): 31-76.

Schaffer, J. (2010b) ‘The Least Discerning and Most Promiscuous Truthmaker’. The
Philosophical Quarterly 60: 307-324.

Schaffer, J. (2012) 'Grounding, Transitivity, and Contrastivity'. In F. Correia and B.

Schnieder (eds.), Metaphysical Grounding: Understanding the Structure of Reality.
Cambridge University Press: 122-138.

Schaffer, J. (2017) 'Social Construction as Grounding'. Philosophical Studies, 174:

2449-2465.

Schaffer, J. (MS) 'Heavy Ontology, Light Ideology'.

Schipper, A. (2019) 'Fundamental Truthmakers and Non-Fundamental Truths'.

Forthcoming in Synthese.

Schulte, P. (2011a) 'Can Truthmaker Theorists Claim Ontological Free Lunches?'.

European Journal of Philosophy 22(2): 249-268.

Schulte, P. (2011b) 'Truthmakers: A Tale of Two Explanatory Projects'. Synthese 181:

413-431.

Schnieder, B. (2006) 'Truth-making Without Truth-makers'. Synthese 151(1): 21-46.

Schnieder, B. (2011) 'A Logic for "Because"'. Review of Symbolic Logic 4(3): 445-
465.

Sider, T. (2001) Four-Dimensionalism: An Ontology of Persistence and Time. Oxford

University Press.

Sider, T. (2007a) 'Neo-Fregeanism and Quantifier Variance'. Proceedings of the

Aristotelian Society 81: 201-232.

Sider, T. (2007b) 'Parthood'. The Philosophical Review 116: 51-91.

272
Sider, T. (2009) ‘Ontological Realism’. In D. Chalmers, M. Manley and R.
Wasserman (eds.), Metametaphysics. Oxford University Press: 384-423.

Sider, T. (2011) Writing the Book of the World. Oxford University Press.

Sider, T. (2013a) ‘Against Parthood’. In K. Bennett and D. Zimmerman (eds.), Oxford

Studies in Metaphysics, Volume 8. Oxford University Press.

Sider, T. (2013b) 'Symposium on Writing the Book of the World'. Analysis 73(4): 751-
770.

Sider, T. (2014) 'Hirsch's Attack on Ontologese'. Noûs, 48: 565-572.

Soames, S. (2009) 'Ontology, Analyticity, and Meaning: The Quine-Carnap Dispute'.

In D. Chalmers, D. Manley, and R. Wasserman (eds.), Metametaphysics. Oxford
University Press: 424-443.

Tahko, T. (2013) 'Truth-Grounding and Transitivity'. Thought: A Journal of

Philosophy 2(4): 332-340.

Tahko, T. (2014) ‘Boring Infinite Descent’. Metaphilosophy 45(2): 257-269.

Tahko, T. (2018) 'Fundamentality'. In E. Zalta (ed.) The Stanford Encyclopaedia of

Philosophy.

Tallant, J. (2011). Metaphysics: An Introduction. London: Continuum.

Thomasson, A. (2007) Ordinary Objects. Oxford University Press.

Thomasson, A. (2009) 'Answerable and Unanswerable Questions'. In D. Chalmers,

M. Manley and R. Wasserman (eds.), Metametaphysics. Oxford University Press:
444-471.

Thomasson, A. (2015) Ontology Made Easy. Oxford University Press.

Thomasson, A. (MS) 'Truthmakers and the Project of Ontology'.

Trogdon, K. (2013) ‘An Introduction to Grounding’. In M. Hoeltje, B. Schnieder and

A. Steinberg (eds.), Varieties of Dependence: Grounding, Supervenience, Response-
Dependence. Munich: Philosophia-Verlag: 97-122.

Turner, J. (2010) 'Ontological Pluralism'. The Journal of Philosophy 107: 5-34.

273
Turner, J. (2016) 'Curbing Enthusiasm About Grounding'. Philosophical Perspectives
30(1): 366-396.

Uzquiano, G. (2004) 'Plurals and Simples'. The Monist, 87: 429-451.

Van Inwagen, P. (1990) Material Beings. Ithaca, New York: Cornell University Press.

Van Inwagen, P. (1998) 'Meta-Ontology'. Erkenntnis, 48: 233-250.

Van Inwagen, P. (2004) 'A Theory of Properties'. in D. Zimmerman (ed.), Oxford

Studies in Metaphysics, Vol. 1. Oxford University Press: 107-138.

Van Inwagen, P. (2005) 'Existence, Ontological Commitment, and Fictional Entities'.

In M. Loux and D. Zimmerman (eds.), The Oxford Handbook of Metaphysics. Oxford
University Press: 131-158.

Van Inwagen, P. (2009) 'Being, Existence, and Ontological Commitment'. In D.

Chalmers, M. Manley and R. Wasserman (eds.), Metametaphysics. Oxford University
Press: 472-506.

Von Solodkoff, T. (2012) 'Straightening Priority Out'. Philosophical Studies 161(3):

391-401.

Von Solodkoff, T. (2014) 'Paraphrase Strategies in Metaphysics'. Philosophy

Compass 9(8): 570-582.

Von Solodkoff, T. and Woodward, R. (2013) 'Noneism, Ontology, and

Fundamentality'. Philosophy and Phenomenological Research 87(3): 558-583.

Williams, J.R.G. (2007) 'Eligibility and Inscrutability'. The Philosophical Review

116: 361-399.

Williams, J.R.G. (2010) ‘Fundamental and Derivative Truths’. Mind, 119: 103-141.

Williams, J.R.G. (2012) 'Requirements on Reality'. In F. Correia and B. Schnieder

(eds.), Metaphysical Grounding: Understanding the Structure of Reality. Cambridge
University Press: 165-185.

274
275