Douglas Edwards - The Metaphysics of Truth (2018, Oxford University Press) - libgen.lc

SAVE OFFLINE

The Metaphysics of Truth

The Metaphysics of Truth Douglas Edwards

1

1 Great Clarendon Street, Oxford, OX2 6DP, United Kingdom Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © Douglas Edwards 2018 The moral rights of the author have been asserted First Edition published in 2018 Impression: 1 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2018931633 ISBN 978–0–19–875869–3 Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

To Alexandra

Contents Acknowledgements

ix

Introduction

1

1. Truth as a Property

6

2. Substantivity and Sparseness

22

3. Deflationism Revealed

41

4. Language–World Connections

60

5. Truth Re-Emerges

83

6. From Truth to Being

105

7. Models of Truth

122

8. Models of Being

142

9. Primitivism and Truth-Making

159

10. Truth-Making and the Nature of Truth

172

Concluding Remarks

183

Bibliography Index

185 195

Acknowledgements There are many people I would like to thank for their help with this book. I would like to start with Peter Momtchiloff, for his advice and guidance throughout, which has made the whole process smooth and enjoyable. I would like to thank two anonymous reviewers for Oxford University Press for their comments on the initial proposal, and, in particular, two anonymous reviewers for their comments on the first draft of the manuscript, and one of the two for commenting again on a revised version. These comments were detailed and exhaustive, and they enabled me to improve the book in many ways. I am very grateful for the diligence and care with which they examined the manuscript. I’d like to thank Christine Ranft for copy-editing the text, and Ranjith Shanmugam for managing the production process. Many people read parts of an earlier draft of the book, and I am grateful for their very helpful comments. Thanks to Filippo Ferrari, Will Gamester, Nathan Kellen, Michael Lynch, Sebastiano Moruzzi, Nikolaj Pedersen, Paul Simard Smith, Cory Wright, and Jeremy Wyatt. I am also grateful to the members of the reading group at the University of Connecticut who discussed an earlier draft, who, in addition to some of those mentioned above, were Jared Henderson, Ken Ito, Robin Jenkins, Junyeol Kim, Joseph Lurie, and Andrew Parisi. I am also grateful to reading groups at the University of Regina, and the University of Bonn/Cologne for their comments, which were kindly passed on to me by Paul Simard Smith and Filippo Ferrari. Thanks also to the participants at ‘Pluralisms Week’ at Yonsei University in June 2016 for their comments and discussion, who, also in addition to some mentioned above, were Colin Caret, Seungrak Choi, Jiwon Kim, Teresa Kouri, Jisoo Seo, Erik Stei, Elena Tassoni, Pilar Terrés Villalonga, Andy Yu, and Luca Zanetti. Earlier versions of some of the material here have been presented at talks or conferences at the University of Manchester, the University of Connecticut, the University of Bologna, the University of Padova, the University of Amsterdam, the University of Aberdeen, University College Dublin, the University of St Andrews, the University of Sydney, the Australian National University, La Trobe University, the University of Edinburgh, Cardiff University, and Queen’s University Belfast. I am grateful to audiences at all these talks for many very helpful discussions. I am also grateful to Aaron Cotnoir, Patrick Greenough, Michael Lynch, and Crispin Wright for many conversations about truth and pluralism over the last few years. I am particularly grateful to Crispin and Michael for encouraging me to write a book on truth, and for the inspiration that their work has been. Earlier versions of some of the material here have been published as articles: Chapter 2 is based on ‘Truth as a Substantive Property’, Australasian Journal of Philosophy, 91 (2), 2013: 279–94, copyright © Australasian Association of Philosophy,

x

Acknowledgements

reprinted by permission of Taylor & Francis Ltd, www.tandfonline, on behalf of Australasian Association of Philosophy. Chapters 6 and 8 incorporate material from my sections of ‘From Truth Pluralism to Ontological Pluralism and Back’, The Journal of Philosophy, 112 (3), 2015: 113–40, co-authored with A.J. Cotnoir, reprinted by permission of The Journal of Philosophy. Chapter 7 includes short extracts from ‘Truth, Winning, and Simple Determination Pluralism’, in N. Pedersen and C.D. Wright (eds) Truth and Pluralism: Current Debates, Oxford University Press, 2013 reprinted by permission of Oxford University Press, ‘Simplifying Alethic Pluralism’, The Southern Journal of Philosophy, 49 (1), 2011: 28–48, reprinted by permission of Wiley-Blackwell, and ‘How to solve the problem of mixed conjunctions’, Analysis 68 (2), 2008: 143–9, reprinted by permission of Oxford University Press. Chapter 9 includes short parts of ‘Truth as a Relational Property’, Synthese, 2016, reprinted by permission of Springer. Thanks to all the publishers concerned for these permissions. My greatest thanks are to my family, and especially my wife, Alexandra Plakias. I am very grateful to her for giving me the time to finish the book (particularly as our son Theo was born towards the end of the process); reading and commenting on multiple versions of the manuscript; talking through many ideas; and keeping my morale up throughout the process. This book is dedicated to her, with love.

Introduction Truth occupies a central, yet elusive, role in our lives. From a young age we are told to tell the truth, and that a good person seeks the truth. We are all familiar with the oath taken in court to tell the truth, the whole truth, and nothing but the truth. Many people spend their lives in pursuit of truth, from the deep to the mundane: an astrophysicist tries to discover truths about nature of the universe; Hercule Poirot tries to find the truth about who committed the crime; and probably every day you try to discover the truth about where your keys are. Given that truth is so familiar, it is natural for a philosopher to ask what truth itself is. This is where the elusiveness of truth comes in, as answering this question is notoriously difficult. This question about the nature of truth is a metaphysical question about truth, in that metaphysics is, at least in part, about the nature of things, with ‘things’ understood in terms of different traditional metaphysical categories, such as objects, properties, and events. Accordingly, metaphysical questions about truth ask what sort of thing truth is, and what roles it plays. This requires looking at our practices involving truth, and discerning features truth is taken to have. It also requires looking at what sorts of things are capable of being true, and the role truth has in connection to other things of interest. Of the three ‘traditional’ theories of truth, correspondence theories of truth1 hold that truth consists in our thought and talk corresponding, in some sense, to reality; coherence theories2 hold that hold that truth is a matter of coherence between beliefs in a system; and pragmatist theories hold that truth is a matter of what we are justified in believing, either because those beliefs are useful and verified,3 or because those beliefs are justified at the limit of inquiry.4 Those preoccupied with these traditional metaphysical questions about truth also often take it that truth has a significant role to play in considering other metaphysical questions. For instance, one key issue that has been traditionally associated with truth is the question of how language and thought relate to the world. In particular, truth has been associated with the question of whether reality is independent of our thought and talk (‘realism’), or whether reality is somehow dependent on our thought and talk (‘anti-realism). Correspondence theories of truth were traditionally associated with realism, and coherence and pragmatist theories with anti-realism. This is because, for 1 2

E,g, Russell (1912), Austin (1950), Tarski (1944) (controversially). 3 E.g. Blanshard (1939), Bradley (1914). James (1907).

4

Peirce (1878).

2

Introduction

correspondence theories, reality comes before truth, in that what is true is dependent on reality, whereas for coherence theories and pragmatism, truth comes before reality, in that the nature of reality depends on what is true. For these views, truth plays a significant role in substantial metaphysical debates about how language relates to the world: debates about the nature of truth concern what sort of property truth is, and how we should conceive of the relationship between language and the world as a result. However, contemporary work on truth tends to have a different focus, with less importance given both the metaphysics of truth, and the role of truth in metaphysical debates. This is due to the rise of two main trends in the study of truth. Firstly, deflationary approaches to truth5 have risen to prominence by claiming that the word ‘true’ performs some interesting logical and expressive functions, but, beyond that, there is little more to say. Deflationists aim to remove truth as a notion of significance for philosophical study, and indeed as something that can play a key role in metaphysical theories. Indeed, many deflationists are sceptical about the viability of many metaphysical debates. Secondly, from another direction, primitivist approaches to truth6 have become increasingly popular. Primitivists are more interested in metaphysical debates than deflationists, but do not see the need to engage in the project of investigating the nature of truth. In particular, whilst there has been an explosion of interest in the notion of truth-making, this has not correlated with an increase of interest in the nature of truth; rather, interest has shifted from theories of truth to theories of truthmaking. This has led to the idea that, whilst truth may have more roles to play than the deflationist allows, there is little of substantive interest to say about it, expressed by the primitivist claim that truth is not open to definition, and investigations into its nature are “folly”, in Davidson’s (1999) terms. These two trends have rather sidelined the traditional debates about the nature of truth. Truth is not seen as central to discussions of the relationship between language and the world, and there is less focus on developing versions of the traditional correspondence, coherence, and pragmatist theories of truth. Moreover, by taking what appear to be the least committed views about truth, deflationism and primitivism are able to lay claim to the high ground in debates about truth. For example, as the logical and expressive functions of ‘true’ that deflationists focus on are taken to be uncontroversial, deflationism is able to pitch itself as the ‘default’ theory of truth, which requires those interested in connecting truth to metaphysics to refute deflationism before constructing theories of their own.7 For the primitivists, if all can be explained without providing a difficult account or analysis of truth, then they can ask why we should think truth needs investigation.

See, e.g. Horwich (1998), Field (1994), Grover (1992), Ayer (1936), Strawson (1950b). See, e.g. Frege (1918), Davidson (1999). 7 Indeed, this idea has become entrenched to the point where, even though I disagree with this way of looking at things, I am still compelled to begin this book with a critique of deflationism! Hopefully what I say will lend support to an anti-deflationist movement which means others will not have to do so. 5 6

Introduction

3

This book has two main aims. The first main aim of this book is to counter these deflationary and primitivist movements, and re-assert the importance of the metaphysics of truth with regard to both the study of truth itself and metaphysical debates more generally. This will involve revealing the true nature of these views, and showing how they depend on more metaphysical claims about truth than they appear to. It will also involve demonstrating that they lack enough explanatory power to explain the things that they wish to explain. However, this is not to say that we should rest content with the traditional approaches. Indeed, we can learn much from the moves that deflationists and primitivists make, even if we do not wish to follow them all the way. In particular, there has been a growing ‘pluralist’ trend8 that recommends that the proper way to think of the relation between language and the world should not be tackled in a general manner, but by looking at specific subject matters on their own merits. The second main aim of this book is to develop a unique and detailed pluralist metaphysical approach, which starts with the diversity of different subject areas, and holds that there are different relationships between language and the world in different areas. In other words, whilst issues concerning the nature of truth and the connection between language and the world are important, we should not think that they are settled in a uniform manner. I construct a pluralist approach that shows how different areas (or ‘domains’) are individuated; which metaphysical frameworks apply in different domains; and how truth plays a key role in the picture. I focus in particular on the differences between domains that have not yet received a great deal of attention in debates about truth, namely the institutional and social domains. In doing so we will see how debates about truth, whilst by their nature are a little abstract, connect to important discussions about social construction. I will also show that this pluralist approach is not just about truth; it is also about existence. We can note that the situation with the study of existence is not dissimilar to that with truth. Debates about the nature of existence have given way to debates about the existential quantifier, and there is little work for metaphysical investigations into the nature of existence to do. The pluralist metaphysical picture I develop shows how pluralism about truth connects to pluralism about being, which allows us to see how what it takes to exist varies from one domain to another. In particular, I will develop the idea that, in some cases, truth depends on being, but, in others, being depends on truth. This allows us to further see the differences between different domains, and develop the differences between the existence of different kinds of entities, including social and institutional entities. The aim of this pluralist metaphysics is thus to give us a rich, varied, and powerful explanatory framework which gives interesting answers to questions about how sentences are true, and how entities exist. In doing so, it demonstrates the vitality of questions about the nature of truth and the nature of existence, and how they play 8 See, e.g. Wright (1992), Lynch (2009), and Sher (2004). This view is sometimes referred to as ‘alethic pluralism’, but I will use ‘truth pluralism’ throughout this book.

4

Introduction

crucial roles in the vindication of metaphysical inquiries. It also takes us across the rich spectrum of subjects that we think and talk about, and invites us to consider the important differences between these subjects. The structure of the book is as follows. The first three chapters are in the service of responding to the challenge laid down by deflationary theories of truth. Chapter 1 introduces and clarifies the claim that truth is a property, and examines ‘ultra deflationary’ views that deny this. Chapters 2 and 3 then build a case against deflationism about truth. Chapter 2 aims to clarify the deflationist’s claim that truth is an ‘insubstantive’ property in terms of the distinction between abundant and sparse properties, which I suggest is the best way to make sense of the claim. In Chapter 3 we then see that deflationists are unable to make use of this distinction due to the radical anti-metaphysical nature of their view. In the process I show that the four main characteristics of deflationism about truth cannot be maintained, and that, at most, deflationism about truth cannot be separated from a larger ‘global’ deflationist view of language and the world which has radical consequences. Critiques of this global form of deflationism are also presented in Chapters 5 and 6. In Chapter 4 I begin to construct a general pluralist picture of the relationships between language and the world. I show how we can distinguish between different kinds of domains, and show how we can go on to ask metaphysical questions about them. Key to this is the distinction between sparse and abundant properties that we made in Chapter 2. Included in this chapter are some case studies of, in particular, social and institutional domains. I close the chapter by developing the notion of a domain and setting out the criteria for domain membership for atomic sentences. In Chapter 5 I show how the general framework developed in Chapter 4 is dependent on views about the metaphysics of truth. I develop a pluralist approach to the metaphysics of truth, develop some examples to show how truth is considered differently in different domains, and defend the idea from some initial objections. In Chapter  6 the view is extended from truth to being, and I show how a pluralist metaphysics is augmented by ontological pluralism: the idea that there are different ways of being. At the end of Chapter 6 I show why the ‘global’ deflationism we ended up with in Chapter 3 is untenable. Chapters 7 and 8 then deal with how truth pluralism and ontological pluralism ought to be formulated, and I discuss how similar models can be constructed for each view. Chapters 9 and 10 deal with two different challenges presented by primitivism, both of which are connected to the notion of truth-making. In Chapter 9 I discuss the first, which comes from the claim that there are truths for which there are no truth-makers, and implies that we should be primitivists about truth, as there are truths for which there are no explanations for why they are true. I show that the pluralist framework has the resources to counter this move, and offers useful resources to those tempted by truth-maker theory. In Chapter 10 I discuss the second challenge, which comes from a commitment to truth-makers, and holds that, once we are committed to there being truth-makers for truths, we do not need a theory of the nature of truth. I show that a theory of truth-making only makes sense if we make key claims about the nature of

Introduction

5

truth, which undermines the primitivist account, and shows another aspect of the importance of the metaphysics of truth. Before moving on, I will just say a little more about metaphysics. I take it that, in general terms, metaphysics is concerned with foundational issues about what things there are, and what features those things have. These issues are foundational in that they are distinct from investigations carried out within a particular field, pursued a ‘step back’ from the field of inquiry being pursued, and not necessarily using the methods of that field. For example, a chemist might use empirical methods to identify and describe a compound, such as sodium chloride, and its various features and interactions. In a certain sense they are talking about what there is and what it is like, but a metaphysician would not see these as foundational issues, for we can still ask, from a position outside the practice, whether there really is such a thing as sodium chloride, and, if there is, how it should be accounted for.9 These questions would draw on foundational questions such as: What is it to exist, and what relationship is there between existence and empirical detection? Are compound entities distinct things from the elements that compose them? These are not questions that a chemist can answer, but they are questions that a metaphysician would tackle. There has been increasing interest in recent years in the connections between metaphysics and fundamentality.10 As will become clear, I do not associate metaphysical inquiry solely with inquiry into what is fundamental, though I do discuss fundamentality at various points in what follows. Truth might not be a fundamental feature of reality, but that this does not mean that there should be any less interest in the metaphysics of truth. Many of the things that are important to us are not fundamental features of reality, but this does not mean that metaphysical inquiry into them is any less substantial. Indeed, as Elizabeth Barnes (2014) emphasizes, it is in these areas that metaphysical inquiry is most useful and urgent. Accordingly, this book contains metaphysical discussion of a range of different areas, including institutional, social, physical, biological, chemical, moral, and mathematical issues. I contend that truth is an important presence in all of these areas, and any metaphysical account of truth needs to consider them all. Finally, in trying to focus this discussion on the metaphysics of truth, there are important aspects of the study of truth that are not discussed in detail, particularly logical, epistemic, and normative issues. This is in a way inevitable if one is trying to focus on one particular area of study, and I do not wish to suggest that these other issues are not of great significance. I hope that the metaphysical issues discussed here will also be of use to further investigations in these other areas.

9 For an example of a discussion of this sort of example, see Lewis (1986b) and Armstrong (1986) on structural universals. 10 See, for example, Sider (2011), Schaffer (2009), and Dorr (2005, 2008).

1 Truth as a Property One of the two main aims of this book is to defend the importance of thinking about the metaphysics of truth, and a key part of this involves responding to views which aim to remove interest in the metaphysics of truth. In the first three chapters our focus will primarily be on the idea that truth is a property, and the viewpoints of deflationary theories of truth with respect to this issue. In this chapter we will encounter three ‘ultra deflationary’ views about truth—redundancy theory, performative theory, and prosententialism—which all deny that truth is a property. We will see why these views are hard to maintain, and that even deflationary theories of truth accept that truth is a property. In Chapters 2 and 3 we will explore how more mainstream forms of deflationism try to accept that truth is a property but deny that there are interesting things to say about the metaphysics of truth. I begin this chapter by stating why thinking about truth as a property is part of a metaphysical investigation into truth. Then I discuss the two main arguments for the claim that truth is a property: the linguistic argument, and the metaphysical argument. In the course of the discussion of the linguistic argument we will discuss the three ultra deflationary views noted above, and in the course of the metaphysical argument we will compare the view that truth is a property to the views that truth is an object, and truth is an event. I will close the chapter by making some clarificatory remarks about concepts and truth-bearers.

1.1 Truth and Metaphysical Investigations We can begin with an ontological question: does truth exist? When all is said and done and we give a settled account of all the existents, will truth be one of them? The problem with this question as stated is that it does not really give us enough to go about answering it. There are a number of sorts of thing that can be said to exist, and truth could be considered in a number of different ways. For instance, consider the following options: 1. 2. 3. 4. 5.

The word ‘true’. The concept of truth. The property of truth. The object of truth. The event of truth.

the linguistic case for truth’s being a property

7

With respect to the idea that metaphysics is about the nature of things,1 one way to describe what metaphysicians are interested in is to say that there is a general distinction between things that are parts of thought and talk (concepts and words) and things that are parts of the world (objects, properties, events). Philosophers of language and mind tend to focus on the things that are part of thought and talk, whereas metaphysicians tend to focus on things that are part of the world. As we will see below, making this distinction does not imply that there are no interesting relationships of dependence between these things; rather there is an initial distinction between what philosophers of language and mind focus on, and what metaphysicians focus on. With this in mind, whilst there is a metaphysical aspect to 1 and 2 (words and concepts are objects of a sort, after all), the more significant metaphysical issues concern 3, 4, and 5. When thinking about truth, then, there are a number of different kinds of investigations one could engage in. One could take the word ‘true’ to be the primary thing of interest, and examine how that word is used. One could investigate the concept of truth and examine what the content of that concept is. These are perfectly fine types of investigations, and, indeed, they will be discussed below, but I contend that if we stop there we do not engage in any metaphysical investigations about truth. Engaging in metaphysical investigations about truth involves going beyond looking at the word ‘true’ and the concept of truth, and asking the question of what in the world (if anything) is truth.2 As I noted above, making a distinction between metaphysical and non-metaphysical issues does not mean that there are not significant connections between the linguistic or conceptual issues and the metaphysical issues. For example, as we will see below, many take it that the kind of metaphysical thing truth is, is to some extent settled by the kind of word that ‘true’ is. That is, ‘is true’ is taken to be a predicate, and predicates are taken to refer to properties, which suggests that truth is a property. This is what I will call the standard view: that ‘is true’ is a predicate, and that truth is a property. I call it the standard view as the majority of views on truth agree on this, though there is dispute about what sort of property truth is, which we will discuss in Chapter 2. In the next section I’ll discuss this standard view in more detail, which makes the case that truth is a property on a linguistic basis, before turning to discuss a more metaphysical case.

1.2 The Linguistic Case for Truth’s Being a Property One who rejects the need to ask any metaphysical questions about truth would be one who thought that all there is to say about truth can be said either about the word ‘true’ or the concept of truth, and that there is no need to talk about any sense in which truth See the Introduction. In what follows I will focus primarily on the predicate ‘is true’, as opposed to the concept of truth. The reasons for this are discussed below. 1 2

8

Truth as a Property

exists beyond these two things. These ‘ultra deflationary’ views hold that, once we analyse the sort of word ‘true’ is, we find that it is not the sort of word that has the function of referring to anything, whether object, property, or event. Examples of these views are the ‘redundancy’ theory of truth, where ‘true’ is essentially meaningless, and hence does not refer to anything; the ‘performative’ theory, which holds that ‘true’ has expressive, and not descriptive meaning, and hence does not refer to anything; and (some forms of) the ‘prosententialist’ theory, which holds that ‘true’ is a prosentence (or a prosentential operator) which is not suited to refer to any sort of entity that could be considered to be truth. These views all attack the view that ‘is true’ is a predicate, so we will couch the debate here about the word ‘true’ in terms of whether ‘is true’ is a predicate. Predicates are related to properties, as opposed to objects and events, so the debate here is more about whether truth is a property, as opposed to an object or an event. We will discuss those issues in more detail later, but for now let’s set up the debate here. The ‘linguistic’ argument for truth’s being a property goes as follows. We have a background view which states that a predicate is an incomplete expression of the form ‘x is F’, which forms a sentence when a singular term is plugged into the x position. Any well-formed predicate, i.e. a predicate that has consistent rules for its application, will have an extension, namely the objects that ‘x is F’ is correctly predicated of. If an object is in the extension of ‘x is F’, then that object has the property of being F. At this stage, nothing more is read into having a property than being in the extension of a predicate (we will examine properties in more detail in Chapter 2). If there are objects in the extension of ‘x is F’, then those objects have the property of being F, and we can also say that there is a property, Fness, that those objects have. For example, ‘x is red’ is a well-formed predicate, and (some) roses are in the extension of ‘x is red’. From this we can say that some roses have the property of being red, and that there is a property— redness—that some roses have. We can thus state the linguistic argument for truth’s being a property as follows: 1. Predicates refer to properties. 2. ‘is true’ is a predicate. 3. Therefore, ‘is true’ refers to a property (truth). As we will see, the action in the truth debate here largely concerns premise two. Let’s briefly say something about premise 1, which is accepted by all parties to this particular debate. Why think that predicates refer to properties? As Lewis (1983) argues, at least for the purposes of constructing a semantic theory, we need to have semantic values for the terms in a language. The semantic value of a singular term is the object that it refers to, and the semantic value of a predicate is the set of objects that are in the extension of the predicate. There is at least one understanding of the word ‘property’ on which to have a property F is just to be in the extension of the predicate ‘F’. Taking this view, we can say that properties are the semantic values of predicates, just as objects are the semantic values of singular terms. We will talk about properties more in Chapter 2, but

the linguistic case for truth’s being a property

9

we can note that this is the most minimal way of thinking about properties, and a commitment to them on this model is the least substantial commitment available.3 This is the background picture accepted by all parties with respect to the linguistic argument for truth’s being a property. The linguistic argument for truth’s being a property states that ‘is true’ is a well-formed predicate, and hence it follows that there is a property of being true. If this argument goes through, then there are some metaphysical commitments incurred straightaway. Those who reject this argument attack the premise that ‘is true’ is a well-formed predicate, on the grounds that there is reason to think that it is not a predicate at all.4 Let’s now turn to discuss the views that make this claim: redundancy theory, performative theory, and prosententialism. There has been a fair amount of discussion of these views, and it is fair to say that they occupy something of a niche in contemporary debates about truth, with even most deflationists these days rejecting them. I will briefly run through them and the reasons why they are problematic, as it is important to explain why such initial rejections of the metaphysics of truth are problematic. I will not go into great detail, as others (particularly Künne 2003: ch. 2) have already discussed these issues extensively, and whilst I have a few comments to add, I am mostly in agreement with what has been said before.

1.2.1 The redundancy theory The first view we’ll look at is the redundancy theory of truth. Here is a quote from A.J. Ayer to describe it: Reverting to the analysis of truth, we find that in all sentences of the form “p is true,” the phrase “is true” is logically superfluous. When, for example, one says that the proposition “Queen Anne is dead” is true, all that one is saying is that Queen Anne is dead. And similarly, when one says that the proposition that “Oxford is the capital of England” is false, all that one is saying is that Oxford is not the capital of England. Thus, to say that a proposition is true is just to assert it, and to say that it is false is just to assert its contradictory. And this indicates that the terms “true” and “false” connote nothing, but function in the sentence simply as marks of assertion and denial. (Ayer 1936: 88–9)

The redundancy of the term ‘true’ on this view is claimed on the basis that, if one sincerely says the sentence ‘Queen Anne is dead’, then one is already asserting ‘Queen Anne is dead’. Adding the words ‘is true’ is redundant, as they simply mark the quoted text as an assertion, which is already clear in most contexts. 3 Though I will not argue for it here (see instead Edwards 2014, ch. 4), it is hard to sustain a view which holds that predicates do not refer to properties. This would be to adopt the ‘ostrich nominalist’ position, often attributed to Quine (1948), which holds that there are no properties—in any sense—which, if correct, would mean that predicates do not refer to properties. However, this is a very hard view to maintain, as Quine himself (1960) noted when he retreated to a more moderate position. 4 One could also object on the grounds that truth is not a well-formed predicate, due to the liar paradox. I do not discuss this issue in this book, but see, for example, for example, Scharp (2013) for an extensive discussion of this issue.

10

Truth as a Property

The redundancy theory of truth is hard to sustain. For one thing, the word ‘true’ occurs in contexts other than the example Ayer uses, and in such a way to make it hard to say that it is redundant. For instance, if I say ‘everything Max said is true’ we cannot read this as me simply asserting the sentence ‘everything Max said’. Examples like this lead even those who are sceptical about truth being of metaphysical substance to hold that ‘true’ is not redundant in the sense the redundancy theorist holds. We will discuss this issue more in Chapter 3. In addition to these contexts, it is not clear that there is no meaning to the word ‘true’ in the original examples. This is because there are cases where a sentence and the attribution of truth to that sentence have different truth-values. For example, consider the sentence, ‘the present King of France is bald’, and its corresponding truth-involving claim, ‘ “the present King of France is bald” is true’. If these have the same meaning, then at least the following biconditional should be true: ‘ “the present King of France is bald” is true’ if and only if the present King of France is bald

However, if we suppose that the sentence ‘the present King of France is bald’ is neither true nor false,5 then this biconditional comes out false, as the right-hand side is neither true nor false, but the left-hand side is false: if it is neither true nor false that the present King of France is bald, then it is not the case that it is true that the present King of France is bald, which is what the left-hand side claims. Regardless of whether there are in fact any sentences that are neither true nor false, the fact that such a difference is intelligible makes it hard to sustain the idea that the word ‘true’ adds nothing to the meaning of a sentence.

1.2.2 The performative theory The ‘performative’ theory of P.F. Strawson (1950b) disagrees with the redundancy theory in that it holds that ‘true’ has some meaning, and adds something to what is said, but agrees with the redundancy theorist in that this does not imply that ‘true’ refers to anything. This is because the meaning of ‘true’ is taken to be expressive, and not descriptive. To say ‘ “Queen Anne is Dead” is true’ is not to say exactly the same thing as ‘Queen Anne is dead’, but rather it is to give additional rhetorical flourish to the claim ‘Queen Anne is dead’. To say ‘ “Queen Anne is Dead” is true’ is to say something with the same descriptive content as ‘Queen Anne is Dead’, but the ‘is true’ adds additional expressive content, which adds to the forcefulness of the assertion. There are parallels between this view of truth and expressivist theories in ethics,6 which hold that moral terms like ‘good’ and ‘bad’ have no descriptive content, but are rather expressions of positive and negative attitudes towards particular acts. The sentence ‘murder is wrong’ on this view would not be interpreted as ascribing some

5 6

See Strawson (1950a) for the classic development of this idea. See, e.g. Ayer (1936), Blackburn (1998).

the linguistic case for truth’s being a property

11

descriptive property of wrongness to murder, but rather as the expression of a negative attitude towards the act of murder. However, there are problems with this approach. Firstly, as Künne (2003: 62) points out, the performative theory of truth inherits the ‘Frege–Geach’ problem for expressivism about morals. For example, consider the following argument, from Künne: If it is true that Caesar was murdered, then a civil war is most likely. It is true that Caesar was murdered. Therefore, a civil war is most likely. (Künne 2003: 62) The problem here is that ‘true’ in the second premise occurs in an expressive context, on Strawson’s view, as especially asserting the proposition that Caesar was murdered, whereas in the first premise it occurs as the antecedent of a conditional, where it is not asserted. This not only threatens to render the argument fallacious, as the meaning of ‘true’ is expressive in the second premise, but not in the first, but also suggest a non-expressive meaning of ‘true’ that explains how we are able to use the term in nonassertive contexts. We find a similar problem when considering examples like ‘the present King of France is bald’, discussed above. The problem for the redundancy theory there was that there was a potential divergence in truth-value between the sentence ‘the present King of France is bald’, and the sentence ‘“the present King of France is bald” is true’. This suggests a difference in meaning between these two sentences, when the only difference between them is the occurrence of the words ‘is true’ in one but not the other. This is also a problem for the performative theory, as even the addition of expressive meaning in terms of the addition of the expressive phrase ‘is true’ does not look sufficient to explain the difference in truth-value between the two sentences. That is, the fact that one is an assertion with additional flourish does not explain why it is possible that one is false and the other neither true nor false. Finally, returning to the analogy with expressivism, it is not the case that expressivists need to deny that moral terms are predicates which express properties. Indeed, some, such as Blackburn, embrace that view: There is no harm in saying that ethical predicates refer to properties, when such properties are merely the semantic shadows of the fact that they function as predicates. A quasi-realist protection of ethical truth protects ethical predicates, and if our overall semantic picture is that predicates refer to properties, so be it. But ethical predication remains an entirely different activity from naturalistic predication, and this is only disguised by thinking of the world of properties as one in which hidden identities may be revealed by the philosopher-as-scientist. (Blackburn 1993: 181)

This suggests that the move to an expressive meaning of ‘true’ alone does not suffice for the rejection of the idea that ‘is true’ is a predicate, and that truth is a property. Indeed, Strawson himself ended up accepting that view: [I have long come to admit that] truth is a genuine property, ‘true’ a genuine predicate. (Strawson (1998): 402, quoted by Kunne (2003): 63)

12

Truth as a Property

1.2.3 Prosententialism The third view we’ll consider is prosententialism. This is the most complex of the three, and requires some set up. Also, as we will see below, there is disagreement between the two main advocates of the view—Dorothy Grover and Robert Brandom—on the key issues of whether ‘is true’ is a predicate and truth is a property. Suppose you say ‘snow is white’, and I say ‘that’s true’. On one way of looking at this, what has happened is that I have taken something you have said—‘snow is white’—and attributed a particular property—being true—to that sentence. On a second way of looking at the situation, I have just simply said the same thing you said, but in different words. This second interpretation is the idea behind the prosentential theory of truth, advocated, in slightly different ways, by Dorothy Grover (1992) and Robert Brandom (1994). Grover sees the word ‘true’ as what she calls a ‘prosentence’. We can explain prosentences by reference to pronouns. We often use pronouns as shorthand to help us say things more easily and quickly. For example, if you say ‘did you see the match last night?’, and I say, ‘yes, it was crazy’, then I have used the word ‘it’ as a pronoun rather than saying in response, ‘yes, the match was crazy’, which would be an unusual thing to say given that you have already made clear what we are talking about. Likewise if I say ‘Rooney hit the bar, and then he hit the post’, I use the word ‘he’ as a pronoun, which is used as shorthand to enable me to avoid having to repeat the name twice given that I have made clear who I am talking about. Words like ‘he’, ‘she’, and ‘it’ are pronouns as they stand in for proper nouns in sentences once it is clear what or who is being talked about. Prosentences work in a similar way: once it is clear what sentence is being discussed, a prosentence can be used in place of the original sentence. So, in our original example, when you say ‘snow is white’, and I say ‘that’s true’, I’m using the phrase ‘that’s true’ as a prosentence: it is something that stands in for the sentence ‘snow is white’, and it allows me to express my agreement with you without simply repeating what you have said, which would be conversationally awkward. This allows for expediency in other cases as well, such as in the expression of arguments. Suppose you say ‘snow is cold’, it is easier for me to then say ‘if that’s true, we should bring coats to Alaska’ than to rather robotically say in response ‘if snow is cold, we should bring coats to Alaska’. The idea of the prosentential theory of truth is that this is the primary function of the word ‘true’ in our language: to serve as a prosentence. This marks a divergence from all the other theories of truth available, as all other theories take ‘is true’ to be a regular predicate, not a prosentence. This feature frees the prosentential theory from having to think of truth as a property, in any sense, as we only need to think of truth as a property if we take ‘is true’ to be a predicate. However, you may have noticed that the examples given to drive the prosentential theory are all contexts of a specific sort: they are cases where one person wants to re-assert something another person has said. What about cases where this does not happen, such as if I simply assert ‘it is true that snow is white’, without any reference to what has been said by anyone else (or myself) beforehand? In cases like these the prosentential theory holds that the attribution of truth is redundant, and can simply be

the linguistic case for truth’s being a property

13

removed without loss of meaning. The words ‘it is true that’ can be removed, and my assertion just considered to be the assertion of the sentence ‘snow is white’. Cases like direct truth ascriptions, the prosententialist thinks, do not call for any analysis of truth, as the word ‘true’ can be removed without loss. What this gives us is a view where the word ‘true’ can either be deleted without loss of meaning, as it is in straightforward truth ascriptions, or, in cases where it cannot be as it does not make grammatical sense to do so, such as in cases like ‘that’s true’, the word serves as a prosentence, not a predicate. There are a couple of things to note here. The first is that if the prosententialist wants to treat standard truth ascriptions in the way that the redundancy theorist does, then this view will inherit the same problems about the meaning of ‘true’ that the redundancy theory had above. The second is that there is disagreement among prosententialists on the issues of whether ‘is true’ is a predicate, and whether truth is a property. Grover, for example, frequently refers to ‘is true’ as a predicate, and is also open to the idea that truth is a property, even on a prosententialist account of truth: My interest in properties (or the concept of a property) was prompted by the fact that a prosentential truth predicate seems to lead to an extension for ‘true’; for if anyone can successfully define an extension for the truth predicate, a prosentential theorist can. This is because a prosentence like ‘It is true’ can be used in the following kind of way: The sentence ‘Sea water is salt’ belongs in the extension of the truth predicate if and only if it is true, the sentence ‘Roses are red’ belongs in the extension of the truth predicate if and only if it is true, etc. (Grover 1992: 22)

Grover then goes on to consider whether an extension is sufficient for a property, landing on the idea that, even though an extension is sufficient for a property, it is not sufficient to show that we have an “interesting” or “substantive” property (Grover 1992: 23). This, at this point, is all we need, as the issue of what an interesting or substantive property is will be taken up in Chapter 2. The basic idea is that, whatever additional functions prosententialists think ‘is true’ has, there is an unavoidable commitment to the idea that ‘is true’ has an extension, and hence there is a property of being true. Brandom, on the other hand, gives negative answers on both counts: The first argument depends on treating ‘ . . . is true’ as a predicate. If it is, then since that expression is used to make claims and state facts, it must, on deflationary accounts, be taken to express a property. But the essence of the anaphoric versions of semantic deflationism is precisely to take issue with this grammatical presupposition. According to the account endorsed here, ‘ . . . is true’ expresses a prosentence-forming operator. Its syntax and grammar are quite distinct from those of predicates, to which it bears only the sort of surface similarity that quantificational expressions bear to genuine singular terms . . . The part of speech ‘ . . . is true’ is assimilated to by these theories does not have a directly denotational semantics but inherits its significance anaphorically, by an entirely distinct mechanism. So when it is claimed here that ‘ . . . is true’ does not express a property, this means that it is not even of the right grammatical form to do so—any more than ‘no one’ is of the right form to pick out an individual, although there are some features of its use that could mislead one on this point. (Brandom 1994: 327)

14 Truth as a Property However, as Båve (2009) and Künne (2003) note, Brandom is a little inconsistent on this issue. Båve (2009) notes that it is not clear that thinking of ‘true’ as a prosentential operator is in conflict with the idea that it is a predicate, noting that Brandom’s own criteria do not rule out ‘is true’ from being a predicate. At the very least, more work needs to be done to show that it is not a predicate. On the issue of misleading surfacestructure, Künne (2003: 84) notes an oddity with regard to Brandom’s claim that we should not trust the appearances of truth predication. Here is what Brandom says: And it is clear that any sentence that has the surface form of a predication of truth of some sentence nominalization can be construed as a conditional propositional quantification. But it is not clear that it is a good idea to assimilate what look like straightforward predications of truth to this quantificational model. To do so is to reject the disquotational treatment of these lazy prosentences, which has no greater ontological commitments and stays closer to the apparent form of such sentences. Otherwise almost all sentences involving ‘true’ must be seen as radically misleading as to their underlying logical form. (Brandom 1994: 304; parts also quoted in Künne 2003: 84)

Here Brandom seems to be suggesting that there are contexts where we should not reinterpret predications of truth in a different way which would render ‘is true’ not a predicate. These contexts are the direct truth attributions found in claims such as ‘“snow is white” is true’, which Brandom—and other prosententialists—suggest should be treated in the way the redundancy theory treats them. Granted, the redundancy theory also does not think that ‘is true’ is a predicate, but, as we have seen, it thinks so in these contexts because it holds that the words ‘is true’ have no content, and not because it takes ‘is true’ to have a linguistic function different to that of a predicate. There are a couple of problems with thinking that prosententialism denies that ‘is true’ is a predicate, then. The first is that the anaphoric account of the word ‘true’ only applies in certain contexts; in others, such as direct truth attributions, the redundancy account is used. If, as suggested above, the redundancy account fails to show that ‘is true’ is not a predicate in these contexts, then prosententialism will inherit that feature. This is really all that we need at this stage to see the difficulty in maintaining the view that ‘is true’ is not a predicate. The second issue is that, even in the contexts where the prosententialist thinks ‘true’ is used anaphorically, there is disagreement whether ‘is true’ still functions as a predicate. On Grover’s view there is no bar to thinking that ‘is true’ is a predicate, which expresses a property, whereas on Brandom’s view there is. As we have already established enough to show that there is difficulty in rejecting that ‘is true’ is a predicate in certain significant circumstances, I will leave this further debate to the prosententialists themselves. We started this section with the observation that ‘is true’ appears to function as a predicate. We subsequently stated the three main lines of objection to this idea, and reviewed the central reasons why they are hard to sustain. I take it then that we are at least prima facie justified in thinking that ‘is true’ is a predicate, and thus that the second premise in the linguistic argument is prima facie justified.

the metaphysical case for truth’s being a property

15

1.3 The Metaphysical Case for Truth’s Being a Property The linguistic argument for truth’s being a property begins with a thesis about the connection between a specific piece of language (a predicate) and a specific entity (a property). I now wish to examine whether truth is a property from a more metaphysical standpoint: in other words, putting language aside for a moment, what metaphysical reasons might we have for thinking of truth as a property? And what are the alternatives? Supposing for the moment that we do think there are metaphysical things to say about truth, we have (at least) the three options given above: that truth is a property; that truth is an object; that truth is an event. I will say more below about properties and objects, but I will take it for granted at this stage that there are such things, and thus that it is intelligible to ask which of these categories a particular entity belongs to. We can put the metaphysical case for truth being a property as follows. Consider the following list of sentences: (1) (2) (3) (4)

‘The atomic number of gold is 79.’ ‘Wales is to the West of England.’ ‘The moon is made of green cheese.’ ‘Monkeys are primates.’

When we have a set of entities like this, we can consider the similarities and differences between them. For example, (1)–(4) have something in common, namely that they are all sentences. We would typically put this by saying that (1)–(4) all have the property of being sentences. In addition, there is something that (1), (2), and (4) have that (3) lacks, namely that (1), (2), and (4) are true, whereas (3) is not. Another way to put this is that (1), (2), and (4) have, whereas (3) lacks, the property of being true. This is part of the general idea that some sentences are true, whereas others are false. This naturally lends itself to the thought that there is some property—being true—that true sentences have that false sentences lack. This is a particular instance of the ‘one over many’ argument for the existence of properties, which holds that properties are needed to explain how distinct objects can form a kind.7 Here the idea is that the property of being true is needed to explain the differences between (1), (2), and (4), and (3). At this stage, I don’t want to imply that this metaphysical motivation means that we need an account of properties any more substantive than the minimal account committed to above in relation to the linguistic argument, but this issue will be discussed in more detail in Chapter 2 and as the book progresses. The orthodox view amongst those who take there to be metaphysical things to say about truth is that truth is a property, and not an object or an event. However, it is worth pausing for a moment to say more about this, particularly as there are significant 7

See Armstrong (1978), and also Edwards 2014: ch. 1.

16

Truth as a Property

examples in the history of philosophy of those who have held the alternative positions. I will first examine each view and discuss whether it represents a significant difference from a view which takes it that truth is a property, before discussing each view in relation to the aforementioned metaphysical motivations for thinking of truth as a property.

1.3.1 Truth as an object Probably the most significant expression of the view that truth is an object is the view of Frege, who held that the True and the False were objects, with the True serving as the referent of all true sentences, and the False serving as the referent of all false sentences. With apologies to Frege scholars, I will aim to explain this view in the context of the distinctions we have made here.8 Frege was sceptical of the idea that truth is a property, partly on the grounds that he didn’t see ‘is true’ functioning as a predicate. This was because, as a precursor to redundancy theory, he didn’t think that adding ‘is true’ to a sentence added any content to it. Unlike the redundancy theory though, Frege did not want to deny the reality of truth, and instead suggested that we treat truth as an object. That is, to be precise, he suggested that we treat the truth-values True and False as objects. The idea is that we treat whole sentences as singular terms, and, as singular terms refer to objects, sentences will refer to objects. The referents of sentences are the truth-values of the True and the False, with true sentences referring to the True, and false sentences referring to the False. All true sentences thus have the same referent on this view, as do all false sentences. This does not mean that all sentences mean the same thing, as they are differentiated by their sense, much as singular terms are. Just as ‘Hesperus’ and ‘Phosphorus’ are singular terms with different senses that refer to the same object (Venus), ‘2 + 2 = 4’ and ‘grass is green’ are sentences with different senses that refer to the same object (the True). On this Fregean view, then, we have Truth as an object, and all true sentences are related to that object by the relation of reference. We have already discussed some problems with the line Frege takes on truth not being a predicate when we discussed redundancy theory, so let’s focus here on the claim that truth is an object, and see whether this represents a distinct view to the claim that truth is a property. Consider the view of properties that holds that properties are Platonic Forms.9 On this view we have a single thing, the Form of the Good, say, and all good things are good in virtue of their being related to that Form in some way, sometimes expressed as the relation of resemblance, other times expressed as the relation of participation. Leaving aside the specifics of the relation for a moment, the view looks structurally very similar to Frege’s proposal about truth. To put it in Plato’s terms, we have the Form of the True, and all true sentences are related to that Form. Evidently the nature of the relation is different to Plato’s account, but the basic structure of the two views is the 8 For full and detailed discussions of Frege’s ideas, see, for example, Burge (1986), Ricketts (1986), and Künne (2003). See Beaney (1997) for a collection of Frege’s writings on the subject. 9 See Russell (1912) for a more recent development of this view.

the metaphysical case for truth’s being a property

17

same: we have a single thing, and multiple things are related to that same thing. It is important to notice that, on Plato’s view, properties do not fall out of the picture; indeed, this view is explicitly intended to give an account of what it is to have a property. To have the property of being good, on this view, is to be appropriately related to the Form of the Good. What all good things share is that they all participate in the same relation to the same object: that is what the property of being good consists in. On Frege’s proposal about truth we can say exactly the same thing. Even if grant that there is object, the True, this does not mean that there is no property of being true. All true sentences participate in the same relation to the same object, and, as a result, we can say that there is a property that they all share: the property of being related to the True by the relation of reference. To keep the parallel with Plato’s view, this would be the property of being true: to have the property of being true is to be related to the True by the relation of reference. This is a property that all true sentences have. This also supplies us with content for the predicate ‘is true’: when ‘is true’ is predicated of a sentence, what it means is that the sentence in question is related to the True by the relation of reference.

1.3.2 Truth as an event For a statement of the idea that truth is an event, we can turn to the following words of William James: The truth of an idea is not a stagnant property inherent in it. Truth happens to an idea. It becomes true, is made true by events. Its verity is in fact an event, a process: the process namely of its verifying itself, its veri-fication. Its validity is the process of its valid-ation. (James 1907)

For James, truth and verification are intimately connected, so much so that the verification of a belief is its truth. That is, James thinks of truth as a process, identified with the process of verification. In this way truth is an event. We can distinguish between two different things that James runs together. Firstly, there is the possession of a property, and, secondly, there is the process by which an object gets to possess that property. For example, there is the property of being a father, which a person can have, and there is the process by the person gets to have that property. Even if we say that the process is necessary for the possession of the property, and hence, inseparable in a sense from the possession of the property, the fact remains that the process and the possession of the property are two different things. We then have the choice of what to identify the relevant phenomenon with. In the fatherhood example, we have the choice of identifying the phenomenon of fatherhood with the property possessed by fathers, or with the process by which an individual comes to possess the property possessed by fathers. Generally speaking, the phenomenon is typically identified with the property, and not the process, even if we acknowledge that the process is necessary for the possession of the property. In the truth case, whilst we may accept that there is a ‘truth-making’ process by which a truth-bearer comes to possess the property of being true (in James’s case, verification), this is distinct from truth itself, which is identified with the property of being true.

18

Truth as a Property

1.3.3 Objects, events, and the metaphysical case Explaining the idea behind the metaphysical case for truth’s being a property in terms of the alternative options above (that truth is an object, and that truth is an event) is very hard to do without invoking properties at some stage of the explanation. On the object view, what is needed is the claim that (1), (2), and (4) are all related to the same object (the True) by the same relation (reference). From this it is natural to suppose that (1), (2), and (4) all have the property of being related to the same object by the same relation, and that it is this property—the property of being related to this object (as opposed to another object) by this relation (as opposed to a different relation)— that explains what it is that (1), (2), and (4) have in common. On the event view, we would need to say that (1), (2), and (4) are similar, if not identical, events (they are all truth events), whereas (3) is a different event (a falsity event). To say this we would need to invoke event-types, which would allow us to say that (1), (2), and (4) are the same event-type, and that (3) is a different event-type. Once we introduce event-types, though, it is hard to avoid reference to properties, as some events have the property of being a certain event-type, whereas others are not of that type because they lack that property. As a result, even when we consider objects and events in relation to truth, it is hard to avoid the idea that truth is a property.

1.4 A Note on the Concept of Truth So far we have briefly stated the case for truth’s being a property from a linguistic and metaphysical standpoint. Before moving on I want to make a couple of notes on some important issues, namely concepts and truth-bearers. Let’s start with concepts. So far I have focused on the relationship between predicates and properties, as opposed to the relationship between concepts and properties. This is, for the most part, how much of the ensuing discussion will be conducted, but I wanted to say a little about concepts and their role in the debate. One often finds concepts discussed in relation to truth,10 with the thought being that we look to features of the concept of truth to discern features of the property of being true. Some features of this type are facts of the sort often referred to as ‘platitudes’ in debates about truth. Some examples of these are: Transparency. That to assert a proposition is to present it as true and, more generally, that any attitude to a proposition is an attitude to its truth—that to believe, doubt, or fear, for example, that p is to believe, doubt, or fear that p is true. Embedding. A proposition’s aptitude for truth is preserved under a variety of operations—in particular, truth-apt propositions have negations, conjunctions, disjunctions, etc., which are likewise truth-apt. 10 See, e.g. Alston (2002), Lynch (2009), Wright (2003), Horwich (1998), Naess (1938), Barnard and Ulatowski (2013).

a note on the concept of truth

19

Straight-talking. For a proposition to be true is for it to correspond to reality, accurately reflect how things matters stand, ‘tell it like it is’, and so on. Contrast.

A proposition may be true but not justified, and vice versa.

Stability. If a proposition is ever true, then it always is, so that whatever may, at any particular time, be truly asserted may—perhaps by appropriate transformations of mood, or tense—be truly asserted at any time. Absoluteness. Truth is absolute—there is, strictly, no such thing as a proposition’s being more or less true; propositions are completely true if true at all. (This list is taken from Wright 2003: 271–2. Some of the names of the platitudes in italics are added.) Objectivity: The belief that p is true if, and only if, with respect to the belief that p, things are as they are believed to be. Warrant Independence: Some beliefs can be true but not warranted and some can be warranted without being true. Norm of Belief: It is prima facie correct to believe that p if and only if the proposition that p is true. End of Inquiry: Other things being equal, true beliefs are a worthy goal of inquiry. (Lynch 2009: 8–12) It is indeed undeniable that whenever a proposition or an utterance is true, it is true because something in the world is a certain way—something typically external to the proposition or utterance. (Horwich 1998: 104) It is [a] mistake to regard the principle that, if a statement is true, there must be something in virtue of which it is true, as peculiar to realism. On the contrary, it is a regulative principle which all must accept. (Dummett 1991: 331)

Now, whilst I introduced these ideas using the word ‘concept’, I am wary of having concepts do too much theoretical work. Part of the reason I want to avoid too much discussion of concepts is that this would require an account of what concepts are, which I do not have. In the absence of this, talk of concepts can be somewhat mysterious. Moreover, there are significant disagreements about what concepts are—in particular, whether they are ‘mental’ entities, or abstract entities11—and the choice one takes here has implications for the sorts of things concepts can correspond to. So, rather than talk explicitly about concepts, I will talk about predicates. Where others talk about concepts I will respect their usage, which is broadly that we can think of concepts as things that are had and understood by competent speakers. They are distinct from predicates in that they are not parts of speech, and distinct from properties in that they are not features of objects.

11 See, e.g. Millikan (2000) and Fodor (2003) as examples of the former, and Peacocke (1992) as an example of the latter.

20 Truth as a Property Returning to the list of features given about truth, some present these as features constitutive of our ordinary concept of truth, but, for the reasons just given, I prefer to steer clear of this way of putting things. Instead, it makes more sense to think of inquiry into truth in terms of the method of reflective equilibrium. We try to settle on some constraints on a theory of truth that are common ground, and then construct theories that locate the nature of truth in some property. The aim is to have the characteristics of that property mesh as best as possible with these constraints. Of course, in the process of reflective equilibrium there may be cases where theories and constraints clash, and in these cases decisions need to be made about rejecting the viability of the theory, or revisiting the status of the original constraint. The ‘platitudes’ thus stated are not unrevisable, non-negotiable principles. They constitute the starting point of substantial philosophical analysis in that they are principles that are, broadly speaking, accepted by philosophers working on truth as constraints on a theory of truth. This is not to say that no philosopher can reject them, but that rejection of them usually involves some substantial theoretical commitment to show why the constraint was mistaken in the first place, as opposed to simply rejecting its application out of hand. We will see that these principles have various roles to play as we go through the book.

1.5 A Note on Truth-Bearers Finally, I just want to clarify an important issue going forward in relation to the things that are taken to be the bearers of truth. As you have probably already noticed from the discussion above, when reading work on truth, one will find a whole host of different sorts of things given as truth-bearers, with beliefs, thoughts, ideas, utterances of sentences, types of sentences, statements, and propositions being the main examples. I am content to take a ‘live and let live’ approach to truth-bearers, and by and large will not quibble below with the chosen truth-bearers that different theorists use. In general, when discussing the views of others I will respect their choice of truth-bearer, unless there are problems in doing so, which I will signify. In neutral cases, and when building my positive proposals, I will use utterances of sentences as truth-bearers in my examples. One reason for this is that I am interested in the relationships between singular terms and predicates and the objects and properties that those terms refer to. As these things are the components of sentences, with a ready relation to connect the terms to their corresponding entities, namely reference, this provides the most intelligible structure for the project. This is not to say though that the proposals on offer are restricted to this view of truth-bearers. For example, if one is interested in taking beliefs or thoughts to be truth-bearers, then one just needs to substitute in the equivalent components for singular terms and predicates (objectconcepts and property-concepts, for example), and the equivalent relation to reference (denotation) and the essentials of the picture should remain the same. When it comes to propositions, things get a lot more complex, partly because it is not clear what exactly propositions are. Some take propositions to be structured

summary

21

entities composed of concepts; others take them to be structured entities composed of objects and properties; others take them to be unstructured entities; others take them to be sets of possible worlds; some take them to be abstract objects; others take them to be concrete. The differences between these views are so vast that it is impossible to give a general account of truth for propositions, full stop: one needs to have a particular view in mind. This is especially so as the account one takes may well have significant implications for the account of truth one can give. For example, if one thinks propositions are abstract objects, then certain correspondence theories of truth that take there to be causal relationships between a truth-bearer and the world are off the table right away. Propositions thus introduce a whole host of complications into the mix. Taking propositional truth alone to be the basis for an investigation into the metaphysics of truth in my view makes the project far less interesting and far too idiosyncratic. It also requires a substantial investigation into what the best theory of propositions is, which is a book-length project in its own right.12 I will repeat the offer made above: if one is able in one’s view of propositions to find equivalents for the components and relations involved in the sentences model, then what is said here may work for certain kinds of propositions: the structured model with components being concepts comes to mind as one possibility. I will also say that the considerations offered about the relationships between the components of sentences and their worldly correspondents is something that any content-bearing entity needs to accommodate, so if there is no way to translate thoughts of these sorts into one’s theory of propositions, or incorporate something like them somewhere in the story, then perhaps something has gone wrong somewhere along the way.

1.6 Summary The main aim of this chapter has been to clarify the reasons for thinking that truth is a property, and to respond to the ultra deflationary views of redundancy theory, performative theory, and prosententialism. We saw that, if we are to think of truth as a thing, then it is most naturally thought of as a property, as opposed to an object or event. We also discussed reasons why it is hard to deny that truth is a property, provided we adopt the liberal notion of a property accepted by all parties to the truth debate. However, this is really just the beginning of debates about the metaphysics of truth. The next issue to consider will be the debate between those who, despite accepting that truth is a property, hold that it is not a ‘substantive’ property, and that there is little of interest to say about the metaphysics of truth, namely deflationists about truth, and their opponents, often called ‘inflationists’, who hold that truth is a ‘substantive’ property, and there are interesting things to say about the metaphysics of truth. It is to this issue that we turn in Chapter 2.

12 See Merricks (2015) and King, Soames, and Speaks (2014) for a couple of recent examples which sustain the disagreements over propositions.

2 Substantivity and Sparseness In Chapter 1 we saw the reasons why most parties in debates about truth take truth to be a property. This presents a challenge for deflationary approaches to truth, as, with the ultra deflationary proposals off the table, they now need to both hold that truth is a property and deny that there are interesting metaphysical things to say about truth. This is where the notion of a ‘substantive’ property comes in, as instead of denying that truth is a property, deflationists typically deny that it is a ‘substantive’ property. The main aim of this chapter is to discern the best way to distinguish between substantive and insubstantive properties, and thus to provide the best possible terms in which to conduct the debate between deflationists and their ‘inflationist’ opponents. I argue that three influential accounts fail, and offer what I take to be a more promising view of property substantiveness in terms of a distinction between abundant and sparse properties.

2.1 Deflationism ‘Deflationism’ is a broad term, which can encompass a number of different views, including the ‘ultra deflationary’ redundancy, performative, and prosentential views identified in the previous chapter. I will use the term ‘deflationism’ here to refer to a  particular kind of deflationism, which is what Damnjanovic (2010) refers to as ‘new wave deflationism’.1 Broadly speaking, these views hold that ‘is true’ is a genuine predicate which refers to a property (contra the redundancy theorists, performative theorists, and prosententialists that we encountered in Chapter 1), but deny that this property is ‘substantive’. We will elaborate on the key features of deflationism below, but we can begin by noting that a key role is afforded to the instances of the following schemas: (D) ‘p’ is true iff p. (E) is true iff p. Which of these schemas one takes to be central depends on one’s view of truthbearers: (D) applies to sentences, and (E) applies to propositions. The two main forms 1 Damnjanovic (2010) cites various versions of this view (Field  1994, Hill  2002, Horwich  1998, Künne 2003, Soames 2003, Williams 1999). As will become clear, it is Horwich’s and Field’s formulations that I will focus on in this book.

deflationism

23

of deflationism take different views on this: ‘disquotationalism’ (Field 1994) uses sentences, and ‘minimalism’ (Horwich 1998) uses propositions. For the most part, what is said below applies to both of these forms of deflationism, so I will generally refer to both, using (D/E) to do this. In places where particular arguments concern particular truth-bearers, I will signal this. Also, while I referred to (D/E) above as schemas, I will speak primarily not of the schemas themselves, but of the instances of the schemas. This is because it is controversial whether the instances can be generalized into a general schema,2 so I will focus on the weakest formulation of both of these views by mostly talking about the instances of the schema. Deflationists hold that ‘is true’ is a genuine predicate that plays two important roles in language: generalization and disquotation. Let’s briefly state these two key features of the truth predicate that deflationists focus on. For simplicity’s sake we will use instances of (D) in the next two sections, but one could translate what is said below into propositions and use (E) instead if one wished.

2.1.1 Generalization Let’s start with the generalization function of the truth predicate. To see why deflationists think the word ‘true’ is useful, we can imagine a language which did not have the word ‘true’ in it, and see some problems that would emerge with a person’s ability to express certain thoughts in that language. Suppose I want to endorse a particular thing that somebody said, but I don’t know exactly what they said. For example, suppose you are at a dinner party and the guests are discussing the wine. Your friend Sarah, who is knowledgeable about wine, remarks on it, but due to the general hubbub you don’t manage to hear what she says. Someone then asks for your opinion, and, not being knowledgeable about wine, you simply want to endorse whatever Sarah said, as you trust her opinion. Without the word ‘true’ in your repertoire, this is very difficult, if not impossible. Because you don’t know what Sarah said, you have to consider all the possible things she could have said, and essentially say something like the following: If Sarah said ‘the wine is sweet’, then the wine is sweet, and If Sarah said ‘the wine is fruity’, then the wine is fruity, and If Sarah said ‘the wine is corked’, then the wine is corked, and . . . Evidently, there are many possible things that Sarah could have said about the wine, perhaps even an infinite number, which makes it nigh on impossible for you to endorse what she said to your interlocutor. The addition of the word ‘true’, governed by (D), comes to your rescue. We can take one of the lines above, such as: If Sarah said ‘the wine is fruity’, then the wine is fruity.

2 See Field (1994), Horwich (1998), Hill (2002), Künne (2003). Despite Horwich’s rejection of a general schema, he does talk about a general “propositional structure” (Horwich 1998: 19–20) (see Chapter 3 for more on this).

24

Substantivity and Sparseness

Using (D) we can transform this into: If Sarah said ‘the wine is fruity’, then ‘the wine is fruity’ is true. We can then replace the singular terms with variables and bind them with a quantifier to get: (∀x)

(If Sarah said x, then x is true)

This means that, for all things, if Sarah said that thing, then that thing is true, which is a more formal way of putting the phrase ‘Whatever Sarah said is true’. This is a much easier way of endorsing what Sarah said than listing all the possible things she could have said.

2.1.2 Disquotation Let’s now consider the function of disquotation. Here are some remarks from Quine on the subject: Yet there is some underlying validity to the correspondence theory of truth, as Tarski has taught us. Instead of saying that ‘Snow is white’ is true if and only if it is a fact that snow is white we can simply delete ‘it is a fact that’ as vacuous, and therewith facts themselves: ‘Snow is white’ is true if and only if snow is white. To ascribe truth to the sentence is to ascribe whiteness to snow; such is the correspondence, in this example. Ascription of truth just cancels the quotation marks. Truth is disquotation.3 (Quine 1990) And Wrenn: For example, given this T-biconditional: (3)

‘Snow is white’ is true if, and only if, snow is white.

we can infer each of these from the other: (4)

Snow is white.

(5)

‘Snow is white’ is true.

That is how the truth predicate is able to play its disquotational role. We can use ‘ “Snow is white” is true’, a sentence about a sentence, to say something about the whiteness of snow. (Wrenn 2015: 162–3) The idea here is that the truth predicate allows us to move from talk of language (talk about sentences) to talk about the world. Deflationists take this feature of ‘semantic 3 It is worth considering here whether this expression of deflationism suggests that the problem the original deflationists had was not so much with correspondence, but with the notion of facts. Perhaps the original intention was just to show how facts are defunct, as opposed to truth in general.

deflationism

25

descent’ to be a key feature of the truth predicate. Disquotation is achieved, as one can see from the examples above, once again by the use of the instances of (D). One can infer one side of the biconditional from the other, and hence can infer a direct claim about the world from a claim about a sentence. For the deflationist, the (D/E) instances play a key role in enabling the truth predicate to have these functions. In addition to the importance of the (D/E) instances, following Horwich (1998), we can distinguish between the deflationist theory of truth, and the deflationist conception of truth. The deflationist theory of truth is composed of the (perhaps infinite) list of instances of (D/E). The deflationist conception of truth contains some additional principles that contextualize the significance of these instances. These are: Basicness: the instances of (D/E) are basic. This means that “we do not arrive at them, or seek to justify our acceptance of them, on the basis of anything more obvious or immediately known”. (Horwich 2001: 559)4 Completeness:

the instances of (D/E) explain all the uses of truth predicate.

Purity : the instances of (D/E) are given without making any essential connections between truth and other concepts.5 Insubstantiality: the instances of (D/E) do not imply that there is a substantial property of being true. The deflationist theory of truth in addition to the deflationist conception of truth yields a picture where no substantial theory of truth is needed. The deflationist conception of truth essentially says that the (D/E) instances contained in the deflationist theory of truth are sufficient to explain all that needs to be explained about truth. It is the Insubstantiality thesis that I am interested in in this chapter (we will examine the others in Chapter 3). We should note that the idea of a substantive property with regard to truth was introduced by deflationists who—looking to avoid the pitfalls of redundancy theory, performative theory, and prosententialism—wanted to allow that truth is a property. However, in doing so, they wanted to avoid claiming that there was any substantial nature to truth. As a result, they admitted that truth is a property, but denied that it is a ‘substantive’ property. The question now becomes what is meant by a ‘substantive’ property, and what does a denial or affirmation of truth having that feature mean? At present, there are number of different accounts of what distinguishes a ‘substantive’ property from an ‘insubstantive’ one. This has made it hard to pin down exactly what is at issue at the metaphysical level between deflationists and anti-deflationists (sometimes referred to as ‘inflationists’), which makes it increasingly hard to see how debates between them are properly phrased. Let’s now delve into the issues. 4 As noted in Chapter 3 below, Horwich expresses the idea by saying that the instances are “epistemologically fundamental” (Horwich 2001: 559). Field (1994) suggests that there is a ‘cognitive equivalence’ between the left and right hand sides of the biconditional. 5 Bar-On and Simmons (2007) refer to this feature as ‘Isolationism’.

26 Substantivity and Sparseness

2.2 Features of Inflationary Truth ‘Inflationary’ views are taken to be those that typically hold that the truth predicate refers to some metaphysically significant property. Of course, there is significant disagreement between inflationists regarding what substantial property truth is, but there is general agreement that there is more to it than the deflationist allows. Some examples of inflationary views are correspondence theories, coherence theories, superassertibility theories, primitivist theories, and pluralist theories, all of which are explored below. Before moving any further, it would be useful to have some idea of what the main points of disagreement are between deflationists and inflationists: for example, what jobs do deflationists deny that truth does that they think makes truth insubstantive? It is worth noting that the characteristics of inflationism discussed here have been given under pressure from deflationism, which we should bear in mind when one considers the origin of the distinction between deflationism and inflationism. Nevertheless, having some ideas about the inflationists’ views on truth will help to guide us through the various conceptions of substantiveness on offer to glean what it is that really separates the views. At this stage the aim is just to report the general approaches, so I will keep the discussion in this section brief, and reserve substantial discussion of deflationism versus inflationism for Chapter 3.

2.2.1 The generalization and disquotation jobs Inflationists thus hold that there is a substantive property of being true that can do explanatory work. How does this relate to the functions of the truth predicate that deflationists focus on? According to Wrenn (2015), inflationists agree with the way that deflationists account for generalization and disquotation, but disagree with deflationists that this is all the work there is to do. This would render inflationism essentially deflationism plus something else. As Wrenn puts it: All sides agree that the truth predicate is useful for disquotation and generalization. Let us call the logic of the truth predicate, which underwrites its role in disquotation and generalization, L. Deflationists basically just adopt L as their theory of truth, but an inflationary theory consists of L + X, where X is some additional set of claims that go beyond the logic of the truth predicate, whether those are the claims of the causal correspondence theory or the claims of a pluralist theory. If L alone accounts for all the work the concept of truth does, the addition of X would be a pointless, idle theoretical curlicue. (Wrenn 2015: 164)6

However, we should be careful about putting things this way. The inflationist may choose to use the deflationist’s explanations of generalization and disquotation,7 but they do not have to. This is because they have ways of explaining these features that use what the deflationist does not have, namely a substantive property of being true. 6 Wrenn uses this idea to argue for ‘methodological deflationism’, which we will discuss in more detail in Chapter 3. 7 Devitt (2001) suggests that inflationists should choose to do so.

features of inflationary truth

27

The inflationist can account for the generalizing function of the truth predicate by using the property of being true. For example, compare a similar case where we have a straightforward property-ascribing predicate, like ‘inflammatory’. Suppose that, at the same dinner party as we discussed above, John says something from the other end of the table, which you do not hear, but causes an uproar amongst those sitting around him. Without knowing what John said, it is perfectly reasonable for you to assert, ‘What John said was inflammatory’. This claim takes the same logical form as the claim about truth noted above, where we effectively say that, whatever John said, that thing was inflammatory. What we have here is a simple attribution of a property to a sentence, even though we do not know what the content of that sentence is. Now consider ‘true’ as a regular descriptive predicate, as the inflationist thinks. If I say ‘What John said was true’, even if I didn’t hear it, I am attributing the property of being true to whatever it was that he said. This does not seem at all problematic, and, if the inflationist thinks that ‘is true’ is just a regular predicate, then it seems like the sort of explanation they ought to give, as they do not think that there are the relevant differences between ‘is true’ and ‘is inflammatory’ that the deflationist does. The inflationist can also accommodate the idea that I am asserting the thing that Sarah said when I say ‘what Sarah said is true’. This can be done by appeal to the feature of the property of being true that concerns the connection between truth and assertion. Inflationists typically hold that it is a feature of the property of being true that an assertion presents a proposition as having the property of being true, crystalized in the Transparency ‘platitude’ we encountered in Chapter 1 of ‘to assert is to present as true’. Conversely, on this view, presenting a proposition as having the property of being true just is to assert that proposition. In the case at issue I am explicitly presenting something as true—what Sarah said—even if I do not know what exactly it was that she said. As a consequence, using this feature of the property of being true, we can infer that I asserted what she said, even though I did not know what she said. The inflationist thus has a different explanation to the deflationist of how the truth predicate has a generalizing function. The deflationist claims that this function is discharged entirely by instances of (D). The inflationist, on the other hand, can claim that this function is discharged by the regular property-ascribing nature of the truth predicate, and an aspect of the property of being true that concerns a connection between truth and assertion. The inflationist also need not use the deflationist’s account of disquotation. The key move here is from the claim ‘ “Snow is white” is true’ to the claim, snow is white. Return to the idea that the inflationist thinks that the truth predicate ascribes a property— being true—to a sentence, in this case. It is plausible to assume that any candidate property for counting as the property of being true will have certain constraints on it, such as that any sentence that has the property of being true must ‘tell it like it is’, ‘accurately state how things are’, ‘reflect reality’, or a similar such constraint (as expressed in our list of ‘platitudes’ in Chapter 1). If so then we can infer the whiteness of snow from the truth of the sentence ‘snow is white’ in the following way. If ‘snow is

28 Substantivity and Sparseness white’ is true, then the sentence ‘snow is white’ reflects reality. If the sentence reflects reality, then things are as the sentence says they are. The sentence says that snow is white, so snow is white. Thus we get the move from ‘ “snow is white” is true’ to ‘snow is white’, or from the truth of the sentence that says that snow is white to the whiteness of snow, which is what the disquotation function of the truth predicate does. Just as with the generalization function, we have two different explanations of the same phenomenon from deflationists and inflationists. Deflationists use the instances of (D), and inflationists use the property of being true. These are important differences between the kinds of explanations offered, and show that we should not consider inflationism to be ‘deflationism plus something else’.

2.2.2 Other explanatory jobs Now let’s move on to other jobs. One important task that inflationists think truth does involves the thought that truth registers a distinctive norm of inquiry. The basic idea here is that truth is a dimension of evaluation for a proposition, which is different from other evaluations, such as of being justified, or warranted, or assertible, or probable on the evidence.8 If so, the idea is that there something in which being true consists which gives content to this norm, and an account of the metaphysics of truth is needed to explain it. This, according to the inflationists, suggests that there is something to be said about truth that a deflationary view is unable to accommodate.9 Connected to this is the perceived need to afford truth a particular explanatory role when it comes to explanations of other phenomena.10 The usual thoughts here involve truth playing a significant role in explaining meaning, or individuating content. For instance, many projects in semantics take the meaning of a sentence to be explicable in terms of the conditions under which that sentence is true.11 Such truth-conditional approaches to meaning and content give truth a significant explanatory role to play. In addition there is the idea that truth—particularly true beliefs—plays an important role in explaining how we are successful in getting what we want, and why certain beliefs are valuable to hold.12 For example, the truth of my belief that there is milk in the fridge partly explains how I am successful at getting what I want, namely a glass of milk. True beliefs like this are also valuable to hold, as they help us navigate the world. This also gives truth some significant explanatory role, which, again, a deflationary account seems unable to explain.

8 The central argument to this effect is Wright’s ‘inflationary argument’ in Wright (1992: ch.1). See also Price (2003). 9 Deflationists, of course, have responses to both of the motivations mentioned, but, as I am aiming to merely describe the inflationists’ motivations, I will not outline them here. See Horwich (1998) for more on how deflationists might defuse these concerns. See also Damnjanovic (2005) for some concerns about the aspect of the second motivation that involves the role of true beliefs in successful action. 10 This is argued for specifically in Lynch (2009: ch. 6). 11 See Davidson (1967) for a classic statement of this view. 12 See, for example, Kitcher (2002).

the opacity conception

29

For inflationists, these reasons suggest that there is something about truth—at the metaphysical level—which deflationary views are unable to account for. Precisely how this is to be spelled out is, of course, the topic of this chapter, but it is worth noting that the reasons require at the very least that there is something that being true consists in, whether to establish a norm (so that there is something in which satisfying that norm consists), or to give truth an explanatory role (for truth to explain meaning, for example, there must be something in which truth consists).13 Moreover, according to some inflationists, we need to go further, and explain what it is that all truths have in common. For instance, both Lynch (2006, 2009) and Wright (2013) argue that establishment of a norm requires that all truths share some significant property in common, even on the ‘pluralist’ conceptions of truth which they favour. These, then, are some of the reasons why inflationists think that there needs to be some substantial nature to truth. They suggest a genuine metaphysical difference between deflationism and inflationism, and I will now turn to explore the three main attempts to capture this difference through various notions of what it is that makes a property ‘substantive’. Note that, given what has been said above, two key dimensions for assessing the suitability of a view of property substantiveness are whether the view guarantees genuine similarities between the bearers of the property, and an explanatory role for the property.

2.3 The Opacity Conception The first is the opacity conception of substantiveness, which makes a distinction between ‘transparent’ and ‘opaque’ properties, and is utilized by Crispin Wright (2001), Michael Lynch (2009), and Nic Damnjanovic (2005,  2010).14 Lynch expresses the notion of a ‘transparent’ property as follows: the best way of characterising deflationism’s metaphysical commitments is to say that according to deflationism, truth is a metaphysically transparent property. Metaphysically transparent properties have no underlying nature that isn’t revealed in our grasp of the concept; grasping the relevant concept tells us the whole, or real essence of the property. (Lynch 2009: 116)15

The property of being a conjunction is taken to be a paradigm case of a transparent property. The thought is that once one grasps the concept of a conjunction—that it is a compound statement which is true if and only if each of its components is true— one grasps all there is to know about the essential features of the property of being a 13 Though, as we will see below, this need not imply that we can specify what this is, if some primitivist views are to remain inflationary, as they are surely intended. 14 Wright and Lynch draw on thoughts initially expressed in Alston (1996). Damnjanovic (2010) uses the term ‘revelatory’ for this idea, but I will continue to use ‘transparent’ here. See also Wyatt (2016) for more on this notion. 15 Immediately following this passage, Lynch suggests that ‘concept’ here is to be understood as ‘ordinary concept’, and I will use this understanding in what follows.

30 Substantivity and Sparseness conjunction. That is, a statement is a conjunction if, and only if, it satisfies the right-hand side of the condition laid down in the concept. Contrastively, the example taken to be indicative of a clear case of an opaque property is water. Suppose we take it that our concept of water is comprised of features such as the clear colourless liquid that flows through rivers and streams, that sustains life, that comes out of taps, and so on. A complete list of features such as these would comprise a full account of the concept of water. It is consistent to hold—and indeed was the case for a long time—that one could have a full grasp of the concept of water without knowing what water is, or what being water consists in. This information only became apparent with the discovery that water has the chemical structure H2O. Up until this discovery, it was possible to have a complete grasp of the concept of water without knowing what the property of being water consists in. This suggests that water (somewhat ironically) is not a transparent property: the nature of water is not clear upon grasping the concept.16 Now, the deflationist view about truth holds that all that needs to be said about truth is given by the (D/E) instances.17 Given that the instances of this schema are taken to be part and parcel of the concept of truth, deflationists do not allow that any corresponding property of truth can reveal anything further about the nature of truth. In this sense, deflationists can (at most) allow that truth is a transparent property. This, then, is one way that a property could be considered substantive: by being opaque. The opacity conception of substantiveness aims to ground a property’s substantiveness in the nature its ability to outstrip our (ordinary) concept of it. One question to ask is whether a property’s being opaque implies that all of the things that instantiate that property have the same ‘nature’, in some metaphysically important sense. Given the reasons above, this is important if showing that truth is opaque on a particular view is supposed to be sufficient to show that that view is not deflationary (as is argued by Lynch (2009: ch. 6, 2013b), for example). There is reason to think, though, that there is no guarantee in this area. Take the example of jade.18 It is plausible to think that the property of being jade is a substantive property, and that there is more to being jade than is revealed in our concept of jade. But, as we all know, jade turned out to come in two distinct kinds, jadeite and nephrite. As a result, whilst jade is substantive in the sense that the opacity conception of substantiveness provides, as there is more to the nature of jade than is revealed in the ordinary concept, there is nothing at the level of chemical structure that all instances of jade share. As a result, on the assumption that, in this case, chemical structure is the 16

here.

See also Locke’s (1700) distinction between real and nominal essences, to which there is a connection

17 This, of course, is phrased in terms of how the Horwichian deflationist sees things, but the same applies for other deflationists too: for example, Künne (2003) holds that the concept of truth is given by the schema ∀x (x is true iff ∃p ((x = ) & p)), and this also gives us a full account of the truth property. 18 I would like to thank an anonymous reviewer for the suggestion of this particular example to replace the slightly more complicated example originally in place here.

truth as a logical property

31

appropriate point to assess genuine similarity, we have a property—being jade—which is opaque, but that there is no genuine similarity between all of its instances. Moreover, a property’s being transparent does not entail that its bearers do not share some genuine feature in common. We can find such examples in a scenario where we imagine that our concept formation is perfect. In this scenario, all properties would be transparent, yet there would be some that mark genuine similarities. That is, if we were able to form concepts in such a way that our concepts reflected perfectly all the features of the relevant properties, then our concepts would capture entirely the nature of the properties they correspond to, even those that mark genuine similarities. The nature of these properties would be entirely transparent on each of the corresponding concepts (for the concept would contain all of the relevant information about the property), so they would count as transparent, but each of the objects that instantiates a property would share some genuine characteristic in common. This suggests that opacity and transparency are not, strictly speaking, metaphysical features of properties, rather epistemic features which concern the ability of our concepts to fully reveal the nature of properties.19 This may imply that the transparency of, say, the property of being a conjunction, and the opacity of the property of being water may just reflect the state of our concepts of conjunction and water, as opposed to grounding some sort of significant metaphysical difference. In addition, we may doubt whether there really are any transparent properties of the sort discussed here. Conjunction is often used as the paradigm case, but even logical notions like conjunction, disjunction, and negation are not so transparent as they first seem. Different logical systems, for instance, have different views of logical connectives, some of which do not offer the standard account of a conjunction being something that is true just when its conjuncts are. If—as seems so—different logical systems provide different legitimate accounts of logical connectives, then logical connectives are not as transparent as they first appeared.20 There are number of reasons to doubt, then, the fitness of the opacity/transparency distinction to serve as a reliable guide to the substantiveness or insubstantiveness of properties in general, and in the truth debate. Its status as a primarily epistemic relation between concept and property precludes it from being able to pronounce authoritatively on issues regarding the metaphysical status of a property.

2.4 Truth as a Logical Property The second view of substantiveness uses the notion of a logical property. Damnjanovic (2010) argues that deflationism is committed to the view that truth is a transparent 19 A point noted by Damnjanovic (2010), when he states that properties are transparent relative to a concept. 20 Below in Chapter 7 I will suggest that different accounts of conjunction can be squared with this intuitive gloss, but this involves the machinery of a pluralist theory of truth so I will not go into it here.

32 Substantivity and Sparseness property and that truth is a logical property. Let us now examine whether the notion of a logical property can offer any progress.21 What could justify the claim that truth is a logical property on the deflationary view? One answer is that truth is a logical property because it is a property that has certain logical functions, just as perhaps a property is a physical property because it has certain physical functions. However, this line is not open to the deflationists we are considering here. This is because all of the logical functions deflationists think are important are discharged by the truth predicate, not the truth property. For instance, the standard functions deflationists point to, such as truth being a device for making generalizations, or of indirectly asserting what another has said, are all done by the truth predicate: by having the phrase ‘is true’ as part of the language. Languages that lack the phrase ‘is true’ lack these expressive powers, and the need for these expressive powers is precisely the reason why deflationists allow the phrase ‘is true’ to be a proper part of language. Notice that nothing in this account of the logical and expressive functions associated with truth—which many take to be a crucial part of deflationary accounts of truth— requires a truth property, just a truth predicate. If we think that a property is a logical property in virtue of the logical tasks it performs, then there is not reason to think that, according to deflationism, truth is a logical property. This is because, for deflationists, the truth property performs no logical tasks, rather the truth predicate does. Perhaps, though, what is meant is that any properties expressed by logical predicates will be ‘logical properties’ merely because they shadow logical predicates, thus inheriting the status of their corresponding predicates. While this could be made to fit somewhat with the motivations for avoiding ‘ultra deflationism’ that we discussed in Chapter 1, this approach ultimately does not look too promising. Suppose that all properties merely shadow their corresponding predicates: does this mean that we get distinct kinds of properties for each kind of predicate (e.g. moral properties that shadow moral predicates, physical properties that shadow physical predicates, and so on)? This would be the wrong way to describe the situation. What we would have would perhaps be different kinds of predicates, but one kind of properties: all properties just shadow their predicates. Granted, the predicates may all have very different roles and functions, but this does not cross over to distinctions at the level of property. It thus is not helpful to consider truth to be a logical property, and Damnjanovic’s suggestion of characterizing deflationism by combining the claim that truth is a transparent property with the claim that truth is a logical property does not offer any progress. Of course, we can highlight the claim that deflationists think that truth is a merely logical predicate, but then we lose the significance of the idea that we are looking 21 Künne (2003: 338) also suggests that that truth is a logical property on his view, but whilst Künne defines truth in terms of purely logical notions, this is all at the level of predicate, or concept, not property. The proper conclusion to draw is thus not that, on his view, truth is a logical property, but rather that truth is a logical predicate, or concept. McGinn (2000) also holds that truth is a logical property, but seems to understand logical properties as transparent properties (McGinn  2000: 84) which does not offer much progress for our purposes.

the constitution conception

33

for a distinction between substantive and insubstantive properties, and with it the thought that it is a metaphysical difference that drives the distinction between deflationism and inflationism.

2.5 The Constitution Conception The third potential way to distinguish between substantive and insubstantive properties is to hold that substantive properties are properties which admit of a ‘constitution theory’. This idea is expressed by Horwich, who, responding to a criticism from Crispin Wright, says: Perhaps, then, Wright has in mind what I have called a ‘substantive’ property: namely, the sort of property for which there might well be a constitution theory of the form x is true = x is F. (Horwich 1998: 143)

On this conception, substantive properties are taken to be those properties that have some, in principle, specifiable nature. An insubstantive property, on this view, is a property which lacks any, in principle, specifiable, nature. Deflationism does not allow for truth to be considered a substantive property in this sense. As we saw above, it holds that all there is to be said about truth is given at the level of concept by the (D/E) instances. Deflationists thus do not think that there is any nature of truth to uncover, and reject a constitution theory for truth.22 But there is good reason to think that having a constitution theory, like opacity, does not entail that all instances of the property share some significant feature in common. Consider again the jade case. Instances of this property would fail to share some significant property in common, but this property would pass the test for having a constitution theory, and would thus be a substantive property on this view. This is because we can construct the following constitution theory: x is jade = x jadeite or x is nephrite The fact that the right-hand side specifies a disjunctive property does not mean that we cannot construct a theory of what it is to be jade. Consequently, on the constitution criterion, this property comes out as substantive, yet its instances fail to have some genuine feature in common. It is also worth noting that many properties that would be considered ‘substantive’ may not admit of constitution theories. For example, if there are some fundamental properties (perhaps fundamental physical properties), then these properties would not admit of constitution theories of the type Horwich envisages. This is because these constitution theories supply a reductive account of the phenomenon in question, but, if 22 Künne (2003: 337) claims that his view differs from Horwich’s because he does allow for some form of constitution theory, expressed as ∀x (x is true iff ∃p ((x = ) & p)). But note that Künne gives a biconditional claim in support of this, not the identity claim which is the focus of Horwich’s discussion, and the latter, stronger, claim does not seem to be available to Künne. The same applies to Hill’s (2002) view.

34

Substantivity and Sparseness

we are dealing with a fundamental property, then, by definition, no such reductive account is, in principle, available. Few crumbs of comfort can be found if we were to admit these features of fundamental properties but deny that they applied to truth, for there are accounts of truth that hold that truth is a substantive property, but that there is no reductive account of its nature forthcoming. Forms of ‘primitivism’ about truth along these lines have been held by Russell (1903, 1904), Davidson (1999), and Merricks (2007); Sosa (2001) also expresses sympathy. Moreover, primitivist approaches which hold that the target notion is substantive yet non-reductively analysable are growing in popularity, with Williamson’s (2000) account of knowledge being perhaps the most influential. These views do not say that we just have not found a constitution theory yet; rather they hold that there is something mistaken about the search for one in these cases: that, in principle, there could not be such an account, because the properties concerned are taken to be fundamental to thought and language. Thus, the constitution theory account of substantiveness also looks problematic. It suffers from one of the defects of the opacity account in that possession of a constitution theory does not serve as a significant guide to the instances of the property sharing some genuine feature in common. It also, if taken to be the sole criterion of substantiveness, would rule fundamental properties as insubstantive, and would class the versions of primitivism about truth noted above as giving an insubstantive truth property, both of which are the wrong results.

2.6 Conceptions of Properties So far in the discussion we have said very little about what we ought to take properties to be. In the remainder of the chapter I will spend some time discussing more general accounts of properties, and will suggest that a distinction between abundant and sparse properties is what we need to capture the metaphysics of deflationary and inflationary truth.23 First let us take the view of properties we encountered in Chapter 1, standardly called ‘predicate nominalism’.24 On this view, there are properties insofar as there are extensions of predicates: to have the property of being red is to be in the extension of ‘red’; to have the property of being human is to be to be in the extension of ‘human’, and that’s all there is to it. On this view, properties are taken to be abundant: there are as many properties as there are extensions of predicates.25 This allows us to be very permissive when it comes to assigning properties to predicates, for any predicate with 23 As will become clear below, whilst the terms ‘abundant’ and ‘sparse’ were introduced in terms of the number of properties of a certain type, they have since taken on a different meaning being technical terms referring to features that properties have. 24 The name is due to Armstrong (1978). 25 Note that this is not the most abundant view of properties available. As Lewis (1983) notes, if we take the view that properties are classes (class nominalism), then properties will be more abundant than on predicate nominalism, as there will be classes to which there is no predicate attached.

conceptions of properties

35

consistent rules for its application will be have an extension: a class of (actual and/or possible) objects that satisfy it. Thus, the predicate ‘is red or is a turkey’ will have an extension—the class of objects that are either red or turkeys—and thus will express a property. On this view of properties, it is not the case that an object is in the extension of a predicate ‘is F’ because it has the property of being F; rather the object has the property of being F because it falls under the predicate ‘is F’.26 At the other end of the spectrum we have views that hold that properties form a distinct ontological kind, such as universals.27 On this view, things that share a property in common instantiate the same universal, and in order for two or more objects to instantiate a universal, it must be shown that they share some significant similarities, and that the universal plays some important causal-explanatory role.28 If we take the only properties to exist to be those that are universals, then this limits the number of properties to the number of universals. As universals require some significant similarities at the metaphysical level, the only properties that exist will have to meet this requirement, which renders properties sparse. However, if we take properties to just be abundant, in the predicate nominalist sense, or sparse in the universalist sense then we cannot ground distinctions between different kinds of properties. Predicate nominalism is unable to demonstrate any interesting differences between properties: as all properties are understood in the same minimal sense, there is not a way to distinguish between, say, the property of being metallic and the property of being a trout-turkey: both are just predicate extensions, and possession of the property is just being a member of the relevant class.29 This leaves the view lacking in general explanatory power. Conversely, with the view that properties are universals, we have the problem of how to account for the semantic values of predicates. We saw in Chapter 1 that this is one job that properties are taken to do, but if the only properties are the universals, then this limits the number of semantic values available, and there will be many predicates that we cannot account for in this way. This limits the explanatory power of this approach to properties. As is now commonplace in contemporary theories of properties, there is room for a mixed approach that allows for both abundant and sparse properties. This view, most prominently held by David Lewis (1983, 1986a), holds that all properties are classes, but denies that—in all cases—there is no account of why particular entities are members of particular classes. In some cases, there won’t be anything more to say about why particular objects are members of a class, but in others, there will.30 Or, to put things in terms of predicates as we have been doing, in some cases there will not be an account of why a particular class of objects falls under a particular predicate, whereas, in other cases, there will. See Edwards (2014: ch. 5) for more on this view. See, e.g. Armstrong (1978). See also Edwards (2014: ch.2). 28 This perhaps requires that universals are taken to be the immanent universals favoured by Armstrong (1978), as opposed to abstract universals, and I will assume that here. 29 This is what we said in the context of the discussion of logical properties earlier. 30 See Edwards (2014: ch. 6) for more on this idea. 26 27

36 Substantivity and Sparseness For example, falling under the predicate ‘is red or a turkey’, and thus being a member of the class of red or turkey things does not imply that all the members need share anything significant in common over and above their membership of the class. Contrast this now with, for example, falling under the predicate ‘is metallic’, and being a member of the class of metallic objects: membership of this class is not arbitrary, and each member of the class must meet some significant constraints in order to be a member of the class. Indeed, it is plausible to think that all metallic objects do share some significant feature in common. Perhaps the best way to capture this is to say that all the members of the class instantiate the same universal, but, either way, what should be clear is that, despite the fact that both properties are classes, there are significant metaphysical differences between them, leading to the former’s status as an ‘abundant’ property, and the latter’s status as a ‘sparse’ property. In particular, sparse properties ground genuine similarities between their bearers, and have a causal-explanatory role, whereas abundant properties do not ground genuine similarities between their bearers, and do not have a causal-explanatory role.

2.7 Abundance and Sparseness, Deflationism and Inflationism Allowing both sparse and abundant properties offers progress in the truth debate. Firstly, we can account for the inflationary view in terms of the idea that truth is a sparse property: a property that grounds genuine similarities and that plays a causalexplanatory role. For the deflationist, the notion of an abundant property clearly allows for truth to be a property: to have the property of being true is to be in the extension of the truth predicate (this is the sort of truth property we saw Grover find it hard to deny in Chapter 1). This would sit well with the deflationist, as it would allow the existence of a truth property without committing them to saying that there is anything interesting to say about it. Moreover, notice that, because of the terms of the view, there is nothing more to say about truth than that: there is no explanation of why certain sentences are members of the class of true sentences. If we use this distinction here, then, perhaps most importantly, we allow both inflationists and deflationists to say exactly what they want to say about the kind of property truth is, and thus get to the heart of the debate between them. For the inflationists, holding that truth is a sparse property gives them a truth property with metaphysical weight. If truth is sparse, then truths form a genuine kind, and they share a property in virtue of which they are truths. Moreover, as was noted above, not only are sparse properties taken to ground genuine similarities, they are also taken to feature in causal-explanatory relations, which would satisfy (at least part of) the second reason for favouring inflationism discussed earlier in the chapter.31 31 Damnjanovic (2005) offers some considerations counter to standard thoughts in this area, suggesting— seemingly—that properties that would be classed as abundant in the above distinction do play

clarifying sparseness

37

For the deflationist, the distinction also serves their purposes well. In denying that truth is sparse, they are able to say all the negative things about truth that they standardly want to say, such as that truths do not form a genuine kind, that truth is not a causal-explanatory property, and that truth does not carry any metaphysical weight. In holding that truth is an abundant property they are able to say exactly the kind of positive thing about the truth property that they wish to say: that essentially that there is very little to say about it. The distinction between abundance and sparseness offers progress from the three distinctions we considered earlier. It succeeds where the transparent/opaque distinction failed primarily because the sparseness or abundance of a property is a purely metaphysical matter, and does not rest on anything epistemic, or on the relation between concept and property. It is better suited than the distinction between logical and non-logical properties, as it does not rest on a distinction between corresponding predicates, and does not collapse the metaphysical claims of deflationism into claims about the role of the truth predicate. Finally, it offers progress from the constitution conception, as sparse properties—unlike constituted properties—necessarily ground genuine similarities and causal-explanatory roles, and also because there is no requirement on sparse properties that they have constitution theories, which allows for primitivist accounts of truth to be considered inflationary.

2.8 Clarifying Sparseness Before moving on, I’d like to briefly clarify the position I will be taking on the nature of property sparseness. The first is how to ground the nature of sparse properties. Lewis (1986a) discusses three main options: universals, tropes, and natural classes. I do not think that there is much to choose between these views (see Edwards 2014, ch. 6), and the choice of one over the other makes little difference in what follows. The second issue concerns the connections between property sparseness, as understood here, and fundamentality. Here is a comment Lewis makes about sparse properties: Sharing of [sparse properties] makes for qualitative similarity, they carve at the joints, they are  intrinsic, they are highly specific, the sets of their instances are ipso facto not entirely miscellaneous, there are just enough of them to characterise things completely and without redundancy. (Lewis 1986a: 60)

In this passage, and the following discussion, Lewis introduces the idea that sparse properties are fundamental properties, in addition to being properties that ground genuine similarities and have causal-explanatory roles. This is in line with Armstrong’s causal-explanatory roles. However, it is worth noting that, for Damnjanovic they only do so in a causallyexplanatorily relevant sense, as opposed to a causal-explanatorily efficacious sense (the distinction is due to Jackson and Pettit (1990)), and the standard inflationary line can be pushed as holding that only sparse properties can discharge the latter function.

38

Substantivity and Sparseness

(1978) view of universals. Armstrong thinks that it is the task of physical science to tell us what universals there are, in particular, that the list of universals will be exhausted by the posits of fundamental physical science. One question we can ask here, then, is whether we should include fundamentality in our features of sparse properties. I think that we are better off separating sparseness and fundamentality, for reasons that Schaffer (2004) gives. Schaffer argues that there is a conflict between fundamentality and the features of grounding genuine similarities and causal-explanatory role, in that fundamental properties are limited in what they can do in these respects. For example, take a compound molecule, such as water. There are genuine similarities between all water molecules, and water molecules have specific causal-explanatory roles, but they are not fundamental. As I have discussed elsewhere (Edwards  2014: ch. 6), I take Schaffer’s criticisms here to be effective, so will not take fundamentality to be a feature of sparse properties. The third issue concerns whether we should think of the distinction between abundant and sparse properties as a matter of degree, in that some properties are sparser than others. In previous work (Edwards 2013a) I suggested that we should think of the distinction as a matter of degree, but I will not be taking that line here, primarily because the idea of grading properties in terms of their sparseness is connected to the idea that sparseness is connected to fundamentality, with the sparser properties being the more fundamental properties. In the absence of a commitment to fundamentality as a feature of sparse properties, it does not make a great deal of sense to grade properties in terms of how sparse they are. This brings out a difference between property sparseness, as understood here, and Lewis’s notion of property naturalness. This notion is explicitly tied to fundamentality, and, indeed, in recent years has become associated primarily with that notion,32 with the idea being that the fundamental properties are ‘perfectly natural’, with other properties being more-or-less natural in terms of how long the chains of definition are from the perfectly natural properties. As a consequence, property sparseness, as understood here, should not be equated with Lewisian property naturalness.33 The fourth issue concerns sparseness and intrinsicality, which is another feature mentioned in Lewis’s quote above. Very briefly for now, we can consider the intrinsic properties of an object to be the properties it has in virtue of how it itself is, and extrinsic properties of an object to be properties an object has in virtue of things other than how it itself is.34 We have seen that a different feature mentioned in Lewis’s quote— fundamentality—conflicts with the two features of sparseness we outlined, namely the grounding of genuine similarities, and causal-explanatory role, which led us to exclude See, e.g. Sider (2011), Dorr and Hawthorne (2013). For discussion of truth in relation to Lewis’s notion of naturalness, see Edwards (2013b) and Edwards, Ferrari, and Lynch (2016). For discussions of the naturalness, of notions closely related to truth, see Weatherson (2003) on knowledge, Williamson (2007: ch. 8) and Sider (2011: ch. 3) on reference, and Sider (2009, 2011) on existence. For a discussion of truth and naturalism, see Brogaard (2016). 34 See Edwards (2016) for more on intrinsicality in relation to truth. 32 33

summary

39

fundamentality form our understanding of sparseness here. I think that there is reason also to exclude intrinsicality, not necessarily because it conflicts with the features of sparseness, but because there are examples of extrinsic properties that exhibit the features of sparseness. For example, compare my mass and my weight. The former is an intrinsic property, and the latter is an extrinsic property. Now consider the extent to which these two properties ground genuine similarities and play causal-explanatory roles. I contend that both ground genuine similarities, and both play causal-explanatory roles, albeit in distinct ways. If we are to include intrinsicality in our definition of sparseness, though, then we should discount my weight as a sparse property, but this is unmotivated in terms of our criteria for sparse properties. Consider also the property of being acidic. The Brønsted–Lowry model makes acidity extrinsic, as being acidic is understood in terms of being a proton donor: possession of this property is not just had in virtue of how a thing is in itself, and the nature of acidity is such that it is defined in terms of interactions between a thing and other things. However, acidity looks like a paradigm case of a sparse property: it is a property that marks genuine similarities between its bearers, and plays a significant causal-explanatory role. If intrinsicality is to be included in the criteria for property sparseness, we need some independent reason for thinking so. Interestingly, if we look at Lewis’s subsequent discussion of the role of intrinsicality, it mostly concerns the relationship between intrinsicality and naturalness. We have already noted that a central difference between our understanding of sparseness and Lewis’s notion of naturalness is that naturalness concerns fundamentality, whereas sparseness does not. On the assumption that the intrinsic properties of an object are more fundamental than the extrinsic properties of an object (perhaps because the extrinsic properties are a consequence of the intrinsic properties plus other objects), it does make sense to think that there is a connection between intrinsic properties and natural properties. However, we cannot use this line of argument here to justify a connection between intrinsicality and sparseness, as we do not take sparseness to involve fundamentality. As a result, whilst there may be reason to connect intrinsicality to naturalness, it is harder to justify a connection between intrinsicality and sparseness as understood here, particularly when there are examples of extrinsic properties that exhibit our characteristics of sparseness.

2.9 Summary We have seen that there is reason to think that the distinction between abundant and sparse properties gives us the materials we need for making a metaphysical distinction between deflationism and inflationism at the level of property. It also highlights the best way to think about the battleground for the metaphysical debate between deflationism and inflationism, with one side holding that truth is an important, significant, explanatory property, which binds all truths together, and the other side denying this. My aim in this chapter has been to provide a framework which best reflects what both deflationists and inflationists want out of the deflationism/inflationism distinction.

40

Substantivity and Sparseness

My claim is that the distinction between sparse and abundant properties is the best way of doing this. Moreover, as we will see throughout the book, this distinction between properties has additional theoretical roles of significance to play. In Chapter 3 I will move beyond constructing a framework to understand the deflationism/inflationism debate, and examine the viability of deflationism itself. Amongst other things, I will argue that this distinction between properties—which deflationists commit themselves to—ends up causing deflationism significant problems.

3 Deflationism Revealed As mentioned earlier, an influential source of resistance to thinking about the metaphysics of truth comes from the rise in contemporary philosophy of deflationism about truth. Deflationism threatens the metaphysics of truth in two ways. Firstly, by offering an austere view of truth, it is a view that, if correct, renders the majority of metaphysical issues with respect to truth obsolete. Secondly, it has inspired a certain methodological approach to the study of truth, sometimes referred to as ‘methodological deflationism’.1 The idea here is that, since deflationism is the most neutral and innocuous of all truth theories, it ought to enjoy the position of the ‘default’ theory of truth, with the onus being on other theories of truth to demonstrate why deflationism is wrong prior to taking up their more substantial metaphysical projects. In this chapter I respond to both of these threats. My main focus will be deflationism about truth itself, and to reveal the true nature and commitments of the view. Once these are exposed, I will address methodological deflationism.

3.1 Some Reminders Recall that the deflationist theory of truth is composed of the (D/E) instances: (D) ‘p’ is true iff p. (E) is true iff p. The deflationist conception of truth is composed of the following features: Basicness:

the instances of (D/E) are basic;

Completeness:

the instances of (D/E) explain all the uses of truth predicate;

Purity : the instances of (D/E) are given without making any essential connections between truth and other concepts; Insubstantiality: the instances of (D/E) do not imply that there is a substantive property of being true. In this chapter I will argue that none of the four main features of the deflationist conception of truth is sustainable, and hence there is no distinctive thesis of deflationism about truth. I will suggest that, at most, we have a global deflationary package that 1

See, e.g. Field (1994), Dodd (2013), Wrenn (2015), and Armour-Garb and Beall (2005).

42

Deflationism Revealed

includes a view of truth as a part, but a deflationary theory of truth alone is not separable from this package. We will see that this involves serious commitments about the relationship between language and the world, which undermines the thesis of methodological deflationism that deflationism is an innocuous and neutral theory of truth. As we saw in Chapter 2, the (D/E) instances are key for the deflationist to be able to explain the uses of the truth predicate that are taken to be important. Just as key is the attitude towards the instances expressed in the deflationist conception of truth, as they ensure that the (D/E) instances are not derived from any more basic principles, which may reveal a presupposed substantial nature of truth. A good point to consider in detail, then, is claim that the instances of (D/E) are fundamental. If they are shown not to be, then this is a significant blow to deflationism. We will thus start by evaluating Basicness, and in the course of this investigation explore how the findings in relation to this feature affect the other principles too. As stated above, ultimately we will find that Basicness, Completeness, Purity, and Insubstantiality are all compromised. I will suggest that what we end up with is, at most, a global deflationary package that includes a view of truth as a part, and a deflationary theory of truth alone is not separable from this package. The evaluation of deflationism must thus be an evaluation not just of a theory of truth, but of a larger controversial picture of the relationship between language and the world.

3.2 Examining Basicness There are a number of different explanations offered to show how the instances of (D/E) are derived, and not basic. I will discuss those offered by Tarski (1944), Ramsey (1927), Lynch (2009), and Wright (1992). They all offer slightly different accounts, and I will discuss each in turn.

3.2.1 Tarski’s derivation Tarski begins his influential (1944) paper ‘The Semantic Conception of Truth and the Foundation of Semantics’ by discussing what he calls ‘the classical conception of truth’, the intuitions behind which he initially attempts to express in three different ways: (i) To say of what is that it is not, or of what is not that it is, is false, while to say of what is that it is, and what is not that it is not, is true. (from Aristotle’s Metaphysics, 1011b25) (ii) The truth of a sentence consists in its agreement with (or correspondence to) reality. (iii) A sentence is true if it designates an existing state of affairs. (Tarski 1944: 343) Tarski rejects these as unsatisfactory expressions of the classical view, and looks for something which is, in his terms, “more precise” (Tarski 1944: 343). Note though—and this is important—that he has not abandoned the classical view, just these expressions of it. Here is what he says immediately following this:

examining basicness

43

Let us start with a concrete example. Consider the sentence “snow is white.” We ask the question under what conditions this sentence is true or false. It seems clear that if we base ourselves on the classical conception of truth, we shall say that the sentence is true if snow is white, and false if snow is not white. Thus, if the definition is to conform to our conception, it must imply the following equivalence: The sentence “snow is white” is true if, and only if, snow is white. (Tarski 1944: 343, bold emphasis added.)

He then goes on to offer the generalized schema: (T)

X is true iff p (Tarski 1944: 344).2

I want to pause to discuss the two phrases highlighted in bold in the quote above. Tarski begins by asking for a statement of truth conditions: he starts his quest by asking for the conditions under which a sentence is true. He then answers this question in the spirit of the classical view discussed earlier. As it is stated in the quote above the exact answer to this question is a little ambiguous, but if we move forward a couple of pages we can find a more precise account. Firstly, Tarski calls truth a ‘semantic concept’: I should like to propose the name “the semantic conception of truth” for the conception of truth which has just been discussed. Semantics is a discipline which, speaking loosely, deals with certain relations between expressions of a language and the objects (or “states of affairs”) “referred to” by those expressions. (Tarski 1944: 345) Later on this page he says: it turns out that the simplest and the most natural way of obtaining an exact definition of truth is one which involves the use of other semantic notions, e.g., the notion of satisfaction. (Tarski 1944: 345) Tarski naming his view the ‘semantic conception of truth’, combined with his account of what semantics is, places his proposal firmly within the classical conception he alludes to at the beginning: truth, on this view, involves connections between language and the world. Moreover, he takes it that this connection can be best explained by using other semantic notions, and this is precisely what he does with the notion of satisfaction. This is the idea that the sentence ‘a is F’ is true just in case the object referred to by ‘a’ satisfies the predicate ‘is F’. Using these ideas we can account for the exact answer to the question about truth conditions above. For Tarski, the conditions under which ‘snow is white’ is true will be properly stated as: (Tar) The sentence ‘snow is white’ is true iff the object referred to by ‘snow’ satisfies the predicate ‘is white’.

2

In (T), ‘X’ stands for the name of a sentence, and p stands for the sentence itself.

44 Deflationism Revealed The right-hand side of the (Tar) biconditional gives the proper explication of the right-hand side of the (T) biconditional first stated above. So what we have here is a derivation of the first (T) biconditional from a more complex one, (Tar): Tarski takes the (Tar), satisfaction-involving, biconditional to provide ‘the simplest and most natural’ definition of truth.3 With regard to sentences, the deflationist takes their (D) instances as an alternative statement of the (T) biconditional, not the (Tar) biconditional. However, the (T) biconditional, on Tarksi’s proposal, is not basic. Firstly, it is tied up with the notion of truth conditions, and is offered as a statement of them, and, secondly, the (T) biconditional is not basic, as it is an abbreviation of a more complex principle, (Tar), involving the notions of reference and satisfaction.4 As a consequence, Tarski’s derivation of the (T) biconditional runs contrary to the deflationist’s claim that the (D) instances are basic. In case one is concerned that this just applies to sentences, let me explain how it applies to propositions too. For Tarski, the idea was that the sentence version of the deflationist’s schema: (D)

‘p’ is true iff p

is derived from: (Tar) The sentence ‘snow is white’ is true iff the object referred to by ‘snow’ satisfies the predicate ‘is white’. The basic idea was that we can only make sense of (D) by reference to the semantic features of sentences, including the notions of reference and predicate satisfaction. We can generate a similar issue for propositions as follows. The idea is that we can only understand an instance of the deflationist’s schema for propositions: (E)

is true iff snow is white

by reference to a more general principle, such as: (PTar) The proposition that snow is white is true iff the object denoted by has the property denoted by . (PTar) is an example which uses a particular view of propositions—that propositions are composed of concepts—to generate a similar structure to Tarski’s analysis for sentences. As I mentioned in Chapter 1, the situation with propositions is generally a lot more complicated than the situation with sentences, because there are such vastly different views about what propositions are, and many reject the structured concept-based model on display here. However, I also said there that, if propositions 3 Interpreting Tarski is a controversial enterprise, and I do not pretend that the account given here is uncontroversial. See, for example, Field (1972), Soames (1984), Etchemendy (1988), and Heck (1997) for discussion. 4 Note that later (359–60) Tarski talks about the ‘folk’ notion of truth giving support to initial biconditionals—but note that what he says there is not inconsistent with line above, as issue is presented to people in terms of truth conditions, not simple equivalences.

examining basicness

45

are to be content bearers, any view of propositions needs to make sense of a connection between a proposition and the things it is about (or purporting to be about), and thus some story of this sort is hard to avoid.

3.2.2 Ramsey’s derivation F.P. Ramsey is often seen as an archetypal early deflationist, and the chief proponent of the ‘redundancy’ theory of truth. However, examination of his views on truth reveals that his views are not consistent with the new wave deflationism under discussion here. Like Tarski, Ramsey also takes the (D/E) instances to be derived, but rather than use truth conditions as his starting point, he looks more to truth-makers. He says: Suppose a man believes that the earth is round; then his belief is true because the earth is round; or generalizing this, if he believes that A is B his belief will be true if A is B and false otherwise. (Ramsey 1927/2001: 437)

Ramsey here uses the word ‘because’ to connect a belief and the thing that makes the belief true. As with Tarski’s initial statement, there is a little ambiguity about what Ramsey means here, but it is plausible to take him to be saying that the belief that the Earth is round is true because the Earth has the property ascribed to it, namely being round. This is substantiated by a further remark he makes: “If you think the earth is round, you think truly” and this amounts to no more than that earth has the quality you think it has when you think it is round, i.e. that the earth is round. (Ramsey 1927/2001: 441, bold emphasis added.)

The truth of a belief, on this view, is determined by whether the object of the belief has the property ascribed to it by the belief. A more explicit statement of Ramsey’s claim in the first quote would be: The belief that the Earth is round is true iff the Earth has the property of being round. This can be generalized as follows: The belief that A is B is true iff A has the property of being B, which is abbreviated by Ramsey as (Ba)

The belief that A is B is true iff A is B.

Interestingly, Ramsey takes statements like this, involving what he calls a particular form of ‘propositional reference’ for a belief, to be the more basic statements about truth. He notes that (Ba) cannot be generalized for all beliefs, as many beliefs do not have the simple structure A is B. His solution is to introduce a more general form: ‘the belief “that p” is true iff p’, which is given as a generalization of the specific propositional structures that different kinds of beliefs can have. Once more, the instances of the form of (D/E) are not basic, but derived from a consideration of truth-making, involving commitments to objects and properties. Furthermore, we also get the

46

Deflationism Revealed

derivation of a generalized schema from a generalization of specific propositional structures. Ramsey also did not see that the (D/E) instances were solely about truth; in fact he saw them as expressing a key connection between truth and reference: And whatever the complete definition [of truth] may be, it must preserve the evident connection between truth and reference, that a belief “that p” is true if and only if p. (Ramsey 1927/2001: 443, bold emphasis added.)

3.2.3 Lynch’s derivation Lynch sees the (D/E) equivalences as being derivable from some intuitive thoughts about what he calls the ‘objectivity’ of truth. Lynch’s basic principle is: Objectivity: The belief that p is true if, and only if, with respect to the belief that p, things are as they are believed to be. (Lynch 2009: 8)

Lynch sees this principle as having the best claim to being a basic, intuitive, thought about truth, or ‘truism’ in his words. For propositions, Lynch derives (E) in the following way. If we have a related principle: With respect to the belief that p, things are as they are believed to be if, and only if, p (Lynch 2009: 9) we can combine this with Objectivity to derive instances of The belief that p is true if and only if p. (Lynch 2009: 9) Then, with the claim that the things that are believed are propositions, we can get to The proposition that p is true if and only if p. (Lynch 2009: 9) So Lynch does not see the instances of (E) as basic, he sees them as derivable from basic thoughts about objectivity, combined with some principles connecting beliefs and propositions. Moreover, he sees this as an explanation for why the instances of (E) are ‘compelling’, which Horwich cannot provide. A sentence analogue for Lynch’s derivation would involve expressing Objectivity as Objectivity S:

The sentence, ‘p’, is true if and only if things are as ‘p’ says them to be.

Followed by: Things are as ‘p’ says them to be if and only if p. To yield: (D)

‘p’ is true iff p.

3.2.4 Wright’s derivation Wright suggests that the (D/E) instances can be derived from platitudes about the ‘transparency’ of truth, which he states as:

deflationist responses

47

that to assert is to present as true and, more generally, that any attitude towards a proposition is an attitude to its truth—that to believe, doubt or fear, for example, that p is to believe, doubt or fear that p is true. (Wright 2003: 271)

Wright states the idea behind this derivation as follows: Why does it seem that any competitive account of truth must respect [(D/E)]? Relatedly, why just that starting point for the deflationary conception? The answer, I suggest, is that standing just behind [(D/E)] is the basic, platitudinous connection of assertion and truth: asserting a proposition—a Fregean thought—is claiming that it is true. The connection is partly constitutive of the concepts of assertion and of truth, and it entails the validity of [(D/E)]. (Wright 1992: 23–4)

If we start with ‘to assert is to present as true’ as the basic claim, we can then say that there will always be an equivalence between asserting ‘p’ and asserting ‘p is true’, for to assert ‘p’ just is to assert ‘p is true’, according to the platitude. This suggests, then, that, in assertoric contexts at least, there will be an equivalence between ‘p’ and ‘p is true’ (or between and is true), which we can express in the form of the familiar (D/E) biconditionals: (D) ‘p’ is true iff p. (E) is true iff p. This indeed supports the way that some forms of deflationism are motivated, where the claim is made that to assert that it is true that p is really just to assert that p. Here Wright uses that claim as a starting point and derives the (D/E) instances from it, as opposed to starting with the (D/E) instances and deriving the claim about truth and assertion from those.

3.3 Deflationist Responses Before discussing the potential implications of these derivations, we need to consider how the deflationist responds to them. Let’s start with Tarski and Ramsey, before moving on to Lynch and Wright.

3.3.1 Deflationism versus Tarski and Ramsey’s derivations Both Tarski and Ramsey’s derivations involve the idea that the more general instances of (D/E) are derivable from more specific basic principles. In particular, the following two: (Tar) The sentence ‘snow is white’ is true iff the object referred to by ‘snow’ satisfies the predicate ‘is white’. (Ba)

The belief that A is B is true iff A is B.

In order to respond to these derivations, the deflationist can try to show that the order of derivation should be reversed, making the (D/E) instances more basic than (Tar) and (Ba). This is precisely the strategy Horwich suggests:

48

Deflationism Revealed

An account [of truth] may or may not be compositional—it may or may not define the truth of an utterance or proposition in terms of the semantic properties of its parts. For example, a theory inflated in this way might involve the principle, (T/R) ‘a is F’ is true iff there exists an object x such that ‘a’ refers to x and ‘F’ is satisfied by ‘x’. The minimalist policy is not to deny such principles relating truth, reference, and satisfaction, but to argue that our theory of truth should not contain them as axioms. Instead, they should be derived from a conjunction of the theory of truth and quite distinct minimalist theories of reference and satisfaction. (Horwich 1998: 10)

But why think that the (D/E) instances are more basic than (T/R)? Horwich addresses this in a way that can also be applied to Ramsey. He says: the Tarskian strategy applies only to those sentences whose logical forms may be represented in first order logic. Or, in other words, it applies only to those sentences whose truth values are determined by the truth values of atomic sentences. But there is no reason to assume that every conceivable truth has such a structure. Consider, for example counterfactual conditionals, probability claims, laws of nature, and modal assertions of various kinds. All of these constructions resist formulation in the language of predicate logic, and so it is not clear that Tarski’s theory can be made to cover them. (Horwich 1998: 111)

From this we can see how a counter-argument to the primacy of (Tar) and (Ba) would go. These principles are limited in application: they only account for a particular class of sentences or propositions, but (D/E) is more general, applying to sentences or propositions of all logical forms. If this is the case, should we not see (Tar) and (Ba) as the result of applying the general form of the (D/E) instances to specific cases? In this way, we could see (Tar) and (Ba) as the result of applying the general (D/E) instances to specific kinds of sentences or propositions, and, consequently, as specific principles derived from more general principles. Horwich makes appeal here to the (D/E) instances being of a more general form than (Tar), and hence that we should consider the (D/E) instances to be more basic than corresponding instances of (Tar). It is worth noting that this appeal to generality is significant. We noted earlier that it is controversial whether general forms of (D/E) can be stated, which is why we have been talking about instances, as opposed to a general schema, but, since Horwich raises the issue, we can now discuss it. Interestingly, it is Horwich himself who has the most trouble with stating a general schema, stating that the most we can have is a general “propositional structure” (Horwich 1998: 18) of the form . Moreover, this general structure is only something that can be inferred from the specific instances, as opposed to being the general principle from which the instances are derived. Let us set aside issues about the legitimacy of this move here, and focus on how we can respond to this argument supposing that there are general statements of (D/E). This is an issue that, as we have seen Ramsey anticipates. He gives an account of the process by which we move from the specific principle (Ba) to the general form of (D/E) as follows:

deflationist responses

49

In order to avoid this infinity we must consider the general form of a propositional reference of which all these forms are species; any belief whatever we may symbolise as a belief that p, where ‘p’ is a variable sentence just as ‘A’ and ‘B’ are variable words or phrases (or terms as they are called in logic). We can then say that a belief is true if it is a belief that p, and p. This definition sounds odd because we do not at first realize that ‘p’ is a variable sentence and so should be regarded as containing a verb; “and p” sounds nonsense because it seems to have no verb and we are apt to supply a verb such as “is true” which would of course make nonsense of our definition by apparently reintroducing what was to be defined. But ‘p’ really contains a verb; for instance, it might be “A is B’ and in this case we should end up “and A is B” which can as a matter of ordinary grammar stand perfectly well by itself.” (Ramsey 1927/2001: 437)

Ramsey’s claim here is interesting as it suggests that we cannot even understand the general (D/E) principles without appeal to (Ba). This is because the general form of the (D/E) instances is ungrammatical, and they only make sense when read with a specific instance in mind (his example involves a sentence of the form of (Ba)). If this is the case, then we have grounds for thinking that the particular claims are prior to the general claims, as we have no way of understanding the general claims without filling them in in some specific way. I think that there is some truth to this, and that the only reason we do not see the general (D/E) schemas as peculiar are because we have been inducted into this way of speaking through the use of specific examples (‘snow is white’, in particular, comes to mind). We could thus view Tarski and Ramsey as providing the basic principle for a specific kind of sentence, which can then be generalized to form principles that govern other kinds of sentences, i.e., the instances of (D/E). The fact that (Tar) and (Ba) are not themselves applicable to all kinds of sentences does not necessarily mean that they are not prior to the instances of (D/E): what they do is generate a general formula that can be applied in other cases. Indeed, as we saw above, this is precisely the method Horwich uses when formulating his own theory. On his view, we start with specific axioms, then generate a propositional structure from them which we can use to infer further axioms. In this case, the fact that the axioms are specific claims about one specific proposition and the propositional structure a general claim about all propositions does not mean that the general is prior to the specific; quite the contrary. We will return to this issue below in relation to Lynch’s derivation, to which we now turn.

3.3.2 Deflationism versus Lynch’s derivation Let’s now consider deflationist responses to Lynch’s derivation. We can view Lynch’s derivation as aiming to provide an answer to the question of why the (D/E) instances hold. For example, compare the following to answers to the same question: Q: Why is it the case that: is true iff water is wet? A1: Because a proposition states that things are a certain way, and a proposition is true if and only if things are the way that it states them to be. The proposition states that water is wet, so it is true if and only if water is wet.

50

Deflationism Revealed Q: A2:

Why is it the case that: is true iff water is wet? Because it is an axiom, which is epistemologically fundamental.5

A1 offers an explanation in line with Lynch’s derivation above. It appeals to a more general principle about propositions, truth, and ways things are. It also may involve commitments to objects and properties, depending on how we interpret the words ‘water is wet’ on the right-hand side of the final biconditional stated. For instance, if we adopt (PTar), then we get the following: A1*: Because a proposition states that things are a certain way, and a proposition is true if and only if things are the way that it states them to be. The proposition is (partly) composed of and , and states that the substance denoted by has the property denoted by , so it is true if and only if the substance denoted by has the property denoted by . A2, on the other hand, aims to avoid such commitments, by asserting the basicness of the (E) instances. A similar line can be given in the case of sentences, by holding that (Tar) and (Ba) are able to offer satisfying explanations of the truth of the instances of (D). For instance, consider the following questions and answers: Q:

Why is it the case that: ‘water is wet’ is true iff water is wet?

A3: Because the word ‘water’ refers to a substance, and the word ‘wet’ is a predicate which the substance referred to by ‘water’ is capable of satisfying; and the sentence in quotation marks on the left-hand side of the biconditional states that the substance satisfies the predicate. The claim on the right-hand side of the biconditional is that the substance does in fact satisfy the predicate, thus that things are as the sentence on the left-hand side of the biconditional states them to be, and, given that a true sentence is a sentence that states things how they are, under these conditions, the sentence ‘water is wet’ is true. Q:

Why is it the case that: ‘water is wet’ is true iff water is wet?

A4: Because ‘ “water is wet” is true’ and ‘water is wet’ are cognitively equivalent. This equivalence is basic, in that there is no explanation of why these two sentences are cognitively equivalent. If presented with A3 and A4, it is obvious which is explanatorily superior. A3 gives a clear account of the truth of a specific instance of (D), whereas all A4 can do is assert the basicness of the instances.6 5 An anonymous reviewer suggested a different response for A2: Because is the proposition that water is wet, and part of what it is to be the proposition that water is wet is to be true iff water is wet. I think this is an interesting line, but I do not discuss it further here, as, for Horwich, propositions need to be understood prior to truth, so it cannot be a constitutive part of a proposition that it has a particular truth condition (see Horwich 1998: 16–17). 6 We may also question whether A4 is true, as the reasons given for the lack of equivalence in meaning between the left-hand side and the right-hand side may also show that there is a lack of cognitive equivalence. See the discussion of redundancy theory in Chapter 1.

deflationist responses

51

This speaks to a more general concern with the move of making the (D/E) instances fundamental. In general, an appeal to a primitive, or fundamental, thing is warranted when there is reason to think that a good reductive explanation of that thing is unfeasible, in some way. This is a principle that Horwich himself endorses. When talking about inflationary theories in general he says: The main objection [to inflationary theories] is that none can meet the explanatory demands on an adequate theory of truth. Specifically, none provides a good account of why it is that instances of the equivalence schema are true. (Horwich 1998: 11–12)

This is given as a central reason in favour of deflationism, in that by reversing the inflationary direction of explanation and holding that the (D/E) instances themselves are basic, and using them to explain all that needs explaining, we gain a new and valuable perspective on truth.7 This move to the fundamentality of the (D/E) instances only works if Horwich is right that no such explanations of their truth are available. Our example of A1 above calls this into question, as we have a perfectly reasonable, and informative, explanation of why a specific instance of (D/E) holds. Granted, Horwich can repeat his concern stated above that this will not be a general explanation of why all the (D/E) instances hold, and this is quite correct. However, why suppose that there must be a unified explanation? If, as Ramsey thought, we are already supposing that the (D/E) instances are generalizations of more specific principles, then we should expect the explanations of why different sorts of instances are true to be different. Those who are squeamish about taking this line also have alternative resources here. A correspondence theorist, for instance, can hold that A1/A3 just is the only explanation for the truth of an (D/E) instance, for the only genuine truths are those that are of subject-predicate form. Such a theorist would simply deny that the examples Horwich cites above of counterfactuals and probability statements are the sorts of things that are capable of being true or false. As a consequence, the fact that they have a structure that cannot fit into (Tar)/(PTar) or (Ba) is not an embarrassment; rather it is precisely what one should expect. This tallies with the well-known fact that the instances of (D/E) cannot allow for truth-value gaps. If we have a problematic class of sentences over which there is doubt over whether they are true or false, then one might be better off denying that there are relevant (D/E) instances in this area, as they are in danger of being false. So, it is not clear that there is a good response to the claims that (Tar)/(PTar) and (Ba) are the basic claims from which the instances of (D/E) are derived. If this is so, then the Basicness component is compromised.

3.3.3 Deflationism versus Wright’s derivation Wright claims that the instances of (D/E) are derivable from the platitude ‘to assert is to present as true’. To reply to this, the deflationist either needs to deny the platitude, or 7

For more on this reversal of explanatory direction, see C.D. Wright (2016).

52

Deflationism Revealed

reverse the direction of the derivation to show that the platitude is derived from the instances of (D/E). The first strategy is unappealing, as I don’t think the deflationist would wish to deny this claim. The second strategy, however, has been recently shown by Bar-On and Simmons (2007) to be a non-starter: How might the deflationist respond? Consider disquotationalism or Horwich’s minimalism. According to this kind of deflationary view, the function of “true” is exhausted by its disquotational or denominalizing role. For example, “true” contributes no more than its denominalizing role to an explanation of why true beliefs engender successful action. Now consider the thesis that to assert is to present as true. The thesis invokes a certain concept—that of presenting as true—in a natural explanation of assertion. It also involves the use of the truth-predicate; in Horwich’s terms, it is a fact about “true” that needs to be explained. With the denominalizing role of “true” in mind, a deflationist might claim that the thesis that to assert that p is to present p as true is equivalent to the thesis that to assert that p is to present p. This commits us to the claim that to present p as true is just to present p; for example, to present as true the thought that aardvarks amble is just to present the thought that aardvarks amble. But this claim is false, for there are many ways to present a thought. I can present a thought as worthy of your consideration, or as a conjecture, or as a remote possibility, or as outrageous—and I can also present it as true. Presenting as true is just one way of presenting. So this cannot be the right way to denominalize away “true” in the locution “present as true.” (Bar-On and Simmons (2007): 71)

Bar-On and Simmons show that the deflationist cannot use the instances of (D/E) to derive the platitude that to assert is to present as true, for to do so we would need to begin with a non-truth involving expression of the principle, and then use (D/E) to show how the principle can be stated as truth-involving. The problem is that we cannot have a true non-truth involving initial platitude, as the only form available is ‘to assert p is to present p’, which is false. This means that the deflationist cannot reverse the direction of derivation presented by Wright. As a consequence, they must either reject the platitude or reject Basicness.8

3.4 The Impact The considerations above suggest that the (D/E) instances are derived, and are not basic. As a result, the Basicness feature of the deflationary conception of truth is compromised. What about the other three features of Completeness, Purity, and Insubstantiality? The issues discussed above also cause an immediate threat for Completeness and Purity. In terms of Completeness, if the (D/E) instances are not basic, then we can identify features of truth that are not captured by them, namely those features which are involved in the derivation of the instances. The features of truth outside of the (D/E) instances will be that, in some cases, there is a connection between truth and reference (from (Tar) and (Ba)), and that there is a connection between truth and assertion. Both of these 8 As Bar-On and Simmons (2007) note, some deflationists, such as Brandom, take the first route and deny the platitude, offering a different account of assertion where truth is not invoked. See also Blackburn (2013) for more on this route. I will not discuss this issue further here, but Bar-On and Simmons question whether Brandom has the resources to offer a non-truth involved account of assertion.

insubstantiality preserved?

53

things are features of truth that are not explained in terms of the (D/E) instances, so Completeness falls. It should also be clear that Purity is lost too, as both of these features connect truth to other notions, namely reference and assertion. Given that these notions are involved in what are given as the basic principles about truth, it is hard to deny that there are essential connections between truth and these notions. If this is the case, then Purity no longer holds, as truth cannot be understood as something which has no essential connections to other concepts. These are significant consequences, and cast enormous doubt on the deflationary project as is so far understood. The loss of Basicness shatters the deflationary conception of truth as stated above, and leads to the loss of Purity and Completeness also. This is a significant result if we take the deflationary conception of truth to involve all of  these features, and requires at least a recasting of how deflationists and nondeflationists conceive of deflationism. In Chapter 2, I discussed the Insubstantiality feature, and how we can understand this in such a way as to distinguish deflationism from inflationism at the metaphysical level. If what has been said above is correct, then this may turn out to be the only way to distinguish deflationism from other views, as it is the only feature of the view left. If Insubstantiality can be preserved, then perhaps the central spirit, if not the letter, of deflationism is preserved if we can show that none of the concerns above force the deflationist to give up on Insubstantiality. But can it be preserved?

3.5 Insubstantiality Preserved? If the concerns above are correct, then deflationists are wrong to hold Basicness, Completeness, and Purity, as the (D/E) instances are not basic, and they are derived from principles that make essential connections between truth and assertion, reference, and satisfaction. To knock out Insubstantiality as well inflationists would need to show that these essential connections force the deflationist to abandon Insubstantiality. What the deflationist needs to do is show that these connections do not make truth substantial, because the concepts that are connected to truth are equally insubstantial. Indeed, the idea that concepts like reference and satisfaction are not substantial is a common view favoured by deflationists, who often see a deflationary approach to truth as naturally complementary with deflationary approaches to these other notions. Horwich, for example, writes: The deflationary conceptions of truth, being-true-of, and reference go hand in hand with one another. These notions are interdefinable, so any substantive analysis of one would imply substantive analyses of the others. And any argument for deflationism with respect to one of the notions . . . will be convertible into an argument for deflationism about the other notions too. (Horwich 2004: 74)9 9 Thomasson (2015) is an example of a deflationist who makes the stronger claim that a rejection of deflationism about one of these concepts is inconsistent with a rejection of deflationism about another. She argues for this in a different way to how I set about it above, and also includes existence

54

Deflationism Revealed

What the deflationist could do, then, is hold that the essential connections between truth, reference, and satisfaction do not imply that Insubstantiality must be lost provided those notions are equally insubstantial. What emerges from this is a global deflationism, where all of the ‘semantic’ notions are treated in a deflationary way. In terms of the debate between deflationists and anti-deflationists, what we would end up with is competing worldviews, not just competing views about truth. With Basicness and Completeness substantially compromised, the only way for deflationists to avoid slipping into the rejection of Insubstantiality is to embrace the rejection of Purity. If they accept that truth is essentially related to other notions, such as reference, belief, and assertion, then they are forced to deny that any of these notions are substantial. This enforces a global deflationism, which severs the connections between language and the world, and rejects the idea that there are interesting things to say about these notions in any contexts. Let’s now examine more closely whether the deflationist really can preserve Insubstantiality with this move. Remember that the Insubstantiality thesis holds that truth (and the associated notions of reference and satisfaction) are not substantive properties or relations. As we noted in Chapter 2, to make sense of this claim, we need to have a distinction between substantial and insubstantial properties (and, presumably, of relations too), which I suggested we understand in terms of the distinction between sparse and abundant properties. As Thomasson, a global deflationist, puts it: the semantic deflationist’s view should not be expressed as denying that there is a property of truth or a relation of reference—and that is a good thing. For, as Horwich says: That truth is a property almost goes without saying. We do, after all, distinguish two classes of statement: those that are true and those that aren’t—so truth is what members of the first class have in common. (Horwich 2004: 71) But if the thoroughgoing deflationist comes to say that there is a property of truth or a relation of reference, how is the deflationary view after all to be distinguished from non-deflationary positions? I have alluded to this in my way of expressing the view already: the semantic deflationist need not (indeed should not) deny that there is a property of truth or a relation of reference. But she does deny that there is a deep and substantive ‘nature’ of the ‘truth property’ or ‘relation of reference’ to be uncovered by a philosophical theory, so that it makes sense to seek some reductive view of what truth or reference ‘really’ is. The feeling that there is such a theory to be found is based on mistakenly treating the property of truth on analogy with properties referred to in scientific theories—such as being magnetic or conductive, where such ‘discoveries’ of an underlying nature make sense. (Thomasson 2014: 208)

Thomasson’s words here echo those of Horwich (1998), who says: The main cause of [inflationary] misconceptions [about truth], I suspect, is linguistic analogy. Just as the predicate ‘is magnetic’ designates a feature of the world, magnetism, whose nature is revealed by quantum physics, and ‘is diabetic’ describes a group of phenomena, diabetes, in the package. We will discuss whether her view is sustainable below in Chapter 6, and existence in general in Chapters 6 and 8.

insubstantiality preserved?

55

characterizable in biology, so it seems that ‘is true’ attributes a complex property, truth—an ingredient of reality whose underlying structure will, it is hoped, one day be revealed by philosophical or scientific analysis. (Horwich 1998: 2)

Both Thomasson and Horwich thus make use of a distinction between properties in order to explain and motivate the deflationary view about truth. The question now is whether, on the global deflationary picture we have been discussing, they have the resources to make such a distinction. There is good reason to think that they cannot. Here is Thomasson again, reporting a significant feature of the view: Once we give up [the representational] picture, it seems that semantic minimalism is radical in another sense: it also should lead us to give up much of traditional metaphysics. For without the representationalist view of language, the metaphysician’s traditional questions, such as ‘Are there distinctive objects or properties referred to by our (social, evaluative, mathematical) terms?’ and ‘What are the truthmakers for our (aesthetic, moral, mathematical) claims?’ do not arise. Once we reject representationalism, we cannot even formulate robust metaphysical positions about what ‘really’ exists, what could serve as the truthmakers for our diverse claims, and so on, and are left with a sort of metaphysical quietism (Price 2011: 235–6). The quietist approach may leave us uttering the same truths as the ordinary person does about what numbers, obligations, or beauties there are. But we do so without raising—or purporting to make any sense of—deep metaphysical questions about whether the relevant terms really refer, whether the statements have truthmakers, or whether the objects described really exist—where the ‘really’ is supposed to avert to some deep and substantive criteria that might fail to be met, even when the standard conditions for reference given by the rules of use for the terms are satisfied. (Thomasson 2014: 203–4)

Thomasson makes reference to existence here too, which we will discuss further in Chapters 6 and 8, but let’s now focus on the rejection of the representationalist picture. The idea is that, once we accept the global deflationist picture, many (if not all) metaphysical debates dissolve. Now, one such metaphysical debate is the debate about the nature of properties, and whether properties are substantive or insubstantive (sparse or abundant, in our terms). This debate is deflated by global deflationism, as the existence and nature of properties is settled by a simple principle about properties that we have already encountered in Chapter 1: that whether there is a property of being F is exhaustively settled by the answering the question of whether there is a predicate ‘is F’ which has an extension. Once again, as Horwich himself puts it: That truth is a property almost goes without saying. We do, after all, distinguish two classes of statement: those that are true and those that aren’t—so truth is what members of the first class have in common. (Horwich 2004: 71)

If this is the case, though, and the nature of properties is settled in a deflationary way, then what grounds do we have for being able to distinguish between substantive and insubstantive properties? This is a distinction which the deflationist needs to maintain Insubstantiality, as it is crucial to the contrast drawn between properties like truth and properties like magnetism and diabetes. However, as we saw in Chapter 2, this is itself a

56 Deflationism Revealed metaphysical distinction, and, according to the terms of global deflationism, the possibility of making such metaphysical distinctions is precluded, as indeed is much of metaphysics. For example, let’s think about how we distinguished substantive from insubstantive properties in Chapter 2. In order to draw a contrast between a property like being magnetic and a property like being true, the deflationist needs to be able to say that the property of being magnetic is sparse, in that it grounds genuine similarities and has a causal-explanatory role. These are features of the property that are not captured by the idea that properties are just what predicates refer to, or that properties are the extensions of predicates. What this suggests is that there is a different relationship in the case of being magnetic between predicate and property: there is a gap between the real, substantive, property and the predicate used to express it. In other words, the property is ‘out there’, doing work separate from our predications of it, and it cannot simply be explained in terms of being the product of predication. In particular, the fact that we have the predicate ‘is magnetic’ is because of some significant interactions between the property of being magnetic and our language and thought, in that being magnetic is some feature of the world which we come to observe and form words and thoughts to describe. These ideas lend themselves to the idea that there is a substantive nature to the property of being magnetic, for it is not something that can be fully accounted for by the extension of a predicate. Moreover, it suggests that there is a more substantial relationship of reference between the property and the predicate: the reason the predicate ‘is magnetic’ refers to the property of being magnetic is not because the property is a product of the predicate (or the predicate’s extension), it is because of some significant relationship between the property of being magnetic and our thought and talk. Indeed, this is precisely the sort of thing Horwich himself says about the difference between the relationship between predicate and property in the case of magnetism as compared to the case of truth: Unlike most other predicates, ‘is true’ is not used to attribute to certain entities (i.e. statements, beliefs, etc.) an ordinary sort of property—a characteristic whose underlying nature will account for its relations to other ingredients of reality. Therefore, unlike most other predicates, ‘is true’ should not be expected to participate in some deep theory of that to which it refers—a theory that articulates general conditions for its application. Thus its assimilation to superficially similar expressions is misleading. The role of truth is not what it seems. (Horwich 1998: 2)

So, this is what the deflationist aims to say to distinguish substantial properties like being magnetic from insubstantial properties like being true. However, the deflationist cannot help herself to this explanation. This is because, as we have seen above, she cannot recognize a more substantial relationship between predicates and properties than the idea that properties are just the extension of predicates. The deflationist also cannot recognize the more substantial relation of reference needed to develop the nature of the relationship between the predicate ‘is magnetic’ and the property of being

where now for deflationism?

57

magnetic, for to do is to acknowledge that, in some cases at least, there are substantial connections between language and the world captured in the way that the predicate refers to the property.10 What this uncovers is an inconsistency with regard to the Insubstantiality feature.11 Global deflationism involves the rejection of metaphysical investigations, including those into the nature of properties, and how we distinguish substantial from insubstantial properties.12 However, the Insubstantiality feature of deflationism about truth requires such a metaphysical distinction in order to preserve what is now the only key feature of deflationism about truth left. Given that deflationism about truth entails global deflationism, the only option is to reject the Insubstantiality thesis. The upshot of this is that the deflationist is unable to state the metaphysical aspect of the view about truth: the deflationist cannot claim that truth is different from the other types of properties she wants to contrast truth with, for to do so would be to make metaphysical distinctions the deflationist cannot make. The only option is to hold that the deflationary approach holds for all properties, which means that this is now not simply a thesis about truth, or even about semantic properties in general: it is a global thesis in every sense, as it covers all properties, including those ‘scientific’ properties that the deflationist originally wished to contrast truth with.

3.6 Where Now for Deflationism? The loss of Basicness, Purity, and Completeness show that there is no such thing as deflationism about truth per se: there is a global deflationist view, of which a deflationary approach to truth is a part. The loss of Insubstantiality makes it very difficult to 10 It has been suggested to me that the deflationist could avoid a commitment to substantive properties by instead claiming that that the notion of a substantive property is something like a useful fiction: a device used to understand the deflationist’s view of truth by contrasting the deflationary approach to truth with the more substantive view of properties that is often taken of properties like magnetism. According to this move, the deflationist would then claim that it turns out that there are no substantive properties, even in cases like magnetism, though we can hypothesize about what they would be like for theoretical purposes. There are a few things to note here. The first is that, as the quotes above show, this is not what deflationists do say, and would thus be a revision to the standard deflationary position, which does make commitments to the existence of substantive properties. The second is that there are questions about whether this is still too much metaphysical wrangling for a committed deflationist to be involved in. The third is that such a response would still undermine the idea that deflationism makes a distinctive claim about truth (or indeed semantic notions in general), for on this view it would remain the case that all existent properties (including both truth and magnetism) are of the same kind: merely the extensions of predicates. Consequently, the deflationist’s desired contrast between truth and properties like magnetism and diabetes would not be preserved, and the argument in the text that deflationism about truth yields global deflationism would not be affected. Thanks to Brad Armour-Garb and the audience at my Joint Session talk on methodological deflationism at the University of Edinburgh in 2017 for discussion here. 11 One may wish to compare the inconsistency discussed here with the inconsistency discussed by Boghossian (1990) in relation to deflationism and truth conditions. 12 Note that, whilst I have developed this point with reference to the sparse/abundant distinction, it applies to the other attempts to ground the substantive/insubstantive distinction discussed in Chapter 2 too. This is because the problem is structural: the substantial/insubstantial distinction itself presupposes an ability to state what substantial properties are, which is against the ethos of global deflationism.

58

Deflationism Revealed

state what exactly deflationism is from a metaphysical point of view, for its very nature is to reject metaphysical investigations and distinctions. This rules out the standard way the view is characterized (in terms of Insubstantiality), and reveals the truly global nature of global deflationism: it must apply to all properties, predicates, and concepts, not only semantic ones like truth. Deflationism is thus not just a view about truth: it is an all-encompassing approach to the way language relates to the world. This reveals the problems with the way that deflationism is often presented: as an innocuous theory of truth, which is neutral on all substantial issues, as it takes what allegedly is common amongst all truth theorists (the (D/E) instances) and aims to explain everything that needs to be explained in terms of them. This breeds a methodological position—‘methodological deflationism’—which states that, due to this innocuousness and neutrality, deflationism should be considered the default view of truth.13 The discussion above shows that methodological deflationism gives a gross misrepresentation of the situation. First of all, the way the deflationist approaches the (D/E) instances is very different to how other theories approach them, with deflationists holding that they are basic, and other theories holding that they are derived, and I have given reasons above to doubt the deflationist’s claim of their basicness. Secondly, deflationism about truth is not innocuous: it is part of a larger, and controversial, picture of language. This needs to be taken into account when considering what view of truth to take. It is key to emphasize, for any prospective truth theorist, that the deflationary view of truth enforces the radical view about language and the world. Deflationism about truth cannot be sold as an innocuous theory of truth, for accepting that theory is tantamount to accepting a radical view about language in general, and its relationship to the world. This also shows that it is not a neutral theory of truth, for it does take a stance on one of the key issues, namely whether or not language is representational. As we have seen above, it is committed to the view that language is not representational. How, exactly, then are we to characterize deflationism? This is where we become in danger of sinking in a terminological quagmire: Engel (2002) interprets it as a form of pragmatism, Thomasson (2015) considers it a form of “simple realism”, David (2015) states that deflationists can be “metaphysical realists”, Price (2011) sees a connection between deflationism and quietism, the view that realism/anti-realism debates are fruitless. If what has been said above is correct, then deflationism is not compatible with metaphysical realism in any meaningful sense that holds that there are connections between our language to a metaphysically real world. ‘Simple realism’ is merely a commitment to the view that entities of a certain disputed class exist, which is in contrast to the view that they do not exist. This is not a particularly helpful characterization, though, as this is also compatible with pragmatism, which, as we will see, does not deny the existence of the entities in question, and also holds that there is little to say 13

See, e.g. Field (1994), Dodd (2013: 301), Wrenn (2015), and Armour-Garb and Beall (2005: 29).

summary

59

about them. This is also not too distinct from quietism. Indeed, if we interpret quietism to be a dismissal of more traditional substantive realist/anti-realist debates about the nature of entities, in particular whether they are mind-independent or not, then simple realism is a form of quietism. Whether the word ‘realism’ should thus be used to characterize the view is thus questionable, as it is a little misleading. If we must choose a label, the best way to think of global deflationism initially is as a form of pragmatic quietism. It is pragmatic, as it is anti-representationalist, and it is quietistic, because, unlike other forms of pragmatism, it has nothing to say about the connections between language and the world. As we will see, some forms of pragmatism do think that there are connections between language and the world, and that language— in a sense to be explained—generates the world. In Chapter 6 I will argue that global deflationism ultimately collapses into this non-quietistic form of pragmatism.14 Where does this leave the notion of ‘inflationism’ about truth? There is undoubtedly a sense in which the very terminology of the deflationism/inflationism debate favours the deflationist. This is because it suggests that there are substantial differences in the kinds of commitments each view takes on, with the deflationist pitching themselves as the most minimal alternative to every other view. Now that this aspect of deflationism has been revealed, inflationists may wish to reject the label ‘inflationism’, and focus instead on the pre-existing distinction between views of truth that are representational in character (correspondence theories of truth), and those that are not (pragmatist and coherence theories of truth). This is not to say that the distinction between sparse and abundant properties falls away; on the contrary, we will see that it has some key roles to play.

3.7 Summary My main aim in this chapter has been to show deflationism for what it really is: to clarify its true nature and commitments, and undermine some of its motivations. Given that deflationism does not enjoy a methodological advantage over other views, in deciding whether or not to adopt deflationism, one must take into account the implications of doing so for a variety of other positions one might hold, and recognize that a commitment to deflationism about truth entails a commitment to global deflationism. I will argue in Chapter 5 that global deflationism is inconsistent, and in Chapter 6 that global deflationism is unable to deliver on even its minimal explanatory promise, but to get to that point we will need to discuss some further issues, so I will defer these arguments until then. Let’s now think about how language relates to the world.

14 Deflationists could also try denying that truth is a property, which would be to retreat to the ultra deflationary views identified in Chapter 1. As we saw, it was not clear that we could maintain a denial that truth is a property on those views, plus those views come with problems of their own.

4 Language–World Connections Deflationism has been shown to throw a “metaphysical wet blanket”1 over all interesting metaphysical discussions, not just those about truth. In the next few chapters I will develop a picture of the relationship between language and the world, and show how truth plays a key role in it. The issue I want to explore in this chapter is the relationship between predicates and properties. We touched on this in Chapter 2 in the discussion of sparse and abundant properties, and this distinction will play a role here. My main interest is in exploring the different kinds of predicates that we find, and seeing if there are differences between the ways that predicates of different kinds relate to their corresponding properties. This is part of a more general interest in the relationship between language and the world: does language always respond to the world, or does it also, in some cases, generate the world? I will begin by discussing the idea that predicates come in different kinds, and discuss the ways that different kinds of predicates can be individuated. I will then develop two different models of the relationships between predicates and properties: the responsive model, involving sparse properties; and the generative model, involving abundant properties. I will suggest that whether a model fits a kind of predicate should be decided on a case-by-case basis, and give examples of kinds of predicates for each model. After briefly noting some parallel issues for singular terms and objects, which will be explored in more detail later on, I close the chapter by defining the notion of a domain, and discuss how sentences are assigned to specific domains. In Chapter 5 I will explore the role that truth plays in all this, but my aim here is to provide a picture that is useful from a number of different points of view, as distinguishing between different kinds of predicates is a project that has broad appeal.

4.1 Kinds of Predicates We talked in Chapter 1 about predicates being linguistic items, typically of the form ‘is F’, which, when combined with a singular term, form a sentence. We also noted that predicates typically refer to properties. There are many different predicates that we use

1 This phrase is attributed to Gareth Evans by Wright (1992: 205) in relation to rule-following considerations.

kinds of predicates

61

to describe many different features of things, and my aim in this section is to substantiate the claim that predicates come in different kinds, and that predicate kinds are distinguished by the kinds of functional roles that predicates have. Distinguishing between kinds of subject matters is something that most philosophers do, even if implicitly, in defining what they primarily work on. For instance, a philosopher of mathematics works on different issues to an ethicist, and we need some way of explaining what these differences are. One such way is to say that the ethicist focuses their attention on various issues to do with moral predicates, whereas the philosopher of mathematics focuses on various issues to do with mathematical predicates. This requires some account of what distinguishes moral from mathematical predicates. More substantially perhaps, some philosophers want to take specific views about specific sorts of subjects that do not necessarily entail the same view elsewhere. For  instance, one may want to be an expressivist about morals, a Platonist about mathematics, and a scientific realist about physics. To hold these views together one needs to say what marks these subject areas out as different. To give a sense of the general strategy I will propose, think of how we would ordinarily aim to distinguish, say, moral properties from physical properties. This is a challenge faced by anyone who advocates either a reductive or non-reductive account of moral properties. The reductivist needs to identify a category of properties that are to be reduced, and the non-reductivist needs to identify a category that will be taken to be sui generis. The general thought is that there is something distinctive about moral properties that renders them distinct from physical properties. One feature might be that moral predicates pick out normative properties, where the recognition of an action or object having a certain moral property entails some reason to act or not to act on that basis. This would be distinct from a physical predicate, which perhaps ascribes a property that has causal, but not normative, powers. Moral predicates will ascribe properties whose recognition gives one a distinctly moral reason to act or not to act. What exactly is meant by ‘moral’ here could perhaps be cashed out in terms of general platitudes concerning what we take to fall under the umbrella of the word ‘moral’. To illustrate what I mean here, it would be useful to quote an example of the sort of approach I have in mind. Here are some words from Simon Blackburn, at the start of his book Ruling Passions: Ethics is about how we live in the world. It separates the things we will do gladly from those we will not do, or not do without discomfort. It classifies those situations we aim for, and those we seek to avoid. It is displayed in our attitudes towards ourselves, such as pride, or self-satisfaction, or guilt, or shame. It is also displayed in our attitudes to others: whether they behaved well, or  went the extra mile, or did their bit, carried their burdens, lived or died in ways we admire. Our ethics is shown in the things we forbid, or tolerate, or require, It is shown in the reactions we have as we think of the characters of people, or the events they cause, or the nature of the societies they form. . . . Ethics puts pressure on our choices, and we use ethical considerations to guide the choices of others. The practical role of ethics is what defines it. This is what ethics is for. (Blackburn 1998: 1)

62

Language–World Connections

This passage continues for a few further pages, which I won’t quote, but the idea I hope is clear: whilst we might not be able to state in a few neat and clean principles what the nature of a certain kind of subject-matter is, we can give a detailed description if need be. Included in this description will be the use of certain predicates distinctive to that subject-matter, and the description will tell us how those predicates function, what features they have, and what roles they play. In a similar vein, Michael Smith (1994) offers a set of platitudes which he takes to serve as analyses of moral concepts. Here is what he says about them: To say that we can analyse moral concepts, like the concept of being right, is to say that we can specify which property the property of being right is by reference to platitudes about rightness: that is, by reference to descriptions of the inferential and judgemental dispositions of those who have mastery of the term ‘rightness’. (Smith 1994: 39)

Smith goes on to list platitudes in the categories of practicality, objectivity, supervenience, substance, and procedures. I will just quote what he says about the ‘substance’ category: For example, there are platitudes concerning the substance of morality: ‘Right acts are often concerned to promote or sustain or contribute in some way to human flourishing’ (Foot. 1958); ‘Right acts are in some way expressive of equal concern and respect’ (Dworkin, 1977: 179–83; Kymlicka, 1989: 13, 21–9; 1990: 4–5), and the like. What these platitudes about substance force us to admit, at the very least, is that there are limits on the kind of content a set of requirements can have if they are to be moral requirements at all, as opposed to requirements of some other kind (Dreier, 1990). (Smith 1994: 40)

As with Blackburn, Smith offers us a detailed account of the kinds of features moral properties have, which demarcates them from other kinds of properties. This process is explicitly carried out by reference to the relevant concepts of these properties, but, for our purposes, we can interpret Smith’s thoughts in relation to moral predicates. Moral predicates are of course just one case: similar accounts could be constructed for predicates of various sorts. For example, here are some remarks from Searle on institutional predicates: I get up in the morning in a house jointly owned by me and my wife. I drive to do my job on the campus in a car that is registered to both of us, and I can drive legally only because I am the holder of a valid California driver’s license. On the way, I illegally answer a cell phone call from an old friend. Once I am in my office the weight of institutional reality increases. I am in the Philosophy Department of the University of California in Berkeley. I am surrounded by students, colleagues, and university employees. I teach university courses and make various assignments to my students. The university pays me, but I never see any cash because my pay is deposited automatically into my bank account. After my lecture I go to a restaurant and I use my credit card to pay the bill. When I get back to campus, I telephone my insurance agent about my homeowner’s policy, and I also call my travel agent to arrange airline tickets for an invited lecture at a professional society. I accept an invitation to a dinner party . . .

kinds of predicates

63

All of the italicized expressions in the previous paragraph refer to institutional reality in its various aspects. Institutional facts range all the way from the informality of friendship to the extreme legal complexities of international corporations. The simplest test for whether a phenomenon or fact is genuinely institutional is to ask, Does its existence imply deontic powers, powers such as those of rights, duties, obligations, requirements, and authorizations? There are (uncodified) rights and obligations of friendship and dinner parties, just as there are (codified) rights and obligations of citizenship and employment. There are deontologies without institutional facts (I am, for example, under a moral obligation to help people who are in desperate need of immediate help and whom I am able to help), but there are no institutional facts without some form of deontology. (Searle 2010: 90–1)

In this passage Searle offers some general thoughts on how we identify distinctly institutional predicates. As with Blackburn and Smith in the moral case, no precise rule is given, but some general considerations are offered which chime with our general ability to separate the institutional from the non-institutional. The suggestion is that we can individuate kinds of predicates in accordance with the general functional roles that those predicates are taken to have. These are intended to mark fairly intuitive distinctions between kinds of subject-matter. The fact that we can often individuate different kinds fairly easily does not mean that there will not occasionally be difficulties in demarcating them. For example, compare religious predicates with moral predicates. Both are concerned with describing how to live well, and how to behave towards others and ourselves. In this way, they have similar functional roles, which may lead one to think that they are essentially the same. It is not much of a stretch to imagine that someone who has been brought up in a religious environment with no study of moral philosophy may come to their first philosophy class with the impression that moral predicates just are religious predicates. Perhaps for instance they think that the predicates ‘is right’ and ‘is wrong’ are equivalent in meaning to the predicates ‘is pious’ and ‘is sinful’, as they have no experience of morality in a secular sense, and only have experience of morality in a religious sense. However, with some philosophical work we can prise apart the religious predicates from the moral predicates. The predicates ‘is pious’ and ‘is sinful’, for example, concern a person’s relationship with God, and to hold that ‘is right’ and ‘is wrong’ are equivalent to those terms is just to hold that actions are right or wrong insofar as they affect an individual’s relationship with God in a particular way. However, we can show that ‘is right’ and ‘is wrong’ have a broader role to play in that we can consider the rightness or wrongness of actions under different background views, which show that the meanings of the predicates ‘is right’ and ‘is wrong’ are not tied to a person’s relationship with God, unlike ‘is pious’ and ‘is sinful’. A more detailed example of this method of predicate individuation in action is race and gender predicates. There are those who have held2 that these predicates are 2

Amongst others, philosophers such as Kant (1777/2007), and, arguably, Hume (1754) held this view.

64 Language–World Connections biological predicates, which distinguish different kinds of human individuals. On this view, race and gender predicates mark genuine, objective, distinctions between people, and play an explanatory role when considering the characteristics an individual has. The opposing position—which I take to be the orthodox position—is that these predicates are not biological predicates, even if they have been masquerading as such.3 Instead, they are social predicates, which are expressive of social categories that people have been placed into, and usually carry implications of privilege or oppression.4 We can use the machinery discussed above to provide a justification of why race and gender predicates are social predicates and not biological predicates. Recall that we said that a kind of predicate is demarcated by its functional role. When we compare the roles of biological predicates and social predicates, we will see that they have different functions. Part of the functional role of biological predicates is to demarcate the key characteristics of individual organisms and relate those characteristics to other organisms in order to discern different kinds in nature. This classification procedure is at least intended to be carried out without imposing any sort of privilege or subjugation of different kinds of organisms. Social predicates, on the other hand, are specifically concerned with how human beings act and behave both towards themselves and each other. They are concerned with describing and explaining power relations between different groups of people, and how these have developed in various cultures. When we consider race and gender predicates, we have to consider what sort of functional role they have, and, accordingly, whether they fit the category of biological predicates or social predicates. The argument for them being biological predicates would claim that they are expressive of genuine biological distinctions between organisms, perhaps pointing to gender predicates being expressive of some distinctions in terms of biological sex, and race predicates being expressive of some distinctions in terms of geographical origin. The case for this is highly questionable. As Appiah puts it with regard to race: Indeed, it turns out that, in humans, however you define the major races, the biological variability within them is almost as great as the biological variation within the species as a whole: put another way, while there are some characteristics that we are very good at recognizing—skin color, hair, skull shape—that are very unevenly geographically distributed, the groups produced by these assignments do not cluster much for other characteristics. (Appiah 1994: 95)

The argument for race and gender predicates being social predicates thus focuses in part on the dissonance between the functions of these terms and the biological distinctions with which they are associated. In the case of gender predicates, it is worth noting that the characteristics associated with the predicates ‘is a man’ and ‘is a woman’ often

3 There is also the view that they are failed biological predicates, which motivates the ‘racial skepticism’ of Appiah (1994, 1995), and Zack (2002). 4 See, e.g. Haslanger (2000) for the idea that oppression and privilege must be incorporated into the definitions of these predicates.

kinds of predicates

65

involve the roles that the bearers of these properties play in a social structure, as opposed to their particular biological features. As Simone de Beauvoir famously said, “One is not born, but rather becomes a woman” (Beauvoir 1973: 301). Haslanger elaborates on this idea as follows: Very roughly, as the slogan goes, gender is the social meaning of sex. The idea is that gender is not a classification scheme based simply on anatomical or biological differences, but marks social differences between individuals . . . To help understand this, consider, for example, the category of landlords. To be a landlord one must be located within a broad system of social and economic relations which includes tenants, private property, and the like. It might have been that all and only landlords had only four toes on their left foot. But even if this were the case, having this physical mark is not what it is to be a landlord. Being nine-toed is an anatomical kind; being a landlord is a social kind. Similarly, we can draw a distinction between sex and gender: sex is an anatomical distinction based on locally salient sexual/reproductive differences, and gender is a distinction between the social/political positions of those with bodies marked as of different sexes. (Haslanger 2012: 249)5

In addition, with both race and gender predicates, we find that they are intertwined with issues of privilege and oppression. Indeed, Haslanger also claims that a central reason to pursue philosophical inquiry into these issues is “The need to identify and explain persistent inequalities between females and males, and between people of different ‘colors’; this includes the concern to identify how social forces, often under the guise of biological forces, work to perpetuate such inequalities” (Haslanger 2000: 36). These factors should tell us that we are not looking for some purely objective distinction between persons, and suggests inquiry in the direction of the thought that these terms are used to express power relations in human cultures.6 We have seen, then, that this method of looking at the function of predicates is useful when considering debates about kinds of predicates. Some examples of kinds of predicates are given in Table 4.1:7 I have focused on a specific aspect of a predicate’s function in general, which is its marking of certain features. I focus on that here just to keep one aspect of a predicate constant while highlighting differences between kinds of predicates. I do not wish to deny that some predicates, such as moral and aesthetic predicates, may also have an evaluative or expressive component, and that this may be a key feature of the kind of predicate that they are. I do not, though, think that this conflicts with the

5 Note also that in making the distinction between sex and gender Haslanger does not deny that the notion of sex itself might be socially constructed. See Haslanger 2012: 249, note 3. 6 Note that some see race and gender predicates as institutional predicates, as opposed to social predicates (see Mallon 2006: 536). The issue of whether there is a distinction between social and institutional predicates is an interesting one that I will not take up here, but we can note that such debates provide further support for the general framework suggested here. 7 Some of the statements of functions and examples below may be controversial, but the general point is that distinctions can be made between different kinds of predicates, even if there is room for debate about exactly how these distinctions are drawn.

66

Language–World Connections

Table 4.1. Examples of kinds of predicates Kind of predicate

Function

Moral

To mark the things we ought to do from the things we ought not to do; Institutional To mark features to do with duties, rights, and obligations in one’s social environment.8 Arithmetical To mark features of numbers. Biological To mark features of living things and their components. Chemical To mark features of molecules and relationships between molecules. Physical To mark features of fundamental phenomena, such as matter, mass, and force. Comic To comment on humour-related responses to a thing or an event Aesthetic To mark a particular kind of the sensory features of an object. Religious To describe God and human beings’ relationship to God. Social To mark features of interactions between persons and groups, particularly those involving hierarchy and privilege. Modal To mark what could be from what could not be.

Examples ‘is right’, ‘is wrong’ ‘is illegal’, ‘is a corporation’, ‘is married’ ‘is prime’, ‘is divisible’ ‘is a catalyst’, ‘is an enzyme’ ‘is metallic’, ‘is acidic’ ‘is a proton’, ‘has negative charge’ ‘is funny’, ‘is hilarious’ ‘is beautiful’, ‘is sublime’, ‘is ugly’ ‘is pious’, ‘is sinful’, ‘is holy’ ‘is working class’, ‘is white’ ‘is possible’, ‘is necessary’

idea that these predicates also mark distinctions between things; indeed, for those who take the evaluative aspects of these predicates seriously, the marking of distinctions may be a product of these evaluative aspects. This issue will be discussed further below. The general method discussed above is applicable to any kind of predicates where a set of predicates is taken to form a kind. Moreover, whilst I will later argue for truth pluralism, this division of predicates into kinds is not intended to be a pluralist-specific idea: it is an account open to theories of any kind, or anyone who wishes, for any reason, to make distinctions between kinds of predicates. As I said above, there may be many different reasons why we might want to make these distinctions, and this account is open to all. The same goes for the way I develop the metaphysical issues below, and the accounts of domains and domain-individuation.

4.2 Predicates and Properties What we have said above in terms of the individuation of predicate kinds is neutral on the substantial metaphysical matter of whether there are any properties that correspond 8 Note that this function is hard to specify in a way that does not appeal to the notion of an institution! The same goes for the function of aesthetic predicates below.

predicates and properties

67

to these predicates, and, if so, what those properties are like. For every predicate kind F that is individuated, the following three options are available: (a) There are objective F properties. (b) There are projected F properties. (c) There are no F properties. Option (a) holds that there are objective properties ‘out there’, so to speak, and this can explain why we have predicates of a certain kind: they are responsive, in some way, to existent properties. Option (b) holds that there are properties that correspond to the relevant predicates, but these properties exist because they are projections of the predicates, as opposed to being already existing things that our predicates correspond to. Option (c) would be the error-theoretical option, where, perhaps due to the specification of the predicate kind, or alternatively because the world is uncooperative, there are no properties that correspond to the predicates of a certain kind. This, for example, would be the view of an error theorist with regard to morality, where there are no properties to correspond to our moral predicates.9 The focus below will be on (a) and (b), but I wanted to say a few words about (c) before moving on. In terms of the error-theoretical option mentioned above, one might think that ending up in this position shows that there is something defective about the way the predicates have been understood, and one’s job is thus to revisit this issue. For example, in the moral case, Mackie’s (1977) queerness argument suggested that there was a conflict in the way moral predicates are understood: they are supposed to both mark objective features of the world and be action-guiding, but how can anything be both? Mackie’s own answer was would be not to settle with the view that there are no moral properties, but instead to revisit the account of moral predicates and revise it until this conflict is resolved. An alternative route to (c) would be to take the view that moral predicates have no descriptive function at all (perhaps in response to Mackie’s argument), just an expressive function, and, as a result, there are no moral properties. However, as I noted in the discussion of expressivism in relation to the performative theory of truth in Chapter 1, this rejection of properties does not follow immediately from the claim that a predicate has an expressive function, and some expressivists do accept that there are moral properties. Such a position would not, in my view, lead to option (c), but rather to option (b), where moral properties are simply the projection of moral predicates. So, the two main motivations for taking option (c) ultimately lead back to taking option (a) or (b). There is a third motivation which may represent a more stable case where (c) applies, which would be a case where a predicate is explicitly intended to be responsive, but there just are no properties that it responds to. That is, a mistake is made in the formation of the predicate in thinking that it responds to something which turns out not to be there. An example here would be supernatural predicates, such as ‘is a witch’ or ‘is 9

Examples of error theory in morality are Mackie (1977) and Joyce (2001).

68 Language–World Connections ghostly’. These predicates are intended to be responsive to certain things—supernatural entities—but if there are no such things then there are no properties for supernatural predicates to refer to. In light of this one could revisit the status of the initial predicates and view them differently—perhaps as terms referring to fictional entities—but one need not do so.10 Unlike in the moral case above where there is pressure to do this in light of the function such predicates play, one may simply decide to leave supernatural predicates as defective responsive predicates, with no properties to refer to. In terms of (a) and (b), we can illustrate the differences between them using the following principle about properties: (P) The object referred to by ‘a’ falls under the predicate ‘F’ iff the object referred to by ‘a’ has the property referred to by ‘F’. Option (a) holds that there is a right-to-left order of determination on this biconditional: it is because A has the property of being F that A falls under the predicate ‘is F’. For example, it is because this rod has the property of being metallic that this rod falls under the predicate ‘is metallic’. We have a property-to-predicate direction of explanation. Option (b), on the other hand, takes the reverse view: there is a left-to-right order of determination on the biconditional as it is because A falls under the predicate ‘is F’ that A is F. One example here is the property of being cool:11 motorbikes have the property of being cool because motorbikes fall under the predicate of ‘is cool’, rather than vice versa. We have a predicate-to-property direction of explanation. These two ways of looking at properties broadly correspond to the distinction between sparse and abundant properties that we discussed in Chapter 2, with sparse properties being those under option (a), and abundant properties being those under option (b). Here we are discussing predicates, so I am going to use the terminology of responsive predicates and generative predicates. Broadly speaking, responsive predicates respond to sparse properties, and generative predicates generate abundant properties.12 This framework is neutral with regard to the specific accounts that should be given of a predicate-kind in terms of whether that predicate should be considered descriptive or expressive. For instance, it is compatible with the framework above that one be an expressivist anti-realist about morals. On this view, moral predicates would be generative, and generate abundant properties that would just be the extensions of moral predicates. This does not mean that the predicates could not be understood in 10 If the temptation to take this option for supernatural entities is too great, consider also the example of phlogiston. Thanks to an anonymous reviewer here. 11 See Haslanger (2012: 89–98) for an extended discussion of coolness. 12 Note that this terminology still applies if we are thinking about properties as classes, and classes as mind-independent. This is because, even if there is a vast number of classes, we still need to make sense of a predicate selecting a particular class, and thus having the particular extension it does, which will be dependent on our practices.

causal-explanatory role

69

expressivist terms: what would be understood in expressivist terms would be the application conditions for the predicate, i.e. how the extension of the predicate is determined, which is consistent with the predicate having an extension, and hence yielding a property.13

4.3 Causal-Explanatory Role Before continuing further, I want to examine one feature of sparse properties in more detail, as it is important in what follows. We mentioned earlier the idea that sparse properties have a causal-explanatory role, and I wanted to put more flesh on the bones of this thought.14 To explain the idea in more detail we can appeal to Wright’s (1992) notion of width of cosmological role. Wright defines this notion as follows: Let the width of cosmological role of a subject-matter of a discourse be measured to the extent to which citing the kinds of states of affairs with which it deals is potentially contributive to the explanation of things other than, or other than via, our being in attitudinal states which take such states of affairs as object. (Wright 1992: 196)

Here is Wright explaining the idea in relation to moral properties: Let’s try an exercise. Compare the Wetness of These Rocks, and the Wrongness of That Act. Reference to the wetness of the rocks can, uncontroversially, contribute towards explaining at least four kinds of thing: (1) My perceiving, and hence believing, that the rocks are wet. (2) A small (prelinguistic) child’s interest in his hands after he has touched the rocks. (3) My slipping and falling. (4) The abundance of lichen growing on them. The wetness of the rocks can be ascribed, that is, each of four kinds of consequence: cognitive effects, precognitive–sensuous effects, effects on us as physically interactive agents, and certain brute effects on inanimate organisms and matter. By contrast, the wrongness of that act, although citing it may feature in a vindicatory explanation of my moral disapproval of the action, and hence of the further effects on the world which my disapproval may generate, would seem to have no part to play in the direct (propositional-attitude unmediated) explanation of any effects of the latter three sorts: precognitive–sensuous, interactive, and brute. (Wright 1992: 197)

The key idea here is that, for a property to have a broad cosmological role, it must feature in explanations of things that cannot be explained by reference to our beliefs that things have that property. If a property does not have this feature, then it will have a narrow cosmological role. For example, take again the property of being cool. Being 13 See the discussion of the performative theory of truth in Chapter 1 here, particularly the quote from Blackburn. 14 We also discussed a few other features of sparseness in Chapter 2, including the idea of genuine similarities. For the most part in what follows I will focus primarily on the causal-explanatory role.

70

Language–World Connections

cool brings a person certain benefits, but is there anything that possession of the property of being cool brings that cannot be explained by the fact that others believe or assert that that person is cool? If not, then coolness has a narrow cosmological role: it has some explanatory power in that we can cite a person’s coolness in explanations of why certain things happen, but that explanatory power is light in that all that explanatory work can be done by reference to the fact that the person is believed to be cool, as opposed to there being explanatory work done by their coolness itself. If narrow cosmological role is a matter of a property’s just exhibiting the belief-based effects, broad cosmological role involves the exhibition of at least one other feature.15 This shows that there is more to the property and its behaviour than what we believe about it, which demonstrates the independence of the property from human thought and language. This feature of width of cosmological role is key when we consider the responsiveness of a predicate. Part of the reason a predicate responds to a sparse property is because there are aspects of that property that are independent from our thought and talk about it: it has explanatory roles to play that are not dependent on our practices. Abundant properties, on the other hand, will only ever have a narrow cosmological role, as defined above. This is because there are not features of those properties that cannot be explained by our beliefs that an object has a particular property, as these properties are generated by predicates. So the distinction between sparse and abundant properties in this context is that abundant properties have a narrow cosmological role, whereas sparse properties have a broad cosmological role. When we consider the relationships between predicates and properties of a certain kind, examining the property’s width of cosmological role is a key factor in determining a property’s responsiveness, and hence, sparseness.

4.4 Applying the Models We have some choices when it comes to applying these models. We could decide in advance that one model is the correct model, and hold that it must apply for all kinds of predicates. That is, we could decide that either the sparse model is the correct model, and hold that it must apply everywhere, or we could decide that the abundant model is the correct model, and hold that it must apply everywhere. We have reviewed some concerns with taking these lines earlier in Chapter 2, and will build on them further in Chapter 5, but I wish to add another worry here. If we agree that there are different kinds of predicates, then it is methodologically odd to decide in advance which model 15 Even though the distinction between narrow and broad cosmological role is not a matter of degree, we can note that there may be different ranges of effects that properties in the broad category have. Properties with the broadest cosmological role in the broad category will be properties that exhibit all four kinds of effects, but there is room for properties that have fewer. For example, observable subatomic particles may have just the first and fourth effects. We can also note that we are not just talking about the fundamental properties of physical science here—see the discussion of intrinsicality and sparseness in Chapter 2.

applying the models

71

is the correct model, and try to force every kind into that. Instead it makes more sense to examine each kind of predicate on its own terms to see which model is best suited to it. This allows us to examine the situation in the most objective terms, without any preconceptions about which model is the correct one. Moreover, if then it turns out that one model fits all kinds, then that would give us a more solid motivation for thinking that there is just one model that fits all. So, it should remain an open question at this point which kinds of predicates are responsive, and which are generative. Indeed, there is room for substantial disagreement about this in particular domains. In the moral domain, for example, moral realists may argue that moral predicates are responsive,16 whereas moral anti-realists may argue that they are generative. I will now turn to examine some kinds of predicates to see whether we can build a case for some kinds being apt for the responsive model, and other kinds being apt for the generative model.

4.4.1 Chemical properties Consider the predicate ‘is metallic’. This is a chemical predicate, in that it concerns the chemical bonding of atoms, and the ways in which substances interact. Substances that have the property of being metallic are generally solid (depending on the temperature) and able to conduct heat and electricity. They also usually appear shiny. Now let’s suppose that we are considering the sentence ‘my ruler is metallic’, and let’s suppose that my ruler is metallic. One of the key things to consider in terms of whether the predicate ‘is metallic’ is responsive or generative is width of cosmological role. Generally speaking, properties that have narrow cosmological role are properties generated by predicates, whereas properties with broad cosmological role are properties to which predicates respond. For metallicness to have narrow cosmological role it would need to be the case that all the relevant explanatory work can be done by reference to an object’s being taken, or accepted, or believed, to be metallic. Of course, there may be some explanatory work of this kind—my belief that the ruler is metallic may explain why I consider it to be sturdier than my wooden ruler—but it looks unlikely that we can do all our explanatory work this way. For example, the ruler’s increase in temperature when placed on the stovetop is not something we can explain by reference to my beliefs about the ruler, and nor is the ruler’s conducting of electricity when jammed into a power socket. This suggests that metallicness has a broad enough cosmological role to warrant us in thinking that the predicate ‘is metallic’ is responsive, not generative. It is a predicate that picks out a property whose behaviour is not explainable simply in terms of beliefs we have about things that bear that property, which suggests that it is a part of reality that our thought and talk respond to, as opposed to generate.

16 An example of a moral realist of this sort would be Boyd (1988). An example of a moral anti-realist of this sort is Blackburn (1993).

72

Language–World Connections

4.4.2 Institutional properties Take the predicate ‘is the Governor of New York’. This is an institutional predicate in that it concerns a structure of rights, obligations, and duties in the context of a clearly specified institution. The Governor of New York evidently has certain specific duties and powers that no other individual has. Their workplace is located in a specific place, and they bear certain relationships to other parts of the state legislature. Let’s now consider whether the predicate ‘is the Governor of New York’ is responsive or generative. Once more, let’s consider width of cosmological role. The key question here is whether the property of being the Governor of New York has any explanatory role distinct from what can be explained by a person’s being taken as, accepted as, or believed to be, the Governor of New York. It is important to distinguish here between the powers that the Governor of New York has qua Governor of New York, and the powers that the particular bearer of that property—Andrew Cuomo, at the time of writing—has. Evidently Andrew Cuomo may have some powers that do not depend on his being believed to be Governor of New York, such as casting a shadow on the ground, which would suggest that there is some width of cosmological role in that case. However, we are just considering the powers that the property of being the Governor of New York bestows, and wondering whether any of those are things that cannot be explained by a person’s being taken, accepted, or believed, to be the Governor of New York. It is difficult to see in this case how there could be any such explanatory work to be done. As is pointed out frequently in discussions of institutional ontology (see, e.g. Searle (1995, 2010) and Thomasson 2003b, 2009), institutional properties depend on the collective acceptance of a community in order to make sense. It is only because we collectively accept and engage in a communal practice which allows there to be presidents and governors that there are presidents and governors. Moreover, it is only because an individual is collectively accepted to occupy an institutional position— usually because they have met some specified requirements—that they do so. If Andrew Cuomo ceased to be taken, believed, or accepted, to be the Governor of New York, then he would cease to have any of the powers of the Governor of New York. If this is so, then the predicate ‘is the Governor of New York’ has narrow cosmological role, as there is no explanatory work for the property to do that cannot be done by a person’s being taken, accepted, or believed, to have that property. Based on our criteria above, this suggests that institutional properties, such as being the Governor of New York, are generated by institutional predicates, as opposed to those predicates being responsive to sparse properties.

4.4.3 Race and gender properties Returning to our examples of race and gender predicates above, according to our methodology, a key issue when considering their corresponding properties is their width of cosmological role. Moreover, this is applicable in this case, as it is reasonable

applying the models

73

to think that biological properties have a broad cosmological role, whereas social properties have a narrow cosmological role. If racial properties were biological properties, then we might have reason to think they have a broad explanatory role. However, the case against this is strong, as Zack notes: If it were true that being black, white, Asian, or Indian caused human beings to have the types of physical traits they do, then there ought to be some physical marker for race, apart from those traits, that scientists could identify. Otherwise, there is nothing general about a race that could cause specific traits in individuals. However, neither biologists, nor anthropologists, nor physiologists, nor geneticists, nor any of the other scientists who have studied physical race have ever identified any general racial characteristics shared by all members of any particular race. No genes or hereditary factors are shared by every member of the main racial groups. (Zack 2006: 9)

This strongly suggests that racial properties do not have a broad cosmological role. Moreover, whilst there may be biological properties that have been typically associated with certain racial properties, racial properties are not reducible to these biological properties. Consider skin colour, for example. The colour of a person’s skin may be a biological property, determined by the level of melanin in a person’s skin, and it is also a property that has been associated with race. However, there is little reason to think that racial properties can be identified with skin colours. As Appiah says, amongst other things,17 skin colour just does not seem to do the relevant explanatory work: Why does biological variation in skin color not correlate more with other characteristics? Partly, because the other characteristics have been selected (as has, say, sickle-cell disease, in parts of West Africa and the eastern Mediterranean) under pressures not highly correlated with the presence of harmful amounts of sunlight. Perhaps, too, because there are mechanisms that have evolved to maintain the stability of the genotype, reflecting, among other things, the fact that certain combinations of genes are adaptive only when they are present together. As a result, even after long periods of geographical separation, human populations do not drift apart significantly with respect to most of their biological properties. And, finally, because there has been continuous exchange of genes among the major geographical areas of human settlement over the hundreds of thousands of years since the first humans set off out of Africa. (Appiah 1994: 97)

This goes for other biological characteristics too: A second candidate18 for the biological referent [of ‘race’] would simply be groups defined by skin color, hair, and gross morphology, corresponding to the dominant pattern for these characteristics in the major subcontinental regions: Europe, Africa, East and South Asia, Australasia, the Americas, and, perhaps, the Pacific Islands. This grouping would encompass many human See, e.g. Appiah (1994: 58–9). See also Zack (2002: 87–8). The first candidate Appiah considers is the notion of a population, which he also rejects (Appiah 1994: 99–101). This idea has been discussed in recent work by Kitcher (2007) and Andreasen (2000). 17 18

74

Language–World Connections

beings quite adequately and some not at all: but it is hard to see of what biological interest it would be, since we can study the skin and gross morphology separately, and there is, at any rate, a good deal of variation within all these areas, in skin, hair color, and the morphology of the skull. Certainly, this referent would not provide us with a concept that was central to biological thinking about human beings. (Appiah 1994: 101)

Whilst the biological characteristics typically associated with race might be genuine biological properties that have broad cosmological roles, racial properties are not reducible to these properties. In addition, if we try to account for a racial property in terms of a collection of different biological properties, what we end up with is a category which is not a genuine biological category at all. These factors suggest that racial properties are not biological properties that have a broad cosmological role. If, instead, we think that they are social properties it is hard to see how they could have anything but narrow cosmological roles, as they are parts of a social system that depends on human beliefs and attitudes for its sustenance. Similar issues arise with gender predicates. Whilst the predicates ‘is a man’ and ‘is a woman’, for example, are associated with individuals bearing the biological characteristics associated with the male and female sex, respectively, we have already noted above that there is a distinction between sex and gender. The biological properties that an individual has may have broad cosmological roles, but the gender properties are not reducible to biological properties. For example, suppose one were to hold that having the property of being a woman explains why a person has a certain level of oestrogen in their system, which shows that being a woman has a broad cosmological role. The mistake here is that it is not the property of being a woman that is responsible for the oestrogen levels in a person’s system, rather it is the person’s having XX chromosomes. Being a woman is not reducible to having XX chromosomes, as there are women who do not have XX chromosomes. The same would be so for other physical characteristics that are typically associated with being a woman. What one will find is that, whilst certain physical characteristics may have a broad cosmological role, this does not show that the property of being a woman has a broad cosmological role. This is because the property of being a woman is not the sort of thing that can be reduced to physical characteristics. These considerations suggest that race and gender properties do not have a broad cosmological role, and hence should be considered on the abundant generative model. Now, there is no doubt that a person’s having a particular race or gender property has significant explanatory roles to play: in New York (and doubtless elsewhere), for example, a person’s being white makes it less likely that they would be stopped and frisked by a police officer than a person who is black, and a woman is likely to receive less in salary than a man, to name just two examples.19 In claiming that race and gender properties do not have a broad cosmological role I do not wish to deny these important 19 See, e.g. https://www.nyclu.org/en/stop-and-frisk-data and http://www.nationalpartnership.org/ research-library/workplace-fairness/fair-pay/4-2016-ny-wage-gap.pdf.

interim summary

75

explanatory roles. The claim is just that these explanations can all be carried out by reference to beliefs that people have about race and gender properties, and those who possess them.20 This is not to downplay the significance of these properties, and their importance in understanding our lives and our world; quite the contrary, it shows that there are properties of crucial importance that are not sparse, but abundant.21

4.5 Interim Summary To sum up what has been said so far, we started with the idea that there are different kinds of predicates, with each kind playing a different functional role. For instance, the role of moral predicates is different from the role of physical predicates, which, in turn, is different from the role of institutional predicates, to take just a few examples. We then saw that, for each kind of predicate, we can ask questions about the sort of property that predicates of those kind pick out. We discussed two examples, the sparse and abundant properties we saw in Chapter 2, and noted that there are different relationships between predicates and sparse properties, and predicates and abundant properties, with predicates responding to sparse properties, and generating abundant properties. We then considered whether we should apply one of these relationships across the board, for all kinds of predicates, or begin on a case-by-case basis to see which model works for a specific kind of predicate. I suggested that the most sensible option is to work on a case-by-case basis, and demonstrated examples of cases where different models apply. I now want to turn to two further issues. The first is whether what has been said above about predicates and properties can also be said about singular terms and objects. This I will only address briefly as we will discuss the issue in more detail in Chapter 6, but I mention it here as it has a minor role to play in the discussion of the second issue. The second, more substantive, issue is to systematize the talk of different kinds of predicates as talk of different domains. I will explain what a domain is, and how sentences are assigned to different domains.

20 Hacking’s (1999) notion of looping and interactive kinds is important here, and we will discuss it in Chapter 5. See also Mallon (2006) and Root (2000). 21 Are there cases of social properties that do not depend on beliefs? Thomasson (2003b: 288) discusses the example of the property of being in a recession, which, she claims, an economic system can have independently of any beliefs that individuals in that system have. The issues here are complex. For instance, there is a sense in which people’s beliefs about whether the economy is in a recession play a key part in whether or not we are in a recession. If we believe that we are in a recession we are likely to behave in a way that makes it more likely that we are in a recession: indeed, one challenge in such cases is to restore confidence in the economy to an extent that it is possible to get out of being in a recession. This is one reason why people are careful to use the term ‘recession’, for if it is stated that a country is in a recession, this will have a large impact on the factors that make it the case whether the country is in a recession. This suggests that recessions might also be examples of Hacking’s (1999) ‘looping kinds’ (again, see Chapter 5 for more on this).

76

Language–World Connections

4.6 Singular Terms and Objects We can consider the general function of singular terms to be to refer to objects,22 and that singular terms can be divided into kinds by sorts of objects they are used to refer to. This will be more closely determined by the features we take the corresponding objects to have. For example, we can distinguish between ‘the number 5’, ‘the US government’, and ‘the guilty party’ as different types of singular terms in a sense, namely mathematical, institutional, and moral. This is partly because of how they are used: ‘the number 5’ is used to refer to a mathematical object; ‘the US government’ is used to refer to an institutional object; and ‘the guilty party’ is used to refer to a moral object. Moreover, the case can be made that there is a distinction between sparse and abundant objects which lines up with the distinction between sparse and abundant properties. Consider the feature of width of cosmological role, which we took to play a key role in distinguishing between sparse and abundant properties. The idea there was that a property has a broad cosmological role if it features in explanations of things that are not mediated by beliefs we have about that property, or about things that bear that property. We can construct a matching condition for objects by saying that an object has broad cosmological role if it features in explanations of things that cannot just be explained in terms of beliefs we have about that object. One way to develop this idea is by explicit reference to what objects are able to cause. We saw in the initial definition of width of cosmological role that things that have broad cosmological role are able to have a variety of effects, and this is the same for objects. For example, the idea that objects have causal powers is an influential idea. Trenton Merricks, in his influential argument against the existence of macroscopic material objects makes explicit use of this idea: If there were baseballs, they would break windows, they would injure batters, they would cause visual sensations (and so be seen), and they would cause tactile sensations (and so be  felt). In general, if there were inanimate macrophysical objects, they would have causal powers. (Merricks 2001: 81)

This sort of causal condition on objects is one way to substantiate the idea that objects that meet it have broad cosmological role: the effects that the baseball has are not dependent on any person’s belief that the object is a baseball. We noted earlier that social and institutional properties have a narrow cosmological role, and social and institutional objects will also have a narrow cosmological role. For example, institutional and social objects would fail the causal condition described above. A government, for example, cannot break a window, be seen, or be felt, even if particular members of a government can satisfy these conditions. Moreover, everything that a government can explain is explained in terms of beliefs that we have about that thing

22

See, e.g. Hale (1994).

domains and domain membership

77

being a government: collective acceptance is a key factor in institutional and social cases, and institutional and social objects would not be the objects they are without collective acceptance, so the idea that they could play explanatory roles outside of this context does not really make sense. We distinguished above between sparse and abundant properties in terms of properties that were responsive or generative, respectively. We can use that terminology again here with reference to the relationship between singular terms and objects. We can say that singular terms can be responsive or generative: responsive singular terms respond to sparse objects, and generative singular terms generate abundant objects. Whether or not an object is responsive or generative will depend on whether the object has a broad or narrow cosmological role. What is at issue here is not what it takes for an object to have a property, but for a singular term to refer to an object. In particular, the idea is that, in some cases, a singular term will refer to an object because it responds to that object; whereas in other cases a singular term will generate an object by referring to it. These ideas will be developed further when we come to discuss the connections between existence and truth in Chapter 6.

4.7 Domains and Domain Membership So far I have used various phrases to talk about the differences between different kinds of predicates: ‘areas’, ‘subject-matters’. Now that we have the view up and running I want to give a more systematic account of this idea, using the term domain. As with the rest of the framework above, the account of domains and domain-membership is  offered to anyone who wants to make distinctions between different kinds of subject matters.

4.7.1 Domains A domain has both a semantic and a metaphysical component, which are closely related. The semantic component of a domain is made up of the singular terms and predicates that belong to that domain. The domain to which a singular term or predicate belongs is determined by the kind of singular terms or predicate that the relevant singular term or predicate is identified to be. This is done in the way outlined above, where we look at the functional role of a singular term or predicate kind to demarcate different kinds of singular terms or predicates. The semantic aspect of a domain will just be identified with a singular term and predicate kind. For example, the semantic aspect of the moral domain will be the singular terms and predicates identified as being moral singular terms and predicates in virtue of their playing the functional role associated with moral singular terms and predicates. The metaphysical aspect of a domain will be composed of the objects and properties that the singular terms and predicates in the semantic aspect of the domain refer to.

78

Language–World Connections

Thus the metaphysical aspect of the moral domain, for example, will be the objects and properties referred to by moral singular terms and predicates. This is not to say that all there is to objects and properties of different kinds is that they are referred to by different kinds of singular terms and predicates. This is because, as we have seen, the relationships between singular terms and objects, and predicates and properties, varies: in some cases the nature of the objects and properties is dependent on the singular terms and predicates, whereas, in other cases, the singular terms and predicates are dependent on the properties. In the rest of the book I will use the word ‘domain’ to talk about different areas, with the term understood as described above. I have not separated out the different components of a domain into distinct terms, as they are so closely related on this proposal that it doesn’t make sense to do so. If there are occasions when it matters whether I am talking specifically about just the semantic or metaphysical component of a domain, I will say so.

4.7.2 Domain membership I talked above about singular terms and predicates belonging to particular domains, but we will also talk about sentences belonging to particular domains. That is, that some sentences will be moral sentences, and others will be social sentences, still others will be mathematical sentences, and so on. How is it that sentences are assigned to a domain? Once again, this is a key issue for anyone who wishes to treat different domains differently. Our example at the beginning of the chapter of a person who wants to be an expressivist about ethics, a Platonist about mathematics, and a scientific realist about physics needs a way to distinguish moral sentences from mathematical sentences from physical sentences. I will focus here on simple atomic sentences, of the form ‘a is F’.23 Let us take it that an atomic sentence is composed of two elements, a singular term and a predicate. A key question is whether it is the singular term, the predicate, or both that determines the domain of the sentence. This is important as, in a lot of cases, singular term and predicate are of different kinds, such as ‘the sunset is beautiful’, ‘the broccoli is delicious’, ‘killing that person is wrong’. We will discuss the nature of objects further in Chapter 6, but for now we can note that, in these examples, we have a difference between the kind of thing the object is and the kind of thing the property predicated of it is, which raises the question of how to understand the domain of the sentence.24 I will suggest that it is the predicate that determines the domain. We can distinguish between two things: what a sentence is about, and what is said about the thing the 23 I will not in this work consider atomic cases more complicated than this, with the idea being that we have to start somewhere, and then develop the account for more complex cases. 24 This is a broader version of the ‘problem of mixed atomics’ raised for truth pluralism by Sher (2004), David (2013), and Wyatt (2013). I keep it more general here as I think it is a problem for any view that holds that there are distinctions between kinds of subject matter.

domains and domain membership

79

sentence is about. A sentence is about its object: ‘snow is white’ is about snow, ‘torture is wrong’ is about torture, ‘the sunset is beautiful’ is about the sunset, so atomic sentences are always about their objects. But what makes these things sentences is that there is more: there is something that is said about the things that the sentences are about. In the examples above it is said of snow that it is white, it is said of torture that it is wrong, and it is said of the sunset that it is beautiful. This ‘saying of ’ occurs due to the attribution of a property to the object. It is this aspect—the attribution of a property to an object—that makes these kinds of sentences sentences in that they are bearers of content. So, it is not what a sentence is about that we should be considering, it is rather what is said about the thing the sentence is about.25 As a result, we should just be considering the predicate, and not the singular term, when making decisions about which kind an atomic sentence belongs to. With this in mind, we can say of our examples that they do in fact belong to a single kind: because deliciousness is an aesthetic predicate, ‘the broccoli is delicious’ is just an aesthetic sentence, and because wrongness is a moral predicate, ‘killing that person is wrong’ is just a moral sentence. They belong to these kinds because of the functions of the predicates involved. Consequently, simple mixes of singular terms and predicates in a sentence should not motivate any move from the view that each sentence belongs to a single kind. Accordingly, I propose that sentences are individuated by the predicate they contain, whereas the singular term is not relevant to kind individuation. This can be substantiated by considering that many different kinds of thing can be said about the same object: a single chair can be blue, solid, beautiful, sad, dangerous, presidential, or singular, to name just a few examples. Given that the nature of the singular term is held fixed, and the only thing that varies is the predicate, this is independent reason to think that kind-membership is determined by the predicate. However, one might worry that this is too easy. If we avoid the problem of mixed atomics by holding that it is the predicate that is relevant for determining kindmembership, then would not problems be caused by predicates that themselves appear to be members of more than one kind? A key class of examples are ‘thick’ predicates, such as ‘courageous’, ‘kind’, or ‘lewd’. So, for example, the sentence ‘Sergio Parisse is courageous’ might be considered mixed because the predicate ‘is courageous’ expresses a property (or properties) with both moral and non-moral aspects. These are contrasted with ‘thin’ predicates like ‘is good’, which do not appear to have any non-moral aspects.

25 In the context of truth pluralism, Lynch (2009, 2013a) does not make this distinction, holding that what matters is what a sentence is about, with aboutness encompassing both object and property, which, as Wyatt (2013) notes, does not help assign a sentence to a domain when object and property are from different domains. Pedersen and Wright (2013b) offer a pluralism-specific line which inspired the more general line taken here.

80

Language–World Connections

This is a problem that afflicts many views, and there is a wealth of literature26 on whether expressivism about morals, in particular, can deal with thick predicates. The worry in that case is that the expressive and descriptive elements are inseparable, thus causing problems for the view that moral language is purely expressive. There are solutions to this problem offered on behalf of expressivists, however, and we can see how they can be adopted to give a general response to the problem. One strategy is to take sentences involving thick predicates to be compound sentences, as opposed to atomic sentences. Williams (1985) puts this approach nicely as follows: A statement using [thick] terms can be analyzed into something like “this act has such-andsuch a character, and acts of that character one ought not to do.” It is essential to this account that the specific or “thick” character of these terms is given in the descriptive element. The value part is expressed, under analysis, by the all-purpose prescriptive term ought. (Williams 1985: 144)

If we take this approach, we can say that we are no longer dealing with a problem of mixed atomics, but rather a problem of mixed conjunctions.27 In this case there will be no problem accounting for the domains of the atomics, as—in line with what was said above—we can use the predicates attributed to identify the domains the sentences belong to. However, this option has some downsides.28 For one thing, it does not answer the question of what kind of sentence it is: is it still a mixture of a moral and physical sentence, for example? Accordingly, it is worth exploring an alternative account, which is inspired by Michael Smith’s (2013) account of thick predicates. Smith suggests that we think of the difference between thick and thin predicates not as a difference in kind, but in degree. One way to explain this is to say that thick predicates seem to be thick because they impose a certain amount of descriptive content on the entity that bears the property. For example, saying ‘Sergio Parisse is courageous’ attributes not only some positive moral feature to Sergio Parisse, but also some descriptive features, namely that he displays some sort of appropriate resistance in the face of danger, for example. This may suggest to some that there is some difference in kind between thick predicates, like ‘is courageous’, and thin predicates, such as ‘is good’, as one imposes descriptive content whereas the other does not. However, this is a mistake. Even thin properties like being morally good impose some descriptive content on their bearers. For example, if an action is deemed to be morally good, this at the very least implies that the action was carried out by an intentional agent, as opposed to

Eklund 2011 provides an overview. See also Elstein and Hurka (2009), Blackburn (1992), and Gibbard (1992). Also see Lynch  2013b for an argument to suggest that expressivists should embrace alethic pluralism. 27 See Chapter 7. 28 Williams rejects the view himself! See Elstein and Hurka (2009) for a detailed study. 26

summary

81

simply being the result of happenstance.29 The movement of the branch of a tree in the breeze, for example, would not be something that could be morally good, on the grounds that it does not meet the requirements for being a bearer of the property of being morally good. This suggests that thin predicates like ‘is good’ do impose some descriptive content on their bearers, even if it is more minimal than the descriptive content imposed by thicker predicates. If we take this line, then we can say that all moral predicates will have a degree of thickness, with the thinner predicates being predicates like ‘is good’, and the thicker predicates being predicates like ‘is courageous’. There will also be predicates thicker than courageousness which impose very specific descriptive features on their bearers, such as the predicate ‘is a bad father’. This imposes a set of descriptive features on the bearer which are much more detailed and specific than those specified in the predicate ‘is courageous’. However, we would say that, despite the variations in degrees of thickness, these are all moral predicates. The fact that some moral predicates imply some descriptive content does not mean that they are not moral predicates, as predicates of pretty much any property imply some constraints on what can bear it. In line with what was said earlier in this chapter, the status of these predicates as moral predicates will be determined by the functional role that these predicates play in the cultivation and evaluation of character, and the way we decide what we ought to do. I have claimed, then, that domains have both semantic and metaphysical components, with the semantic component being understood in terms of predicates, and the metaphysical component understood in terms of properties. Later I will talk more about singular terms and objects to see how they fit into the picture. My suggestion in terms of domain membership is thus to say that predicates determine the domains of atomic sentences, and that, in cases where the predicate seems mixed, we look to the primary functional role of the predicate to determine the domain.

4.8 Summary To sum up this chapter, we started with the idea that there are different kinds of predicates, with each kind playing a different functional role. For instance, the role of moral predicates is different from the role of physical predicates, which, in turn, is different from the role of institutional predicates, to take just a few examples. We then noted that, for each kind of predicate, we can ask questions about the sort of property that predicates of those kind pick out. We discussed two examples, the sparse and abundant properties we saw in Chapter 2, and saw that there are different relationships between 29 Compare this thought to the idea of truth-bearers: no matter what theory of truth you have, truth is not a property that can be borne by shirt buttons, which suggests some descriptive content common to all theories of truth.

82

Language–World Connections

predicates and sparse properties, and predicates and abundant properties, with predicates responding to sparse properties, and generating abundant properties. We then considered whether we should apply one of these relationships across the board, for all kinds of predicates, or begin on a case-by-case basis to see which model works for a specific kind of predicate. I suggested that the most sensible option is to work on a case-by-case basis, and demonstrated examples of cases where different models apply. Finally, I introduced another idea to systematize these thoughts, namely the notion of a domain, and explained how sentences are assigned to domains. I mentioned above that the model proposed here is open to anybody who wants to make distinctions between different domains, and it is now time to ask: what does all this have to do with truth?  

5 Truth Re-Emerges In Chapter 4 I set out a model for thinking about different domains, and we were left with the question of what it all has to do with truth. This chapter answers this question: everything! In doing so, I develop an argument for truth pluralism. I begin by discussing in more detail the connections between predicates and properties, and suggest that truth is heavily involved in understanding these relationships. As a consequence, lurking behind the picture I outlined in Chapter 4 are substantial commitments to views about truth. In particular, I suggest that different understandings of truth are involved in the distinction between abundant and sparse properties, and, as a result, that distinction entails there being at least two different approaches to truth: a representational and a non-representational approach. This is an initial commitment to truth pluralism, and I go on to develop options in both the representational and non-representational approaches, applying them to the examples we discussed in Chapter 4. I close the chapter by considering and critiquing in more detail the monist alternatives to pluralism, namely the ideas that all domains are understood on the abundant model, or all domains are understood on the sparse model. I also show why an alternative pluralist model—correspondence pluralism—is inadequate. I should note that an existing argument for truth pluralism is the ‘Scope Problem’ (Lynch 2001, 2009; Sher 2004), which holds that different theories of truth should be limited in scope, as they each perform well in some domains, but not others. I endorse this argument, and will discuss it further in Chapter 6, but the argument made in this chapter offers a different, and stronger, argument for truth pluralism. This argument claims that it is not just that different theories of truth are best suited for different domains, but that truth has to be understood differently in different domains, for we cannot get the general distinctions we made between different domains in terms of sparseness and abundance up and running without a pluralist approach to truth. Now, the idea that the differences between approaches to the relationship between language and the world depends on different approaches to truth may be surprising in the context of the rise of deflationary theories of truth, but it is not at all surprising in  the context of the close associations made between truth and realism and antirealism in the history of debates about truth. Realism, for example, often just used to be understood as the adherence to the correspondence theory of truth, and anti-realism as adherence to a pragmatic or coherence theory of truth. Part of the impact of the rise

84

Truth Re-Emerges

to prominence of deflationary theories of truth has been the separation of truth from the realism/anti-realism issue. However, as we saw earlier, we should be careful about affording deflationism this sort of role.

5.1 Sparseness, Abundance, and Truth Let’s start with the principle about properties and predicates that we have taken to be key so far in understanding the difference between sparse and abundant properties: (P) The object referred to by ‘a’ falls under the predicate ‘F’ iff the object referred to by ‘a’ has the property referred to by ‘F’. In terms of the relationship between predicates and properties, we identified two different ways of explaining it: (PA) the object referred to by ‘a’ has the property referred to by ‘F’ because the object referred to by ‘a’ falls under the predicate ‘F’ (PS) The object referred to by ‘a’ falls under the predicate ‘F’ because the object referred to by ‘a’ has the property referred to by ‘F’. In the case of abundant properties, (PA) is the key claim, as it is because an object falls under a predicate that it has a property, and in the case of sparse properties it is (PS), where an object falls under a predicate because it has a property. In (PA) we are left with a key thing to explain: what it takes for an object to fall under a predicate. With (PS) this isn’t a problem, as an object’s falling under a predicate is explained in terms of an object’s having a property. We cannot do this with (PA), however, as an object’s having a property is explained in terms of its falling under a predicate. A key resource here is the idea of predicate satisfaction, and, in particular, the idea that predicate satisfaction is connected to the idea that predicates are ‘true of ’ objects. That is, an object falls under a predicate just in case the predicate is true of the object. This is expressed in the following biconditional: (FT) The object referred to by ‘a’ falls under the predicate ‘F’ iff ‘F’ is true of the object referred to by ‘a’. This helps (PA) because we have the resources to explain how an object satisfies a predicate that does not rely on the idea that the object has a property. That is, for (PA) we will read (FT) as having a particular order of determination: it is because ‘F’ is true of the object referred to by ‘a’ that the object referred to by ‘a’ will have the property referred to by ‘F’. Conversely, for (PS), we would have the opposite reading where an object falls under a predicate because it has a property. There is a clear connection between a predicate’s being true of an object, and a sentence involving the predicate and corresponding singular term being true. That is, the following principle holds: (TO)

‘F’ is true of the object referred to by ‘a’ iff ‘a is F’ is true

sparseness, abundance, and truth

85

Once again, we will have different readings of this biconditional for (PA) and (PS). For (PA), we need to explain how a predicate is true of an object, which we can do by saying that the predicate ‘F’ is true of the object referred to by ‘a’ because the sentence ‘a is F’ is true. For (PS), we already have an explanation of why ‘F’ is true of the object referred to by ‘a’, and we can use this to explain why ‘a is F’ is true. What this shows is that we have two different explanatory paths through these various principles. One starts with the claim that an object has a property and goes on to show how a sentence is true. The other starts with the truth of a sentence and shows how we can derive the possession of a property by an object. We can now trace these two different lines of connection between these principles as follows: (A) 1. ‘a is F’ is true. 2. If ‘a is F’ is true, then ‘F’ is true of the object referred to by ‘a’. 3. If ‘F’ is true of the object referred to by ‘a’, then the object referred to by ‘a’ falls under the predicate ‘F’. 4. If the object referred to by ‘a’ falls under the predicate ‘F’, then the object referred to by ‘a’ has the property referred to by ‘F’. 5. The object referred to by ‘a’ has the property referred to by ‘F’. (S) 1. The object referred to by ‘a’ has the property referred to by ‘F’. 2. If the object referred to by ‘a’ has the property referred to by ‘F’, then the object referred to by ‘a’ falls under the predicate ‘F’. 3. If the object referred to by ‘a’ falls under the predicate ‘F’, then ‘F’ is true of the object referred to by ‘a’. 4. If ‘F’ is true of the object referred to by ‘a’, then ‘a is F’ is true. 5. ‘a is F’ is true. From (A) we get the conditional: if ‘a is F’ is true, then the object referred to by ‘a’ has the property referred to by ‘F’. From (S) we get the conditional: if the object referred to by ‘a’ has the property referred to by ‘F’, then ‘a is F’ is true. From these we can get the biconditional: (TP)

‘a is F’ is true iff the object referred to by ‘a’ has the property referred to by ‘F’.

Interestingly, this is close to the principle (Tar) that we met in Chapter 3: (Tar)

‘a is F’ is true iff the object referred to by ‘a’ satisfies the predicate ‘is F’.

As we have seen above, (TP) is a way to account for predicate satisfaction, and thus ground the right-hand side of (Tar). What we have done is shown how we can derive both conditionals of (TP) by considering the relationship between predicates, properties, and truth. There is an important difference though between what I want to do with this principle and a Tarski-inspired approach. A Tarski-inspired approach would use

86

Truth Re-Emerges

(Tar) to show how truth could be defined in terms of predicate satisfaction, and (TP) to show how predicate satisfaction can be explained in terms of property possession. Whilst of course the Tarskian acknowledges both conditionals as part of the (TP) biconditional, there is an order of determination on the biconditional, which has it that the conditional derived from the (S) derivation above is the primary conditional: If the object referred to by ‘a’ has the property referred to by ‘F’, then ‘a is F’ is true’ That is, in terms of discharging the conditionals, for the Tarskian, we always take the possession of a property by an object to be the initial thing, and then we infer the truth of a sentence from that. Of course, as (TP) is a biconditional, once we have the truth of the sentence we can then go back and infer the possession of the property by the object, but that is only after we have shown the initial sentence to be true due to that property being possessed by the object. In thinking that truth is dependent on predicate satisfaction, this interpretation of (TP) has it that we cannot start with the truth of a sentence and then infer the possession of a property by an object in any interesting sense. Because predicate satisfaction is prior to truth, any derivation of the (A) form cannot be used to show that predicate satisfaction is dependent on truth—at most we can trace back through the initial reasoning used. This is important, as it shows that the original Tarskian reading of (TP), which takes truth to be dependent on predicate satisfaction, is compatible only with the sparse conception of properties previously discussed, and not with the abundant one. It precludes the abundant view because the claim of the abundant view is that we cannot understand predicate satisfaction in terms of property instantiation, so we have to understand it some other way. The suggestion is that we understand it in terms of a predicate being true of an object. In other words, we determine the extension of a predicate not by a property that objects possess, but by the objects the predicate is true of. The objects the predicate is true of is determined by the truth of sentences containing the relevant singular term and predicate.1 However, on the Tarskian reading the way to determine the truth of a sentence is by looking at whether the property is possessed by the object, which we cannot do here, as the truth of the sentence is the very thing we need to determine in order to establish that the object has the property! The abundant view thus requires an understanding of the way that sentences are true that is distinct from the Tarskian interpretation of (TP). In other words, we need an understanding of truth that does not make truth dependent on predicate satisfaction, as, on the abundant view, predicate satisfaction needs to be understood in terms of truth. What this shows is that, if we want to say that there are both abundant and sparse properties, we need to accept that both the (A) and (S) derivations are good. Given that abundant and sparse properties are appropriate models in different domains, the (A) and (S) derivations will apply in different domains: in some domains 1 See, for example, Schiffer (1996, 2003), and Hale and Wright (2009). See Schiffer (2003: 67) for a discussion of how to respond to potential problems with properties like the property of being a property that doesn’t instantiate itself.

sparseness, abundance, and truth

87

we start with a true sentence and derive the possession of a property by an object, and in other domains we start with the possession of a property with an object and derive a true sentence. We thus need two different ways of reading (TP) if we are to understand the distinction between sparse and abundant properties, (R) and (NR): (R) ‘a is F’ is true because the object referred to by ‘a’ has the property referred to by ‘F’. (R) is derived from the following claims on the assumption that ‘because’ is transitive: 1. ‘a is F’ is true because ‘F’ is true of the object referred to by ‘a’. 2. ‘F’ is true of the object referred to by ‘a’ because the object referred to by ‘a’ falls under the predicate ‘F’. 3. The object referred to by ‘a’ falls under the predicate ‘F’ because the object referred to by ‘a’ has the property referred to by ‘F’. (NR) the object referred to by ‘a’ has the property referred to by ‘F’ because ‘a is F’ is true (NR) is derived from the following claims on the assumption that ‘because’ is transitive: 1. The object referred to by ‘a’ has the property referred to by ‘F’ because the object referred to by ‘a’ falls under the predicate ‘F’. 2. The object referred to by ‘a’ falls under the predicate ‘F’ because ‘F’ is true of the object referred to by ‘a’. 3. ‘F’ is true of the object referred to by ‘a’ because ‘a is F’ is true. We can connect each reading to representational and non-representational understandings of truth, and call (R) the representational reading, and (NR) the non-representational reading. (R) is the representational reading as it states that what is true is dependent on the possession of a property by an object, which is taken to be a state of affairs that exists prior to the truth of the sentence. (NR) is the nonrepresentational reading, as the possession of a property by an object is dependent on the truth of a sentence, and hence is not a state of affairs that exists prior to the truth of the sentence. This tells us something interesting about the connection between generative predicates, abundant properties, and non-representational truth. Generative predicates are so because they do not respond to pre-existing sparse properties; instead they generate abundant properties. However, they are only able to generate abundant properties if a sentence involving that predicate is true. Note though that, by the nature of the predicate, a sentence that involves it cannot be true by means of representational truth. This is because representational truth requires the pre-existence of the objects and properties that are referred to. In other words, representational truth would require the predicate to be responsive, but, by stipulation, it is not—it is generative. As a

88 Truth Re-Emerges consequence, generative predicate are necessarily paired with non-representational kinds of truth, and the existence of abundant properties is secured by the successful pairing of generative predicates and non-representational truth.2 Conversely, responsive predicates and sparse properties are naturally paired with representational accounts of truth. The direction of the truth principle above favoured by representational theories holds that the objects and properties that sentences are about exist prior to the truth of the sentence itself. This suggests that the predicates in question are responsive, since if they were generative then the properties in question would not exist prior to the truth of the sentence. If the predicates in question are responsive, then the truth of the sentence will depend on the existence of the relevant sparse properties which give rise to the predicates. In Chapter 4 I gave some general reasons for thinking that there is something to be said for distinguishing between different domains, and asking metaphysical questions about domains on a case-by-case basis. I presented both the sparse and abundant models as options, and discussed why they fit different domains. What I gave was essentially an argument for pluralism about subject-matters, with the idea being that different sorts of subject-matter need to be understood differently in terms of the metaphysics of the properties in question, and the ways that predicates relate to properties. I offered this as a general discussion, without at that point being partisan to any particular account of truth. What I am now suggesting is that, if we want to take the models outlined in Chapter 4 on board, we need to accept that truth will come in different forms. This is because, as shown above, there are close connections between predicates, properties, and truth, and the only way to preserve the availability of both the sparse and abundant models in different domains is to hold that truth comes in at least two general forms. One form, which connects with the sparse account, is that truth is representational in that truth depends on predicate satisfaction. The other form, which connects with the abundant account, is that truth is non-representational in that predicate satisfaction depends on truth. We will now explain representational and non-representational accounts of truth in more detail, and show how they apply to our examples of responsive and generative predicates from Chapter 4.

2 Are there ways to get the connection between predicates and abundant properties without reference to truth? This is challenging. Suppose that we take abundant properties to just be classes, so having an abundant property is just being a member of a class. This is all well and good, but how then do we attach a predicate to a specific class? We have to say that we attach a predicate to a class because that class is the extension of the predicate. This leads us to have to say what it is that determines the extension of the predicate. We could say that we set out rules for the use of the predicate, and that an object is in the extension just in case it meets the conditions set out in those rules. However, to make good on this move one would firstly need to show that meeting the conditions set out in the rules is distinct from the predicate being true of an object. Secondly, even if this can be done, one would need to show that this is the best way to understand the operation of these predicates. See particularly sections 5.2.2 and 5.4.3 below for more on this, along with Schiffer (1996, 2003, Forthcoming), Hale and Wright (2009), and Thomasson (2015, ch.3).

forms of truth

89

5.2 Forms of Truth 5.2.1 Representational theories of truth Correspondence theories of truth tend to be associated with ‘realist’ views concerning the relationship between mind and language on the one hand, and the world on the other. We have a divided picture, with thought and language on one side, and reality on the other, and beliefs or sentences are true when they successfully bridge the divide, and represent reality as it really is. ‘Correspondence’ is the name given to the relation by which true beliefs and sentences bridge the gap to reality. Exactly how the relation of correspondence is cashed out varies amongst correspondence theories. There are two main kinds that I will mention here. On the view proposed by Bertrand Russell (1912), beliefs are the relevant items on the mind/language side of the divide, and facts are the relevant items on the worldly side. A belief is true, for Russell, when there exists a fact to which it corresponds, and false when there is no such fact. This fact-based correspondence can make sense of the principle (Tar) in the following way. If, following Armstrong (1997) we hold that facts—or states of affairs in his terms—are the basic units of reality, then an object’s having a property will be inferred from a fact. That is, we can have the following biconditional (but for sentences, not beliefs): (SA) The object referred to by ‘a’ has the property referred to by ‘F’ iff it is a fact that a is F. The idea here will be that there is an order of determination on (SA) which has it that the right-hand side is more basic than the left-hand side. The viability of this view depends on whether facts are taken as genuine entities. However, following Strawson (1950b), there has been scepticism of this notion of a fact, with it being seen as a piece of metaphysical extravagance that does not do the theoretical work required, and, moreover, does not fit into a naturalistic picture. However, I do not wish to discount this view here, and it has had some recent defenders (see, e.g. Fumerton (2002), Vision (2004), and Rasmussen (2014)). A second account, which is perhaps more influential in contemporary correspondence theories, uses the notion of reference. This strategy is inspired by Tarski’s work on truth, that we touched on in Chapter 3, and, in particular, by Field’s (1972) development of Tarski’s theory.3 The idea here is to show how truth can fit into a naturalistic system. By showing that truth can be explained in terms of reference, and that reference can be explained naturalistically, we can show how truth is a naturalistically acceptable notion. The first step is that the view is committed to the idea that the truth of a sentence is defined in terms of its components, an idea we encountered with the principle (TP) earlier: (TP): The sentence ‘a is F’ is true if and only if the object referred to by ‘a’ has the property referred to by ‘F’. 3

See Grover (1990) and Sher (1999) for more discussion of Field’s view.

90

Truth Re-Emerges

Correspondence theories in general read (TP) as having a right-to-left order of determination: reference is taken to be prior to truth. To develop our particular example we need to supplement (TP) with a theory of reference. One strategy here, developed by Field, is to use a causal theory of reference, which is key to the project of showing how truth is naturalistically acceptable. In doing so, according to Lynch, we get: CC (Causal-Correspondence): The sentence ‘a is F’ is true if and only if the object causally mapped by ‘a’ has the property causally mapped by ‘F’. (Lynch 2009: 27. Note that Lynch’s original principle is for beliefs, but I have adapted it for sentences.) Taking CC as the statement of our correspondence theory of truth, Lynch states that it has two further separate components: 1. True beliefs map objects that exist and have their properties mind-independently. 2. The objects and properties that are so mapped are capable of entering into at least indirect causal interaction with our minds. (Lynch 2009: 33) Claim (1) here is a re-statement of the ‘realist’ idea that we met earlier: that there is a divide between mind and world, and that the things in the world are the way they are independently of how we think them to be. Talk of facts is replaced by talk of objects and properties. This represents a difference between semantic correspondence and fact correspondence in that fact correspondence takes facts to be more basic than objects and properties, whereas semantic correspondence (if it wants to accept talk of facts) takes objects and properties to be more basic than facts. Claim (2) is an artifact of the specific theory of reference adopted in order to make this a fully-fledged theory, which, in this case, requires causal relations to understand reference. I will pause for a moment to say a little about why the causal theory of reference is appropriate here in terms of the motivations for the correspondence theory of truth. The causal theory of reference, as developed by Kripke (1980) and Putnam (1975) is opposed to a descriptivist theory (e.g. Russell (1905), Searle (1983). On the descriptivist theory, the referent of a term is determined by features that are believed to be true of the referent. For instance, Gödel is believed to be the discoverer of the incompleteness theorem, and this descriptive feature is one of the things used to pick out the referent of the name ‘Gödel’. Likewise, water is believed to be a clear colourless liquid, and this is part of what picks out the referent of ‘water’. The causal theory of reference, however, rejects this picture. It holds that these descriptive features are not what determine the referent of a term; rather what determines the referent is that the use of the term is appropriately causally constrained by a particular entity. In the Gödel case, the referent of the name ‘Gödel’ is not determined by any particular descriptive information, but by the individual to whom the name ‘Gödel’ was given. In the water case, the referent of ‘water’ is not determined by any descriptive beliefs about water, but by the chemical composition of the stuff ‘water’ was initially used to refer to, namely H2O. In both the Gödel and water cases there is some language-independent entity which constrains the use of a particular term in such a way that, when it comes to

forms of truth

91

determining the referent of the corresponding term, this entity takes priority over any descriptive content we associate with the term: it is possible for most (if not all) of our descriptive beliefs about the referent to be false. What our words refer to is thus not something over which we have much control: what is important is how the use of our words is constrained by non-linguistic entities. This fits naturally with the motivations for the correspondence theory of truth, which holds that truth is a matter of a relationship between language and the non-linguistic world, with the world being the thing that determines truth, as opposed to truth being the thing that determines the world. This picture of truth fits naturally with the idea of responsive predicates discussed above. The property of being metallic is a sparse property with a broad cosmological role. Our use of the predicate ‘is metallic’ is responsive to this property, as there are causal interactions between that property and our thought and talk about it. As a consequence, when we utter sentences such as ‘the rod is metallic’, the truth of that sentence will be dependent on the possession of the property of being metallic by the rod, which is a property that the rod has mind-independently.

5.2.2 Non-representational theories of truth There are two main forms of non- representational theories of truth, namely coherence theory and pragmatism. Coherence theories of truth are sceptical of the correspondence theorist’s ability to separate the world from our abilities to represent it. Accordingly, they think that what makes a sentence true is its coherence with other sentences as part of a system. Different coherence theories cash this idea out differently, but the broad idea is that, when considering the truth of a sentence, we do not look for it to match the nonpropositional world; instead we look to see how it fits with other sentences taken to be true. I will not talk much more about coherence in this book, preferring instead to use the notion of superassertibility in my examples, but coherence theories have been applied to particular domains in a number of cases: see Wright (2003) on arithmetic, and Dorsey (2006) on ethics. Pragmatist theories share the coherence theorist’s concerns about correspondence theory, but look in slightly different places to understand truth. Pragmatists are concerned with the practical value of having true beliefs, and think that true beliefs must play a particular practical role in order to qualify as true. Different pragmatist theories develop this idea differently, but many pragmatist views give a key role to the notion of verification: true beliefs are those that we can verify to be true.4 This also ties truth to the notion of justification, as true beliefs must be beliefs that we are justified in believing, through a process of verification. Whether this verification is something that can take place now, or at some idealized point of inquiry is a matter about which different pragmatist theories disagree. 4 See, for example, James (1907) and Peirce (1878) for discussion of this. James is an interesting case, though, as in other works he is also interested in cases where we hold beliefs without verification, such as in his discussion of religious belief in James (1897).

92

Truth Re-Emerges

Pragmatism is a natural companion to the idea of abundance and generation discussed above. Consider, for example, the following excerpt from Peirce’s influential paper ‘How to Make Our Ideas Clear’: The opinion which is fated to be ultimately agreed upon by all who investigate, is what we mean by the truth, and the object represented in this opinion is the real. That is the way I would explain reality. (Peirce 1878)

The key idea here is that there is not some ‘independent’ reality out there, which makes beliefs true. Rather, it is the truth of certain beliefs that determines the contents of reality. If this is the case, then, unlike the correspondence theory, truth cannot be a matter of correspondence with mind-independent reality, for reality is not mind-independent, and reality is dependent on what is true, as opposed to what is true being dependent on reality. For our purposes, we will focus on the development of pragmatism given by Crispin Wright, who uses the notion of superassertibility as a candidate for the nature of truth. Superassertibility is a form of durable warrant to assert a sentence; a warrant which is available at some stage of inquiry, and then persists through all subsequent stages. The idea is defined by Wright as follows: A statement is superassertible, then, if and only if it is, or can be, warranted and some warrant for it would survive arbitrarily close scrutiny of its pedigree and arbitrarily extensive increments to or other forms of improvement of our information. (Wright 1992: 48)

There is a reason for using warrant in the domains we have identified as fitting the abundant model. This is because, due to the narrow cosmological role exhibited, there is no nature to the properties involved beyond what is believed about them and their bearers. As a consequence, truth in these domains cannot outstrip what we have warrant to believe or assert, as there is no nature to the properties in question beyond this. As Wright (1992: 57–61) suggests, superassertibility is only a viable model of truth in domains where all truths are knowable, as, if all truths are knowable, then there will be warrant available for all truths on the assumption that knowledge requires warrant. We can now show how truth can be understood in terms of superassertibility in the institutional and social domains. Consider the following explanation of the truth of the sentence ‘Andrew Cuomo is the Governor of New York’. The property of being the Governor of New York is an abundant property, so the possession of that property is not a matter of an object possessing some property mind-independently that causally interacts with our thought and talk. Instead, we should explain Andrew Cuomo’s possession of that property in a different way, and hence understand the truth of the sentence ‘Andrew Cuomo is the Governor of New York’ in a different way. The superassertibility model of truth gives us this account. The sentence is true just in case there is warrant available for it, and that warrant persists through future stages of inquiry. That warrant will involve the various steps that established Andrew Cuomo as the Governor of New York, involving the electoral process and his subsequent swearing in. Provided that all stands up, the sentence is true.

forms of truth

93

This account does justice to the idea that institutional truths depend on a complex system of collective acceptance and procedures, as opposed to a reflection of preexisting reality. Note that all institutional sentences will need to be considered in the context of the time that they are uttered, so we will be considering the superassertibility of ‘Andrew Cuomo is the Governor of New York’ as uttered in 2017. Considerations of durable warrant will concern whether we are warranted in asserting the sentence at the time the sentence was uttered, which makes sense given that institutional objects and properties are dependent on particular beliefs in particular cultural contexts. The social cases of race and gender are a little trickier than institutional cases, as, whilst in institutional cases there are usually clearly prescribed and publicly accessible rules for an object’s having a property, things are more opaque in social cases more generally. For instance, while there are clear conditions set out for being a Governor, there are not for being black or white. What we can do though is draw on the work of philosophers who aim to provide definitions of these predicates. Let’s consider a concrete example, namely Sally Haslanger’s (2000) analyses of gender and race:5 S is a man iff i) S is regularly and for the most part observed or imagined to have certain bodily features presumed to be evidence of a male’s biological role in reproduction; ii) that S has these features marks S within the dominant ideology of S’s society as someone who ought to occupy certain kinds of social position that are in fact privileged (and so motivates and justifies S’s occupying such a position); and iii) the fact that S satisfies (i) and (ii) plays a role in S’s systematic privilege, i.e., along some dimension, S’s social position is privileged, and S’s satisfying (i) and (ii) plays a role in that dimension of privilege. (Haslanger 2000: 42) S is white iff: i) S is observed or imagined to have certain bodily features presumed to be evidence of ancestral links to a certain geographical region, Western Europe; ii) S’s having (or being imagined to have) these features marks them within the dominant ideology of S’s society as someone appropriately occupying a privileged social position (and so motivates and justifies their occupying such a position); and iii) S’s satisfying (i) and (ii) plays (or would play) a role in their systematic privilege, i.e., S is along some dimension systematically privileged when in C, and satisfying (i) and (ii) plays (or would play) a role in that dimension of privilege.6 5 6

I use Haslanger’s view as an example. See Mikkola (2009, 2011) for critiques of Haslanger’s view. This is derived from Haslanger’s following analysis of racialized groups:

A group G is racialized relative to context C iff members of G are (all and only) those: i) who are observed or imagined to have certain bodily features presumed in C to be evidence of ancestral links to a certain geographical region (or regions); ii) whose having (or being imagined to have) these features marks them within the context of the background ideology in C as appropriately occupying certain kinds of social position that are in fact either subordinate or privileged (and so motivates and justifies their occupying such a position); and

94

Truth Re-Emerges

Given that we are understanding social truth in terms of superassertibility, we can view these conditions as being the conditions under which we are warranted in asserting the sentences ‘S is a man’ and ‘S is white’. Recall that superassertible claims are those that one has warrant to assert, and one will retain that warrant throughout future stages of inquiry, so in order for a particular sentence to be true we would need to have not just warrant for it, but also for that warrant to be durable. The question of whether ‘Angela Merkel is white’ is true will be answered by whether I am durably warranted in asserting the sentence ‘Angela Merkel is white’. This would be settled using Haslanger’s analysis, and determining whether I am durably warranted in thinking that Angela Merkel meets the conditions specified. If I am, then the sentence is true, and if I am not, then the sentence is false. As with institutional truths, we will have to assess social sentences from within a particular social context, as is made clear in the conditions Haslanger gives. Thus, we will need to assess the claim ‘Angela Merkel is white’ as a sentence uttered in 2017, and examine whether it is durably warranted in that context.7 The route from truth to property instantiation here is as follows. If I have durable warrant to assert ‘Angela Merkel is white’, then ‘Angela Merkel is white’ is true. Now that we have the truth of ‘Angela Merkel is white’, we can infer that the predicate ‘is white’ is true of Angela Merkel, and hence that Angela Merkel falls under the predicate ‘is white’. From this we can infer that Angela Merkel has the property of being white.8

5.3 Interim Summary At this stage, then, we have reached the point where we have distinguished between kinds of predicates, and stated that, for each kind of predicate, we can ask whether it fits the responsive model or the generative model. We also discussed an example of a predicate kind that fits each, suggesting that it is not the case that only one of the proposed models holds across the board. We then connected these ideas with truth by showing that each model comes with its own account of truth, and that, if we want to allow for there to be different relationships between language and the world of this sort, then we need to allow truth pluralism. iii) whose satisfying (i) and (ii) plays (or would play) a role in their systematic subordination or privilege in C, i.e., who are along some dimension systematically subordinated or privileged when in C, and satisfying (i) and (ii) plays (or would play) a role in that dimension of privilege or subordination. (Haslanger 2000: 44) 7 Thus, even if our social conditions change and the inequalities associated with race and gender subside, leaving us with no use for the terms involved, we will still be durably warranted in thinking that Angela Merkel is white in 2017, as the context of 2017 is a time when those inequalities were in place. 8 We can note here that in this case the only reasons we would have to reject the truth of the sentence ‘Angela Merkel is white’ would come from consideration of aspects internal to the investigation. It is not as though we are trying to map onto some independent notion of what there is, leaving us open to failure if someone demonstrates for independent reasons that there are no white people. If there are no truths about whiteness, it will be because no adequate definitions of what it is to be warranted in asserting sentences of the form ‘S is white’ can be given.

problems with monist approaches

95

The situation we end up in is one where truth pluralism is licensed. For now, we can define the view by saying that it is the view that different theories of truth are required for different domains, and we will go into the specifics of the view in Chapter 7.

5.4 Problems With Monist Approaches The argument so far has been premised on the idea that we do wish to hold that there are different connections between language and the world in different areas. To close this chapter I want to consider the situation if we reject this, and enforce monism. That is, suppose one wants to adopt the sparse model everywhere, or the abundant model everywhere: what (if anything) is wrong with doing so? I will consider abundant monism first, and argue that there is an inherent problem with this model. As a consequence, if one is tempted by the abundant model anywhere, one also needs to adopt the sparse model in at least one other area. This essentially converts any argument for the abundant model into an argument for pluralism. In the case of sparse monism, I will suggest that there is not an inherent problem in taking it to apply globally, but it does force one into some fairly radical positions.

5.4.1 The problem with abundant monism 5.4.1.1 global deflationism As previously discussed, when it comes to abundant properties, truth plays an important role in their characterization. This is because abundant properties are just the extensions of their corresponding predicates. When it comes to the question of what determines the extension of a predicate, we use truth to say that the extension of a predicate is determined by the true sentences in which it features. This gives truth a key role in the generation process for abundant properties, and raises a problem for thinking of truth itself as an abundant property. Here is the concern. If truth is an abundant property, then to have the property of being true is just to be in the extension of the predicate ‘is true’ (this is the definition of an abundant property). But, the extension of the predicate ‘is true’ is determined by the truth of sentences involving the predicate ‘is true’. The italicized instance of truth in the previous sentence indicates a role for the property of being true in the generation process of the abundant property of being true. This suggests that having the property of being true cannot simply be a matter of being in the extension of ‘is true’, because what is in that extension is determined by what is true. As a result, truth cannot be an abundant property: there has to be more to it than the extension of ‘is true’, as the extension of ‘is true’ is determined by what is true. This argument suggests that we cannot think of truth as an abundant property, as its role in the general account of abundant properties precludes it from being an abundant property itself. This is an interesting result, as it shows that there is an incompatibility between this view of abundant properties and the idea that truth is an abundant

96 Truth Re-Emerges property. In order to get the idea of abundance off the ground, we need there to be at least one non-abundant property—the property of being true—so we cannot generate global abundance. This is a problem for a deflationary form of abundant monism, for it shows that truth cannot be an abundant property. This is a problem even if deflationists cannot accept the distinction between sparse and abundant properties, as deflationism’s account of properties just is the abundant account of properties. The fact that they are barred from really calling it this does not mean that the properties themselves are any different. This highlights a significant problem for global deflationism, and calls into question whether the view as a whole can be sustained. 5.4.1.2 pragmatist monism We also now have the resources to show that a monist form of pragmatism which takes truth to be superassertibility is also untenable. This pragmatist view is one that holds that all properties are abundant, and that that truth is superassertibility. We can now run a similar argument to the one above about how truth cannot be an abundant property to show that we get the same problem with superassertibility. Suppose that we have maintained the abundance of truth by showing that the extension of the predicate ‘is true’ is determined by what has the property of being superassertible. Now we have to consider the property of being superassertible. By the terms of abundant monism, this needs to be understood just as the extension of the predicate ‘is superassertible’, which is determined by truths involving the predicate ‘is superassertible. However, on this pragmatist picture, truth just is superassertibility, so we have the extension of the predicate ‘is superassertible’ being determined by the things that are superassertible. But, being superassertible, on this view, can only be being in the extension of ‘is superassertible’, so we have no way of explaining how it is that things are superassertible. This problem would apply again if we tried to ground the extension of ‘is superassertible’ by another property, as the abundant monism of properties means that we would have no way to ground that further property. The only solution is to admit that there is at least one non-abundant property of being true, which is to reject the pragmatist form of abundant monism in favour of a form of pluralism about properties and truth.

5.4.2 Sparse monism Sparse monism takes it that the only properties that exist are sparse properties, and that any properties we have described above as ‘abundant’ properties do not exist. There is no process of generation, and the only relationship between predicate and property is the responsive one. It also means that any sentences involving predicates that we have classed as generative need to be re-examined.9 9

An example of sparse monism is Armstrong’s (1978, 1997) ‘a posteriori realism’.

problems with monist approaches

97

One option would be to find a way to show that they are actually responsive. For instance, one could try to show how a predicate we have taken to be generative could be shown to be responsive after all, perhaps because we can show that the property in question can be reduced to a sparse property. This is a strategy one could take in ethics, for example, where one would try to show that a moral property, such as being good, is reducible to a sparse property, such as being pleasurable. However, the history of attempts to do this in the moral case has been riddled with problems,10 and in other cases it looks like a non-starter. In the social case above, for example, it was shown to be mistaken to think that race and gender properties can be identified with sparse properties, and the same argument could be made for institutional properties too. For instance, one could try to reduce the property of being the Governor of New York to the property of being Andrew Cuomo, but this does not look promising as they have different modal features (e.g. it is true that Andrew Cuomo might not have been the Governor of New York, but it is not true that Andrew Cuomo might not have been Andrew Cuomo). The reductive strategy is a difficult one to pull off, and it would need to be done in a variety of different areas if one wanted to preserve the existence of a wide variety of properties and a wide variety of true claims. Supervenience is also a tricky option here, as, even if one says that the problematic properties supervene, or are dependent, on the sparse properties in some way, they still exist as supervenient or dependent, and not sparse, entities.11 A second option would be to say that the properties are fictional entities.12 That is, to distinguish between the real things (the sparse things) and the fictional things (the abundant things). This still technically violates global sparseness unless we say that fictional properties do not exist. If we say that fictional properties do not exist, though, then there is no gain made by making the distinction between real and fictional entities. On the other hand, if we allow fictional entities to exist, then we are stuck with a number of non-sparse existing things, which is exactly where we started when considering the rejection of abundant properties. Another alternative is to say that there are no non-sparse properties, and hence every sentence that positively attributes a non-sparse property to an object is false. This is an extreme move, and one that renders many sentences of ordinary communication false. For example, we would want to distinguish between the sentence ‘Andrew Cuomo is the Governor of New York’ and ‘Stone Cold Steve Austin is the Governor of

See, e.g. Moore (1903), and Horgan and Timmons (1991). See Edwards (2014: ch. 6) for more on this. See also Schaffer (2009) for the idea that dependent entities exist. 12 For discussions of fictionalism about mathematics, see Field (1980); ethics, see Kalderon (2007); social entities, see Thomasson (2003b); modality, see Rosen (1990). For a general discussion of fictionalism, see Sainsbury (2009). 10

11

98

Truth Re-Emerges

New York’ in that the former is true and the latter false. But, if there is no such property as being the Governor of New York, then both of these sentences are false.13 Perhaps one could apply the reductive strategy in whatever domains it can work for, and then deny the existence of anything that does not fit into those domains, and hence deny the truth of any claim that involved a non-responsive predicate. At this stage one would need to seriously question the motivations for adopting sparse monism in the first place. If what is driving it is the idea that the only existing things can be those that have the characteristics of sparse entities, then the fact that such a view leads to this conclusion may require revisiting the motivations for this view. We will examine this idea in relation to existence later on in Chapter 6.14

5.4.3 Correspondence pluralism? Finally, one might worry that I have set things up unfairly, and missed an option that shows that we can maintain one theory of truth for both sparse and abundant properties. That is, why associate the correspondence theory of truth exclusively with the sparse model? Instead, we might think that we can always account for truth in the same way—correspondence—but that there are differences in the entities that true sentences correspond to. As Shapiro (2009) says, there may be independent reasons for correspondence theorists to pursue this route, as the theory faces its own problem of scope: Given the way the account is presented and motivated, it does not even extend to propositions about non-medium-sized physical objects, nor to even to mildly theoretical statements about medium-sized physical objects. Consider the propositions expressed by the following sentences: My tire is inflated with nitrogen. The gravitational field around the earth is several times as strong as the gravitational field around the moon. Clearly, nothing like the above “toy” accounts of denotation will work for concepts like and . To belabor the obvious, nitrogen does not “cause, under appropriate conditions, mental tokenings of ” . . . For one thing, we are (almost) always in the presence of nitrogen. Similarly, mental states with gravitational-field-ish content are not causally responsive to an external environment that contains gravitational fields. Every external environment contains a gravitational field. (Shapiro 2009)

13 Compare also this approach in the race and gender case. On this view, the sentence ‘Angela Merkel is a white woman’ would be false, as there are no white people and there are no women. The sentence ‘Angela Merkel is a black man’ would also be false, for the similar reason that there are no black people and there are no men. 14 One might wonder why I do not discuss the idea that sentences in the problematic domains are not even truth-apt, i.e., why not take a non-cognitivist approach? There are two issues with this. The first is that, as noted in Chapter 4, the non-cognitivist approach is compatible with the proposal for abundant properties given above, so it would not be an option for a sparse monist. It is also compatible with the idea that the relevant sentences are true: for more on this, see Blackburn (1993), Lynch (2013b), and Smith (1994). Secondly, whilst non-cognitivism may have support in some domains, like ethics, it is hard to make sense of it in the institutional domain, for example.

problems with monist approaches

99

To elaborate, take as an example the idea of the causal reference relation serving as the relevant connection that we discussed earlier. In terms of beliefs, there is reason to favour this sort of model for simple perceptual beliefs, such as the belief that the tree is solid, where we take there to be a causal relation between the tree’s solidity and my belief that the tree is solid in the event that that belief is true. Even if one has some suspicions about the precise details of such an account, there is a powerful thought stemming from representational theories of the mind that the reason our perceptual beliefs are true is because reality is a certain way, and because our minds accurately represent reality when we have true beliefs. This simple causal model does not work so well when it comes to other sorts of beliefs about the material, or natural, world. Take the belief that there is a star fifteen million light years from Earth. This belief seems to be something that should be true in virtue of how things are independently of us, but, if true, it is not true because of some causal relationship between the star and my belief, as it appears that there is no such relationship at present. If this is so, then correspondence theorists would do well to consider a variety of different reference relations as candidates for the connection between truth and reality in these different settings. This has led some to suggest that the way forward for correspondence theory is to consider the idea that there are different relationships of reference or satisfaction. Here is Shapiro, discussing Michael Lynch’s book Truth as One and Many (Lynch (2009)): I suggest that the overall [pluralist] theory be re-written as a functionalist account of satisfaction, a relation between sequences of objects and relations. I don’t know if it is possible to come up with a list of core truisms for the folk concept of satisfaction, since there may be no such folk concept—unless it is the concept underlying the locution “true of ”. Nevertheless it should be possible to articulate its functional role and to speculate on the range of relations that play this role in various stretches of discourse. I do not know how amenable this suggestion is to the intuitions that underlie the treatment in [Lynch’s] book, but, for what it is worth, I don’t see any other way of bringing mathematics and science into the fold. In any case, the spirit of [Lynch’s] book would be maintained. Since truth can defined in terms of satisfaction, a functionalist pluralism about satisfaction would be a functionalist pluralism about truth. (Shapiro 2009)

And Gila Sher: According to this approach, the plurality of truth is rooted not in differences between complete propositions but in differences between sub-propositional units: names, predicates, and functions. As a result, the plurality of truth is reduced to a plurality of reference, satisfaction, and fulfilment. [Causing pain is bad] is realised by a combination of three distinct forms of reference/satisfaction/fulfilment: M-reference, P-fulfilment, and R-satisfaction, where M, P, and R are the forms of reference/satisfaction/ fulfilment applicable to mental, physical, and moral concepts, respectively. (Sher 2004: 322)

Sher (2004, 2005, 2015, 2016) develops this idea further, and holds the view that, in order to make sense of correspondence, we need to hold that there is a variety of different correspondence relations. I want to now compare Gila Sher’s (2016) theory of ‘manifold

100

Truth Re-Emerges

correspondence’ to my account given above to see whether plural correspondence is as effective as pluralism proper. Sher lists the three main components of manifold correspondence as: A. Correspondence is, in a deep sense, a matter of both world and mind. B. Correspondence need not have a simple form, like mirror, copy, or isomorphism; it might have a far more intricate form. C. Correspondence might take several forms, varying from field to field. (Sher 2016: 189, italics in original) To elaborate on feature C, she earlier says: The forms [correspondence] takes depend both on what aspect of reality a given thought targets and on the cognitive resources available to us for reaching it. Abstracting from differences, this connection holds between a given thought and reality when the aspect of reality it targets is, directly or indirectly, yet systematically, as it says it is. (Sher 2016: 186)

Sher’s motivations for moving to a pluralist form of correspondence are partly due to the thought that the standard causal model of correspondence is unsuitable to explain truths about mathematics and logic. She develops detailed examples to show how such accounts of correspondence can be given in these areas. The idea is to retain the unity of truth through truth always being correspondence, but to acknowledge that there are different correspondence relations in different domains. This seems on the face of it to constitute a rival form of pluralism to the view outlined above, where correspondence theory works in some domains, with superassertibility operational in others. The question I want to explore here is whether Sher’s form of plural correspondence constitutes a threat to this approach. The first thing to note is that, given that Sher’s view is a representational theory of truth, it is—like all representational theories—committed to the idea that the truth of a sentence is dependent on the existence of the objects and properties that the sentence is about. Correspondence cannot avoid this idea, even if we take it that parts of reality are constructed by us. As a consequence, this view cannot avail itself of the idea that some properties are instantiated because their corresponding predicates are true of objects, as this would require cases where truth is not dependent on how reality is, due to some aspects of reality being dependent on truth. As a result, such a view needs an independent account of abundant properties if it is to be able to capture both sparse and abundant properties. This view would thus need an entirely separate account of how aspects of reality are constructed that does not depend on truth in the way I have described above, or indeed how I will describe in subsequent chapters below. One issue is how broad the scope of this view is, and whether it can offer the same level of flexibility as the approach outlined above. One of the virtues of my approach is that it can appeal to a wide variety of theories that have vastly different views of a domain. For example, my approach is consistent with an expressivist approach to ethics, and anti-realist approaches more generally, as they can adopt the abundant model for

problems with monist approaches

101

the moral domain. Is this the case for plural correspondence, though, or is the view unable to account for genuinely anti-realist options? In her remarks on ethics, Sher suggests that a commitment to plural correspondence enforces a particular view of moral discourse: What about other areas?—Composite correspondence carries some promise for other areas as well. Take ethics, for example. Our approach suggests that in investigating truth in ethics we don’t start by trying to fit ethics into the standard template of correspondence, inspired by Tarski’s theory, with its emphasis on individuals and their properties. Instead, we try to understand morality in its own terms. We try to figure out whether there is, or should be, something objective in our moral judgments, something that transcends our subjective and communal emotions (preferences, etc.), and if there is, what kind of thing it is. And we try to understand how humans reach this objective ground of morality, what standard (pattern) of correspondence is involved, and how it relates to the standards (patterns) of correspondence in other fields. In this way, if there is truth in ethics it will be based on correspondence—composite correspondence, like all other fields. (Sher 2015: 209)

The talk of correspondence here only comes in in relation to talk of an objective ground to reality, which implies transcendence of subjective and communal emotions and preferences, which is not talk that is compatible with anti-realist approaches to ethics; indeed, it indicates the exact opposite of those approaches! One of the virtues of my approach is that it is neutral on these issues, and taking a stance on truth does not rule out in advance a way of thinking about a domain, but this is not something available on Sher’s view. Indeed, it is notable that her two key examples of areas where composite correspondence is useful are mathematics and logic, two areas where realist intuitions are powerful. However, perhaps this is unfair to Sher’s view: after all, she does not make any explicit claims about the social and institutional domains that we have been considering in detail. Let’s make the case for plural correspondence as strong as we can, and give the plural correspondence theorist even more than perhaps Sher wishes to allow, and reject the idea that all aspects of reality are independent from thought and language.15 Suppose that we take the idea that there is natural reality and constructed reality. Constructed reality is formed by stipulating definitions, like those Haslanger gives above. These definitions give the conditions under which an object has a certain property, and hence satisfies a predicate. Thus, when it comes to truth, we can apply the same representational reading of (TP), and hold that truth is dependent on predicate satisfaction because the truth of a sentence will be determined by whether an object has a property; the only difference between the natural and constructed domains will be the way in which the object gets to have that property. In the natural domains the 15 See, e.g. David (2015) and Fumerton (2013) for claims that the correspondence theory of truth is able to have these features. Though, note that, after claiming this, David says: “Keeping this point in mind, one can nevertheless acknowledge that advocacy of a correspondence theory of truth comes much more naturally when combined with a metaphysically realist stance and usually signals commitment to such a stance”. (David 2015).

102

Truth Re-Emerges

object will have that property because of its participation in certain causal relations (say), whereas in the constructed domains the object will have the property because it meets some stipulated definition. When it comes to saying what reference is, we can say that truth depends on reference, but reference is understood differently in different domains: in the natural domains it is understood causally, whereas in the constructed domains it is understood in some other way. The key issue here is going to be this non-causal reference relation and whether there is room for it to be understood in a way that both allows it to serve as part of a correspondence theory of truth and avoid the problems for the causal theory of reference in constructed domains. Let’s first examine in more detail why social and institutional entities may not be suited to a causal theory of reference. Possession of social and institutional properties depends on collective acceptance by a community, and the beliefs that are accepted in one’s community are part of what sustains the background necessary for an object to bear a social or institutional property. As Thomasson (2003a: 591–2) argues, this makes a causal theory of reference problematic in the social and institutional domain for the following reason.16 We can begin by noting that, on a causal theory of reference, it is possible for all the descriptive beliefs that a person or a community has about the referent of a term to be false, and yet reference to still succeed (witness, for example, Kripke’s (1980) Gödel/Schmidt case, where it turns out that we are still able to refer to Gödel even in a hypothetical situation where all our beliefs about what Gödel did are false). This cannot be the case here, though, as beliefs about institutional and social properties are necessary for the possession of those properties in the first place: there is no room for widespread error, as there is no gap between what is believed about these properties and the properties themselves. The causal theory of reference allows a gap between what is believed about a referent and the referent itself, but this gap does not exist in social and institutional cases.17 An additional concern relates to the idea of responsiveness used in the causal theory of reference, namely that the causal theory of reference is designed to help language track entities that are already there, and have their nature independent of language. Social and institutional entities, with their reliance on collective acceptance and belief, are not such entities. Now, even though we are stipulating that a causal account of reference is not in play, for a correspondence model to make sense in any way, we need 16 See also Appiah’s arguments concerning race and the causal theory of reference in Appiah (1994; 2007: 25–6). See Glasgow (2017) for a discussion of the role of reference in debates about race. 17 It is worth noting that this argument is not uncontroversial. As we saw in Chapter  4, Haslanger (2000), for example, advocates an ‘analytical’ approach to race and gender where the stated aim does not involve giving an account of what is generally believed about who bears certain properties (indeed, it is consistent with her account that some ‘ordinary’ beliefs about race and gender are false). Those who favour these sorts of accounts may wish to look to the additional considerations given below for why the responsiveness model is inappropriate in the social and institutional domain, which, as noted below, hold regardless of the view we take of the connection between descriptive beliefs about a property and the nature of that property.

problems with monist approaches

103

to have the idea that the portion of reality in question is there prior to reference being made to it. This is what Sher means above by the idea that our thoughts ‘target’ that reality. You can’t hit a target that isn’t there, and truth cannot depend on reference if the entities being referred to do not exist prior to the claim being made. As a consequence, it has to be the case that social and institutional entities exist prior to the assertion of sentences about them. The problem with this way of thinking about these domains is that the creation of social and institutional properties is not a one-time thing, where a property is created by stipulation and then is there for all to refer to whenever they wish; rather the existence and nature of the property is only sustained by the practice of continuing to refer to it, attributing it to things, and maintaining collective beliefs about it. Moreover, the nature of the entity can change depending on how those who bear it view it and view themselves. For example, take Hacking’s example of being a genius, and the ‘looping’ phenomenon he discusses: People of [certain] kinds can become aware that they are classified as such. They can make tacit or even explicit choices, adapt or adopt ways of living so as to fit or get away from the very classification that may be applied to them. These very choices, adaptations or adoptions have consequences for the very group, for the kind of people that is invoked. The result may be particularly strong interactions. What was known about people of a kind may become false because people of that kind have changed in virtue of what they believe about themselves. I have called this phenomenon the looping effect of human kinds (Hacking 1995). Looping effects are everywhere. Think what the category of genius did to those Romantics who saw themselves as geniuses, and what their behavior did in turn to the category of genius itself. Think about the transformations effected by the notions of fat, overweight, anorexic. If someone talks about the social construction of genius or anorexia, they are likely talking about the idea, the individuals falling under the idea, the interaction between the idea and the people, and the manifold of social practices and institutions that these interactions involve: the matrix, in short. (Hacking 1999: 34)

On the correspondence model described above, the property of being a genius would come into existence via a stipulation or definition of some sort, and then sentences attributing this property to individuals would be true insofar as the individual met the conditions stated in the definition. This reflects the idea that constructed reality is there, constructed by us, complete with all its entities, and thus something that fits the representational model of truth, even if reference is not causal. The problem is that this is a poor account of constructed reality, and it does not properly account for the continued existence of institutional entities. In particular, our reference to these properties, practices of attributing them to things, and beliefs about them, is part and parcel of what sustains their continued existence and nature.18 As a consequence, even this 18 This is so even on the Haslangerian account of race and gender mentioned in the note above, as, even if we relax the tie between the nature of race and gender and descriptive beliefs about these properties, the existence of races and genders is still dependent on a sustained variety of social practices where people are treated in certain ways depending on the race or gender properties they are taken to have.

104 Truth Re-Emerges liberal correspondence model looks like a poor fit, and plural correspondence does not provide a viable alternative to the pluralist approach developed in this book.

5.5 Summary The main aim of this chapter has been to argue for truth pluralism. The main idea was to show that, if one takes the sparse and abundant models in Chapter 4 seriously, and the idea that each model is appropriate in different domains, then pluralism about truth follows. This is because, I have argued, the models are committed to different ways of thinking about truth. I then developed some of these ways in more detail, and showed how they apply to our previous examples. Finally, I discussed some issues with global abundance and global sparseness which add further weight to the case for pluralism. We now move in Chapter 6 to discuss how this picture can be augmented with ontological pluralism.

6 From Truth to Being In Chapter 5 I motivated truth pluralism by thinking about the relationship between predicates and properties, arguing that this relationship varies across domains, and that this requires that truth vary across domains. In Chapter 4 I mentioned the other components of atomic sentences—singular terms—and their corresponding entities— objects—and said that we would come back to them. It is now time to do so. I will suggest that consideration of singular terms and objects will lead us from truth pluralism to ontological pluralism, the idea that things exist in different ways. In this chapter I explore the connections between truth pluralism and ontological pluralism, and develop the features of a global pluralism, which includes pluralism about truth and existence. I begin by noting that some motivations for truth pluralism can also be applied to ontological pluralism, before demonstrating how a method similar to my argument in Chapter 5 for truth pluralism can also be used to give an argument for ontological pluralism. I then discuss how the views complement each other, and how ontological pluralism can add to our understanding of domains by highlighting differences between the ways things exist. Once we have the pluralist metaphysical picture up and running, I demonstrate its explanatory power by comparing it to global deflationism. In doing so, I expose further problems for global deflationism.

6.1 Ontological Pluralism Ontological pluralism, as it is to be understood here, is the thesis that things exist in different ways.1 The purpose of this section is consider whether there are motivations for the view which spring from the project of giving an analysis of the nature of existence. In particular, I will argue that there are striking parallels between debates about existence and debates about truth, and that these parallels suggest that if we are motivated to be truth pluralists, we should also be motivated to be ontological pluralists. Ontological pluralism is usually formulated in reference to objects, as opposed to other sorts of entities. This is because of the emphasis on quantification in debates about existence, which we will explore in more detail in Chapter  8. If existence is 1 See, e.g., McDaniel (2009, 2010a, 2010b, 2017), and Turner (2010, 2012) for canonical contemporary discussions of ontological pluralism.

106

From Truth to Being

understood in terms of quantification, and quantification is objectual, in the sense that the variables bound by quantifiers stand for objects, then the question of what it is to exist is just is the question of what it is to be an object. As a consequence, different accounts of existence are primarily applied to objects, but, as I will suggest below, once we free ourselves from thinking about what existence is exclusively in terms of quantifiers, we can also think about the different ways entities of other sorts exist, such as properties.

6.2 Two Scope Problems My suggestion is that the ontological pluralist begins by asking the following question: ‘What can’t we do if we don’t admit different ways of being?’, and gives the answer that we cannot provide a satisfactory account of existence. If it turns out that ontological pluralism is required in order to give a satisfactory account of existence, then evidently that will count in favour of the view. Let’s now examine a metaphysical motivation for ontological pluralism that connects back to arguments for truth pluralism. In particular, we can develop similar ‘scope problems’ for both the study of truth and the study of being.

6.2.1 The scope problem for truth In Chapter 5 we discussed the idea that different theories of truth are appropriate for different kinds of subjects. This general idea is also motivated by what is often referred to as the ‘Scope Problem’ (see Lynch 2009, Sher 2004), which we can briefly rehearse here. Take the correspondence theory of truth, which holds that to be true is to correspond to the facts. Such a view is plausible for the truth of beliefs about objects in the material world, such as trees and mountains, but is poorly suited to account for the truth of mathematical or moral beliefs. This is due to the fact that correspondence is generally understood in broadly causal terms as a relation between an observer and a mind-independent realm of objects and properties. In the mathematical case, it seems implausible that the truth of mathematical beliefs consists in a causal relation between an individual and a fact, whereas in the moral case problems often arise when we conceive of the facts in question as mind-independent. On the other side of the coin we have more broadly anti-realist accounts of truth, such as the idea that to be true is to be superassertible, with social and institutional truths being two examples. Whilst this account may serve reasonably well for institutional and social truth (and perhaps even moral and mathematical truth), it is often thought to be poor as an account of truths about the material world. After all, truths about the nature and state of the material world depend on the material world itself, as opposed to what we have warrant to believe about it. In light of these concerns, it is hard to see how we can have a substantive account of the nature of truth that will hold across the board. This provides motivation for thinking

two scope problems

107

about truth pluralistically. That is, the way to respond to these concerns is not to give up on the project of accounting for the nature of truth but rather to embrace the idea that there may be different things to say about the nature of truth in different domains. This opens up the idea that truth may best be thought of as correspondence for discourse about the nature and state of the material world, but understood differently— perhaps in terms of superassertibility—for discourse about mathematics or ethics. So there is a way to turn the failures of the traditional debates into positive directions for the analysis of the nature of truth by thinking of truth plurally.

6.2.2 The scope problem for existence Do we find the same thing with existence? Yes: just as correspondence and superassertibility represent realist and anti-realist ways of thinking of truth, there are parallels in existence. Let us take the realist account first. This approach is summed up by the Eleatic Principle: Eleatic Principle:

To exist is to have causal powers.2

This reflects a broadly naturalistic worldview: it is the job of science to tell us what things exist, and these things will be the things that participate in the push-and-pull of spatiotemporal life. However, it should be clear that this account will fare badly when it comes to accounting for the existence of abstract objects, such as numbers. Abstract objects are usually defined as the very kind of entities which lack causal powers, and spatiotemporal location, and, as a result the Eleatic Principle cannot serve as an adequate account of their existence. Of course, the advocate of the Eleatic Principle as a general account of existence is free to deny that there exist any such entities on the grounds that they fail the satisfy the criterion, but this leads to a problematic form of sparse monism that we discussed in Chapter 5. Coming at things from the other direction, let us look at an alternative account of existence, which we can think of as the development of a broadly anti-realist approach. This account is a development of Berkeley’s famous principle: Esse Est Percipi: To be is to be perceived.3

2 The principle is referred to as the ‘Eleatic Principle’ by Armstrong (1978), as the idea can be traced back to Plato’s The Sophist where the Eleatic Stranger, in an attempt to state the mark of being, says “My notion would be, that anything which possesses any sort of power to affect another, or to be affected by another, if only for a single moment, however trifling the cause and however slight the effect, has real existence; and I hold that the definition of being is simply power.” Plato, The Sophist (trans. N.P. White), p. 247e. For a detailed examination of this idea in the context of The Sophist, see Brown 1998. It is also referred to as ‘Alexander’s Dictum’ due to the work of Samuel Alexander (1920), and this name was popularized by Kim (1993). 3 Berkeley, of course, said that the existence of an idea consists in its being perceived. But, given that he took objects to be ideas, we feel licensed to give this more general formulation here.

108

From Truth to Being

Modern applications of this principle have been developed in various constructivist proposals in various fields, some of which have already examined in relation to the existence of social and institutional entities, where there the broad thought is: Intuitionist Principle:

To be is to be constructed.4

This view has been developed in relation to the intuitionist programme in the philosophy of mathematics, which will provide us with a useful case study to focus on. While I will follow the mathematics example, as we will see, the model can also be applied to social and institutional objects (though, as we will see, there are important differences too). The thought here is that mathematical practice consists in the development of proofs, and the existence of mathematical objects, and the properties they have, comes about by way of proof. We can develop this idea further by looking at a particular interpretation of the ‘Neo-Fregean’ work of Bob Hale and Crispin Wright.5 We can take them to substantiate the general constructivist thesis by offering the following principle: Neo-Fregean Principle: in a true sentence.

To be is to be the referent of a singular term that appears

This approach began as an attempt to ground the existence of mathematical objects, but it has also been applied as a global account of existence:6 The lynch-pin of Frege’s platonism, according to our interpretation, is the syntactic priority thesis: the category of objects . . . is to be explained as comprising everything which might be referred to by a singular term, where it is understood that possession of reference is imposed on a singular term by its occurrence in true statements of an appropriate type. (Wright 1983: 53) [O]bjects, as distinct from entities of other types (properties, relations, or, more generally, functions of different types and levels), just are what (actual and possible) singular terms refer to. (Hale and Wright 2005: 171)

The basic idea that I will explore here is that we can secure reference to a mathematical object—the number 5, say—by noticing that its associated singular term appears in at least one true sentence. Thus, the truth of the sentence ‘the number 5 is prime’ is sufficient to secure the existence of a referent for the singular term ‘the number 5’: the number 5. It should be clear on this model that truth is understood in terms of proof, See, for example, Shapiro (2000: 180), and Pedersen (2014). The constructivist interpretation of Hale and Wright explored here is not uncontroversial, as many see the Neo-Fregean project as a way to vindicate a realist approach to mathematical entities. I do not wish to engage in a dispute about Hale and Wright exegesis here, and merely note that this is one potential way to develop the proposal which I take to be consistent with their remarks. See Tennant (2003) and MacBride (2003) for more on these interpretative problems. 6 See Sider (2007), Hawley (2007), and Eklund (2006) for more discussion here. 4 5

two scope problems

109

and that what makes the sentence ‘the number 5 is prime’ true is the availability of a proof for the claim that the number 5 is prime (i.e. a proof which shows that the number 5 is divisible only by 1 and itself).7 This idea connects with the notion of an abundant object discussed in Chapter 4. If we hold that an object is abundant, then it cannot be that ‘a’ refers to a because a exists, as we are explicitly denying that a has existence prior to ‘a’ referring to a. As a consequence, we need some grounds to establish in virtue of what it is that ‘a’ refers to a. Hale and Wright offer an account here in terms of truth: ‘a’ refers to a if ‘a’ appears in a true sentence. Now, whatever the prospects for the Neo-Fregean Principle as an account of the existence of mathematical objects, it should be clear that it has some serious limitations. Consider material objects like trees, mountains, and the Eiffel Tower. Is it plausible that the existence of the Eiffel Tower consists in the truth of a sentence which contains its associated singular term, such as ‘the Eiffel Tower is in Paris’? It seems not, and, indeed, such a view would be an even more extreme version of the Esse Est Percipi principle; for the existence of the Eiffel Tower is supposed not simply to consist in its being perceived, but rather in the appearance of a singular term in a true sentence. But surely, the existence of the Eiffel Tower is not, in any sense, grounded in the singular term of some true sentence. In short, whilst the Neo-Fregean Principle might be a promising approach for the existence of mathematical objects, it is not plausible for physical objects. Such problems suggest that we can construct an analogue to the Scope Problem for existence. Indeed, this should hardly be surprising, as it is very natural to pair the accounts of existence we have considered with the accounts of truth we have considered. For instance, the Eleatic Principle fits well with the correspondence theory of truth, with the thought being that everything that exists has causal powers, and the truth of a belief consists in there being certain corresponding causal relations between the believer and the existent objects and properties. Likewise, the superassertibility account of truth fits nicely with the Neo-Fregean Principle: if truth is connected intimately with proof, which yields durable warrant, then the existence of objects will be secured by their being referred to in a proof. These findings give support to a pluralist account of existence: that there is more to say about the nature of existence, and what there is to be said varies from one kind of object to the next. This is indeed the conclusion we saw drawn in the parallel case of truth, where the Scope Problem shows that we need to think of truth differently in different domains.

7 For thoughts on the Neo-Fregean treatment of mathematical properties, see Hale and Wright (2009). For more discussion of these issues in relation to properties more generally, see Chapter  7, and also Edwards (2014).

110 From Truth to Being

6.3 Truth Pluralism + Ontological Pluralism I now wish to connect this discussion back to the way that we motivated truth pluralism in Chapter 5. The key issue here is the relationship between truth and being. The view that I will now develop holds that the relationship of dependence between truth and being varies, so that in some cases truth depends upon being, and in other cases being depends upon truth. This maps onto the different kinds of truth and being identified above to demonstrate how the ways of being that an ontological pluralist posits connect to the ways of being true that the truth pluralist posits. What we get out of this is a global pluralist metaphysics, incorporating pluralist views of both truth and existence. A good place to start is the principle we previously identified in Chapter 5 as (TP): (TP)

‘a is F’ is true iff the object referred to by ‘a’ has the property referred to by ‘F’.

We saw in Chapter 5 that this principle is often taken to be expressive of representational approaches to truth, where the right-hand side of the biconditional is taken to explain the left-hand side. In this case, where truth is explained in terms of reference, there is an order of determination on the biconditional that makes the right-hand side prior to the left-hand side. This standard reading of (TP) not only shows how truth is dependent on reference, but also how truth is dependent on being, for the objects and properties mentioned on the right-hand side need to exist in order to be referred to, and to subsequently make the candidate sentence on the left-hand side true. Reading (TP) in this way naturally pairs a representational account of truth with an account of existence which makes existence distinct from what we think or believe things to be. On this view, existence is understood in terms of the Eleatic Principle, and to be is to have causal powers. This gives us a class of entities that exist prior to the formation of any thoughts or statements about them, and which are ready to serve as the referents of our singular terms and predicates. When we form words and make statements, we are latching onto this pre-existing class of things, and what we say is true just in case we ascribe the correct property to an object. Truth, on this picture, only comes into the story after the existence of the entities in question has been secured. An example to work with here would be ‘this lump of iron is metallic’. The thought would be that the truth of this sentence is dependent on the existence of mind-independent objects and properties (the lump of iron, and being metallic), and on the relevant terms ‘this lump of iron’ and ‘metallic’ referring to these entities. In examples like this of physical truths, we have a set of pre-existing objects and properties, and sentences are true insofar as they successfully represent the arrangement of these entities. If this were the only way to read (TP), then what we would have would be a uniform picture of existence, reference, and truth, with truth being dependent on reference, and reference being dependent on existence. However, if we take both ontological pluralism and truth pluralism seriously, then there is another legitimate reading of (TP) which reverses this explanatory chain. On this reading we take the left-hand side of the

back to truth

111

biconditional to be prior to the right-hand side, and take truth to be more fundamental than reference, and subsequently truth to be more basic than existence. This is in line with the thought expressed by Peirce about truth and reality that we encountered in Chapter 5, namely (paraphrasing somewhat) that what is real is what is described by our true sentences.8 To illustrate, consider our example of ‘the number 5 is prime’ discussed above. If we think of the truth of this sentence in non-representational terms—as superassertibility, for example—then the existence of the number 5, and the successful reference of the phrase ‘the number 5’ is secured by the truth of the sentence, as opposed to the truth of the sentence being secured by the existence of the object and property. That is, the superassertibility of ‘the number 5 is prime’ ensures the truth of the sentence, and, if we are thinking of arithmetical existence in accordance with the Neo-Fregean Principle, then the existence of the number 5 is secured. The differences between these two ways of reading (TP) demonstrate a way to combine truth pluralism and ontological pluralism to get a unified pluralist account of truth and existence. The basic idea is that there is one approach to truth which holds that what is true is dependent on what exists, and an associated form of existence which is not dependent on what is true. There is also another approach to truth according to which what is true is not dependent on what exists, and an associated form of existence which is not dependent on what is true. This matches the distinction between representational and non-representational approaches to truth that we developed in Chapter 5.

6.4 Back to Truth We can now see that what we said in Chapter 4 about predicates and properties can also be said about singular terms and objects. Some singular terms will be responsive, in that they respond to what we can call sparse objects; and others will be generative, in that they generate what we can call abundant objects. Sparseness and abundance of objects, whilst it will not be understood in terms of the grounding of objective similarities, can be understood in terms of the causal-explanatory relationships objects have as determined by their width of cosmological role. What emerges, then, is a picture where we have a distinction between two different ways of thinking about both the relationships between, firstly, predicates, properties, and truth, and, secondly, singular terms, objects, and truth. These can, of course, be seen as two competing pictures, with only one ultimately being capable of being correct, and this is one way to view classical realist/anti-realist debates. However, if we are sympathetic to a pluralist approach, we can show how these different pictures can coexist to explain the differences between the metaphysics of different regions of thought and talk.

8 See Ball (2017) for a version of this argument in relation to reference in a metasemantic context. The idea has an early formulation in Lynch (2009, ch. 7, section 3).

112 From Truth to Being Moreover, both of the explanations involving (TP) given above suggest a crucial role for truth in understanding the metaphysics of any kind of subject-matter. If we are interested in broadly realist or anti-realist approaches to an area, then these questions are usually about the existence of entities of a particular kind. If, as I have claimed above, the existence of entities is tied up with the nature of truth, then we cannot separate issues concerning the nature of truth from metaphysical issues in the realism/antirealism debate more generally. We can also connect truth to existence more explicitly by firstly noting that we have identified a number of prospective ways of being, including: The Eleatic Principle: To be is to have causal powers. Esse Est Percipi: To be is to be perceived. The Intuitionist Principle: To be is to be constructed. The Neo-Fregean Principle: sentence.

To be is to be the referent of a singular term in a true

We have also related these ways of being to the different directions of explanation on the principle (TP); in doing so relating these different ways of being to different ways of being true. There is an additional dimension available to the picture we have been constructing, which is that we now have the resources to explain the differences between kinds of objects and properties in terms of how they exist. Broadly speaking, the responsive model that is paired with the representational approach to truth will be open to the account of existence given by the Eleatic Principle. By contrast, the generative model, paired with a non-representational approach to truth, will yield entities that exist in the form dictated by the Intuitionist Principle and the Neo-Fregean Principle. This adds a useful further dimension to our understanding here: we now have an account of the existence of the entities, as well as how claims get to be true, and can see how the two issues are related. In the next two sections I will say a little bit more about some issues with the ways of being under discussion here, before moving on at the end of the chapter to compare the global pluralist package with global deflationism. More details on the precise formulation of truth pluralism and ontological pluralism are given in Chapters 7 and 8, respectively.

6.5 Abstract and Concrete? Let us call the type of existence yielded by the Eleatic Principle sparse existence, and the type of existence yielded by the Neo-Fregean principle abundant existence. Would it be helpful here to connect these ways of being with the familiar terminology of ‘concrete’ and ‘abstract’ existence, and thus to say that sparse objects and properties are concrete and abundant ones abstract? It is hard to say as the standard distinction made between concrete and abstract objects might be too blunt to be of any use here.

abstract and concrete?

113

The feature of having causal powers expressed in the Eleatic Principle is often taken to be a hallmark of concrete existence, and a vital tool in negatively defining abstract objects (as the things that do not have causal powers). Also a key factor is the definition of concrete objects as objects that have spatiotemporal properties, whereas abstract objects do not. Whilst these features might hold for some of the entities that fit the generative model, it is not clear that fits them all. McDaniel (2010a: 701–5) examines the idea that there are two different kinds of being, ‘temporally relativized existence’ and ‘atemporal existence’, with the idea being that some entities exist in space and time, and others do not. This is one potential way to distinguish between concrete and abstract entities, as it seems as though having causal powers requires a location in spacetime (see Thomasson  2015: 120), which would line up with the idea of concrete entities. Abstract objects like numbers, on the other hand, do not have spatiotemporal existence, they are atemporal, and exist outside of space and time. However, does this distinction give us what we need to distinguish between sparse and abundant existence? I don’t think so. For example, consider institutional entities. At first glance, these sorts of entities do not appear to fit into the standard distinction between abstract and concrete objects. We could consider them to be abstract, but the problem with this proposal is that they are things that come into being at particular times (a marriage and a goal in a football match have not existed eternally), and can be destroyed (a marriage can be ended by divorce). They can also be said to come into being at particular places (a marriage in Sheffield, for example, and a goal at Wembley). Moreover, institutional and social entities are tied to specific institutional and social contexts, which are often embedded in historical situations. For example, there does not exist a King of France now, as there is no institution of the French monarchy, but there did exist a King of France in 1415, as there was such an institution then. Similarly, whilst in 2017 there are relationships of inequality between men and women that may give us reason to posit the existence of men and women, were there to be a point in the future where such inequalities subsided, at that point men and women may cease to exist. If abstract objects are things that, by definition, do not have spatiotemporal properties, then social and institutional entities cannot be abstract entities. Are they concrete, though? This is also not straightforward, as one cannot physically interact with a marriage, or a goal, in the way that one can interact with, say, a table or a tree. They also, as we saw in Chapter 4, fail the width of cosmological role test for entities, in that, due to their nature as institutional entities, there is no explanatory work that they do that cannot be explained in terms of our believing that the things in question exist and have the properties they do. There are different ways to go here. One could try to shoehorn social and institutional entities into either category, but I think they show that the standard abstract/ concrete distinction is not helpful in some cases, and other accounts should be sought. For example, in her discussion of fictional entities, Thomasson (1999) discusses institutional entities as ‘cultural artifacts’, which share many characteristics in common

114

From Truth to Being

with regular artifacts, such as chairs, in that they are created by intentional agents at  particular times and in particular places, and are capable of destruction. I take Thomasson’s view to be instructive here, but it is worth separating these claims about cultural artifacts from the claim that fictional characters are also part of this category.

6.6 Quine’s Objection I now want to consider a famous objection of Quine’s to the idea of ontological pluralism, which questions whether we need different ways of being: If Pegasus existed he would indeed be in space and time, but only because the word “Pegasus” has spatio-temporal connotations, and not because “exists” has spatio-temporal connotations. If spatio-temporal reference is lacking when we affirm the existence of the cube root of 27, this is simply because a cube root is not a spatio-temporal kind of thing, and not because we are being ambiguous in our use of “exist”. (Quine 1948: 23)

The idea here is that whilst the differences we have noted so far between sparse and abundant entities might lead us to think that we have differences between kinds of entities, this does not mean that there are differences in how these entities exist. That is, what it is to exist can remain constant, with the idea being that there are different kinds of entity that exist. This can be applied to both our accounts of properties and objects by saying that, even if we grant that we have differences between sparse and abundant properties and objects, why think that they are two kinds of existence? Why not think that there is one kind of existence, and two kinds of things that exist? We can respond to this by noting the fact that existence is intimately involved in the way that we characterize the distinction between sparse and abundant properties and objects. Here is one way to display this: Abundant Properties: properties exist because they are instantiated. Sparse Properties: properties are instantiated because (in part) they exist. Abundant Objects: objects exist because they are referred to. Sparse Objects: objects are referred to because (in part) they exist. That is, we can undercut Quine’s concern by saying that it is part of the distinctions we make between different kinds of objects and properties that they exist in different ways: for some, existence is dependent on truth, whereas for others it is not. If different ways of being are required to understand different kinds of objects and properties, then we cannot separate the idea that there are different kinds of objects from the idea that there are different ways of being.9

9 Indeed, ironically, from a Quinean point of view, this should not be too surprising given Quine’s connection between quantification and existence noted above and below.

pluralism and deflationism

115

6.7 Pluralism and Deflationism In the final section of this chapter I will discuss the relationship between the pluralist account I have given here and a deflationary approach. In Chapter 3 I said that global deflationism was undesirable as it lacks explanatory power. I will now demonstrate one key instance of this, showing that its lack of explanatory power leads to an inconsistency in the view. I will focus on Amie Thomasson’s (2014, 2015) global deflationism. After noting the similarities and differences between my approach and Thomasson’s, I will argue that global deflationism cannot be sustained. Thomasson holds that existence claims are trivial, in that the question of whether a particular kind of thing, call them Ks, exist, is settled by the following principle: E: Ks exist iff the application conditions actually associated with ‘K’ are fulfilled. (Thomasson 2015: 86) That is, the existence of, say, tables, is settled by whether the ordinary application conditions for the predicate ‘is a table’ are fulfilled. Given that they are, it trivially follows that tables exist. There are apparent similarities between this strategy and one of the two alternative ways I have been thinking about the relationship between language and the world in the past few chapters. This is the abundant, generative, view, where the thought was that the existence of a property is determined by the extension of a predicate, and the existence of an object is determined by its corresponding term appearing in a true sentence. Thomasson also takes there to be key connections between truth, being, and reference. Indeed, as we saw in Chapter 3, and will explore further below, she holds that deflationism about one notion entails deflationism about all. As should be clear from what I have said above, I take it that pluralism about one notion entails pluralism about all (and likewise think that monism about one notion would entail monism about all). There are, however, a couple of important differences between Thomasson’s view and the view put forward here. The first is that, whilst I have suggested a similar line with the abundant, generative view, I have suggested that it applies only in certain domains, whereas Thomasson applies her view globally. That is, she doesn’t just think that her approach works in certain domains, she thinks that it works for all existence claims, across the board. As a consequence, she is unable to make the sorts of distinctions we have been making between different domains, and, indeed, she does not want to, as she thinks that there is no room for such distinctions. On Thomasson’s view, there is no scope to examine the different relationships between language and the world, for there are no interesting distinctions to be made. This is an aspect of the view that we have already discussed in relation to deflationism in Chapter 3. The second difference concerns the relationship between existence and truth. Thomasson holds that there are key connections between truth and existence, and that a deflationary theory of truth is all that is needed to serve as the foundation for existence

116 From Truth to Being claims.10 I also hold that there are key connections between existence and truth, but that these can only adequately be explained by a pluralist theory of truth. In the account I have given above, truth plays a key role in the foundation of some existence  claims, and part of that key role involves theories of truth that give specific accounts of what it is for a sentence to be true. That is, in the Neo-Fregean line discussed above, for example, truth is understood in terms of superassertibility, and this gives us an important explanation for why the initial sentence ‘the number 5 is prime’ is true, which then secures the existence of the number 5. Such an explanation is not available on the deflationary view. We have already raised problems for global deflationism, and how it cannot sustain a deflationary account of truth, and I won’t repeat them again here. Instead I will raise a different problem, which concerns the inability of a deflationary theory of truth to generate the existence claims Thomasson wants. The basic worry is that, if existence claims are inferred from truth claims, we need some grounds for the initial truth claims in order to get the existence claims, and the deflationist does not have the resources to give us these. This leaves the view too weak to ground existence claims, and suggests that there is an incompatibility between deflationism about truth and deflationism about existence. Consider the following argument Thomasson gives for how we can derive existence claims from truth claims: (1) is true (2) is true iff n is P (3) n is P (4) n is P iff ∃x ((n=x) & Px) (5) ∃x ((n=x) & Px) (6) ∃x ((n=x) & Px) → ∃x (n=x) (7) ∃x (n=x) (8) ∃x (n=x) iff n exists (9) n exists (10) ∃x ((n=x) & Px) → ∃x(Px) (11) ∃x (Px) (12) ∃x (Px) iff Ps exist (13) Ps exist (Thomasson (2014: 195–6) This argument aims to give us steps to infer an existence claim from a truth claim, and I have no complaint with the claim that it achieves this. However, there are problems with giving this argument from a deflationary point of view if what we ultimately want are actual existence claims. The problem is that deflationism lacks the resources to yield an actual existential conclusion, as it cannot discharge the assumption made in (1). 10

Compare Thomasson 2014 with Thomasson 2015, section 3.2.

pluralism and deflationism

117

As a consequence, we cannot get existential claims for deflationism out of this sort of argument. Let me explain why. Thomasson’s argument above shows us a way to infer an existence claim from a truth claim, which means that we start by treating (1) as an assumption with which to begin. This is all well and good, but what this means is that we have the conclusion that Ps exist only on the assumption that is true. In order to reach an actual existential conclusion—the claim that Ps exist—we would need to discharge this assumption, i.e. we would need independent grounds to think that it is the case that is true. The problem for the deflationist is that there are no resources to do that on their view. For example, the standard line for a deflationist to prove that is true would be as follows: (i) is true iff n is P (ii) n is P (iii) Therefore, is true. We cannot do that here, though, as the reverse direction is already used in the proof: (1)/(i) is used in conjunction with (2)/(iii) to prove (3)/(ii) by modus ponens, which means that (2) cannot be used to prove (1). This means that the deflationist cannot use an (E) instance—(2)—in the standard line to support (1). This means that they have no resources to discharge (1). As a result, whilst a deflationist can show that one can infer an existence claim from a truth claim, they cannot get an existence claim proper as they are unable to demonstrate how the assumption of the truth claim can be discharged. We can note that the pluralist is in a stronger position here, as they can discharge the assumption by saying that is true because is superassertible, which gives independent grounds for the truth of that allow the assumption to be discharged, and an actual existential conclusion to be reached. This demonstrates a significant advantage that pluralism has over deflationism. Immediately after this argument, Thomasson says the following. I will quote her at length as the issues here are important. Since we can always derive existence claims (singular and general) from truth claims, there is some substantive across-the-board criterion for existence only if the same is available for truth. But if (as the deflationist holds) there is no general substantive condition required for a proposition to be true, then we may derive different existence claims from the truth of diverse propositions—without any common condition holding, and so without any single substantive criterion for existence being fulfilled across all cases. So, since the deflationist about truth denies that there is any across-the-board criterion for a proposition to be true, she is also led to deny that there is any across-the-board criterion for existence. The truth deflationist should thus also embrace deflationism about existence. In neither case does this mean denying that it takes something of the world for a particular claim to be true, or for a thing (or things of a certain kind) to exist. What it does mean, however, is that we should expect no robust across-the-board criteria for truth or for existence in either

118 From Truth to Being case. Instead (for extensional first-order predicates), we can simply move from a true proposition to derive the relevant existence claim. So if we have a true proposition such as we can move from there to infer that tables (and brown things) exist; from we can infer that the number five (and an odd thing) exists. Given these entailments, the only way to doubt the existence claim is if one also doubts the truth of the original propositions ( or ). The original propositions certainly seem true, however. It seems that doubts about their truth are only raised (except by the mathematically or perceptually incompetent) for metaphysical reasons: doubts, for example, about what the truthmakers for the relevant claims could be or whether we should countenance the alleged truthmakers (given that they may not be causally relevant, etc.) But those worries are out of place if we accept a deflationary theory of truth, which rejects across-the-board demands such as the requirement that truths have truthmakers. On this deflationary view, we should thus give up the search for some ‘criterion of existence’ telling us what it is for something to exist, just as semantic deflationists give up the search for the nature of reference, meaning or truth. (Thomasson 2014: 196–7)

One thing that is interesting about this passage is that the way Thomasson uses ‘deflationism’ here is such that one could replace ‘deflationism’ with ‘pluralism’ and have a good account of some of the key characteristics of pluralism, with its rejection of a general substantive criterion for both truth and existence. These features attributed to deflationism, though, do not help us with the argument above, as, whilst there is no across-the-board criterion for truth, there is still no local criterion either! On the deflationist view the most we can get out of a local criterion for the truth of a proposition is the relevant instance of the equivalence schema. So, we may think, for example, that the left-hand side of is true iff snow is white is somehow explained in terms of the right-hand side, and this yields a local criterion for the truth of just as we could get a local criterion for by looking at the right-hand side of is true iff grass is green. The viability of this strategy will be discussed in Chapter 10 below, but, for now, we just need to note that it is of no help here, as we have already established that we cannot use the relevant instance of the equivalence schema to discharge (1). Thomasson’s gloss on the argument works if we have resources to establish local, rather than global, criteria for the truth of specific propositions, but these resources are not available on a deflationist view. They are, however, as stated above, available on a pluralist view. An alternative suggestion we can draw from Thomasson’s work is the idea that the initial claims are true by ordinary criteria. That is, that we notice that, in ordinary discourse, the sentence ‘there is a table in the room’ is true if there is an object in the room that meets the ordinary application conditions for ‘table’. This is a simple matter of noting the ordinary meaning of ‘table’ combined with a basic empirical observation  of whether there is anything in the room that meets the conditions for  ordinary uses of ‘table.11 Indeed, this is one sense in which, for Thomasson, 11

See Thomasson 2015: ch. 2.

pluralism and deflationism

119

ontology is ‘easy’, as we can infer existence claims from our ordinary practices. The issue for us here is whether this line involves a commitment to a specific way of thinking about truth. If it does, then we have grounds to think that we are not really dealing with a deflationary theory. Returning to her quote above, she says that the propositions and “seem true”, and that, metaphysical doubts aside, their truth would only be doubted by the “mathematically and perceptually incompetent”. This suggests that competence with the terms involved, along—in some cases—with the ability to perceive one’s environment properly is all that is needed to establish the truth of these claims. That is, in the case, if one is competent with the words ‘table’ and ‘brown’, and one has standard perceptual faculties, then one should be able to work out whether it is true that the table is brown. It is not unreasonable to suppose that, in this story, there is a connection between truth and what we are warranted in asserting. In particular, it is hard to see how, if we are dealing with ‘ordinary criteria’, truth could outstrip warrant in these cases. This is because ‘ordinary criteria’ will be standards that ordinary speakers are aware of, and, more importantly, will need to be aware when they have been met, otherwise they do not work as ordinary standards. The whole point of distinguishing between ordinary standards and the more loaded potential metaphysical standards is that it is easy to see when the former have been met, whereas it is questionable whether the latter have been met. Thomasson’s move is to use the former when establishing truth, and forget about the latter. So, the idea of ordinary standards involves a commitment to the idea that, if a sentence meets the ordinary standards, warrant is available for thinking that the sentence has met them. However, as Thomasson says, this does not mean that warrant alone is sufficient to establish truth. She says the following: [Application conditions] are not merely conditions under which we would have evidence that the term applies, but rather conditions under which the term would be correctly applied. Not much more needs to be said here. The point is only to make clear that ‘application conditions’ should not be understood as conditions under which it would be warranted or generally accepted to apply the relevant term, or in which we would have some evidence that it applies. Instead they are conditions under which it (really) would be proper to apply the term. Notice that this distinction applies wherever we have a rule: when we have evidence that someone has scored a touchdown, or committed a crime, is not the same as when it would be correct to say that someone scored a touchdown or committed a crime. (Thomasson 2015: 93–4)

If we have a distinction, then, between having mere warrant for applying a term and correct application of a term, we need to understand truth as more than mere warrant. An obvious amendment is to use the notion of superassertibility: it is not mere warrant that suffices for truth, but durable warrant that would survive through subsequent stages of inquiry.

120 From Truth to Being We can bolster this point by using Wright’s explanation of ordinary criteria in terms of “canonical grounds”: Say that a type of ground for a statement is canonical just in case the supposition that that statement is true entails the availability, at least in principle, of grounds of that type for believing it. Thus determining that all cats in Michigan weigh less than twenty pounds is supplying oneself with canonical grounds for the statement that all cats in the Great Lakes weigh less than twenty pounds. So canonical grounds need not be conclusive. But they can be. If it is true that 241 x 73 = 17, 593, a correctly conducted calculation will disclose as much. So correctly performing that calculation and getting the result, 17, 593, is acquiring canonical grounds for the statement that 241 x 73 = 17, 593. (Hale and Wright 2001: 158)

According to Wright, canonical grounds can also be conclusive grounds for the truth of a statement, as in his example above. This, I take it, is very much in the spirit of Thomasson’s suggestion about ordinary criteria above, namely that there are canonical grounds for the truth of and , the former involving competence with ‘table’, ‘brown’, and some perceptual abilities, and the latter involving competence with ‘five’ and ‘odd’, and that these grounds are conclusive. We can understand conclusiveness here in terms of superassertibility: the grounds are conclusive because they warrant assertion of the claim in question, and will persist in warranting the assertion of that claim as inquiry progresses. Indeed, this is precisely what Thomasson wants to claim about these propositions in the face of potential metaphysical doubts about their truth: no matter what the metaphysicians say, our canonical grounds for the truth of and , based on ordinary criteria, persist, and are hence conclusive. Examination of the idea that these claims are true by ordinary criteria thus leads us to the view that they are true in virtue of being superassertible. We are now in a position to see the full details of the relationship between truth and existence on Thomasson’s view. While Thomasson wants to have a deflationary account of truth, this cannot yield actual existence claims, as the deflationist has no way to ground the initial truth claims from which the existence claims are derived. The alternative approach open to Thomasson was to emphasize the idea that the initial truth claims are true by ordinary criteria. Examination of this idea led us to the conclusion that this is effectively the view that the initial truth claims are true in virtue of being superassertible. This means abandoning the deflationary theory of truth if the route from truth to existence is to be maintained.12 What this shows is that there is no space for a deflationary theory of truth and existence. This is because the connections between them that deflationists wish to draw are unable to yield the non-conditional existence claims deflationists want without appeal to a non-deflationary account of truth. 12 It should also be clear that truth cannot be understood in a representational way here, for a hallmark of representational truth is the view that what is true depends on what there is, and the whole idea behind Thomasson’s inference above is that we can infer what there is from what is true.

summary

121

6.8 Summary In this chapter I have argued that we can combine truth pluralism with ontological pluralism to give an expanded pluralist account of the relationships between language and the world. I have also contrasted it with global deflationism, and shown how global deflationism is unsustainable. The issues of how exactly we should formulate truth pluralism and ontological pluralism remain outstanding, and it is to those issues that we now turn, starting with truth.

7 Models of Truth Chapters 5 and 6 developed a general pluralist picture of the relationship between language and the world, with pluralist views of both truth and existence playing key roles. The discussion so far has been deliberately general on the specifics of truth pluralism and existence pluralism, and it is now time to focus on the details. There are a number of different ways to develop each view, and a number of different problems that each view faces. My aim in this chapter and in Chapter 8 is to give the strongest formulation of each view and defend it from objections. In doing so, I will inevitably have to critique different ways that each view has been formulated, but I should say up front that my main commitment is to the general pluralist picture that has been developed, as opposed to any specific formulation of truth pluralism or existence pluralism. As a consequence, my critiques of alternative formulations from my pluralist comrades should be viewed in a friendly way, in the interest of finding the best way to express the view. In this chapter I will focus on truth pluralism, and in Chapter 8 I will focus on existence pluralism. I will begin by outlining my ‘determination pluralism’ about truth, before comparing it to alternative formulations of truth pluralism.

7.1 Determination Pluralism Determination pluralism draws on the useful analogy between truth and winning.1 When playing a (competitive) game,2 the object of that game is to win, and this tells us something important about the practice of playing games. Likewise, when believing or asserting the object is to believe or speak truly, and this tells us something important about the practice of believing or asserting.3 It also, of course, tells us something important about truth, just as the observation about games tells us something about winning. In this section I want to explore this analogy to demonstrate one way that 1 The classic statement of this idea is due to Dummett (1978). The analogy is also discussed in Glanzberg (2004) and discussed in relation to truth pluralism in Edwards (2011) and Wright (2013). 2 Consider what follows to have competitive games in focus, rather than games in general. See Wittgenstein (1953) for a famous discussion of games. 3 This is not to say that the aiming need be conscious in each case. Also, for the purposes of this chapter, I will bracket the issue of where other forms of cognitive achievement, such as justification and knowledge, fit as norms of assertion and belief.

determination pluralism

123

we can arrive at an attractive formulation of pluralism about truth, which I call ‘determination pluralism’. Where winning is typically the goal of playing a game, truth is typically the goal of asserting or believing. In a particular game, the players will typically be trying to win that game, and in asserting or believing, an assertor or believer typically aims to hit the truth. Given the clear multiplicity of games, there is a strong sense of pluralism about winning: what it takes to win will change from one game to the next. It is also plausible to think that we have an understanding of winning that is not tied to any particular game, expressed by the thoughts that winning is the general aim when playing any game, and that winning is desirable. These are general features of winning that transcend any particular features regarding what it takes to win any particular game. We also often ascribe a general property of winning, such as when we say on Sports Day ‘these are the winners’ when gesturing towards the children holding sweets in their hands. There is a sense in which we would want to say that they share a property in common—being winners—which is distinct from the various ways in which they have become winners. There is a strong sense of unity with truth: truth is what all assertions and beliefs aim at, and is the property that all true sentences have. Truth is a distinctive—and unified4—norm of assertion or belief formation in the way that winning is a distinctive— and unified—norm of certain kinds of activity. Moreover, if one takes what has been said so far in this book seriously, then there is also a strong sense of plurality to truth: even if we take it that truth is a single property, there may be very different things to say about how different kinds of sentences get to be true. This thought is supported by the idea that what there is to say about the truth of institutional sentences may be very different from what there is to say about the truth of chemical sentences, which in turn is different from what there is to say about mathematical truths. With these thoughts in mind, suppose we were to give a theory of winning. To satisfy the twin constraints of unity and plurality, I suggest that we need to make sense of the idea that that there is a single property of winning, and that that there are a number of different ways to get to have this property. I contend that the best way to think of this structure is to hold that, for each game, there will be a winning-determining property, the possession of which by a player will determine possession of the general winning property. First of all, what can we say about the nature of the general winning property? We need not look far to answer this question as we can begin by describing winning as the property that one aims to achieve when playing a game. There may also be other features we can use to describe winning, such as that if one has the property of winning the game is over; the property of winning is a desirable property; if one has the property of winning one has been engaged in some form of competitive activity; and winning is a form of success. A full list of features like these should give us a 4

For a pluralist’s argument for this claim, see Lynch (2006).

124 Models of Truth complete specification of the property of winning: they are used as descriptions which characterize the nature of the winning property, and are intended to do so exhaustively. Consequently, there will be no reductive account of the property of winning that attempts to identify winning with any other property or properties. To establish what the winning-determining property is for a game, the natural place to look is at the rules of the game in question. For the kinds of games we are interested in, there must be some move that one makes, or some achievement in the context of the game, which determines that one has won that game. For a game to have this kind of structure, it is imperative that rules of the game are specified, which establish the permissible and impermissible moves, numbers of players permitted, and, of course, the specifications for winning that game. When we have examined these things, we will find that a conditional can be constructed of the form: (Cx) When playing game x: if one possesses property F then one has won (has the property of winning). Specific examples of conditionals of this form for chess and tennis, respectively, are plausibly: (Cc) When playing chess: if one has the property of having checkmated one’s opponent’s king then one has won. (Ct) When playing tennis: if one has the property of amassing a majority of the allotted sets then one has won. The rules of each game thus specify a property the possession of which determines possession of the property of winning. On this view, we treat the game-specific property as the property that determines the possession of the separate property of winning, and the nature of the game-specific property will be established by the rules of the game in question. The key distinction, then, is between winning and what it takes to win, or between the property of winning and the properties that determine winning. This account explains both the unity and the plurality involved in winning. We have a single property of winning, which is shared by all winners, and is the property that one aims to achieve when playing a game. We also have an explanation of how this property is attained, which fits nicely with the intuitive thought that it is the rules of the game in question which establish the property that determines winning. If one takes truth pluralism seriously, then truth, like winning, has claims to both unity and plurality. We can now consider how a pluralist theory of truth analogous to our account of winning looks, which I call ‘determination pluralism’ about truth. What is the property of being true on this account? In Chapter 1 we discussed some initial principles about truth that gave us an indication of what the constraints on a theory of truth are, namely that it must give an account of truth which exhibits these features. The determination pluralist takes this idea seriously and holds that these

determination pluralism

125

features describe the property of being true. The key move we need to make here is to rewrite them so that they make reference to a property. What we end up with is a list of claims like the following: Truth is the property that is the goal of inquiry.5 Truth is a property that is distinct from justification. Truth is a property that is distinct from warranted assertibility. To have the property of being true is to tell it like it is. To assert p is to present p as having the property of being true. ‘a is F’ has the property of being true iff the object referred to by ‘a’ has the property referred to by ‘F’. When asked to state what the property of being true is, we will be able to say things like ‘truth is the goal of inquiry’, ‘ “a is F” has the property of being true iff the object referred to by “a” has the property referred to by “F” ’, ‘truth is distinct from merely being warranted’, ‘to assert “p” is to present “p” as true’. These sorts of statements will be descriptions of the property of being true. These descriptions give a rich account of what truth is, and the relations truth has to other properties. There is no reductive account of truth available in that truth cannot be reduced to another property: we treat these descriptions as characteristics of a distinct and unique property. All true sentences, then, will possess this truth property. This property is truth, and this ensures that the unity constraint is met: there is one property that is truth, and this property is possessed by all true sentences. Now there is the significant explanatory project of explaining how sentences get to have this property. To explain this we are going to need to say some more about domains. Take it that a domain is like a game in that there is a goal for those participating in talk in that domain: to hit the truth. The idea is that truth is attained in virtue of the possession of a distinct property which, in accordance with the nature of the domain, determines truth in that domain. To establish which property determines truth in a domain we need to examine carefully the domain in question. What we will need to do is to examine the rules of a domain to generate conditionals analogous to (Cx), (Cc), and (Ct). This is essentially the task we carried out in Chapters 5 and 6. There I said that we need to determine whether the predicates in a domain are responsive or generative. Once we know that, we will know whether the properties in the domain are sparse or abundant, and hence what general approach to the way that they are true is appropriate, i.e. representational or non-representational. Once we know that, we will look at what particular account within that category is best suited to serve as the truth-determining property for that domain. For example, if the domain is 5 Sub-versions of this platitude would include aforementioned claims of the form approximating ‘truth is the property that is the goal of assertion’ and ‘truth is the property that is the goal of belief ’.

126 Models of Truth non-representational, then we need to look to superassertibility,6 and if the domain is representational, then we need to decide between different correspondence theories. Using this idea in conjunction with the work done in the book so far, we are able to get conditionals of the form: (Cdx) In domain x: if ‘p’ has property F then ‘p’ is true (has the property of being true). Two examples of these are: (Cc) In the chemical domain: if ‘p’ causally corresponds then ‘p’ is true. (Cs) In the social domain: if ‘p’ is superassertible then ‘p’ is true. On the supposition that that there is only one truth-determining property in a domain,7 these will form one direction of biconditionals of the form: (Bdx) In domain of discourse x: ‘p’ is true (has the property of being true) iff ‘p’ has property F.8 Using our examples, we can construct the following: (Bc) In the chemical domain: ‘p’ is true iff ‘p’ causally corresponds. (Bs) In the social domain: ‘p’ is true iff ‘p’ is superassertible. There is an order of determination on these biconditionals from right-to-left which reflects the explanatory primacy of the original conditionals. In the chemical domain, for example, it is because ‘p’ causally corresponds that ‘p’ is true, whereas it is not because ‘p’ is true that ‘p’ causally corresponds. The nature of each domain will thus specify a property the possession of which determines the possession of the separate truth property. Hence the name, ‘determination pluralism’.9 The structure of determination pluralism is thus as follows. Truth is given as the property that is exhaustively described by the truth features. This property is the property possessed by all true sentences, regardless of domain. For each domain there will be a property that determines possession of the truth property, and these properties are held fully distinct from the truth property itself. The relationship between the truth-determining properties and truth is underwritten by the conditionals of the form (Cdx) above, which in turn ground the order of determination on the biconditionals of the form (Bdx). Or, perhaps, coherence, though that option is not discussed in detail here. Without thus supposition there is a risk of contradiction, as, within a domain, one sentence may possess one truth-determining property and lack another. 8 Analogous biconditionals are more complicated in the games case due to circumstances like forfeiture. 9 As should be clear, the name ‘determination pluralism’ thus does not imply that truth is to be understood in terms of the determinate/determinable relation. Unfortunately there has been some confusion to this effect, but I think it is too late to change the name now. 6 7

other forms of truth pluralism

127

We will develop further features of determination pluralism below, but now I will briefly introduce other available forms of truth pluralism.

7.2 Other Forms of Truth Pluralism 7.2.1 Predicate pluralism There are three general models of truth pluralism that we can consider. The first, which we can call predicate pluralism holds that there is more than one truth predicate. In the way that we have been developing things, this view would hold that the predicate ‘is  superassertible’ and the predicate ‘causally corresponds’ would each be distinct truth predicates, each operating in specific domains. There is no general predicate ‘is true’ on this view.10 If we take this view, then any occurrence of the predicate ‘is true’ needs to be explained in terms of either of these other two predicates. For example, our principle (TP): (TP):

‘a is F’ is true iff the object referred to by ‘a’ has the property referred to by ‘F’,

would be explained by holding that one can substitute either ‘is superassertible’ or ‘causally corresponds’ in place of ‘is true’. Strictly speaking, this principle is really about superassertibility and causal correspondence, as are all other apparent claims about truth. As there is more than one truth predicate on this view, there is also more than one property of being true: in some domains truth is superassertibility, and in other domains truth is causal correspondence.

7.2.2 Strong pluralism Another form of pluralism is strong pluralism (see e.g. Wright 2003).11 Strong pluralism acknowledges that there is a single truth predicate, whose meaning is captured by a set of basic principles about truth,12 and that this predicate picks out different properties in different domains. On this view the predicate ‘is true’ has both a fixed and variable aspect: the fixed aspect is the meaning of the predicate, and the variable aspect is the property that the predicate picks out. One way to understand this view, due to Lynch (2006), is to think of ‘is true’ as a definite description, such as ‘the colour of the sky at noon’. This description has some fixed meaning, in that we understand it as meaning the same thing whether it is uttered on Tuesday or Wednesday, but it has a Wright (1992) can be interpreted as a predicate pluralist, but he later clarified his position as a form of strong pluralism, discussed below. See Kölbel (2013) for another example. 11 The term ‘strong pluralism’ comes from Pedersen and Wright (2013). It marks an opposition to ‘moderate pluralism’, which covers views that admit of a general truth property. It really should be ‘strong property pluralism’ to clarify that it is distinct from predicate pluralism, but I will just use ‘strong pluralism’ here. 12 Wright’s list is included in the list of constraints on a theory of truth given in Chapter 1. 10

128

Models of Truth

different referent on different days: on Tuesday it may refer to blue, whereas on Wednesday it may refer to grey.

7.2.3 Second-order functionalism and disjunctivism Lynch’s (2001, 2004) ‘second-order functionalist’ proposal and Pedersen and Wright’s (2013a) ‘disjunctivist’ proposal are similar views. Both begin with strong pluralism and then add an extra property into the mix. If we start with strong pluralism, we will have a set of sentences that has the property of being superassertible, and a set of sentences that has the property of causally corresponding. From these sets of sentences we can define a new property. On the disjunctivist proposal the idea is that we define the new property as the property of being either superassertible or causally corresponding, and on the second-order functionalist view we define the new property as the property of having some domain-specific truth property. Each view then identifies truth with this new property: on the disjunctivist view truth is the property of being either superassertible or causally corresponding; and on the second-order functionalist view truth is the property of having some domain-specific truth property. On both accounts, the general truth property is derived, in some way, from the domain-specific properties like superassertibility and correspondence, making these ‘derivative’ forms of pluralism.

7.2.4 Manifestation functionalism Lynch’s (2009) manifestation functionalism begins with a list of ‘truisms’ about truth: Objectivity: The belief that p is true if, and only if, with respect to the belief that p, things are as they are believed to be. Warrant Independence: Some beliefs can be true but not warranted and some can be warranted without being true. Norm of Belief: It is prima facie correct to believe that p if and only if the proposition that p is true. End of Inquiry: Other things being equal, true beliefs are a worthy goal of inquiry. (Lynch 2009: 8–12) According to Lynch, we ought to take these truisms to specify the nature of the truth property: the functionalist . . . can claim that there is a single property and concept of truth. The property being true (or the property of truth) is the property that has the truish features essentially or which plays the truth-role as such. (Lynch 2009: 74)13

13 The determination pluralist’s account of the property of being true is thus similar to the manifestation functionalist’s account, but shorn of the ideas of a property ‘playing the truth role as such’, or ‘having the truish features essentially’.

other forms of truth pluralism

129

Lynch does not want to disregard the ‘domain-specific’ properties, such as correspondence and superassertibility, with which the strong pluralist identifies truth. But Lynch does not want to identify truth with these properties; rather he wants to say that truth is manifested in these properties: this approach allows the functionalist to claim that truth is, as it were, immanent in ontologically distinct properties. Let us say that where property F is immanent in or manifested by property M, it is a priori that F’s essential features are a subset of M’s features. (Lynch 2009: 74)

Lynch’s view thus requires that the truth features—as expressed in the characterization of the truth property—are a subset of the features of the particular domainspecific properties. That is, each of the properties that, in Lynch’s terms, ‘manifest’ truth in a domain, must contain the truth property as a proper part. So, for example, for superassertibility to manifest truth in a domain, it would have to be part of the  features of superassertibility that the belief that p is superassertible if, and only if, things are as they are believed to be; that some beliefs can be superassertible but not warranted and some can be warranted without being superassertible; that it is prima facie correct to believe that p if and only if the proposition that p is superassertible; and that, other things being equal, superassertible beliefs are a worthy goal of inquiry. We can represent the different pluralist views as follows in Figure 7.1: As I mentioned at the beginning of this chapter, my main aim in this book is to argue for a general pluralist approach to metaphysics. As a result, I am happy with whichever pluralist view of truth is the strongest. However, I do think there are reasons to think that determination pluralism is the superior view, and will now explain why.

Is there a general truth predicate? YES

NO Predicate Pluralism

Is there a general property of being true? YES

NO Strong Pluralism

Is the general truth property derived from the domain-specific properties? YES Disjunctivism

NO Determination Pluralism

Second-Order Functionalism

Manifestation Functionalism

Figure 7.1 The different pluralist views.

130

Models of Truth

7.3 Determination Pluralism and Other Forms of Pluralism I will now discuss why determination pluralism is superior to these other forms of pluralism. I will begin by comparing determination pluralism with other views that hold that there is a general property of being true, before turning to predicate pluralism. Strong pluralism will be discussed in the next section.

7.3.1 Determination pluralism, second-order functionalism, and disjunctivism Both second-order functionalism and disjunctivism hold that the general truth property is derived from the specific truth properties. One problem with this view is that it is hard to see what reason we have to think that these derived properties have anything to do with the property of being true. The method of strong pluralism involves the idea that the features of truth are established by thought about what property truth would have to be. These features are then used to identify the property of being true, and it just turns out that there is more than one such property. The derivative form of moderate property pluralism accepts this general methodology, but bolts on an additional property at the end.14 A second, and perhaps more significant problem, is that the whole methodology supposes that we have a prior understanding of what truth qua property is, and this is what is used to identify truth properties. We cannot both hold this idea and then also  say that truth is derived from these properties previously identified as truth properties. A derived property just doesn’t look to be the right sort of property to be identified with truth given the general methodological picture. This leads us towards an additional piece of explanatory work that truth does on the determination pluralist model. The aim of an assertion in the physical domain, for example, is to causally correspond, and this to some extent may explain certain investigative techniques and methods employed by inquirers in that domain. But ultimately the reason why an assertion having the property of causally corresponding in that domain matters is because that property leads to the assertion having the property of being true. If causally corresponding were not the truth-determining property in that domain, then it would not have the significance it does. Truth thus provides an explanation of why certain properties in certain domains matter, and without truth being in the picture a significant part of the explanation is missing. In this way, truth does have a significant explanatory role to play in the account of inquiry in specific domains.

14 See Pedersen and Wright (2013b) for discussion of disjunctivism on this, and also Edwards (2012) and Lynch (2009).

determination pluralism and other forms of pluralism

131

7.3.2 Determination pluralism and manifestation functionalism A number of problems have been raised for Lynch’s manifestation relation (see, for example, David (2013) and Wright (2013)). I will just raise one (originally raised in Edwards (2011)). Lynch’s specification of the property of being true requires that the ‘truish features’ that compose it are read unrestrictedly; that is, it is a crucial part of Lynch’s view that the truth property itself is not annexed to any specific domain. But, if the domain-specific truth-manifesting properties exhibit the truth features, they cannot exhibit them unrestrictedly, because none of the properties manifest truth in all domains. Consequently, as they only manifest truth in specific domains, they can only be said to exhibit the truth features for those domains in which they manifest truth. For example, suppose that superassertibility manifests truth only in the moral domain, and consider the ‘Norm of Belief ’ platitude. It is not part of the features of superassertibility that it is prima facie correct to believe that p iff the proposition that p is superassertible. On the view we are supposing, this is only part of the features of superassertibility in the moral domain; in other domains, different properties will ground the norm of belief. What we should say instead is that it is part of the features of superassertibility that, in the moral domain, it is prima facie correct to believe that p  iff the proposition that p is superassertible. But, of course, this is to restrict this feature of superassertibility to a specific domain, which is to restrict the feature of truth in question to a specific domain. At the very most, then, the ‘truish features’ which superassertibility, in this case, can manifest must be restricted to a specific domain. Consequently, at most, the property of superassertibility can manifest a restricted truth property (perhaps moral truth, in this case). But this is not the result Lynch needs; he needs superassertibility—and all the domain-specific properties—to manifest the generic truth property. That is, he needs them to contain the features of the domain-free truth property as a proper part. The generic truth property, however, is composed of the unrestricted readings of the truth platitudes, and, thus, to manifest truth, a property must contain these features as a proper part. But this cannot be done: at most they can manifest a property composed of restricted readings of these platitudes which. The problem for Lynch’s view, then, is that the claim that truth is manifested in the domain-specific properties ends up in tension with the claim that truth is a property independent of any domain-specific annexing. Because determination pluralism does not hold that truth is part of the truth-determining properties, it does not have this problem.

7.3.3 Determination pluralism and predicate pluralism Predicate pluralism has problems with explanation. We have acknowledged that there are some general principles about truth, of which (TP) is one. The predicate pluralist aims to explain these by holding that they are simply convenient generalizations of the idea that this principle holds for both superassertibility and causal-correspondence.

132

Models of Truth

Rather than having to say ‘ “a is F” is superassertible/causally corresponds’ all the time, one can simply say ‘ “a is F” is true’. The problem is that we have to consider why it is that both the predicates ‘is superassertible’ and ‘causally corresponds’ have this feature. A natural way to explain the idea would be that there are general conditions associated with being a truth predicate, including being involved in (TP), and that any predicate that meets these conditions is a truth predicate. Given that ‘is superassertible’ and ‘causally corresponds’ meet these conditions, they are truth predicates. Predicate pluralism cannot avail itself of this explanation, though, as this presupposes that we have an understanding of the predicate ‘is true’ independently of the predicates ‘is superassertible’ and ‘causally corresponds’: in order to say that they meet the conditions for being a truth predicate, we have to have some prior understanding of what it is to be a truth predicate, which means that ‘is true’ cannot just be a generalization meaning ‘is either superassertible or causally corresponds’. Predicate pluralism thus looks to get things the wrong way around: we are not interested in truth because it is a notion constructed from these predicates, rather we are interested in these predicates because they are truth predicates.

7.4 Determination Pluralism, Strong Pluralism, and Mixing Problems The final remaining contender to determination pluralism is strong pluralism. I will now explore the reasons why determination pluralism has advantages over strong pluralism by considering the series of ‘mixing problems’ that have been the subject of much discussion in the truth pluralism literature. I begin with the series of mixing problems that were initially raised for strong pluralism, which holds that there are distinct truth properties in different domains, but no general property of being true shared by all truths. These are the problem of mixed inferences, the problem of generalizations, the problem of normativity, and the problem of mixed compounds. I show how determination pluralism offers better solutions to all of these problems than strong pluralism before discussing the truth of truth attributions.

7.4.1 Mixed inferences and generalizations The problem of mixed inferences (Tappolet (1997)) challenges a truth pluralist to make sense of the idea that validity is necessary preservation of truth. We often make inferences where premises and conclusion are from different domains of discourse, such as: Torture causes pain. What causes pain is wrong. Therefore, torture is wrong. The worry is that, if premises and conclusion are from different domains of discourse, then what truth property is preserved across the inference? This is a challenge for strong

mixing problems

133

pluralism as there is no general truth property. However, determination pluralism can make short work of it, because determination pluralism holds that there is a general property of being true that is shared by all truths, regardless of domain. As a consequence, there is only one property—truth—that is preserved across a valid inference.15 The problem of generalization is the problem of how to make sense of claims such as ‘Everything Socrates said is true’. Again, this is a problem for strong pluralism if there is no general property of being true, as it seems hard to make sense of what exactly I would be attributing to all the things that Socrates said if there is no general property of being true. This is not a problem for determination pluralism, as there is a general truth property which can be ascribed to all the things that Socrates said, in the example above.

7.4.2 Normativity Lynch (2006) objects to strong pluralism on the grounds that it does not provide a general norm of inquiry; rather, instead, what we have are distinct norms within different regions of inquiry: moral truth being the norm for moral inquiry, mathematical truth being the norm for mathematical inquiry, and so on. Determination pluralism does not have this problem, as it is a key part of the view that the general truth property is a general norm of inquiry. Indeed, this one thing that the analogy with winning aims to demonstrate. Just as with all games there is a single thing that one aims at—winning— there is a single thing that all beliefs and assertions aim at—truth. Of course, how one gets to truth will vary according to the subject-matter, but this should not affect the general norm of truth, just as the fact that one needs to do different things to win different games does not affect the idea that one is always aiming to do one thing—win.

7.4.3 Mixed compounds If mixed inferences and generalizations are easily dispatched, the problem of mixed compounds poses a stiffer challenge. However, in responding to it we will see some further motivations for holding that there is a general, domain-free, property of being true, which gives the determination pluralist additional advantages in the debate with the strong pluralist.16 Tappolet (2000) posed the problem of mixed conjunctions for the strong pluralist. The challenge is to explain how conjunctions of sentences from different domains are true. Her example is: (MC)

‘This cat is wet and it is funny’ (Tappolet 2000: 384).

15 See Beall (2000), Pedersen (2006), and Cotnoir (2013a) for responses to the mixed inference problem. See Strollo (2016) for a solution to the mixed inferences problem for strong pluralism which draws upon my (Edwards (2008)) solution to the problem of mixed conjunctions. 16 Note that I will just talk here about truth-functional compounds. See Chapter 9 for a short discussion of other examples.

134

Models of Truth

(MC) is capable of being true, but what truth property would it possess? The first conjunct, ‘this cat is wet’, reports a state of the world, and thus fits the notion of truth as correspondence to a fact, whereas the second conjunct, ‘this cat is funny’, Tappolet supposes, may be better suited to the notion of truth as grounded in social agreement. But, Tappolet asks, what property of truth would apply to the conjunction as a whole, which seems to neither be a case or correspondence nor a case of social agreement? In Edwards (2008) I suggested a solution for strong pluralism.17 I suggested that the relationship between conjuncts and conjunction can be expressed by the following biconditional, derived from the standard truth-table for conjunction: (C)

p is true and q is true iff p & q is true.

From (C) we can discern a key feature of conjunctions: the truth of conjunctions ought to be treated differently from the truth of the conjuncts, because there is an order of determination on the biconditional (C). That is, the conjunction, p & q, is true in virtue of the conjunct p being true and the conjunct q being true, but not vice versa. As the truth table for conjunctions suggests, conjunctions are true in virtue of each of their conjuncts being true.18 Using this idea we can solve the initial problem Tappolet set for pluralist theories of truth. Her worry was that the truth pluralist cannot explain the truth of a conjunction where each of the conjuncts were true in different ways (i.e. were from different domains of discourse). The answer is that p & q is true in virtue of p being true and q being true. We formulate the biconditional (C): (C)

p is true and q is true iff p & q is true.

We read (C) as having an order of determination from left to right which means that p & q is true because p is true and q is true, but not vice versa. This explains the way that p & q is true: the way that p & q is true is by being true in virtue of each of its conjuncts being true. The fact that the conjuncts are true in different ways does not affect our diagnosis of the truth of the conjunction: what matters at this stage is that the conjuncts are true, not the specific ways in which they are true. This solution can also be put in terms of determination pluralism: what determines the truth of a conjunction is that each of the conjuncts are true; and what determines the truth of a disjunction is that at least one disjunct is true. This, of course, means that the truth of a compound will be determined differently from the way that the truth of the atomic sentences that compose it are determined. However, this is exactly what we would expect, given that the analysis above treats compound sentences as 17 As noted in Edwards (2008), this is not just a problem for truth pluralism: correspondence theories face this issue as well. 18 Compare the logical atomists’ view that there are no conjunctive or disjunctive facts, with the truth of a conjunction or disjunction derived from the truth of the atomic sentences that comprise its conjuncts or disjunct(s). See Russell 1956 and Wittgenstein 1922. See also Tarski 1944 §11 for a distinction between the account of the truth of atomic sentences and the account of the truth of non-atomic compound sentences, which utilizes recursion procedures.

mixing problems

135

different to atomic sentences: they have the same property of being true, it is just that possession of this property is determined by different things. Whilst both strong pluralism and determination pluralism appear to be able to use  the same initial solution, determination pluralism has an advantage when we consider some deeper issues, particularly those concerning domain-membership for compounds. Remember that, for the strong pluralist, all truth properties are tied to domains, which will cause problems when, as I will argue, compounds do not belong to domains. Let’s look at these issues now. As the strategy above suggests, on my view conjunctions should be treated differently from atomic sentences. In particular, there is a difference when we consider them in relation to domains. As we saw above, domain membership is only relevant when we consider atomic sentences. This is because domains are composed of objects and properties, and atomic sentences are composed of singular terms and predicates that refer to those objects and properties. Accordingly there will be a direct connection between the content of a domain and an atomic sentence. Compound sentences, on the other hand, form an additional layer above this framework. They are composed of atomic sentences, and the characteristic of a particular compound that makes it a particular compound is not the kinds of terms that are contained in its atomic components, but the particular connective involved, determined by the conditions under which the compound is true. It is the truth conditions of a conjunction, for example, which mark it out from a disjunction. This is what makes it a conjunction, as opposed to another form of compound, and this has nothing to do with the kinds of atomic sentences that compose it. With this idea in mind, we do not face the same sorts of questions about domain-membership with compounds as we do with atomics. The essence of a compound sentence which makes it a compound sentence is not the kinds of atomic sentences that compose it, but the truth conditions it has.19 Compound sentences form an additional layer above atomic sentences. This is because compound sentences are essentially manipulations of atomic sentences into new structures: they take atomic sentences as ingredients, and with the additional of some specified rules, they turn them into new, compound sentences. As this process involves operations on atomic sentences, as opposed to a different kind of atomic sentence, compound sentences do not reside in a domain of similar status to other domains. As a consequence, if, as suggested earlier, we understand the metaphysical aspect of a domain as being the objects and properties to which singular terms and predicates respectively refer, then compound sentences do not have domains in this sense.

19 In previous work (Edwards (2009), in response to Cotnoir (2009)) I suggested that strong pluralists can say that compounds belong to the logical domain, in that the conditions under which each compound is true are determined by one’s preferred logical system. On this account, each compound, whether mixed or not, is a member of the logical domain. However, I will stick here with the account given in the main text, particularly as the Edwards (2009) approach was specifically designed to fit with strong pluralism, whereas I am more interested in determination pluralism here.

136

Models of Truth

This makes compounds an interesting class of sentences for a pluralist, as they are sentences that are capable of being true, but they do not have a domain. Determination pluralism is much better placed to handle this idea than strong pluralism. Strong pluralism’s truth properties are all tied to specific domains, which means that Tappolet’s challenge re-emerges here in a different way: how can compounds be true (have a truth property) if they are domain-free, and all truth properties are domain-specific? Determination pluralism, however, has an explanation ready: compounds are true in that they have the general truth property, but the way that they get that property is different from the ways that atomic sentences get to have it.

7.4.4 Truth attributions Mixed compounds give us examples of domain-free sentences, which the determination pluralist is able to handle better than the strong pluralist. There is another case worth examining in this area, which is the truth of truth attributions, such as ‘ “snow is white” is true’. To start with truth attributions in general, there are two broad options one can take.20 The first is to say that a truth attribution is true in the same way that the original sentence to which truth is attributed is true. On this account, the truth attribution inherits the domain of the original sentence, as it were, so the sentence ‘ “snow is white” is true’ would be true in the same way as the sentence ‘snow is white’ is true. The second option is to say that truth attributions are true in a different way from the way the original sentence is true.21 A problem with the first option concerns the reasons for assigning a sentence to a domain in the first place. Above I suggested that sentences are assigned to domains in terms of the predicate contained in the sentence. Thus, for example, ‘murder is wrong’ would be a moral sentence because it contains the moral predicate ‘is wrong’. In truth attributions, there is a difference between the truth attribution and the original sentence, because the relevant predicate in the truth attribution is the predicate ‘is true’, not the predicate in the original sentence. There is a difference between ‘snow is white’ and ‘ “snow is white” is true’, as the relevant predicate in the former is ‘is white’, and in 20 See Lynch (2013a) for more discussion of how a pluralist should approach truth attributions. Lynch and I take similar views on the general approach to truth attributions, but we differ on the details: Lynch utilizes his notion of ‘plain truth’ to explain the way in which they are true, whereas I take a different line. 21 Whilst the first option might have some advantages in terms of simplicity, it runs into the same sorts of problems that we saw in the problem of mixed conjunctions. The problem stems not from pluralism, but from general considerations about truth. For instance, take a correspondence theory of truth, and suppose that the sentence ‘snow is white’ is true because it corresponds to some worldly fact. If we are to say that the sentence ‘ “snow is white” is true’ is true in the same way, then this sentence would also have to correspond to some worldly fact. We could have a view that once a sentence is true, it is a fact that it is true, and this fact is added into the set of initial facts we had, but this runs counter to the idea that reality is composed of a set of facts which are independent from us. More seriously, it would suggest that there was a potentially infinite proliferation of facts, as we can iterate the predicate ‘is true’ as much as we want, and if each time we do so we generate a new fact, this multiplies the number of facts in existence rather dramatically.

mixing problems

137

the latter is ‘is true’. If this is the case, then the truth attribution cannot simply inherit the domain of the original sentence, as we need to consider ‘is true’ as the relevant predicate to determine the domain, if indeed there is one. This raises two issues for truth attributions: (a) in what way are truth attributions true; and (b) to what domain do truth attributions belong? Let us take each in turn. With regard to the way in which truth attributions are true, we can take inspiration from the analysis of mixed conjunctions, and say that a truth attribution is true in virtue of the truth of the original sentence. This is to say that a truth attribution’s truth is dependent on the truth of the original sentence, much as the truth of a conjunction is dependent on the truth of its conjuncts. The specific way that a truth attribution is true is that it is true in virtue of the original sentence being true. This suggests that truth attributions, like conjunctions, are not ‘base-level’ (or ‘atomic’, in our terms) sentences, as they are not about the objects and properties that make up domains, but rather about the sentences about those things. Likewise, they are not directly true in virtue of how the world itself is, but rather true in virtue of the truth of some other sentence whose truth is more fundamental. We may also wish to say that this way that truth attributions are true is determined by some basic principles about truth which dictate the circumstances under which truth attributions are true, much as the way that conjunctions are true is determined by some basic logical principles. A candidate for this would be the principle: (TPA) ‘ “a is F” is true’ is true iff the object referred to by ‘a is F’ has the property referred to by ‘is true’ which would be an instance of (TP) applied to truth. We will explore this in more detail below. If adopted, this principle could be used for any number of iterations of ‘is true’ on the left-hand side of the biconditional. When it comes to the issue of what domain truth attributions belong to, we are forced to think carefully about how to characterize the domain (if any) that truth itself belongs to. I think that the most plausible line to take here is that truth is a property that does not belong to any specific domain, and thus that truth attributions are domain-free sentences, much like the compound sentences we discussed above. The versatility of truth attributions suggests that truth is a property that can be applied to any sentence, regardless of domain. The only real requirement for bearing truth is that the bearer be some vehicle of descriptive content, such as a sentence, as opposed to some non-content bearer, such as a shirt button. This gives truth a great deal of versatility, much more so than other properties. We can thus think of truth as being at a general level above the level of all the specific domains we have considered so far. Whilst truth can be applied to sentences in many domains, truth itself is something not bound to these domains. Truth attributions, insofar as we take ‘is true’ to be the predicate that determines domain-membership, are thus domain-free in virtue of truth’s being domain-free. Truth is a valuable property which is possessed by sentences in different

138

Models of Truth

domains, and thus has wide-ranging versatility, without being tied to a specific domain. There are not many properties like this, though we will see below that existence also has this characteristic.22 The aim of this section has been to show that determination pluralism has advantages over strong pluralism, and our discussion of the problems above has revealed that a commitment to a domain-free truth property, in addition to the domain-specific truth-determining properties, gives determination pluralism a key explanatory edge. In light of what has been said above, determination pluralism offers the best framework in terms of explaining what needs to be explained about truth.

7.5 Truth, Sparseness, and Abundance In the final section of this chapter I want to discuss one further feature of determination pluralism. Discussion of truth attributions leads us to important questions about the property of being true. So far we have used truth as a tool to understand differences between different domains, particularly in terms of whether predicates in a domain respond to sparse properties or generate abundant properties. But what about truth itself? Does the predicate ‘is true’ respond to a sparse property of being true, or does it generate an abundant property of being true? Indeed, can we make sense of this question if truth is used to make sense of the distinction between sparse and abundant properties? Considering truth attributions allows us to get some traction on this issue. Let’s consider an example. Suppose ‘a is F’ is a true social claim. We take our instance of (TP) for truth attributions: (TPA) ‘ “a is F” is true’ is true iff the object referred to by ‘a is F’ has the property referred to by ‘is true’. And our biconditional for the truth of social claims: ‘a is F’ is true iff ‘a is F’ is superassertible. More precisely here:23 ‘a is F’ has the property referred to by ‘is true’ iff ‘a is F’ is superassertible. If we suppose that ‘a is F’ is superassertible, then this gives us the right-hand side of the biconditional (TPA) above, which allows us to infer the left-hand side: 22 Discussion of truth attributions naturally leads us towards the notorious liar paradox. The liar paradox is a massive topic in the study of the logic of truth, and, whilst there are important connections between the logic of truth and the metaphysics of truth, I cannot here discuss the paradox and its possible solutions in the detail they deserve. Extensive accounts of the paradox can be found in Beall (2009), Field (2008), Scharp (2013), and Cook (2013), to give just four examples. For discussions specifically on the liar paradox in relation to truth pluralism, see Beall (2013), Cotnoir (2013b), and C.D. Wright (2017) who offers a determination pluralism-specific solution to the liar paradox. 23 This is to clarify that ‘is true’ is always referring to the same property.

truth, sparseness, and abundance

139

‘ “a is F” is true’ is true. This account of truth attributions displays a peculiarity about truth with regard to the  sparse/abundant distinction. In the account of truth attributions given, the biconditional (TPA) is read with the representational direction, in that we infer the left-hand side from the right-hand side, not vice versa: (R) ‘ “a is F” is true’ is true because the object referred to by ‘a is F’ has the property referred to by ‘is true’. Not: (G) The object referred to by ‘a is F’ has the property referred to by ‘is true’ because ‘ “a is F” is true’ is true24 (R) is typically associated with sparse properties, and (G) with abundant properties. This would suggest that truth is a sparse property. However, (R) states that the reason that things are a certain way (the truth attribution’s being true) is because a sentence is true, and an object’s having a property being dependent on the truth of a sentence is typically a hallmark of the property in question being an abundant property! What this suggests is that the abundant/sparse distinction gives out when we consider truth itself. There are additional reasons to think that this is the case. We can start by noting a feature of truth attributions, which is that truth is a property that can be possessed by sentences in all domains. This required generality of truth means that it is difficult to see how it could be either an abundant or a sparse property. Here is why. If truth itself were an abundant property, then the predicate ‘is true’ would be a generative predicate, and all sentences involving it would be true in virtue of superassertibility. This would effectively render all truths to be true in virtue of superassertibility, and enforce abundant monism about truth. Conversely, if truth itself were a sparse property, then ‘is true’ would be a responsive predicate, and all sentences involving it would be  true in virtue of correspondence. This would effectively render all truths to be true in virtue of correspondence, and enforce sparse monism about truth. This means that any moderate pluralist view cannot hold that the general property of being true is either sparse or abundant, on pain of collapsing into a form of monism. This should not be a surprise, as, given the account above, to ask whether truth is abundant or sparse is effectively to ask whether truths are true in virtue of superassertibility or correspondence, which is not a question a moderate pluralist can answer in a straightforward way. We can also note that strong pluralism is in the same boat. Even though it denies that there is a single property of being true, it still holds that there is a single truth predicate. If this predicate is generative, then all sentences involving it are true in virtue of being superassertible, and, if it is responsive, then all sentences involving it are true in virtue of corresponding. As a consequence, the 24

Compare the order of determination on conjunctions noted above.

140

Models of Truth

general truth predicate cannot be considered either a generative or responsive predicate on strong pluralism. In fact, the only pluralist view that is not immediately affected is predicate pluralism, as it denies that there is either a general truth predicate or a general property of being true, so no decisions have to be taken across the board. This should be expected given the way that we have constructed the general metaphysical picture. So far we have used the notions of abundance and sparseness to characterize properties within certain domains. Part of this process has involved truth, as we have seen. In order to play the roles it needs to, truth itself cannot be tied to any specific domain; instead it must be a domain-free property. Putting these pieces together we can make the case for the view that the abundant/sparse distinction does not apply to truth itself. Notice that when we have considered the sparse/abundant distinction so far, we have applied it to whole domains, as opposed to individual properties. When we have considered the abundance or sparseness of a property it has been against the general backdrop of the kind of property it is taken to be, and the kind of domain we are considering. Indeed, this is a condition built into the idea of the width of cosmological role criterion that we have been using: Let the width of cosmological role of a subject-matter of a discourse be measured to the extent to which citing the kinds of states of affairs with which it deals is potentially contributive to the explanation of things other than, or other than via, our being in attitudinal states which take such states of affairs as object. (Wright 1992: 196, underline added this time)

This means that being asked to assess the sparseness or abundance of truth is a different sort of question to the questions we have considered so far about the sparseness or abundance of other properties. Truth is not a domain-restricted property, and it is also a key instrument used to make the distinction between sparseness and abundance. This is another reason why truth itself cannot be considered a sparse or abundant property. Finally, I want to comment on the significance of the issue of whether truth is an abundant or sparse property. The initial relevance of this distinction when we introduced it in Chapter 2 was to specify a difference between deflationary and inflationary approaches to truth. As we saw in Chapter 3, though, due in part to the fact that deflationism cannot make sense of the distinction, being a deflationist is not to take a view about the property of being true, but rather to subscribe to a general worldview about the connection (or lack of it) between language and the world. In Chapter 4 the distinction between sparse and abundant properties then took on a different role in the discussion, being central in the distinction between metaphysical approaches to different domains. The pluralist, anti-deflationist, approach that I have developed thus makes use of the sparse/abundant distinction in a way that a deflationist cannot. It also, as we saw in Chapter 6, has significantly more explanatory power than a deflationist view. In the context of these developments, the explanatory power of a theory of truth

summary

141

does not rest on the claim that truth itself is a sparse property, and it is not a requirement of an anti-deflationist view about truth that truth itself be a sparse property.

7.6 Summary In this chapter I have aimed to give a sense of the different ways that truth pluralism can be formulated, and suggested one way that I think is preferable. We will now turn to discuss how to formulate ontological pluralism, and the models developed here will aid us in that discussion.

8 Models of Being In this chapter we will explore how ontological pluralism should be formulated. We will see that similar issues arise to those that we saw in the formulation of truth pluralism, and that the models from the truth pluralism debate are applicable here as well. I will begin by discussing in more detail the role of quantifiers in the existence debate, before turning to ontological pluralism itself. I will look at strong ontological pluralism, evaluating the extent to which it suffers from analogues of the mixing problems we saw Chapter 7, before looking at other forms of ontological pluralism. After examining McDaniel’s (2009, 2010a) formulation, I will develop a form of determination pluralism for existence, before examining whether existence is abundant or sparse.

8.1 Existence: All About Quantifiers? Before we try to formulate ontological pluralism, we should say more about the word ‘exists’. The general view is that ‘exists’ should not be understood as a predicate, but rather as a quantifier. It is important to clarify what this means for us here, as it will affect the way we think about ontological pluralism. What we can do is identify a kind of—for want of a better term—‘quantificationism’ about existence. Quantificationism about existence holds that there is little to say about the nature of existence because debates about existence are just debates about the behaviour of the existential quantifier. This comes from the idea that Quine’s famous claim about existence—“To be is, purely and simply, to be the value of a variable” (Quine 1948: 32)—tells us all we need to know. As influential as an account of existential commitment as this has been, it is important to note that does not offer an analysis of existence in the traditional sense, as Quine himself acknowledged. What is at issue here is not existence, but rather what it is for a theory to be committed to the existence of an entity. As Quine says: We look to bound variables in connection with ontology not in order to know what there is, but in order to know what a given remark or doctrine, ours or someone else’s, says there is; and this much is quite properly a problem involving language. But what there is is another question. (Quine 1948: 35)

Evidently, Quine notices a separation between what a theory says exists, and what the extension of ‘exists’ is, but we can also extend the point to a separation between what a

existence: all about quantifiers?

143

theory says exists, and what existence itself is. At least on a non-Quinean understanding of existence, what it is to exist is not simply a problem involving language, and not simply reducible to the question of what it is to be committed to the existence of an entity. However, despite this, Quine’s work had radical consequences for work on existence. There was a substantial shift in interest towards issues of quantification and ontological commitment, and work on existence was just considered to be work on the relevant quantifiers and their behaviour. As a consequence, relatively little attention has been given since to questions such as ‘What is the nature of existence?’ that we discussed in Chapter 6.1 The prevalence of the existential quantifier in contemporary discussions of existence reflects this trend towards quantificationism. As Kit Fine puts it: [T]he commonly accepted view . . . is that ontological questions are quantificational questions. (Fine 2009: 158)

Moreover, even contemporary ontological pluralists tend to define their position in terms of the view that there are multiple quantifiers, such as, for example, Turner: To put ontological pluralism in a nutshell: the true fundamental theory uses multiple existential quantifiers. (Turner 2010: 9)

According to this view, existence is understood in terms of the existential quantifier in a fundamental language, and if there are multiple existential quantifiers in the fundamental language, then existence is plural. There are similarities here with debates about deflationism about truth, and this is perhaps no surprise given Quine’s work in that field too. As we saw earlier, the deflationary trend in the study of truth aimed to shift focus away from the nature of truth towards the behaviour and uses of the truth predicate, with the thought being that, once those are fully understood, there really is no further nature of truth to uncover. Returning to existence, we can note that the shift from thinking about the nature of existence to thinking about the behaviour of the quantifiers is similarly satisfying to those who favoured the general deflationary move. We see that there is a parallel method of dissolving mysterious metaphysical questions, and thinking that the real issues of interest in the area are those involving language and logic. The route to deflation in the existence case is that, once the quantificational issues are resolved, there is nothing more to say about the nature of existence. In the case of existence, anti-quantificationist voices are harder to come by than anti-deflationary voices in the study of truth, but we can find some thoughts in the 1 Thomasson (2015) disputes the claim that Quine is a deflationist about existence, offering instead her own version of existence-deflationism, which is intended to be part of the global deflationist picture (detailed in Thomasson 2015) that we discussed in Chapters 3 and 6. However, her reasons for rejecting Quine’s view as a deflationary view were that Quine takes a pragmatic line, and not a deflationary one. However, if the argument against deflationism given in Chapter 6 is correct, there is not a great deal of difference between these two categories.

144

Models of Being

general direction. Kit Fine, for example, having distinguished what he calls ‘ontological’ questions from quantification questions, argues that the former are not reducible to the latter: [M]y broader point is that these excursions into the semantics of quantification, whatever their independent interest, are largely irrelevant to the understanding of ontology . . . The critical and distinctive aspect of ontological claims lies not in the use of the quantifier, but in the appeal to a certain concept of what is real;2 and it is only by focusing on this concept, rather than on our understanding of quantification, that further clarification is to be achieved . . . (Fine 2009: 171)

Nathan Salmon also gives expression to anti-quantificationist thoughts in response to the Quinean movement: Quine gave substance to the idea that what exists is what is covered by the universal quantifier with his equally famous slogan ‘To be is to be a value of a variable’. Taken as a response to the question ‘What is existence?’, Quine’s slogan seems at least extensionally correct. Every existing individual is indeed the value of some variable or other, under some cooperative assignment of values to variables, and it would seem that everything that is assigned to a variable as its value is “one of everything”, i.e., it exists. But it cannot be seriously maintained that being, in the sense of ‘existence’, simply is the state or condition of being the value of a variable, under some assignment of values to variables. When Hamlet (pretending the play were non-fictitious) agonized over the question of whether to be or not to be, he was preoccupied with weightier matters than the question of whether or not to be the value of a variable. (Salmon 1987: 51)

Salmon suggests that there is some pre-existing interest in the nature of existence that cannot be satisfied by talk of quantifiers and variables.3 Whilst (as we will see later on) we may think that quantifiers and variables are part of the story, they cannot be the whole story, as they leave the weighty question of what it is to exist completely untouched. Indeed, this should come as no surprise; as we saw above, Quine himself noticed this. However, the issue is whether taking the Quinean view on board removes the need to ask the question of what the nature of existence is. As we saw in Chapter 6, the type of ontological pluralism under discussion here depends on taking the metaphysical project of giving an account of the nature of existence seriously. This suggests some dissatisfaction with the quantificationist trend towards taking questions of existence to be questions of quantification. However, as we saw above, many contemporary discussions of ontological pluralism do take questions of quantification to be important, and as a result formulate ontological pluralism as the thesis that there are multiple existential quantifiers.4 What should we make of this? Fine uses the word ‘real’ for a ‘thick’ sense of ‘existence’. Salmon himself goes on to define existence in terms of identity: to exist is to be identical to something. We will discuss this idea further below. 4 See e.g. Turner (2010) and (2012). McDaniel (2010a) and (2010b) states that it is optional for ontological pluralists to look deeper than the multiple quantifiers, but, if what is said here is correct, this is not optional: they must look deeper. An exception is the Heideggerian approach explored in McDaniel (2009), where Heidegger’s claim that being is not itself a being precludes taking being to be anything deeper. 2 3

ontological pluralism and ontological debates

145

The answer is that we shouldn’t make much of it, provided the relevant parties are willing to take certain things on board. There is no harm in thinking that ontological pluralists are committed to there being multiple existential quantifiers, and that there are things that need explaining about the quantifiers’ behaviour as a result. The key thing to take on board is that this will not be the end of the story, and that the fact that there are different quantifiers does not get to the heart of ontological pluralism. The heart of the view will be that different things constitute existence in different cases, and this is what is required to meet the anti-deflationary motivations of the view. That this gives rise to multiple quantifiers may be a consequence of the main thesis, but it is not itself the main thesis. Ontological pluralism is about more than quantifiers, as existence is about more than quantifiers. Or, at least, that is what a form of ontological pluralism that wishes to use the motivations above should contend. For the kind of ontological pluralism wanted here, then, we need to move beyond just talking about quantifiers. Precisely where we go beyond quantifiers is a matter for dispute. Whilst it is controversial whether existence is a property, below I will provide one route for thinking about things in that way. Prior to this, I will speak in more neutral terms of ‘ways of being’.

8.2 Ontological Pluralism and Ontological Debates We can bring this point out further by showing how ontological pluralism vindicates debates in metaphysics by contrasting it with the view of quantifier variance (Hirsch 1993, 2009). Quantifier variantists hold that there are multiple quantifiers in operation in the same domain, which explains why people talk past one another certain metaphysical debates due to different sides using different quantifiers. A central example is the debate over the existence of composite material objects, where nihilists (e.g. Merricks 2001, van Inwagen 1990) deny that the claim ‘there are tables’ are true, whereas their opponents (e.g. Lewis 1986a) hold that sentences like ‘there are tables’ are true. The quantifier variantist claims that such debates are somewhat fruitless, as the two sides mean different things by ‘there are’. In other words, they use different quantifiers, and, as such, talk past one another when one side claims that there are tables, and the other side denies this. Quantifier variance can thus be interpreted as a form of relativism, which holds that many debates about existence are unanswerable outside of a particular interpretation of ‘exists’.5 Ontological pluralism is not quantifier variance. The first reason for this is that quantifier variance is a form of quantificationism: it holds that discussion of existence just takes place at the level of quantifiers, and the reason why people talk past one another is that they use different quantifiers, which range over different entities. On 5 Though this is not exactly how Hirsch himself sees it, as he thinks that debates are settled by ordinary language. See, e.g. Hirsch (2009).

146 Models of Being this view, the difference between uses of ‘exist’ is just the range of the quantifier, not anything substantive concerning different views of what it takes to exist. Ontological pluralism, as we have seen, holds that there are deeper explanatory issues concerning existence, which underwrite the use of different quantifiers: different quantifiers correspond to different ways of being. This represents a difference in metaphysical depth between the views. Moreover, whilst the quantifier variantist claims that there are multiple quantifiers within a domain, ontological pluralists claim that there is only one quantifier in each domain. As a consequence, there will not be scope for talking past one another in the way the quantifier variantist claims, as there is only one way of being that is attributed to things in each domain. Ontological disputes will thus concern whether the things in question have the existence-determining property in that domain. For example, in the case of composite material objects, if we take the relevant way of being to be having causal powers, then the debate is about whether composite material objects have causal powers, which is indeed the background presupposed in some arguments for the nonexistence of material objects.6 If we are thinking about a different domain—the social domain, say—then existence disputes will be about whether a relevant singular term appears in a true sentence, which is to say, whether any sentences involving the term are superassertible.7 Ontological pluralism thus offers scope for significant metaphysical debate on at least two levels. The first is the question for a domain of what the relevant way of being is, which is a significant metaphysical issue. The second is that, after a particular way of being has been established in a domain, what exactly it is that has that feature. There is room for significant debate about this too.

8.3 Formulating Ontological Pluralism We will now turn to see how ontological pluralism can be formulated. If existence, for ontological pluralists motivated by the considerations given above, must be a more substantial matter than what we find at the level of quantifiers, then we have similar choices to make as we had with regard to the models of truth pluralism. In particular, we have quantifier pluralism, which would hold that there are multiple quantifiers and multiple ways of being; strong pluralism, which would hold that there are multiple ways of being, but no general way of being had by all existing things. We would also have a derivative form of pluralism, which would hold that there is a general way of being derived from the multiple ways of being, and we would have explanatory form

6 E.g. Merricks’s (2001) ‘overdetermination’ argument, which takes as a premise that composite material objects, if they existed, would have causal powers. Merricks argues that they do not have causal powers, hence they don’t exist. 7 Note that, if superassertibility is to meet the standards required of truth, then there must be only one answer to the question of whether a sentence is superassertible.

formulating ontological pluralism

147

of pluralism, which would hold that there is a single way of being which plays a role in the identification of the multiple domain-specific ways of being. As we will see, there is some discussion in the debates about ontological pluralism about whether a general notion of existence is required, or even unavoidable. Interestingly, these debates often concern analogues of the mixing problems for truth pluralism that we discussed in Chapter 7. We will also be led to think about how the problems for specific formulations of truth pluralism may cross over to the existence debate. Let’s discuss them now and see what we can find out. We start with a formulation of ontological pluralism that holds that there are multiple meanings of ‘exist’ (multiple quantifiers if we are so inclined), corresponding to multiple ways of being.8 We will examine some additional models below, but let’s  for now examine some problems commonly raised for this ‘strong’ form of ontological pluralism.

8.3.1 Mixed inferences Turner (2010) discusses a problem which appears to be parallel to the mixed inferences problem for truth pluralism for ontological pluralism. He considers the following inference: There are chairs. There are numbers. Therefore, there are chairs and numbers If the meaning of ‘there are’ is not held constant—as Turner supposes it is not according to strong ontological pluralism—then there is a danger of this argument coming out as invalid due to equivocation on ‘there are’. Even though Turner attempts to provide a solution to this problem, we can show that there really is not a problem here at all. There would be an analogue problem of mixed inferences if the argument had the following form:     (There exists1 something that is a chair.) ∃1 x (Chair(x)) ∃2 x (Number(x))     (There exists2 something that is a number.) ∴ ∃? x (Chair(x) ∧ Number(x)) (There exists? something that is a chair and a number.) 8

An example of the multiple meanings idea is expressed by the famous remarks of Gilbert Ryle:

It is perfectly proper to say, in one logical tone of voice, that there exist minds and to say, in another logical tone of voice that there exist bodies. But these expressions do not indicate two different species of existence . . . They indicate two different senses of ‘exist’, somewhat as ‘rising’ has different senses in ‘the tide is rising’, ‘hopes are rising’, and ‘the average age of death is rising’. A man would be thought to be making a poor joke who said that three things are now rising, namely the tide, hopes and the average age of death. It would be just as good or bad a joke to say that there exist prime numbers and Wednesdays and public opinions and navies; or that there exist both minds and bodies. (Ryle 1949: 24) Note that the analogy is not exact, as Ryle denies that there is a connection between there being multiple meanings of ‘exist’ and multiple ways of being, which is not something we would subscribe to given the motivations for ontological pluralism given in Chapter 6.

148

Models of Being

We would have a problem as it would not be clear which quantifier applied to the conclusion, and, moreover that there would be a danger of equivocation. But, for this to be the correct logical form of the argument, we would need to interpret the conclusion as saying that there is one thing that is both a chair and a number. But this is not what the original argument says. The correct form of the argument is rather:       (There exists1 something that is a chair.) ∃1 x (Chair(x)) ∃2 x (Number(x))       (There exists2 something that is a number.) ∴ ∃1 x (Chair(x)) ∧ ∃2 x (Number(x)) (There exists1 something that is a chair and there exists2 something that is a number.) This argument is perfectly fine, and the conclusion is just a conjunction of two different quantified statements. The fact that each employs a different quantifier ought not to be a problem: indeed, it follows directly from the premises. We can thus identify one disanalogy between truth pluralism and ontological pluralism: whilst truth pluralists need to provide an answer to a mixed inference problem, there appears to be no equivalent problem for ontological pluralists.9

8.3.2 Mixed conjunctions Turner (2010, §3) also poses a problem analogous to the problem of mixed conjunctions. Indeed, he phrases the argument we considered above as a problem of conjunction. However, for the same reasons we noted above, it looks like it should not be a problem. When we get compounds which include existence claims pertaining to two different kinds of object, we do not get the same problems we get with truth. This is because mixed statements, such as: There are chairs and numbers can be broken down into conjunctions of two different existence statements: ∃1 x (Chair(x)) ∧ ∃2 x (Number(x)). However, there is no problem here. There is a disanalogy with the truth case in that there we needed an account of what the truth of the conjunction as a whole consisted in, whereas in the existence case, we do not need to account for the existence of the conjunction as a whole. Everything that needs to be said can be said using the two different existential quantifiers, and the only way to generate an equivalent problem would be to find a single object which seemed to fall under the range of both of the two different quantifiers. In that case, we would have a genuine problem where it would not be clear in what sense the object in question existed. However, this is not provided by the problem as stated, and indeed seems to be a very difficult case to plausibly generate.

9 There may be such a problem if plural quantifiers are introduced satisfying certain constraints, however. But the matter is too involved to discuss here.

formulating ontological pluralism

149

8.3.3 Van Inwagen and number Peter van Inwagen (1998) offers the following influential argument against ontological pluralism as follows: No one would be inclined to suppose that number-words like ‘six’ or ‘forty-three’ mean different things when they are used to count different sorts of object. The very essence of the applicability of arithmetic is that numbers may count anything: if you have written thirteen epics and I own thirteen cats, then the number of your epics is the number of my cats. But existence is closely tied to number. To say that unicorns do not exist is to say something very much like saying that the number of unicorns is 0; to say that horses exist is to say that the number of horses is 1 or more. And to say that angels or ideas or prime numbers exist is to say that the number of angels, or of ideas, or of prime numbers, is greater than 0. The univocacy of number and the intimate connection between number and existence should convince us that there is at least very good reason to think that existence is univocal. (Van Inwagen 1998: 236)

The idea is that, if there is a connection in meaning between ‘0’ and ‘does not exist’, and ‘1’ and ‘exists’, then multiple meanings of one should imply multiple meanings of the other, and uniformity of meaning of one should imply uniformity of meaning of the other. Van Inwagen claims that number terms are univocal in meaning, so ‘exists’ must be univocal too. Turner (2010) responds to van Inwagen’s argument on behalf of the ontological pluralist by distinguishing between number terms, like ‘0’ and ‘1’, and the term ‘the number of ’. Turner argues that van Inwagen is mistaken to think that the ontological pluralist is committed to there being different meanings of number terms, rather they are just committed to different meanings of the term ‘the number of ’. If the term ‘the number of ’ refers to a numbering relation, then the ontological pluralist’s claim is not that there are multiple meanings of number terms, rather it is the claim that there are multiple numbering relations that different meanings of the term ‘the number of ’ refer to. There is thus an option for a strong ontological pluralist to resist van Inwagen’s argument here, but, as we will now see, the idea that there is some generality to existence is hard to avoid.

8.3.4 Mixed disjunctions Turner (2010, §2) also poses the ‘disjunctive quantifier argument’. Suppose that ontological pluralism is committed to the idea that there are different quantifiers applying to sparse and abundant objects. Surely, one might think, even with this idea in play, we can say things like the following: (DE)

If something exists, it either exists sparsely or abundantly.

The basic thought here is that we can say that, if existing sparsely or abundantly exhaust the ways there are to be, then, for any existing thing, it will either exist sparsely or abundantly. The worry should be clear: what is the meaning of the first instance of ‘exists’? It seems to be a more general term than either existing sparsely or abundantly,

150 Models of Being and it is applied to all the objects that those two terms range over. If this is the case, why exactly do we need the terms ‘existing sparsely’ and ‘existing abundantly’ at all? The problem of mixed disjunctions has seen less discussion in the truth pluralism literature, aside from on occasions where it has been used to yield potential solutions to the more frequently discussed problem of mixed conjunctions.10 However, Pedersen (2010) considers the following variant for truth pluralism. Suppose we have two truth properties, true1 and true2. Can we not simply construct a disjunctive truth property, trueD, which will be possessed by all sentences that are either true1 or true2? And, if so, if we have this general truth property, why do we need the plural truth properties?11 The ontological pluralist can adopt the reply Pedersen gives to this worry. In the truth case, the disjunctive truth property does not render the plural truth properties obsolete, as it is a property that is dependent on those properties, both for its existence and for its instantiation. The disjunctive truth property would not exist if it were not for the plurality of truth properties that compose it, and it depends on them for its instantiation in that the only way a sentence gets to possess it is by possessing one of the plural truth properties (with some qualifications, see below). As a consequence, even if we admit the disjunctive truth property it does not harm the truth pluralist’s thesis, as the plural ways of being true are ontologically prior to the disjunctive truth property. Likewise, the ontological pluralist can admit the disjunctive quantifier, but maintain that it is a derivative quantifier, dependent on the plural quantifiers in much the same way as the disjunctive truth property is dependent on the plural truth properties. It depends on them for its existence, since the disjunctive quantifier would not exist were it not for the plurality of quantifiers that form its disjuncts, and the entities that fall within its range only do so because they fall under the range of one of the plural quantifiers. So if we do admit the general quantifier, it will not render the plural quantifiers obsolete; far from it, for it will be dependent on the plural quantifiers. (Indeed, McDaniel (2009: 300–5) offers a similar line of argument—see below.) What this discussion suggests is that, whilst there are no mixed inferences or mixed compounds problems for ontological pluralism, the ability to express sentences like (DE) make it difficult to avoid the introduction of a general way of being, even if we can mitigate the effects of that admission. We will now explore what happens if we go in that direction.

8.4 Derivative Pluralism and Explanation If we take the disjunctive quantifier argument seriously, then, even if we do not think that the domain-specific ways of being are obsolete, we are left with the idea that there is some general way of existing which can be inferred from the specific ways of existing. 10 Cotnoir (2009) uses this thought and De Morgan facts about negation are used to suggest a solution to the problem of mixed conjunctions. 11 As we saw in Chapter 7, Pedersen and C.D. Wright have endorsed a disjunctive account of truth, in Pedersen and Wright (2013a). On the other hand, Cotnoir (2013b) argues that the semantic paradoxes suggest that such a definition is not available on pain of contradiction.

derivative pluralism and explanation

151

That is, everything that exists exists in some way, even if some things exist in way 1, and other things exist in way 2. This is a derivative kind of existence, and a yields a view which is similar to the views of disjunctivism and second-order functionalism in the truth debate. Kris McDaniel (2009, 2010a), presents such a view, and explores the possibility that we can explain existence by thinking of ‘exist’ as an “analogous” term. Here is how he expresses the idea: Consider being healthy. I am healthy, my circulatory system is healthy, and broccoli is healthy. Let us suppose that there is a common property that we share. Even so, that in virtue of which we exemplify this common property differs from case to case. I am healthy in virtue of being a flourishing organism, my circulatory system is healthy in virtue of functioning properly, and broccoli is healthy in virtue of its contributing to the flourishing of organisms like me. Being healthy is something like a mere disjunction whose disjuncts include being a flourishing organism, being a properly functioning part of an organism, and being something that contributes to the flourishing of an organism. Each of these properties is more natural than being healthy. Being healthy is an analogous feature: each of the ‘specifications’ of being healthy just listed is more natural than the ‘generic’ feature. But being healthy is not a mere disjunction: the various specifications of being healthy are related in such a way to ensure some kind of unity. (Unlike, say, being an electron or a female sibling.) Analogous features are something akin to disjunctive properties, but they aren’t merely disjunctive. Analogous features enjoy a kind of unity that merely disjunctive features lack: they are, to put it in medieval terms, unified by analogy. (McDaniel 2010a: 695–6)

McDaniel suggests that ‘exists’ is analogical in this sense. On this view, we have a number of specific ways of being, and a general way of being, which, whilst being a disjunction of the specific ways of being, is not a mere disjunction, as there are connections we can make between the disjuncts.12 One question to consider at this point is whether we get the same sort of problem with the derivative existence view as we had with the derivative truth view. The shape of the problem here would be that we have some domain-specific ways of being, and a general feature which we derive from the domain-specific ways of being. But why now think that the derived feature is really a way of being at all? Presumably the domainspecific ways of being were identified because we had some prior understanding of what a way of being is, but why think that this derived feature has anything to do with this? The idea is that this sort of view leaves a key thing unexplained: what is to be a way of being, and why think that this derived feature has anything to do with being a way of being? Even if we can show that it is something that all existing things have, this does not mean that it has the chops to be considered a way of being: all existing things might also have the feature of being on God’s big list of existents, but this does not mean that being on God’s big list of existents is a way of being. 12 See Edwards (2012) for more discussion of distinctions between kinds of disjunctive properties. See also Clapp (2001).

152 Models of Being McDaniel (2009) presents a slightly different view in his development of a version of ontological pluralism specifically tailored to Heidegger’s account of being. On this view we have a specification of a generic sense of being in addition to the many different ways of being: The generic concept of being is represented in formal logic by the unrestricted existential quantifier. This quantifier ranges over whatever there is, regardless of which kind of being the thing enjoys. For absolutely every thing there is, i.e. for all x, we can truly say that ∃y (y = x). We can adequately represent the generic sense of “being” with the unrestricted quantifier of formal logic. (McDaniel 2009: 301)

This generic way of being is still taken to be derived from the specific ways of being, however, which are understood as restricted quantifiers. This invites the worry that the restricted quantifiers must be understood in terms of the unrestricted quantifier, for, if they are restrictions, they must be restrictions of a more general quantifier. McDaniel responds to this concern by holding that the restricted quantifiers are “semantically primitive restricted quantifiers” (McDaniel 2009: 303). This returns to the idea that the specific ways of being are more basic than the generic way of being, and that we can only understand the generic way of being by reference to the specific ways of being. More specifically, McDaniel suggests that one way to think of things is that we start understanding existence in specific contexts, and only after that do we deduce the more general, generic, way of being. However, it is not clear if this move helps. If we take the generic way of being to be a mere disjunction of the specific ways of being, then it looks like we have a derived way of being (what it is to be is just to be in the range of a quantifier defined as the disjunction of at least two other quantifiers). However, on the analogical view the idea is that we are not dealing with a mere disjunction, as there are connections between the different ways of being. This connection now calls out for explanation. One way to explain the connection is that there is a generic way of being, which all the specific ways of being somehow relate to, which explains why they are all ways of being. On this view, the generic way of being is prior, as the idea is that it explains the unity of all the specific ways of being. On such a view, the generic way of being cannot be understood as derived from all the specific ways of being, as it is in the mere disjunction view. The structural problem for the analogical view is the tension between the claim that the different ways of being share some features in common—which lead us to think that there is a generic way of being—and the claim that the generic way of being is derived from the specific ways of being. The first claim is needed to distinguish the analogical view from mere disjunctivism, but this introduces an appeal to a generic way of being in order to explain the unity of the specific ways of being, which is in conflict with the second claim. If we take it that the generic way of being is prior to the specific ways of being, then we need a model which accounts for the need for the specific ways of being if we are

determination pluralism about existence

153

to do justice to the motivations for ontological pluralism given in Chapter 7. I will now suggest that one way to do this is by giving a determination pluralist account of existence.

8.5 Determination Pluralism about Existence Recall that the structure of determination pluralism about truth is as follows. Truth is given as the property that is exhaustively described by the truth platitudes. This property is the property possessed by all true sentences, regardless of domain. For each domain there will be a property that determines possession of the truth property, and these properties are held fully distinct from the truth property itself. The relationship between the truth-determining properties and truth is underwritten by the conditionals of the form (Cdx): (Cdx) In domain of discourse x: if ‘p’ has property F then ‘p’ is true (has the property of being true), which in turn ground the order of determination on the biconditionals of the form (Bdx): (Bdx) In domain of discourse x: ‘p’ is true (has the property of being true) iff ‘p’ has property F. How does this help in the case of existence? The idea is that we have a general existential quantifier, which will be applicable to all existing things. A useful suggestion here that I will adopt comes from Salmon (1987: 63), who defines existence by saying that to exist is to be something; or, more precisely, to be identical to something.13 He also holds that this yields a first-order property of existing, possessed by all existing individuals, which is the property of being something—or, more precisely, the property of being identical to something. In this general sense of existence then, to be is just to be something, and this property—the property of being something—is shared by all existing objects. In more formal terms, the basic idea is that we can define existence in terms of the principle ∃y (y = x),14 which is the formalization of the idea that to exist is to be identical to something. Then, using lambda abstraction, we can get the formal analogue of the idea that to exist is to have the property of being identical to something, which is expressed as (λx)(∃y) [y=x].15 This general understanding of existence is a minimal one, and perhaps not particularly interesting. This is perhaps hardly surprising, as, as Quine (1948) notes, the question of what there is is answered fairly trivially: everything! Moreover, there is a peculiarity with existence in that, for most properties we can distinguish between things that See also Sider (2009, 2011). See also the passage from McDaniel quoted earlier (from McDaniel (2009): 301). 15 For more details on this, see Salmon (1987: 63–4). The connection between existence and identity is also encapsulated in the Quinean slogan, ‘no entity without identity’—see Quine 1948. 13 14

154 Models of Being have the property and things that lack it, but, in the case of existence, there is no such distinction, as there are no things that do not exist!16 The interesting distinction here is thus not between things that exist and do not exist, but rather between the different ways that things get to exist, and there are interesting questions about the relationships are between the ways of existing and the general property of existence. The idea that there are different ways of being will kick in when it comes to considering the question of how it is that a particular object gets to have this general property of existing. Consider again the analogy with winning. If one wants to win, one needs to know which game one is playing, and, also, what it takes to win that game, which, of course, changes from one game to the next. We used this analogy in Chapter 7 to understand truth pluralism insofar as the ‘goal’ of an assertion (speaking very loosely) is to be true, but in order for a sentence to be true we need to know which domain of discourse it is from, and, also, what it takes to be true in that domain. Now let’s see how the analogy helps explain ontological pluralism, using the distinction between sparse and abundant existence as our model. In the case of existence we can construct the analogy as follows. Let’s limit our attention to objects for now, and suppose we are considering the existence of an object. For a determination pluralist about existence, in order to answer this question we need to know what kind of object the object is—sparse or abundant in this case. The idea then is that there are different things that it takes to exist in each of these ways. For example, we might think that if a sparse object exists, it must have causal powers: this is what it takes for sparse objects to get to exist. Contrastingly, we might think that if an abundant object exists, its associated singular term must appear in a true sentence: this is what it is for abundant objects to get to exist. Thus, the thought goes, there are different properties that determine the existence of a potential object in different domains—different ways that objects must be in order for them to exist. For instance, take Merricks’s argument against the existence of macroscopic objects. Here is an important point he makes about them: If there were baseballs, they would break windows, they would injure batters, they would cause visual sensations (and so be seen), and they would cause tactile sensations (and so be felt). In general, if there were inanimate macrophysical objects, they would have causal powers. But given the Overdetermination Argument and the schema of which it is an instance, if there were such objects, they would not have causal powers. So there are no such objects. Arguments linking existence to causal powers are often controversial. Consider such arguments against Platonic Forms or moral properties. But this is no problem. For I do not rely on the entirely unrestricted thesis that to be is to have causal powers. I claim only that, for macroscopic objects, to be is to have causal powers. Macroscopic objects are exactly the sort of things about which this causal commitment seems to be true. There should be no controversy on this point. The controversy, instead, is about which other sorts of things are like macroscopic objects in this way. (Merricks 2001: 81) 16 Assuming we do not adopt a form of Meinongianism of course. See Berto (2012) for an extended discussion.

determination pluralism about existence

155

Whilst Merricks is not committing himself to ontological pluralism in this passage, we can adopt his idea for our purposes: when considering whether an entity exists, we need to consider what sort of entity it is supposed to be, and how entities of that sort are supposed to exist. If an entity fails to meet its conditions for existing in its own way, then it does not exist. This is indeed the idea behind the framework given here. To take another example, suppose you are assured that numbers exist. As things stand, this doesn’t really tell you very much, and to properly understand the situation you need to know more about the way in which numbers are taken to exist: it could be that you are being told that numbers exist because they exhibit the same sort of features as people and elephants, or it could be that numbers exist because they are the referents of singular terms in true sentences. On the first interpretation, we suppose that numbers need to have causal powers in order to exist, and thus do not exist, as they do not have causal powers. On the second account, we suppose that there need to be true sentences making reference to numbers for them to exist, in which case they would. In terms of the connection to the idea that to exist is to be something, we can say that, for sparse objects, to be something is to have causal powers, and, for abundant objects, to be something is to be referred to by a singular term in a true sentence. We can push the analogy with determination pluralism about truth further by noting that we will get the same kind of conditionals in the case of existence. (Cspa) If x exists, then x has causal powers. (Cabu) If x exists, then x’s associated singular term appears in a true sentence. Moreover, assuming that there is one existence-determining property in each domain, we will also get the associated biconditionals: (Bspa)

In the physical domain: x exists iff x has causal powers.

(Babu) In the social domain: x exists iff x’s associated singular term appears in a true sentence. If an ontological pluralist adopts determination pluralism for existence, then they have an answer to the question of how it is that objects exist: what it takes for an object to exist varies from one kind of object to the next. This allows access to a general way of being, whilst maintaining the importance of the idea that the way this way of being is attained is different for different kinds of objects. One might still ask whether the general, unrestricted, way of being, expressed by the unrestricted existential quantifier is more basic than the domain-specific existence determining properties. Determination pluralism holds that the general way of being is more basic than the domain-specific properties. This is because, as we noted earlier, on the determination pluralist model, the general way of being comes first, and then we understand the domain-specific properties as ways that entities get to have that general way of being. This gives determination pluralism an advantage over derivative pluralism. It has a clear account

156 Models of Being of which is more basic between the general way of being and the specific ways of being, and thus does not get caught in the problems that the analogical view did.17

8.6 Existence, Sparseness, and Abundance Finally, as we did with truth in Chapter 7, we will address the issue of whether existence itself is sparse or abundant. We can note that we find the same issues that we had with truth in Chapter 7, in that, due to existence’s role in the formation of the distinction between sparseness and abundance, existence itself cannot be sparse or abundant. Recall that questions of sparseness or abundance are asked of objects and properties of certain kinds, within certain domains. On the determination pluralist model, though, existence is not bound to a domain, much as we saw truth is not bound to a domain in Chapter 7. As a consequence, existence itself must remain neutral on the sparse/abundant issue, as it is something that both sparse and abundant entities can have: if existence were either sparse or abundant, this would force us to adopt a form of  monism about existence.18 Setting aside for a moment the peculiarity of asking whether existence exists, we can illustrate this as follows. If existence exists sparsely, then the following two conditionals must be true: If existence exists, then existence has causal powers If existence has causal powers, then existence exists. If this is the case, then we have a view where existence across the board is to have causal powers: anything that exists must have causal powers, which rules out the idea of abundant existence. On the other hand, if existence exists abundantly, then the following two conditionals must be true: If existence exists, then ‘exists’ appears in a true sentence. If ‘exists’ appears in a true sentence, then existence exists. If this is the case, then existence is tied to ‘exists’ appearing in a true sentence across the board, which rules out the idea that there is a connection between existence and causal powers in some domains. Remember also the connections we have made between responsive predicates, sparse properties, and representational truth, and between generative predicates, abundant 17 Note also that determination pluralism does not fall foul of the disjunctive quantifier argument, for there is still significant explanatory work for the domain-specific properties to do, namely explain why different kinds of entities exist in the general sense. Whilst the general way of being is not dependent on the specific ways, the specific ways are still essential to the picture. 18 Compare Sider’s (2009,  2011) arguments for the idea that the unrestricted quantifier is ‘natural’. His arguments rest on the idea that there is an independent structure to reality, which the unrestricted quantifier latches onto. It is part of ontological pluralism to deny that this is the case in all domains.

existence, sparseness, and abundance

157

properties, and superassertibility. With these in mind, the following two options are available: (a) If existence is sparse, then all sentences of the form ‘x exists’ will be responsive, and true in virtue of correspondence. (b) If existence is abundant, then all sentences of the form ‘x exists’ will be generative, and true in virtue of superassertibility. Option (a) enforces a sparse monism about existence, and option (b) enforces an abundant monism. Thus, if we want to maintain pluralism with a general property of existence, that property cannot be abundant or sparse. As was the case with truth, it should not come as a surprise that we cannot think about existence in these terms, for existence is a crucial part of the sparse/abundant distinction. If we cannot appeal to the idea that existence is abundant or sparse, does this mean we have no account of existence claims? I don’t think so. Recall how existence claims are treated on this view. As we saw with the example of whether numbers exist above, the evaluation of an existence claim requires the following process. First of all, one needs to know what sort of entity we are talking about, and hence what way of being is appropriate. Once we have settled this, we look to the appropriate biconditional for the domain in question. For example, take the sentence ‘the number 5 exists’. This would be false if we took the following biconditional to be appropriate: x exists iff x has causal powers This is because it is not the case that the number 5 has causal powers. However, if we apply this biconditional: x exists iff ‘x’ appears in a true sentence then the sentence ‘the number 5 exists’ is true, as ‘the number 5’ appears in a true sentence. Existence claims are responsive in the terminology used in the book so far, as they are dependent on other things, namely the possession of the existence-determining properties. In our discussion of determination pluralism about truth, we found that the truth of truth attributions is responsive because they depend on a sentence having the relevant truth-determining property. Likewise, existence claims are responsive as they are dependent on an entity having the relevant existence-determining property. This is not inconsistent with the idea that the existence of some entities is dependent on truth, whereas the existence of others is not, however, as this concerns the particular ways that entities get to have this general way of being. As in the truth case, the connections between truth and existence make it such that this does not mean that existence is straightforwardly a sparse property. This is because some existence claims will be dependent on the truth of sentences. In particular, all existence claims that concern abundant entities will be true because there are true

158 Models of Being sentences involving their corresponding singular terms. As in the case of truth, this is a key feature of abundant entities! The versatility of existence is such that it cannot be subject to the sparse/abundant distinction as understood here.

8.7 Summary In this chapter we have explored the various ways that ontological pluralism can be formulated, and discussed some key features of the view. In doing so, we have brought together the general pluralist framework that we get from our considerations about domains, truth, and existence. In Chapters 9 and 10 I will turn to two challenges from primitivist theories of truth, and demonstrate some additional uses of the pluralist framework.

9 Primitivism and Truth-Making In Chapters 9 and 10 we will explore some applications of the pluralist framework, and show how responses can be given to two different threats from primitivist approaches to truth. Primitivist approaches distinguish themselves from deflationism by holding that truth has important explanatory roles to play, but deny that there is any informative theory of truth to be given. The different threats from primitivism are from two different directions, as they are from two different forms of primitivism. Interestingly, they both involve the notion of truth-making, but in opposite ways. The first threat comes from the claim that there are truths for which there are no truth-makers, which implies that we should be primitivists about truth, as there are truths for which there are no explanations of why they are true. The second threat comes from a commitment to truth-makers, and holds that, once we are committed to there being truth-makers for truths, we do not need a theory of the nature of truth. Whilst it may be a little bizarre that we find both approaches under the banner of primitivism, it does creates a pincer-movement to squeeze the pluralist approach I have given: either there are no truth-makers for certain truths, in which case the pluralist is wrong to think that a theory of truth has explanatory power, or there are truth-makers for truths, in which case the pluralist theory of truth is obsolete. In this chapter I will address the first threat, and in Chapter 10 I will address the second threat. In doing so I will demonstrate some additional explanatory work that the pluralist theory can do, and also re-emphasize the importance of the metaphysics of truth.

9.1 Truth-Making Various theories have been offered of truth-making, but the basic idea is that truths need to be made true, and that, for every truth, there will be a truth-maker, something that makes it true. The thing that makes a sentence true is usually thought of as some existing entity, such as an object, or object-property state of affairs combination. As Armstrong puts it: p (a proposition) is true if and only if there exists a T (some entity in the world) such that T necessitates that p and p is true in virtue of T. (Armstrong 2004: 17)

160 Primitivism and Truth-Making The truth-making relation is supposed to be something special: a distinctive relation which holds between a truth and a truth-maker, quite unlike any other relation. In Armstrong’s case, there is taken to be a relationship of necessitation between a truthmaker and its truth: the existence of the truth-maker necessitates the truth of a sentence. As you may have guessed from the way Armstrong defines the idea, Armstrong originally introduced the notion as a way of developing what he took to be the key insight behind the correspondence theory of truth.1 The relationship between theories of truth-making and theories of the nature of truth is a vexed one, though. Some see truth-making as an extension of a particular theory of the nature of truth (the correspondence theory); others see truth-making as compatible with a variety of approaches to the nature of truth; whereas others see a theory of truth-making as a replacement for a theory of the nature of truth.2 In this chapter I aim to clarify this relationship, and argue that truth-making is compatible with a variety of approaches to the nature of truth. I begin by examining Merricks’s (2007) arguments against truth-maker theory, and how they connect to conceptions of the nature of truth. I then argue that a pluralist theory of truth has the resources to provide a response to Merricks’s objections to truth-maker theory. In doing so, I explore potential pluralist accounts of counterfactuals, claims about the past, and negative existentials, which are areas not previously discussed by pluralists.

9.2 Merricks on Truth-Making Trenton Merricks (2007) is a primitivist about truth who rejects theories of truth-making. Merricks argues that the theory of truth-making is mistaken, and uses his reasons for thinking that the theory of truth-making is mistaken to show that theories of the nature of truth are also mistaken. As he also rejects deflationism, he ends up with a primitivist view of truth. Next I will explore the objections Merricks makes to truth-maker theory. My aim is to show that the pluralist metaphysics I have developed has the resources to respond to these objections. Insofar as we want to preserve the idea that truths are made true, this illustrates an important use of the pluralist framework. It also shows that we are not forced into a primitivist view, and nor do we have to reject outright the idea that all truths are true in virtue of something.

9.2.1 Merricks on the truth-maker principle and truth A core claim of truth-maker theory is that every truth is made true by something. Merricks’s central objections to truth-maker theory are that there are classes of truths 1 There are many different accounts of truth-making aside from Armstrong’s. The details do not matter for what is said below, but see, for example, the papers in Beebee and Dodd (2005) for an influential selection. 2 Influential examples of these views include, of the first, Armstrong (2004); the second, Dummett (1991); and the third, Lewis (2001) and Lowe (2008).

merricks on truth-making

161

for which this principle fails. His main examples are negative existentials (such as ‘there are no hobbits’), counterfactuals (such as ‘If Queen Elizabeth II had been born in seventeenth-century Japan, she would have been a samurai warrior’), and, assuming presentism, claims about the past (such as ‘the Trojans were conquered’). Merricks argues that many negative existential claims are true, but there is nothing that exists to make them true. In the case of counterfactuals, Merricks contends that it is too hasty to rule out that counterfactuals are true, but, again, almost by definition, there could be nothing in existence that makes them true, as they are about ways things could have been, as opposed to ways things are. For presentists, past entities do not exist, so there is nothing to make the sentence ‘the Trojans were conquered’ true. Examples such as these, Merricks thinks, are sufficient to show that the general truth-maker principle is false, and should be rejected. Furthermore, he things that these reasons carry over to show why particular views about the nature of truth are mistaken. In the case of the correspondence theory of truth, for example, he says the following: So Russell, Austin, and Moore all say that each truth corresponds to some existing thing that that truth is about. In fact, they seem to agree that a claim’s corresponding to something that exists, something that that claim is about, is what it is for that claim to be true. Thus they offer an analysis of being true, and so offer a theory of truth. But their theory of truth is false. It is false because there are some truths that are not true in virtue of being related to some existing thing that that truth is about. For example, there is no existing thing that is what that hobbits do not exist is about (in any good sense of ‘about’), and is such that being appropriately related to it is what it is for that hobbits do not exist to be true; and so it goes generally for true negative existentials. Moreover, it is false that there exists something that is what that there might have been a dozen more fundamental particles is about, much less something whose being related to that claim is what it is for that claim to be true. Likewise, given presentism, for that the Trojans were conquered. The same goes for many true subjunctive conditionals. Indeed, this goes for any truth that is not about the mere existence of something. (Merricks 2007: 173–4)

This problem for the correspondence theory,3 Merricks thinks, demonstrates that we cannot think of truth as a ‘relational property’: Some truths are not true in virtue of how they are related to any existing entity or entities. So we should reject any theory of truth that implies that every truth is true in virtue of how it is related to some existing entity or entities. So we should reject any theory that implies that what it is to be true is to be appropriately related to some existing entity or entities. And so we should reject any theory of truth that takes being true to be a relation between each true (primary) truth-bearer and some entity or entities. Therefore, we should conclude that being true is not a relation between a truth and some entity. (Merricks 2007: 181–2) 3 Merricks also discusses the coherence theory of truth and the identity theory of truth. I will not address Merricks’s comments about the coherence theory, as they are made in the context of a realist approach to truth, which is not the context in which I have mentioned the coherence theory previously.

162 Primitivism and Truth-Making If we cannot think of truth as a relational property, then, Merricks claims: Being true is monadic. But it is not intrinsic. For whether a proposition exemplifies being true is often a matter of how things are around it. For example, whether that dogs bark is true is not only a matter of what that proposition is like, but also a matter of whether there are any dogs and what those dogs do. (Merricks 2007: 182)

Merricks thus holds that being true is a non-relational extrinsic property. It is worth pausing for a moment to consider this claim, particularly as relational properties are often confused with extrinsic properties.4

9.3 Extrinsic Properties, Relational Properties, and Truth Extrinsic properties are opposed to intrinsic properties. For example, take Bader’s (2013) analysis of an intrinsic property:5 A property F is an intrinsic property iff F is always had solely in virtue of how a thing itself is or F is a fundamental property. (Bader 2013)

This asks us to look at how the property is instantiated in order to determine whether or not it is intrinsic or extrinsic. The main thing to say about the distinction between intrinsic and extrinsic properties is that the former are possessed entirely in virtue of the way the object itself is, and the latter are not.6 Relational properties, on the other hand, are defined in terms of the instantiation of some specific relation: A property, F, is a relational property if and only if there is a relation, R, such that the holding of R is what a’s instantiating F consists in.7 In terms of debates about truth, one often comes across the claim that the idea that truth is a relational property is a characteristic claim of the correspondence theory of truth.8 I think it is clear, though, that the idea that truth is a relational property is also compatible with theories of truth beyond the correspondence theory, including its 4 For examples of this in the truth debate, see Wright (2003: 247–8), Engel (2002: 11), and Rasmussen (2014: 2). For more general cases (though not of confusion!), see Francescotti (1999, 2014) who aims to analyse extrinsicality in terms of relationality. 5 For more details on how this account is arrived at, see Edwards (2016). 6 We can also note that the relation marked by the phrase ‘in virtue of ’, often referred to as grounding, or ground, is taken as a technical notion here. We will follow that standard line that its main features are that that it is asymmetric (if a grounds b, then b does not ground a), irreflexive (a cannot ground a), and transitive (if a grounds b, and b grounds c, then a grounds c). There is an extensive literature on grounding and the ‘in virtue of ’ relation. Some main examples include Fine (2012), Schaffer (2009), Jenkins (2011), and Rosen (2010). 7 I just use dyadic relations as an example here, but the analysis can clearly be extended to include relations between different numbers of entities. See Edwards (2016) for more details. 8 See, e.g. David (2005, 2015), Fumerton (2013: 198), Vision (2004: 4, 12, 53), and Wrenn (2015: 73).

extrinsic properties, relational properties, and truth

163

main competitors. This is important in the context of Merricks’s argument, and we will come back to this issue below. To illustrate, we will need some specific versions of these theories to work with, so let us use the following accounts: CORR:

‘a is F’ is true iff ‘a is F’ corresponds to a fact

PRAG: ‘a is F’ is true iff ‘a is F’ is, or can be, warranted by some state of information and some warrant for it would survive arbitrarily close scrutiny of its pedigree and arbitrarily extensive increments to or other forms of improvement of our information.9 COH: the belief that a is F is true iff the belief that a is F coheres with the maximally coherent system of beliefs Each of these accounts holds that truth is an extrinsic property, according to our account of extrinsic properties given above. They all hold that the bearers of truth are distinct from the things that make those bearers true. For CORR, a sentence is true in virtue of things distinct from it—the object and property that the sentence is about. For PRAG, a sentence is true in virtue of something distinct from it—its assertion being warranted by some state of information. For COH, a belief is true in virtue of its  cohering with the maximally coherent system of beliefs. Whilst the maximally coherent system includes that belief within it, is distinct from it insofar as a whole is distinct from one of its individual parts. According to the three ‘traditional’ theories of truth, and not just the correspondence theory, then, truth is an extrinsic property. We can also show how, for these three views, truth is a relational property. We can specify the specific relations, in the instantiation of which the possession of the property of being true consists, as follows: CORR:

truth is a relation (correspondence) between a sentence and some fact.

PRAG: truth is a relation (durably warranted by) between a sentence and some state of information. COH:

truth is a relation (coherence) between a belief and a system of beliefs.

On each of these three views—the three traditional theories of truth—truth is a relational property, on the grounds that each takes it to consist in the instantiation of a particular relation which is the source of a sentence or belief ’s truth. So, the idea that truth is a relational property is not an exclusive claim of the correspondence theory, rather it is a view shared by all the three main traditional views of truth. It is important to note that these kinds of properties are distinct: intrinsic properties are not coextensive with non-relational properties, and extrinsic properties are not 9 This is based on Wright’s notion of superassertibility that we discussed earlier. It is by no means common to all pragmatists as they conceive themselves (e.g. Peirce, Dummett, and Putnam would not explicitly endorse it), but Wright argues that this is the account pragmatists ultimately ought to end up with. See Wright 1992, ch. 2 for discussion.

164 Primitivism and Truth-Making coextensive with relational properties. Weatherson and Marshall (2012) note that the property of having longer legs than arms is an intrinsic property—it is a property one has independently of how things are outside of oneself—but it is a relational property in that it involves the relation longer than. This is a property that is intrinsic, yet relational. Another example here is the example of identity: I am identical to myself, and this involves a relation (identity) but does not concern anything outside of myself. Further examples are the relations I bear to myself, such as loving or hating myself: if I have the property of being someone who loves themselves, this is a property that I have that has nothing to do with any other objects, but it is a property I have due to a relation. In the other direction, Merricks (2007, 2015) gives the example of being the only object in the universe as an example of a non-relational extrinsic property. It is extrinsic as its instantiation depends on how things are beyond the object, but it is not relational as it does not consist in the holding of any particular relation. There is thus conceptual room for the kind of property Merricks wishes to identify truth with, and we will return to this issue below in relation to pluralism.

9.4 Primitivism and Pluralism Let’s now return to Merricks’s argument. His idea is that there are truths in areas where there are no entities to make those truths true. Hence, he argues, truth cannot be a relational property, and we must consider truth to be a non-relational extrinsic property. This is then combined with a primitivist approach which holds that there is no substantive explanation available for why truths are true. This primitivist approach is in contrast with the pluralist approach I have recommended in that it denies, whereas pluralism asserts, that there are explanations of why truths are true. We can now explore how these differences come about. On a pluralist approach, the correspondence theory’s failure to accommodate certain kinds of truths is not a problem; indeed, it is to be expected. I will first consider this response in terms of the theory of truth, before returning to consider Merricks’s objection to truth-maker theory in light of this discussion. A pluralist can agree with some of Merricks’s claims, but reject the primitivist aspect of the view. It is key to note that Merricks’s arguments are based on the idea that we are approaching truth in a realist way: that is, that we are thinking that truth always depends on being. If truth always depends on being, then, if there are no relevant beings in an area, then there are no truths in that area. It should be clear, though, that if we adopt the pluralist framework developed in this book, we do not have to accept this argument. This is because we reject the assumption that truth always depends on being, and, as a consequence, have alternative understandings of areas where truth does not depend on being. This means that we do not have to accept the primitivist idea that there is no substantive explanation for why truths are true, because, in principle, there will be an explanation available in all areas for why truths are true, it is just that not all of these explanations will involve truths depending on the existence of

modal truths

165

entities. In some cases, for example, it will involve durable warrant being available for the sentence in question. Thus the pluralist can accept Merricks’s claim that there are truths that are not true in virtue of any existing entity, but deny that this means that there is no substantive explanation available for their truth. Let’s now explore in more detail how a pluralist might go about this. As should be clear from our discussions of pluralism so far, when faced with a class of truths that do  not fit a particular model of truth, the obvious thing to do is look for other approaches to truth which may fare better. By identifying classes of truths that do not fit the correspondence model, Merricks has done the pluralist a favour in that the classes of truths themselves are already identified. His examples fall into three main classes: modal truths, truths about the past, and negative existentials. Let’s consider each in turn, and see what a pluralist might suggest in each case.

9.5 Modal Truths In the case of modal truths, we will be looking to demarcate them in terms of the presence of modal terms in the sentences expressing the claims, such as ‘would’, ‘could have’, ‘had been’, ‘possible’, and ‘necessary’. These signify that the claim is not about how things are, but how things could have been. It is worth noting too, as we will discuss below, that whilst all of these terms express modal concepts, they may express different sub-kinds of modal concepts. Of course, it is not part of a pluralist view that one has to validate the truth of these claims: nothing in pluralism itself rules out taking, say, an error theoretical account of modal discourse, but let’s explore the issue nonetheless. Remember also that we are playing along with Merricks’s objections, so we cannot aim to use a correspondence theory combined with Lewis’s modal realism, for example, although, in the absence of causal relations between the actual world and other worlds, it is hard to see how such a view could get off the ground in any case.10 Let us start with simple, ‘atomic’, modal sentences about possibility; that is, sentences that contain a monadic modal operator, like ‘it is possible that’. The modal domain is a complex place, with many different kinds of possibility (logical, metaphysical, epistemic, nomological) in play. It is not unrealistic to suppose that each kind of possibility and necessity involves a different source of modal truth. Let us begin with logical possibility. If we interpret modal terms as expressing logical possibility, in virtue of what are they true? A natural answer here is one that is given simply in terms of logical consistency. A modal sentence involving logical possibility is true just in case it does not contradict the basic laws of logic. Suppose we take the basic laws of logic to be the law of non-contradiction, the law of excluded middle, and the law of identity. A modal sentence about the possibility of something obtaining will be true just in case it coheres with these laws, with coherence here being understood simply as consistency. Claims 10 Nor will we use any forms of ‘ersartz’ modal realism, such as Plantinga (1974) or Stalnaker (1984), or the modal fictionalism of Rosen (1990).

166 Primitivism and Truth-Making expressed in terms of the other kinds of possibility will be true just in case they cohere (are consistent with) the basic laws specified by those kinds of possibility. For metaphysical possibility, a sentence will need to be consistent with whatever metaphysical principles are added to logical possibility; with nomological possibility a sentence will need to be consistent with the laws of nature in the actual world; and with epistemic possibility a sentence will need to cohere with what is known.11 So, for basic statements about what is possible, we can think of the truth of a modal sentence as being determined by its coherence with the basic principles of the kind of possibility being considered. However, Merricks’s concern was with specific sorts of cases which look more complicated, such as ‘If Queen Elizabeth II had been born in seventeenth-century Japan, she would have been a samurai warrior’. One might think that this is a different sort of example, in that the account of modal truths above is given for atomic claims, in the sense that a monadic modal operator (such as ‘it is possible that’) is applied to an atomic sentence, whereas counterfactual claims like the example given are compound claims, in that a dyadic modal operator (such as ‘if X had occurred then Y would have occurred’) is applied to two different sentences.12 As I suggested in Chapter 7, a pluralist (like perhaps other theorists of truth) has reason to consider the truth of compound claims differently from the truth of atomic claims. The truth-value of the compound is determined by the truth-values of the components, in conjunction with the rules for the truth of that specific compound set out by the truth-table for that compound in the adopted logical system. With this in mind, the analysis of the truth of Merricks’s candidate counterfactual could be determined by how we choose to account for the truth of compounds in general. However, this move is too quick. One problem with taking counterfactuals as standard compounds is that the antecedent and consequent cannot easily be taken as separately evaluable claims. For instance, ‘Queen Elizabeth II had been born in seventeenth-century Japan’ and ‘she would have been a samurai warrior’ are not obviously evaluable in their own terms for their truth or falsity in the way that ‘it is raining’ and ‘the ground is wet’ are from the conditional, ‘if it is raining, then the ground is wet’. With this in mind, we have reason to treat counterfactuals more like atomic sentences than compound ones, as they cannot be broken down into separable component parts.13 11 Is it a problem that we are understanding modal truths in terms of consistency when consistency itself arguably needs to be understood in modal terms (e.g. for two statements to be consistent it cannot be possible that one contradicts the other)? This is a version of Russell’s famous (1912) objection to coherence theories of truth, namely that we must presuppose an understanding of truth in order to understand coherence, so coherence cannot be an account of the nature of truth. Remember here though that I am not proposing that consistency be an account of the nature of truth, rather an account of the property that determines truth for modal claims. Also, as this is part of a pluralist theory there are other aspects of truth that are not part of this analysis, such as the other truth-determining properties and the general truth property. Finally, there is the general issue of whether a reductive account of modal notions is possible. This may just be an instance of a non-reductive approach to modality. Thanks to an anonymous reviewer for raising this issue. 12 See Williamson (2007) for more on the relationship between these two kinds of modal operators. 13 For influential discussions of the truth of counterfactuals, see Lewis (1973), Jackson (1977), and McDermott (1996, 1999).

presentism and truths about the past

167

Here is one line a pluralist could take in light of what was said above about modal truth. One way to think about how we evaluate counterfactuals is that we think about how likely the thing described is on the basis of what we know. In the case of the Queen Elizabeth example above, we would think about what we know of Queen Elizabeth’s character and interests, and what we know about seventeenth-century Japan, and consider whether, if given the right opportunities, a samurai would have been Queen Elizabeth’s path in seventeenth-century Japan. The method also works with simpler examples, such as ‘if the match were to have been struck, it would have lit’. In evaluating the truth of this counterfactual we would think about what we know of the match, the conditions, the laws of nature, and the reliability of matches in general to make a judgement. If this is correct, one way to view counterfactuals is as statements of epistemic possibility: they are true or false in relation to what we know of the situation. Given what we know about matches and the conditions, we would say that it is true that if the match were to have been struck, it would have lit, yet it is false that if the match were to have been struck, it would have turned into an elephant. If this is the case, then we can treat the truth of counterfactuals in the same way as we treat the truth of statements of epistemic possibility: they are true in case they cohere with what we know. In fact, we would need to tweak this a bit: they are true in case they would have been the most likely outcome, given what we know. This is a more complex form of coherence, as it involves coherence with what we know, along with judgements about what is likely. This is evidently a broadly ‘anti-realist’ approach to counterfactuals, as the truth of a counterfactual is dependent on judgements we make, as opposed to the way things are.14 On this view there is no gap between the judgement and the truth: it is not the case, for example, that there is a fact of the matter about whether the match would have lit that would prove or disprove our judgement that the match would have lit. This accords with the assumptions we made to begin the project here: that there is something mistaken about thinking that there are entities out there that could make such claims true.

9.6 Presentism and Truths about the Past With regard to sentences about the past from a presentist perspective, the pluralist can provide a useful option here. Presentism claims that the only existing things are things that exist at the present time. Evidently, this makes the truth of sentences about the past difficult to substantiate, as, if they are about things that previously existed but that do not now exist, then there are no things in existence now to make those sentences true. One option to explore here is Dummett’s (2004) account of past truths. On Dummett’s view, sentences about the past are true only insofar as we have present 14 To follow a theme noted above, this is also not, as is probably clear, a reductive account of the  truth-conditions of counterfactuals, as locutions like ‘would have’ are used in the statements of the truth-conditions.

168 Primitivism and Truth-Making means of verifying them. We can understand verification here in terms of the notion of  superassertibility we discussed earlier. The idea would be that a sentence about the past will be true just in case we have warrant for asserting that sentence, and that warrant would persist through subsequent stages of inquiry. Methods of obtaining warrant may vary: some may involve direct recollection of past experiences, and others may involve corroboration of historical records. For instance, the sentence ‘I had eggs for breakfast this morning’ may be made true by my direct recollection of that experience, whereas the truth of the sentence ‘the Trojans were conquered’ may involve a more complex process involving the examination of historical records. This account of past truth is consistent with presentism, because all the materials used to establish past truths exist at the present time, whether they be direct recollections, occurring now, or corroboration of historical records, also occurring now. So, there is an account of truth available to the presentist, which could fit into a general pluralist framework, which does not require the existence of past entities to make the required sentences true. This is not to say that, on this view, that past entities do not exist, for our general pluralist approach to metaphysics allows for the existence of entities to depend on truth. What we would hold in this case is that past entities are abundant entities, generated by truths about the past, with those truths understood in terms of superassertibility. The existence of past entities would be different from the existence of present entities in that what makes past entities exist would be different from what makes present entities exist. This would also fit with the idea that past entities depend on the present for their existence: things in the past only have significance, and existence, insofar as we continue to discuss them and think about them.

9.7 The Truth of Negative Existentials Negative existential sentences like ‘there are no hobbits’ pose a difficult challenge. One reason for this is that they seem to be sentences of the same kind as non-negative existentials: they concern what exists, and should surely be afforded the same sort of treatment. This is what leads to the problem Merricks identifies, for surely positive existential sentences are true in virtue of what there is, but it is hard to see how negative existential sentences could be true in virtue of what there is, for they are about what is not. On the pluralist model I have been developing in this book existence claims are about a wide variety of different things; they are not limited to just physical, or concrete objects. Moreover, we have substantial accounts of what it is to exist, which may offer some help. Taken at face value, the sentence ‘there are no hobbits’ can be interpreted in a number of ways, and on some interpretations it is true, and on others it is false. It is true if we are talking about physical entities that exist in the same way that you and I do, but it is false if we are talking about fictional characters. To explain the idea, we can develop Thomasson’s (2007: ch. 2.5) approach in a pluralist way. Think back to our evaluation of the claim that hobbits exist, and Merricks’s argument against the existence of baseballs that we mentioned in Chapter 8. The idea there was that in order to evaluate an existence claim we need to know more about

the truth of negative existentials

169

what sort of way of being the entity is being claimed to have. In Merricks’s example we have the sentence ‘baseballs exist’. In order to evaluate it we need to know what way of being baseballs are supposed to have. If, as we supposed, baseballs are supposed to have the sparse way of being, then we can note that the following biconditional applies: baseballs exist iff baseballs have causal powers. From this we can also infer a nonexistence biconditional: baseballs do not exist iff baseballs do not have causal powers. The truth of the nonexistence claim is thus determined by establishing the truth of the right-hand side of the nonexistence biconditional. If we accept Merricks’s arguments against the causal powers of baseballs, then we can infer that baseballs do not exist. We can disambiguate the issue of the existence of hobbits in the same sort of way. If the claim is that hobbits are not material objects, then the truth conditions of the nonexistence claim are: hobbits do not exist iff hobbits do not have causal powers. The right-hand side of this biconditional is presumably true, so the sentence ‘hobbits do not exist’ is also true. On the other hand, if the claim is that hobbits are fictional characters, then the nonexistence claim can be put as follows: hobbits do not exist iff there are no true sentences about hobbits being referred to in works of fiction. The right-hand side of this biconditional, however, is false, as, for example, the sentence ‘Hobbits are characters in The Lord of the Rings’ is true, so the nonexistence claim is also false.15 The problem of negative existentials, as Merricks understands it, presents the challenge of how to explain the truth of nonexistence claims when there is nothing in existence to make them true. My claim here is that we can understand the truth of such claims when we consider what it must take for the entity in question to exist. If the existence condition fails, then any sentences saying that it does not exist will be true. Is this strategy self-defeating, for don’t we have to refer to an entity in order to attribute nonexistence of it? I think not, if we understand that we are dealing with conditional claims above. For example, Merricks’s argument against the nonexistence of baseballs doesn’t presuppose the existence of baseballs only to deny their existence, as he is dealing with conditional claims of the form ‘if there were such things, they would have causal powers’. Showing that the consequent of this conditional cannot be true allows him to infer the negation of the antecedent by modus tollens. This is what the method above suggests: we deal with conditional claims concerning what it would take for such a thing to exist, once we know what sort of thing we are dealing with. This does not presuppose the existence of the thing itself. 15

Thanks to an anonymous reviewer for discussion here.

170

Primitivism and Truth-Making

9.8 The Truth-Maker Principle Revisited Returning now to the truth-maker principle that Merricks finds problematic, we can find a way to rescue it if we wish. If we interpret the claim that every truth is made true by the existence of some thing as a claim about the existence of objects that make truths true, then Merricks is correct that this fails as a global principle. This is what is to be expected on a pluralist approach, as we hold that the existence of some objects is dependent on the truth of sentences. However, we can also interpret this principle in a milder way, which takes ‘thing’ to refer to a much broader class of entities. For instance, take the idea that there are some truths that are true because they have been proved to be true. On the face of it, this suggests that there are truths that do not depend on existing entities. However, if we look closer, this seems wrong: they do depend on the existence of something, namely a proof. If the proof did not exist, then the claim would not be true. Likewise for truths that are true in virtue of being superassertible. These require a durable warrant for their truth, and if the particular durable warrant that makes them true did not exist, then the claims would not be true. Likewise for truths that are true in virtue of coherence. If there did not exist the coherent set with which they cohere, or the coherence relation that allowed them to do so, then the claim would not be true.16 If we interpret the truth-making principle in this broader sense, where the notion of ‘some thing’ is kept very broad, then we can rescue the global applicability of the truthmaking principle. Here we would do well to remember Dummett’s words on the subject: It is [a] mistake to regard the principle that, if a statement is true, there must be something in virtue of which it is true, as peculiar to realism. On the contrary, it is a regulative principle which all must accept. (Dummett 1991: 331) To summarize, my claim is that a pluralist theory of truth provides a way of responding to Merricks’s objections to both truth-making and theories of truth, and, in doing so, suggests a way to preserve the global truth-making principle in a very general form.

9.9 Pluralism and the Relationality and Extrinsicality of Truth I will close this chapter by considering whether truth is a relational and/or extrinsic property on pluralist views. We can note that there is a further difference between strong pluralism and determination pluralism on this issue. Strong pluralism holds that there are different properties of being true, and each truth property is to be identified with a different property in different domains. There is no general, domain-independent, property of being true. On this proposal, because different kinds of truth are identified with different properties in each domain, the 16 This does not impact the idea that some forms of existence depend on truth, as there we are talking about the existence of certain entities, such as objects and properties, which do depend on truth for their existence.

summary

171

kinds of truth will inherit whatever features those properties have. So, if we use the examples of correspondence, superassertibility, and coherence in the different domains mentioned, there will be multiple truth properties, and each will be extrinsic and relational. A strong pluralist thus has a ready account of the extrinsicality and relationality of truth, in all domains, provided, of course, that all the properties they identify different kinds of truth with are extrinsic and relational. Determination pluralism, on the other hand, holds that truth is an extrinsic nonrelational property. Determination pluralism holds that there is a general, domainindependent, property of being true, and that this is a property shared by all true sentences from all domains. Truth is extrinsic on this view as possession of the truth property is ultimately dependent on things other than the sentence: it is dependent on another property that the sentence has that, crucially, the possession of which is determined by things other than the sentence itself. In the case of ‘snow is white’, for example, on the determination pluralist proposal that sentence will have the general truth property just in case it causally corresponds, and this will involve matters beyond the sentence itself. If we follow the chain of things that the truth of the sentence is dependent on, then we will find that it is dependent on things other than the sentence, thus rendering truth extrinsic. So, according to determination pluralism, truth is extrinsic, but it is not relational because there is not a relation in which being true consists. Being true is a property that a sentence has in virtue of possessing some other property, not in virtue of a relation it has to anything.17 It turns out, then, that determination pluralism offers a similar truth property to Merricks’s view. It is important to note, though, that this does not mean that determination pluralism is a form of primitivism. As Merricks himself notes (2007: 183), primitivism does not lie in the claim that truth is a non-relational extrinsic property, rather it lies in the idea that there is no satisfactory account of truth, and how truths get to be true. It should be clear that, on these issues, determination pluralism is not a primitivist view.

9.10 Summary In this chapter I have explored the threat to pluralism from Merricks’s primitivism, and argued that a pluralist has the resources to not only respond to the threat, but also provide others who are interested in truth-maker theory with some useful resources to respond to Merricks’s arguments. In Chapter 10 we will turn to the second threat to pluralism from primitivism and truth-making, which comes from a very different direction.

17 The closest one could get would be to say that it consists in the instantiation of some other property, but this does not offer much help. Firstly, it is controversial whether instantiation is a relation, but, even if it is taken to be, then taking instantiation to be the grounds for a property’s being relational would trivialize the notion of a relational property, for then all properties would turn out to be relational. See, e.g. Armstrong (1978), Lewis (1983), and Edwards (2014: ch. 2).

10 Truth-Making and the Nature of Truth In Chapter 9 I responded to a move towards primitivism about truth prompted by the perceived failure of theories of truth-making. In this chapter I will respond to a move towards primitivism which is based on the perceived success of theories of truth-making! I have claimed above that a way to preserve ideas about truth-making is to think of truth differently for different sorts of claims. However, what, exactly is the connection between theories of truth and truth-making? Might we be able to make do with the idea that there are different sorts of truth-makers without committing to the idea that there are different sorts of truth? In this chapter I consider this issue in detail, and argue that the notion of truth-making only makes sense in the context of views about the nature of truth. I argue that some stance on the nature of truth is a necessary prerequisite of a theory of truth-making. This causes problems for any theorist who wishes to combine a theory of truth-making with a deflationary or a primitivist approach to truth. The central idea here is that theories of truth-making do not make theories of the nature of truth obsolete. Instead, as we will see, they give another demonstration of the paramount importance of metaphysical investigations into the nature of truth.

10.1 Pluralism and Explanation As a starting point for this chapter I will use the debate between pluralists about truth and primitivists about truth. A recent objection to pluralism from the point of view of primitivism is due to Asay (2018), who argues that a theory of truth-making renders the distinctive claim of truth pluralism—that there are multiple ways of being true— obsolete. I begin by offering a rebuttal to his objection, showing that the proposed account of truth-makers without ways of being true that is intended to offer a more parsimonious alternative to truth pluralism fails. I go on to demonstrate how this rebuttal uncovers serious problems for the viability of primitivist views of truth, along with any approach to truth which aims to use truth-makers in the absence of a theory of the nature of truth. I claim that this clarifies the relationship between theories of truth-making and theories of the nature of truth: a theory of truth-making only makes sense against the backdrop of a theory of the nature of truth.

pluralism and explanation

173

I should note that whilst I use the objection to pluralism as a starting point, the points developed below about why a theory of truth-making needs a theory of the nature of truth are generalizable to non-pluralist views such as correspondence theory and pragmatism.

10.1.1 Truth pluralism As we have seen, truth pluralism is the view that different properties are relevant when considering the truth of sentences in different domains. The idea is that the property that explains why physical claims are true (correspondence) is different from the property that explains why social claims are true (superassertibility). Earlier I suggested that we understand these properties as ‘truth-determining’ properties, and will continue to use that expression here, though what I have to say will also apply to other formulations of truth pluralism that use different locutions. The key thing to keep in mind for the purposes of this chapter is that, on these views, these properties are both part of an explanation of the nature of truth, and part of the explanation of how sentences get to be true. A sentence is true on any of these models just in case it possesses the truth-determining property for its domain. To show how this works, let’s consider a physical sentence, say ‘snow is white’, on a correspondence model of truth for physical claims. This sentence will be true just in case it has the truth-determining property for its domain. In this case that property is correspondence, so it will be true just in case it corresponds to the facts. The sentence will, in turn, correspond to the facts just in case there exists a relevant fact—that snow is white, in this case—and the sentence corresponds to it in the relevant way.

10.1.2 The primitivist’s objection Asay (2018) objects to truth pluralism by attempting to undercut the need for the different truth-determining properties that form the heart of the pluralist view. He claims that pluralist explanations like the one just given above do not give us reason to think that we need different truth-determining properties, for a simpler, non-truth-determining-property involving, explanation is also available. This explanation cuts out the middleman, so to speak, and holds that all we need to explain a sentence’s truth is the ground-level truth-maker, and not the truth-determining property itself. That is, in our example above, Asay claims that all we need to explain the truth of ‘snow is white’ is the fact (in this case) that snow is white. The existence of this entity alone is sufficient to secure the truth of ‘snow is white’, and thus we do not need the additional layer of explanation that the truth pluralist gives, which goes via a relevant truth-determining property. This is given as a reason to favour primitivism over pluralism, a claim that we will evaluate below. If this is correct, it undercuts a central motivation for truth pluralism. This is because the positing of different truth-determining properties in different domains is perhaps the key distinguishing feature of the view, and one of the key reasons for introducing these properties is to provide the sorts of explanations like the one given above. If these

174 Truth-Making and the Nature of Truth explanations are shown not to require these truth-determining properties at all, then the motivations for truth pluralism are in serious question. In the next few sections I will argue that Asay’s argument is unsuccessful. In particular, I will claim that we cannot simply cut out the middleman in the explanation, because we cannot understand truth-makers without a prior understanding of truth. In addition, consideration of these issues will lead us into a more general discussion of the respective merits of pluralism and primitivism about truth. I will start by clarifying the relationship between truth-makers and truth using an analogy with the property of fatherhood.

10.2 Fatherhood and Truth Suppose that John is the father of Jimmy. Fatherhood is the sort of property the possession of which requires the existence of some specific thing other than the individual in question: a ‘father-maker’, if you will. John can only be a father if there exists (or has existed) something that makes him a father. In his case, it is Jimmy. Now, what is the relationship between the property of fatherhood and the idea of a father-maker? There is a very close relationship, in that the reason Jimmy is marked out as the entity that serves as the father-maker for John stems from facts about the property of fatherhood itself. For example, consider Jimmy, John’s dog Rover, and John’s house. These are three different entities: how do we know which (if any) serves as the father-maker for John? The answer is pretty straightforward: we turn to the nature of fatherhood to give us the conditions under which an entity counts as a father-maker. When we examine the nature of fatherhood we will find conditions laid down for an entity to count as a father-maker, and it is only in virtue of meeting these conditions that Jimmy counts as a father-maker for John. It is also via these conditions that we understand why John’s dog and John’s house are not viable father-makers for John.1 On this line of explanation, which is natural when considering fatherhood, there is an intimate connection between the nature of fatherhood and the very idea of a father-maker; indeed, the latter does not make sense without the former. Returning to truth, the same considerations apply, and these show that one cannot appeal to truth-makers without resting on some substantial facts about truth. For instance, take the state of affairs of my cup being on the table. This may be said to serve as the truth-maker for the sentence ‘my cup is on the table’. But why? First of all, why are states of affairs singled out as the kinds of entities that can serve as truth-makers for this sort of sentence, as opposed to, say, the tree outside my window, or the clouds in the sky? Moreover, even if we just consider states of affairs, why should this state of affairs be singled out and not a different state of affairs, like the state of affairs of my foot being on the floor? This is the analogue of the question of why Jimmy, and not Rover, is 1 One could develop this example further by considering one’s potential response to a person who claims that they are a father because they own a dog. Presumably any debate about whether dogs are legitimate father-makers will be a debate about the proper understanding of the nature of fatherhood.

fatherhood and truth

175

a father-maker for John, and, just like in the fatherhood case where the only way to give an explanation is to appeal to the nature of fatherhood, we must appeal to the nature of truth in order to give an explanation for why one entity is the relevant truth-maker and the other is not. Here is how that would go if we adopt a standard truth-making account coupled with a correspondence theory of truth: A truth is a sentence that is made true by something (a truth-maker). In order to identify the truth-maker for a truth we need to consider that truth is a relation between a sentence and the world, such that a sentence is true just in case it accurately represents the world. A specific sentence will accurately represent the world just in case the state of affairs that it represents exists.

We can use this explanation—which involves substantial claims about the nature of truth—to show why the state of affairs of my cup being on the table is the truth-maker for ‘my cup is on the table’. First of all, the explanation gives a clear account of what sorts of things are capable of being true, and what sorts of things are capable of serving as truth-makers. Both of these features come from this explanation concerning the nature of truth. With respect to why the specific state of affairs of my cup being on the table is the relevant truth-maker whereas the state of affairs of my foot being on the floor is not, this is because ‘my cup is on the table’ represents a specific state of affairs as obtaining, and is only true if that very state of affairs obtains. The state of affairs of my cup being on the table is that state of affairs, whereas the state of affairs of my foot being on the floor is not. This is the sort of explanation one would need to justify the claim that the state of affairs of my cup being on the table is the truth-maker for ‘my cup is on the table’. Notice that it involves direct appeal to the nature of truth, and, moreover, it must make this appeal in order to make sense, just as one needs to appeal to the nature of the nature of fatherhood in order to make sense of the idea that Jimmy is the fathermaker of John. In this example I have used an account of truth that is evidently in line with the sort of correspondence theory of truth traditionally aligned with truth-maker theory. The thought is that, in order to identify relevant entities, and even kinds of entities, as truth-makers, we need a prior understanding of truth to do so. This explanation is not necessarily tied to this correspondence-influenced account of truth, and it should hold for all the accounts of truth a pluralist wishes to use. For example, if we take a mathematical example, we can wonder what the truth-maker for ‘2+2=4’ is. To answer this question we need to take a view on what the mathematical truth-determining property is, and only then will we have a sense of what kinds of entities—mathematical proofs, say—are candidates to serve as the truth-maker for ‘2+2=4’. Again, the key idea is that to make sense of the kind of thing that can serve as a truth-maker in a domain one needs to have a sense of how to understand the truth-determining property in that domain, otherwise we have no way to understand what kinds of entities, let alone what particular entities, serve as truth-makers for the sentences at issue.

176 Truth-Making and the Nature of Truth My claim is thus that Asay’s argument against truth pluralism does not succeed. He claims that we can remove truth properties from the picture and just talk of truthmakers, but my counter-argument is that we cannot understand truth-makers without truth-determining properties, so the move he wishes to make cannot be made. This not only rebuts the objection to pluralism, but also points to a significant fact about the relationship between theories of truth and theories of truth-making, which is that a theory which aims to use truth-makers must have both substantive and informative things to say about the nature of truth. This vindicates the importance of the metaphysics of truth, and shows that the project of explaining how truths get to be true cannot be separated from the project of explaining what truth itself is. We will now turn to explicitly note the problems this causes for primitivism.

10.3 Truth-Making and Truth The considerations above undermine any attempt to talk about truth-makers without talking about truth. In particular, they demonstrate the peculiarity of holding a position where truth-makers are given substantial theoretical roles in a picture where there is no substantial nature of truth. Consider again the fatherhood case. It would be very strange if we acknowledged the existence of father-makers, like Jimmy is to John, but denied that there was anything substantial to say about the nature of fatherhood. Indeed, I suggested above that such a position would not make sense as we could not understand father-makers without an appeal to facts about fatherhood, but here I  suggest that the motivation for even thinking this in the first place looks very odd indeed. Before turning to primitivism itself, I will briefly note why this causes further problems for deflationary theories of truth.

10.3.1 Deflationism and truth-making What the above considerations show at the outset is that any attempt to combine a deflationist theory of truth, where there is no substantial property of being true on offer, with a substantial theory of truth-making would be unsuccessful.2 Deflationary theories of truth explicitly deny that there is anything interesting to say about the nature of truth, and that truth plays any interesting theoretical roles. If what is said above is correct, then they would be unable to avail themselves of truth-makers, for they lack the resources to explain either how a particular kind of entity serves as the truth-makers for a certain kind of sentences, or how a specific entity serves as the specific truth-maker for a sentence. In addition, we can see further problems by examining how Horwich aims to show that deflationism is able to give expression to the idea that truths depend on the world. Here is Horwich’s argument in his own words:3 2 This is attempted by Lewis (2001). Further objections to Lewis’s approach are given by Vision (2003) and MacBride (2013). 3 Note that Horwich uses the terms ‘minimalism’ and ‘the minimal theory’ in this passage, but, for our purposes, we can use these interchangeably with ‘deflationism’ and ‘the deflationary theory’.

truth-making and truth (1)

177

’s being true is explained by snow’s being white.

That is to say, (2)

is true because snow is white.

But these intuitions are perfectly consistent with minimalism. In mapping out the relations of  explanatory dependence between phenomena, we naturally and properly grant ultimate explanatory priority to things such as the basic laws of nature and the initial conditions of the universe. From these facts we attempt to deduce, and thereby explain, why, for example,

(3)

Snow is white.

And only then, invoking the minimal theory, do we deduce, and thereby explain, why (4)

is true.

Therefore, from the minimalistic point of view, (3) is indeed explanatorily prior to (4), and so (1) and (2) are fine. Thus we can be perfectly comfortable with the idea that truths are made true by elements of reality. Since this follows from the minimal theory (given certain further facts), it need not be an explicitly stated part of it. (Horwich 1998: 104–5)

As you can see, the conclusion that Horwich wants to draw here is that there is an order of dependence in instances of the equivalence schema which make the truthinvolving claims on the left-hand side dependent on the non-truth-involving claims on the right-hand side. However, as Crispin Wright (1992: 26–7) has pointed out, Horwich’s argument above does not license that conclusion. If—and this is questionable—Horwich’s claims about laws and initial conditions of the universe are correct, what it does license is an explanation of why snow is white, and then a transferal of that explanation across the instance of the equivalence schema to also explain why it is true that snow is white. All Horwich has shown here is that, whatever explanation is good for the whiteness of snow will also be good as an explanation for the truth of the proposition that snow is white,4 but this does not show that there is a dependence of the left-hand–side of the equivalence schema on the right-hand side. Moreover, we can question whether the deflationist is licensed to even attempt this move, given what we have said about deflationism earlier. They would need to show, firstly, that the description given in (1) does not appeal to an object’s instantiation of a property to explain the truth of a sentence, for this is one of the things that deflationists wish to deny when asserting the primacy of the instances of the equivalence schema. Secondly, they would need to show that the explanation of (3) does not appeal to a substantial notion of reference: that is, we would need to explain why 4 Note that there are independent reasons to doubt that such explanations are good for both, but that is a separate issue. Horwich (2010: 310 n. 12) aims to respond to Wright by saying that there is an order of explanatory dependence: one first infers that p, and only then infers that p is true. However, Horwich does not give us reasons for thinking that this must always be the case (see, for example, our discussion of the varying relations between being and truth in Chapters  6 and  8). Moreover, this still does not seem to adequately speak to Wright’s objection if we are still saying that the same explanation ultimately holds for both. See also Künne (2003: 156–7).

178

Truth-Making and the Nature of Truth

snow  is white without any appeal to the idea that there are substantial—perhaps causal—relationships between ‘snow’ and snow, and ‘white’ and whiteness. Even though the argument fails, these important issues in its construction are left untouched. It is thus doubtful that deflationists can deliver on the idea that truths are made true. It is also puzzling why they would even want to: as we have noted, deflationists about truth are required to be global deflationists, which precludes any substantial metaphysical investigation, including the notion of some sort of dependence between sentences and the world. The idea then that there needs to be some explanation of the apparent dependence of truths on the world seems to run counter to the very ethos of the deflationary position, and the motivations for combining deflationism with truth-making are unclear. If some deflationists are uncomfortable with this, and see the need to provide explanations in these areas, then perhaps they should not be deflationists.

10.3.2 Primitivism and truth-making The next question to address is whether primitivism faces the same sort of problem. We encountered primitivism in Chapter 9, and, whilst it comes in various forms,5 there is a common core to the view. Primitivism denies that there is an informative analysis of truth available, whether we are considering the predicate ‘is true’, the concept of truth, or the property of being true. (For reasons noted in Chapter 1, I have so far avoided talking about the concept of truth, but primitivist views discuss the concept of truth in detail, so it may be necessary to talk in those terms briefly here, just for the purposes of evaluating primitivism.) Primitivists, following the lead of Davidson (1999), hold that there are interesting connections to be drawn between truth and other connected phenomena, but they hold there is no informative analysis of truth itself that can be given, as either a predicate, a concept, or a property. Primitivists thus aim to distinguish themselves from deflationists by holding that truth has interesting theoretical roles to play, either as a concept (Davidson, Asay), or as a property (Merricks), usually encapsulated by the connections drawn between truth and other related phenomena alluded to earlier. As Asay puts it: According to the primitivist conception of truth, truth is an indefinable conceptual primitive. That is to say, the concept of truth cannot be defined, without circularity, in terms of other concepts (like correspondence, coherence, or assertibility). Taking truth as primitive is consistent with saying that truth is an explanatorily useful concept, one that can be employed in giving analyses of other philosophical notions such as belief, meaning, and assertion (as Davidson, for one, does). (Asay 2018: 189)

The interesting question for our purposes here is whether primitivism is compatible with truth-making. Whilst, as we saw in the previous chapter, Merricks does not look 5 See, e.g. Davidson (1999), Sosa (2001), Merricks (2007), and Asay (2013). Frege (1918) is also sometimes interpreted as a primitivist.

truth-making and truth

179

for such compatibility, other primitivists, such as Asay, do, and claim it as a virtue of primitivism. I do not think it is compatible, for nothing in the primitivist view is able to do the work necessary to identify both the kinds of entities needed for truth-makers, and the specific entities for specific truth-makers. Let’s explore why. Consider again our example of ‘my cup is on the table’: what in the content of the primitivist theory is able to determine what sorts of entities are candidate truthmakers for this sentence? The answer is, nothing: we cannot appeal to the truth concept, for it has no specifiable content, and we cannot appeal to the truth property, for there is no informative property of being true on the primitivist account. The claim about the truth property here is fairly straightforward: given the primitivist’s denial that there is an informative property of being true, there is not a property that can provide the kind of explanation we need. The claim about the concept requires a little clarification, so let’s look at it in more detail. We noted just above that the primitivist claims that the truth concept is ‘indefinable’. The claim has both a negative and a positive component. The negative component is that we cannot give an elucidation of the concept of truth in terms of necessary and sufficient conditions, as, for example, some traditional forms of the correspondence theory may try to do by analysing the concept of truth as the concept of corresponding to the facts. The positive component is that we can nevertheless think of truth as a substantive concept that can play significant explanatory roles. So, presumably, the primitivist thinks that claims such as ‘truth is the norm of belief ’, and ‘the meaning of a sentence is given by its truth conditions’, and ‘to assert is to present as true’ are true statements, even though there is no reductive analysis of the concept of truth. Indeed, if we take the appeal to Davidson seriously, we might think that these principles help us to informally elucidate the concept of truth by connecting truth to other concepts, without giving a reductive analysis of truth. However, it is difficult to see an explanation of how these principles are true. We can try out some options to illustrate this. One option would be to say that they are true by mere stipulation, as they are taken to be expressive of foundational and basic connections between equally foundational and basic concepts. This may be an option, but it does not look consistent with what the primitivist says. In the quote above Asay suggests that truth is a concept “that can be employed in giving analyses of other philosophical notions such as belief, meaning, and assertion”. If this is the case, then these do not look to be equi-fundamental concepts: rather truth is taken to be the primitive that is then used in part to explain the other concepts. Indeed, this looks to be a key aspect of truth having an explanatory role. An alternative route would be to suppose that the principles connecting truth with these notions are true not in virtue of the nature of truth, but in virtue of the nature of the other concepts connected to truth. For instance, it might be that meaning is such that it is analysed in terms of truth, and that this is a fact about meaning, rather than truth. This is evidently an option, but it is explanatorily peculiar if we are thinking of truth as playing a significant role in analysing these other concepts. What we would be doing is attempting to explain the

180 Truth-Making and the Nature of Truth connection between truth and meaning in terms of the nature of meaning, and not truth, whilst also holding that the nature of meaning must be analysed in terms of truth. This looks too perilously circular to be an attractive explanation, and we are back in the vicinity of mere stipulation of the principle that meaning is analysed in terms of truth. Now, whilst the issues here are interesting, I do not wish to get too side-tracked by discussion of these connections. This is because, whatever the issues with primitivism’s ability to explain the connections between truth and other concepts, things are worse when we examine the notion of truth-making. I noted above that there is a burden on a viable theory of truth-making to explain at least the following two things: firstly, what kinds of things serve as truth-makers for a particular sort of truth-bearer; and, secondly, what thing would serve as the specific truth-maker for a specific truth-bearer. I suggested above that a theory of the nature of truth supplies explanations of these things. How can the primitivist explain them? She can’t. By the terms of the primitivist view, we do not have an informative property of being true to appeal to, nor do we have an analysis of the concept of truth to appeal to. This leaves nothing left: the only option left is mere stipulation. The primitivist can stipulate that, for physical truths, states of affairs are truth-makers, and that, due to what ‘my cup is on the table’ is about, the state of affairs of my cup being on the table would serve as the relevant truth-maker, but there is no explanatory grounding for this. To highlight how unsatisfactory this is, consider trying to justify the aforementioned account of the truth-making of physical truths to someone who thought that visual proofs, not states of affairs were the relevant kinds of things to serve as truthmakers for physical sentences. One may want to say, for example, that the first account is superior to the second, because physical truths are a matter of the relation between sentences and the world, not between sentences and sense-experience. This is indeed the sort of explanation I imagine a truth pluralist would give, but the primitivist cannot say this, as it involves an appeal to the nature of physical truth, whether conceived as a statement about the concept of truth or the property of being true. Any explanations of this sort which attempt to justify one account of truth-making over another are unavailable to the primitivist, as they all involve appeal to some facts about truth, which primitivism is unable to utilize. This marks a major problem for primitivism, as the nature of the position is such that it is unable to sustain the explanatory demands of a theory of truth-making. Asay suggests an alternative for the primitivist: turning to a metaphysical account of the nature of the phenomenon one is considering in order to discern the truth-makers for claims about that phenomenon. For instance, in the physical case, the thought would be that, once we establish the correct account of the metaphysics of the physical world, an account of the relevant truth-makers follows. Here is how the suggestion is put: For standard truthmaker theory, which truth-bearers instantiate truth depends on what’s going on in the world, ontologically speaking. The alethic properties aren’t themselves doing the truthmaking work; instead, we explain the presence of truth by way of the world’s ontology. (Asay 2018: 185)

truth-making and truth

181

The idea here seems to be that once we establish the correct metaphysics of x, we are in a position to identify the truth-makers of claims about x. In our case of the physical world, the story would be that once we establish that the nature of the physical world is independent of our sense experience, this suffices as an explanation for the superiority of the view that truth-makers for sentences about the physical world are states of affairs over the view that the truth-makers are items of sense experience. If we can do this, then we can identify the relevant truth-makers without appealing to the nature of truth. However, this route also fails, as there remains no explanation of why the underlying metaphysics of x has anything to do with the truth of claims about x unless we are assuming something about the nature of truth. Let me try to explain this using our two candidate explanations of the truth-makers for physical sentences. Suppose we do establish that the metaphysics of the physical world is such that the nature of the physical world is separate from our sense experience of it. One question to ask is what—if anything—this tells us about what would make sentences true. For suppose someone accepts this metaphysical picture, yet denies that this has anything to do with what makes sentences true. This might be because they think that the nature of the physical world is unknowable, and the truth-makers for sentences must be things that we are capable of knowing. On this view, the truth-makers for physical claims may well be items of sense-experience, as these are things that we are capable of knowing whether or not they obtain. I am not suggesting that this is an attractive view, but what is interesting is how we go about showing it to be mistaken. One natural thing to say is that it is mistaken because it involves a mistaken view of physical truth, which would involve claiming that physical truth is a matter of a relation between sentences and non-sentential reality, and not reliant on items of sense-experience. This would be a claim in part about the metaphysics of physical reality, but it would also be a claim about truth, involving a bridge principle from the claim about the nature of physical reality to the subsequent nature of truth. If this is the case, though, the primitivist cannot use this sort of explanation, as they do not think that there is any informative nature to physical truth. This means that they are once again left unable to warrant the claim that the truth-makers for physical claims are states of affairs as opposed to items of sense experience. A metaphysical account of the nature of the physical world takes us part of the way, but we can only use this metaphysical picture to yield a view about truth-makers if we combine it with a theory of truth. However, once again, a theory of truth is not something that the primitivist can appeal to, so we are left with no means to justify the choice of truth-makers in the theory. Finally, we can also note independently of this that the arguments of Chapters 4, 5, and 6 suggest that it is mistaken to think that we can establish the metaphysics of a domain without making reference to truth. There we showed that establishing the sort of existence objects and properties have in a domain is intimately related to the way we understand truth (or a truth-determining property) in a domain. This is because the kinds of entities that anti-realists typically posit depend on the truth of certain

182 Truth-Making and the Nature of Truth sentences in order to exist. If these entities depend on these truths, then truth cannot be understood representationally, in contrast to domains where we take a more realist perspective. If this is the case then we cannot simply examine the metaphysics of a domain without taking a stance on what sort of truth-determining property is in play.

10.3.3 Primitivism and truth-making: the verdict The primitivist theory of truth cannot avail itself of truth-makers, and thus cannot use them to provide a rival explanation to pluralism as some primitivists claim. Not only does this leave truth-making friendly forms of primitivism inferior to pluralist views in terms of explanatory power, it also leaves the view itself in an unstable position: one can have primitivism or the theory of truth-making, but not both. In Chapters 9 and 10 I hope to have shown how pluralism can avoid the primitivist pincer movement, and, in the process, offer some useful resources to those interested in truth-maker theory. In this chapter I have argued that there is a serious problem for any view that attempts to give truth-makers a substantial theoretical role, yet deny that there is a substantial nature to truth itself. This shows that there is no middle ground: if you want to endorse a theory of truth-makers, you must admit that there is a substantive and articulable nature of truth. This clarifies the vexed relationship between theories of truth-making and theories of the nature of truth. Theories of truth-making do not make theories of the nature of truth obsolete; on the contrary, they demonstrate the vital importance of the nature of truth for the very viability of the idea that there are things—of any sort—that make sentences true. If we take this idea seriously, then we should take the project of investigating the metaphysics of truth seriously too.

Concluding Remarks I began this book with two main aims: first, to counter deflationary and primitivist movements to re-assert the importance of the metaphysics of truth; and, second, to develop a distinctively pluralist account of the metaphysics of truth. With regard to the first aim, I hope to have shown that there is no presumptive advantage to being a deflationist or a primitivist, and to have loosened the hold that these views tend to have on discussions of truth. Moreover, I have noted that successfully defending these views would require overcoming some significant explanatory challenges. In identifying these challenges, I hope to have demonstrated that the metaphysics of truth is a topic worthy of discussion, and that there are interesting and important issues to consider in studying it. With respect to the second aim, I constructed a pluralist view of the connections between language and the world, demonstrating the crucial role that truth plays in this. This yields a pluralist approach to truth, and also to being, which offers a variety of metaphysical explanations, enabling the exploration of key metaphysical issues in a variety of areas. I also hoped to show that considering social and institutional truth and ontology is a key part of understanding the relationship between human thought and reality. Finally, I do not of course pretend to have conclusively settled the issues discussed, but no piece of philosophy worth its salt should do so. I will be satisfied if I have provided enough of interest to stimulate further discussion of this fascinating and elusive subject.

Bibliography Alexander, S. 1920. Space, Time, and Deity. Vol. 2. London: Macmillan. Alston, W.P. 1996. A Realist Conception of Truth. Ithaca: Cornell University Press. Alston. W.P. 2002. Truth: Concept and Property. In R. Shantz (ed.), What is Truth? Berlin and New York: Walter de Gruyter: 11–26. Andreasen, R. 2000. Race: Biological Reality or Social Construct? Philosophy of Science 67 (Supplement): 653–66. Appiah, K.A. 1994. Race, Culture, Identity: Misunderstood Connections. The Tanner Lectures on Human Values. San Diego: University of California: 53–136. Appiah, K.A. 1995. The Uncompleted Argument: DuBois and the Illusion of Race. In L. Bell and D. Blumenfeld (eds), Overcoming Racism and Sexism. Lanham, MD: Rowman and Littlefield. Appiah, K.A. 2007. Does Truth Matter to Identity? In Jorge J.E. Gracia (ed.), Race or Ethnicity? On Black and Latino Identity. Ithaca: Cornell University Press. Armstrong, D.M. 1978. Universals and Scientific Realism., Cambridge: Cambridge University Press. Armstrong, D.M. 1986. In Defence of Structural Universals. Australasian Journal of Philosophy 64 (1): 85–8. Armstrong, D.M. 1997. A World of States of Affairs. Cambridge: Cambridge University Press. Armstrong, D.M. 2004. Truth and Truthmakers. Cambridge: Cambridge University Press. Asay, J. 2013. The Primitivist Theory of Truth. Cambridge: Cambridge University Press. Asay, J. 2018. Putting Pluralism in its Place. Philosophy and Phenomenological Research. Vol. 96, Issue 1: 175–191. Ayer, A.J. 1936. Language, Truth and Logic. London: Camelot Press. Austin, J. L., 1950. Truth. Reprinted in Philosophical Papers, 3rd ed., Oxford: Oxford University Press, 1979, 117–33. Bader, R. 2013. Towards a Hyperintensional Theory of Intrinsicality. Journal of Philosophy 110 (10): 525–63. Ball, B. 2017. Alethic Pluralism and the Role of Reference in the Metaphysics of Truth. Southern Journal of Philosophy 55 (1): 116–35. Bar-On, D. and Simmons, K. 2007. The Use of Force Against Deflationism. In D. Greimann and G. Siegwart (eds), Truth and Speech Acts: Studies in the Philosophy of Language. Routledge: 61–89. Barnard, R. and Ulatowski, J. 2013. Truth, Correspondence, and Gender. Review of Philosophy and Psychology 4 (4): 621–38. Barnes, E. 2014. Going Beyond the Fundamental: Feminism in Contemporary Metaphysics. Proceedings of the Aristotelian Society 114 (3): 335–51. Båve, A. 2009. Why is a Truth-predicate like a Pronoun? Philosophical Studies 145 (2): 297–310. Beall, J.C. 2000. On Mixed Inferences and Pluralism about Truth Predicates. The Philosophical Quarterly 50 (200): 380–2.

186 Bibliography Beall, J.C. 2009. Spandrels of Truth. Oxford: Oxford University Press. Beall, J.C. 2013. Deflated Truth Pluralism. In N.J.L.L. Pedersen and C.D. Wright (eds): Truth and Pluralism: Current Debates. New York: Oxford University Press. Beall, J.C. and Armour-Garb, B. (eds) 2005. Deflationary Truth. Open Court. Beaney, M. (ed.) 1997. The Frege Reader. Blackwell. Beebee, H. and Dodd, J. 2005. Truthmakers: the Contemporary Debate. Oxford: Clarendon Press. Berto, F. 2012. Existence as a Real Property. Synthèse Library, Springer. Blackburn, S. 1992. Morality and Thick Concepts II: Through Thick and Thin. Proceedings of the Aristotelian Society Suppl. Vol. 66: 285–99. Blackburn, S. 1993. Essays in Quasi-Realism. Oxford University Press. Blackburn, S. 1998. Ruling Passions. Oxford University Press. Blackburn, S. 2013. Deflationism, Pluralism, Expressivism, Pragmatism. In Nikolaj Jang Lee Linding Pedersen and Cory D. Wright (eds), Truth and Pluralism: Current Debates. Oxford University Press: 263–278. Blanshard, B. 1939. The Nature of Thought. London: George Allen and Unwin. Boghossian, P.A. 1990. The Status of Content. The Philosophical Review 99 (2); 157–84. Beauvoir, S. de. 1973. The Second Sex. New York: Vintage Books. Boyd, R. 1988. How to be a Moral Realist. In G. Sayre-McCord (ed.), Essays on Moral Realism. Ithaca: Cornell University Press: 81–228. Bradley, F. 1914. Essays on Truth and Reality. Oxford: Clarendon Press. Brandom, R. 1994. Making It Explicit. Cambridge, MA: Harvard University Press. Brogaard, B. 2016. Against Naturalism about Truth. In K.J. Clark (ed.) The Blackwell Companion to Naturalism, Oxford: Blackwell: 262–76. Brown, L. 1998. Innovation and Continuity: The Battle of Gods and Giants, Sophist 245–9. In J. Gentzler (ed.).Method in Ancient Philosophy. Oxford: Clarendon Press. Burge, T. 1986. Frege on Truth. In L. Haaparanta and J. Hintikka (eds), Frege Synthesized. D. Reidel: 97–154. Clapp, L. 2001. Disjunctive Properties: Multiple Realizations.  Journal of Philosophy  98 (3): 111–36. Cook, R.T. 2013. Paradoxes. Cambridge: Polity Press. Cotnoir, A.J. 2009. Generic Truth and Mixed Conjunctions: Some Alternatives. Analysis 69. (2): 473–9. Cotnoir, A.J. 2013a. Validity for Strong Pluralists. Philosophy and Phenomenological Research 86 (3): 563–79. Cotnoir, A.J. 2013b. Pluralism and Paradox. In N.J.L.L. Pedersen and C.D. Wright (eds), Truth and Pluralism: Current Debates. New York: Oxford University Press: 339–50. Damnjanovic, N. 2005. Deflationism and the Success Argument. The Philosophical Quarterly 55 (218): 53–67. Damnjanovic, N. 2010. New Wave Deflationism. In N. Pedersen and C.D. Wright (eds), New Waves in Truth. New York: Palgrave Macmillan: 45–58. David, M. 2005. Review of G. Vision, Veritas. Notre Dame Philosophical Reviews (10). David, M. 2013. Lynch’s Functionalist Theory of Truth. In N.J.L.L. Pedersen and C.D. Wright (eds): Truth and Pluralism: Current Debates. New York: Oxford University Press. David, M. 2015. The Correspondence Theory of Truth. The Stanford Encyclopedia of Philosophy (Fall 2015 edition), Edward N. Zalta (ed.), http://plato.stanford.edu/archives/fall2015/entries/ truth-correspondence.

Bibliography

187

Davidson, D. 1967. Truth and Meaning. Synthese 17 (3): 304–23. Davidson, D. 1999. The Folly of Trying to Define Truth. Journal of Philosophy 93 (6): 263–79. Devitt, M. 2001. The Metaphysics of Truth. In M.P. Lynch (ed.), The Nature of Truth, Cambridge, MA: MIT Press: 579–611. Dodd, J. 2013. Deflationism Trumps Pluralism! In Nikolaj Jang Lee Linding Pedersen and Cory D. Wright (eds), Truth and Pluralism: Current Debates. Oxford University Press. Dorr, C. 2005. What We Disagree About When We Disagree About Ontology. In M. E. Kalderon (ed.), Fictionalism in Metaphysics. Oxford: Oxford University Press: 234–86. Dorr, C. 2008. There Are No Abstract Objects. In T. Sider, J. Hawthorne and D.W. Zimmerman (eds), Contemporary Debates in Metaphysics. Oxford: Blackwell: 32–64. Dorr, C. and Hawthorne, J. 2013. Naturalness. In Karen Bennett and Dean Zimmerman (eds), Oxford Studies in Metaphysics 8. Oxford University Press. Dorsey, D. 2006. A Coherence Theory of Truth in Ethics. Philosophical Studies 127 (3): 493–523. Dummett, M. 1978. Truth. In his Truth and Other Enigmas. Oxford University Press. Dummett, M. 1991. The Logical Basis of Metaphysics. Harvard University Press. Dummett, M. 2004. Truth and the Past. Columbia University Press. Edwards, D. 2008. How to Solve the Problem of Mixed Conjunctions. Analysis 68 (2): 143–9. Edwards, D. 2009. Truth-conditions and the Nature of Truth: Re-solving Mixed Conjunctions. Analysis 69 (4): 684–8. Edwards, D. 2011. Simplifying Alethic Pluralism. The Southern Journal of Philosophy 49 (1): 28–48. Edwards, D. 2012. On Alethic Disjunctivism. dialectica 66 (1): 200–14. Edwards, D. 2013a. Truth as a Substantive Property. Australasian Journal of Philosophy 91 (2): 279–94. Edwards, D. 2013b. Naturalness, Representation, and the Metaphysics of Truth. European Journal of Philosophy 21 (3): 384–401. Edwards, D. 2014. Properties. Cambridge: Polity Press. Edwards, D. 2016. Truth as a Relational Property. Synthese. Edwards, D., Ferrari, F., and Lynch. M.P. 2016. Truth and Naturalism. In K.J. Clark (ed.) The Blackwell Companion to Naturalism. Oxford: Blackwell: 246–61. Eklund, M. 2006. Neo-Fregean Ontology. Philosophical Perspectives 20 (1): 95–121. Eklund, M. 2011. What are Thick Concepts? Canadian Journal of Philosophy 41 (1): 25–49. Elstein, D. and Hurka, T. 2009. From Thick to Thin: Two Moral Reduction Plans. Canadian Journal of Philosophy 39 (4): 515–36. Engel, P. 2002. Truth. Acumen. Etchemendy, J. 1988. Tarski on Truth and Logical Consequence. Journal of Symbolic Logic 53 (1): 51–79. Field, H. 1972. Tarski’s Theory of Truth. The Journal of Philosophy 69 (13): 347–75. Field, H. 1980. Science Without Numbers. Oxford: Blackwell. Field, H. 1994. Deflationist Views of Meaning and Content. Mind 103 (411): 247–85. Field, H. 2008. Saving Truth from Paradox. Oxford: Oxford University Press. Fine, K. 2009. The Question of Ontology. In David J. Chalmers, David Manley and Ryan Wasserman (eds), Metametaphysics. Oxford: Oxford University Press: 157–76. Fine, K. 2012. The Pure Logic of Ground. Review of Symbolic Logic 5 (1):1–25. Fodor, J. 2003. Hume Variations. Oxford: Oxford University Press.

188 Bibliography Francescotti, R. 1999. How to Define Intrinsic Properties. Noûs 33: 590–609. Francescotti, R. 2014. Intrinsic/Extrinsic: A Relational Account Defended. In R. Francescotti (ed.) A Companion to Intrinsic Properties. De Gruyter 175–98. Frege, G. 1918. Der Gedanke. Eine Logische Untersuchung. In Beiträge zur Philosophie des deutschen Idealismus, I (1918–1919): 58–77. Translated as ‘Thoughts’, by P. Geach and R. Stoothoff, in B. McGuinness (ed.), Collected Papers on Mathematics, Logic, and Philosophy, Oxford: Blackwell 1984: 351–72. Fumerton, R. 2002. Realism and the Correspondence Theory of Truth. Rowman & Littlefield. Fumerton, R. 2013. Alethic Pluralism and the Correspondence Theory of Truth. In Nikolaj Jang Lee Linding Pedersen and Cory D. Wright (eds), Truth and Pluralism: Current Debates. Oxford University Press. Gibbard, A. 1992. Morality and Thick Concepts I: Thick Concepts and Warrant for Feelings. Proceedings of the Aristotelian Society, Suppl. Vol. 66: 267–83. Glanzberg, M. 2004. Against Truth-Value Gaps. In Liars and Heaps: New Essays on Parado, (ed. J.C. Beall). Oxford University Press: 151–94. Glasgow, J. 2017. A Metatheory of Race. In Naomi Zack (ed.), The Oxford Handbook of Philosophy and Race. Oxford: Oxford University Press. Grover, D. 1990. Truth and Language-world Connections. Journal of Philosophy 87 (12): 671–87. Grover, D. 1992. A Prosentential Theory of Truth. Princeton: Princeton University Press. Hacking, I. 1999. The Social Construction of What? Harvard University Press. Hale, B. 1994. Singular terms. In Brian McGuinness and Gianluigi Oliveri (eds), The Philosophy of Michael Dummett. Kluwer Academic Publishers: 17–44. Hale, B. and Wright, C.J.G. 2001. The Reason’s Proper Study. Oxford: Oxford University Press. Hale, B. and Wright, C.J.G. 2005. Logicism in the 21st Century. In Stewart Shapiro (ed.), The Oxford Handbook of Philosophy of Mathematics and Logic. Oxford: Oxford University Press. Hale, B. and Wright, C. 2009. The Metaontology of Abstraction. In D. Chalmers, D. Manley and R. Wasserman (eds), Metametaphysics. Oxford: Oxford University Press: 178–212. Haslanger, S. 2000. Gender and Race: (What) are They? (What) do We Want Them to Be? Noûs 34 (1): 31–55. Haslanger, S. 2012. Resisting Reality: Social Construction and Social Critique. Oxford University Press. Hawley, K. 2007. Neo-Fregeanism and Quantifier Variance. Aristotelian Society Supplementary Volume 81: 233–49. Heck, R. 1997. Tarski, Truth, and Semantics. Philosophical Review 106 (4): 533–54. Hill, C. 2002. Thought and World. Cambridge: Cambridge University Press. Hirsch, E. 1993. Dividing Reality. Oxford University Press. Hirsch, E. 2009. Ontology and Alternative Languages. In D. Chalmers, D. Manley and R. Wasserman (eds), Metametaphysics: New Essays on the Foundations of Ontology. Oxford University Press. Horgan, T. and Timmons, M. 1991. New Wave Moral Realism Meets Moral Twin Earth. Journal of Philosophical Research 16: 447–65. Horwich, P. 1998. Truth. Oxford: Oxford University Press. Horwich, P. 2001. A Defense of Minimalism. In M.P. Lynch (ed.), The Nature of Truth: Classic and Contemporary Perspectives. Cambridge, MA: MIT Press.

Bibliography

189

Horwich, P. 2004. From a Deflationary Point of View. Oxford University Press. Horwich, P. 2010. Truth-Meaning-Reality. Oxford University Press. Hume, D. 1754. Of National Characters. In Essays: Moral, Political, and Literary. Indianapolis, IN: Liberty Fund, Inc. Jackson, F. 1977. A Causal Theory of Counterfactuals. Australasian Journal of Philosophy 55: 3–21. Jackson, F. and Pettit, P. 1990. Causation in the Philosophy of Mind. Philosophy and Phenomenological Research 50 (Supplement): 195–214. James, W. 1897. The Will to Believe. Philosophical Review 6: 88. James, W. 1907. Pragmatism’s Conception of Truth. The Journal of Philosophy, Psychology and Scientific Methods. 4 (6): (14 March): 141–55. Jenkins, C.S. 2011. Is Metaphysical Dependence Irreflexive? The Monist 94 (2): 267–76. Joyce, R. 2001. The Myth of Morality. Cambridge University Press. Kalderon M. 2007. Moral Fictionalism. Oxford: Clarendon Press. Kant, I. 2007. Of the Different Races of Human Beings (1775). In Anthropology, History, and Education. Cambridge University Press. Kim, J. 1993. Supervenience and Mind. Cambridge University Press. King, J., Soames, S. and Speaks, J. 2014. New Thinking About Propositions. Oxford University Press. Kitcher, P. 2002. On the Explanatory Role of Correspondence Truth. Philosophy and Phenomenological Research 64. (2): 346–64. Kitcher, P. 2007, Does ‘Race’ Have a Future? Philosophy and Public Affairs 35(4): 293–317. Kölbel, M. 2013. Should We Be Pluralists about Truth? In Nikolaj Jang Lee Linding Pedersen and Cory D. Wright (eds), Truth and Pluralism: Current Debates. Oxford University Press. Kripke, S. 1980. Naming and Necessity. Harvard University Press. Künne, W. 2003. Conceptions of Truth. Oxford: Oxford University Press. Lewis, D., 1973, Counterfactuals. Oxford: Basil Blackwell. Lewis, D. 1983. New Work for a Theory of Universals. Australasian Journal of Philosophy 61 (4): 343–77. Lewis, D. 1986a. On the Plurality of Worlds. Oxford: Blackwell. Lewis, D. 1986b. Against Structural Universals. Australasian Journal of Philosophy 64 (1): 25–46. Lewis, D. 2001. Forget About the ‘Correspondence Theory of Truth’. Analysis 61 (272): 275–80. Locke, J. 1700. An Essay Concerning Human Understanding. Oxford University Press. Lowe, E.J. 2008. An Essentialist Approach to Truth-Making. In E.J. Lowe and A. Rami (eds), Truth and Truth-Making. Acumen. Lynch, M.P. 2001. A Functionalist Theory of Truth. In M.P. Lynch (ed.), The Nature of Truth: Classic and Contemporary Perspectives. Cambridge, MA: MIT Press: 723–49. Lynch, M.P. 2004. Truth and Multiple Realizibility. Australasian Journal of Philosophy 82 (3): 384–408. Lynch, M.P. 2006. ReWrighting Pluralism. The Monist. 89 (1): 63–84. Lynch, M.P. 2009. Truth as One and Many. Oxford University Press. Lynch, M.P. 2013a. Three Questions for Truth Pluralism. In N.J.L.L. Pedersen and C.D. Wright (eds): Truth and Pluralism: Current Debates. New York: Oxford University Press.

190

Bibliography

Lynch, M.P. 2013b. Expressivism and Plural Truth. Philosophical Studies 163 (2): 385–401. MacBride, F. 2003. Speaking with Shadows: A Study of Neo-logicism. British Journal for the Philosophy of Science 54 (1): 103–63. MacBride, F. 2013. For Keeping Truth in Truthmaking. Analysis 73: 686–95. McDaniel, K. 2009. Ways of Being. In David J. Chalmers, David Manley and Ryan Wasserman (eds), Metametaphysics, Oxford: Oxford University Press: 290–319. McDaniel, K. 2010a. A Return to the Analogy of Being. Philosophy and Phenomenological Research 81 (3): 688–717. McDaniel, K. 2010b. Being and Almost Nothingness. Noûs 44 (4): 628–49. McDaniel, K. 2017. The Fragmentation of Being. Oxford: Oxford University Press. McDermott, M. 1996. On the Truth Conditions of Certain ‘If ’-Sentences. The Philosophical Review 105: 1–37. McDermott, M. 1999. Counterfactuals and Access Points. Mind 108: 291–334. McGinn, C. 2000. Logical Properties. Oxford: Clarendon Press. Mackie, J.L. 1977. Ethics: Inventing Right and Wrong. Penguin. Mallon, R. 2006. Race: Normative, Not Metaphysical or Semantic. Ethics 116 (3): 525–51. Merricks, T. 2001. Objects and Persons. Oxford University Press. Merricks, T. 2007. Truth and Ontology. Oxford University Press. Merricks, T. 2015. Propositions. Oxford University Press. Mikkola, M. 2009. Gender Concepts and Intuitions. Canadian Journal of Philosophy 9: 559–83. Mikkola, M. 2011. Ontological Commitments, Sex and Gender. In C. Witt (ed.), Feminist Metaphysics, Dordrecht: Springer. Millikan, R. 2000. On Clear and Confused Ideas. Cambridge: Cambridge University Press. Moore, G.E. 1903. Principia Ethica. Dover Publications. Naess, A. 1938. ‘Truth’ as Conceived by Those Who Are Not Professional Philosophers. Oslo: Det Norske Videnskaps-Akademi i Oslo, Skrifter. Peacocke, C. 1992. A Study of Concepts. MIT Press. Pedersen, N. 2006. What Can the Problem of Mixed Inferences Teach us About Alethic Pluralism? The Monist 89 .(1): 102–17. Pedersen, N. 2010. Stabilizing Alethic Pluralism. The Philosophical Quarterly 60 (238): 92–108. Pedersen, N. 2014. Pluralism x 3: Truth, Logic, Metaphysics. Erkenntnis 79 (2): 259–77. Pedersen, N.J.L.L. and Wright, C.D. 2013a. Pluralism about Truth as Alethic Disjunctivism. In N.J.L.L. Pedersen and C.D. Wright (eds), Truth and Pluralism: Current Debates. New York: Oxford University Press. Pedersen, N.J.L.L. and Wright, C.D. 2013b. Pluralist Theories of Truth. The Stanford Encyclopedia of Philosophy Edward N. Zalta (ed.). http://plato.stanford.edu/archives/ spr2013/entries/truth-pluralist/. Peirce, C.S. 1878. How to Make Our Ideas Clear. Popular Science Monthly 12 (Jan.): 286–302. Plantinga, A. 1974. The Nature of Necessity. Oxford: Clarendon Press. Price, H. 2003. Truth as Convenient Friction. Journal of Philosophy 100: 167–90. Price, H. 2011. Naturalism Without Mirrors. Oxford: Oxford University Press. Putnam, H. 1975. The Meaning of ‘Meaning’. Minnesota Studies in the Philosophy of Science 7: 131–93.

Bibliography

191

Quine. W.V.O. 1948. On What There Is. Review of Metaphysics 2: 21–48. Quine, W.V.O. 1960. Word and Object. Cambridge, MA: MIT Press. Quine, W.V.O. 1990. The Pursuit of Truth. Cambridge: Harvard University Press. Ramsey, F.P. 1927/2001. The Nature of Truth. Episteme 16: 6–16, 1927. Reprinted in M.P. Lynch (ed.), The Nature of Truth. MIT Press, 2001. Rasmussen, J. 2014. Defending the Correspondence Theory of Truth. Cambridge University Press. Ricketts, T. 1986. Objectivity and Objecthood: Frege’s Metaphysics of Judgment. In J. Hintikka and L. Haaparanta (eds), Frege Synthesized. Kluwer Academic Publishers: 65–95. Root, M. 2000. How We Divide the World. Philosophy of Science 67 (Supplement): 628–39. Rosen, G. 1990. Modal fictionalism. Mind 99 (395): 327–54. Rosen, G. 2010. Metaphysical Dependence: Grounding and Reduction. In Bob Hale and Aviv Hoffmann (eds), Modality: Metaphysics, Logic, and Epistemology. Oxford University Press: 109–36. Russell, B. 1903. The Principles of Mathematics. London and New York: W.W. Norton. Russell, B. 1904. Meinong’s Theory of Complexes and Assumptions (III). Mind 13 (2): 509–24. Russell, B. 1905. On Denoting. Mind 114 (456): 873–87. Russell, B. 1912. The Problems of Philosophy, The Home University Library, Williams & Norgate. Russell, B. 1956. The Philosophy of Logical Atomism. In his Logic and Knowledge: Essays 1901–1950, 175–283. London: George Allen & Unwin. Ryle, G. 1949. The Concept of Mind. Hutchinson & Co. Sainsbury, R.M. 2009. Fiction and Fictionalism. Routledge. Salmon, N. 1987. Existence. Philosophical Perspectives 1: 49–108. Schaffer, J. 2004. Two Conceptions of Sparse Properties, Pacific Philosophical Quarterly 85 (1): 92–102. Schaffer, J. 2009. On What Grounds What. In David Manley, David J. Chalmers and Ryan Wasserman (eds), Metametaphysics: New Essays on the Foundations of Ontology. Oxford University Press: 347–83. Scharp, K. 2013. Replacing Truth. Oxford: Oxford University Press. Schiffer, S. 1996. Language-Created Language-Independent Entities. Philosophical Topics 24 (1): 149–67. Schiffer, S. 2003. Things We Mean. Oxford University Press. Schiffer, S. Forthcoming. Amie Thomasson’s Easy Approach to Ontology. Philosophy and Phenomenological Research. Searle, J. 1983. Intentionality, Cambridge: Cambridge University Press. Searle, J. 1995. The Construction of Social Reality. New York: The Free Press. Searle, J. 2010. Making the Social World. Oxford: Oxford University Press. Shapiro, S. 2000. Thinking about Mathematics: The Philosophy of Mathematics. Oxford: Oxford University Press. Shapiro, S. 2009. Review of Truth as One and Many. Notre Dame Philosophical Reviews. Sher, G. 1999. What is Tarski’s Theory of Truth? Topoi 18: 149–66. Sher, G. 2004. In Search of a Substantive Theory of Truth. Journal of Philosophy 101: 5–36. Sher, G. 2005. Functional Pluralism. Philosophical Books 46: 311–30.

192

Bibliography

Sher, G. 2015. Truth as Composite Correspondence. In T. Achourioti, H. Galinon, J. Martínez Fernández, and K. Fujimoto (eds.), Unifying the Philosophy of Truth. Springer Verlag: 191–210. Sher, G. 2016. Epistemic Friction. Oxford: Oxford University Press. Sider, T. 2007. Neo-Fregeanism and Quantifier Variance. Aristotelian Society Supplementary Volume 81: 201–32. Sider, T. 2009. Ontological Realism. In D. Chalmers, D. Manley and R. Wasserman (eds), Metametaphysics. Oxford: Oxford University Press: 384–423. Sider, T. 2011. Writing the Book of the World. Oxford: Oxford University Press. Smith, M. 1994. Minimalism, Truth-Aptitude and Belief. Analysis 54 (1): 21–6. Smith, M. 2013. On the Nature and Significance of the Distinction between Thick and Thin Ethical Concepts. In S. Kirchin (ed.), Thick Concepts. Oxford University Press. Soames, S. 1984. What is a Theory of Truth? Journal of Philosophy 81 (8): 411–29. Soames, S. 2003. Understanding Deflationism. Philosophical Perspectives. 17: 369–83. Sosa, E. 2001. Epistemology and Primitive Truth. In M.P. Lynch (ed.), The Nature of Truth. Cambridge, MA: MIT Press: 641–62. Stalnaker, R. 1984. Inquiry, Cambridge, MA: MIT Press. Strawson, P.F. 1950a. On Referring. Mind, New Series, 59 (235): 320–44. Strawson, P.F. 1950b. Truth. Proceedings of the Aristotelian Society 24: 129–56. Strawson, P.F. 1998. Intellectual Autobiography. In L. Hahn (ed.), The Philosophy of P.F. Strawson. Chicago: Open Court: 1–21. Strollo, A. 2016. A Simple Notion of Validity for Alethic Pluralism. Synthese. Tappolet, C. 1997. Mixed Inferences: a Problem for Pluralism About Truth Predicates. Analysis 57: 209–10. Tappolet, C. 2000. Truth, Pluralism and Many-Valued Logics. The Philosophical Quarterly 50 (200): 382–3. Tarski, A. 1944. The Semantic Conception of Truth and the Foundations of Semantics. Philosophy and Phenomenological Research 4: 341–76. Tennant, N. 2003. Bob Hale and Crispin Wright. The Reason’s Proper Study: Essays Towards a Neo-Fregean Philosophy of Mathematics. Philosophia Mathematica 11 (2): 226–40. Thomasson, A. 1999. Fiction and Metaphysics. Cambridge: Cambridge University Press. Thomasson, A. 2003a. Realism and Human Kinds. Philosophy and Phenomenological Research 67 (3): 580–609. Thomasson, A. 2003b. Foundations for a Social Ontology. ProtoSociology 18: 269–90. Thomasson, A. 2007. Ordinary Objects. Oxford: Oxford University Press. Thomasson, A. 2009. Social Entities. In Robin Le Poidevin (ed.), The Routledge Companion to Metaphysics. Routledge. Thomasson, A. 2014. Deflationism in Semantics and Metaphysics. In A. Burgess and B.  Sherman (eds) Metasemantics: New Essays on the Foundations of Meaning. Oxford University Press. Thomasson, A. 2015. Ontology Made Easy. Oxford: Oxford University Press. Turner, J. 2010. Ontological Pluralism. The Journal of Philosophy 107 (1): 5–34. Turner, J. 2012. Logic and Ontological Pluralism. Journal of Philosophical Logic 41 (2): 419–48. Van Inwagen, P. 1990. Material Beings. Cornell University Press. Van Inwagen P. 1998. Meta-ontology. Erkenntnis 48 (2/3): 233–50. Vision, G. 2003. Lest we Forget ‘The Correspondence Theory of Truth’. Analysis 63 (278): 136–42.

Bibliography

193

Vision, G. 2004. Veritas: The Correspondence Theory and Its Critics. Cambridge, MA: The MIT Press. Weatherson, B. 2003. What Good are Counterexamples? Philosophical Studies 115 (1): 1–31. Weatherson, B. and Marshall, D. 2012. Intrinsic vs. Extrinsic Properties. The Stanford Encyclopedia of Philosophy (Fall 2014 edition). Edward N. Zalta (ed.), http://plato.stanford. edu/archives/fall2014/entries/intrinsic-extrinsic/. Williams, B. 1985. Ethics and the Limits of Philosophy. Routledge: Routledge Classics Edition. Williams, M. 1999. Meaning and Deflationary Truth. Journal of Philosophy 96 (11): 545–64. Williamson, T. 2000. Knowledge and its Limits. Oxford: Clarendon Press. Williamson, T. 2007. The Philosophy of Philosophy. Oxford: Blackwell. Wittgenstein, L. 1922. Tractatus Logico-Philosophicus. London: Routledge & Kegan- Paul. Wittgenstein, L. 1953. Philosophical Investigations. Wiley-Blackwell. Wrenn, C. 2015. Truth. Polity Press. Wright, C.D. 2016. Truth, Explanation, Minimalism. Synthese. DOI: 10.1007/s11229-016-1129-6. Wright, C.D. 2017. Pluralism and the Liar. In B. Armour-Garb (ed.), Reflections on the Liar. Oxford University Press. Wright, C.J.G. 1983, Frege’s Conception of Numbers as Objects. Vol. 2. Aberdeen: Aberdeen University Press. Wright, C.J.G. 1992. Truth and Objectivity. Cambridge, MA: Harvard University Press. Wright, C.J.G. 2001. Minimalism, Deflationism, Pragmatism, Pluralism. In M.P. Lynch (ed.), The Nature of Truth. Cambridge, MA: MIT Press: 751–87. Wright, C.J.G. 2003. Truth: A Traditional Debate Reviewed. In his Saving the Differences: Essays on Themes from ‘Truth and Objectivity’. Cambridge, MA: Harvard University Press: 241–87. Wright, C.J.G. 2013. A Plurality of Pluralisms. In N.J.L.L. Pedersen and C.D. Wright (eds), Truth and Pluralism: Current Debates. New York: Oxford University Press. Wyatt, J. 2013. Domains, Plural Truth, and Mixed Atomic Propositions. Philosophical Studies 166 (1): 225–36. Wyatt, J. 2016. The Many (Yet Few) Faces of Deflationism. The Philosophical Quarterly 66 (263): 362–82. Zack, N. 2002. Philosophy of Science and Race. New York: Routledge. Zack, N. 2006. Thinking about Race. Thomson Wadsworth.

Index abstract/concrete distinction 106–7, 112–14 alethic pluralism 3n 8 see also truth pluralism Alexander, S. 107n 2 Alston, W.P. 18n 10, 29n 14 Andreasen, R. 73n 18 Appiah, K.A. 64, 73–4, 102n 16 Aristotle 42 Armour-Garb, B. 41n 1, 57n 10, 58n 13 Armstrong, D.M. 5n 9, 15n 7, 34n 24, 35nn 27 and 28, 37–8, 89, 96n 9, 107n 2, 159, 160, 171n 17 Asay, J. 172–4, 176, 178–80 Ayer, A.J. 2n 5, 9–10 Austin, J. L. 1n 1, 161 Bader, R. 162 Ball, B. 111n 8 Bar-On, D. 25n 5, 52 Barnard, R. 18n 10 Barnes, E. 5 Båve, A. 14 Beall, J. C. 41n 1, 58n 13, 133n 15, 138n 22 Beauvoir, S. de 65 Beebee, H. 160n 1 being see existence Berkeley, G. 107 Berto, F. 154n 16 Blackburn, S. 10n 6, 11, 52n 8, 61–3, 69n 13, 71n 16, 80n 26, 98n 14 Blanshard, B. 1n 2 Boghossian, P.A. 57n 11 Boyd, R. 71n 16 Bradley, F. 1n 2 Brandom, R. 12–14, 52 Brogaard, B. 38n 33 Brown, L. 107n 2 Burge, T. 16n 8 causal-explanatory role 35–9, 56, 69–70 causal theory of reference 90–2, 99–100, 102–3, 178 Clapp, L. 151n 12 coherence theory of truth 1–2, 26, 83, 91, 126n 6, 161n 3, 163, 165–7, 170–1, 178 Cook, R.T. 138n 22 concepts 6, 18–20, 21, 25, 31, 41, 44, 47, 53, 58, 62, 98–9, 165, 178–80 correspondence theory of truth 1–2, 21, 24, 26, 42, 51, 59, 83, 89–92, 98–104, 106–7, 109, 134n 17, 136n 21, 160–5, 173, 175, 179

Cotnoir, A.J. 133n 15, 135n 19, 138n 22, 150nn 10 and 11 counterfactuals 51, 160–1, 166–7 Damnjanovic, N. 22, 28n 9, 29, 31–2, 36–7n 31 David, M. 58, 78n 24, 101n 15, 131, 162n 8 Davidson, D. 2, 28n 11, 34, 178–9 deflationism about existence 115–20 about truth 2, 4, 6–59, 95–6, 115–20, 176–8 and methodology 26n 6, 41–2, 57n 10, 58–9 and pluralism 115–20 and truth-making 176–8 arguments against 41–59, 95–6, 114–20 characterizing 22–40, 57–9 descriptivist theory of reference 90–1, 102–3 determination pluralism about existence 153–6 about truth 122–38 Devitt, M. 26n 7 disquotation 14, 23–8, 52 disjunctivism about existence 151–2 about truth 128–30, 151 Dodd, J. 41n 1, 58n 13, 160n 1 domains 3–4, 60, 66, 71, 75, 77–83, 86–8, 91–2, 95, 98, 100–5, 107, 109, 115, 125, 127, 130–40, 154, 156, 158, 170 Dorr, C. 5n 10, 38n 32 Dorsey, D. 91 Dummett, M. 19, 122n 1, 160n 2, 163n 9, 167, 170–1, 173, 182 Eklund, M. 80n 26, 108n 6 eleatic principle 107, 109–10, 112–13 Elstein, D. 80nn 26 and 27 Engel, P. 58, 162n 4 error theory 67, 102, 165 Etchemendy, J. 44n 3 existence 3, 5, 38n 33, 53n 9, 55, 97–8, 100, 103, 105–21, 122, 138, 142–58, 160–1, 164, 167–70, 181–2 Ferrari, F. 38n 33 Field, H. 2n 5, 22n 1, 23, 25n 4, 41n 1, 44n 3, 58n 13, 89–90, 97n 12, 138n 22 Fine, K. 143, 144, 162n 6 Fodor, J. 19n 11 Francescotti, R. 162n 4 Frege, G. 2n 6, 11, 16–17, 47, 108, 178n 5

196 index Fumerton, R. 89, 101n 15 functionalism about truth 128–31, 151 gender predicates 63–5 properties 72–5 truths about 93–4, 97–8, 102n 17, 103n 18 Gibbard, A. 80n 26 Glanzberg, M. 122n 1 Glasgow, J. 102n 16 grounding 162n 6 Grover, D. 2n 5, 12–14, 36, 89n 3 Hacking, I. 75nn 20 and 21, 103 Hale, B. 76n 22, 86n 1, 88n 2, 108–9, 120 Haslanger, S. 64n 4, 65, 68n 11, 93–4, 101, 102n 17, 103n 18 Hawley, K. 108n 6 Heck, R. 44n 3 Hill, C. 22n 1, 23n 2, 33n 22 Hirsch, E. 145n 3 Horgan, T. 97n 10 Horwich, P. 2n 5, 18n 10, 19, 22n 1, 23, 25, 28n 9, 30n 17, 33, 46–56, 176–7 Hume, D. 63n 2 Hurka, T. 80nn 26 and 27 inflationism about truth 26, 28–9, 33, 36–7, 39–40, 53, 59 institutional objects 76–7, 108, 113 predicates 62–6, 81 properties 72, 75, 97, 98n 14 truth 3–5, 92–4, 101–3, 106, 123, 183 Jackson, F. 37n 31, 166n 13 James, W. 1n 3, 17, 91n 4 Jenkins, C.S.I. 162n 6 Joyce, R. 67n 9 Kalderon, M. 97n 12 Kant, I. 63n 2 Kim, J. 107n 2 King, J. 21n 12 Kitcher, P. 28n 12, 73n 18 Kölbel, M. 127n 10 Kripke, S. 90, 102 Künne, W. 9, 11, 14, 16n 8, 22n 1, 23n 2, 30n 17, 32n 21, 33n 22, 177n 4 Lewis, D. 5n 9, 8, 34n 25, 35, 37–9, 145, 160n 2, 165, 166n 13, 171n 17, 176n 2 liar paradox 9n 4, 138n 22 Locke, J. 30n 16 looping kinds 75n 20, 103 Lowe, E.J. 160n 2

Lynch, M.P. 3n 8, 18n 10, 19, 28n 10, 29–30, 38n 33, 42, 46–7, 49–50, 79n 25, 80n 26, 83, 90, 98n 14, 99, 106, 111n 8, 123n 4, 127–9, 130n 14, 131, 133, 136n 20 MacBride, F. 108n 5, 176n 2 McDaniel, K. 105n 1, 113, 142, 144n 4, 150–2, 153n 14 McDermott, M. 166n 13 McGinn, C. 32n 21 Mackie, J.L. 67 Mallon, R. 65n 20 Marshall, D. 164 manifestation functionalism 128–9, 131 metaphysics, conceptions of 1–7 methodological deflationism 26n 6, 41–2, 57n 10, 58–9 Merricks, T. 76, 145, 146n 6, 154–5, 160–71 Mikkola, M. 93n 5 Millikan, R. 19n 11 minimalism about truth 23, 52, 55, 176n 3, 177 see also deflationism about truth mixed atomics, problem of 78–81 mixed compounds, problem of for existence 148–150 for truth 132–6, 166 mixed inferences, problem of for existence 147–8, 50 for truth 132–3 modal truths 165–7 monism 95–8, 107, 139, 156–7 Moore, G.E. 97n 10, 161 moral predicates 5, 10–11, 32, 55, 61–3, 65–9, 71, 75–81, 97, 99, 101, 106, 131, 133, 136, 154 Naess, A. 18 negative existentials 160–1, 165, 168–9 neo-fregean principle 108–9, 111–12, 116 objects abundant and sparse 76–7, 109–15, 149–50, 154–5 and singular terms 76–7 ontological pluralism and quantifiers 142–6 formulations of 146–56 motivations for 105–14 objections to 114, 147–50 past, truths about 160–1, 165, 167–8 Peacocke, C. 19n 11 Pedersen, N.J.L.L. 79n 25, 108n 4, 127n 11, 128, 130n 14, 133n 15, 150 Peirce, C.S. 1n 4, 91n 4, 92, 111, 163n 9 performative theory of truth 6, 8, 10–11, 22, 25, 67, 69n 13

index Pettit, P. 37n 31 Plantinga, A. 165n 10 Plato 16–17, 107n 2 platitudes 18–20, 27, 46–7, 51–2, 61–2, 125n 5, 131, 153 pluralism about existence 3, 4, 105–21, 142–58 about objects 107–14, 154–6 about properties 34–6, 66–70, 76–7 about truth 3, 83–141, 159–82 see also truth pluralism pragmatism about truth 2, 59, 91–2, 173 global pragmatism 96 Price, H. 28n 8, 55, 58 predicate pluralism about truth 127 about existence 147 predicates and properties 15–16, 66–75 functional roles of 60–6 responsive and generative 60, 67–72, 87–8, 91, 94, 96–8, 125, 139–40, 156–7 primitivism about truth 2, 4, 34, 159–82 properties abundant and sparse 4, 22, 34–40, 54–5, 57n 12, 59, 60, 68–70, 74–7, 81, 83–4, 86–8, 96, 98, 100, 104, 125, 138–40 and fundamentality 37–9 and naturalness 38–9 and predicates 15–16, 66–75 conceptions of 34–6 extrinsic and intrinsic 38–9, 162–4, 170–1, 178 relational and non-relational 161–4, 170–1 substantive and insubstantive 13, 15, 21, 22–40, 41, 53–9 propositions 9–11, 18–21, 22–3, 27–8, 44–50 prosentential theory of truth 12–14 Putnam, H. 90, 163n 9 quantifier variance 145–6 quietism 55, 58–9 Quine. W.V.O. 9n 3, 24, 114, 142–4, 153 race predicates 63–5 properties 72–5 truths about 93–4, 97–8, 102n 17, 103n 18 Ramsey, F.P. 42, 45–9, 51 Rasmussen, J. 89, 162n 4 redundancy theory of truth 9–10 reference 16–18, 20, 36n 33, 44–6, 48–9, 52–6, 89–90, 99–103, 108, 110–11, 114–15, 118, 177 reflective equilibrium 20

197

representation 55, 58–9, 83, 87–91, 99–103, 110–12, 120n 12, 125–6, 139, 156, 182 Ricketts, T. 16n 8 Root, M. 75n 20 Rosen, G. 97n 12, 162n 6, 165n 10 Russell, B. 1n 1, 16n 9, 34, 89–90, 134n 18, 161, 166n 11 Ryle, G. 147n 8 Sainsbury, R.M. 97n 12 Salmon, N. 144, 153 satisfaction 43–4, 48, 53–4, 61, 84–8, 99, 101 Schaffer, J. 5n 10, 38, 97n 11, 162n 6 Scharp, K. 9n 4, 138n 22 Schiffer, S. 86n 1, 88n 2 scope problem 83, 106–9 Searle, J. 62–3, 72, 90 sentences atomic 8, 20–1, 77–81 compound 132–8 Shapiro, S. 98–9, 108n 4 Sher, G. 3n 8, 78n 24, 83, 89n 3, 99–104, 106 Sider, T. 5n 10, 38nn 32 and 33, 108n 6, 153n 13, 156n 18 Simmons, K. 25n 5, 52 Smith, M. 62–3, 80, 98n 14 Soames, S. 21n 12, 22n 1, 44n 3 social objects 76–7, 108, 113 predicates 62–6, 81 properties 72, 75, 97, 98n 14 truth 3–5, 92–4, 101–3, 106, 123, 183 Sosa, E. 34, 178n 5 Speaks, J. 21n 12 Stalnaker, R. 165n 10 Strawson, P.F. 2n 5, 10–11, 89 Strollo, A. 133n 15 strong pluralism about existence 142, 146–9, 170–1 about truth 127–30, 132–40 superassertibility 26, 91–4, 96, 107, 109, 111, 116, 119–20, 126–9, 131, 139, 146n 7, 157, 163n 9, 168, 171, 173 Tappolet, C. 132–6 Tarski, A. 1n 1, 24, 42–5, 47–9, 85–6, 89, 101, 134n 18 Tennant, N. 108n 5 thick concepts 79–81 Thomasson, A. 53n 9, 54–5, 58, 72, 75n 21, 88n 2, 97n 12, 102, 113–21 Timmons, M. 97n 10 truth and assertion 9–11, 13, 27, 47, 52–4, 120, 122n 3, 123, 125n 5, 130, 133, 154, 178–9 and the norm of belief 19, 28, 130, 133, 122n 3, 123, 125n 5, 128, 131, 133, 179

198 index truth (cont.) and inquiry 1, 5, 19–20, 28, 91–2, 94, 119–20, 125, 128–30, 133, 168 as an event 17–18 as an object 16–17, 18 as a property 6–40, 122–41 concept of 6–7, 18–20 explanatory jobs 22–9 ‘platitudes’ about 18–20, 27, 46–7, 51–2, 125, 131, 153 truth attributions 14, 132, 136–9, 157 truth-bearers 6, 18, 20–1, 23, 81n 29, 180 truth-making 159–82 truth pluralism formulations of 122–38 motivations for 83–104, 106–7, 110–13 objections to 132–8 Turner, J. 105n 1, 143, 144n 4, 147–9

Van Inwagen, P. 145, 149 Vision, G. 89, 162n 8, 176n 2

Ulatowski, J. 18n 10

Zack, N. 64n 3, 73

Weatherson, B. 38n 33, 164 width of cosmological role 69–77, 91–2, 111, 113, 140 Williams, B. 80 Williams, M. 22n 1 Williamson, T. 34, 38n 33, 166n 12 winning and truth 122–7 Wittgenstein, L. 122n 2, 134n 18 Wrenn, C. 24, 26, 41n 1, 58n 13, 162n 8 Wright, C.D. 51n 7, 79n 25, 128, 130, 138, 150 Wright, C.J.G. 3n 8, 18n 10, 19, 28n 8, 29, 33, 42, 46–7, 51–2, 60n 1, 69, 86n 1, 88n 2, 91–2, 108–9, 120, 122n 1, 127, 131, 140, 162n 4, 163n 9 Wyatt, J. 29n 14, 78n 24, 79n 25