New Thinking About Propositions

NEW THINKING ABOUT PROPOSITIONS

Jeffrey C. King, Scott Soames, & Jeff Speaks King020513OUK.indd ii 11/23/2013 12:58:33 PM New Th inking about Propositions

King020513OUK.indd i 11/23/2013 12:58:28 PM King020513OUK.indd ii 11/23/2013 12:58:33 PM N e w Th inking about Propositions

J e ff rey C. King Scott Soames

J e ff Speaks

King020513OUK.indd iii 11/23/2013 12:58:33 PM 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 © Jeff rey C. King, Scott Soames, and Jeff Speaks 2014 Th e moral rights of the authors have been asserted First Edition published in 2014 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: 2013944506 ISBN 978–0–19–969376–4 Printed and bound in Great Britain 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.

King020513OUK.indd iv 11/23/2013 12:58:33 PM Contents

I n t r o d u c t i o n 1 Jeff S p e a k s Part I. Common Ground 1. What Role do Propositions Play in our Th eories? 5 Jeff rey C. King 2. What's Wrong with Semantic Th eories Which Make no Use of Propositions? 9 Jeff Speaks 3. Why the Traditional Conceptions of Propositions Can't be Correct 25 Scott Soames

Part II. Th ree Th eories of Propositions 4. Naturalized Propositions 47 Jeff rey C. King 5. Propositions are Properties of Everything or Nothing 71 Jeff Speaks 6. Cognitive Propositions 91 Scott Soames

Part III. Critical Essays 7. Criticisms of Soames and Speaks 127 Jeff rey C. King 8. Representational Entities and Representational Acts 147 Jeff Speaks 9. Propositions vs Properties and Facts 166 Scott Soames

Part IV. Further Th oughts 10. Responses to Speaks and Soames 185 Jeff rey C. King 11. Representation and Structure in the Th eory of Propositions 215 Jeff Speaks

King020513OUK.indd v 11/23/2013 12:58:33 PM vi CONTENTS

12. Clarifying and Improving the Cognitive Th eory 226 Scott Soames

B i b l i o g r a p h y 245 Index 251

King020513OUK.indd vi 11/23/2013 12:58:33 PM I n t r o d u c t i o n

J e ff S p e a k s

Beginning in the 1970s and 1980s, many philosophers of language found themselves in an increasingly diffi cult situation. On the one hand, many came to believe that in order to do semantics properly as well as to give an adequate treatment of the attitudes, one needed to posit certain entities—propositions—which could be the meanings of sentences (relative to contexts), the contents of mental states, and the primary bearers of truth and falsity. But largely due to the arguments of Scott Soames1 many also came to distrust the standard theoretical account of the nature of propositions, which treated them as sets of worlds, and came to think of them instead as structured entities of some sort. Th ere was however no consensus about what these structured entities could be. A standard way of talking around the problem was to point out that propositions could be represented as ordered pairs. But it is pretty plain that this is just a way of talking around the problem—to say that propositions can be represented as ordered pairs is not to say what they are. Th e unsatisfactory situation persisted. Jeff King’s 2007 book, Th e Nature and Structure of Content, had a decisive impact on this situation in two ways. First, it made clear just how unsatisfactory the situation was, by stating the desiderata on a theory of propositions—desiderata which thinking of propositions as ordered pairs, for instance, plainly did not meet. Second, and perhaps more important, it made clear that the situation did not have to remain unsatisfactory; the book presented a novel view of propositions which both provided a clear metaphysical account of their nature, and made the case that propositions, so understood, could play the roles for which philosophers of language and mind wanted them in the fi rst place. It made clear that progress on the question of the nature of the proposition, the question which so puzzled Frege, Russell, and the early Wittgenstein, is possible. Th is book is a contribution to the ongoing discussion to which King’s book, as well as the work of others, gave rise.2 Th e three co-authors of this book agree on the views which led to the unsatisfactory situation described above. Th e aim of the fi rst Part is to make the case for this

1 See especially Soames (1988). 2 We can’t, of course, hope to discuss all of this recent work. See, among other places, Collins (2011), Gaskin (2008), Hanks (2009), Hanks (2011), McGlone (2012), Moltmann (forthcoming).

King020513OUK.indd 1 11/23/2013 12:58:33 PM 2 INTRODUCTION

constellation of views in two ways: fi rst, by arguing that we need propositions in our accounts of language and the mind, and in our semantic theories; and, second, by arguing that traditional accounts of propositions—the classical theories of Frege and Russell, and the view of propositions as sets of possible worlds—are not up to the task. A subsidiary aim of this part of the book is to make plain some of the constraints which a successful theory of propositions should have to meet. By seeing how certain views of propositions, or attempts to do without them altogether, fail—for example by failing to provide suitable objects of the propositional attitudes, or to give an adequate semantics for a certain class of sentences—we get clearer on the roles propositions must play. Each of us thinks that we should be able to give some account of the entities which play these roles. None of us is committed to the claim that this account should respect commonsense or folk intuitions about what these entities are (to the extent that such intuitions exist). But, on the other hand, the role these entities play—with respect to our mental states and our language use—means that they must be the sorts of entities to which ordinary users can refer and so, in that sense, they must not be completely beyond the ken of non-philosophers. At the end of Part I the agreement ends. In Part II, each of us lays out and defends his favored theory of propositions. King elaborates and refi nes his view of propositions as facts; Soames extends his view (fi rst defended in his 2010 book, What is Meaning? ) of propositions as cognitive event types; and I defend the view that propositions are a kind of property. Th e essays in Part III criticize the views defended in Part II; and in Part IV each of us presents some further thoughts on the task of giving a theory of propositions.

King020513OUK.indd 2 11/23/2013 12:58:33 PM P A R T I Common Ground

King020513OUK.indd 3 11/23/2013 12:58:33 PM King020513OUK.indd 4 11/23/2013 12:58:33 PM 1 What Role do Propositions Play in

our Th eories?

J e ff rey C. King

Why do many philosophers believe in propositions? Th e answer is that propositions are thought to perform a variety of jobs where it isn’t clear what would do those jobs were there no propositions. In the fi rst instance, sentences of natural languages (taken relative to contexts of utterance) seem to encode pieces of information. Competent hearers seem able to extract the encoded information from utterances of sentences by speakers. If sentences (relative to contexts of utterance) express propositions, these propositions can simply be identifi ed with the information contents of sentences (taken relative to contexts). By so doing, we can make sense of the pretheoretical idea that diff erent sentences of the same or diff erent languages can encode the same piece of information. Th is simply amounts to the claim that such sentences (relative to contexts) express the same proposition. Since presumably understanding a sentence (relative to a context) is a matter of grasping the piece of information it encodes, we can also say that understanding a sentence (relative to a context) is a matter of grasping the proposition it expresses (relative to the context). In addition, we can identify the meaning of a sentence relative to a context—its semantic content relative to that context—with the proposition it expresses relative to the context. Further, sentences taken relative to contexts are true or false. Th ey are so, presumably, in virtue of the information they encode. Here again, if we identify propositions with the information sentences encode relative to contexts, we can say that sentences are true or false relative to contexts in virtue of expressing propositions that are true or false. So here propositions play the role of being the information encoded by sentences relative to contexts and the things that determine whether a sentence relative to a context is true or false. Th at propositions play this latter role is oft en put by saying that propositions are the primary bearers of truth and falsity, since the truth and falsity of sentences relative to contexts is explained by reference to the propositions they express in those contexts being true or false.

King020513OUK.indd 5 11/23/2013 12:58:33 PM 6 JEFFREY C. KING

Propositions are also thought to be the things that possess modal attributes such as being possible, necessary and impossible. It certainly does seem as though the primary bearers of truth and falsity are also the things that possess these modal attributes. We do, aft er all, talk of something being necessarily true, thereby attributing both being true and being necessary to one and the same thing. Th en there are the things we believe, doubt, assume and suspect. It is thought that propositions are the things we bear these various attitudes to. Hence, e.g. believing that snow is white is a matter of bearing a certain sort of cognitive relation to the proposition that snow is white. Indeed, this view is so ingrained among philosophers that attitudes such as belief and doubt are generally called propositional attitudes . Again, this fi ts nicely with the other roles played by propositions, since it does seem that the things we believe and doubt are true or false and possess modal attributes like being necessary, possible and so on. Our perceptual experiences seem to represent the world as being certain ways. My current perceptual experience in some sense seems to represent the shirt I am wear- ing as black. Further, it seems that my perceptual experience can represent the world as being a certain way accurately or inaccurately. But this suggests that perceptual experiences have accuracy conditions. Th is leads many to think that perceptual experiences have contents that are accessible for accuracy. And this in turn leads many to believe that perceptual experiences have as their contents some sorts of propositions. So here, propositions play the role of being the contents of perceptual experiences and thus what explains how perceptual experiences can represent the world accurately or inaccurately. Th e relation between the propositional contents of perceptual experiences and of natural language sentences is a matter of some controversy and will come up in the sequel. For philosophers who countenance possible worlds, propositions off er an appealing way to understand possible worlds.1 Possible worlds can be identifi ed with complete, consistent sets of propositions. 2 Th ere are various ways to defi ne completeness and consistency, but the idea behind a set of propositions being complete is that any proposition is to be true or false relative to a complete set of propositions. As to consistency, the idea is that the members of the set could have been jointly true. We turn now to some of the roles played by propositions in philosophy of language. Having mentioned earler that propositions are the things we believe, doubt and so on, they are also the things we assert when we utter sentences. Asserting, like believing, is a relation between an individual and a proposition. It is widely agreed that to explain a variety of phenomena we need some notion of context that gives us something like the record of the conversation to that point. What exactly this record must contain is a matter of some dispute, but at least since the pioneering work of Stalnaker [1978] it is widely agreed that the record must include the material that the conversational participants are taking for granted in the conversation and recognize each other to be

1 See Adams [1974] 2 Or, as in Soames [2007], possible worlds could be identifi ed with the property of making all the propositions in a suitably defi ned set of propositions true.

King020513OUK.indd 6 11/23/2013 12:58:33 PM THE ROLE OF PROPOSITIONS IN OUR THEORIES 7

so doing. Th is material will include things that have been asserted and accepted in the conversation. But such things are propositions. Hence propositions will play a signifi - cant role in characterizing context/conversational record. We have mentioned that propositions serve as the semantic contents of sentences relative to contexts. But propositions will fi gure in semantic theory in other ways as well. In particular, we explain the semantics of “that” clauses by saying that they designate propositions. 3 Th us, the standard view is that a belief ascription such as 1. Amy doubts that snowboarding is hard. asserts that Amy stands in the doubting relation to the proposition designated by “that snowboarding is hard,” viz., the proposition that snowboarding is hard. On this account, “doubt” expresses a two-place relation between individuals and propositions, with “Amy” and “that snowboarding is hard” designating an individual and a proposition, respectively. As many have noted, this account looks able to explain why inferences such as the following are valid

Amy believes everything Carl said. Carl said that skiing is hard. So Amy believes that skiing is hard. For on the account in question, this argument has the following structure:

For every x such that Carl said x, Amy believes x. Carl said P. So, Amy believes P. Advocates of propositions hold that “that” clauses designating propositions are used to attribute other properties to propositions as well. Consider the following: 2 . Th at Harry is ignorant is true. 3 . Th at Harry is ignorant is likely. 4 . Th at Harry is ignorant is certain. 5 . Th at Harry is ignorant is possible. 2 attributes the property of being true to the proposition that Harry is ignorant, whereas 3, 4 and 5 attribute the properties of being likely, being certain and being possible to that proposition. Th ose who believe in propositions will also hold that they sometimes are designated by demonstratives. For example, suppose I say “Skiing is hard.” You may reply with either of the following: 6 . Th at’s true. 7. Glenn said that too.

3 I cautiously use the word “designate” to avoid committing myself to the claim that “that” clauses refer to propositions. See King [2002a] and King [2007] Chapter 5. Further, it may be that “that” clauses sometimes designate things other than propositions—facts—e.g. in so-called factive contexts such as “Jeff resents that the ski season is over.” King and Soames debate the issue in subsequent chapters.

King020513OUK.indd 7 11/23/2013 12:58:33 PM 8 JEFFREY C. KING

Given propositions, it is immensely plausible that the demonstrative “that” in both sentences designates the proposition that skiing is hard. 6 then predicates the truth of that proposition, while 7 asserts that Glenn stood in the saying relation to it. Th ere are still other cases in which advocates of propositions will claim that propositions are the semantic values of expressions. First, they will claim expressions like “what John said/asserted/assumed” and so on designate propositions.4 Second, they will claim that certain expressions function like names of propositions, such as “Gödel’s theorem,” “Goldbach’s conjecture” and “logicism.” Th ey will also hold that quantifi ers sometimes range over propositions, as in the sentence discussed above “Amy believes everything Carl said.” It should now be clear that propositions play a great number of roles in philosophy generally and philosophy of language in particular. To summarize, propositions are pieces of information we pretheoretically hold sentences to encode (relative to contexts). We thereby hold that propositions are the meanings of sentences taken relative to contexts and what is grasped in understanding a sentence in a context. Diff erent sentences may express the same proposition (relative to contexts), thereby capturing the intuition that diff erent sentences (relative to contexts) can “say the same thing.” Propositions are the primary bearers of truth and falsity and the possessors of modal attributes. Th ey are the things we believe, doubt and assert. Propositions of some sort are also widely held to provide the contents of perceptual experience. Some philosophers wish to construct possible worlds out of propositions. In philosophy of language, propositions are crucial for characterizing the important notion of context/conversational record. Th ey also serve as the semantic values (relative to a context) of a wide variety of expressions that occur in a variety of constructions and we oft en quantify over them. Th ose who wish to do without propositions need alternative accounts of all these matters. Providing such would be far from trivial.

4 Again, some expressions of this form might designate facts. For example, “what Glenn regrets.” See previous note.

King020513OUK.indd 8 11/23/2013 12:58:33 PM 2 What’s Wrong with Semantic Th eories Which make no Use

of Propositions?

J e ff S p e a k s

In this paper I’ll focus on two arguments against semantic theories which wish to avoid commitment to propositions. Each has been discussed in various forms in the literature. Th e fi rst holds that on the most plausible semantics of a class of natural language sentences, the truth of sentences in that class requires the existence of propositions; and some sentences in that class are true. Th e second holds that, on the best understanding of the form of a semantic theory, the truth of a semantic theory itself entails the existence of propositions.

Propositions as the referents of natural language expressions In chapter 1, King noted that one of the roles traditionally assigned to propositions is to serve as the referents of various natural language expressions. Th e purpose of this section is to say why propositions are standardly given this job—and why it is hard to see how we could give an adequate treatment of the relevant expressions without them. One way into this issue begins with a class of sentences which seems far removed from the topic at hand:

Th e apple sentences John ate something. Th e thing John ate was a delicious apple. Th ough what John ate was delicious, it would have been rotten if he’d waited another day before eating it. What John ate is what Mary gave him.

King020513OUK.indd 9 11/23/2013 12:58:33 PM 10 JEFF SPEAKS

It is the job of a semantic theory for a language to give the meanings of expressions of the language, part of which task is revealing, in a sense which I will leave informal, the logical structure of sentences of the language. It seems plausible that, at a fi rst pass, the logical forms of the sentences of the apple sentences are, respectively,

ූ x John ate x . [the x : John ate x ] delicious apple( x ) [the x : John ate x ] delicious apple( x ) & (John waits another day before eating x ☐→ x is rotten) [the x : John ate x ] Mary gave x to John.

Some evidence for these interpretations of the apple sentences is that the apple sentences seem to jointly entail:

(1) Th ere is something which John ate, which was a delicious apple, which could have been rotten, and which Mary gave him.

the logical form of which seems to be, roughly,

(1F) ූ x (John ate x & x is a delicious apple & x could have been rotten & Mary gave x to John)

We should want our interpretations of the logical structures of sentences to explain the logical relationships between those sentences; and the assignments of logical form given to the apple sentences does explain why they entail (1)/(1F).1 Let’s suppose for now that the interpretation of the logical forms of the apple sentences above is correct. Its correctness does not seem to depend on anything about the particular subject matter of those sentences—for example, that they are about apples rather than bananas, or that they concern John and Mary rather than other people. In fact, its correctness does not seem to depend upon anything which distinguishes the apple sentences from the sentences below which, for reasons which are probably obvious, I will call the proposition sentences :

John said something. What John said was true. Th ough what John said was true, it would have been false if things had gone diff erently. What John said is what Mary believed.

Th ese sentences seem, at a fi rst pass, to have something like the following as their logical forms:

1 I’m ignoring for simplicity worries about incomplete descriptions, and the assumption that the antecedent of the above counterfactual is possible.

King020513OUK.indd 10 11/23/2013 12:58:33 PM SEMANTIC THEORIES WITHOUT PROPOSITIONS 11

ූ x John said x . [the x : John said x ] x is true [the x : John said x ] x is true & (things go diff erently ☐→ x is false) [the x : John said x ] Mary believed x . As with the apple sentences, we can defend these interpretations of the proposition sentences by noting that the proposition sentences seem to jointly entail:

(2) Th ere is something which John said, which was true, which could have been false, and which Mary believed. the logical form of which seems to be, roughly,

(2F) ූ x (John said x & x is true & x could have been false & Mary believed x ) And the logical forms assigned to the proposition sentences explain why they entail (2F). If (2F) is true, then there are things which are said and believed, which are the bearers of truth values and have modal properties like being possibly true. So if (2F) is true, there are propositions. Hence the claim that (2F) is true is a claim which anyone who wants to develop a semantic theory without commitment to propositions must deny. Th e problem is that a plausible argument can be made that our semantic theory should be committed to the truth of (2F). We can think of this argument as having two independent premises. First, that sentences like the proposition sentences are sometimes true, and second, that the correct semantic analysis of these sentences is (roughly) the one given above. So one who wants to deny that (2F) is true must either deny that the analyses of the proposition sentences given above is correct—and so deny that the proposition sentences really do jointly entail (2F), even if they do entail (2)—or must deny that the proposition sentences are true. Before considering these two strategies for blocking the argument for (2F), let’s pause for a moment to consider the claim that (2F) entails the existence of propositions. Th ough this claim might seem obvious, there is one complication here worth noting. What (2F) and related claims say is that there is a type of entity which is what speakers say, what subjects believe, and is the bearer of truth and falsity as well as possible truth and falsity. Couldn’t one say that there is such a type of entity—but that it is not a proposition? One could, because standardly propositions are taken not just to have the character- istics attributed by (2F), but also to be language-independent abstract objects. We can bring out the complication by considering a “sententialist” view according to which sentences, rather than propositions, are the objects of attitudes like saying and believing, and the bearers of truth and falsity. According to this sort of sententialist view, (2F) is true—but still, one might think, we’re not thereby committed to the existence of propositions.

King020513OUK.indd 11 11/23/2013 12:58:34 PM 12 JEFF SPEAKS

Fortunately for our purposes, we can avoid this complication by pointing out that this version of sententialism is false. 2 Were this theory true, then the truth of “Violet believes that the sky is blue” would entail that Violet stands in the belief relation to the sentence, “Th e sky is blue.” But this ascription could be true even if Violet were a monolingual French speaker who stood in no special relation at all to this sentence— and indeed could be true even if Violet spoke no language at all. And given that sententialism is false, there don’t seem to be any other plausible candidates for objects which are not language-independent abstract objects which could satisfy (2F). (Th is, of course, leaves completely open the question of the nature of the abstract objects which satisfy sentences like (2F)—but at this stage, that’s what we want.) Hence I’ll proceed by ignoring the sententialist alternative, and assuming that if (2F) is true, then there are such things as propositions. Let’s turn to our fi rst strategy for blocking the argument from the proposition sentences to (2F): denying the argument’s validity. One question for the the proponent of a semantic theory which accepts the truth of the proposition sentences and (2) but denies (2F) is whether they want to say similar things about the apple sentences, (1), and (1F). Either way we go here, we seem to get in trouble. On the one hand, it is pretty hard to deny that (1F) gives an accurate rendering of (1). 3 On the other hand, it is pretty hard to deny that the relationship between the apple sentences and (1F) appears to be the same as the relationship between the proposition sentences and (2F). Semantic theories, like other theories, are to be judged in part by their capacity to give explanations which unify apparently similar phenomena. For this reason alone, we have some reason to doubt a semantic theory which off ers radically diff erent explanations of the fact that the apple sentences entail (1) and the fact that the proposition sentences entail (2). But, setting aside the point that the apple sentences and the proposition sentences seem, on their face, to demand parallel treatment, there remains the problem of how—if not via the analyses above or some notational variant thereof—to understand the logical form of sentences like the proposition sentences and (2), while capturing entailments like that between the proposition sentences and (2). Many attempts have been made to provide analyses of sentences like the proposition sentences which, unlike (2F), don’t entail the existence of propositions. Here I’ll focus on the most

2 Th e locus classicus for the following objection is Church (1950). Not every view which is standardly classifi ed as “sententialist” fi ts this mold. For example, the paratactic analysis of Davidson (1968), which I discuss below, is oft en thought of as a sophisticated sententialist view, but according to the paratactic analysis (2F) does not give the form of (2), and is not true. Hence I discuss it under the heading of views which deny that the proposition sentences entail (2F), rather than as a view which aims to reconcile (2F) with the denial of the existence of propositions. 3 Th ough, to be fair, some who doubt the existence of apples because of their views about mereology are inclined to off er diff erent interpretations of the apple sentences than the ones given above. (See, for example, van Inwagen (1990), ch. 10.) Someone with metaphysical scruples about propositions—whether or not they have doubts about apples—might on parallel grounds off er a reconstrual of the proposition sentences. I will consider some candidate reconstruals below.

King020513OUK.indd 12 11/23/2013 12:58:34 PM SEMANTIC THEORIES WITHOUT PROPOSITIONS 13

well-worked-out attempt to give a semantics which captures the logical properties of the proposition sentences and their ilk without commitment to propositions: the paratactic analysis of that-clauses fi rst proposed by Davidson (1968). On this view, a sentence like

Violet believes that the sky is blue. has the logical form of

Violet believes that. Th e sky is blue. where “that” is functioning as a demonstrative referring to the subsequent utterance of “Th e sky is blue.” What, on this account, does it take for a belief ascription like the above to be true? Not, of course, for Violet to endorse the sentence “Th e sky is blue” to which “that” is supposed to refer. As noted earlier, Violet might not speak English (or any language for that matter). Rather, what is required is that Violet stand in some belief-like relation to some sentence token which, as it’s standardly put, “samesays” the relevant utterance of “Th e sky is blue.” Th is sort of account gives rise to two immediate questions: What is samesaying? And what is the relation in which the belief ascription requires Violet to stand to a sentence which samesays the utterance of “Th e sky is blue”? Neither question is easy to answer. Intuitively, of course, for a pair of utterances to samesay each other is for the utterances to have the same content; the question is what this comes to if we don’t believe that there are such things as contents. One can’t say, for instance, that two sentence tokens samesay each other if they express the same proposition without undercutting the central aim of the paratactic analysis which is to avoid commitment to propositions. Some advocates of the paratactic analysis seem inclined to take the samesaying relation as primitive. LePore and Loewer (1989), for example, say that “it may prove impossible to explicate this relation in other terms” (103). To friends of propositions, this looks a bit suspicious. Th e samesaying relation is exactly the point in the paratactic analysis of attitude ascriptions which we should expect to force the theorist to appeal to propositions; hence it’s reasonable to wonder whether it is an acceptable primitive for the proponent of the paratactic analysis. 4 Let’s now turn to the second question raised by our rough sketch of the paratactic analysis: what is the relation in which Violet must stand to some sentence which samesays “Th e sky is blue”? Th e fact that we’re focusing on what Violet believes—as opposed to, for example, what she says—seems to raise special problems here. Isn’t it possible, aft er all, that Violet believes that the sky is blue without having ever uttered a sentence which samesays “Th e sky is blue”? Indeed, isn’t this possible even if Violet has not yet

4 Note that that it’s not enough for the proponent of the paratactic analysis to point out that competent speakers of a language have some grip on when a pair of utterances samesay each other. It is quite plausible that they do. Th e relevant question is not whether there is such a thing as the samesaying relation, but whether this relation can be understood in a way which does not appeal to propositions.

King020513OUK.indd 13 11/23/2013 12:58:34 PM 14 JEFF SPEAKS

learned the words to express her belief about the sky, and hence is not even disposed to accept any sentence which samesays “Th e sky is blue”?5 For the friend of the paratactic analysis, the best move here is to appeal to token belief states—thought of as states of or events involving the brain—rather than natural language sentences.6 One can then say that the above ascription is true iff Violet stands in the relation expressed by “believes” to some brain state which samesays “Th e sky is blue.” One might object that this view of belief ascriptions commits the paratactic analysis to a controversial theory of the mind—one according to which, necessarily, for every belief of every subject, there is a corresponding belief state which (intuitively) has the same content as the belief. Now this commitment by itself might not seem so bad—aft er all, this sort of theory of belief is widely, even if not universally, held to be plausible. What seems to me a bit worse is the fact that the paratactic account is not only committed to this theory of the mind, but also builds it into the meaning of belief ascriptions. On this sort of theory, when ordinary speakers talk about belief, they’re quantifying over the brain states of subjects of ascriptions, and making claims about brain states samesaying particular utterances. I’m sympathetic to Stalnaker’s worry that this sort of theory

makes a belief attribution carry more weight than it is plausible to assume that it carries. If it were correct, belief attributions would be far more speculative, and believers far less authorita- tive about their beliefs, than they seem to be. While theoretical and experimental developments in cognitive psychology may someday convince me that I store my beliefs in a form that is structurally similar to the form in which they are expressed and described in English, I don’t think that my ordinary belief attributions commit me to thinking that they will. 7 One way to bring out the oddness here is to imagine a philosopher expressing Stalnaker’s skepticism by saying:

It is possible for a subject to have the belief that the sky is blue without having any particular belief state with the content that the sky is blue. On the paratactic analysis, what this sentence means is, roughly,

Possibly ූ S ((ූ x (x is a belief state of S & x samesays “Th e sky is blue”)) & (¬ූ y (y is a belief state of S & y samesays “Th e sky is blue”))) Maybe our imaginary philosopher is making a false claim about what is possible for believers; but could he really be uttering an outright contradiction? But let’s set to one side these doubts about samesaying, and about the relation between belief ascriptions and belief states. Th e main source of objections to the

5 For discussion see, among other places, Schiff er (1987). 6 See, for example, LePore and Loewer (1989), 112. 7 Stalnaker (1990), 230.

King020513OUK.indd 14 11/23/2013 12:58:34 PM SEMANTIC THEORIES WITHOUT PROPOSITIONS 15

paratactic analysis is that the analysis seems unable to capture all of the entailments which we should want our semantics for belief ascriptions to capture. Consider, for example, the following inference:

1. Violet believes that the sky is blue. 2. Th e sky is blue. ———————————————— C. Violet believes something true. Th is argument certainly appears to be valid—in the (usual) sense that it is impossible for its premises to be true and its conclusion false. But according to the paratactic analysis, it is invalid. For consider a world w in which “Th e sky is blue” means that the sky is red, and suppose that in w Violet has a belief state which samesays (in w ) “Th e sky is blue”—and suppose that in w, as in @, the sky is blue. Th en the second premise of our argument is true; and on the paratactic analysis the fi rst must be as well, since in w Violet does have a belief state which samesays “Th e sky is blue.” But the conclusion of our argument will be false—since, in w, Violet’s belief is not true. And this contradicts our claim that the above argument is valid. One might reply that this argument overlooks the fact that the paratactic analysis of the fi rst premise attributes to Violet a belief state which samesays a particular utterance of “Th e sky is blue”—not the sentence type of which that utterance is a token. But this doesn’t seem to help; just as sentences could have had diff erent meanings than they actually have, particular utterances or inscriptions could have had diff erent meanings than they actually have. To suppose the opposite would be to make the quite surprising claim that particular sounds, or marks on a page, could have their meanings essentially. Given that they don’t have their meanings essentially, we can let w be a world in which the relevant utterance of “Th e sky is blue” has a diff erent meaning than it actually does, and the argument proceeds as above. A second reply would be to borrow a page from the post-Kripke descriptivist playbook, and introduce rigidifi cation into the analysis. One might say, roughly, that “Violet believes that the sky is blue” is true iff Violet has a belief state which samesays the utterance of “Th e sky is blue” in @ (at a particular time and location). But this is open to what seems to me to be a decisive objection raised by Soames (2002). Consider the ascription “Amelia knows that Violet believes that the sky is blue.” Th is could be true at a world w in which Amelia has no thoughts about @, the actual world. (Indeed, she may be in no position to refer to @, as opposed to worlds similar to @.) But on the proposed modifi cation of the paratactic analysis, this will be impossible, since the analysis of “Violet believes that the sky is blue” will make reference to @. A quite diff erent line of reply, defended by LePore and Loewer (1989) (110), is to concede the conclusion that, according to the paratactic analysis, the above argument is invalid, and to try to explain away its seeming validity. Th eir strategy is to say that

King020513OUK.indd 15 11/23/2013 12:58:34 PM 16 JEFF SPEAKS

the argument seems valid to speakers because they are assuming the truth of an extra premise like:

“ Th e sky is blue” is true iff the sky is blue. As competent English speakers, we know this premise to be true. And if we add this premise to the argument this is enough to make the argument, given the truth of the paratactic analysis, valid. Th is seems to me unsatisfactory. In general, it is simply not true that arguments which would be valid were we to add a premise which we all agree to be true seem valid to us. Consider, for example,

Blue is my favorite color. — — — — — — — — — — — Th e color of the sky is my favorite color. Th is argument does not even seem valid; students just introduced to the concept of validity can see that it is not. Does that mean that such students harbor some doubts about whether the sky is blue? Surely not. Rather, despite the fact that they know—and know that each other knows—that the sky is blue, they are able to evaluate the validity of this argument without importing this assumption as an extra premise. It is therefore mysterious why we should not be able to do this with the argument above.8 Summing up: it seems to me plausible that even if we set aside doubts about the samesaying relation, and about the use of belief states in the semantics of belief ascriptions, the paratactic analysis fails to do the job which we wanted it to do: namely, to account for the validity of inferences involving sentences like the proposition sentences without appealing to propositions. 9 Let’s set aside attempts to explain the truth of proposition sentences, and the fact that they entail (2), without commitment to (2F) and the existence of propositions. A quite diff erent line of response to the inference from the truth of the proposition sentences to the existence of propositions is not to deny its validity but simply to deny its premise—namely, that sentences like the proposition sentences are true. Th is is, strictly speaking, an issue outside the scope of semantic theory, since semantic theories are committed to claims about what it takes for a sentence of the object language to be true rather than claims about which sentences of that language are true. 10 But it is tempting to say that if any view deserves an incredulous stare, it is this one. Th ose who deny that any proposition sentence can be true are committed to denying

8 One might reply that the diff erence is that while the students know that the sky is only contingently blue, we mistake the contingent truth that “Th e sky is blue” is true iff the sky is blue for a necessary truth. But of course philosophers are very familiar with the fact that biconditionals of this sort are only contingent, and the argument about Violet’s beliefs still seems obviously valid to us. 9 For an excellent summary of other objections, along with replies to those objections, see LePore and Loewer (1989). 10 With the exception of logical truths and, in the case of a possible worlds semantics, truths which are true in every index.

King020513OUK.indd 16 11/23/2013 12:58:34 PM SEMANTIC THEORIES WITHOUT PROPOSITIONS 17

that, strictly speaking, anyone ever says anything, believes anything, or wants anything. Indeed, it is one of those views that is, unless you’re being careful, hard to state without self-refutation—as the preceding sentence illustrates. How might one make plausible the denial that the proposition sentences are true? One currently popular strategy would be to endorse a fi ctionalist theory of our proposition-talk. On this view, we should take our discourse about propositions— i.e., our use of proposition sentences—as involving something other than outright assertions of the propositions corresponding to the surface form of the sentences uttered. Instead, in uttering those sentences we’re engaging in a kind of pretense or pseudo-assertion. What exactly this comes to depends on the fi ctionalist view in question. On some views, it will be a matter of asserting some proposition other than the one corresponding to the surface form of the sentence uttered—like the proposition that that sentence is true according to a certain fi ction. On other views, it will be a matter of speakers standing in some attitude other than assertion to the relevant proposition. 11 Like the paratactic analysis, fi ctionalist approaches to various discourses have generated a large literature to which I can’t aim to do justice here. It should be admitted that, in some ways, our proposition talk seems ripe for a fi ctionalist treatment—one might, to borrow an example from Yablo (2000), feel impatient with someone worrying about whether we have beliefs on the basis of worries about whether propositions exist in much the same way as one might feel impatient with someone worrying about whether “creatures of metaphorical make-believe” like “the green-eyed monster” really exist. Surely, one wants to say, such skepticism misses the point of our talk about belief! On the other hand, discourse about what we and others believe and desire, and what is true, can seem like the very paradigm of genuine assertion, and hence a quite implausible target for fi ctionalist treatment.12 Th e question of whether fi ctionalism about our use of the proposition sentences and the like can give an adequate treatment of their (seeming) truth conditions and logical properties is one which can only be answered by consideration of detailed fi ctionalist proposals.13 But, independent of such consideration, one might wonder whether whether fi ctionalism about proposition talk faces a special sort of self-referential

11 On either construal, this would be a “hermeneutic” rather than a “revolutionary” fi ctionalism, since it purports to describe proposition talk rather than to reform it. Th e distinction is due to Burgess (1983) and is applied to the case of fi ctionalism in Stanley (2001). Fictionalism about propositions is defended in Balaguer (1998). Crimmins (1998) and Kroon (2004) defend related but weaker theses. Of course, one might worry in the present context that this description of fi ctionalism immediately entails the existence of propositions. More on this below. 12 Th is sort of objection to various fi ctionalist views is voiced in Stanley (2001), who objects that at least some fi ctionalist treatments of areas of discourse which we take to be perfectly literal involve an implausible attribution of “a novel and quite drastic form of failure of fi rst-person authority over one’s own mental states” (47). 13 For two excellent examples, see Richard (2000) and Stanley (2001).

King020513OUK.indd 17 11/23/2013 12:58:34 PM 18 JEFF SPEAKS

problem which is not faced by fi ctionalism about other types of discourse. To see this, consider Mark Richard’s well-known explanation of the notion of “piggy-backing,” which is part of the explanation of how one sort of fi ctional use of a sentence might work. Piggy-backing is, he says,

making an utterance u within a pretense in which u has a real world truth condition c , thereby actually asserting a proposition which is (in fact) true iff c obtains14 Th e salient point, for our purposes, is that this account of what piggy-backing is—and hence, in part, of what the relevant sort of fi ctionalist theory says—involves claims about the assertion of propositions and hence can’t be endorsed by anyone who denies that there are such things. And this is not just true of Richard’s way of setting the issue up. Fictionalism is oft en introduced via a distinction between two diff erent attitudes toward a class of propositions—one which involves commitment to the propositions’ truth, and one which does not. But no such way of explicating the doctrine can work if the goal is fi ctionalism about propositions—we can’t use talk of propositions in our (presumably non-fi ctional) explication of the theory while explaining them away with the theory so explicated. 15 Similar remarks apply to Kendall Walton’s fi ctionalist account of discourse about fi ctional characters, as Walton is well aware. As he says, “I have shamelessly helped myself to properties and propositions in the preceding chapters, and will use them now in explaining away fi ctional entities.”16 And the same goes for any fi ctionalist account which either explicitly makes reference to contents or uses sentences like the proposition sentences, since there’s at least a serious suspicion of circularity about the method of stating a fi ctionalist theory of the proposition sentences by (presumably, non-fi ctionally) using those very sentences. It’s important not to overstate the importance of this point—not every fi ctionalist account makes use of piggy-backing, defi ned as Richard defi nes it, or is modeled on Walton’s fi ctionalism about fi ctional characters. But it’s at least not obvious that fi c- tionalism about proposition talk can be explicated in a way which does not smuggle propositions in via the back door. I will conclude this section with two general thoughts about the view that the proposition sentences are false, whether or not this is accompanied by a fi ctionalist story about our ordinary use of those sentences. First, the best motivation for this view seems to be, in eff ect, to run the present argument in reverse and say that, precisely because the proposition sentences do entail (2F), and because admitting propositions to our ontology is an unacceptably high cost to pay, we are forced to deny the truth of the proposition sentences. Answering this

14 Richard (2000), 214. Richard introduces this notion in the process of explicating and criticizing the fi ctionalism about modes of presentation defended in Crimmins (1998). 15 Related points are made in Richard (2000), §5. 16 Walton (1990), 390.

King020513OUK.indd 18 11/23/2013 12:58:34 PM SEMANTIC THEORIES WITHOUT PROPOSITIONS 19

sort of argument involves showing that admitting propositions to our ontology is not an unacceptably high cost to pay, and this is best done by developing a positive theory of propositions—hence, to that extent, evaluation of this case for denying the proposition sentences must wait until the sketches of our positive theories of propositions in Part II. Second, there is a sense in which this strategy might be unstable. If the theorems of our best semantic theory are themselves relevantly like proposition sentences, then the semantic theorist, who is in the business of constructing a semantic theory, can hardly avoid commitment to propositions by denying all of the proposition sentences. (Unless, of course, she regards her own semantic theorizing as itself a kind of pretense.) As we’ll see in the next section, a good case can be made that the most plausible semantic theories which do not make explicit reference to propositions do, in fact, have proposition sentences among their theorems.

Propositions and the theorems of semantic theories Th e aim of a semantic theory is to state the meanings of expressions of the target language. So the output of such a theory should include, for every well-formed expression of the target language, a theorem which—in some sense or other—states that expression’s meaning. But how exactly should we understand these theorems? One possibility is that we should take the theorems of a semantic theory to provide a pairing between sentences of the language and the propositions expressed by those sentences.17 Th is, obviously, is not a view of the theorems of a semantic theory of which a theorist who wants nothing to do with propositions can avail himself. So how should such a theorist think of the theorems of her semantic theory? Th e main alternative to propositional semantics is due to the work of Donald Davidson.18 For our purposes, the salient aspect of Davidson’s approach to semantic theory is his view that the theorems of semantic theory should be, not pairings of sentences with propositions, but rather T - sentences of the form:

“Amelia sings” is T (in the language) if and only if Amelia sings. Th e claim that sentences of this sort could be the theorems of semantic theories is, on the face of it, puzzling. Why should a theory which issues T-sentences, but makes

17 Here, as above, I’m ignoring for simplicity the need to relativize the propositions expressed by sentences to contexts. Put in these terms, what a semantic theory should provide is a pairing between sentences and characters—where the latter are functions from contexts to propositions. 18 See Davidson (1967). One might think that another main alternative is the view that a semantic theory should provide not a pairing between sentences and propositions, but rather a pairing between sentences and a set of indices in which those sentences are true. I’m setting this aside for two reasons. First, I take this to be not an alternative to a propositional semantics, but rather a version of it according to which propositions are sets of worlds, or situations, or whatever. Second, this view of semantic theory is (in part) the topic of Ch. 3.

King020513OUK.indd 19 11/23/2013 12:58:34 PM 20 JEFF SPEAKS

no explicit claims about meaning or content, count as a semantic theory? Davidson’s answer was that knowledge of such a theory would be suffi cient to understand the language. If Davidson were right about this, then he would have a plausible argument that a semantic theory could take this form. Aft er all, it is plausible that someone who understands a language knows the meanings of the expressions in the language; so, if knowledge of a Davidsonian semantic theory for the language were suffi cient to understand the language, then knowledge of what that theory says would be suffi cient to know all the facts about the meanings of expressions in the language, in which case it seems that the theory would state all the facts about the meanings of expressions in the language. Let’s assume that we have in hand a complete Davidsonian semantic theory for a language—which entails, for every sentence of the language, a true T -sentence—and ask whether knowledge of such a theory would be suffi cient to understand the language. Th ere are two reasons—both ultimately due to Foster (1976)—for thinking that it would not. Following Larson and Segal (1995), I’ll call these the extension problem and the information problem . Th e extension problem stems from the fact that it is not enough for a semantic theory whose theorems are T -sentences to yield true theorems; the T -sentence

“Snow is white” is T in English iff grass is green

is true, but tells us hardly anything about the meaning of “Snow is white.” Rather, we want a semantic theory to entail, for each sentence of the object language, exactly one interpretive T-sentence: a T -sentence such that the sentence used on its right-hand side gives the meaning of the sentence mentioned on its left -hand side. Our theory must entail at least one such T -sentence for each sentence in the object language because the aim is to give the meaning of each sentence in the language; and it must entail no more than one because, if the theory had as theorems more than one T-sentence for a single sentence S of the object language, an agent who knew all the theorems of the theory would not yet understand S, since such an agent would not know which of the T-sentences which mention S was interpretive. Th e extension problem is the problem of designing a theory which can meet both of these requirements. One reason why the extension problem is diffi cult is that it seems that any theory which implies at least one T -sentence for every sentence of the language will also imply more than one T -sentence for every sentence in the language. For any sentences p,q , if the theory entails a T-sentence

S is T in L iff p,

then, since p is logically equivalent to p & ฏ ( q & ฏ q ), the theory will also entail the T-sentence

S is T in L iff p & ฏ( q & ฏ q ),

King020513OUK.indd 20 11/23/2013 12:58:34 PM SEMANTIC THEORIES WITHOUT PROPOSITIONS 21

which, if the fi rst is interpretive, won’t be. But then the theory will entail at least one non-interpretive T -sentence, and someone who knows the theory will not know which of the relevant sentences is interpretive and which not; such a person therefore would not understand the language. Th e information problem is that, even if our semantic theory entails all and only interpretive T-sentences, it is not the case that knowledge of what is said by these theorems would suffi ce for understanding the object language. For, it seems, I can know what is said by a series of interpretive T-sentences without knowing that they are interpretive. I may, for example, know what is said by the interpretive T -sentence

“Londres est jolie” is T in French iff London is pretty but still not know the meaning of the sentence mentioned on the left -hand side of the T-sentence. Th e truth of what is said by this sentence, aft er all, is compatible with the sentence used on the right-hand side being materially equivalent to, but diff erent in meaning from, the sentence mentioned on the left . In response to these objections, Davidsonians typically modify the bare sketch of Davidsonian semantic theory given above in two related ways. First (in response to the extension problem) they specify a set of canonical rules of inference which don’t permit the derivation of non-interpretive T -sentences. Second (in response to the information problem) they add to the theory a rule of inference which takes us from canonically derived T -sentences to a diff erent sort of sentence, like:

A theory meeting such-and-such formal and empirical constraints entails that S is T iff p or a meaning theorem, like:

S means that p .19 Th is extra rule of inference helps with the information problem, because it seems that someone who knows what is said by, for example, a meaning theorem will know the meaning of S . Let’s take a step back here. Th e initial idea was that Davidson’s approach to semantics might provide a viable avenue for the semantic theorist who wants to avoid commitment to the existence of propositions; the reason for thinking this was that the theorems of a Davidsonian semantic theory— T -sentences, rather than pairings of sentences with propositions—don’t involve any commitment to the existence of propositions. But now we’ve given up the idea that T -sentences are the theorems of a Davidsonian semantic theory and replaced these with sentences which are relevantly similar to the “proposition sentences” discussed in the previous section. A plausible case can be made, as noted earlier, that proposition sentences and others of the same form entail the existence of propositions. And I noted earlier, also that, for

19 Th ese responses are defended in, respectively, Davidson (1976) and Kolbel (2001).

King020513OUK.indd 21 11/23/2013 12:58:34 PM 22 JEFF SPEAKS

this reason, one might avoid commitment to propositions simply by saying that no sentence of this form is true. But this is plainly not something which our neo-Davidsonian semantic theorist can say, since she would then be denying the theorems of her own theory. Th e proposed modifi cation of Davidsonian semantics thus ends up closing off one route for the semantic theorist who wants to avoid commitment to propositions. 20 Setting this aside, there are independent problems with the idea that a semantic theory could include a rule of inference like one which takes us from T -sentences to meaning theorems—at least when this is conjoined with the idea that a theory is a satisfactory semantics for a language iff knowledge of the theory would suffi ce to understand sentences of the language. Th e problem is that the sort of explanation of semantic competence that we get from a neo-Davidsonian semantic theory can also be given by a certain sort of translation manual; and giving a translation manual, which maps sentences from one language onto their translations in the other language, is not an adequate semantic theory for either. Hence the sort of explanation of semantic competence provided by a neo-Davidsonian theory of the sort sketched above is not suffi cient to justify those theories.21 Th e point that providing a translation manual between two languages does not suffi ce for providing a semantic theory of either has been defended elsewhere, and I won’t go through the arguments again here.22 Given this, it is worth noting that a translation manual of this sort might provide an explanation of a kind of a speaker’s competence with a language. Aft er all, given my knowledge of English, if I am given, and suffi ciently internalize, a translation manual which maps sentences of Urdu onto their English translations (and vice versa), this will give me an understanding of the relevant Urdu sentences. So a translation manual can explain semantic competence with a language—so long as we take for granted the speaker’s understanding of another language. Let’s call this a derivative explanation of semantic competence, since it explains competence with one language in terms of competence with another. Given that translation manuals are not semantic theories, if Davidsonian semantic theories are to be defended on the grounds that knowledge of them would suffi ce for competence, the sort of explanation they off er had better not be a derivative one. But this is just the sort of explanation of semantic competence which can be provided by a Davidsonian semantics supplemented by the sort of extra rule of inference mentioned above. To see this, consider how, precisely, the crucial extra rule of inference from canonically derived T-sentences to the theorems of the theory might be formulated. Where L is the target language, we could try to formulate it as:

I f ໠ S is T in L iff p ໡ is a theorem of the theory, then ໠ S means that p ໡ is a theorem of the theory.

20 Th is still leaves the Davidsonian the option, discussed in the previous section, of endorsing the truth of proposition sentences while denying that these sentences entail claims like (2F). 21 Th is argument is inspired by Harman (1975). I develop this line of argument in more detail in Speaks (2006). 22 See, among other places, Lewis (1970), LePore and Loewer (1981), and Speaks (2006).

King020513OUK.indd 22 11/23/2013 12:58:35 PM SEMANTIC THEORIES WITHOUT PROPOSITIONS 23

But this will plainly not provide knowledge of the meaning of S . It will, perhaps, provide knowledge that a certain sentence about the meaning of S is a theorem of a true theory, and hence true; but knowledge that a sentence about the meaning of S is true doesn’t tell one what the meaning of S is unless one understands the sentence about S. In this case, that sentence—a sentence of the form ໠ S means that p ໡—is stated in the language of the theory. Hence explaining competence via a theory which depends crucially on this rule of inference relies upon the speaker’s prior competence with the language of the theory—and this is the same sort of derivative explanation of semantic competence as is provided by a translation manual. We could try to avoid the problem by formulating the rule of inference in terms of what is said by the theorems of the theory, without mentioning expressions in the language of the theory, as follows:

If it follows from the theory that S is T iff p , then S means that p . But there is a problem with understanding this formula. “ S ” can be understood as a universally quantifi ed variable over sentences, but how is “ p ” to be understood? One idea is that “p ” is a sentence letter, and that the above is a schema; on this formu- lation, to master this rule of inference is to know that every instance of the schema is true. But an instance of a schema is a sentence, and if what this rule of inference gets us is that a certain sentence is true, then we are back in the problem we were trying to avoid. Knowledge that a certain meaning theorem is true yields knowledge of the meaning of the sentence mentioned in the theorem only if one understands the theorem; but if we are making use of competence with the language in which the meaning theorems are stated, we can give no non-derivative explanation of semantic competence. One might object that this argument attacks a straw man. Any theory has to be stated in some language or other; hence any explanation of semantic competence based on knowledge of a semantic theory will presuppose understanding of the language of the theory. To argue that knowledge of a neo-Davidsonian theory for L cannot give a non-derivative explanation of semantic competence with L is thus only to point out that this theory can’t do the impossible. If this were true, this would be a problem for Davidsonian semantics rather than for the present argument; for if no semantic theory can give a non-derivative explanation of semantic competence, then Davidson’s principal argument in favor of his approach to semantics—namely, that it can give such an explanation—fails. But, more importantly, the objection is based on a mistake. Consider a semantic theory whose theorems pair sentences with the propositions which are their semantic contents. Th ere’s no reason why one can’t know the theorems of such a theory without knowing the language of the theory, since in general there’s no reason why one can’t know a proposition without knowing the truth of some sentence which expresses that proposition.23 Th e troubles into which neo-Davidsonian explanations of semantic competence fall

23 Th ough such explanations of competence might face other problems. See Soames (1992).

King020513OUK.indd 23 11/23/2013 12:58:35 PM 24 JEFF SPEAKS

are not the result of the fact that they are stated in a language. Th ey are the result of the particular rules of inference on which those theories rely to solve the information problem. And in fact we can show that there is something fi shy about the inclusion of these extra rules of inference without even bringing in issues about the explanation of semantic competence. Suppose for reductio that a Davidsonian semantic theory, supplemented with one of the sorts of rules of inference above, can serve as a satisfactory semantic theory for L . Th en any theory which provides as much information about the meanings of expressions of L (and no extra, false information) must also be satisfactory semantics for L. But consider a translation manual of the sort mentioned above, which pairs synonymous sentences of two languages. All should agree that this sort of translation manual is a semantic theory of neither of the two languages; it says which expressions mean the same as which other expressions, but doesn’t say what any of those expressions do mean. But, as Gilbert Harman suggested, we can imagine adding to this translation manual a rule of inference parallel to the rules of inference which are supposed to take us from T- sentences to meaning theorems, e.g.

I f ໠໠ S໡ in L means the same as ໠S*໡ in L* ໡ is a theorem of the theory, then ໠໠ S໡ in L means that S* ໡ is a theorem of the theory.24 But if we add this sort of rule of inference to our translation manual, we are able to use it to derive as theorems all of the theorems of our neo-Davidsonian theory. Hence our modifi ed translation manual must be a satisfactory semantic theory for L . But it isn’t. So our original supposition, that a Davidsonian semantic theory supplemented with these sorts of extra rules of inference, might be a satisfactory semantic theory for a natural language, must be rejected.25 In “Truth and Meaning,” Davidson remarked that, “paradoxically, one thing that meanings do not seem to do is oil the wheels of a theory of meaning—at least as long as we require of such a theory that it non-trivially give the meaning of every sentence in the language. My objection to meanings in the theory of meaning is not that they are abstract or that their identity conditions are obscure, but that they have no dem- onstrated use.”26 Th e foregoing provides some reason for thinking that Davidson was wrong about this—indeed radically wrong, if the preceding arguments show that meanings as entities are not just useful, but necessary, for the construction of a semantic theory. But, as Davidson’s quote suggests, there is another powerful motivation for non-propositional semantics: this is the thought that, however useful propositions may be in semantic theory, there can be no adequate metaphysical account of what sorts of things these entities are. An adequate answer to this sort of skepticism requires an answer to the question: What are propositions? Th is is the question to which much of the following is addressed.

24 See Harman (1974). Here I am supposing that L* is the language of the theory. 25 For a less compressed exposition of this line of argument, see §II of Speaks (2006). 26 Davidson (1967), 20–21.

King020513OUK.indd 24 11/23/2013 12:58:35 PM 3 Why the Traditional Conceptions of

Propositions Can’t be Correct

S c o t t S o a m e s

Th e two leading traditional conceptions of propositions are those found in classical theories of structured propositions descending from Frege and Russell, and those growing out of more recent theories of propositions as sets of possible worlds (or functions from such to truth values). I will begin with the former.

Th e classical theories of Frege and Russell According to Frege and Russell, propositions are meanings of sentences, objects of attitudes, and the primary bearers of truth. Since meaningful sentences are grammatically structured complexes of meaningful constituents, it was natural for Frege and Russell to take propositions to be similarly structured complexes composed of the meanings of those constituents. In identifying propositions as objects of attitudes such as judgment and belief, Frege and Russell implicitly distinguished what is judged or believed from cognitive events of judging and mental states of believing. When speaking of John’s judgment or belief that someone exists as being true, even though it could have been false, or as following from the judgment or belief that John exists, we are referring to propositions, and predicating certain logical properties or relations of them. By contrast, when speaking of a judgment or belief as irrational , ill-conceived, or unshakable , we are speaking of a cognitive act or state, and attributing psychological properties to it. Judgment and belief are themselves relations between one who judges or believes and the thing so judged or believed. In this way, these attitudes are like seeing. Just as when one sees a tree, some object is seen, so when one believes or judges that the tree is green, some proposition is believed or judged. Since propositions are the primary bearers of truth for Frege and Russell, other things—sentences, utterances, acts of judgment, and states of believing—are true only in virtue of the relations they bear to propositions that are true. If John’s prepared

King020513OUK.indd 25 11/23/2013 12:58:35 PM 26 SCOTT SOAMES

statement, which lasted for over ten minutes, is entirely true, it is so because his ora- tion resulted in the assertion of one or more propositions, each of which is true. Propositions themselves then are taken to be timeless, unchanging, platonic entities with which we are acquainted by a kind of passive intellectual awareness. As Russell put it in (1904),

Suppose, for the sake of defi niteness, that our judgment is ‘A exists,’ where A is something that does as a matter of fact exist. Th en A’s existence [by which he means the proposition that A exists ], it seems plain, subsists independently of its being judged to subsist . . . In this case the Objective [i.e. the proposition] of the judgment—at least in the view of common sense—is as truly independent of the judgment as is A itself. But the peculiarity of the cognitive relation, which is what we wish to consider, lies in this: that one term of the relation is nothing but an awareness of the other term.1 According to this conception, which Frege shared, the fact that propositions represent things as being one way or another, and so are true (false) iff those things are (are not) the way they are represented to be, is not derivative from, or attributable to, conceptually prior cognitive activities of agents who entertain them. On the contrary, since propositions are the primary bearers of intentionality, the intentionality and truth conditions of cognitive acts or states must be explained in terms of quasi-perceptual relations we bear to propositions. For Frege and Russell, all intentionality originates and is grounded in an abstract “third realm.” It is this diffi cult doctrine, more than any other, that generates the fatal diffi culty they encountered —known as “the problem of the unity of the proposition”. Since the two philosophers struggled with the problem in diff erent ways, I will deal with them separately. Russell on propositional unity Th e problem of the unity of the proposition is to explain what propositions are in a way that makes clear how they can have the intentional properties they do. Russell starts with simple sentences like (1), one part of which—the predicate—is used to say, or assert, something about the referent designated by its other part—the subject. 1. Socrates is human. Corresponding to this, Russell thought, one part of the proposition expressed by (1)— the concept/property humanity—is applied to, or asserted of, the other part of the proposition—the man Socrates. Th is was his basic model for propositions.

In every proposition . . . we may make an analysis into something asserted and something about which the assertion is made.2

1 Russell (1904), “Meinong’s Th eory of Complexes and Assumptions,” Mind , 13: three installments: 204– 219, 336–354, 509–524; reprinted in Russell , Essays in Analysis , New York : George Braziller), 1973 , cited at p. 60. 2 Russell (1903), p. 43.

King020513OUK.indd 26 11/23/2013 12:58:35 PM THE INADEQUACY OF TRADITIONAL CONCEPTIONS 27

Although there are problems with this model, there is also something revealing about it. When we assert that Socrates is human we may be said to assert the property being human (a.k.a. humanity ) of him. In so doing we represent him as human. Hence, our assertion has truth conditions; it is true iff Socrates is the way he is represented to be. Th is is clear enough. However, Russell’s task of explaining the intentionality of our speech act in terms of the conceptually prior intentionality of the proposition expressed was more daunting. Roughly put, he needed to translate commonsense talk about what we do—assert of things that they are so-and-so, and thus represent them as being a certain way—into talk about what propositions “do,” in a way that reveals them to be the fundamental bearers of intentionality. However, the idea needs fi ne tuning. As the examples in (2) illustrate, the notion Russell needs is not assertion, but predication. 2a. Socrates is human. b. If Socrates is human, then Socrates is mortal. c . Th at Socrates is human is widely believed. d. I wonder whether Socrates is human. e. Is it true that Socrates is human? f . I s S o c r a t e s h u m a n ? Whereas one who assertively utters (2a) asserts that Socrates is human, one who assertively utters (2b), (2c), or (2d) does not assert this, while one who uses (2e) or (2f) to ask a question doesn’t assert anything. Nevertheless, there is something common to the cases. In each case, the speaker uses the name “Socrates” to refer to Socrates, and the predicate “is human” to represent him as being a certain way. In (2a) Socrates is asserted to be that way; in (2b) it is asserted that if he is that way, then he is mortal; in (2c) it is asserted that he is widely believed to be that way; in (2d) I indicate that I wonder whether he is that way, and in (2e) and (2f), I ask whether he is. In each case, part of what a speaker does in performing his or her overall speech act (of asserting, questioning, etc.) is to represent Socrates as being human by predicating the property humanity of him. Since, for Russell, the intentional properties of the resulting speech acts are derived from the propositions they express—which include the proposition that Socrates is human—the relation he needs to “unify” that proposition is predication. In the proposition that Socrates is human, humanity is predicated (not asserted) of Socrates. Th e problem is that while predication, as I have used it here, is something that agents do, what Russell needs is “a logical sense of predication.” It is, admittedly, diffi cult to see what this might amount to. Perhaps because of this he was drawn to assertion, rather than predication, as the crucial “unifying” relation. Since for him propositions are entities the intentional properties of which are conceptually independent of us, it was natural for him to look to “assertion in the logical sense”—which he took to be linked in a mysterious way to truth—to do the job.3 Th e obvious problem, of course, is that no

3 At this stage, Russell was strongly tempted by the idea that just as the copula, “is,” in “Socrates is human” signals that “human” is functioning as a predicate of the referent of “Socrates,” so the proposition expressed

King020513OUK.indd 27 11/23/2013 12:58:35 PM 28 SCOTT SOAMES

relation that fails to apply to false propositions can possibly “unify” them. In addition, since, for Russell at this stage, there was no intentionality without propositions, and hence no conception of truth prior to an account of its bearers, the unifying notion he needs is explanatorily prior to truth. For these reasons, we are better off shift ing the burden of “unifying” propositions to predication, while waiting until later to decide whether propositions so conceived can have whatever ontological independence from agents turns out to be needed. Th is brings us to Russell’s most famous remark on propositional unity.

Consider, for example, the proposition “A diff ers from B.” Th e constituents of this proposition, if we analyze it, appear to be only A, diff erence, B. Yet these constituents, thus placed side by side, do not reconstitute the proposition. Th e diff erence which occurs in the proposition actually relates A and B, whereas the diff erence aft er analysis is a notion which has no connection with A and B. It may be said that we ought, in the analysis, to mention the relations which diff erence has to A and B, relations which are expressed by is and from when we say A is diff erent from B. Th ese relations consist in the fact that A is referent and B relatum with respect to diff erence. But A, referent, diff erence, relatum, B, is still merely a list of terms, not a proposition. A proposition, in fact, is essentially a unity, and when analysis has destroyed the unity, no enumeration of constituents will restore the proposition. Th e verb, when used as a verb, embodies the unity of the proposition, and is thus distinguishable from the verb considered as a term, though I do not know how to give a clear account of the precise nature of the distinction.4 (my emphasis)

A central point here is that there is more to the proposition that A diff ers from B than the fact that its constituents are A, B, and diff erence. In addition, there is both the manner in which these constituents occur, and how their occurring as they do represents A and B as being diff erent.Modifying Russell, we may put this by saying that in the proposition, diff erence is predicated of A and B, with the result that they are represented as being diff erent . In a mere list, nothing is predicated of anything, so the list doesn’t represent the items as being one way rather than another. Consequently propositions are true or false, while lists are neither. Although we are presently taking predication to be primitive, one can still reasonably ask for an account of what, in a proposition, indicates which constituent is predicated of which things. Since adding predication as an extra propositional constituent would do nothing to confer intentionality on what is otherwise a mere list, there seems

by the sentence contains something functioning as “a verb” that somehow unites Socrates and humanity into a coherent propositional whole. Th is “verb” is thought to be a special sort of assertion/predication relation by which the concept humanity is brought to bear on the man Socrates, rendering the whole representational. See section 53 of Russell, Th e Principles of Mathematics, 1903. Of course, it wouldn’t do for this “verb” simply to occur as another constituent of the proposition, since that would generate a further unity problem. Rather it must “really apply to,” and hence really relate, its arguments. Russell’s struggle in Th e Principles of Mathematics with “the logical notion of assertion” as the element needed to explain the unity of the proposition is discussed in section 3.5 of chapter 7 of Volume 1 of Soames, Th e Analytic Tradition, Princeton: Princeton University Press. 4 Russell (1903), pp. 49–50.

King020513OUK.indd 28 11/23/2013 12:58:35 PM THE INADEQUACY OF TRADITIONAL CONCEPTIONS 29

to be only one answer to this question that is roughly Russellian in spirit. 5 J u s t a s t h e structural relations holding among syntactic constituents of a sentence show how they are to be understood, so the structural relations holding among the constituents of the proposition must show what it predicates of what. 6 So far, so good. What structural features of a proposition do show what is predicated of what? Consider the proposition expressed by (3), the constituents of which are the relation identity together with the relation diff erence occurring twice over.

3. Identity is diff erent from diff erence.

In this proposition, the diff erence relation is predicated of identity and diff erence . What feature of the proposition shows this? Consider some candidates for being that proposition.

4a. < diff erence, > b. {{diff erence}, {diff erence, {{identity}, {identity, diff erence}}}} c. < , diff erence > d. {{{identity},{identity, difference}},{{{identity},{identity, difference}}, diff erence}} e . < d i ff erence, > f. {diff erence, {diff erence, {diff erence, {diff erence, identity}}}} g. < , diff erence > h. {{{difference}, {difference, identity}},{{{difference},{difference, identity}}, diff erence}}

Any of these could be used as a formal model of the Russellian proposition expressed by (3), as could any number of tree structures, two of which are pictured in (5).

5 It is worth contrasting this example— A diff ers from B —with Russell’s earlier example— Socrates is human. In the earlier case, Russell was inclined to import something he called “a verb” corresponding to the copula into the proposition. Th e function of this new element was to really relate, via assertion or predication, the man Socrates to the concept humanity. However, in the present case we have no copula, but rather what is grammatically already a verb—“diff ers.” Th is compounds Russell’s problem, since here the job of relating the constituents of the proposition falls to the relation diff erence itself. And how can diff erence do this job unless A really does diff er from B? Russell has no answer. My suggested revision on Russell’s behalf assigns the job of relating propositional constituents to one another (in a way that forms a representational whole) not to something that is itself a propositional constituent, but rather to the structural relationships the constituents bear to one another in the proposition. 6 Frege’s answer—that some constituents of propositions are essentially predicative (and so cannot themselves be subjects of predication)—was regarded by Russell with deep suspicion, believing it to lead to “inex- tricable diffi culties.” (Section 49 ofTh e Principles of Mathematics. ) His idea, developed in that section and following, is essentially that such a doctrine is self undermining. Since to state it we have to be able to refer to purely predicative propositional constituents and predicate properties of them, stating the doctrine requires recognizing propositions in which they occur not predicatively, but as predication targets. Th at said, it must be admitted that Russell himself found it diffi cult to avoid such undermining himself. See, for example, section 52 of Th e Principles of Mathematics.

King020513OUK.indd 29 11/23/2013 12:58:35 PM 30 SCOTT SOAMES

5a. Prop

difference

identity difference

5b. Prop

identity difference difference

Which of the structures of the sort illustrated by (4) and (5) is the proposition expressed by sentence (3)? Expressed in this direct and uncompromising way, the question is absurd. Th e problem is not that any of these could serve, and hence that there is no determinate answer. Th e problem is that it is hard to see how any formal structure of this, or any similar, sort could possibly be the proposition we are looking for. Proposition (3) is something that represents the relation of identity as being diff erent from the relation of diff erence by virtue of the fact that diff erence is predicated of the two relations. But there is nothing in the sets or sequences of (4), in the tree structures of (5), or in any other formal structure we might construct to organize the constituents of the Russellian proposition which, by its very nature, indicates that anything is predicated of anything. Hence, there is nothing inherent in such structures that makes them representational, and so capable of being true or false. Structures of this sort can’t possibly be the primary bearers of intentionality. We could , if we wished, adopt rules that would allow us to read off the needed information about predication from such structures, and so interpret them. To do this would be to endow the structures with representational meaning or content, thereby making them bearers of truth and falsity. Th is wouldnot , however, make them Russellian propositions. For Russell, propositions are not things that have meanings , or get interpretations from us; they are the meanings that sentences come to express when we initially endow them with meaning, or that we discover when we come to understand sentences previously so endowed. Th e real problem with Russell’s conception of propositions is that it makes it diffi cult or impossible to answer a question that can’t be avoided: “ What makes propositions representational, and, hence, bearers of truth, objects of the attitudes, and meanings of sentences?” Frege on propositional unity Frege had his own problem with “the unity of the proposition.” He too had to “unify” the constituents of a proposition without either adding a further constituent relating the others, or relying on a mysterious, unspecifi ed structural confi guration to do the

King020513OUK.indd 30 11/23/2013 12:58:35 PM THE INADEQUACY OF TRADITIONAL CONCEPTIONS 31

job. It is easy to understand his predicament. According to the traditional conception, all there is to a structured proposition is its structure and constituents . Since it will not solve the problem to appeal to set theoretic, hierarchical, or any other formal structure we have an independent conception of, it makes sense to reexamine the constituents. Although we know that predication must come into the story somehow, we also know that adding predication as an extra constituent won’t help. So it is natural that Frege should be drawn to the idea that some constituents are inherently predicative—in the sense of always occurring in a proposition in a predicative role—while other constituents are inherently non-predicative—and so always occurring in the complementary role of an argument to something predicative (or functional). For him, this diff erence is supposed to allow predicative constituents—which are “unsaturated” and so in need of completion by something else—to combine with either non-predicative or higher-order predicative constituents to form complete representational unities, while disallowing other combinations. Frege treated this propositional story as the analogue to his story about sentences, according to which predicate expressions—whether they be simple constants or complex (open) formulas—are expressions with gaps, which when fi lled in by singular terms yield complete sentences. Th e worry, of course, is that this story of constituents with holes to be fi lled by pieces designed to fi t seems to be more mystery and metaphor than genuine explanation. Frege expresses his doctrine by saying that it is impossible for the sense of a predicative expression to be the same as the sense of a singular term, or Fregean “proper name.” He argues for this by noting that substitution of “proper names” (singular terms) for predicates in sentences doesn’t preserve sense. His arguments fail because they neglect the possibility that substitution changes grammatical structure, which in turn may have semantic signifi cance. So, in addition to being mysterious, his so-called “solution” isn’t independently motivated. Moved by the idea that a sense is nothing more than a mode of presentation of a referent, he makes his position worse by extending the distinction between predicative and non-predicative senses to a distinction between predicative and non-predicative referents—i.e. between predicative “concepts” (functions from objects to truth values) and non-predicative “objects.” Th is leads to the dis- astrous conclusion that the concept horse is not a concept, and more generally, to the thesis that the referent of a predicate can never be designated by a singular term, and in that way made the subject of further predication. It also contradicts his own analysis of quantifi cation as predication of a higher-order quantifi cational concept of the referent of the predicative expression to which the quantifi er is attached . Simply put, the thesis that the referent of a predicate can’t be designated by any singular term is self-refuting.7 At this point, we have reached an impasse. If propositions are to be structured entities that represent things as being certain ways, and so have truth conditions, they must somehow predicate some things of other things, which means that some constituents

7 For detailed discussion, see chapter 2 of Soames (2010), also chapter 2 of Soames, Volume 1 of Th e Analytic Tradition .

King020513OUK.indd 31 11/23/2013 12:58:36 PM 32 SCOTT SOAMES

of propositions must play predicative roles. Of course, entities of certain kinds can never be predicated of anything; but entities of other kinds can. Since those that can may themselves be subjects of further predications, Frege’s idea of inherently predicational constituents can’t solve the problem of propositional unity. Th is brings us back to the Russellian idea of a structure of constituents, at least one of which is predicated of the others, even though that which is predicated can itself be the target of other predications.

Th e nature of the problem Th e problem we need to solve—which neither Frege nor Russell had a way of solving— is not to fi nd some relation born by the constituents of a proposition to one another that “holds them together” as parts of a single complex entity; the problem is to explain the intentionality of propositions. Th e former, misconceived, problem stems from the idea that for any complex entity there must be some relation in which its parts stand by virtue of which they are all parts of a single thing. If this idea is correct, then there is such a relation involving being members of a set the only elements of which are a, b, and c that “unites” the set {a,b,c}; there is another relation involving membership and order that “unites” the ordered triple , and there is still another “uniting” relation involving hierarchical domination and linear order that “unites” the tree structure:

One might, of course, wonder what these relations amount to, and try to determine whether they have informative analyses. But since the same questions arise for all complex entities, there is no special “uniting” problem of this sort for propositions. What is special is that propositions must be—inherently and without further interpretation by us —capable of being true or false. Since it would seem absurd to characterize any set, sequence, or abstract tree as inherently representing things as being certain ways—and so as being true or false—the idea that propositions are any of these structures is a non-starter. Of course, propositions can’t be sentences either, since it is only by virtue of expressing propositions that sentences are supposed to be bearers of truth conditions themselves. So the problem remains. In my opinion, the key to solving it is to recognize the obvious fact that predication is something that agents do. Properties don’t predicate themselves of anything; nor, unless we have it explained to us, do we understand what it is for a complex of which various properties are constituents to predicate one of them of the others. Th is is what Frege and Russell were up against. Th ey needed predication to make sense of propositions, but their conception of propositions made it impossible for them to

King020513OUK.indd 32 11/23/2013 12:58:36 PM THE INADEQUACY OF TRADITIONAL CONCEPTIONS 33

fi nd appropriate agents for the needed predications. Th e solution to their problem is to retain their idea of propositions as structurally complex entities that are inherently intentional, and hence the bearers of truth conditions, while giving up their idea that propositions are the primary bearers of intentionality. Instead of explaining the intentionality of the cognitive activity of agents in terms of an imagined conceptually prior intentionality of the propositions they entertain, we must explain the intentionality of propositions in terms of the genuine conceptually prior intentionality of the cognitive activity of agents who entertain them. Th is can’t be done on the traditional Frege-Russell model of robustly platonic propositions passively apprehended by agents. It also can’t be done by tying propositions too closely to agents. In my view, we need a conception of propositions that (i) recognizes unentertained propositions, including the truth or falsity of propositions at world states at which no propositions are entertained, while (ii) explaining the intentionality of propositions in terms of the intentionality of the cognitive acts of possible agents who entertain them. Th is will be my approach in chapter 6.

Th e possible worlds conception of propositions Th ough the possible worlds approach suff ers from a plethora of problems of its own, it too suff ers from a version of the problem of the unity of the proposition . I begin with it.

Possible worlds and the unity of the proposition Th e most general form of the problem is to explain how propositions—whether essentially structured or not—manage to represent the world, and so have truth conditions. Although many things have truth conditions—including sentences, utterances, and mental states—the standard explanation for this is that they do so because they express propositions that are inherently representational, and so have truth conditions independent of us . Th e mystery is how any entities could satisfy this condition. To be sure, if there are propositions, then sentences and utterances are representational because agents use them to express propositions. But recognizing this only makes it more urgent that we explain how propositions get their intentional properties. Th e problem with all traditional theories—including those invoking possible worlds (or world-states)—is that they don’t, and can’t, do this. Since the entities these theories identify as propositions are not inherently intentional, they can come to have truth conditions only if we decide to interpret them in certain ways. But this undermines the role they are supposed to play in theories of language and mind. Propositions are the meanings of (some) sentences, as well as being what we typically mean by uttering (declarative) sentences; they are not themselves things we endow with meaning by interpreting them . With this in mind, consider the view that propositions are sets of possible worlds. Th e problem is immediate. No set is intrinsically representational, no matter what its

King020513OUK.indd 33 11/23/2013 12:58:37 PM 34 SCOTT SOAMES

members. What does the set containing worlds 1, 2, and 3 represent? Is it true or false? Th ese questions are bizarre. If we wanted, we could use the set to represent the actual world as being in the set—and so to make the claim that no world outside the set is actual. But we could equally well use it to represent the actual world as not being in the set—and so to make the claim that no world inside the set is actual. Independent of interpretation by us, the set doesn’t represent anything, doesn’t make any claim, and so doesn’t have truth conditions. What about the function f that assigns worlds 1-3 truth and all others falsity? On the face of it, f would seem not to be a proposition either, for surely if sets aren’t inherently representational then neither are their characteristic functions. Suppose we replaced the values truth and falsity with the North and South Poles. What does the function that assigns worlds 1-3 the North Pole and all other worlds the South Pole represent? Independent of interpretation by us, it doesn’t represent anything. Why, then, should the original function assigning truth and falsity be representational? What, aft er all, are truth and falsity but properties we grasp primarily through their application to propositions? But surely, if propositions are needed to illuminate truth and falsity, they can’t be among the building blocks for constructing propositions. Th e illusion that propositions can reasonably be construed as functions from worlds to truth values is fed by a natural way of thinking about worlds—as maximal states (properties) the universe might conceivably be in (or have). 8 Under this conception, each assignment of a truth value to a world-state w can be correlated with the proposition that predicates w of the universe, and is thus true or false depending on whether or not w is the maximal state the universe really is in. A function from world-states to truth values can then be associated with the (possibly infi nite) disjunction of the propositions correlated with its assignments of truth to world-states. Th is picture is fi ne, if one already has independent accounts of propositions and truth . But it doesn’t provide a foundational account of what propositions are. Instead, it presupposes propositions by appealing to them as things that predicate properties, in this case a world-state, of other things, in this case the universe—while further presupposing that propositions are true when their predication targets have the properties predicated of them. When (wrongly) taken as a foundational story, this account also (i) inverts the conceptual order relating propositions and truth by taking the latter to be an unexplained primitive involved in the construction of the former, while (ii) merely correlating functions from world-states to truth values with propositions, without providing a basis for identifying the two. In addition, the story is inconsistent with

8 S e e S t a l n a k e r , Inquiry , Cambridge : MIT Press , 1984 , and Ways a World Might be , Oxford: Oxford University Press, 2003. See also, pp. 200–209 of Soames Reference and Description, Princeton and Oxford, 2005 ; Soames, “Actually,” in Mark Kalderon, ed., Proceedings of the Aristotelian Society, supplementary volume, 81, 2007, 251–277, reprinted in Soames , Philosophical Essays, Vol. 1: Natural Language: What it Means and How we Use it , Princeton and Oxford : Princeton University Press , 2009 ; also chapters 5 and 6 of Soames , Philosophy of Language , Oxford and Princeton: Princeton University Press, 2010 .

King020513OUK.indd 34 11/23/2013 12:58:37 PM THE INADEQUACY OF TRADITIONAL CONCEPTIONS 35

the parallel attempt to identify properties with functions from world-states to sets of individuals. Since it is part of the story that world-states are properties attributed to the universe, these properties can’t be functions from world-states to sets. But if properties aren’t reducible to functions from world-states to extensions, there is no reason that propositions should be reducible to functions from world-states to their extensions. So far I have made two negative points: (i) the possible-worlds conception of propositions fails to explain how propositions can be representational, and so have truth conditions, and (ii) it wrongly takes what it calls “worlds” and “truth values” as unexplained primitives from which it tries to construct properties and propositions, when in fact properties and propositions are needed to explain and illuminate both truth and worlds as world-states. I now turn to related points about the conceptual foundations of possible-worlds semantics. Th e conceptual foundations of possible-worlds semantics Possible-worlds semantics standardly identifi es propositions with functions from worlds (or world-states) to truth values. Th ese are intensions of sentences, derived from intensions of their sub-sentential constituents, which are functions from worlds (world-states) to extensions. Since necessarily equivalent propositions are identifi ed, “possible-worlds propositions” are coarse grained, which makes them ill-suited to be objects of hyperintensional attitudes. But the fundamental problem starts much earlier. Th e possible-worlds analysis of propositions prevents us from answering the questions: How do possible-worlds semantic theories provide any information about meaning? In particular, how do we learn anything about the meaning of S from a statement of the conditions under which S is “true at a world-state?” Details aside, the correct answer goes roughly like this. For S to be meaningful is for S to express a proposition that predicates something of something, and so represents it as being a certain way. Th e truth conditions of S follow from the truth conditions of the proposition S expresses; if S expresses a proposition that represents A as being B (and nothing more), then S is true iff A is B (provided that the property B is defi ned for A). 9 Th is explanation invokes monadic truth . How do we get a grip on that? Just as one can get a sense of the property designated by “red” by being told, “Something is red if it looks like this (pointing at some red things), and it isn’t red if it looks like those (pointing at non-red things),” so one can get a sense of the property designated by “true” by being told, “Th e proposition that the earth is round is true if the earth is round, and it isn’t true if the earth isn’t round, and so on.” We gain further information by learning that the proposition that p is true is necessarily and a priori equivalent to p, and that any warrant for asserting, believing, or denying one is warrant for taking the same attitude toward the other. Complications aside, truth’s connection to meaning is given by our a priori knowledge that a sentence that means (expresses the proposition)

9 In this paragraph, as in the previous paragraph, “S” is a metalinguistic variable. “A” and “B” are used as schematic letters.

King020513OUK.indd 35 11/23/2013 12:58:37 PM 36 SCOTT SOAMES

that so-and-so is true iff so-and-so. Th is allows us to derive information about meaning from statements about truth conditions. For example, we derive ⌜“S” doesn’t mean that R ⌝ from ⌜ “S” is true iff Q⌝ , when Q and R are known to diff er in truth value (e.g. when they are inconsistent). 10 Modalized truth conditions provide further information about meaning. 11

NT. Necessarily, the proposition that S is true iff S.

Assuming the usual connection between modal operators and world-states, we derive (6a,b). 6a. ∀w [at w (the proposition that S is true iff S)] b . ∀w [at w, the proposition that S is true iff at w, S)] Th e truth predicate is monadic, “at w” is a sentential operator with the force if w were instantiated it would be the case that , and (6c) and (6d) come to the same thing. 6c. ∀w [at w, the proposition that Plato philosophized is true iff at w, Plato philosophized] d . ∀w [the proposition that Plato philosophized is true at w iff Plato philosophized at w ] Th is is the starting point for understanding statements like (7) about the truth conditions of sentences made by possible worlds semantics. 7. Th e English sentence “Plato philosophized” is true at w iff Plato philosophized a t w . Claim (7) carries information about the meaning of the sentence it mentions in virtue of our antecedent understanding of what it is to be true, and what it is to philosophize, at a world-state. To say that x philosophizes at w is to say that if w were instantiated, then x would philosophize. What is it for S to be true at w? Th e possible-worlds semanticist can’t quite say that for the English sentence S to be true at w is for it to be such that if w were instantiated, then S would be a true sentence of English (i.e. one which expresses a true proposition ). In possible-worlds semantics, S can be true at w even if S means nothing at w, or means something diff erent from what it actually means. Th is shows that the dyadic truth predicate of possible worlds semantics is a technical substitute for our ordinary notion. Using our ordinary notion, we say that S is true at w iff at w, S expresses a proposition that is true. Since what S could have meant (or expressed) is no help in illuminating what S actually means (expresses), the possible-worlds semanticist doesn’t follow us in this. Instead, he dispenses with the ordinary notion

10 Here, “Q” and “R” are metalinguistic variables along with “S.” 11 “S” is used as a schematic letter in NT and (6), while being used as a metalinguistic variable later in the paragraph.

King020513OUK.indd 36 11/23/2013 12:58:37 PM THE INADEQUACY OF TRADITIONAL CONCEPTIONS 37

of sentential truth, and introduces the technical predicate of sentences “is true at w” to mean the proposition p that S actually expresses is true at w —otherwise put: S expresses a proposition at @ and that proposition is true at w. In short, the dyadic truth predicate of possible-worlds semantics is parasitic on the prior notions: the proposition actually expressed by a sentence and the monadic property truth of propositions. It is by taking these for granted that we extract useful information about meaning from the truth conditions provided by such a semantic theory. We derive ⌜”S” doesn’t mean that R ⌝ from ⌜ ∀w “S” is true at w iff at w, Q⌝ when Q and R aren’t necessarily equivalent—tacitly assuming ⌜if “S” means that P, then necessarily the proposition “S” actually expresses is true iff P⌝ . 12 Although this doesn’t identify what S does mean, it does so up to necessary equivalence, thereby providing information about meaning that restricts the range of acceptable alternatives. Without the prior notions of truth and propositions here employed, even this limited information about meaning extracted from the semantic theory would be lost. Th is is the heart of the problem with the analysis of propositions as functions from world-states to truth values. If one takes world-states and truth values to be unexplained primitives, with the goal of using them to provide reductive analyses of properties, propositions, and meaning, then one can’t use the method just given for extracting claims about meaning from possible-worlds semantic theories. Without prior accounts of propositions, truth, and the connection between meaning and truth, the theorems of such a semantic theory won’t carry any information about meaning. Having denied themselves such accounts, proponents of this approach, like Bob Stalnaker in his book Inquiry, are reduced to telling us that any two objects—0 and 1, or the North and South Poles—can serve as values assigned by the intensions of sentences to unexplained indices called “worlds.” 13 But, surely, we don’t learn anything about the meaning or genuine truth conditions of a sentence by associating it with a function that assigns some unexplained indices the North Pole and others the South Pole. As I explain more fully in chapter 6, the proper explanation starts with the idea that agents predicate properties of objects in cognition and perception, thereby entertaining propositions. Agents do this before they have the concept proposition. Focusing on similarities and diff erences in our experience, agents like us are able to acquire the concept, aft er which we are able to make propositions objects of thought and subjects of predication. Th is, in turn, allows us to acquire the concept of truth. Given truth, properties can be conceptualized as things true of other things. With the concepts truth, property, proposition and modality (what could be but isn’t) in place, we can characterize world-states as ways for things to be—maximally informative properties that the world could have had. Such a world-state w can be defi ned as the property of

12 Again, “P,” “Q,” “R,” and “S” are metalinguistic variables. 13 On p.2 we are told, “Th ere are just two truth values—true and false. What are they: mysterious Fregean objects, properties, relations of correspondence and non-correspondence? Th e answer is that it does not matter what they are; there is nothing essential to them except that there are exactly two of them.”

King020513OUK.indd 37 11/23/2013 12:58:37 PM 38 SCOTT SOAMES

making true a set w* of basic propositions that tell a complete world-story. Roughly put, a proposition p is true at w iff p is an a priori consequence of w*. So, we can come to know that p is true at w by deriving p from w*. As for the actual world-state @, we can come to know p to be true at @ , given knowledge of p, by noting that since p is true, it must be true at this very world-state —the one that is instantiated. Th is is a satisfying foundational picture of how the notions proposition, property, truth, and world- state are related to one another, and to cognition, including how we are able to know the things about them that we do. It is also the conceptual background needed to extract useful information about meaning from possible-worlds semantic theories. Th e problem for the philosophically ambitious proponent of such theories— who wants to tell us what propositions and properties are, but not what his primitive dyadic notion of truth amounts to—is that he can’t adopt this foundational picture, since to do so he would have to invoke a rival conception of propositions as bearers of monadic truth, antecedent to what his theory provides. His project of using what he calls “truth” and “worlds” to defi ne propositions prevents him from, at the same time, taking real propositions and ordinary truth to be prior to those supposed primitives. Since he typically characterizes properties as functions from worlds to sets of individuals, he also can’t presuppose an antecedent conception of worlds as properties. With no way of explaining the crucial notions taken to be primitive, the reductive possible-worlds semanticist lacks the conceptual resources needed to explain how his own elaborate technical machinery yields any information whatsoever about the genuine semantic properties of sentences and other expressions. Coarse-grainedness of possible-worlds propositions: A corollary Th ese are the fundamental diffi culties that defeat the conception of meaning and propositions given by the philosophically ambitious possible-worlds semanticist. By contrast, the long-celebrated coarse-grainedness problem—which makes possible-worlds propositions ill suited to be objects of hyperintensional attitudes—merely adds a further debilitating corollary: sets of world-states can’t even model propositions . Historically, the three main attempts to deal with this problem in intensional semantics have been (i) to substitute fi ner-grained circumstances for world-states, (ii) to distinguish the proposition semantically expressed by S from what one asserts by uttering S, and (iii) to adopt a 2D semantic theory that associates pairs of coarse-grained propositions with sentences. Th e high-water mark for the fi rst strategy was the system of situation semantics developed by Jon Barwise and John Perry. 14 It failed because the hyperintensionality problem is reconstructable for situations, and indeed for all theories satisfying the following conditions, no matter how fi ne-grained truth-supporting circumstances are taken to be.

14 Situations and Attitudes , Cambridge: MIT Press, 1983.

King020513OUK.indd 38 11/23/2013 12:58:37 PM THE INADEQUACY OF TRADITIONAL CONCEPTIONS 39

A1. Th e semantic content of a sentence/formula S (relative to a context C and assignment A of values to variables)—a.k.a. the proposition S semantically expresses (relative to C,A) = the set of circumstances E such that S is true at E (relative to C, A). A2. (i) ⌜P & Q ⌝ is true at a circumstance E (relative to C,A) iff both conjuncts are true at E (relative to C,A). (ii) ⌜ූ x Fx ⌝ is true at E (relative to C,A) iff Fx is true of some object o at E (relative to C,A). (iii) ⌜An F is G⌝ is true at E (relative to C,A) iff for some object o at E, Fx and Gx are jointly true of o at E, (relative to C,A).

A3. If S 1 and S 2 are nonintensional sentences/formulas with the same grammatical structure, which diff er only in the substitution of constitutents with the same

semantic contents (relative to C,A), then the semantic contents of S 1 and S2 will be the same (relative to C,A). A4. Semantic contents of variables, indexicals, or names (relative to C,A) are their referents (relative to C,A). A5. Propositional attitude ascriptions report relations to the semantic contents of their complement clauses (relative to C,A); so ⌜ x v’s that S ⌝ is true at E (relative to C,A) iff at E, the value of “x” relative to A bears R to the semantic content of S (relative to C,A). When v is “believes,” R is the relation believing , when v is “says” or “asserts” R is the relation saying or asserting, and similarly for other attitude verbs. A6. Many attitude verbs, including “say,” “assert,” “believe,” “know,” and “prove” dis- tribute over conjunction. For these verbs ⌜x v’s that P & Q ⌝ is true at E (relative to C,A) only if ⌜x v’s that P ⌝ and ⌜ x v’s that Q ⌝ a r e t o o . Assumption A1 is the truth conditional construction of propositions to be refuted, while (i)-(iii) of A2 are uncontroversial corollaries of that approach. Although some technical counterexamples to A3 can be constructed, they don’t come into play here, so A3 can be granted. A4 is accepted by nearly everyone for variables, while being widely (though not universally) accepted for names and/or indexicals. (Th e reductio of propositions as sets of truth-supporting circumstances based on these assumptions requires that A4 holds for at least one of these categories of terms.) While A5 is arguable, the chief alternatives, which are metalinguistic in nature, are highly problematic, and have found little currency among theorists pursuing the truth-supporting-circumstance approach to meaning. 15 Moreover, rejecting A5 constitutes a diff erent strategy for dealing with hyperintensionality than the strategy, presently being considered, of making truth-supporting circumstances more fi ne-grained; so A5 can be accepted. Finally,

15 For criticism of infl uential metalinguistic accounts of belief ascriptions see chapter 7 of Soames, Beyond Rigidity , New York: Oxford University Press, 2002.

King020513OUK.indd 39 11/23/2013 12:58:37 PM 40 SCOTT SOAMES

A6 cannot be denied. Th us, it is natural to take the unacceptable results generated by A1-A6 to constitute a reductio ad absurdum of the strategy of dealing with attitude ascriptions by making truth-supporting circumstances more fi ne-grained than sets of possible world-states. Here is one version of the reductio . (i) Th ere is a planet x seen in the morning sky and a planet y seen in the evening sky such that the ancients believed that x was seen only in the morning sky and y was seen only in the evening sky. (ii) Th e (unique) planet seen in the morning sky = the (unique) planet seen in the evening sky. (iii) Th ere is a planet x such that the ancients believed that x was seen only in the morning sky and x was seen only in the evening sky . ( i v ) Th ere is a planet x such that the ancients believed that x was seen only in the morning sky and x was seen only in the evening sky and there was some one thing that was both seen only in the morning sky and seen only in the evening sky . (v) Th e ancients believed that there was some one thing that was both seen only in the morning sky and seen only in the evening sky. Since (v) is empirically false, while (i) and (ii) are empirical truths, we must reject any set of semantic principles according to which the truth of (v) is guaranteed by the truth of (i) and (ii). But, the truth of (iii) follows from the truth of (i) and (ii) by A1, A2, A3, A5, and A4 for variables; the truth of (iv) follows from that of (iii), since (by A2) the complement clauses of (iii) and (iv) are true at the same truth-supporting circumstances, and so (by A1) express the same proposition; and the truth of (v) follows by the undeniable A6 from the truth of (iv). Th us, making truth-supporting circumstances fi ne-grained won’t solve the problem posed by propositional attitudes for theories that take sets of such circumstances to model propositions.16 Th e high-water mark of the second strategy for dealing with hyperintensionality used sets of metaphysically possible world-states to model propositions, while combining the distinction between semantic and assertive content with the pragmatic 2D strategy of “diagonalization” to breathe informativeness into utterances of necessary truths. 17 Since the model recognizes only one necessary proposition, which every agent already believes, it must identify the information communicated by an utterance of any sentence semantically expressing that proposition with some contingent proposition. Th e technique for doing this is an operation called “diagonalization” that takes as

16 For details, see Soames , “Direct Reference, Propositional Attitudes, and Semantic Content,” Philosophical Topics 15 , 1987 , 44–87 ; and “Why Propositions Can’t be Sets of Truth-Supporting Circumstances,” Journal of Philosophical Logic 37 , 2008 , 267–276 , both reprinted in Philosophical Essays, Vol. 2, Princeton: Princeton University Press, 2009. 17 Robert Stalnaker , “Assertion,” in Peter Cole , ed., Syntax and Semantics, vol. 9 Pragmatics , New York : Academic Press , 1978 ; reprinted in Stalnaker, Context and Content: Essays on Intentionality in Speech and Th ought , New York : Oxford University Press , 1999 , 78–95 .

King020513OUK.indd 40 11/23/2013 12:58:38 PM THE INADEQUACY OF TRADITIONAL CONCEPTIONS 41

input (i) the meaning of a sentence S that is uttered, and (ii) the possible world-states compatible with everything known or presupposed in the conversation at the time of utterance. Th ese, in turn, are supposed to generate a contingent proposition that is true (false) at any world-state w in (ii) (and so compatible with all presuppositions in the context of utterance) iff the proposition S would express, if w turned out to be the actual world-state, is true (false) at w. Th is, it is alleged, is the contingent content asserted and informatively communicated by the utterance of S. Th e problem with this strategy is that there are many crucial cases in which it cannot generate the required content. It will generate such a content only if there are world-states w compatible with everything known or presupposed by the conversational participants at which the proposition that would be expressed by S, if w were actual, is false at w. With this in mind, consider a case in which I say to you ⌜He/she/it/that is F ⌝, rigidly designating an individual x and predicating a property of x that is, in fact, one of x’s essential properties (without which x could not exist). Suppose further that the situation in which I make this informative remark is one in which both of us are intimately acquainted with x, and have many singular thoughts about x—including the thought that x exists, is the subject of our conversation, and is the individual about whom we are speaking. In this scenario, every world-state in (ii) is one in which S would express a proposition attributing F-hood to x (and nothing else). Since x exists at all these world-states and F-hood is essential to x’s existence, the propositions that would be expressed, if any of them were actual, are true at those states. Th us, we are left with the possible-worlds proposition that is true at all relevant world-states, which means that diagonalization has failed.18 As indicated above, the source of the problem is the inability of the model to accommodate attitudes to singular propositions that predicate essential properties of objects—even though its chief proponent, Robert Stalnaker, recognizes both that some properties are essential in the relevant sense, and that the model requires singular thought about world-states. 19 Since these states are specifi ed in terms of objects and properties, this should mean that singular thought about world-states bottoms out in singular thoughts about objects and properties. Being at the center of the model, such singular thoughts can’t legitimately be excluded in cases, like those just discussed, in which they lead to indigestible results. Th e problem can be shown to persist even when epistemically possible world-states replace metaphysically possible world-states as the truth-supporting circumstances.20 Th us, combining this failed strategy with the fi rst failed strategy for dealing with hyperintensionality doesn’t save the ability to model propositions within the possible-worlds framework.

18 For discussion, see Soames , “Understanding Assertion,” in J. Th omson and A. Byrne , eds., Content and Modality , Oxford : Clarendon Press , 2006 ; reprinted in Philosophical Essays, Vol. 2. 19 Stalnaker’s reply to my argument in “Understanding Assertion” is also in Content and Modality . 20 Th is result is also established in “Understanding Assertion.”

King020513OUK.indd 41 11/23/2013 12:58:38 PM 42 SCOTT SOAMES

Th e high-water mark for the fi nal strategy for dealing with hyperintensionality while retaining propositions as sets of possible world-states is a semantic view I have previously called “strong two-dimensionalism”—suggested by certain infl uential writings of Frank Jackson and David Chalmers. 21 Th e view was prompted in part by the problem posed by the Kripkean necessary a posteriori for the view that propositions are sets of metaphysically possible world-states. Since, on this view, there is only one necessary proposition, which everyone knows a priori, no proposition can be both necessary and knowable only a posteriori. Th e alleged “illusion” to the contrary is in failing to recognize that context-sensitive sentences are associated not with one proposition but with two. One of these propositions, called “the primary intension” of S, is the set of possible world-states at which S expresses a truth. Th is proposition is true at w iff S’s Kaplanian character maps w (considered as a context) onto a proposition that is true at w. In this sense, the primary intension of S is more or less equivalent to the claim that S’s Kaplanian character expresses a truth. Th e other proposition associated with S, called “the secondary intension of S” at a given context C, is the proposition expressed by S at C, which is the set of possible world-states w such that S is true relative to C,w. Given that primary and secondary intensions come apart when S is context sensitive, strong two-dimensionalists analyze names and natural kind terms as context-sensitive, rigidifi ed descriptions ⌜the actual D ⌝. Since (8a) and (8b) express truths in precisely the same contexts, their primary intensions are identifi ed, even though their secondary intensions will, typically, be diff erent. 8a. Th e D is F (if the D exists) b . Th e actual D is F (if the actual D exists) Since (8b) is the analysis of (9), when N is a name or natural kind term, the primary intension of (9) is the primary intension of (8a) while its secondary intension is the secondary intension of (8b). 9. N is F (if N exists) Th us, when F expresses an essential property of the individual or kind designated by N, and ⌜the D ⌝ is nonrigid, the secondary intension of (9) is necessary, and knowable a priori, while the primary intension of (9) is typically contingent, and knowable only a posteriori. In this way, the thesis that propositions are sets of metaphysically possible world-states while also being objects of the attitudes was defended against the challenge posed by the Kripkean necessary a posteriori. Th is defense depends on taking the secondary intensions of sentences to provide the arguments of modal operators and the primary intensions of those sentences to

21 Chalmers (1996), Th e Conscious Mind , Oxford University Press; Jackson , From Metaphysics to Ethics , Oxford University Press , 1998 . In Reference and Description, I reconstruct several precise and explicit versions of two-dimensionalism based on their work. Th e simplest, and, I believe, truest to the original motivat- ing idea, is the one I there call “strong two-dimensionalism.”

King020513OUK.indd 42 11/23/2013 12:58:38 PM THE INADEQUACY OF TRADITIONAL CONCEPTIONS 43

provide the arguments of epistemic operators (“assert,” “believe,” “know,” etc.) Th e incorrectness of such semantics can easily be shown by embedding a single sentence under both modal and epistemic operators. When we do this, we fi nd that in natural language such operators require a single proposition to be supplied as argument to operators of both types. For example, consider the (10a) and (10b).

10a. It is a necessary truth that if the actual husband of Stephanie Lewis was the actual author of COUNTERFACTUALS and Mary believes that the actual husband of Stephanie Lewis was the actual author of COUNTERFACTUALS, then Mary believes something true. b . I t i s a n e c e s s a r y t r u t h that if the actual husband of Stephanie Lewis was the actual author of COUNTERFACTUALS and Mary believes that the husband of Stephanie Lewis was the author of COUNTERFACTUALS, then Mary believes something true.

Although (a) is true, strong 2D semantics wrongly take it to express the same proposition as the false (10b). Th e same is true for the pair in (11).

11a. It is a necessary truth that if Mary believes that the actual husband of Stephanie Lewis was the actual author of COUNTERFACTUALS, and if that belief is true, then the actual husband of Stephanie Lewis was the actual author of COUNTERFACTUALS. b . I t i s a n e c e s s a r y t r u t h that if Mary believes that the husband of Stephanie Lewis was the author of COUNTERFACTUALS, and if that belief is true, then the actual husband of Stephanie Lewis was the actual author of COUNTERFACTUALS.

Th e sentences in (12) bring names (or natural kind terms) into the picture.

12a. Although John truly believes that n is D, had the world been in state w, n would not have been D and John would not have believed that n was D. b. Although John truly believes that the actual D is D, had the world been in state w, the actual D would not have been D and John would not have believed that the actual D was D. c. Although John truly believes that the D is D, had the world been in state w, the actual D would not have been D and John would not have believed that the D was D. Here we let o be uniquely denoted by the nonrigid description ⌜the D⌝ , we let n be a name of o, and we let the strong 2D analysis of n be ⌜ the actual D ⌝. We stipulate that ⌜John truly believes that n is D ⌝ is true (which assures the truth of what he believes). We further take w to be a world-state at which some object other than o is uniquely denoted by ⌜ the D ⌝, and in which John does not believe of o that it has the property expressed by D, though he does believe the proposition expressed by ⌜the D is D⌝ .

King020513OUK.indd 43 11/23/2013 12:58:38 PM 44 SCOTT SOAMES

Given all this, (12a) should be true, even though strong 2D semantics wrongly assimilates (12a) to (12b), which is in turn assimilated to the false (12c). Th is is the same problem that was illustrated in (11), but put in a diff erent form. Finally, I extend the point to include the relationship between proper names and variables of quantifi cation. Here we let n be a name, and F be a predicate such that the truth of ⌜ n is F⌝ guarantees the truth of ⌜ Th ere is such a thing as n⌝ . I further assume that for any context C and world-state w, if the following (a) sentences are true at C,w, then the (b) sentences are too. 13a. John truly believes that n is F, but had the world been in state w, n would not have been F. b . Th ere is an x such that John truly believes that x is F, but had the world been in state w, x would not have been F. 14a. John truly believes that n is F, but had the world been in state w, John would not have believed that n was F. b . Th ere is an x such that John truly believes that x is F, but had the world been in state w, John would not have believed that x was F. Putting these facts together we see that if (15a) is true at C,w, then (15b) must also be true there. 15a. John truly believes that n is F, but had the world been in state w, n would not have been F and John would not have believed that n was F. b . Th ere is an x such that John truly believes that x is F, but had the world been in state w, x would not have been F and John would not have believed that x was F. Strong 2D semantics misses the entailment of (15b) by (15a), Th is results from a pair of facts: (i) that whereas the primary and secondary intensions of variables (relative to assignments) are identical, the primary and secondary intensions of names (and natural kind terms) are diff erent in 2D semantics, and (ii) that primary intensions are taken to be the arguments of propositional attitude verbs, but not modal operators, in the system. Th e lesson of this argument is easily summed up. Let n rigidly designate o. Th en for any world-states w, ⌜ n is F⌝ is true at w iff at w, o has the property expressed by F. So, it ought to be the case that for any world-state w, ⌜ John’s belief that n is F ⌝ stands for a belief about o—one that comes out true when evaluated at w only if at w, o has the property expressed by F. Surprisingly, this is not so on strong two-dimensionalist semantics, which wrongly allows something other than what we might call ⌜the fact that n is F ⌝ to verify (at w) the truth of what strong two-dimensionalism identifi es as ⌜ the belief that n is F⌝ . In and of itself, this refutation of strong two dimensionalism doesn’t refute other varieties of two-dimensionalism. But those either don’t identify propositions with sets of world-states, or haven’t been fully enough specifi ed in order to determine what view of propositions they presuppose. Hence, they need not be considered here.22

22 See pp. 267–282 of Reference and Description for a battery of arguments, including those given here, against strong semantic two-dimensionalism. Other versions of two-dimensionalism are defi ned and rejected on pages 290–325.

King020513OUK.indd 44 11/23/2013 12:58:38 PM P A R T I I Th ree Th eories of Propositions

King020513OUK.indd 45 11/23/2013 12:58:38 PM King020513OUK.indd 46 11/23/2013 12:58:38 PM 4

Naturalized Propositions

J e ff rey C. King

Th at propositions represent— that is, that they have truth conditions—is something that needs to be explained. According to “classical” conceptions of propositions of the sort championed by Frege and Russell, propositions have truth conditions by their very natures and independently of minds and languages. But no one has ever been able to explain how anything could have truth conditions by its very nature and independently of minds and languages. Th us, the classical conception of propositions leaves unexplained something very much in need of explanation. As a result, in King [2007, 2009] I rejected classical conceptions of propositions. To be clear, I am of course not saying that any time a thing has a property, there must be some explanation for how or why it has that property. I seriously doubt, for example, that there is any substantial explanation of why/how I have the property of being identical to Jeff King. But certain sorts of properties are such that we feel compelled to give an account of how/why something manages to possess them and perhaps even what the possession of them consists in. Some things in the world—names, maps, mental states, perceptual experiences and so on—represent other things. Properties of this sort—representing something else or representing it as being a certain way—are precisely the sort of properties the possession of which we rightly feel compelled to explain. Perhaps it is not entirely clear what it is that makes a property such that possession of it is something that needs to be explained. But it seems utterly clear that the property of having truth conditions is the sort of property whose possession is in need of explanation. Hence, again, since the classical conception of propositions as things that have truth conditions by their very natures and independently of minds and languages is incapable of explaining how or why propositions have truth conditions, it is unacceptable. Th ese same considerations show that propositions are neither sets (of possible worlds or anything else), nor n-tuples nor any other such things. “Formal” entities of this sort are just not the kinds of things that have truth conditions. For example, there is nothing about the 3-tuple consisting of me, the loving relation and Annie that makes

King020513OUK.indd 47 11/23/2013 12:58:38 PM 48 JEFFREY C. KING

it true iff I love Annie (as opposed to true iff Annie loves me). 1 In general, there is nothing about a set or n-tuple that determines that it is true under this or that circumstance. Having for these sorts of reasons rejected classical conceptions of propositions, as well as views according to which propositions are sets or n-tuples, in King [2007, 2009] I formulated a new account of propositions according to which they are endowed with representational powers—truth conditions—by thinking agents. I call the account an account of naturalized propositions. In formulating this theory, I assumed a broadly Russellian view of propositions on which they are complex, structured entities with individuals, properties and relations as constituents. Th e proposition that Michael swims, for example, has Michael and the property of swimming as constituents; the proposition that Barry loves Michelle has Barry, the loving relation and Michelle as constituents; and so on. As I made clear in King [2007] 2 and King [2009], the challenge in formulating this new account of propositions is to say exactly what relation holds together the constituents of a proposition; and, most importantly, why/how this complex—the proposition—consisting of the constituents standing in the relation in question manages to have truth conditions. At the time those works appeared, the account of propositions articulated in them was the only account in which there was an explanation of how/ why propositions have truth conditions. I consider this to be a very signifi cant strength of the view sketched in those works. Nonetheless, since the publication of these works, I have come to think that a slightly diff erent version of my view is superior to the one I had been defending. One of the goals of the present chapter is to articulate this new view. A second goal is to consider and respond to some objections to the view. In so doing, I hope to facilitate a better grasp of my account. Because the new view I’ll be sketching here is in many respects similar to the view in King [2007, 2009] and is the result of amending the latter in a number of ways, the best way to explain the new view is by sketching the account of propositions given in King [2007, 2009] and then explaining how the new view diff ers from this one. Idealizing a lot, let’s begin by considering the sentence “Michael swims” with the syntactic structure as follows:3

1 I made this point in King [2007] p. 8. More generally, the points just made regarding why propositions cannot have truth conditions by their very natures and independently of minds and languages, as they were classically thought to, and why no merely “formal” entity such as an n-tuple can by itself have truth conditions are covered in greater detail in King [2007, 2009 and 2012]. Many of these points are repeated by Soames in Chapter 3 of the present work. 2 Pp. 3–4, 25–26 and 59–64. 3 John Collins [2007] has recently argued that my pretending syntax is much simpler than it is for exposi- tory purposes is far from innocent, since the real complexity of syntax ends up being a problem for me. I respond to Collins below.

King020513OUK.indd 48 11/23/2013 12:58:38 PM NATURALIZED PROPOSITIONS 49

Michael swims

Call the syntactic relation that obtains between “Michael” and “swims” in the sentence here R. I call relations like R that lexical items stand in to form sentences sentential relations . It is worth pointing out, and we’ll return to this later, that the syntactic relation R itself has a certain semantic signifi cance in English. atTh is, English speakers interpret R in a certain way: they take R to ascribe the semantic value of “swims” to the semantic value of “Michael.” Th is is part of the reason that the English sentence is true iff Michael possesses the property of swimming. Further, it is a contingent matter that R is interpreted by English speakers in the way it is in the sense that there might have been a language that included the sentence 1, but whose speakers took the sentence to be true iff Michael doesn’t swim. In so doing, they would have been interpreting R diff erently from the way English speakers do. It is worth saying a bit more about the idea that English speakers interpret R and syntactic concatenation generally. Th at English speakers interpret R as ascribing the semantic value of “swims” to the semantic value of “Michael” results in the fact that they take 1 to be true iff Michael possesses the property of swimming. Similarly, when English speakers confront other cases of syntactically concatenated expressions, they spontaneously and unrefl ectively compose the semantic values of the concatenated expressions in characteristic ways. For example, when English speakers confront “brown cow” they do something like conjoin the properties that are the semantic values of the two expressions; when they confront “some man,” they do something like saturate an argument of the relation expressed by “some” with the property expressed by “man,” resulting in the (relational) property of properties that is possessed by a property A iff some man has A. Th at speakers interpret syntactic concatenation in the ways they do consists in the fact that they spontaneously and unrefl ectively compose the semantic values of the concatenated expressions in the ways described. Hence, this is how my talk of R above being interpreted by English speakers as ascribing the property of swimming to Michael should be understood. I’ll put the fact that speakers of English so interpret R by saying that R encodes ascription in English . But why do English speakers interpret syntactic concatenation in the small handful of ways they do? I suspect it will turn out that speakers of diff erent natural languages interpret syntactic concatenation in the same small handful of ways. Th is would make it a reasonable hypothesis that doing so is part of our biologically endowed language faculty. Th at this is so would make language acquisition signifi cantly easier. When encountering concatenated expressions, speakers would be hardwired to compose the semantic values of the concatenated expressions in a small handful of ways. Hence, speakers would only need to learn which way to do it in specifi c cases.

King020513OUK.indd 49 11/23/2013 12:58:38 PM 50 JEFFREY C. KING

Returning to the main theme, in virtue of the existence of the English sentence 1, there is a two-place relation that Michael stands in to the property of swimming. Th e relation is this: ___is the semantic value of a lexical item e of some language L and ___ is the semantic value of a lexical item e’ of L such that e occurs at the left terminal node of the sentential relation R that in L encodes ascription and e’ occurs at R’s right terminal node. Because we also wish to talk about the two-place relation that Michael stands in to the property of swimming in virtue of the existence of the English sentence “I swim” taken in a context with Michael as speaker, we should really suppose that in virtue of the existence of sentence 1, Michael stands in the following relation to the property of swimming (boldface indicates new additions): there is a context c such that ___is the semantic value in c of a lexical item e of some language L and ___ is the semantic value i n c of a lexical item e’ of L such that e occurs at the left terminal node of the sentential relation R that in L encodes ascription and e’ occurs at R’s right terminal node .4 Th is relation, I claimed, is the relation that holds Michael and the property of swimming together in the proposition that Michael swims. As such, I’ll call it the propositional relation of the proposition that Michael swims. As I did in King [2007, 2009], I’ll call an object possessing a property, or n objects standing in an n-place relation, or n properties standing in an n-place relation or etc. a fact . Th en the proposition that Michael swims is the fact consisting of Michael and the property of swimming standing in the two-place relation mentioned above: there is a context c such that Michael is the semantic value in c of a lexical item e of some language L and the property of swimming is the semantic value in c of a lexical item e’ of L such that e occurs at the left terminal node of the sentential relation R that in L encodes ascription and e’ occurs at R’s right terminal node . 5 Note that this fact is distinct from the fact that is Michael possessing the property of swimming. Th e latter fact makes the former fact qua proposition true. One might complain that the account I have just given of how constituents are held together in a proposition simply trades in one problem for another. I have answered the question of what holds the constituents of propositions together by specifying the relations that I claim do that job. But, one might complain, this leaves unanswered the general question of what holds together a relation and its relata when they are so related.6 It is true that I haven’t answered this question, and in this sense I have traded in the question of what holds together the constituents of a proposition for the question of what holds together the components of a fact. My excuse is that I think that anyone who believes that things stand in relations and possess properties must face the question, if only to dismiss it, of what holds an object and a property together when the

4 Th e quantifi cation over contexts here is over possible contexts of utterance. See King [2007] pp. 42–45. More on this below. 5 I’ll qualify this slightly below. 6 Jim Higginbotham raised this sort of worry at an Author Meets Critics Session on King [2007] at the Pacifi c Division Meetings of the American Philosophical Association in Vancouver on April 11, 2009.

King020513OUK.indd 50 11/23/2013 12:58:39 PM NATURALIZED PROPOSITIONS 51

object possesses the property or what holds an n-place relation and n objects together when the objects are so related, etc. So I claim to have reduced the mystery of what holds propositions together to a mystery that all of us who think that objects possess properties and stand in relations need to face in any case. Reducing two mysteries to one seems like progress to me. I have claimed that the proposition that Michael swims is the fact described above consisting of Michael standing in the two-place relation mentioned to the property of swimming. We must now face the question of how/why this fact has truth conditions. In general, such facts aren’t the sorts of things with truth conditions. Consider the fact consisting of me standing in the two-place sitting in front of relation to my computer. Th is fact, of course, obtains but it doesn’t have truth conditions. So how is it that the fact that I claim is the proposition that Michael swims does have truth conditions and so is the sort of thing that is true iff Michael possesses the property of swimming? One of the most radical and provocative features of the account of propositions in King [2007, 2009] is the idea that it is something speakers do that endows the fact that is the proposition that Michael swims, and propositions generally, with truth conditions. Th is will explain why this fact has truth conditions, while many other facts do not. Th ough the two-place propositional relation binding together Michael and the property of swimming is highly complex (e.g. it has the sentential relation R of 1 as a component or “part”), let’s suppress that complexity for a moment and simply focus on the idea that on the present view the proposition that Michael swims is a fact consisting of Michael standing in the (complex) two-place propositional relation to the property of swimming. We can represent this fact/proposition thus:

1P. 2007 2007 2007 2007 20072007

Melbourne 2007 Melbourne 20 Melbourne 200

Melb 20 Melbourne Melbourne 2007Melbourne 200

King020513OUK.indd 51 11/23/2013 12:58:39 PM 52 JEFFREY C. KING

(where the picture on the left is Michael; that on the right is the property of swimming and the branching tree structure is the propositional relation). Now one way this fact could have truth conditions is if speakers interpreted the propositional relation here as ascribing the property of swimming at its right terminal node to Michael at its left terminal node. Th en the fact would be true iff Michael possessed the property of swimming. Recall that the sentential relation of the sentence 1 is interpreted by English speakers as ascribing the property that is the semantic value of “swims” to the semantic value of “Michael,” which we expressed by saying that the sentential relation R of 1 encodes ascription in English. What we are now saying is that if the propositional relation of 1P were interpreted as ascribing the property at its right terminal node to the individual at its left terminal node, and so itself encoded ascription, the fact/ proposition would have truth conditions. Encoding ascription understood in this way, note, is a relational property of the propositional relation itself: the property of being interpreted as ascribing what is at its right terminal node to what is at its left terminal node. So henceforth, let’s understand the proposition that Michael swims to be the fact described above, taken together with the propositional relation having the relational property of encoding ascription (this means that the fact that is the proposition that Michael swims is a slightly “larger” fact than we have taken it to be to this point, since it now includes the propositional relation possessing a certain relational property). In so doing, we can explain why the proposition/fact has truth conditions. But the explanation is still preliminary and unsatisfying until we explain what constitutes our interpreting the propositional relation of 1P as ascribing the property of swimming to Michael. What exactly makes it the case that we so interpret the propositional relation? Let me sketch my explanation, which comes in two steps. Call the fact that I claim is the proposition that Michael swims FAST . What we fi rst need to explain is why it is FAST, rather than some other fact, whose propositional relation we interpret as ascribing the property of swimming to Michael so that it is true iff Michael swims. I believe that there are a number of conditions a fact must satisfy in order to be the one whose propositional relation we so interpret, including being a fact consisting of Michael standing in a two-place relation to the property of swimming. 7 But a crucial condition is that we must be able to make sense of the idea that speakers have some sort of cognitive connection to the fact in question. Surely it would be bizarre to hold that speakers are interpreting the propositional relation of a fact in a certain way, where we claim that they have no cognitive connection or access to it. Further, since we want speakers of diff erent languages to in some cases grasp the same proposition, we must be able to make sense of speakers of diff erent languages interpreting the propositional relation of the same proposition/fact. And this requires them to be cognitively connected to the same fact in order that we can make sense of their interpreting its propositional relation. In addition, it seems reasonable to hold that the required cognitive connection to the fact that is the proposition that Michael swims comes about in virtue of speakers

7 See King [2007] pp. 62–64 and King [2009] p. 268 for discussion.

King020513OUK.indd 52 11/23/2013 12:58:44 PM NATURALIZED PROPOSITIONS 53

deploying sentences of their languages. For by the time speakers deploy sentences of their languages, they presumably must have propositional attitudes whose contents are the semantic contents of the sentences they are using. But this means that propositions must exist by that time. Th at in turn means that speakers must be interpreting the propositional relations of the facts that are propositions in certain ways by that time. And in turn, this means that speakers at that time must be cognitively connected to the relevant facts. Th e most straightforward explanation of why speakers have cognitive connections to the facts that are propositions as soon as they deploy sentences of languages is that by deploying sentences of their languages they thereby have cognitive access to the relevant facts. 8 To summarize, then, for a fact to be the proposition that Michael swims, we must be able to make sense of the idea that speakers of diff erent languages all have cognitive access to it and do so in virtue of deploying the relevant sentences of their languages. I’ll now argue that FAST is preeminently a fact of this sort. To see this, note fi rst that sentences (types) themselves are likely facts in my sense. For it seems plausible that word types are properties and hence that sentences are properties standing in sentential relations. Obviously, speakers of e.g. English and German have cognitive access to the facts that are sentences in their languages, like “Michael swims” and “Michael schwimmt.” More importantly, as a result, they also have access to the following “interpreted sentences”: Th ese are just the sentences, together with the semantic relations the lexical items bear to their semantic values (including the semantic values themselves—these relations are represented by vertical lines connecting “Michael” to Michael and “swims”/“schwimmt” to the property of swimming). Hence these interpreted sentences are just “bigger” facts than the sentences themselves in virtue of including the semantic relations between lexical items and their semantic values, as well as the semantic values themselves. We can describe the fact 1IE as follows: there is a context c such that Michael is the semantic value of “Michael” in c, which occurs at the left terminal node of the syntactic relation R that in English encodes ascription and the English word “swims” occurs at the right terminal node of R and has as its semantic value in c the property of swimming.9 It seems to me that by having cognitive access to the sentences “Michael swims.” and “Michael schwimmt.” and being competent with them, English and German speakers thereby have cognitive access to the facts that are the interpreted sentences IE and IG respectively.

8 Of course the explanation cannot be that they have cognitive access to the facts that are propositions because they are expressed by the sentences they are deploying. For we are now trying to explain how certain facts came to be propositions (by having their propositional relations interpreted in certain ways, etc.) and so we cannot appeal to the fact that they already are propositions expressed by sentences of the relevant languages. 9 Th e fi gures IE and IG fail to capture that “Michael swims.”/“Michael schwimmt.” is English/German and that Michael is the semantic value of “Michael” relative to a context of utterance (this qualifi cation is unnecessary here, but would be crucial if we considered the sentence “I swim” in a context with Michael as the speaker).

King020513OUK.indd 53 11/23/2013 12:58:44 PM 54 JEFFREY C. KING

1IE.

Michael swims 2007 2007 2007 2007 20072007

Melbourne 2007 Melbourne 20 Melbourne 200

Melb 20 Melbourne Melbourne 2007Melbourne 200

1IG.

Michael schwimmt 2007 2007 2007 2007 20072007

Melbourne 2007 Melbourne 20 Melbourne 200

Melb 20 Melbourne Melbourne 2007Melbourne 200

Let’s say that the fact of object o possessing property P is a witness for the fact of there being P’s (i.e. the fact of the property P having the property of being instantiated); similarly for the fact of o bearing R to o’ and the fact of there being an x and y such that xRy, and so on. Th e facts IE and IG are both witnesses for the fact that I claim is the proposition that Michael swims, namely, FAST. FAST is the result of “existentially generalizing” on the words in IE/IG and the languages involved.

King020513OUK.indd 54 11/23/2013 12:58:44 PM NATURALIZED PROPOSITIONS 55

Th e crucial point for current concerns is that that having cognitive access to a witness for a fact is a way of having cognitive access to the fact witnessed: having cognitive access to the fact of o possessing P is a way of having cognitive access to the fact of there being P’s. But then having cognitive access to IE or IG above suffi ces for having cognitive access to FAST, the fact I claim is the proposition that Michael swims. Th us we can see how English and German speakers can all have access to FAST in virtue of deploying the relevant sentences of their languages. Hence, we are in a position to make sense of their interpreting its propositional relation as encoding ascription, and so make sense of the claim that it is FAST whose propositional relation we so interpret. But even if we are now convinced that it is FAST’s propositional relation that we interpret as ascribing the property of swimming to Michael, we need to say what constitutes our so interpreting it. Th at is, what is it we do that amounts to our so interpreting it? It is simply that we compose the semantic values at the terminal nodes of the propositional relation in the way we do. In the end, this is just a refl ex of the sentential relation R having the semantic signifi cance it does. When we entertain a proposition, we work our way up the propositional relation, combining semantic values to yield new semantic values for further combining. Obviously we must combine or compose those semantic values in some way. In the case of FAST, were we to do anything other than ascribe the property of swimming to Michael, we would not be combining semantic values in a manner that is consistent with the way we interpret the syntax of the sentence 1. It just isn’t coherent to interpret the sentential relation R as ascribing the semantic value of “swimming” to the semantic value of “Michael,” while composing the semantic values Michael and the property of swimming in some other way as one moves up the propositional relation of FAST. Semantic values only get composed once in understanding the sentence 1, and hence entertaining the proposition FAST. We either do so in the way dictated by the way we interpret the sentential relation R or not. To do so in the way dictated by our interpretation of the sentential relation R just is to interpret the propositional relation as encoding ascription. To summarize, FAST has truth conditions because speakers interpret its propositional relation as ascribing the property of swimming to Michael. Th e account of what constitutes speakers doing this is in two steps. First, reason was given for thinking that it is FAST’s propositional relation that gets interpreted as ascribing the property of swimming to Michael. Second, an account was given of what so interpreting FAST’s propositional relation consists in. Th ere are facts closely related to FAST that probably satisfy these conditions as well, so here we would have to claim that FAST is the most eligible to be the proposition that Michael swims of the facts satisfying all relevant conditions. Having sketched the account of propositions I endorsed in previous work, let me now note a couple ways in which I would alter that account. Th e fact that I claimed is the proposition that Michael swims, FAST, is very roughly the result of composing the syntactic relation R that obtains between “Michael” and “swims” in the sentence “Michael swims” with the semantic relations obtaining between “Michael” and

King020513OUK.indd 55 11/23/2013 12:58:54 PM 56 JEFFREY C. KING

Michael and “swims” and the property of swimming (relative to a context c), while existentially generalizing away the lexical items “Michael” and “swims.” In this way, semantic relations between words and their semantic values play a role in binding together the constituents of a proposition. In order that there be propositions containing constituents that have never actually been referred to (even using demonstratives or etc.), I had to make use of semantic relations between lexical items and their semantic values in possible contexts of utterance. Th at is why propositions are facts of the following sort: Th ere is a (possible) context c and lexical items a,b of some language L such that.. . I now think that I was hasty to take this route. In so far as the relation ___ being the semantic value of ___ relative to context ___ is a perfectly acceptable semantic relation, so too is ___ being the semantic value of___ relative to assignment___ and context ___, where assignments are functions assigning objects to variables. Hence, if the latter is an acceptable semantic relation, it can play a role in binding constituents together in propositions, just as I claimed the semantic relation ___being the semantic value of ___ relative to context ___ did. For a language containing as singular terms names, contextually sensitive singular terms and variables, we can defi ne the relevant relation as follows.

For any context c, assignment g, and singular term e, o is the semantic value of e relative to g and c iff (i) e is a contextually sensitive singular term and o is the referent of e in c; or (ii) e is a name and o is the bearer of e; or (iii) e is a variable and o is g(e). We can now say that the proposition that Michael swims is the following fact: there is a context c, assignment g and language L such that for some lexical items a and b of L, Michael is the semantic value of a relative to g and c and the property of swimming is the semantic value of b relative to g and c and a occurs at the left terminal node of syntactic relation R that in L encodes ascription and b occurs at R’s right terminal node. Several points should be emphasized about this characterization of propositions. First, as before, the existence of the English sentence “Michael swims” suffi ces for the existence of the above fact/proposition. For given the existence of this sentence, that the lexical items “Michael” and “swims” have as their semantic values relative to any c and g Michael and the property of swimming, respectively, and the semantic signifi cance of the syntactic relation R the lexical items stand in in the English sentence “Michael swims,” there is a context c, assignment g (namely, any context and assignment) and language L (English) such that for some lexical items a and b of L (“Michael” and “swims”), Michael is the semantic value of a relative to g and c and the property of swimming is the semantic value of b relative to g and c and a occurs at the left terminal node of the syntactic relation R that in L encodes ascription and b occurs at R’s right terminal node. But of course, this is just to say that the fact that I claim is the proposition that Michael swims exists. Second, it suffi ces for the existence of this fact/proposition that there be a syntactic structure of the following sort in English:

King020513OUK.indd 56 11/23/2013 12:58:54 PM NATURALIZED PROPOSITIONS 57

1a.

x swims

where the left terminal node is occupied by a variable, the right terminal node is occupied by “swims” and the syntactic relation obtaining between them as before has the semantic signifi cance of encoding ascription. For given that this is so, again, there is a context c, assignment g (namely, any context and an assignment that maps “x” to Michael) and language L (English) such that for some lexical items a and b of L (“x” and “swims”), Michael is the semantic value of a relative to g and c and the property of swimming is the semantic value of b relative to g and c and a occurs at the left terminal node of syntactic relation R that in L encodes ascription and b occurs at R’s right terminal node. Th ird, we no longer need to make use of semantic relations between lexical items and entities in merely possible contexts of utterance. Recall that we needed to do that before in order to secure the existence of propositions that contained constituents that had never actually been the semantic value of any expression (relative to a context). But now syntactic structures like 1a insure e.g. that the proposition that o swims exists even where o is an object that has never been the semantic value of any expression relative to any actual context of utterance. 10 For even if that is so, there is a context c, assignment g (namely, any context and an assignment that maps “x” to o) and language L (English) such that for some lexical items a and b of L (“x” and “swims,” respectively), o is the semantic value of a relative to g and c and the property of swimming is the semantic value of b relative to g and c and a occurs at the left terminal node of the syntactic relation R that in L encodes ascription and b occurs at R’s right terminal node. And this is just to say that the proposition that o swims exists. 11 Th is means that our account of propositions only commits us to the existence of actual contexts of utterance and so doesn’t presuppose or require the existence of possible worlds. Th at in turn means that on the new account, we leave open the option of analyzing possible worlds in terms of propositions, (see e.g. Adams, 1974). On the old account, we could not do that since in characterizing propositions we appealed to possible contexts of utterance, which were eff ectively centered possible worlds. Hence, the old account was precluded

10 I don’t mention assignments here precisely because, as mentioned above, on the old view the only semantic relation that played a role in binding constituents together in proposition was ___being the semantic value of ___ relative to context___ . 11 What about propositions with properties as constituents that have never been expressed by any predicate nor been actually designated by contextually sensitive expressions? Well, given sentences like “Several properties are had by John” will have syntactic structures like “x was had by John,” where the “x” takes properties as values. Given what was said in the main text, this is enough for the proposition that P was had by John to exist, where P is a property that no predicate expresses and that has never been actually referred to by contextually sensitive expressions.

King020513OUK.indd 57 11/23/2013 12:58:55 PM 58 JEFFREY C. KING

from analyzing possible worlds in terms of propositions. 12 Th at the new account leaves open this possibility the old account precluded strikes me as a point in its favor. Finally, the explanation for why our facts/propositions have truth conditions goes through as before.13 Let’s now turn to a second emendation of the view defended in King [2007, 2009], henceforth assuming the above emendation is accepted. As was said before, on the current view of propositions, the facts that are propositions very roughly result from composing syntactic relations of sentences with semantic relations between the lexical items in the sentences and their semantic values (relative to a context and assignment), while “existentially generalizing away” the lexical items (context, language and assignment). But the syntactic relation between the lexical items is not existentially generalized away and so remains a component of the fact that is the proposition. Th is has the result that any syntactic diff erence between sentences results in the sentences expressing distinct propositions. Th ough some may think that this results in propositions being individuated too fi nely, elsewhere I have argued that this isn’t so. 14 Nonetheless, there is a way to alter the account with the result that propositions are individuated somewhat less fi nely.15 Th e idea is simply to “existentially generalize away” the syntactic relations like R that to this point have been components of the facts that are propositions. Th us we now claim that the proposition Michael swims is the following fact: there is a context c, language L, syntactic relation R, assignment g and lexical items a and b of L such that the semantic values of a and b relative to c and g are Michael and the property of swimming, respectively, and a and b occur at the terminal nodes of R, where R ascribes the semantic value of b relative to g and c to the semantic value of a relative to g and c in L. Th is account, when extended to other sentences and propositions, has the result that sentences containing lexical items with the same semantic values (relative to parameters) but e.g. whose lexical items occur in diff erent orders, may nonetheless express the same proposition. For example, suppose there were a language, Renglish, in which the following is a sentence that is true iff Michael swims:

1b.

swims Michael

12 Soames [2011] criticizes the old account for ruling out analyzing possible worlds as sets of propositions. 13 Th e explanation needs to be slightly complicated in so far as we have to say something about how cognitive access to a syntactic structure like 1a (which may or may not be sentential) gives us cognitive access to the fact of there being a context c, assignment g and language L such that for some lexical items a and b of L, o is the semantic value of a relative to g and c and the property of swimming is the semantic value of b relative to g and c and a occurs at the left terminal nod of syntactic relation R that in L encodes ascription and b occurs at R’s right terminal node. 14 King [2011] 15 I might owe this idea to Jeff Speaks.

King020513OUK.indd 58 11/23/2013 12:58:55 PM NATURALIZED PROPOSITIONS 59

Th en on the new account, 1b and 1a above express the same proposition, despite the syntactic diff erences between the two sentences. Henceforth, this is the account of propositions we’ll be working with. I’ll now turn to objections to this view of propositions. I called the fact with which I identifi ed the proposition that Michael swims on my old view FAST; let’s call the fact I now identify with the proposition that Michael swims FAST . I’ll begin by considering two closely related objections to the explanation as to how/ why propositions have truth conditions. Th e fi rst objection runs as follows. According to the explanation off ered, e.g. English comes into existence bringing into existence facts like FAST above. But in order for the fact FAST to have truth conditions and so be a proposition, the relation binding together its components—its propositional relation—must be interpreted by speakers of a language, thereby endowing it with semantic signifi cance. Th is, however, will only happen aft er languages exist and speakers are using and understanding sentences. Th us, the account of propositions on off er postu- lates a time when English sentences existed, had truth conditions and were being used and understood by speakers, but at which time no propositions existed and so speakers had no propositional attitudes. Th is means speakers were using and understanding language without having any beliefs or intentions, and indeed without asserting anything (since making assertions and having beliefs and intentions requires bearing relations to propositions). Surely this is implausible.16 Th is objection rests on a misunderstanding of my view. I do not claim that there was a time when language existed and speakers were using and understanding sentences, but propositions did not exist. Quite the contrary. I claim that languages (at least some—see below), propositions and propositional attitudes came into existence at the same time. Once speakers are deploying sentences of their languages, they thereby are cognitively connected to the facts that I claim are propositions and are interpreting their propositional relations in such a way that the structured contents of those sentences have truth conditions. In so doing, language users endow the propositions with truth conditions. It is true that my explanation of why speakers interpret the propositional relation of the proposition that Michael swims in the way they do appeals covertly to the way in which they interpret the sentential relation in the sentence “Michael swims.” Speakers work their way up the propositional relation composing semantic values in the way they do, thus interpreting the propositional relation, because it is the way the semantic signifi cance of the syntax dictates that they compose these semantic values. In this way, the explanation of why speakers interpret the propositional relation the way they do appeals to how they interpret sentential relations. (Recall that the explanation of how and why speakers interpret sentential relations in the way they do appeals to our biologically endowed language faculty.) So in this sense, speakers interpreting syntactic

16 Soames [2011] raises a version of this objection.

King020513OUK.indd 59 11/23/2013 12:58:55 PM 60 JEFFREY C. KING

concatenation in the way they do is explanatorily prior to speakers interpreting propositional relations in the way they do. But nothing in this account requires language to exist temporally prior to propositions. And, contrary to what was stated in the objection, I deny that it does. 17 Th e second objection, which is closely related to the objection just considered, can be stated in the following way. Consider a time before language existed. On the view of propositions under discussion, at that time prelinguistic agents had no propositional attitudes. Aft er all, on the present account propositions didn’t exist then. But then how could these prelinguistic agents bring language into existence? Surely, that would require believing certain things were in one’s environment, desiring to say something about them, intending that a certain sound be understood as a sign for an object, and so on. But having such beliefs, desires and intentions would require the existence of propositions. So the present account of propositions can give no plausible, noncircular account of how languages, and hence propositions, came into existence. I raised this worry in King [2007] and responded by noting that our prelinguistic ancestors could have had “proto intentional” states that weren’t relations to propositions and that these mental states could have been suffi cient for them to bring language, and hence propositions, into existence. It seems to me quite plausible that various creatures have such proto intentional mental states and so the idea that our prelinguistic ancestors did as well is reasonable. So perhaps that story is fi ne so far as it goes, but it overlooks an important point. I believe that many things have content other than sentences of natural languages. Maps, diagrams, perhaps pictures and, most importantly for present purposes, perceptual experiences have contents.18 In the case of each sort of thing that has content, there will be an account of those contents in the spirit of the present account of the contents of natural language sentences. Due to limitations of space, time and knowledge, the details of the theory of the contents of perceptual experiences that is in the spirit of the present account of contents of natural language sentences cannot be sketched here. However, it is plausible to suppose that the contents of perceptual experiences have truth conditions. Finally, it seems reasonable to suppose that the contents of perceptual experiences can be the objects of attitudes like belief, desire, etc. But then our prelinguistic ancestors could have had beliefs and desires whose objects are the contents of perceptual experiences. Th ese attitudes could then fi gure in the account of how language, and the contents of natural language sentences, came into existence.19 So it seems that the present account of the contents of natural language sentences can give a plausible, noncircular account of how language, and the contents of natural language sentences. Let me now turn to a very diff erent sort of objection raised by John Collins [2007]. Collins argues that syntactic structure/sentential relations cannot be used to construct

17 I had already made this explicit in King [2009] pp. 266–268. 18 Greenberg [2011] argues that certain pictures have semantic contents. 19 All this suggests that a better title for King [2007] would have been Th e Nature and Structure of Linguistic Content .

King020513OUK.indd 60 11/23/2013 12:58:55 PM NATURALIZED PROPOSITIONS 61

propositional relations as I claim. More specifi cally he argues that if current syntactical theories are correct, syntax cannot do the work I want it to do. I think Collins’s argument is important and serious, based as it is on details about syntax that I take very seriously. Collins begins by stating some ground rules. He writes:

I take all relevant parties to be minimally naturalistic in the sense of holding to the principle that philosophical elaborations of content impose no a priori constraint on how syntactic theory will or should develop.20 I absolutely agree with this. Second, Collins writes

I shall take the philosophers at their word in their explicit commitments to current or the “best” syntactic theory in the generative camp.21 Again, fair enough. So I agree to Collins’s ground rules. With these ground rules in place, let’s see exactly what Collins is arguing. Looking at a few quotes in which he addresses this is instructive:

In a sense, then, my position is conservative just to the extent that I am arguing for a mismatch between linguistic and propositional structure, much as Frege and Russell argued... In short, as revealed by generative inquiry, it looks as if language narrowly construed is just not in the business of expressing propositions.22 In the fi rst part, I shall consider the claim that syntactic structure is the actual structure of content. In a series of articles and a recent book, King has argued that it is. My riposte will be that there is just far too much structure in the syntax for this to be plausible. 23 My substantive complaint against this proposal [King 2007] is that syntax provides far too much structure, much more than can be accommodated as the values that determine a sentence’s truth conditions, the proposition it expresses. Th is, further, is not a mere quirk of some constructions: the surplus structure is generally exhibited and refl ects the fact that current syntactic theory is not in the business of providing structures that answer to philosophical conceptions of propositions. 24 Th ese quotes suggest that Collins misunderstands my view in a way that matters to the debate. He seems to think that I have some philosophical conception of propositions and their structures, and that I think that somehow syntax just happened to deliver structure exactly answering to this conception. 25 Such a view appears to require a miraculous convergence and so does seem implausible; but it isn’t my view. To anticipate, I think that there are these syntactic structures that developed the way they are

20 p. 806 21 p. 806 22 p. 807–808 23 p. 808 24 p. 810 25 Another quote suggesting the same thing occurs on p. 810: “Th e proposal under discussion presumes that syntactic theory arrives at the same point, from a diff erent direction, as the philosopher is aiming at.”

King020513OUK.indd 61 11/23/2013 12:58:55 PM 62 JEFFREY C. KING

for who knows what reason and we press them into service as providing the structures of propositions. Let’s now consider the details of Collins’s objection. Th e fi rst worry Collins raises for my view concerns movement. 26 Th e problem is that even in quite simple sentences, there is movement; and movement is understood as copying (and deletion at PF but not LF). Collins assumes that movement is driven by the requirement that uninterpretable features be valued (and uninterpretable features must be valued if a syntactic structure is to be interpretable). Collins provides the following illustrative derivation for “Bill sleeps” (where +/- mark interpretable and uninterpretable features, respectively; and underlining marks the valuing of uninterpretable features):27 (3)a. Bill sleeps

b. [VP Bill{+1st, +sing, -Case} sleep] (I think Collins intends +3rd and –Nom here)

c . [ T ’ {+Pres, -3rd Per , -Sing Num, -EPP} [VP Bill{+3rd, +sing, -Nom } sleep]]

d. [TP Bill{+3rd, +sing, -Nom } [T’{+Pres, -3rd Per , -Sing Num , -EPP } [VP sleep]]] In line c, the uninterpretable features of T’ are valued by matching the interpretable features of “Bill.” Th e movement of “Bill” to TP in d is driven by the requirement that the uninterpretable EPP feature of T be valued. Th is all means that for a sentence like “Bill sleeps,” we get the following syntax:

3e.

+Pres, etc. Bill sleeps

Collins writes:

Th ere are two crucial things to note here. Firstly, the derivation is driven to value uninterpretable features, not to provide a propositional structure. Th is motive is semantic in nature, for an uninterpretable feature is precisely a feature that has no semantic signifi cance, but the structure produced departs from what we have been imagining is required for propositional structure. Th is is the second point. Th e copying of Bill higher up the structure creates an item that looks to be surplus relative to property instantiation; that is, the lower copy of Bill exhausts what propositional structure appears to be interested in, namely, Bill is fi xed as the agent of sleep , the instantiater of

26 Actually, prior to this Collins raises another objection. He begins by making the point that syntax has jettisoned the notion of a sentence ( S) in favor of thinking of it as the projection of tense (T ) or infl ection ( I ) (pp. 811–812). Collins seems to think this creates problems for philosophers, apparently including me (on p. 812 he writes “As we’ll see, it is precisely this well motivated move in linguistics that causes trouble for the philosophers.”). But I don’t know why he thinks this. I agree that what we call sentences are TPs or IPs; and I don’t see at all why that is any sort of problem for me. Since this objection seems to me not to have any force, I don’t discuss it in the main text. 27 P. 8 1 4

King020513OUK.indd 62 11/23/2013 12:58:56 PM NATURALIZED PROPOSITIONS 63

the property of sleeping. Further, aft er the introduction of tense, we have a temporal dimension. Still, Bill gets copied for a reason that is not recorded in the proposition, as it were (perhaps Case valuation or EPP elimination). Of course, per the fi rst point, this is no problem or mystery at all from the perspective of the syntax. Th e syntax operates to value uninterpretable features, and copying is part of that mechanism, a displacement of items to meet interface demands, not to create a propositional structure. 28 Again here, Collins suggests that I claim that syntactic derivations are driven to create propositional structure. Again let me say that my view is that we exploit syntactic structure, however derived for whatever reason, and make use of it to produce propositional structure. So I can happily agree with Collins that “the derivation is driven to value uninterpretable features” and that “the syntax operates to value uninterpretable features, and copying is part of that mechanism, a displacement of items to meet interface demands.” Th e crucial question is: can I build the propositional relation of the proposition expressed by “Bill sleeps” out of the “sentential relation” in 3d above? I think it is clear that I can. Here is the proposition: Th ere is a context c, assignment g, lexical items a,b and sentential relation R of language L such that copies of a and b occur at the terminal nodes of R, where the semantic value of a relative to c and g is Bill and the semantic value of b relative to c and g is the property of sleeping and R ascribes the semantic value of b to a. So I think that taking account of what Collins takes to be real syntactic complexity (including copied material) will complicate my account, (indeed, that is why I ignored real syntactic complexity in outlining my view), but I do not see that it refutes or even damages it. 29 Th at said, it should be remembered that syntactic theory is likely far from fi nished. My account requires that propositional relations can be built out of the syntactic structures the true syntactic theory assigns to sentences of natural languages. However, my point is that even if, as unlikely as it may be, syntax as described by Collins is the true syntactic theory, this requirement can be met. Collins considers another example (4a-j p. 815) and writes:

Th e details of these constructions and movement/copying in general are quite complex, but a generalization can be readily extracted: copying occurs aft er everything relevant to a propositional (truth conditional) structure has been determined. But copying creates new structure, and so there is more structure than appears to be demanded by the requirements of the encoding of structure relevant to propositions. Again, the crucial point is that, according to recent theory, copying is motivated by semantic requirements, that is, the valuing of uninterpretable features on functional heads, such as Tense. So, all of the structure is playing a semantic role in one sense, just not the sense apparently required for the encoding of propositions.30

28 P. 816 29 I have ignored tense here, but there is no reason to think this will cause problems (I just don’t know exactly what the tense head looks like nor exactly how the semantics of tense works). 30 Pp. 815–816

King020513OUK.indd 63 11/23/2013 12:58:56 PM 64 JEFFREY C. KING

A bit later, he says:

Th e apparent mismatch of structure, however, poses a serious challenge to philosophers minded to take syntax to encode propositional structure.31 I just don’t see an argument here against my view. Collins seems to take me to claim that syntax must produce only as much structure as is required to get enough and the right sort of propositional structure. But I claim no such thing! Again, you might think this if you thought there was some pretheoretical notion of propositional structure and you thought it was the job of syntax to deliver it. But again that isn’t my view. To repeat, my view is that syntax produces structures, driven by whatever mechanisms, and we exploit these structures by building propositions out of them. Th ere is no claim that we get exactly the degree/amount of structure required for structured propositions. Th e claim is just that syntactic structure gives us enough structure and so can be made to do the job. From my perspective, then, it is as though I had claimed that people press pianos into service as tables and Collins argued that I can’t be right because pianos have more structure than tables require. At one point, Collins does seem to touch on something like my view.32 Th e problem he raises for this view is that “...if propositions are structured by syntax, then propositions appear to contain elements that play no role in the determination of truth conditions. Any bullet can be bitten, but if this one is, I cease to understand the position.” 33 But here I don’t understand what Collins doesn’t understand! If Collins is right about the real complexity of syntax, on my view propositions contain elements/structure that go beyond what would be strictly required to determine their truth conditions. I don’t see anything hard to understand about that position; I don’t see that it involves biting any bullet; and I don’t see that Collins has shown any problem with it. I conclude that Collins’s objection to my account of propositions based on the real complexity of syntax can be successfully countered. 3 4 , 35 Finally, let me turn to an objection to my view raised by Soames [2011]. Because Soames states the objection so clearly and concisely, let me quote it in full:

Since English contains both ⌜ the fact that S⌝ and ⌜ the proposition that S ⌝ —which for King designate diff erent things—his view seems to require “that”-clauses to be ambiguous between the readings they bear in ⌜ Pam regrets (the fact) that S⌝ and ⌜ Pam believes(the proposition) that S⌝ .

31 P. 816 32 First paragraph of section 2.2 p. 818—he calls this “biting the bullet,” but I have no idea why. 33 P. 818 34 On pp. 819–825 Collins considers several responses to his objection. I want no part of his suggested responses, as I think his objection can be countered in the way I have just done. Collins also complains that “biting the bullet” (which seems to be something like my view, though again I can’t at all see what bullet I am biting) has bad empirical consequences. Specifi cally, he argues that my view individuates propositions too fi nely (see p. 820). Since I respond to this objection in King [2011], I won’t discuss it here. 35 Sometimes Collins’s point seems to be that according to current syntactic, the syntactic structures of sentences (TPs, IPs) aren’t appropriate for any sort of truth conditional semantics (see pp. 805–806 and 807). Since this isn’t an objection to my account of propositions specifi cally, I haven’t tried to address it here.

King020513OUK.indd 64 11/23/2013 12:58:56 PM NATURALIZED PROPOSITIONS 65

But he neither gives any linguistic argument that this ambiguity exists, nor rebuts seeming evidence to the contrary—e.g., “Pam regrets that she is pregnant. Although her parents don’t realize it yet, in time they will come to believe it .” Here, the fact regretted is described as something that will eventually be believed—a proposition. How, given the supposed diff erence between the two, can that be?36 Two things need to be said in response to this objection. First, Soames assumes here without argument that expressions like “the fact that Annie is smart” designate the things I call facts. If that were so, then e.g. “the fact that Rebecca swims” and “the proposition that Rebecca swims” would designate diff erent things on my view (the former would designate the fact (in my sense) that makes the latter true) and Soames’s objection would be up and running. However, it is important to see that it is not at all clear that expressions like “the fact that Annie is smart” do designate the things I call facts . I picked out a class of things of the following sort: n objects standing in an n-place relation, n properties standing in an n-place relation, and so on. I then stipulated that I would call these things facts . But then clearly it is a very substantial claim that the ordinary language English locution “the fact that Annie is smart” designates the sort of thing that I stipulatively called a fact.37 Perhaps it is easier to see this if we imagine that I had called the things I call facts states of aff airs instead. Th en clearly the question of whether the expression of ordinary English “the fact that Annie is smart” designates a state of aff airs in my sense is a substantial question. I really don’t know what arguments might be given that such expressions do designate what I call facts . But the important point for our purposes is that Soames’s objection here depends on this claim; but Soames has given no reason to accept it, nor is it clear what reason might be given for accepting it. So Soames is correct that in King [2007] I didn’t off er evidence that “the fact that Rebecca swims” and “the proposition that Rebecca swims” designate diff erent things, nor did I rebut seeming evidence to the contrary. But that is because, as I explicitly said in King [2007], I was not assuming that expressions like “the fact that Rebecca swims” designate the sort of thing that I call a fact.38 As just indicated, that this is so is a substantial claim that I do not know how to argue for (or against). Since I was not assuming that these expressions designate diff erent things, I was not committed to off ering an argument that they do (or to rebutting seeming evidence that they don’t). But let’s waive this worry and simply assume that expressions like “the fact Annie is smart” really do designate the things that I call facts . Th en Soames is correct in claiming that on my view e.g. “the fact that Rebecca swims” and “the proposition that Rebecca swims” would designate diff erent things. Soames then objects that I give no reason for thinking that these expressions do designate diff erent things nor do I rebut what Soames off ers as seeming evidence to the contrary. Unfortunately for Soames’s

36 Soames [2011] p. 9 37 I actually made this point already in King [2007] p. 149 note 30 and p. 150 note 31. 38 S e e n o t e 3 7.

King020513OUK.indd 65 11/23/2013 12:58:56 PM 66 JEFFREY C. KING

objection here, there is lots of evidence that such expressions do designate diff erent things.39 Further, Soames’s seeming evidence to the contrary can be explained on the hypothesis that the expressions in question designate diff erent things. Let’s begin with the evidence that expressions of the form “the fact that...” and “the proposition that...” designate diff erent things; and that “bare” that-clauses can designate either sort of thing. Th e fi rst bit of evidence concerns one of the diagnostics for so-called factive contexts . Generally speaking, factive contexts allow that-clauses that begin “the fact that,” whereas non-factive contexts do not:40 4a. Isabel regretted the fact that she didn’t buy insurance. b . Th e fact that Amy won the race surprised us. c . Th e fact that the moon creates the tides is well known. d. Isabel believed [*the fact] that she didn’t buy insurance. e . [ * Th e fact] that Amy won the race is likely. f . [ * Th e fact] that the moon creates the tides is true. On the view that “the fact that...” and “the proposition that...” designate diff erent sorts of things, this data is easy to explain. Factive contexts predicate properties of the sort of things designated by expressions of the form “the fact that...,” which aren’t propositions but rather are facts. Non-factive contexts predicate properties of propositions. Th e properties appropriately predicated of facts are diff erent from the properties appropriately predicated of propositions. If it isn’t already clear, the same sort of data suggests that it can’t just be that while expressions of the form “the fact that...” designate true propositions, expressions of the form “the proposition that...” designate propositions: 5a. ?Isabel regretted the true proposition that she didn’t buy insurance. b . ? Th e true proposition that Amy won the race surprised us. c . ? Th e true proposition that the moon creates the tides is well known. d. Isabel believed the true proposition that she didn’t buy insurance. e . ? Th e true proposition that Amy won the race is likely. f . Th e true proposition that the moon creates the tides is true. To the extent that 5a-c are coherent, they clearly don’t mean the same thing as 4a-c. But if expressions of the form “the fact that...” designated true propositions, it is utterly unclear why there would be this diff erence between 4a-c and 5a-c. Similarly,

39 Th ere is lots of evidence that the expressions in question designate diff erent things, contrary to what Soames suggests. What there isn’t, to my mind, is evidence that expressions like “the fact that Rebecca swims” designate what I call facts. 40 Th is is discussed in Parsons [1993]. As with many diagnostics, this one is not perfect as Parsons himself notes.

King020513OUK.indd 66 11/23/2013 12:58:56 PM NATURALIZED PROPOSITIONS 67

5d-f are fi ne.41 But if expressions of the form “the fact that.” designate true propositions, then we would expect 4d-f to be fi ne as well. Finally, since sentences like 4a-f are all fi ne with “bare” that clauses (“that...”), in so far as we are convinced that expressions of the form “the fact that...” and “the proposition that...” designate diff erent things and that they must designate facts in sentences like 4a-c in order that they are felicitous whereas they must designate propositions in sentences like 4d-f in order that they are felicitous, we should hold that “bare” that-clauses may designate facts or propositions. A second bit of evidence that expressions of the form “the fact that...” and “the proposition that...” designate diff erent things, presented in Parsons [1993], is that when we attempt to quantify across both a factive and a nonfactive context, the results are oft en quite anomalous: 6a. *Everything John says Glenn discovers. b. *Everything John believes is tragic. c. *Everything John knows is likely. By contrast, as Parsons notes, quantifying over two factive or non-factive contexts is generally fi ne: 7a. Everything John says Glenn believes. b. Everything Sue regrets John discovers. c. Everything John says is likely. d. Everything John discovers is tragic. If the things that factive contexts appropriately predicate properties of, and the things that nonfactive contexts appropriately predicate properties of, are diff erent, the anomolousness of 6a-c would be explained.42 A fi nal bit of evidence that expressions of the form “the fact that...” and “the proposition that...” designate diff erent things is given by sentences such as: 43 8a. Th at the rock hit the window caused the window to break. b . Th e fact that the rock hit the window caused the window to break. c . * Th e true proposition that the rock hit the window caused the window to break. d . * Th e proposition that the rock hit the window caused the window to break.

41 5e’s oddity appears to have a straightforward pragmatic explanation. Use of “Th e true proposition that Amy won the race” commits the speaker to the claim that Amy won the race. To then say that her winning the race is likely is odd. 42 Below we’ll consider the fact that sometimes sentences that quantify into factive and non-factive contexts sound fairly good. Examples include “Th ere is something that Glenn knew and Cris doubted” and “John believes everything he knows.” Presumably many philosophers would even agree with the latter. 43 Such evidence is discussed in Harman [2003] and Asher [2000], both of whom conclude that the things designated by expressions of the form “the fact that...”—facts—are distinct from the things designated by expression of the form “the [true] proposition that...”—[true] propositions.

King020513OUK.indd 67 11/23/2013 12:58:56 PM 68 JEFFREY C. KING

8b suggests that the things designated by expressions of the form “the fact that...” are things that can cause other things. 8c and d suggest that neither propositions nor true propositions can do so. Th is strongly suggests that the things designated by expressions of the form “the fact that...” are distinct from the things designated by expressions of the form “the [true] proposition that....” Finally, the felicity of 8a strongly suggests that the “bare” that-clause in it designates the same thing that “Th e fact that the rock hit the window” designates in 8b. Hence, contrary to what Soames suggests, there is ample evidence that expressions of the form “the fact that...” and expressions of the form “the proposition that...” designate diff erent things; and that that-clauses are capable of designating both kinds of things. However, there are two pieces of counter-evidence to these claims. First, there is Soames’s example:

9. Pam regrets that she is pregnant. Although her parents don’t realize it yet, in time they will come to believe it .

Here the occurrences of “it” appear to have “that she is pregnant” as their antecedent. If the latter designates a fact (“Pam regrets...” is a factive context), then so should the former. But then why is it felicitous to say that this fact is believed in the fi nal clause? Second, there are examples in which quantifying across a factive and a non-factive context seems felicitous:

10. Th ere is something that John knows and Glenn doubts.

But this would seem to require that there be some one thing that both the factive and non-factive contexts predicate a property of, contrary to what has been claimed. Before saying something in response to these considerations, there are a couple points that should be stressed. First, as indicated above, we have seen quite substantial evidence that expressions of the form “the fact that...” and “the [true] proposition that...” designate diff erent things and that “bare” that-clauses are capable of designating either kind of thing. Hence in encountering two bits of counter-evidence to these claims, the weight of the overall evidence is still very much in favor of these claims. Th is strongly suggests that we should explain the apparent counter-evidence while maintaining the claims in question. Second, because I think the considerations here are subtle and complex, I won’t be trying to provide a defi nitive rebuttal to the counter-evidence. Rather, I’ll indicate some general strategies of rebuttal. Th e fi rst strategy, due to Parsons [1993], takes as its starting point the idea that e.g. the fact that gasoline is lighter than water and the true proposition that gasoline is lighter than water are intimately related. Following Parsons, let’s say that the relevant fact in this case corresponds to the true proposition that gasoline is lighter than water.44 N o w

44 Here we come up against one of the subtleties alluded to above. For any that-clause designating a true proposition, it seems we can form a phrase of the form “the fact that...,” which apparently designates the corresponding fact. Does this means that for every true proposition, there is a corresponding fact? Th ose skeptical about the claim that every true proposition is made true by a corresponding fact (e.g. because they don’t think true negative existential propositions are made true by corresponding facts; or they don’t think that true propositions about long past things (e.g. that Socrates was a philosopher) are made true by corresponding facts) presumably would also be skeptical about answering this question affi rmatively.

King020513OUK.indd 68 11/23/2013 12:58:56 PM NATURALIZED PROPOSITIONS 69

let’s suppose that we have an object language in which sentences refer to truth-values.45 We then add the expression “^” that fronts sentences (and perhaps other expressions too) with the stipulation that “^[S]” refers to the proposition that S. 46 Finally, letting f be a function that maps true propositions to their corresponding facts, we let “c” name f. Th en “c^[S]” refers to the fact that S. Th e idea is that Soames’s example can be represented as follows: 9. Pam regrets c^ [she is pregnant]. Although her parents don’t realize it yet, in time they will come to believe it . where the fi nal occurrence of “it” is anaphoric on a term of the from “c^[S].” 47 N o w instead of having the entire term as its antecedent, and hence referring to a fact, the pronoun could have as its antecedent the embedded term “^[S]” and so refer to the proposition this expression refers to. Th is would explain the felicity of the example.48 A second strategy for explaining Soames’s example while maintaining that expressions of the form “the fact that...” and “the proposition that...” designate diff erent things (and that bare that-clauses can designate either) is to appeal to other cases in which something similar seems to happen. Again, we begin by assuming that e.g. the fact that snow is white and the proposition that snow is white are intimately related. Now consider the following example, inspired by examples discussed by Chomsky [2000]: 11. Th e book I just stole from the library is on my desk. It was written in 1801 and has been translated into many languages. It is hard to resist the idea that the occurrence of “Th e book I just stole from the library” in the fi rst sentence designates the particular book (token) on my desk. It, aft er all, is the thing I stole from the library and that is on my desk. However, equally, on refl ection it is hard to resist the idea that the second sentence predicates a property of a quite diff erent thing: a thing that was written in 1801 and has been translated into many languages. Surely, this thing isn’t the particular book on my desk that I just stole. And this is so despite the fact that the pronoun in the second sentence is anaphoric on “Th e book I just stole from the library” in the fi rst sentence. In such cases, it looks like the antecedent of a pronoun designates one thing and the pronoun anaphoric on it designates something else, albeit something intimately connected to the former. But if this can happen in a mundane case of this sort, there is no reason to think it can’t happen in Soames’s example 9. Concerning our example 10

45 Parsons [1993] also supposes that names refer to individuals and predicates refer to functions from n-tuples of objects to truth values. 46 So that-clauses referring to propositions are terms of the form “^[S] .” 47 Here we suppose that bare that-clauses that refer to facts are terms of the form “c^[S] .” 48 On such a view, both expressions of the form “the fact that...” and “bare” that-clause that designate facts would be thought of as having the form “c^[S] .” See previous note.

King020513OUK.indd 69 11/23/2013 12:58:56 PM 70 JEFFREY C. KING

10. Th ere is something that John knows and Glenn doubts. that might be thought to show that there is some one thing that both the factive and non-factive contexts predicate a property of; though I won’t go into great detail, essentially the two strategies just discussed can be applied here too. On the one hand, we could think of 10 as having the following structure (using “c” as above): 1 0 R . ( ූp)(John knows c(p) and Glenn doubts p) where our quantifi er ranges over propositions. Here the object of John’s knowledge is a fact and that of Glenn’s doubt is a proposition. Of course some explanation would have to be given of why examples quantifying across factive and non-factive contexts are sometimes so bad. On the other hand, we do say things such as the following and take them to be true in the appropriate circumstances 1 1 Q . Th ere is a book that is on my desk and was written in 1801 on parchment paper. Th is is so even though there appears to be no one thing that both is on my desk and was written in 1801 on parchment paper. Hence, that we take 10 to be true, in the same way and for more or less the same reasons, need not require there to be some one thing that is both known by John and doubted by Glenn.

King020513OUK.indd 70 11/23/2013 12:58:56 PM 5 Propositions are Properties of

Everything or Nothing

J e ff S p e a k s

Two kinds of theories of propositions One way of dividing the space of possible views of propositions is between those who take propositions to be a sui generis category of entity, and those who take propositions to be members of another ontological category. Let’s call these, respectively, non - reductive and reductive views of propositions. Well-known reductive views include the view of propositions as functions from worlds, or indices of some other sort, to truth-values. Th ere are general reasons for preferring reductive views. Th ey have a built-in advantage of ontological parsimony, since they countenance one less category of entity than ontologies which contain sui generis propositions. 1 Th ey also promise greater explanatory power—if we can explain otherwise puzzling features of propositions in terms of features of the ontological category to which propositions are assimilated. Th ere are also reasons to be doubtful of the existence of the sui generis propositions posited by non-reductive theories. One such challenge was raised by Russell. In On the Nature of Truth and Falsehood , Russell asked us to consider some false sentence, like “Gore won in 2000” and asked how, given that there is no such thing as Gore’s having won in 2000, there could be such a thing as that Gore won in 2000 . 2 Russell’s idea seems to have been that when we ascribe a property to an object, there is the object, the property ascribed, the act of property ascription, and, if the object instantiates the property,

1 Th ough this is complicated a bit by the fact that proponents of the non-reductive views might analyze entities of some other type as a kind of proposition, as with views which take facts to be true propositions, and further complicated by the fact that proponents of reductive views might as well—so long, of course, as the category to be reduced to propositions isn’t the same as the category to which the reductive theorist aims to reduce propositions. 2 See Russell’s discussion of “that Charles I died in his bed,” in Russell (1910), p. 151.

King020513OUK.indd 71 11/23/2013 12:58:57 PM 72 JEFF SPEAKS

the fact of the thing’s having the property; but there is no room for some other thing, the proposition that the object has that property. To be sure, Russell’s worry here rests on a bare metaphysical intuition: that once we cross the relevant objects, properties, and facts off the list, there’s no room left for entities of some other kind, which are “about” the relevant situation, to squeeze their way in. But, unargued as it is, the intuition seems to me to be a powerful one—and the best way to answer it seems to be to identify that Gore won in 2000 with some object, property, fact, or event in which we have independent reason to believe. A second reason to be skeptical about sui generis propositions, separate from but related to Russell’s, comes from the oddness of supplementing an ontology of objects, properties, facts, events, etc. with another category of entity whose sole roles are to be the contents of mental states and sentences, and the bearers of truth and falsity. It seems oddly anthropocentric to add a category to our ontology just to play these roles; surely it would be better, all things being equal, to fi nd a class of entity to play these roles among the types in which we already have independent reason to believe. 3 My aim will be to advance a reductive view of propositions. In my view, propositions are a kind of property. As is well known, defenders of (close relatives of) this view have included Roderick Chisholm and David Lewis—and, more recently, Peter van Inwagen has defended the view that propositions are 0-place properties.4 My view is in many ways similar to these. But as will become clear, my view also borrows a lot from an apparently quite diff erent view of propositions—Jeff King’s view that propositions are a kind of fact. 5

Propositions, their constituents, and syntax Since propositions are, among other things, what are expressed by sentences, one way into the question of what propositions are is via the question: what sort of thing does a sentence express? What a sentence expresses—its semantic content, relative to the relevant context—is closely related to the contents of the expressions which compose the sentence. Let’s consider for illustrative purposes a simple monadic predication, “Amelia talks.” Let’s call the proposition expressed by this sentence PROP, and let’s suppose that the content of “Amelia” is a certain person, Amelia, and the content of “talks” is the monadic

3 Th ough here I am setting aside issues about the nature of the constituents of propositions, this is one place in which these issues come to the fore. Th is is because a defender of a Fregean view of content—on which the constituents of propositions are modes of presentation of objects and properties rather than those objects and properties themselves—might reasonably claim to have an answer to Russell’s skeptical argument; aft er all, she might say that true propositions are, in eff ect, modes of presentations of facts, whereas false propositions are modes of presentation of nothing—but which could be modes of presentations of facts, were the facts diff erent. But this won’t help with the second worry, which focuses on the oddness of postulating a class of entities to serve as the contents of mental states and sentences. 4 Lewis (1979), Chisholm (1981), van Inwagen (2004). 5 See King (2007) and Chapter 4 above.

King020513OUK.indd 72 11/23/2013 12:58:57 PM PROPERTIES OF EVERYTHING OR NOTHING 73

property of talking.6 In this sort of case, let’s call Amelia and the property of talking the constituents of PROP; and, in general, we can say that if e is a semantically simple expression which is a part of some sentence S , then, in our sense, the content of e will be a constituent of the proposition expressed by S . Th is gloss on “constituent” is intended to make the claim that propositions have constituents pretty metaphysically unexciting, and therefore common ground between various views of propositions; it does not, for example, build in any assumptions to the eff ect that the relationship between propositions and their constituents is analogous to the relationship between composite material objects and their parts.7 Whatever view we take of the nature of PROP , it seems that it should bear some close relation to its constituents, namely Amelia and the property of talking. Aft er all, there are some interesting necessary connections between the two, like the following:

Necessarily, anyone who entertains a thought with content PROP entertains a thought about Amelia. Necessarily, anyone who entertains a thought with content PROP entertains the thought of an object that has the property of talking. Th e existence of necessary connections of this sort indicates that, in some sense or other, the constituents of a proposition are part of its identity. But, on the other hand, it would be a mistake to respond to the necessary connections between propositions and their constituents by simply identifying propositions with the collection of their constituents, or saying that propositions are “nothing over and above” their constituents. Th at this would be a mistake is shown by the fact that there are distinct propositions with the same constituents. Th e propositions expressed by:

Jane loves John. and

John loves Jane. each have John, Jane, and loving as their constituents, but are plainly not identical. Th e same lesson can be drawn from Russell’s observation in the Principles of Mathematics (§52) that substitution of one expression for another with the same content can transform a sentence—which expresses a proposition relative to a

6 Here, for simplicity, I am setting aside some qualifi cations—fi rst, the qualifi cation that sentences typically express propositions only relative to contexts and, second, that perhaps sometimes sentences express only fragments of propositions (even relative to contexts). Th is discussion also obviously presupposes a particular (Russellian) view about what sorts of thing the constituents of propositions are. But the foregoing could be restated to accommodate other views; for example, a Fregean will want to replace talk about objects as constituents of propositions with talk about individual concepts, and will want to replace talk about properties and relations with talk about modes of presentation of those properties and relations. Nothing in what follows will depend upon one choice or another here, though I will stick with the Russellian view for simplicity. (And because I think that it is true.) 7 For some discussion of whether we should take “constituents” talk more seriously than this, see Ch. 11 below.

King020513OUK.indd 73 11/23/2013 12:58:57 PM 74 JEFF SPEAKS

context—into a string of words which does not express a proposition. To adapt his point to the present example, “loves” and “loving” are apparently both terms for the relation of loving; but, while

John loves Jane expresses a proposition,

John loving Jane expresses nothing. If propositions really were nothing over and above their constituents, it is puzzling why this should be so. Why should switching out one term for another with the same content change a string of words from one which is proposition-expressing to one which is not? Th is all seems to indicate that propositions are—speaking somewhat metaphori- cally—their constituents plus some extra ingredient. Th e problem of giving a theory of the nature of propositions then becomes the problem of saying what this extra ingredient is. Jeff King’s theory of propositions is based in part on the idea that this extra ingredient has something to do with the syntax of proposition-expressing sentences. Th is seems plausible especially when we think about examples like Russell’s; it seems quite plausible that the diff erence between “John loves Jane” and “John loving Jane” is to be explained in terms of the fact that strings with the syntactic form of the former—those consisting of a name, concatenated with a two-place predicate and another name— express propositions in English whereas strings with the syntactic form of the latter— those consisting of a name, followed by an abstract singular term, followed by another name—do not. But the “extra ingredient” we’re looking for can’t just be a syntactic relation—a point which King brings out nicely via the example of a possible language, Nenglish, which is like English but for the fact that concatenation of a name and a predicate expresses a proposition which is true iff the referent of the name does not instantiate the property expressed by the predicate. Hence “Amelia talks” would express a diff erent proposition in Nenglish than it does in English—despite the fact that in both languages Amelia and the property of talking are the constituents of the proposition expressed by the sentence, and that in both languages the syntactic form of the sentence is that it is a name concatenated with a monadic predicate. What this example brings out is that syntactic relations, just like linguistic expressions, can make diff erent semantic contributions in diff erent languages. Perhaps, then, our extra ingredient has something to do with the semantic contribution of the syntactic form of the sentence which expresses the relevant proposition. So far, so good. But this view raises some further questions. Exactly how should we think about the semantic contribution of syntax? And what does this tell us about what propositions are?

King020513OUK.indd 74 11/23/2013 12:58:57 PM PROPERTIES OF EVERYTHING OR NOTHING 75

Simplifying a bit, King answers these questions as follows. He suggests that we think of the semantic signifi cance (in English) of the relation between “Amelia” and “talks” in “Amelia talks” as the following instantiation function from objects, properties, and worlds to truth values: the function which, given as argument an object o and property F, determines the truth value true at w iff o instantiates F at w. He then embeds this view of the semantic signifi cance of syntax in a view that propositions are a certain kind of fact. We can, in King’s view, describe the proposition expressed by “Amelia talks” as follows: it is the fact of there being words x and y of some language such that x has Amelia as its content, y has the property of talking as its content, R(x,y ), and R encodes the instantiation function. 8 Th ough I think that this view has many attractive features, it’s here that King and I part company. So long as we stick with the view that propositions are a kind of fact, then it is hard to avoid the view that the relevant facts are in part about the linguistic items that express propositions. But I think that a simpler view, which avoids the detour through existential quantifi cation over linguistic expressions, is available if we think of propositions as properties, rather than as facts. Consider again the example of “Amelia talks.” If we think of the semantic content of this sentence as a property, one natural view is that the property is the property of being such that Amelia talks. On this kind of view, what is contributed by the syntax of a simple predication—the semantic signifi cance (in English) of this bit of syntax, in King’s terms—is something like the three-place relation corresponding to the open sentence “__ is such that __ instantiates __.” In the case of the sentence “Amelia talks,” the contents of the name and predicate fi ll in the second two slots to deliver the monadic property expressed by “__ is such that Amelia instantiates the property of talking.” Th is view accommodates the two pieces of data about the relationship between propositions and their constituents mentioned above. As on King’s view, the diff erence between “John loves Jane” and “John loving Jane” is explained by the fact that the syntactic form of the former has, in English, a semantic content, whereas the syntactic form of the latter does not. And we get an explanation of the distinctness of the propositions that John loves Jane and that Jane loves John in terms of the distinctness of the properties of being such that John loves Jane and being such that Jane loves John. 9 Th ough the view that propositions are a sort of property may sound odd at fi rst, it actually fi ts rather naturally much of our talk about propositions. We might say that believing a proposition, for example, is taking the world to be a certain way. But if, as

8 See, e.g., King (2007), 37. 9 One might object that this is not much of an explanation; isn’t this just a relabeling of the fact which was supposed to need explanation? I don’t think so. Anyone who believes in the existence of complex properties, like the property of being such that John loves Jane, must already accept the fact that the property of being such that John loves Jane is distinct from the property of being such that Jane loves John. It’s an explanatory gain if we can show that the distinctness of the propositions that John loves Jane and that Jane loves John is nothing over and above this fact about properties. Maybe this fact about the identity conditions of properties is itself deeply mysterious, and in need of explanation; but even if this is so, it is so whether or not we identify propositions with properties—and it is better to have one mystery than two.

King020513OUK.indd 75 11/23/2013 12:58:57 PM 76 JEFF SPEAKS

it seems, “ways things are” are properties, this indicates that having a belief is taking a certain attitude toward a property. Parallel points might be made about mental states other than belief, and speech acts like assertion; we can hope, intend, and desire that the world be a certain way; and we can say, assert, and suggest that the world is a certain way. Again, if ways things can be are properties, this suggests that the objects of mental states and speech acts—i.e., propositions—are properties.

Properties, truth, and representation Supposing that propositions are monadic properties of this sort, what are these monadic properties properties of ? Th ey are properties of, if anything, everything. Aft er all, given that Amelia does instantiate the property of talking, everything instantiates the property of being such that Amelia talks. Hence the proposition that Amelia talks—i.e., the property of being such that Amelia talks—is true iff it is instantiated. So, for example, it is suffi cient for this proposition to be true that I, or my house, or a rock, instantiate the property of being such that Amelia talks. At this point, I think that the following sort of objection might occur to many:

“What could it be for me to instantiate the property of being such that Amelia talks? Th is property has nothing to do with how I am. Surely if I do instantiate this property, this is just for me to exist while the proposition that Amelia talks is true; which means that any attempt to analyze the truth of this proposition in terms of the instantiation of this property gets the order of explanation backwards.” But why hold this view about the correct order of explanation? We can all agree— mod- uloworries about the existence of the property of being such that Amelia talks—that I instantiate this property iff I exist and the proposition that Amelia talks is true. Taking the left hand side of this biconditional as basic has the advantage that we need postu- late no new category of entities, in addition to properties. And it is hard to see what advantage taking the right hand side as basic could have; aft er all, any analysis of what it is for the proposition that Amelia talks to be true will equally serve as an analysis of what it is for any existent to instantiate the property of being such that Amelia talks. Th ere are several diff erent ways in which this view of propositional truth can be generalized to an account of truth with respect to a world (or arbitrary circumstance of evaluation), but the simplest is as follows. Propositions are properties which are true iff they are instantiated. Propositions are true with respect to a world w iff , were w actual, that property would be instantiated—or, equivalently, iff , were w actual, the proposition would be true. 10 Given this view of truth at a world, thinking of propositions as

10 Th ere are, however, reasons not to go for this simple generalization, especially if one thinks that (a) propositions can’t be true without existing, (b) singular propositions can’t exist unless their constituents do, and (c) some singular propositions have contingently existing objects as constituents. I discuss these issues in Speaks (2012).

King020513OUK.indd 76 11/23/2013 12:58:57 PM PROPERTIES OF EVERYTHING OR NOTHING 77

properties does not seem to require any serious revision in the way that we think about entailment relations between propositions, or semantics more generally. Propositions are necessary iff they are true with respect to every possible world; just so, on the present account, the propositions are necessary iff the properties which they are, are instantiated in every possible world. One proposition F would entail another proposition G iff any world in which F is instantiated is also a world in which G is instantiated.11 I mentioned at the outset that views of propositions which assimilate propositions to members of an ontological category—in this case, properties—in which we have independent reason to believe, enjoy the advantage of (relative) ontological parsimony. While this is, I think, a genuine advantage, it is worth noting that—given the account of truth, and truth at a world just sketched—it comes with a signifi cant string attached. If propositions are properties which are true iff they are instantiated, then it seems clear that we must think that there are uninstantiated properties, and indeed properties which could not be instantiated. Otherwise, there would be no account of the propositions expressed by false, and necessarily false, sentences. 12 Even if this view of propositions does make available clear explanations of truth and truth at a world, there is one aspect of the traditional theory of propositions which it does not capture. Th is is the view that, as Scott Soames puts it, propositional attitudes are representational “because of their relations to inherently representational propositions.” 13 Properties like the property of being such that Amelia talks are not inherently

11 An anonymous reader suggested the following alternative view: A proposition is necessary iff every actually existing thing necessarily instantiates it. Th e idea is that this better captures the intuitive connection between truth and necessary truth. I doubt that any pre-theoretic intuition favors this view and, in any case, it (plus the assumption that some actually existing things could have failed to exist) entails the falsity of the plausible principle (oft en called “Serious Actualism”) that, necessarily, if an object instantiates a property, that object exists. One might try instead A proposition is necessary iff every actually existing thing instantiates it in every world where it exists. but this is equivalent to the view defended above, given the premise that some actually existing things exist necessarily. A diff erent view in the neighborhood of the view that I defend (and which I used to hold) is the view that properties are properties of worlds, like the property corresponding to the open sentence Were w actual, it would be the case that Amelia talks. On this sort of view, truth of course can’t be identifi ed with instantiation, on pain of making possible truth entail truth. On this sort of view, a proposition is necessarily true iff it is instantiated by every world. I don’t have the space here to discuss my reasons for favoring the present version of the property view over the “properties of worlds” version. 12 Th ough it is worth noting that the view is not committed to the existence of simple necessarily uninstantiated properties, which may well seem more troubling than complex necessarily uninstantiated properties. A view of properties which I think would suit my purposes is outlined in van Inwagen (2004). But I think that the view of propositions I am developing would be consistent with various views of what properties are, so long as those views are not committed to any sort of principle of instantiation (or possible instantiation). 13 Soames (forthcoming).

King020513OUK.indd 77 11/23/2013 12:58:57 PM 78 JEFF SPEAKS

representational things; hence if propositions are properties of this sort, this aspect of the view of propositions common to Frege and (the early) Russell must be rejected. Th is may seem like a cost; but there is also a benefi t here. Th e idea that an entity can be intrinsically representational has seemed to be a puzzling one to many. If we can give an account of truth and propositional attitudes (about which more below) without making use of entities of this sort, this is a good thing.

Propositional attitudes My sketch of the view that propositions are properties has been, in one respect, quite diff erent than the way this topic is discussed in, for example, Chisholm and Lewis. Both Chisholm and Lewis approach the topic not by asking what sorts of things are expressed by sentences, but rather by asking what sorts of things are the objects of propositional attitudes. Th is is more than a diff erence in order of exposition. Both Chisholm and Lewis— as well as contemporary defenders of their view 14—advance the view that the objects of the attitudes are properties without identifying propositions with properties. Th is view can sound paradoxical, since it sounds like the denial that propositional attitudes are relations to propositions. But this is a superfi cial objection; there’s nothing incoherent in the idea that one sort of a thing, a proposition, is expressed by sentences, while another sort of thing, a property, is the content of mental states. In this respect, the Chisholm/Lewis view about properties is less ambitious than the view that I’m defending; they think that properties can play some of the roles standardly assigned to propositions, whereas I think that properties can play all of the roles—including, crucially, the role of being the things expressed by sentences. Th ere are some general reasons to prefer the “pure” view that properties can play all of the roles assigned to propositions over a “mixed” view of the sort defended by Lewis and Chisholm, according to which properties are the objects of the attitudes, and some other sort of thing is what is expressed by sentences.15 One sort of reason for preferring the pure property theory over a mixed theory is based on the sort of ordinary language considerations oft en used to introduce talk about propositions in the fi rst place. Th e mixed view is forced, for example, to deny face value readings of apparently true bits of ordinary language discourse like:

Th at sentence expresses my belief perfectly. Th is sentence certainly seems to entail

What that sentence expresses is something I believe.

14 See especially Feit (2008, 2010) and Turner (2010). 15 Of course, Chisholm and Lewis had diff erent views about the nature of the propositions expressed by sentences—it’s just that they agreed that these entities were distinct from the properties that were the objects of the attitudes.

King020513OUK.indd 78 11/23/2013 12:58:57 PM PROPERTIES OF EVERYTHING OR NOTHING 79

which in turn seems to entail:

ූ x (that sentence expresses x & I believe x ) which seems to be inconsistent with the mixed view that propositions are expressed by sentences, properties are the objects of belief, and propositions ≠ properties.16 A second reason for preferring the pure property theory is simply that the mixed theory leaves us without a theory of propositions. Th is is, obviously, not a pressing objection if one feels able to give an independent account of propositions, or if one is happy to take them as a primitive category in one’s ontology. But if one is (perhaps for the reasons sketched at the outset) unhappy with non-reductive views, and (perhaps for reasons like those given in chapter 3 above) dissatisfi ed with the reductive alternatives to the view that propositions are properties, these considerations will push one towards a pure version of the property theory of propositions. Even if a pure theory is to be preferred, and even if Lewis and Chisholm each went for a mixed theory, that doesn’t show that we can’t learn from their treatment of the attitudes, since there is no reason (for all we’ve said) why one couldn’t integrate their theory of the attitudes into a pure theory. So let’s turn to the Lewis/Chisholm theory of propositional attitudes. If the objects of the attitudes are properties, then it is natural to think that belief must be believing of something that it is a certain way; supposing must be supposing that something is a certain way; guessing must be guessing that something is a certain way; and so on. A basic question for defenders of this view of the attitudes is: what is this “something”? Lewis and Chisholm gave basically the same answer to this question: in their view, propositional attitudes are ascriptions of properties to oneself. Chisholm expressed the theory like this:

“Believing must be construed as a relation between a believer and some other thing . . . What kind of thing, then? . . . Th e simplest conception, I suggest, is one which construes believing as a relation between a believer and a property—a property which he may be said to attribute to himself.”17 It is understandable why the view was introduced in this way, since its principal initial motivation was the explanation of the distinction between fi rst-personal beliefs and third-personal beliefs about oneself—or, as Chisholm put it, between the emphatic and non-emphatic refl exive. Th ere seems to be a distinction, to use one of many standard examples, between my believing of myself that I am on fi re, and my believing that Jeff Speaks is on fi re. It seems that I might have either belief without the other, if I am suffi ciently confused about my identity. And this point is not limited to belief; parallel remarks could be made

16 See also the discussion of the “proposition sentences” in Ch. 2 above. 17 Chisholm (1981), 27. For a similar statement of the view, see Lewis (1979).

King020513OUK.indd 79 11/23/2013 12:58:57 PM 80 JEFF SPEAKS

about intuitively “fi rst-personal” vs. “third-personal” desires, intentions, etc. But, as is well-known, it is notoriously diffi cult for the proponent of the view that beliefs are relations to propositions to fi nd a distinction between propositions to correspond to this intuitive diff erence—at least as long as we aren’t willing to let our propositions include exotica like rigidly designating essentially fi rst-personal modes of presentation. Th e proponent of the view that beliefs are self-ascriptions of properties, by contrast, can readily capture this sort of distinction between mental states by saying that fi rst-personal belief that I am on fi re is the self-ascription of the property of being on fi re, whereas my third-personal belief that Jeff Speaks is on fi re is the self-ascription of the property of being such that Jeff Speaks is on fi re. Th ese are, plainly, self-ascriptions of distinct properties—which is just what we should want. And once we make this move for beliefs about oneself, it seems that the apparatus can be extended to account for beliefs about what is happening here , or now , and other sorts of indexical belief.18 Th e view that attitudes are self-ascriptions of properties can thus be given a powerful motivation. But it also faces some serious problems. For one, it seems to violate seeming platitudes about the truth conditions of beliefs, like:

A’ s belief is true iff what A believes is true On the property theory, what A believes—the content of A ’s belief—is a property. Th is in itself need not be problematic if we adopt the account of what it is for a property to be true sketched in the preceding section. But consider what sort of properties are assigned as the contents of beliefs in the key case of fi rst personal belief. In the case of the fi rst-personal belief that one is on fi re, the Lewis/Chisholm theory assigns as the content of the belief the property of being on fi re. What would it take for this property to be true? We can’t say that this property is true iff it is instantiated, since (given the above platitude) this would entail that my belief that I am on fi re can be made true by someone other than me being on fi re. And we can’t require for its truth that everyone instantiate the property, since the truth of my belief doesn’t require that everyone be in fl ames. And we can’t require that I instantiate the property, since that would make the truth of others’ fi rst-personal beliefs that they are on fi re hostage to my temperature rather than their own. We could, of course, get round this sort of problem by letting the property self-ascribed in the case of my fi rst-personal belief that I am on fi re be the property of being such that Jeff Speaks is on fi re, since then we could say that what the subject believes is true iff the property is instantiated—but this would remove the wanted contrast between fi rst- and third-personal beliefs, and hence sacrifi ce the motivation for

18 Th e idea is that we can analyze beliefs about what is happening here as beliefs about what is in the same place as me, and beliefs about what is happening now as beliefs about what is happening at the same time as the time at which I exist—though the latter analysis requires us to think of the subjects of the attitudes as time-slices of persons. See the discussion of the example of the insomniac in Lewis (1979), §VII.

King020513OUK.indd 80 11/23/2013 12:58:57 PM PROPERTIES OF EVERYTHING OR NOTHING 81

the view that propositional attitudes are self -ascriptions of properties.19 Th e only plausible move here seems to be to abandon the platitude, and say that talk about the truth of a subject’s belief comes apart from talk about the truth of what the subject believes; but this, I think, is hard to accept. A second sort of problem for the Lewis/Chisholm theory—which has been developed by Markie (1988) and Nolan (2006)—stems from the observation that, while in many cases nothing is lost by thinking of a propositional attitude with the content p as a belief that I am such that p is the case, in other cases, understanding the content of the belief requires that we consider worlds where p is the case, but in which I am not such that p is the case, because I do not exist. Th e most striking case is perhaps the example of the desire that I not exist. On the Chisholm/Lewis theory, this is interpreted as the desire that I instantiate the property of nonexistence. But this seems wrong, since it is impossible that I have the property of nonexistence, and the desire that I not exist is not a desire for something impossible. 20 How might the defender of the Chisholm/Lewis theory of the attitudes reply to this criticism? One interesting suggestion, defended in diff erent ways in Feit (2010) and Turner (2010), is that we can defuse this objection by appealing to something like the distinction, employed in a diff erent context by Kit Fine, between truth in a world and truth at a world—where the latter does not require existence at that world. 21 If we accept this distinction, then we might say that my desire that I not exist is true at w iff I can instantiate the property of not existing at w iff the proposition that I don’t exist is true at (not in) w . I’m sympathetic to Fine’s distinction and to the idea that a defender of the idea that properties are the objects of the attitudes might use this distinction to avoid the result that the desire that I not exist be a desire for something impossible. But this move comes at a cost. Instantiating a property at a world is defi ned in terms of the truth of propositions at a world; hence it seems that anyone who appeals to this notion of having a property at a world must adopt what above I called a “mixed” theory, according to which, while properties are the objects of the attitudes, propositions still exist as members of some other ontological category. For the reasons given above, I think that it would be better to avoid this result.22

19 Of course, the proponent of this view might try to distinguish the properties of being such that Jeff Speaks is on fi re and the property of being such that I am on fi re. But then all the work is being done by this distinction between fi rst-personal and third-personal properties, and none by the idea that the attitudes are self-ascriptions of properties. 20 Here I’m setting aside the view of Williamson (2001) that everything which exists does so necessarily. 21 See, among other places, Fine (1985). 22 Th is needn’t be the end of the story. One might adapt the essentials of Turner’s response to Nolan and then try to give a theory of what it is for an object to have a property at a world which did not go via some proposition’s being true at that world. I think that it is plausible that this could be done for the case of the desire that I not exist. But, as Turner recognizes, a full solution to Nolan-type problems requires an account not just of overtly fi rst personal mental states, but also for desires like the desire that there be world peace— since we don’t want this to entail a preference, ceteris paribus, for worlds in which I exist and there is world peace over worlds where there is world peace without me. And though it is somewhat plausible that I can have the property of nonexistence at a world without existing in that world, it seems less plausible that this is so for every property, and, in particular, it seems implausible that I can instantiate the property of being such that there is world peace at a world without existing in that world.

King020513OUK.indd 81 11/23/2013 12:58:57 PM 82 JEFF SPEAKS

Finally, the foregoing objections aside, I think that the idea that all of my mental states involve attributions of properties to myself should strike us as odd. Imagine someone engaged in purely abstract reasoning—say, a mathematician. It’s very counterintuitive to think of the mathematician as trying to fi gure out what she is like—it seems like what she cares about is not which properties she has, but what the world is like, or what numbers are like. It should be possible for thinking to be less self-involving than the Lewis/Chisholm view makes it out to be. 23 Fortunately, there is no reason why the view that propositions are properties must be tied to the view that all thought is self-ascription. But if we reject this view, this leaves us with the question raised earlier this section: given that belief is believing something to be a certain way, if this “something” is not oneself, then what is it? Th ere is a sense, I think, in which this question has no answer. To believe that Amelia talks, there is no special object which one must believe to be such that Amelia talks; it is enough if one takes there to be something which is such that Amelia talks. Of course, oft en one takes a property to be instantiated by something because one takes it to be instantiated by something in particular, and this is as true of the property of being such that Amelia talks as it is of other properties. One might, for example, believe the proposition that Amelia talks by taking the world to be such that Amelia talks; this fi ts nicely with the intuitive idea that in, for example, the case of belief, one “takes the world to be a certain way.” One might, however, wonder whether Nolan’s objection to the Chisholm/Lewis theory of the attitudes can be resurrected as an objection to the present view. Just as one can desire that one never have existed, can’t one desire that the world never have existed, or that there should have been nothing rather than something? Th is seems to be a coherent desire; but on the present interpretation, this would be the desire that the property of there being nothing be instantiated—which, given that a property is instantiated only if it is instantiated by something, is impossible. However, I think that this objection can be defused by distinguishing two interpretations of the desire that there be nothing rather than something. On the fi rst interpretation, the desire that there be nothing rather than something is something like the desire that there be no concrete things, or no material things. But (presuming that it is possible that there be no concrete things, or no material things) the properties assigned as contents to this desire are possibly instantiated, and so the desire comes out—as it should, on this interpretation—as a desire for something possible. Aft er all, in a world in which there are no concrete things, there will be things—necessarily existing abstract objects, for example—which instantiate the property of being such that there are no concrete things.

23 Th ough see Nolan (2006), 678–679, for some interesting discussion of how the view that all thought is self-involving in the relevant sense seems to follow from Lewis’ views about modality.

King020513OUK.indd 82 11/23/2013 12:58:57 PM PROPERTIES OF EVERYTHING OR NOTHING 83

On the second interpretation, the desire that there be nothing rather than something really is the desire that there be nothing —no concrete things, no abstract things, no things of any sort. It is true that, on the present view, the content of this desire turns out to be a necessary falsehood (if we assume, as I do, that a property cannot be instantiated unless something instantiates it). But this consequence does not seem to be objectionable, since it seems independently plausible that some things—like at least some abstract objects—exist necessarily. 24 So on neither interpretation of Nolan’s problematic desire does it pose a problem for the view that propositions are properties which, if instantiated, are instantiated by everything. Now it should be noted that the fact that my view avoids this sort of puzzle for the Chisholm/Lewis view—as well as the other objections raised against that view—comes at a cost. Th at theory was explicitly developed to explain the distinction between fi rst-personal and third-personal beliefs and, more generally, the distinction between indexical and non-indexical beliefs. And, as it stands, the view of the attitudes just sketched off ers no explanation of this distinction. In the next section I’ll return to the question of whether the sort of “property theory” I’ve been developing can capture some of the explanatory advantages of the Chisholm/Lewis theory.

Th e semantics of attitude ascriptions If propositions are properties, then it is natural to think that propositional attitudes are, not binary relation to a proposition, but ternary relations between subjects, properties, and the thing to which the subject attributes the property. Were this true, this might lead us to expect that ordinary belief ascriptions of the form

A believes that S should not express binary relations between subjects and propositions, but rather ternary relations, and hence to be of the form

A believes of o that it is F . where “ F ” stands for the property expressed by “ S ” in the context. However, there is no plausible candidate for the value of “o, ” for two reasons. First, (as noted above) there is nothing to stop two diff erent subjects from each believing that Amelia talks, but to do so by believing the property of being such that Amelia talks to hold of distinct things. Second, we would get into trouble with the modal profi les of attitude ascriptions if we supplied as value for “ o ” anything whose nonexistence was consistent with the truth of the ascription.

24 Th ough, even if it is not objectionable that the desire that there be nothing come out as impossible on this second interpretation, there still may be some oddness in treating it as the desire that a certain property be instantiated. But this oddness seems to be shared with the traditional view of desire as a propositional attitude, since on such a view the desire that there be nothing will be equivalent to the desire that a certain proposition be true.

King020513OUK.indd 83 11/23/2013 12:58:57 PM 84 JEFF SPEAKS

We can get around this problem by analyzing attitude ascriptions as existential gen- eralizations of the form

ූx A believes of x that it is F . which are true with respect to a world w iff the referent of “A ” in w believes of some object in w that it is F (where F is the proposition expressed by the complement of the ascription). Th is sort of view also stays a bit closer to the standard semantics for belief ascriptions, in that it does not take them to predicate a ternary relation of a subject, a world, and a proposition, but rather a binary relation between a subject and a proposition—albeit a binary relation defi ned by existential generalization on a ternary relation. 25 But this view also faces a serious problem. It appears that, given the truth of an ascription

A believes that S this view would license us to infer the truth of

ූx A believes of x that it is such that S . Th e problem is that (setting aside the, at present, irrelevant point that languages don’t contain singular terms for every object) such examples of quantifying into attitude ascriptions seem to entail, for some singular term n , the truth of

A believes that n is such that S . But then, if we re-apply our analysis of ordinary attitude ascriptions, we fi nd that this entails the truth of

ූx A believes of x that it is such that n is such that S .

which in turn, for some singular term n* , entails the truth of

A believes that n* is such that n is such that S .

and we’re off on a regress, the upshot of which is the implausible conclusion that having a single belief entails having infi nitely many beliefs of ever-increasing complexity.26

25 In this respect, the view of belief is similar to that defended in Salmon (1986). 26 Alexis Burgess suggested to me the possibility that this regress could be blocked by taking note of the fact that the initial existential generalization is a sentence of the meta-language, and that when we “re-apply our analysis” we’ve ascended to the meta-meta-language—and from the truth of some sentence of the meta-meta-language we are not automatically licensed to infer the truth of the corresponding sentence of the object language. Th ere may well be something to this idea—and it would be very nice, for my purposes, to have a response to this regress argument. But there’s an intuitive worry about this strategy. Th e sentences of the meta- (and meta-meta-) language will, like the sentences of the object language, presumably express propositions. So, for example, some proposition P will be expressed by the sentence “ ූx A believes of x that it is such that n is such that S” of the meta-meta-language. And some proposition Q will be expressed by just this string of words in the object language. Th e problem, it seems to me, is that it looks quite plausible that P will entail Q. But, if it does, then the regress argument goes through—even if we pay attention to the distinction between object and meta-language.

King020513OUK.indd 84 11/23/2013 12:58:57 PM PROPERTIES OF EVERYTHING OR NOTHING 85

Th e problem is that every de re ascription entails a corresponding de dicto ascription, so if we also analyze de dicto ascriptions as disguised de re ascriptions, we have, in eff ect, two valid rules of inference which can be alternated to generate, from a seemingly true belief ascription, ascriptions of arbitrarily complex beliefs. Th is shows that the idea that ascriptions should be analyzed as disguised existential quantifi cations into the complements of the ascription is a mistake.27 We can do better by setting aside the initial idea that we should think of attitude ascriptions as reporting ternary relations between subjects, properties, and the thing to which the subject attributes the property. Instead, we can—sticking more closely to the traditional view that ascriptions report dyadic relations between subjects and sui generis propositions—think of ascriptions as reporting binary relations between subjects and properties. Just as, on the traditional view, one might give an informal gloss on the belief relation by saying that it is “taking a proposition to be true” one might, on the present view, gloss that relation by saying that it is “taking a property to be instantiated.”28 Parallel things could be said about other propositional attitudes, which will be diff erent relations in which a subject can stand to a property; one might take it to be instantiated, desire it to be instantiated, suppose it to be instantiated, etc. Once we have this view of the propositional attitudes, it’s open to us to recognize the existence of the relation between subjects and properties with which Chisholm and Lewis identifi ed belief—namely, self-attribution. (Th is would be the “self-predicating” attitude analogous to belief; we could also recognize self-desiring, self-intending, and so on.) Recognizing the existence of such a relation is, of course, independent of the various claims about its relation to belief, and ascriptions of belief, which led to the troubles with the Chisholm/Lewis theory of belief. Taking the existence of this relation seriously allows us to capture some—though not all—of the explanatory advantages of the Chisholm/Lewis theory. Consider the problem of fi nding a distinction between mental states which are “fi rst-personal” and those which are not. On the present view, this shows up as a distinction between dif- ferent attitudes in which one can stand to properties. In the case of belief, this would be the distinction between my standing in the belief relation to the property of being such that JS is on fi re and my standing in the self-attribution relation to the property of being on fi re.29

27 We could solve this problem by adopting a suffi ciently coarse-grained view of properties to block the claim that the beliefs ascribed at various stages in the regress are genuinely distinct. But this would plug one hole in the theory only to introduce problems elsewhere, since we’d then face all the problems with belief ascriptions faced by coarse-grained theories of propositions. See, for discussion, Soames (1988) and Ch. 3 above. 28 Th ough this is just an informal gloss; the idea is not that belief is to be analyzed as a three-place relation between subjects, properties, and the property of being instantiated, since this would lead to the same sort of regress just discussed. 29 One might think that this is overkill: once we have the diff erence in attitudes, why also distinguish between their contents? Why not, in other words, also let the content of the fi rst-personal state be the

King020513OUK.indd 85 11/23/2013 12:58:58 PM 86 JEFF SPEAKS

Th is does not, by itself, solve parallel problems about indexicality which arise at the level of propositional attitude ascriptions—in particular, it does not explain why it seems to many competent speakers that, out of my mouth, “I believe that Jeff Speaks is F ” can diff er in truth value from “I believe that I am F .” But it might still help. One popular style of explanation of these speaker intuitions explains them away in terms of a confusion between the truth of the semantic content of a sentence and the truth of some proposition which would, in the relevant context, be pragmatically conveyed by an utterance of that sentence. Proponents of this strategy owe some account of the relevant pragmatically conveyed propositions, as well as an account of the mechanism by which those propositions are pragmatically conveyed by the relevant utterances. If we recognize the existence of the attitude of self-predication, then this might give us the beginnings of such an account. Perhaps when I utter an ascription ໠I believe that I am F ໡, I pragmatically convey the proposition that JS self-attributes the property denoted by “ F. ” 30 In the relevant cases, the truth-value of this proposition will come apart from the truth-value of the proposition that JS believes-to-be-instantiated the property of being such that JS is F . One might object that this supposed explanatory advantage is one shared with any reasonable theory, since any theory—whatever it says about the nature of propositions—can recognize the existence of the relation between subjects and properties which I’m calling “self-attribution.” But, to do this, proponents of some other view of propositions—whether they defend a non-reductive view, or aim to reduce propositions to an ontological category other than properties—must introduce a distinction between diff erent sorts of contentful mental states, treating some as relations to propositions, and others as relations to properties. Echoing Lewis, I “protest that the advantages of uniform objects are not to be lightly forsaken.”31 Th e fact that the property theory can recognize the existence of the attitude of self-attribution, while preserving a consistent treatment of contentful mental states as relations to properties, is a point in its favor.

Two unresolved puzzles I conclude by sketching two puzzles for the view sketched so far, which point out dif- ferent respects in which that view is incomplete.

property of being such that JS is on fi re? Th e reason is simply that I might self-attribute the property of being such that JS is on fi re without, intuitively, taking myself to be on fi re, if I’m suffi ciently confused about my identity. Th anks to an anonymous reader for pressing this concern. 30 One might also take this to be part of the semantics of attributions of fi rst-personal beliefs; I’m skeptical about this sort of view, since it would treat “believes” as ambiguous, which would lead to problems with capturing certain valid inferences from pairs of fi rst-personal and third-personal ascriptions. Th e problems here are analogous to the problems for Quine’s view that notional and relational senses of “believes” should be distinguished which are pointed out in Salmon (1995). 31 Lewis (1979), 532.

King020513OUK.indd 86 11/23/2013 12:58:58 PM PROPERTIES OF EVERYTHING OR NOTHING 87

A puzzling substitution failure Even if the view of attitude ascriptions sketched in the previous section is fi ne as far as it goes, it falls well short of a full treatment of that-clauses and reference to propositions more generally. Th is can be shown by consideration of a puzzle about attitude ascriptions—a puzzle which, I think, faces any attempt to treat propositions as denizens of some other ontological category (whether propositions, or facts, or mental event types, or sets). If “that”-clauses refer to propositions, and if propositions are properties, then it seems as though the following inference should be valid:

Bob believes that Amelia talks. Bob believes the property of being such that Amelia talks. But the conclusion seems barely coherent, let alone entailed by the premise. One might think that this is a place in which the present theory of propositions can come to its own rescue. Aft er all, given that loving is the relation denoted by “loves,” one might have thought that the following should be valid:

A loves B .

A loving B . But this is not valid, since the conclusion fails to express a proposition—and this fact (as noted above) is a fact which the view that propositions are properties promises to explain. Might we off er a similar explanation of the invalidity of the fi rst argument in terms of the fact that “Bob believes the property of being such that Amelia talks” fails to express a proposition? Perhaps, but it is not quite this easy. We can’t simply say that

Bob believes the property of being such that Amelia talks. fails to express a proposition because sentences of the form A believes the F . are ungrammatical since, in general, they aren’t—we can meaningfully and truly say Bob believes the proposition expressed by “Amelia talks.” Th is suggests that the conclusion of the problematic argument above is not ungrammatical, but simply false. Th is places a constraint on the sort of semantics for attitude ascriptions which the defender of the present view—who will have a hard time denying that “that Amelia talks” and “the property of being such that Amelia talks” refer to the same thing—can accept. In particular, it looks like the proponent of the view of propositions that I have been sketching will have to deny the following principle:

If an attitude ascription ໠ A V ’s x ໡ is true—where A is the name of the subject of the ascription, V is the attitude verb, and x is some term which refers to a proposition—then any other well-formed

King020513OUK.indd 87 11/23/2013 12:58:58 PM 88 JEFF SPEAKS

ascription which diff ers from this only by the replacement of x with another term for the same proposition must also be true. It can’t be denied that this principle has a great deal of initial plausibility. However, there is some independent reason to deny it.32 Consider, for example, the pair of sentences Joe fears that the Mets will win the World Series this year. Joe fears the proposition that the Mets will win the World Series this year. Th ese are both grammatical, but the fi rst is true, and the second is false—Joe may be afraid of many things, but propositions are not among them. And this is so despite the fact that, on any view, “that the Mets will win the World Series this year” and “the proposition that the Mets will win the World Series this year” refer to the same proposition. Th e proponent of the view of propositions as properties might seize on examples like this, and say that whatever explains the fact that these sentences about Joe’s fears diff er in truth-value can also explain the fact that our initial pair of sentences,

Bob believes that Amelia talks. Bob believes the property of being such that Amelia talks. can diff er in truth-value, despite the fact that “that Amelia talks” and “the property of being such that Amelia talks” both designate the same thing. To be sure, it is not obvious that this line of response is satisfactory, since it is not obvious that the explanation of the “fears” substitution failures will carry over to the example we are interested in; but the two sorts of examples do seem similar, so it is perhaps not unreasonable to hold that the right treatment of the “fears” examples will solve our own problem.33

32 For a much more in-depth discussion of these issues, see King (2007), chapter 5. 33 One plausible explanation of the examples involving “fears,” which is defended in King (2007) (153–163) is that some attitude verbs, including “fears,” are ambiguous. Th is will only help with our “believes” example if “believes” is also ambiguous; but it may not be implausible to say that it is. Consider, for example: Joe believes that the Reds will win the World Series this year. Joe believes Dusty. If we take these sentences at face value, then it looks like “believes” expresses diff erent (though obviously related) relations in these two sentences. Perhaps the problem with our troublesome pair of “belief” sentences is that Bob believes the property of being such that Amelia talks. forces the interpretation of “believes” exemplifi ed by “Joe believes Dusty.” Th is is, obviously, a view on which “believes” is ambiguous (though the two meanings are of course related). But it is nowhere near a systematic account of this ambiguity. A systematic account would have to explain why the fi rst interpretation of “believes” is triggered not just by belief ascriptions involving that-clauses, but also by sentences like the following: Bob believes what Amelia said. Bob believes the proposition expressed by what Amelia said. Bob believes the fi rst thing he hears every day. and so on. For an interesting discussion of some alternatives to this view that “believes” is ambiguous in this way, see King (2007), p. 157, note 39.

King020513OUK.indd 88 11/23/2013 12:58:58 PM PROPERTIES OF EVERYTHING OR NOTHING 89

A demarcation problem On the present view, propositions are properties. But are all properties propositions? Presumably not. Redness is a property, but, unlike the property of being such that Amelia talks, not a proposition. Th is view about the relationship between propositions and properties has, of course, many analogues in philosophy. Proponents of the view that mental states are brain states don’t for that reason think that every brain state is a mental state; and proponents of the view that propositions are functions—e.g. from worlds to truth-values—don’t think that addition, just because it is a function, must also be a proposition. But this leads to a question for the proponent of the view that propositions are properties: if not all properties are propositions, what distinguishes the ones that are from the ones that are not? A fi rst answer which suggests itself is that the properties which are propositions are the ones which are such that, if they are instantiated at all, are instantiated by everything. 34 But a moment’s thought shows that this won’t work, since the property of existence (if such there be) and the property of being self-identical are, if instantiated at all, instantiated by everything, and yet these properties don’t seem to be propositions. One might reply by pointing out a diff erence between the property of being such that Amelia talks and the property of self-identity: the latter, but not the former, is such that it is necessarily instantiated. So maybe we could say that propositions are such that (i) if they are instantiated at all, are instantiated by everything, and (ii) they are possibly not instantiated. Th e obvious objection, though, is that though (i) and (ii) may work for the property of being such that Amelia talks, they won’t work for the property of being such that 2 + 2 = 4, since this, like the property of being self-identical, satisfi es (i) but not (ii). A quite diff erent approach would be to demarcate the propositions not using an intrinsic criterion, but an extrinsic one, e.g. by saying that a certain property is a proposition iff it can be expressed by a sentence. But this looks similarly hopeless, partly because it is not obvious that every proposition is possibly expressed by a sentence, and partly because it seems to get the order of explanation backwards. Why, one wonders, is “My pants are on fi re” a sentence, whereas “Pants” is not? Presumably because the former expresses a proposition, and the latter does not. 35

34 Th is is parallel to the way that Chisholm demarcates propositions from states of aff airs; see Chisholm (1976), 122–124. 35 For an attempt to explain the distinction between sentences and non-sentences which does not appeal to propositions, see Davidson (2005). Other “extrinsic criteria” might be more plausible. One might (as Alexis Burgess suggested to me) try to explain which properties are propositions in terms of certain, perhaps modal, relations holding between the instantiations of properties. For the use of a strategy like this in demarcating the laws of nature, see Lange (2009).

King020513OUK.indd 89 11/23/2013 12:58:58 PM 90 JEFF SPEAKS

So, while it would be very nice to have a demarcation criterion, it’s not at all easy to see how to state one.36 But, while this is a vice, it is not a vice unique to the view that propositions are a kind of property. Proponents of the view that propositions are a sui generis category of entities also owe a statement of a demarcation criterion—and, typically, fail to provide one. Suppose we ask a proponent of that view why that grass is green is a proposition while being green is not. What should they say? Th ey might say: that grass is green belongs to the sui generis category of propositions, whereas being green does not. But is this any better than the proponent of the property theory of propositions taking the distinction between those properties which are propositions, and those which aren’t, as primitive?37 And, in fact, lots of views which identify the F ’s with some subset of the G ’s do so without identifying a criterion for distinguishing the G ’s which are F ’s from the G ’s which aren’t—a case in point is the view that mental properties are a subset of the physical properties. Proponents of that view typically don’t also provide an explicit criterion to divide the physical properties which are also mental properties from those which aren’t. If this is no decisive objection to that view, it’s not easy to see why it should be a decisive objection to the view that propositions are properties—even if this is something we should want a fully satisfactory theory of propositions to provide. 38

36 Th is would be one reason to prefer van Inwagen’s version of the property theory over mine. van Inwagen takes propositions to be 0-place properties; this provides a neat way of distinguishing the propositions that are properties from those that are not. Despite this, I prefer my version because I don’t understand what a 0-place property could be. 37 Friends of the view that propositions are sui generis entities might try to explain this distinction in terms of the fact that that grass is green is true or false, whereas being green is not—an explanation of which I cannot avail myself, since I take truth for propositions to be instantiation, and these properties are, equally, instantiated. But this move does not seem plausible since it involves taking truth as explanatorily prior to propositions—see for discussion Ch. 3 above. 38 Th anks to Alexis Burgess, Ben Caplan, Lorraine Juliano Keller, Matthew Lee, and Juhani Yli-Vakkuri for comments on previous versions of this essay, and to the participants in my graduate seminars at Notre Dame in the spring of 2008 and fall of 2009 for discussion of issues surrounding the metaphysics of propositions.

King020513OUK.indd 90 11/23/2013 12:58:58 PM 6

Cognitive Propositions

S c o t t S o a m e s

Th e conception of propositions I defend is based on the recognition that although propositions are needed to play the central roles assigned to them in theories of language, thought, and perception, they can’t do so as they traditionally have been conceived. 1 Th us, we need a new conception of what propositions are.

Th e role of propositions in language, thought, and perception Traditionally, propositions have been taken to play three interconnected roles: bearers of truth and falsity, objects of attitudes like belief and assertion, and meanings, or semantic contents, of sentences. Sentences are used to talk and think about things. For this to be so, they must represent those things as being various ways—as being so-and-so, as being not such-and-such, as being either such-and-such or so-and-so , and the like. To represent things as being a certain way is to impose conditions that must be satisfi ed if those things are to be as they are represented. Since a use of a sentence is true when things are the way they are represented to be, and false when they are not, these conditions are truth conditions. For a use of a sentence to be true or false is for what it is used to assert or express to be true or false. Since propositions are what is asserted or expressed, they are what, in the fi rst instance, represent things as being one way or another, and so have truth conditions. Th e meaning of sentences is explained through their connection with propositions. Sentence meaning is information conventionally associated with a sentence that constrains the propositions the sentence is used to assert or express across varying contexts. To put the same point another way, the meaning of a sentence provides the building blocks that combine with diff erent contextually relevant information to

1 Th e conception in this chapter elaborates, extends, and modifi es the one introduced in Soames, What is Meaning , Princeton and Oxford: Princeton University Press, 2010.

King020513OUK.indd 91 11/23/2013 12:58:58 PM 92 SCOTT SOAMES

yield the propositions the sentence is used to assert or express in diff erent contexts. Sometimes little or no supplementary information is required to reach these propositions. Sometimes, no proposition is expressed without it. Sorting out the contributions made by conventionally-encoded versus contextually-varying information to the propositions asserted or communicated by a use of a sentence is the business of semantics and pragmatics. A rough and ready criterion distinguishing the two is this: the meaning of an expression is information about it that an ideally rational agent would have to know in advance—independent of the varying information available in dif- ferent contexts—in order to reliably grasp what uses of sentences containing it assert, express, or convey. Since propositions are pieces of information that are asserted or expressed, they are the primary bearers of truth conditions, which are inherited by sentences or utterances that express them. As explained in chapter 3, the fundamental connection between truth, meaning, and propositions is expressed by the simple schema (1) (where “S” is a metalinguistic variable over sentences, and “P” is a schematic letter replaceable by sentences). 1. If a sentence S of L means that P, then S is true iff P. Roughly put, for S to mean that P is for S to express the proposition that P, and for S to be true is for the proposition S expresses to be true. Hence, the a priori obviousness of instances of (1) is reducible to that of the corresponding instances of (2). 2. Th e proposition that P is true iff P. Diff erent sentences can, of course, express the same propositions, which may be assumed, asserted, known, or believed. What are propositions, and what is the relation of expressing that sentences or utterances bear to them? A naturalistic account must avoid characterizing propositions as inhabitants of a Platonic third realm beyond mind and matter, with no explanation of how we come to bear attitudes to them, as well as how we are acquainted with, and come to know things about, them. It would be nice to be able to regard talk of the propositions expressed by sentences as simply talk about sentences at a level of abstraction that groups them together in terms of their representational features. However, since propositions have a life beyond language, this can’t be the whole story. One challenge facing any plausible theory of propositions is to make good on their independence from language without turning them into eternal, unchanging, representational contents that somehow become attached to sentences by an otherworldly expressing relation. Th is is one of the challenges I will address. Th ese general remarks about the role of propositions in theories of language hold whether or not the language under investigation contains singular terms referring to propositions, quantifi ers ranging over them, operators operating upon them, or predicates taking them as arguments. Whether or not a language contains such expressions, propositions are needed to explain the uses to which speakers of the language put it. In other words, propositions are needed to state the goals of semantic and pragmatic

King020513OUK.indd 92 11/23/2013 12:58:58 PM COGNITIVE PROPOSITIONS 93

theories for any language. Of course, when natural languages like English are involved, they are also needed within semantics proper as referents of that -clauses, arguments of attitude verbs, referents of some names and uses of indexicals, members of the domains of some quantifi ers, and so on. Propositions are also crucial to cognitive theories. To think about something is to think about it as being a certain way. So propositions, which represent things as being one way or another, are the contents of many cognitive states—such as one’s belief, doubt , or uncertainty that the economy will recover soon. Each of these attitudes is a relation in which an agent stands to a proposition. Although the picture generalizes, it is complicated by syntactic variation among verbs designating relations to propositional objects expressed by their complement clauses. One axis of variation divides verbs according to whether these clauses are fi nite (tensed) or non-fi nite (infi nitival). Further variation among verbs taking fi nite clauses separates those that also take complex noun-phrase objects such as the proposition/claim/ statement that the Earth is round, and the proposition/claim/statement that Martin asserted . Attitude verbs that take both fi nite clauses and complex nominal objects include assert, believe, know, deny, accept, reject, doubt, assume, refute, prove, establish; verbs that take fi nite clauses but not complex nominal objects include say, think, judge, see, perceive, desire, hope, expect, anticipate, suppose, hypothesize, imagine ; verbs that take fi nite clauses plus a few restricted complex nominal objects—but not the proposition/claim that S —include predict (that/the result that), regret , realize (that/the fact that); verbs that take both fi nite and non-fi nite clauses include believe, expect, assume, suppose, imagine, desire, prefer, wish . Th ere are also verbs like want that take only non-fi nite clauses. Despite this variation, it is plausible to suppose that these verbs have readings in which they express cognitive relations that hold between agents and propositions. If so, each should correspond to a cognitive state-type with propositional content. Propositions are also central to our understanding of perception. Seeing and hear- ing are relations between an agent and something else, oft en an object or event. Th ese perceptual states also represent the agent’s immediate environment, or things in the environment, as being certain ways. Imagine seeing a poster on the wall as red —in one case because it is red and the lighting is normal, and in the other case because it is illuminated by light that makes it appear red, even though it is white. If one’s phenomenal experience in the two cases is the same, then one’s visual experience represents the poster as red in both cases, even though it is actually red in only one. Th e fact that one’s perception represents it in this way is independent of whether or not one forms the perceptual belief that it is red—which one might do in either case, both cases, or neither. So we can specify the content of the perception, and evaluate its veridicality, whether or not the agent forms a belief with that content. Th e agent’s perception is veridical—i.e. accurate or truthful—iff the poster is the way that perception represents it to be—i.e. iff it is red. Th is suggests that although what one perceives is typically an object or event, the content of one’s perception is a proposition or set of propositions

King020513OUK.indd 93 11/23/2013 12:58:58 PM 94 SCOTT SOAMES

representing things as being certain ways. Since, perception, like cognition, is representational, it is a bearer of propositional content.2 In this way, we come to see propositions as the common thread tying together language, thought, and perception. Th eir essential feature is that they represent things as being certain ways, and so have truth conditions, which allow them to serve as contents not just of sentences, but also of thoughts and perceptions. Perceptual and cognitive contents are also among the things we use language to think and talk about. Th e fact that the same propositions can simultaneously function as linguistic, perceptual, and cognitive contents provides us with a systematic way of doing this. In the simplest case, we choose a sentence that expresses a proposition p that is part of the content of the perceptual or cognitive state we wish to characterize. Using a complement clause (e.g. a that clause in English) to designate p, we characterize the cognitive or perceptual content as having certain properties and standing in certain relations, e.g. to agents. In this way, we use sentences to express complex cognitive contents that represent other cognitive or perceptual contents as satisfying various conditions. As important as it is to recognize the commonality in linguistic, cognitive, and perceptual content, it is also important not to overlook their diff erences. Propositions are (or at least can be) bite-sized bits of information; they are the minimal units of their representational type. Individual sentences are thus their natural vehicles. Visual perception, the content of which is inherently holistic, stands at the other end of the spectrum. Although individual propositions can be abstracted as constituents of that content, the content of a visual state is most closely approximated by a network of propositions the elements of which are counterfactually connected to one another. Removing or changing one may be impossible without drastically modifying the rest, and hence the overall picture. Th e contents of belief and other cognitive states stand somewhere between atomistic language and holistic visual perception. Although belief reports are typically atomistic, the propositions correctly said to be believed are oft en parts of larger cognitive structures that exhibit some of the holism and counterfactual interconnection exhibited by perceptual states. Finally, the atomistic character of language—with propositions assigned to sentences (or utterances) one by one—is balanced by the vast syntactic and semantic resources that language puts at our dis- posal. As individual sentences become more abstract in content, and more syntactically and semantically complex, they come to express many propositions that cannot be constituents of the contents of any perceptual state, and—due to our cognitive limitations—that cannot be cognized by us except when presented linguistically. A good account of propositions should make room for all of this.

2 A temporal element is needed for the content of a perceptual state to be a complete proposition -- typically the time at which the perceptual experience occurs. Since that moment isn’t seen it may not be strictly part of the content of one’s (purely) visual experience. Th us, we may need a slightly broader conception in which perceptual and cognitive experiences are temporally supplemented. Th anks to Francois Recanati for raising this issue.

King020513OUK.indd 94 11/23/2013 12:58:58 PM COGNITIVE PROPOSITIONS 95

Before leaving perception, a further point should be noted. Perception, like cognition, is, I think, a cognitive activity in which we do something that results in the world being represented in one way or another. Th ink of Wittgenstein’s duck/rabbit example, or of an Escher drawing of a complex geometric structure. In the former case, a curved line can look either duck-shaped or rabbit-shaped. First we see it one way, then another. Once we realize that it can be seen both ways, we may try, with varying degrees of success, to move at will from a perception with one representational content to a perception with the other. Th e same is true when an Escher drawing of a building with a set of stairs that appears at one moment to be descending, and at another to be ascending. Similar experiences can be had in specially constructed rooms designed to create perceptual anomalies. To see what is before us fi rst one way, and then the other, is to fi rst predicate one property of what we see, and then to predicate a diff erent property of it. Somehow perception makes properties and relations available to us to put together in diff erent predicative patterns. How we see things— the predication we make—is usually automatic, unconscious, and so better described as a kind of cognitive operation than as a species of intentional action . But sometimes our experience makes multiple properties or relations available for predicating of the same things, either unconsciously or with a degree of conscious control—in which case our predications occasionally qualify as intentional. Either way, the important point is that putting together representational structures in perception and cognition is always a cognitive operation of some kind. 3 Th e simplest cases are those in which we predicate properties or relations of things that are given to us in perception or cognition, and thereby entertain a simple proposition, like the proposition that o is red, or

that o1 is bigger than o2 .

Toward a theory of propositions We now have two guiding ideas about propositions. (i) Th ey are pieces of information that represent things in the world as being certain ways; thus they have truth conditions. Since the proposition that o is red represents a certain object as red (while doing no further representing) it is true iff o is the way it is represented to be—red. (ii) To entertain a proposition in perception or cognition is to do something; to entertain the proposition that o is red is to predicate redness of o, and thereby to represent it as red. According to these two ideas, both the proposition that o is red and the agent who sees or thinks of o as red represent it as being red. Surely, these facts—that the

3 Th us, in what follows it should be understood that when I speak of “acts of predicating” properties of things I am not assuming that such “acts” are intentional.

King020513OUK.indd 95 11/23/2013 12:58:58 PM 96 SCOTT SOAMES

proposition represents and that the agent does too—are related. Presumably, one is the basis for the other. Which is fundamental? Is the fact that the agent represents o as red explained by (a) the fact that the agent has a certain attitude to the proposition that o is red, plus (b) the fact that the proposition—in and of itself and without interpretation by us—intrinsically represents o as red? Or is it the other way around? Is the fact that the proposition represents o as red explained by the fact that for an agent to entertain it is for the agent to represent o as red? The traditional answer is that propositional representation is primary and the agent’s representation is to be explained in terms of it. In What is Meaning? I argued that this approach is unsustainable. Here I will elaborate a theory based on the opposite approach. The guiding ideas are (i) that the perceptual and cognitive activity of agents is the conceptual basis of all representation and (ii) that propositions are representational in virtue of the relations they bear to this representational activity. Th e key move is to defi ne what propositions are in a way that makes the derivation of their representational properties from the representational activities of agents plausible. Th ink again about the proposition that o is red. It is the content of an occurrent perceptual or cognitive state whenever the agent predicates being red of o. Whenever an agent does this, a concrete event occurs, at a specifi c time and place, in which the agent predicates this property of that object. Th is suggests that the proposition that o is red is simply the minimal event type in which an arbitrary agent predicates being red of o. Th is event-type is representational because every conceivable instance of it is one in which an agent represents something as being a certain way. What it represents is what is representationally common to all such instances. Since every such instance is one in which o is represented as being red, we speak, derivatively, of the proposition itself representing o as red. Since nothing else is representationally common to all conceivable instances of the proposition, representing o as being red exhausts its representational content. Otherwise put, to entertain the proposition that o is red is to predicate redness of o, which is to do something that results in an instance of the event type that the proposition is. Th e representationality, and hence truth conditions, of the proposition are due to the representational features of these possible instances. Th e proposition represents o as red, and nothing further, because what it is to entertain it is simply to predicate redness of o, and so to represent o as red. From this we derive its truth conditions: the proposition is true iff whatever (namely o) it represents to be a certain way is that way (red); it is false iff o isn’t red. Th ese conditions can be modalized. For every metaphysically or epistemically possible world-state w the proposition that o is red is true at w iff at w, o is red—which is just to say that for each such world-state, if it were instantiated, then the proposition that o is red would be true iff o were red. Th us, we explain how it is that the proposition that o is red has its truth conditions essentially.

King020513OUK.indd 96 11/23/2013 12:58:58 PM COGNITIVE PROPOSITIONS 97

In this way, we solve the real problem of the unity of the proposition that defeated Frege and Russell—which, as I argued in chapter 3, also defeats (along with other problems) the possible-worlds conception of propositions. As indicated there, the problem is to explain how propositions manage to represent the world, and so have truth conditions from which the truth conditions of sentences, utterances, and cognitive states can be inherited. While traditional accounts correctly recognize that propositions must have their intentional properties inherently — independent of any need for further interpretation by us—such accounts err in taking this to mean that their intentionality can’t be explained by the relation they bear to the actual and possible cognitive activity of agents. The conception of propositions as cognitive event types saves us from this error by identifying the intentionality-explaining relation that propositions bear to agents, not with being interpreted by us (to mean so and so), but with having instances in which we represent things (as being so and so). Since the proposition that o is red is the event type in which an agent predicates redness of o, it represents o as being red because all conceivable instances of it are events in which an agent does so. The intentionality of the event type is inherent to it in the sense that the event-type couldn’t be what it is without bearing its intentional properties, even though it does so by virtue of a relation it bears to agents. Being inherently intentional, it can be the interpretation of sentences and utterances, without itself being the sort of thing for which an interpretation is needed.

Complex propositions and attitudes Entertaining a proposition is the most basic attitude we bear to it. It is the attitude on which the others are, in one way or another, based. For example, to judge that o is red is to predicate redness of o while affirming or endorsing that predication. To believe that o is red is to judge, or be disposed to judge, that it is. To know that o is red is, roughly, for o to be red, to believe that o is red, and to be justified in so believing. To assert that o is red is to commit oneself, by uttering something, to treating the proposition that o is red as something one knows. There are, of course, more complex attitudes, like denying, refuting, proving, and more. Rather than discussing these, I will go into further detail about what it is to entertain complex propositions. To entertain the proposition that it is not true that o is red is (i), to predicate redness of o, and thereby to entertain the proposition that o is red (ii), to negate the property being true, and (iii) to predicate the resulting property not being true of that proposition. Th is can be done by thinking “Th at’s not true,” referring to the result of the initial predication—provided that one can so refer. Many, but not all, agents capable of entertaining the original proposition can do this. Th ere is nothing inherent in the ability to

King020513OUK.indd 97 11/23/2013 12:58:58 PM 98 SCOTT SOAMES

entertain p that guarantees that one can think thoughts about p. Th e minimal form of acquaintance with propositions is the ability to cognitively or perceptually represent the world by predicating properties of objects, thereby generating tokens of event-types corresponding to those predications. To gain a more robust form of acquaintance one must be able to make propositions targets of predication. Th is requires the ability to focus on the concrete events in one’s cognitive life, recognize their similarities, and group together those bearing relevant similarity relations into units or types. Since the proposition that o is red is an event type in which one predicates redness of o, one, who can focus on particular events in one’s cognitive life, and reliably group together those in which one predicates this property of this object, is in a position to make the proposition that is the event type of which they are instances, an object of thought. Given the means both of thinking of o as red , and of becoming aware in this way of so doing, one can then make further predications about the proposition that o is red, which was the content of one’s initial thought. For example, one may think, “Th at’s not true,” thereby predicating untruth of the proposition that is the type of cognitive event one has just experienced. So far, I have mentioned two operations involved in proposition formation, negating properties and predicating them of objects. In addition to negating properties, agents also conjoin them. We entertain the proposition that o is red and round by conjoining being red and being round , and predicating the result of o. In these cases, functions are applied to properties. In others they are applied to objects, or to other functions. Th us, cognitively primitive agents don’t need to predicate properties of propositions to believe that o isn’t green or that o is red and round . Th ink of the function Neg as 2-place relation in which the identity of its fi rst argument determines the second. When P is a property, NegP is a property uniquely true of the property which is P’s negation. An agent acquainted with NegP can predicate the property it determines of an object. Property conjunction is similar. What about conjunctive and disjunctive propositions ? One way of approaching the problem would be to start with relations R& and RV. Predicating these of a, redness, and b, roundness represents a as red and b as round (and only this), and a as red or b as round , respectively. To perform this predication is to do something that approximates entertaining the conjunctive and disjunctive propositions that a is red and b is round, and that a is red or b is round. To believe the propositions one entertains is to be disposed to endorse the predications. To believe their negations is to believe propositions in which one negates R& and RV to get the relations ~R& and ~RV, which are then predicated of the relevant arguments, just as R& and RV were. None of these beliefs requires making propositions predication targets. By taking R& and RV to be 2-place relations each argument of which is an n-place property followed by an ordered n-tuple, one can get the eff ect of embedding propositions formed using them under R& and RV themselves, thereby making the equivalent of full truth-functional cognition possible for agents that can’t, for whatever reason, refl ect on their own cognitive acts or experiences.

King020513OUK.indd 98 11/23/2013 12:58:59 PM COGNITIVE PROPOSITIONS 99

However, we don’t have to rest content with this approximation of more familiar thoughts about truth-functional cognition. R& and RV are complex relations predicated of pairs of n-tuples of the constituents of arbitrary pairs of propositions, where each such proposition is itself the predication of a property or relation of its other constituents. Using this model, one can, for each truth functional compound of propositions, generate an equivalent proposition that is itself a predication of such a pair of n-tuples. However, we can also generate propositions that are genuinely truth-functional compounds of other propositions. Let p be a proposition that represents things as being so-and-so (and nothing more) and q be a proposition that represents things as being such-and-such (and nothing more). Next consider a certain disjunctive operation the application of which to p and q represents things as being so-and-so or things as being such-and-such (and nothing more). To entertain this proposition is to entertain p, to entertain q, and to operate on them in this way—where operating on them isn’t predicating anything of them. Let the result be a disjunctive proposition. Conjunctive propositions can be treated in the same way. So can negations of propositions; when p represents things as being so-and-so , to negate p is to represent things as not being so-and-so. On this model, all propositions involve predications at some level, but some propositions are properly characterized as operations on constituents that themselves are, or depend on, predications. In what follows, I will sometimes ignore this complication, e.g. when speaking of entertaining an arbitrary proposition as predicating a property of certain things. In all such cases, a more complicated statement involving predicating a property of those things or operating on them can be supplied, without aff ecting the larger point at issue. We are now ready for a more general sketch of propositions. Th e simplest are those in which properties are predicated of objects. Complex propositions may involve other operations such as conjoining, disjoining, and negating properties or propositions, as well as operating on, for example, a two-place relation R to form the refl exive, one-place property self-R-ing . Th ey may also involve applications of functions to objects, or to properties (or propositional functions). In addition, some complex propositions involve the ascription of higher-order properties to lower-order properties (or propositional functions) as in quantifi cation. Propositions of any sort may also be arguments of further predications, which we fi nd in modal propositions and attitude ascriptions. For example, the proposition that necessarily it is not the case that Kripke is Kaplan is the event type of (i) predicating identity of the pair of Kripke and Kaplan (ii) predicating untruth of, or applying the negation operation to, the event type of which the previous predication is an instance, and (iii) predicating being necessarily true of the complex event type of which the second predication or operation is an instance. Th e proposition that John believes that Kripke is Kaplan is the event type of (i) predicating identity of the Kripke and Kaplan, and (ii) predicating the belief relation of the pair consisting of John and the event type of which the fi rst predication is an instance. Further detail is provided by the following illustrations. Th e proposition that Cicero is wise is the event type of predicating being wise of Cicero; the proposition that he is

King020513OUK.indd 99 11/23/2013 12:58:59 PM 100 SCOTT SOAMES

eloquent and wise is the event type of fi rst conjoining being eloquent and being wise , and then predicating the result of Cicero; the proposition that Tully shaved Cicero is the event type of predicating the shaving relation of Cicero and Cicero; and the proposition that he shaves himself is the event type of operating on the shaving relation to get the property being one who shaves oneself, and predicating it of Cicero. Th e proposition that 6 cubed is greater than 14 squared is the event type of applying the cubing function to the number 6 and the squaring function to 14, and predicating being greater than of what results from these applications. Functional application is also at work with Fregean defi nite descriptions, which are singular terms formed from attaching “the”

to a formula. “Th e” denotes a function f the that maps a propositional function g onto

the unique object to which g assigns a true proposition, if there is one; otherwise fthe

is undefi ned. Th e proposition that the G is H is the event type of applying fthe to g, and predicating being H of whatever results from that application. Th e proposition that

all Gs are H is the event type of (i) applying the function fall to g, yielding the property being true of all objects to which g assigns a truth, and (ii) predicating this property of the propositional function h . 4 More generally, the proposition that some G is so and so — expressed by the sentence ⌜ (Some x: Gx) (. ..x. . .x. . .)⌝ —is the event type in which one predicates the property being true of some object to which g assigns a truth (which results

from applying f some to g) of a certain semantic value associated with (. ..x . . .x . . .) . Which

value? Consider the proposition p o expressed by the formula relative to an assignment of object o to “x” (which, for simplicity, I stipulate to be the only variable with free

occurrences in the formula). P o will be an event type consisting of a sequence of event types involving predications and other cognitive operations, where event types i and j in the sequence involve operations on o corresponding to the free occurrences of “x” 5 in the formula. Let f so-and-so be the function that assigns to any object o’ the proposition

p o’ that diff ers from op only (if at all) in that the ith and jth event types in the sequence

of event types that comprise po ’ involve cognitive operations on o’ (rather than o). Th e proposition that some G is so and so —expressed by ⌜(Some x: Gx) (. ..x. . .x. . .) ⌝—is the event type in which being true of some object to which g assigns a truth is predicated of

the propositional function f so-and-so . Th e same basic mechanism accounts for propositions expressed by sentences involving lambda abstraction, as well as those containing anaphora of arbitrarily long distance. 6

4 Th ese functions are properties, not ordered sets. An n-place function is an n+1-place property R such

that o’ = o*, if R o1 . . .o n o’ and Ro 1 . . .on o*. Th e cubing function is a 2-place property that combines with n to determine a property being the product of n times n times n. Th is determines the subject of the predication in

the proposition that n cubed is odd. With Fregean defi nite descriptions, the function fthe is a 2-place property that combines with an argument g to determine a property being an object that is unique in determining a true proposition when taken as argument of g. Th is determines the subject of predication in the proposition

the F is G. With quantifi cation, applying the function f all to g gives us the property being true of all objects to which g assigns a truth, which is the property predicated (rather than merely determining that property). Th ese points are connected with distinctions made in the fi nal section of this chapter. 5 When both occurrences of “x” are in the same simple clause, event type i = event type j. 6 My use of propositional functions, rather than complex properties, is merely a convenience. I take no stand on which way of fi lling out the theory is to be preferred.

King020513OUK.indd 100 11/23/2013 12:58:59 PM COGNITIVE PROPOSITIONS 101

At this point, a word must be said about how I am using the verb “predicate.” I begin here with the account previously given in What is Meaning? According to that account, the verb “predicate”, needed by the conception of propositions as cognitive-event types, is analogous to the intensional transitive “look for.” If Bill is looking for Maria, and Maria is Mary, who, in turn, is the chief of police, then Bill is looking for Mary, but it doesn’t follow (on one reading) that he is looking for the chief of police. It also doesn’t follow from the fact that he is looking for the fountain of youth that there is such a thing. Analogously, if Bill predicates P of x, and x is identical to y, which, in turn, is the unique F, then Bill predicates P of y, but it doesn’t follow that he predicates P of the F. It also doesn’t follow from the fact that he predicates P of the F that there is an F.7 Like an intensional transitive (which expresses a cognitive relation between an agent and a content) the verb “predicate” expresses a cognitive relation between an agent, a property, and a content. So, if we treat defi nite descriptions as singular terms, the proposition that the king of France is wise will be the event type predicating being wise of the king of France, even though there is no king. Th e truth of the proposition depends on there being something of which being wise is predicated, but its existence doesn’t. (Th is account of the content of “predicate” will suffi ce until the fi nal section of this chapter when it will be modifi ed to accommodate special examples considered there.) More can be said about the existence conditions of propositions and other event types. First, if an event type E has instances that exist, then E exists. For example, since I can (directly) refer to Socrates even though he no longer exists, if I do so refer, then a concrete event e exists that is an instance of the minimal event type in which one refers to Socrates—which must also exist. Ditto for the (minimal) event type in which one predicates no longer existing (directly) of Socrates—which is the proposition that Socrates no longer exists. Since, in certain cases, one can also (directly) refer to merely possible individuals, there exist propositions—event types of predicating properties of those individuals—the “constituents” of which have never existed and never will. 8 To understand this, one must not confuse failing to refer with referring to a non-existent. “Th e present king of France” fails to refer, and so has no referent; “Socrates” has a referent, just one that doesn’t exist. Th e view endorsed is not Meinongian— there is no such thing as the golden mountain, whether existent or non-existent. Th is non-Meinongian view eliminates an alleged problem for Millians: namely, that if one of the so-called “constituents” of a singular proposition fails to exist then the proposition also fails to exist. Th is false claim relating the existence of a proposition to the existence of its constituents comes from thinking of propositions in the wrong way.

7 “P” is here used as a variable over properties, while “F” is used as a schematic predicate letter. 8 For more on referring to, or quantifying over, the non-existent, see Nathan Salmon , “Existence,” Philosophical Perspectives, 1 , 1987 , 49–108 ; Scott Soames , “Actually,” in Mark Kalderon , ed., Proceedings of the Aristotelian Society , supplementary volume, 81, 2007 , 251–277 , reprinted in Soames, Philosophical Essays, Vol. 1: Natural Language: What it Means and How we Use it , Princeton and Oxford : Princeton University Press , 2009 ; and pp. 128–129 of Soames, Philosophy of Language , Oxford and Princeton: Princeton University Press, 2010.

King020513OUK.indd 101 11/23/2013 12:58:59 PM 102 SCOTT SOAMES

When a proposition is the event type of predicating a property of an object o, o may be a constituent of the proposition—in the sense that the proposition is defi ned in terms of o—without o’s existence being necessary for the existence of the proposition. Although this is progress, it doesn’t provide solutions to all problems posed by so-called “empty” names. For example, if the name “Vulcan” fails to refer (as opposed to picking out a real, existing character in a story/theory/legend), then, since the content of a name is its referent, we can’t correctly characterize any agent as “predicating a property of Vulcan”—in which case, either the sentence “Vulcan is a planet” fails to express a proposition, or it expresses one that is radically incomplete. By contrast, we can correctly characterize an agent as “predicating a property of the present king of France,” because that claim relates the agent, not to an individual, but to the meanings of “the” and “present king of France.” Th us, sentences containing the (Fregean) description express propositions.9 Here is another plausible principle about the existence conditions of propositions if (i) R is an n-place property for which there have been events in which an agent predi-

cates R of things, and (ii) o 1 . . .on are objects for each of which there have been events in which an agent thinks of or refers to it, then (iii) there exists a proposition p which is

the (minimal) event type of targeting o 1 . . .on and predicating R of them— even if no one has ever performed that predication, and hence there exist no instances of p . 10 Consider the analog with sentences. If R is an n-place predicate that has been used by an agent, ⌜ ⌝ and t 1 . . .tn are names, each of which has been used, then the sentence type R t 1 . . .tn exists—even if it has never been uttered or inscribed. If we take sentences to be complex event types in which agents produce auditory, visual, or tactile tokens, then the principle needed to guarantee the existence of the usual infi nity of sentences of English will be an exact analog of the one suggested for propositions as event types. On this view, since propositions and sentences are complex event types that involve the performance of certain basic acts, their existence is guaranteed by the existence of events in which those acts are performed. In this way, we come to recognize the existence of many propositions that have never been entertained. Still, one might worry, if propositions are event types, some propositions won’t exist that should. One might have supposed that for each molecule in the universe, the proposition that it is a molecule exists and is true. Since many molecules have never been thought of or referred to by any agent, nothing guarantees the existence of these propositions. Th is would be a problem if propositions had to exist to be

9 Although Fregean defi nite descriptions are singular terms in some possible languages, I do not assume that defi nite descriptions in English are Fregean. Rather, I take them to be generalized quantifi ers. 10 In a complete account, this principle—which covers only atomic propositions—would be extended to include complex propositions as well. Th e idea behind the extension is this: Let the event-type p be a proposition; let the basic acts the performance of which defi ne the sequence of event types that make up p be targeting a given object o, predicating a particular simple property p, applying a certain function f, targeting a specifi c argument g, negating something, conjoining a pair of things, etc.; then p will exist if each of those acts has been performed.

King020513OUK.indd 102 11/23/2013 12:58:59 PM COGNITIVE PROPOSITIONS 103

true. But they don’t. Although many properties require things that have them to exist, some don’t. An individual can have the properties being dead, being referred to by me, and being admired by someone despite not existing. Similarly, a pair of individuals— Plato and Aristotle—can instantiate the relation of non-identity without existing. By the same token, a proposition can represent something as being a certain way, and so be true because the thing is that way, whether or not the proposition exists. Th us, there is nothing to prevent the nonexistent proposition that m is a molecule from being true. Since we can quantify over the merely possible, we can quantify over possible propositions, and say that if p predicates being so-and-so of o, then p is true (at world-state w) iff (at w) o is so-and-so, whether or not p exists (at w). Propositions that couldn’t be entertained by any agent might require a further story, as do metaphysically impossible objects generally, but even here it is not obvious that there are irresolvable diffi culties. Perhaps we will have to draw a line excluding propositions that couldn’t conceivably be entertained by any agent, but if so, would that really be a loss? Although a proposition p can be true at a world-state w without existing at w, p can’t be entertained at w, accepted at w, asserted at w, denied at w, or judged at w to be true without existing at w. To bear any of these attitudes to p at w an agent must entertain p at w. Since in each case, this involves producing an instance of the event type that p is, bearing any of these attitudes guarantees the existence of p. So, apart from a few minor complications—e.g. to believe p doesn’t strictly require one to have entertained p, but only to be disposed to bear the judging relation to p—when we really need the existence of propositions as objects of attitudes they are (nearly enough) guaranteed to exist.

Some attractions of the view Th at is the basic view, which also nicely complements an attractive version of defl a- tionism about truth. According to the conception of propositions as event types, the proposition that o is red predicates redness of o, and so represents o as red, while the claim that the proposition that o is red is true says, in eff ect, that o is as the former proposition represents it to be. Not only are these claims obviously equivalent, any warrant for accepting one is warrant for accepting the other. Defl ationism about the truth of propositions is a natural addition. According to this view (i) p and the proposition that p is true are necessary and a priori consequences of one another, (ii) any warrant for believing, asserting, or assuming one is warrant for taking the same attitude toward the other (subject to certain minor complications), and (iii) theses corresponding to (i) and (ii) also hold for the negation of p and the proposition that p is not true . 11 Since parallel theses about sentences don’t hold, sentential truth is not defl ationary.

11 See Soames , “Understanding Defl ationism,” Philosophical Perspectives, 17 , 2003 , 369–383 , also in Soames , Philosophical Essays: Vol. 2: Th e Philosophical Signifi cance of Language , Princeton : Princeton University Press , 2009 .

King020513OUK.indd 103 11/23/2013 12:58:59 PM 104 SCOTT SOAMES

Rather, nondefl ationary sentential truth is defi ned in terms of defl ationary propositional truth. A sentence is true just in case it (semantically) expresses a proposition that is true—where for a sentence to express a truth requires certain conventions to hold among language users. However, at this point we are faced with a puzzle. Why, if p and the claim that p is true are so symmetrically related, are we inclined to think that the proposition that snow is white is true because snow is white, while resisting the thought that snow is white because the proposition that snow is white is true? Well, why is the proposition that snow is white true? It is true because (a) it represents snow as white, and (b) snow is white. Since the fact that it represents snow as white is no part of the explanation of why snow is white, we rightly reject the claim that snow is white because the proposition that snow is white is true. Since the fact that snow is white is part of the explanation of the fact that the proposition is true, we say that the proposition is true because snow is white. Th e present conception of propositions contributes to this solution by explaining what this talk of propositional representation amounts to, in terms of how agents who entertain it represent things. Th e second attractive feature of this conception is that it provides a naturalistic account of the epistemic relations we bear to propositions. Unlike the Platonic epistemology required by traditional theories of propositions, the present account demys- tifi es our acquaintance with, and knowledge of, propositions by taking both to be grounded in concrete cognitive experience. Th e explanation starts with the idea that we predicate properties of objects in cognition and perception, thereby entertaining propositions. Th is is done before we have the concept proposition. Focusing on similarities and diff erences in our experience, we eventually acquire the concept, making propositions objects of thought and subjects of predication. Th is allows us to acquire the notion of truth, in part by being given numerous examples—“the proposition that o is red is true if o is red, the proposition that o is red isn’t true if o isn’t red,” etc.—and in part by coming to recognize the general point that a proposition is true iff things are as it represents them to be. Given truth, properties can be conceptualized as things true of other things . With the concepts truth, property, proposition and modality (what could be but isn’t) under our belts, we can characterize world-states as ways for things to be— maximally informative properties that the world could have had. Such a world-state w can be defi ned as the property of making true a set w* of basic propositions that tell a complete world-story. A proposition p is true at w iff p is an a priori consequence of w*. So, we can come to know that p is true at w by deriving p from w*. As for the actual world-state @, we can come to know p to be true at @ , given knowledge of p, by noting that since p is true, it must be true at this very world-state—the one that is instantiated. 12 Th e third advantage of the cognitive event-type conception of propositions is that it opens the door to demystifying the relationship between propositions and the

12 See Soames, “Actually,” and chapters 5 and 6 of Philosophy of Language for fuller explanations.

King020513OUK.indd 104 11/23/2013 12:58:59 PM COGNITIVE PROPOSITIONS 105

sentences that express them. Just as propositions are event types in which agents perform certain representational cognitive operations, so sentences may plausibly be taken to be event types instances of which are utterances and inscribings—thought of as concrete events occurring at particular times and places in which agents produce auditory, visual, or tactile tokens endowed with semantic and syntactic properties. On this picture, since both sentences and propositions are event types, they can share common instances: e.g., cases in which the event that is one’s referring to (targeting) o and predicating P of o is the event of one’s using expression E to refer to (target) o and expression F to predicate P of o. Th e cognitive acts or operations— referring to (targeting) o and using E to do so —are diff erent, as are the acts or operations of predicating P of o and using F to do so . Th us, the event type p that consists of one’s performance of the propositional acts or operations (of referring and predicating) diff ers from the event type S that consists of one’s performance of the sentential acts or operations (of using E to refer and F to predicate)—even though some instances of one are also instances of the other. When an event is an instance of both a sentential and a propositional type, there is no extra inner event of “grasping the proposition” over and above using the sentence meaningfully. So, when S expresses p, one who understands S can entertain p by tokening S. For some propositions, this may be our only feasible way of entertaining them. In such cases what distinguishes p from S is the possibility that an event could be an instance of one but not the other. More generally, the heretofore mysterious expressing relation holding between a sentence and a proposition may be grounded in something like the by relation that holds between two things that are done when an agent can do one of those things (entertaining the proposition) by doing the other (uttering or inscribing the sentence). Th e fourth advantage I will mention here is the prospect for illuminating otherwise puzzling semantic phenomena. Here is an example. 13 3a. Russell defended the proposition that arithmetic is reducible to logic. b. Russell defended logicism. 4a. Mary believes that Russell defended the proposition that arithmetic is reducible to logic. b. Mary believes that Russell defended logicism. “Logicism” is a Millian proper name for the proposition that arithmetic is reducible to logic , which is also designated by the directly referential that-clause. Nevertheless sentences (3a) and (3b) express diff erent propositions, and the truth of (4a) guarantees the truth of (4b), but not vice versa. “Logicism” and the that -clause contribute the same proposition L to those expressed by the sentences in (3) and (4). But the clause somehow

13 Th is example is discussed in Mark Richard , “Articulated Terms,” Philosophical Perspectives , 7 , 1993 , 207–230 ; and Soames , “What are Natural Kinds?,” Philosophical Topics , 1 and 2 , 2007 , 329–342 .

King020513OUK.indd 105 11/23/2013 12:58:59 PM 106 SCOTT SOAMES

also contributes something else to the propositions expressed . Th e view of propositions as cognitive event types explains what and why. According to it, understanding sentence (3b) and entertaining the proposition it expresses requires one to think of L, and to predicate having defended of the pair consisting of Russell and L. Since one can think of L simply by possessing the name “logicism,” without knowing much about its referent, one who is competent with the name, and accepts sentence (3b), can entertain, and even believe, the proposition it expresses without being able to state, or informatively identify, L. By contrast, in order to understand sentence (3a) and entertain the proposition it expresses, one must fi rst predicate being reducible of the pair consisting of arithmetic and logic—thereby entertaining the proposition L expressed by the that-clause. Next, one predicates having defended of the pair of Russell and L. Th is diff erence carries over to (4a) and (4b), with the result that the truth of the former requires the truth of the latter, but not vice versa. Because propositions are event types that involve thinking of things and predicating properties of them, two propositions can place diff erent constraints on how an agent thinks about their common predication targets, even if the truth conditions of the two propositions result from predicating the very same properties of the very same targets. Although (3a) and (3b) predicate the same thing of the same targets, the former is an event type in which the propositional coor- dinate must be cognized by entertaining it, while the latter is an event type that doesn’t require this. Th e diff erence in truth value between (4a) and (4b) is sensitive to this. In this way, taking propositions to be cognitive event types brings together the two related but distinct aspects of linguistic and cognitive content. On the one hand, such content faces the world—imposing conditions that must be satisfi ed, if the world is to conform to the way it is represented to be. On the other hand, this content also faces the mind, imposing conditions on what it takes for an agent to entertain it. Whereas the worldly aspect of content has long been accommodated in semantics, it has been diffi cult to do justice to the mental aspect of content when integrating the two. Th e conception of propositions as representational cognitive event types provides us with a natural way of achieving this. Being representational, the truth conditions of propositional event types– in virtue of which they “face the world”—are essential to them. Being event types in which one performs cognitive acts or operations, propositions can impose diff erent conditions on the cognitive operations it takes to entertain them, even when they are representationally identical in the sense that their truth conditions are derived from predicating the very same properties of the very same things. Th e fact that we need to recognize propositions that do diff er in this way strongly supports a metaphysics of propositions that explains how this is possible.

New light on attitudes de se versus attitudes de re Having the right metaphysics of propositions also opens up a new line of research on a central problem in semantics and cognitive science involving what have been called

King020513OUK.indd 106 11/23/2013 12:58:59 PM COGNITIVE PROPOSITIONS 107

de se attitudes. 14 Although these attitudes initially appear to be special cases in which an agent asserts, believes, or knows an ordinary proposition, every attempt to identify that proposition has proved problematic. Here is a sketch of one well-known version of the problem. 15 (i) Pushing his cart down the aisle at the supermarket, John Perry notices a trail of sugar leading around the corner to the next aisle. Looking up at the anti-theft mirror attached to the ceiling, he sees the somewhat distorted image of a shopper leaving a trail coming from a torn bag at the bottom of his cart. Perry says to himself “He is making a mess,” and tries to fi nd and inform the man. Pushing his cart faster and faster through the aisles leads to more and more sugar, but not to the messy shopper—until, in a fl ash of recognition, Perry comes to realize that he is the messy shopper, which he expresses by exclaiming “I am making a mess.” (ii) What did Perry say, and come to know (that he didn’t know already)? It can’t be the ordinary singular proposition that predicates being one who is making a mess of him—since he came to know that when he saw (but didn’t recognize) his own image in the mirror. It can’t be the metalinguistic proposition that the sentence “I am making a mess’ is true in my context, since that too is a singular proposition that Perry should already have known by virtue of knowing that the sentence “I am a mess” is true in his [pointing at his refl ection in the mirror] context. Th e propositional object we are looking for also can’t be the descriptive proposition that the man with the torn bag at the bottom of his cart is making a mess, since Perry already knew that too. Might it be the proposition that the man named “John Perry,” who teaches philosophy at Stanford, got his PhD at Cornell, and grew up in Nebraska is making a mess? Th at’s not likely either, since the essential problem would remain even if Perry had forgotten his name, misre- membered some details of his biography, or even had amnesia. 16 Similar objections can be mounted against other candidates for the crucial proposition that he came to know and assert in his moment of epiphany. Nor will it help to include reference to that moment, since the problem of identifying propositions expressing genuine fi rst-person knowledge, belief, or assertion arises in a new form when we try to identify the proposition we come to know when, in a moment of insight, we exclaim at time t “My word, the meeting starts now,” having known all along the ordinary singular proposition that the meeting starts at t. Th e fundamental problem illustrated in this scenario arises (in part) from the common, but oft en implicit, assumption A.

14 For classic discussion, see David Lewis , “Attitudes De Dicto and De Se,” Th e Philosophical Review , 88 , 1979 , 513–543 . 15 J o h n P e r r y , “ Th e Essential Indexical,” Nous , 13 , 1979 , 3–21 . 16 See also, John Perry , “Frege on Demonstratives,” Th e Philosophical Review, 86 , 1977 , 474–497 .

King020513OUK.indd 107 11/23/2013 12:58:59 PM 108 SCOTT SOAMES

(A) All there is to a proposition is its representational content; hence propositions the truth conditions of which arise from predicating precisely the same properties of precisely the same things, are identical. One reasonable (though not theoretically neutral) description of all de se cases is that they turn on systematically diff erent ways of believing/asserting representationally identical things—in particular on believing or asserting in the special fi rst person, or immediate present tense, way versus believing/asserting in a person-time neutral way. Although believing/asserting in the fi rst way is generally thought to guarantee believing/asserting in the second way, the converse doesn’t hold. De re cases are those in which we have the latter without the former. In the presence of (A), this means that the agent’s de se epiphany can’t be a matter of coming to assert, believe, or know any proposition not already asserted, believed, or known. Th us the conventional wisdom about the supermarket example has been that there is no proposition that Perry came to assert, believe, and know that he hadn’t already asserted, believed, and known. Th is “wisdom” comes in two opposing forms. According to Perry, the epiphany is not one of coming to believe or know a new proposition; it is one of coming to believe or know an old proposition in a new (fi rst-person/present-tense) way. According to David Lewis, it involves coming to bear the primitive attitude s elf-ascribing to a certain property P—where this primitive attitude must not be confused with the ordinary attitude ascribing to an individual who happens to be oneself (which Perry already bore to P when he saw his image in the mirror). Th e property P is, of course, being one who is making a mess , which, according to Lewis, is what Perry came to know and believe in his moment of epiphany. Both the views of Perry and those of Lewis are revisionary, in the sense of explaining away, rather than preserving, some of our pre-theoretic thoughts on the matter. Whereas Perry’s view is theoretically conservative, it fl ies in the face of the irresistible urge to describe the messy shopper as coming to learn (know, believe) something he didn’t know (believe) before. Whereas Lewis’s view respects our judgment about this, it does so at the cost of reconstruing all cognitive attitudes previously taken to be relations between agents and propositions—the representational nature of which is readily explainable—as relations between agents and properties (some ger- rymandered). Unfortunately, we are given no explanation of how a property like being one who is making a mess can truly or falsely represent anything as being one way or another. Rather than dwelling on the challenges facing these views, I will try to enlarge the space of alternatives. Since we know already from the discussion of (3) and (4) that assumption (A) is false (because representationally identical propositions can diff er in the cognitive requirements for entertaining them), the tacit premise from which Perry and Lewis derive their common conclusion—that in de se cases no new proposition is asserted, believed or known—is no longer available. Without it, their conclusion no longer follows. If propositions are cognitive event types, the possibility remains that at the moment of epiphany the agent does come to assert, believe, and know a special de

King020513OUK.indd 108 11/23/2013 12:58:59 PM COGNITIVE PROPOSITIONS 109

se proposition not previously asserted, believed, or known—despite having previously borne those attitudes to a non- de-se proposition with the same representational content. How might this happen? Recall Frege’s famous observation in “Th e Th ought.”17

“Now everyone is presented to himself in a special and primitive way, in which he is presented to no one else. So when Dr. Lauben has the thought that he was wounded [which he expresses using “I was wounded”], he will probably be basing it on this primitive way in which he is presented to himself. And only Dr. Lauben himself can grasp thoughts specifi ed in this way.” Th e idea is that each person p has a (fi rst-person) way of thinking of p that no one else can use to think about p. Th is idea is plausible—as is the idea that for each time t there is a special way of thinking about t, at that very time, that is not available at any other time. Th e diffi cult point for the neo-Fregean is to show that these ways of thinking of oneself or the present time are Fregean senses: one that uniquely picks out p and is a constituent of propositions entertained by p when p thinks about p in this fi rst-person way, and one that uniquely picks out t, and is a constituent of propositions about t that are entertainable only at t. John Perry and David Kaplan have explained why ordinary Fregean senses of defi nite descriptions can’t be identifi ed with these special ways of thinking. 18 Although Saul Kripke has recently argued that special “acquaintance-based senses” can be so identifi ed, this appears to be false, as I have argued elsewhere. 19 Th is suggests that there simply are no propositions, as traditionally conceived, that can be objects of newly acquired beliefs in de se cases. But propositions are not what they have traditionally been conceived to be; they are cognitive event types. Consider the de re proposition entertainable by anyone who predicates being one who is making a mess of John Perry. Its constituents are the man Perry and the property being one who is making a mess. Th ere are no special constraints on how one must think of Perry in order to entertain this proposition, beyond the ability to refer to him directly. In this respect, the de re proposition about Perry is analogous to the proposition (3b) that Russell defended logicism. By contrast, if there is a de

17 Frege , “Der Gedanke,” Beitrage zur Philosophie des deutschen Idealismus, 1 , 1918 , 58–77 ; trans. “Th e Th ought” by A. and M. Quinton, in Mind 65, 1956, 289–311; trans. “Th ought,” in Beaney, Th e Frege Reader , Oxford: Blackwell, 1997, 325–345, quoted at p. 333. 18 From Perry we get the example of Rip Van Winkle, who awakens on October 20, 1823 aft er sleeping 20 years, and thinks (wrongly) “Today is October 20, 1803.” Here, the belief is about the day on which it occurs, no matter what day, if any, satisfi es the qualitative temporal description Mr. Van Winkle has in mind. See Perry , “Frege on Demonstratives,” Philosophical Review , 86 , 1977 , 474–497 , at p. 487 . From Kaplan we get the example of Castor and Pollux, raised in qualitatively identical environments to be molecule for molecule identical, and to associate the same purely qualitative descriptions with their corresponding uses of the same terms. Despite this, each refers to himself and not to the other when he uses “I.” See Kaplan , “Demonstratives,” in Joseph Almog , John Perry , and Howard Wettstein , eds., Th emes from Kaplan , New York : Oxford University Press , 1989 , 481–563 ;, at p. 531 . 19 K r i p k e , “ F r e g e ’ s Th eory of Sense and Reference,” Th eoria 74 , 2008 , 181–218 ; Soames, chapter 2 section 7 of Th e Analytic Tradition in Philosophy , Vol. 1.

King020513OUK.indd 109 11/23/2013 12:58:59 PM 110 SCOTT SOAMES

se counterpart of the de re proposition about Perry, it is analogous to the proposition (3a) that Russell defended the proposition that arithmetic is reducible to logic. Like proposition (3a), the putative de se proposition must place a further requirement on how the agent cognizes the predication target. Th e requirement in the de se case is that the agent predicate being one who is making a mess of Perry, thinking about Perry in the fi rst-person way. Th us, if there is a de se proposition that Perry came to believe at his moment of epiphany, it is one that only he can entertain. As in the case of (3a) and (3b), the truth conditions and constituents of the de se and the de re propositions are the same. Moreover, although entertaining, asserting, believing, or knowing the former always counts as bearing the same attitude to the latter, it is possible to entertain, assert, believe, or know the de re proposition without bearing that attitude to de se proposition.20 A similar story can be told about the relationship between the de re proposition that the meeting starts then —directly referring to a precise time t (3 PM on a certain day)—and the de se proposition that the meeting starts now (entertained at t). Here, the cognitive event type that is the de se proposition is identical with the event type that is the de re proposition, except for adding the constraint that t be thought of in the special present-tense way in which one thinks of the time at which one’s thought process is occurring. In this way, the conception of propositions as cognitive event types makes room for propositions that are entertainable only by subjects of those propositions, as well as those that are entertainable only at times that fi gure crucially in the propositions themselves. Th is, I contend, opens up a potentially viable alternative to both Lewis- and Perry-style accounts of the de se . In order to pursue it, one must, of course, answer a great many questions. Let us focus on one. How might this new alternative explain the ability of one agent to report, or have thoughts about, the de se attitudes of others? Th e fi rst step in answering this question is to realize that although entertaining propositions expressed by attitude reports sometimes requires one to entertain a proposition expressed by its complement clause, in other cases it does not. For example, consider a use of a sentence of the form (5a)—understood along the lines of (5b)—to report the de re beliefs of every member of a certain class.21 5a. Every F believes that he or she is G. b. Every x: Fx (x believes that x is G)

20 For example, an agent who self-ascribes F-hood and thereby believes the de se proposition expressed by a use of “I am F,” also counts as believing the de re proposition expressed by “x is F” relative to an assignment of the agent to the variable “x.” Th is is evidenced by the fact that whenever an agent A believes that A is F ( de se ) there is some individual x such that A believes that x is F ( de re ). Although it may oft en be the case that when A has the de se belief, A also believes ( de re ) the de re proposition he would use ⌜t is F⌝ to express, for some name, indexical, or demonstrative used to refer to A, this need not always be so. Whether or not it is so in a particular case depends on the semantics and pragmatics of ⌜t is F⌝ , as used by A. 21 For simplicity in specifying the proposition expressed I will adopt Russell’s technique of letting propositional functions stand in for complex properties, and so will take the quantifi ed proposition to predicate a property of such a function, rather than taking it to predicate a higher-order property of a lower-order one. Nothing hinges on this. “F” and “G” are used as schematic letters.

King020513OUK.indd 110 11/23/2013 12:58:59 PM COGNITIVE PROPOSITIONS 111

Th e proposition (5 de re) predicates the property assigning a true proposition to every

F of the propositional function B de re that assigns to each individual o the proposition

(5Matrix/o de re ). Th at proposition, in turn, predicates the believing relation of the pair of

o and the proposition (5Complement/o de re)—which itself predicates being G of o (without placing further restrictions on how o must be cognized in order to be entertained). Obviously, entertaining proposition (5 de re) doesn’t require one to entertain all propositions in which being G is predicated of an individual, or even all propositions in which believing that one is G is predicated of an individual of which F is true. It is enough that

the reporter be able to think about, and refer to, the propositional function B de re (or the complex property it represents). Although I haven’t discussed what this amounts to, no special problems are presented by examples like (5). Th e same is true when (5) is understood as reporting de se attitudes. Proposition (5 de se ) predicates the property assigning a true proposition to every F of the propositional

function B de se that assigns to each individual o the proposition (5Matrix/o de se ). Th is proposition, in turn, predicates the believing relation of the pair consisting of o and

the proposition (5 Complement/o de se )—which itself predicates being G of o, while requiring any agent who entertains it to think of o in the fi rst-person way . Th us, in making simple

propositions like (5Complement/o de se) available, the conception of propositions as cogni-

tive event types also makes propositional functions like B de se and complex propositions like (5 de se) available. Since agents may use (5 de se) to report or think about the de se attitudes of others— thereby entertaining, asserting or believing the proposition they use the belief ascription to express, without entertaining the de se propositions it represents others as believing—there is no general problem reporting or thinking about the de se attitudes of other agents. Next consider (6a), in which “he” is anaphoric on its antecedent, the name “John Perry.” 6a. John Perry came to believe that he was the messy shopper. Th e semantic eff ect of the anaphoric relation between antecedent and pronoun is to introduce a variable-binding operation making (6a) equivalent to (6b).22 6b. λx (x came to believe that x was the messy shopper) John Perry Th us, to use (6a) on its de se understanding is to use it to assert the proposition (6 de se )

in which being an individual to which the propositional function Bλ de se assigns a truth is

predicated of John Perry—where Bλ de se assigns an individual o the proposition (λ Matrix/o de se) that predicates the believing relation of the pair consisting of o and the proposi-

tion (λ Complement/o de se ). Finally, this proposition predicates being the messy shopper of

22 Th is account of anaphora is defended on independent grounds in Soames , “Attitudes and Anaphora,” Philosophical Perspectives , 8 , 1994 , 251–272 ; reprinted in Philosophical Essays: Vol. 2: Th e Philosophical Signifi cance of Language .

King020513OUK.indd 111 11/23/2013 12:58:59 PM 112 SCOTT SOAMES

o, while requiring any agent who entertains it to think of o in the fi rst-person way . 23 Th e end result is that the proposition expressed by (6a) predicates the property believing of oneself (in the special fi rst-person way) that one is the messy shopper of John Perry. As before, the de se proposition reported to be believed doesn’t have to be entertained by the reporter. 24 Perry can, of course, report his own de se attitude using (6c), taking the complement clause to express the de se proposition he entertains. 6c. I believe that I am the messy shopper. If he does this, his utterance will assert both that he believes the de se proposition only he can entertain, and that he believes the de re proposition, entertainable by all, that predicates being the messy shopper of him. His hearers will understand the de re assertion in the normal way, while also realizing that he endorses the being-a-messy-shopper predication thinking of its target, Perry himself, in the special fi rst-person way. Th is, it may be argued, is our pretheoretic way of describing the de se proposition he believes, without having to entertain it ourselves. A further twist is provided by a version of the case in which Perry’s epiphany comes in two stages. In this version he is accompanied on his trip around the supermarket by his daughter, who is also intent on fi nding the messy shopper. At the crucial moment, her face lights up with the realization that her father is the culprit. Noticing the shock of recognition on her face, Perry mutters (7a) under his breath, wondering if she is right. 7a. She thinks that I’m the messy shopper. Th is is Perry’s fi rst epiphany, in which he seems to believe a new proposition (about his daughter’s beliefs) that he had not previously believed. Since it is a new belief, all the old arguments can be recycled to show that the proposition believed can’t be an ordinary (non-de se ) proposition. What proposition is it? Although “I” occurs in the complement clause of (7a), Perry’s use of (7a) surely doesn’t attribute to his daughter

23 If we conduct the analysis in terms of complex properties rather than propositional functions, the de se proposition (6b) predicates, of Perry, the property being one who believes the proposition that (i) predicates being the messy shopper of one , while (ii) requiring any agent who entertains it to think of oneself in the fi rst-person way . 24 One way of developing these ideas about (5a,b) and (6a,b) in a semantic theory would be to allow two kinds of occurrences of variables—normal ones, and ones that result from adding “*” to occurrences in complement clauses that represent pronouns anaphoric on matrix subjects of attitude ascriptions. Although adding “*” would not aff ect the referent of the variable relative to an assignment, a complement clause . . .x*. . . containing a *ed occurrence of “x” would express a singular proposition, relative to an assignment of o to “x”, that can be entertained only by the referent of “x” relative to the assignment. Th e diff erence between de se and de re readings of attitude ascriptions in cases like (5a,b) and (6a,b) could then be captured by a diff erence between *ed and non-*ed occurrences of variables representing pronouns in the complement clause. In eff ect, pronouns anaphoric on matrix-subject antecedents will be ambiguous between de se and de re inducing occurrences of variables. By contrast, the phonologically empty syntactic constituent PRO -- as in “He expects PRO to win” (meaning “He expects to win”) could be treated as unambiguous, and thus as always corresponding to a *ed occurrence of a variable bound by a higher subject.

King020513OUK.indd 112 11/23/2013 12:58:59 PM COGNITIVE PROPOSITIONS 113

a belief in a de se proposition that she couldn’t possibly entertain. Nor, if Perry’s use of (7a) is to express a new belief, can one maintain that his use of “I” is purely de re . Th ere is, however, another option. Perry’s new belief can be identifi ed with the de se proposition expressed (from Perry’s point of view) by (7b). 7b. λx (she believes x is the messy shopper) me Understood de se (from Perry’s perspective), (7b) expresses the proposition that is the cognitive event type of predicating being someone believed by her [his daughter] to be the messy shopper , of Perry, thinking of the predication target, Perry, in the special fi rst- person way. Here, I have used lambda abstraction to give the fi rst-person pronoun what is, in eff ect, wide-scope over “believe.” Th e mechanism used to achieve this is not crucial—there are multiple ways of encoding scope distinctions for terms of various types (not just quantifi ers). What is needed is simply that one’s semantic or pragmatic theory provides a principled way of applying one of them here. Th is completes my preliminary sketch of the possibilities for the analysis of de se attitudes that are opened up by the conception of propositions as cognitive event types. By providing a natural explanation of how a proposition p can constrain the way one of its constituents (e.g. a predication target) must be cognized by an agent who entertains p, the conception makes available de se (fi rst-person, present-tense) propositions distinct from those that have previously been recognized. Since these special ways of cognizing a predication target do not , for the reasons indicated above, involve any new predications of that target, the new propositions are representationally identical to ordinary de-re propositions. Of course, my sketch falls far short of a complete analysis. However, if it seems promising, the general lesson to take from it is worth emphasizing; coming to understand what propositions are can be important not only in providing philosophical foundations for linguistic and cognitive theories, but also in enhancing their explanatory potential by providing new tools for empirical analyses.

Articulated terms, the worldly de se , and the failure of a priority to close under a priori consequence Th e next step is to combine the lesson learned from (3) and (4)—about the diff erence between the propositions expressed by sentences containing proper names vs. those containing articulated terms—with the new way of thinking about the de se opened up by propositions as cognitive event types. Suppose, as I do, that a world-state is a property, attributable to the universe, of making true a set of basic propositions that tells a complete world story—e.g., the property making it true that the earth is the third planet from the sun in the solar system, that the earth is round, that the earth is largely covered with water, that the earth is inhabited by humans and other animals, etc., etc., etc. Imagine defi ning a world-state w by fi lling this out so as to provide such a complete story—where by “complete” I mean one that answers all questions relevant to a

King020513OUK.indd 113 11/23/2013 12:58:59 PM 114 SCOTT SOAMES

contextually determined inquiry. Let “n” be a proper name of w, and let “PW” abbre- viate an articulated term “the property making it true . . .” that articulates each of the basic propositions used in defi ning w. Finally, let “p” designate any proposition true at w—where for a proposition to be true at w is for the proposition to be an a priori consequence of the set of basic propositions used to defi ne w.25 Th en consider (8). 8a. p is true at PW. b. p is true at n. c. p is true at this very world-state—said at w referring to w Th e relationship between the propositions expressed by (8a) and (8b) is like the relationship between propositions (3a) and (3b). In both cases, the (a) and (b) propositions predicate the same properties/relations of the same things; in both cases the (a) proposition diff ers from the (b) proposition only in adding the constraint that in order to entertain it the agent must also entertain its propositional constituent (in the case of 3a), or its propositional sub constituents (in the case of 8a). Moreover, in both cases, anyone who knows or believes the (a) proposition knows or believes the (b) proposition, but not conversely. Now consider the relationship between the propositions expressed by (8b) and (8c). I have already made room for a special fi rst-person way of thinking about, and referring to, an individual x, plus singular propositions about x that can be entertained only by x when thinking about himself or herself in the fi rst-person way. I have similarly made room for a special present-tense way of referring to a time t—as now, or this very time—plus singular propositions about t that can be entertained only by one who thinks of t in that way at t. Having done so, I introduce a parallel hypothesis about world-states which, although I neither accept nor reject it, is, I think, worth further examination. Th e hypothesis is that there is a special world-bound way of referring to a world-state w—as this very world-state —plus singular propositions about w that can be entertained only by those at w who think about it in this special world-bound way. On this hypothesis, the relationship between proposition (8b) and proposition (8c) parallels the relationship between (i) the ordinary singular proposition that Scott Soames wrote What is Meaning? and the de se proposition that I wrote What is Meaning? (entertained by me in the fi rst-person way), and (ii) the ordinary singular proposition about the present time t that Martha is working at t and the temporally de se proposition I express to myself by saying Martha is working now (at this very time). Remember, these special ways of thinking of things are not special descriptive ways of thinking about them. Castor and Pollux can be in qualitatively identical cognitive states when thinking about themselves in the fi rst-person way, even though each refers to himself and not the other. Rip Van Winkle can be in qualitatively identical cognitive

25 See Soames, “Actually,” summarized in chapter 6 of Philosophy of Language . Th ere is, of course, no requirement that the propositions in terms of which a world-state is defi ned exist at w.

King020513OUK.indd 114 11/23/2013 12:59:00 PM COGNITIVE PROPOSITIONS 115

states at diff erent times when thinking of, and referring to, those diff erent times in the same special present-tense way. By the same token, the world-state to which I actually refer in the hypothesized special world-bound way—by saying to myself “this very world-state”—is diff erent from the world-state to which I refer at a world-state that is merely possible, even though my cognitive state at that world-state is identical with the one I am actually in. On the hypothesis under investigation, all of these cases involve special ways of thinking about and referring to things the identities of which are not determined by any description imposed by the agent. On the contrary, the items picked out are determined by who the agent is , when the thoughts are occurring , and the world state at which the agent entertains the proposition . Hence, the hypothesis maintains, including these special ways of thinking about or referring to the predication targets of propositions do not introduce any new descriptive content, or any new truth-condition-determining predications, into the ordinary singular propositions to which they are added. Given all this, we can distinguish propositions (8a), (8b), and (8c). Although all are singular propositions in which being something at which p is true is predicated of the world-state w, propositions (8a) and (8c) impose constraints on how agents who entertain them must think about w, while proposition (8b) imposes no such constraint. Proposition (8a) requires the agent to think of w by entertaining the propositions that are themselves constituents of the property that w is. Proposition (8c) requires thinking about w in the special actual-world-state way that parallels the present-tense way of thinking about a time and the fi rst-person way of thinking about an agent. We have seen that knowing or believing proposition (8a) guarantees knowing or believing proposition (8b), but not vice versa. Th e same relation holds between proposition (8c) and proposition (8b). As in the ordinary de se case, where knowing or believing de se guarantees knowing or believing de re , but not the other way around, so knowing or believing the worldly-de-se proposition (8c) guarantees knowing or believing (8b), but not conversely. We can now connect this with an interesting result about the a priori. On the account of world-states indicated above, the propositions true at a world-state are those that are a priori consequences of the basic propositions that defi ne it. With this in mind, let w be the actual world-state, and let the proposition that Plato was a philosopher be an a priori consequence of the basic propositions defi ning w. We then get the following results: (i) Th e proposition that it is true at PW that Plato was a philosopher (corresponding to (8a)) is knowable a priori but the proposition that it is true at this very world-state that Plato was a philosopher (corresponding to (8c)) is not knowable a priori; (ii) Th e proposition that Plato was a philosopher iff it is true at this very world-state that Plato was a philosopher is knowable a priori but the proposition that Plato was a philosopher iff it is true at PW that Plato was a philosopher is not knowable a priori;

King020513OUK.indd 115 11/23/2013 12:59:00 PM 116 SCOTT SOAMES

(iii) It follows from (i) that the proposition that that Plato was a philosopher is true at w is knowable a priori. It follows from (ii) that the proposition that that Plato was a philosopher iff it is true at w that Plato was a philosopher is knowable a priori.26 But neither their conjunction nor the proposition that Plato was a philosopher is so knowable. Since we have two propositions that are knowable a priori, even though their conjunction isn’t, the set of a priori truths is not closed under a priori consequence. Although this result isn’t new, the simple explanation of it provided by the hypothesis that this-very-world-state cognition parallels special fi rst-person and present-tense cognition is, as is the recognition that this result is of a piece with those discussed in the previous two sections.27 Th e fact that the conception of propositions as cognitive event types allows us to tie these phenomena together as three aspects of the same thing provides some reason for taking the hypothesis seriously.

Semantic relationism Another case in which there may turn out to be a productive interplay between a satisfying metaphysical conception of what propositions are and an empirically-informed account of the kinds of propositions needed in semantics and cognitive science comes from Kit Fine’s fascinating work in Semantic Relationism.28 Although he doesn’t give a metaphysics of propositions there, he does argue that we must recognize a class of coordinated propositions that place special constraints on how an agent is required to think of their predication targets in order to entertain them. It is central to his conception that these propositions share the structure, constituents, and truth conditions of their uncoordinated counterparts, which lack special constraints on how their predication targets are cognized. Th ough his theory doesn’t apply to examples like (3a) and (3b), what he says about the diff erence between coordinated and uncoordinated propositions is importantly similar to what I have said about (3a) and (3b). Th is raises the question of whether the conception of propositions as cognitive event types is capable of accommodating Fine’s coordinated propositions. I noted earlier that the proposition that Tully shaved Cicero is the event type of predicating the shaving relation of Cicero and Cicero, and that the proposition that

26 From (ii) we get the a priority of the proposition that λy [Plato was a philosopher iff it is true at y that Plato was a philosopher] this very world-state, which guarantees the a priority of the proposition that λy [Plato was a philosopher iff it is true at y that Plato was a philosopher] w, which guarantess the a priority of the proposition that Plato was a philosopher iff it is true at w that Plato was a philosopher. 27 A version of this result is explained and established in chapter 6 of Soames, Philosophy of Language. Th e failure of closure of a priority under a priori consequence does not upset the claim that for a proposition to be true at w is for it to be an a priori consequence of the basic propositions defi ning w. Failures of closure always involve propositions about world-states, which aren’t among the propositions that defi ne world-states. 28 Kit Fine, Semantic Relationism , Malden, MA: Wiley-Blackwell, 2007.

King020513OUK.indd 116 11/23/2013 12:59:00 PM COGNITIVE PROPOSITIONS 117

he shaves himself is the event type of refl exivizing the shaving relation to get the property being one who shaves oneself and predicating it of Cicero. According to Fine, the (coordinated) proposition that Cicero shaved Cicero is diff erent from both of these. Expressed in my terms, it is the event type of predicating the shaving relation of Cicero and Cicero, thinking of the two as the same. What is it to do this? It is not to predicate the shaving relation while assuming that the individual one’s predication represents as shaver is the same as the individual one’s predication represents as being shaved. To predicate a relation of a pair, one must think of the relation and the pair; one doesn’t also have to make a higher-order judgment about what one’s predication represents. Surely there are agents who predicate properties of things, and thereby have propositional attitudes, without bearing attitudes to propositions about their own cognitive activities. Given the importance for thought and action that Fine takes coordination to have, he would, I am confi dent, not wish to exclude such agents from the benefi ts of bearing attitudes to coordinated propositions. Nor can the coordinated proposition be the event type in which one predicates shaving of Cicero and Cicero, while judging Cicero to be identical with Cicero . Th e content of that judgment can’t be the uncoordinated proposition that Cicero is Cicero (which is just the proposition that Cicero is Tully). Nor can it be the coordinated proposition, since that would involve using coordination to explain coordination. So Fine must take thinking of the members of a pair as the same to be primitive, understanding that to predicate shaving of Cicero and Cicero while bearing this attitude to them is diff erent from predicating self-shaving of Cicero, and also that predicating being F of o, and being G of o while taking them to be the same is diff erent from predicating being F and G of o. 29 Suppose, for the sake of argument, that there is such a primitive attitude of taking things to be the same. Since it is a kind of cognition, it may seem that there must be propositions the entertaining of which requires one to cognize things in this way. Are there really such propositions? Th is question can be taken in two ways: (i) Is there a need within semantics and cognitive science to recognize such propositions? (ii) If there is a need, does the conception of propositions as cognitive event types make room for them? Elsewhere I have argued that the case for coordination in natural language semantics (and cognition) remains inconclusive (at best). 30 Th us question (i) remains open. Here, I will confi ne myself to saying a word about (ii).

29 Th is is Fine’s view. On page 59 he says: “But the coordinative aspect of the coordinated content of a sentence, such as ‘Cicero wrote about Cicero’ is entirely lacking in any special descriptive or truth conditional character and relates entirely to how its truth conditions. . .are to be grasped [entertained]. It is a signifi cant feature of the traditional Fregean view that there can be no diff erence in what it is to grasp [entertain] the sense of an expression without there being a diff erence in how the sense has application to [or represents] the world.. . . But under the relational view, these two aspects of sense come completely apart. Th ere is no difference in what it takes for the sentences “Cicero wrote about Cicero” and “Cicero wrote about Tully” to be true, even though there is a diff erence in their coordinated content .” 30 Soames , “Two Versions of Millianism,” in Michael , O’Rourke , ed., Topics in Philosophy, Vol. 10: Reference and Referring , Cambridge : MIT Press , 2012, 83–118.

King020513OUK.indd 117 11/23/2013 12:59:00 PM 118 SCOTT SOAMES

Surely, there are limits on what cognitive acts or operations propositions can encode. Th inking of a certain tune while predicating redness of an object is a (complex) cognitive activity of some sort, as is predicating a relation of a pair, while feeling aff ection toward its members. Presumably, the event types of doing these things are not propositions because one of their cognitive components is orthogonal to how the agent represents things to be. By contrast, propositions (3a) and (3b) diff er— despite the fact that their truth conditions arise from predicating the same relation of the same things—because the condition placed on a predication target in (3a) is that it be entertained, which is itself a cognitive activity in which certain things are represented to be a certain way. Since the cognitive activities that make up (3a) are all representational, it is a genuine proposition that diff ers from (3b). Th is suggests that whether there are (or could be) Finean coordinated propositions depends on whether the putative attitude of taking objects to be the same is appropriately representational. How should we think about it? It may help to ask about cases in which one takes non-identical objects to be the same . Certainly non-identical objects can appear the same way identical objects do, and so, one would think, provoke the same cognitive responses . Are we then to suppose that there are propositions we can entertain only by predicating something of a pair of non-identical things, mistakenly taking them to be the same? Suppose an agent mistakenly takes Cicero and his brother (each of whom shaves the other but neither of whom shaves himself) to be the same , while predicating the shaving relation of them. If this sequence of cognitive acts is encoded by a genuine “coordinated” proposition, what is its truth value? To say it is true ignores the fact that entertaining it requires one to be disposed to mistakenly judge non-identical things to be identical, and so to think of them as related in a way they are not—which seems very much like representing them falsely or incorrectly. To say that the putative proposition is not true is a non-starter if, as I have indicated, the truth conditions of a proposition are derived from what it predicates of what (which in Fine’s system is refl ected by the requirement that coordinated propositions share the truth conditions of their uncoordinated counterparts). For this reason, it is unclear that there are (or could be) propositions in which non-identical objects are coordinated—which, in turn, casts some doubt on the existence of propositions in which identical objects are coordinated.I am not sure whether this doubt can be overcome. If it can, then the conception of propositions as cognitive event types doesn’t, as far as I can see, create any further diffi culties for Fine’s view. Th us, I tentatively conclude that the metaphysics of propositions off ered here provides important philosophical grounding for Fine’s general conception of representationally identical but cognitively distinct propositions without raising any new problems for his particular conception of coordinated propositions that it doesn’t already have.

King020513OUK.indd 118 11/23/2013 12:59:00 PM COGNITIVE PROPOSITIONS 119

Structure, cognition, and predication: A fi nal amendment Th e conception of propositions as cognitive event types diff ers from other accounts of structured propositions in two ways. First, the structure of a proposition is provided by the truth-condition-determining cognitive operations performed on its constituents by an agent who entertains it. Second, in addition to constituents and structure in this sense, some propositions impose further constraints on how its constituents are cognized by the agent. Th e structural operations considered in this chapter have been limited to functional application (which provides structure to some propositional constituents), operations on propositions (which give us truth-functional compounds) and predication (which provides propositional structure). I close by deriving a further important lesson about predication from a notorious historical example. Th e example comes from what Russell took to be the central argument of “On Denoting,” which led him in the spring of 1905 to his celebrated theory of descriptions.31 Th e conclusion of the argument is that complex singular terms are impossible, so no language, natural or artifi cial, can contain defi nite descriptions as singular terms. One of the key examples he used in coming to this conclusion was (9), in which “M” is used as a Millian name for the meaning of the description of “the fi rst line of Gray’s Elegy” (taken to be a singular term). 9a. Th e fi rst line of Gray’s Elegy is “Th e curfew tolls the knell of parting day.” b . “ Th e fi rst line of Gray’s Elegy” means M. c . “ Th e fi rst line of Gray’s Elegy” means the fi rst line of Gray’s Elegy. Since, by hypothesis, “M” and “the fi rst line of Gray’s Elegy” mean the same thing, Russell reasons that propositions (9b) and (9c) must have the same structure and constituents, and so be identical. But they can’t be; for if the description really does mean M, (9b) must be true, while (9c) must be false—since it says that the meaning of the description is the thing it designates (the sentence “Th e curfew tolls the knell of parting day”). 32 At this point, Russell appeals to R. R I f d e fi nite descriptions are meaningful singular terms, (i) they must express meanings that denote unique objects satisfying them (if such objects there be),

31 R u s s e l l , “ O n D e n o t i n g ,” Mind 14 , 1905 , 479–493 . Th e argument comes from “On Fundamentals,” written in June 1905, fi rst published in Russell, Foundations of Logic , London: Routledge, 1994. Exposition and criticism of the argument, along with an account of its role in leading Russell to his theory of descriptions is found in chapters 7 and 8 of Th e Analytic Tradition in Philosophy, Vol. 1. See also Richard Cartwright, “On the Origins of Russell’s Th eory of Descriptions,” in Philosophical Essays, Cambridge: MIT Press, 1987, 95–133, and Nathan Salmon , “On Designating,” Mind , 114 , 2005 , 1069–1133 . 32 I here assume that (9b) unambiguously expresses the true proposition indicated above, and that (9c) unambiguously expresses the false proposition there indicated. I will return to question of ambiguity below.

King020513OUK.indd 119 11/23/2013 12:59:00 PM 120 SCOTT SOAMES

and (ii) these meanings can occur in propositions only in the role of present- ing their denotations as the ultimate subjects of predication in the propositions. Th erefore, these meanings can never be subjects of predication in any proposition in which they occur. Accepting R leads Russell to deny that (9b) is true and to conclude that there is no true singular proposition in which being what “the fi rst line of Gray’s Elegy” means is predicated of anything. From here it is a short step to the conclusion that no one can know of anything that it is what the description means. But surely, Russell thinks, in order for an expression E to mean something it must be possible to know of what E means that E means it. Th us, he concludes, it is impossible for meaningful defi nite descriptions (of any language) to be singular terms. (Th e argument generalizes to all complex singular terms formed by combining a function symbol with one or more terms serving as arguments—e.g. “3 2 .”) Since Russell’s conclusion is clearly false, and R is the likely culprit, we need a conception of propositions that provides an explanation of why it is false. Although purely technical moves could be made, no conception of propositions I know of provides a plausible explanation of why this should be so. I believe that the conception of propositions as cognitive event types can do better. Let us start by taking proposition (9a) to

be the event type of (i) thinking of the function fthe and the function g (that assigns an

object a truth iff it is a line in Gray’s Elegy preceding all others), (ii) combining fthe and

g into a single constituent in which g is understood to play the role argument-of-fthe , and (iii) predicating the identity relation of the pair consisting of “the curfew tolls the

knell of parting day” plus the result of applying f the to g. Let it be part of the theory that

the constituent f the -plus-g is the meaning of the defi nite description. Let it also be part of the theory that this constituent occurs in the propositions expressed by each of the sentences (9a,b,c). Since “the curfew tolls the knell of parting day” is uniquely deter-

mined by f the -plus-g, proposition (9a) comes out true. Similar reasoning gives us the falsity of (9c). What about proposition (9b)? It seems that it should be the event type of

(i) predicating means of the pair “the fi rst line of Gray’s Elegy” and f the -plus-g. However, since these are also the (major) constituents of proposition (9c), the assumption that in both cases the same relation is predicated of the same arguments requires the two propositions to have the same truth value. 33 Th is is what leads to Russell’s false conclusion. So far, the problem is still with us. Th e conception of propositions as cognitive event types does, correctly, block the conclusion that propositions (9b) and (9c) are identical. Th e case is analogous to (3a)

33 Th e major constituents of a proposition p are those in terms of which its truth conditions are directly defi ned. Sometimes a major constituent of p itself has constituents that are sub constituents of p in a weaker sense; they are elements that have to be cognitively accessed in the process of identifying, and putting to use, the major constituent of which they are constituents. Th e constituents of the argument of defending in proposition (3a) are sub constituents of that proposition, but not of proposition (3b); the functions that are constituents of M are sub constituents of proposition (9c), but not of proposition (9b). See chapters 8 and 9 of Th e Analytic Tradition, Vol. 1 for further discussion.

King020513OUK.indd 120 11/23/2013 12:59:00 PM COGNITIVE PROPOSITIONS 121

and (3b)—with “M” in (8b) playing the role of “logicism” in (3b). Applying that lesson here, we see that one who entertains proposition (9c), but not one who entertains prop-

osition (9b), must think of f the, and of g, and combine them into a function-argument structure to be used to determine the predication target. However recognizing this diff erence is not enough, since nothing I have said up to now explains how proposition (9b) can be true (which it must be), while proposition (9c) is false. Since the two propositions have the same major constituents, and since a diff erence in the way that one of those constituents is cognized in the two propositions can’t , by itself, aff ect their truth values, they must also diff er in structure . Th e structure of a proposition is the manner in which its constituents are related to one another. Since propositions are cognitive event types, the structural relationships in which their constituents stand to each other are not relationships in which things occupying certain positions in an n-tuple stand to those occupying other positions; nor are they relationships that certain nodes in an abstract tree structure bear to other nodes. Although nothing prevents using formal constructions of these or other sorts to model propositional structures, the structures being modeled are something else. Th e structural relationships between the constituents of a proposition are given by the roles the constituents play in the sequence of cognitive operations performed by an agent who entertains it—roles like being predicated (of certain things), being targets (of certain predications), being applied (to certain arguments), being arguments (to which certain things are applied), as well as being sub constituents of larger constituents which may themselves play these roles, or of propositions that are constituents of larger propositions. Th is is the sense in which propositions (9b) and (9c) must diff er in structure . Th e required diff erence is a diff erence in the sense in which the relational property being what “the fi rst line of Gray’s Elegy” means is predicated of its argument in the two propositions. In (9b) it is directly predicated of the complex that is the meaning of the description; in (9c) it is indirectly predicated of whatever is determined by that complex (which is the referent of the Fregean description). In order for the relation direct predication to hold of an agent A (who entertains a proposition p), a property F (to be predicated of something) and an item x (of which F is predicated), A must have x in mind as the thing to be represented as having F. By contrast, the indirect predication relation holds between A, F, and an item x which is the kind of thing (e.g. a function-argument complex) that determines something else (e.g. a value) . In order for this relation to hold, A must have x in mind, and use it to represent whatever, if anything, is determined by x as having F. Th e direct predication is veridical iff x has F; the indirect predication is veridical iff there is something uniquely determined by x and that thing has F.

Let “Pred D ” and “Pred I ” express the relations direct predication and indirect predication, respectively, and let “T” be a schematic letter to be replaced by a singular term. When I say something of the form (10), the proposition I express is one that directly

predicates Pred D of the triple consisting of A (or p), the property so-and-so, and the referent , if any, of the term replacing “T”.

King020513OUK.indd 121 11/23/2013 12:59:00 PM 122 SCOTT SOAMES

10. Agent A (or proposition p) directly predicates property so-and-so of T. When I say something of the form (11), the proposition I express is one that directly

predicates PredI of the triple consisting of A (or p), the property so-and-so, and the content (meaning) of the term replacing “T”. 11. Agent A (or proposition p) indirectly predicates the property so-and-so of T. With these understandings, (12a) is true, while (12b) is false, because what entertaining (9b) requires is having M in mind and aiming to represent it—not whatever, if anything, it determines (namely, “Th e curfew tolls the knell of parting day”)—as being what the description means . 12a. Proposition (9b) directly predicates being what “the fi rst line of Gray’s Elegy” means of M. b. Proposition (9b) indirectly predicates being what “the fi rst line of Gray’s Elegy” means of M. By contrast, (13a) is false because entertaining (9a) doesn’t require having “Th e curfew tolls the knell of parting day” in mind and intending to represent it as having the property being identical with “Th e curfew tolls the knell of parting day” ;34 (13b) is true, because what entertaining (9a) does require is having M in mind and intending to

represent whatever, if anything, M determines—the value of fthe at g—as having the property in question (which may or may not involve having any idea of what object is determined). 13a. Proposition (9a) directly predicates being identical with “Th e curfew tolls the knell of parting day” of the fi rst line of Gray’s Elegy. b. Proposition (9a) indirectly predicates being identical with “Th e curfew tolls the knell of parting day” of the fi rst line of Gray’s Elegy.

Examples (14a) and (14b) are like (13a) and (13b). 14a. Proposition (9c) directly predicates being what “the fi rst line of Gray’s Elegy” means of the fi rst line of Gray’s Elegy. 14b. Proposition (9c) indirectly predicates being what “the fi rst line of Gray’s Elegy” means of the fi rst line of Gray’s Elegy. (14a) is false because entertaining (9c) doesn’t require having “Th e curfew tolls the knell of parting day” in mind and aiming to represent it as having one property or another; (14b), on the other hand, is true (even though proposition (9c) is, of course, false). Finally, notice that (15a) expresses a falsehood, while (15b) expresses a truth about the proposition expressed by (9d).

34 Remember, to predicate a property of something one doesn’t have to think that it really has the property.

King020513OUK.indd 122 11/23/2013 12:59:00 PM COGNITIVE PROPOSITIONS 123

9 d . Th e denoting complex identifi ed by Soames in the fi nal section of chapter 6 of New Th inking about Propositions is what “the fi rst line of Gray’s Elegy” means. 15a. Proposition (9d) directly predicates being what “the fi rst line of Gray’s Elegy” means of the denoting complex identifi ed by Soames in the fi nal section of chapter 6 of New Th inking about Propositions. 15b. Proposition (9d) indirectly predicates being what “the fi rst line of Gray’s Elegy” means of the denoting complex identifi ed by Soames in the fi nal section of chapter 6 of New Th inking about Propositions. Th e distinction between direct and indirect predication replaces my earlier univocal characterization of predication as being a 3-place relation between an agent, a property, and (as I put it) “a content.” Th ough indirect predication has some of the features I mentioned in connection with that characterization, neither direct nor indirect predication has all of them. Th us, my earlier use of “predicate” must be recast in terms of these two new notions. To that end, one should understand claims made in previous sections by sentences of the form (16a) in which “T” is replaced by a Fregean defi nite description, or other function-argument singular term, as making the claim expressed by (16b). 16a. Agent A (or proposition p) predicates the property so-and-so of T b. Agent A (or proposition p) indirectly predicates property so-and-so of T In all other cases, claims made by sentences of the form (16a) that I have used earlier should be understood as claims involving direct predication. Th is reconstrual requires making explicit something I have so far taken for granted without comment. When an n-place predicate is paired with n arguments—some of which may be Millian and some non-Millian—we must think of the predication as proceeding in stages. Th is technique, familiar from Montague, treats the proposition expressed by a sentence of the form

(17) A loves B as arising fi rst by combining the two-place relation loves with the content/referent of the term replacing “B,” and then predicating the resulting one-place property of the content/referent of the term “A.” 35 When “B” is replaced by a Millian singular term the content and referent of which is x, the resulting one-place property is loving x , which may then be predicated directly, or indirectly, of the referent or content of the term that replaces “A,” depending on whether that term is Millian or non-Millian. When “B” is replaced by a non-Millian singular term—e.g. something the content of which is a

complex consisting f the combined with an argument g—the resulting one-place prop-

erty is loving whomever is the value of f the at g —which may, of course, also be predicated

35 R i c h a r d M o n t a g u e , “ Th e Proper Treatment of Quantifi cation in Ordinary English,” in J. Hintikka and J. Moravcsik , eds., Approaches to Natural Language , Dordrecht : Reidel , 1973 ; reprinted in Richmond Th omason, ed., Formal Philosophy: Selected Papers of Richard Montague , New Haven : Yale University Press .

King020513OUK.indd 123 11/23/2013 12:59:00 PM 124 SCOTT SOAMES

directly, or indirectly, of the referent or content of the term that replaces “A.” Th us the operation, call it “reduction,” that maps an n-place relation plus an argument to the relevant n-1 place relation subdivides into direct and indirect reduction, on analogy with direct and indirect predication. Although this sketch of propositions as cognitive event types is far from complete, I hope to have given some idea of its promise and fl exibility. What makes the conception natural for resolving Russell’s problem involving (9b,c) is that the distinction required to solve the problem—between what an agent who entertains proposition (9b) aims to represent as having a certain property vs. what an agent who entertains proposition (9c) aims to represent as having that property—is a cognitive diff erence encoded in the cognitive acts that provide the structure of the event types with which propositions are identifi ed. By contrast, what made the problem seem insoluble to Russell in 1905 was his conception of propositions as platonic objects the intentional properties of which are prior to, and independent of, the agents who entertain them. Th inking of propositions in this way, and asking himself the question “What does M do in the proposition expressed by (9c) that it doesn’t do in the proposition expressed by (9c)?” he naturally answered “Nothing!”—which led him astray. Although the purely platonic conception of propositions has lasted for a very long time, we are now, I hope, beyond that. Armed with a more accurate conception of propositions, we pose the crucial question diff erently—“What do agents use M to do when entertaining the proposition expressed by (9c) that they don’t use it to do when entertaining the proposition expressed by (9b)?” Since the answer to this question is obvious, what had been a problem ceases to be.36, 37

36 I return to the question of whether sentences (9b) and (9c) are ambiguous. Sentence (9b) is not; it can’t express the false proposition that (9c) does because entertaining the latter requires one to cognize the constituents of M, whereas no proposition expressed by sentence (9b) does. Nor does it make sense to suppose that (9b) expresses a proposition the truth conditions of which are the same as those of the false proposition expressed by (9c) even though it doesn’t require cognizing the constituents of M. It is less clear whether (9c) has a reading in which it can express the true proposition that (9b) unambiguously expresses. People do say things like “red” means red , and “on the counter” means on the counter . Whether these are legitimate uses or mistakes is not entirely clear to me. But if they are correct, then there may be a similar use of (9c) in which it expresses a truth—indeed, the proposition that results from adding to the proposition expressed by (9b), the requirement that the constituents of M be cognized. 37 I am indebted to Brian Bowman for helpful comments on this chapter.

King020513OUK.indd 124 11/23/2013 12:59:00 PM P A R T I I I Critical Essays

King020513OUK.indd 125 11/23/2013 12:59:00 PM King020513OUK.indd 126 11/23/2013 12:59:00 PM 7

Criticisms of Soames and Speaks

J e ff rey C. King

Soames’s account of propositions As I hope was made clear in Chapter 4, among the various features of my view of propositions, there is a radical negative feature and a radical positive feature. Th ese features are closely connected, and both were novel to the account I defended in King [2007] at the time it appeared. Th e negative feature is that the traditional view that propositions are things that are representational, and so have truth conditions, by their natures and independently of minds and languages is rejected as ultimately mysterious. As I indicated in Chapter 4, we simply have no idea how anything could have truth conditions by its nature and independently of minds and languages. Hence we need a diff erent sort of account of propositions on which we can explain how/why propositions manage to have truth conditions and so represent things as being a certain way.1 I claim the account given in King [2007] and modifi ed in Chapter 4 is such an account. Th e radical positive feature of that account is that it claims that the representational powers of propositions depend on the cognitive activities of agents. In recently developing his own account of propositions, Scott Soames [2010, Chapter 6 of the present work] embraces the negative and positive features just discussed as well. Like me, Soames now thinks that the traditional view of propositions, as entities that represent by their very natures and independently of minds and languages, is to be rejected as mysterious. He has also now come to believe that the fact that propositions are representational must ultimately be explained in terms of the representational capacities of agents. Of course, I cannot help but be pleased to have Soames side with me on these important matters. And that his view embraces these points makes me sympathetic to it. However, Soames’s positive account of propositions and

1 Th e demand that an explanation be given of how/why propositions have truth conditions actually goes all the way back to King [1995], which defended an earlier version of the view defended in King [2007].

King020513OUK.indd 127 11/23/2013 12:59:00 PM 128 JEFFREY C. KING

his explanation of exactly how they manage to have truth conditions diff ers signifi - cantly from mine. Despite my sympathy to his overall approach, my purpose here is to criticize Soames’s account of propositions and his explanation of how they manage to have truth conditions. Soames begins with the notion of the mental act of predication , which he takes to be primitive. 2 However, by way of illustration, if an agent perceives an object o as red, and so has a perceptual experience that represents o as being red, the agent predicates redness of o.3 Similarly, if an agent “thinks of” o as red 4 , or “form[s] the nonlinguistic perceptual belief that o is red.” 5 For Soames, predicating redness of o does not amount to believing that o is red. To believe that o is red, one must predicate redness of o and do something like endorse the predication.6 Unfortunately, it is hard to say precisely what predicating amounts to since the notion of predicating is primitive for Soames. 7 An agent predicating redness of o is an event token. Of course there may well be many event tokens of agents predicating redness of o by an agent perceiving it as red, an agent thinking of it as red, a diff erent agent thinking of it as red and so on. Soames claims that the proposition that o is red is the event type of an agent predicating redness of o. Other more complex propositions are identifi ed with event types of agents performing sequences of primitive mental acts on the various constituents.8 One entertains a proposition by performing these sequences of acts on the relevant constituents. With only this much of Soames’s view on the table, I can raise the fi rst problem with his account. But before turning to that let me note as an aside that it is odd that Soames appears to assume with little comment or argument that the contents of perceptual experiences are expressible by sentences of natural languages.9 Th is claim is controversial among those who work on perceptual experience. Some deny that perceptual experience has content at all. Some allow that it has content, but hold that the content is not propositional. And some allow that perceptual experience has propositional content, but deny that the content is expressible by means of sentences of natural language. Hence, it is a bit surprising that with so little comment or argument Soames adopts the view that the perceptual experience has contents that are expressible by natural language sentences.

2 Soames [2010] p. 81 3 Chapter 6 of present work p. 96 4 Soames [2010] p. 103. 5 Soames [2010] p. 81 6 Chapter 6 of present work p. 97 7 Soames ends up with two notions of predication: direct predication and indirect predication . See Chapter 6 pp. 120–123. 8 Hence Soames needs a number of primitive mental acts beyond predication. See [Soames 2010a] p. 115, 122; and Chapter 6 of present work pp. 97–99 9 Chapter 6 pp. 93–94

King020513OUK.indd 128 11/23/2013 12:59:00 PM CRITICISMS OF SOAMES AND SPEAKS 129

In any case, Soames claims that one way of entertaining the proposition that o is red is to see o as red.10 Having the perceptual experience of seeing o as red constitutes predicating redness of o. When I think the thought that o is red, or understand the sentence “o is red,” I likewise entertain the proposition that o is red and thereby predicate redness of o according to Soames. Th is means that in all three cases—perceiving, thinking, and understanding a sentence—the same primitive cognitive act of predication is occurring. Indeed, since the propositions that are the contents of perceptual experiences in general involve primitive mental acts all of which also occur in entertaining propositions in thought and understanding sentences of natural language, Soames is committed to the claim that all the primitive cognitive acts that occur in perception when one entertains a proposition by seeing things a certain way, also occur in entertaining the proposition in thought, and understanding a natural language sentence that expresses that proposition. I have three related concerns with this. First, what sort of claim is the claim that e.g. the same cognitive act of predication occurs in perceptual experience, thinking and understanding sentences of natural language? It sounds like a (high level) empirical claim; and I am not sure how else to take it. But then the claim seems highly speculative and I wonder what evidence there is for it. Second, the claim that the same cognitive act of predication occurs in thinking and in perception seems quite dubious. In perception, except in some very limited range of cases, I have no choice as to what things appear to have what properties. In thinking, I have complete freedom in what properties and relations I think about things as possessing. Th is would seem to me evidence right off that the cognitive “acts” by means of which properties are attributed to objects in perception and the cognitive acts by means of which properties are attributed to objects in thought are very diff erent kinds of acts, contrary to what Soames claims. Perhaps Soames would admit that at this lower level the cognitive acts in perception and thinking diff er, but that at a higher level of abstraction, the same cognitive act is being performed. But then I am back with the fi rst worry: this claim seems highly speculative and I wonder what evidence there is for it. Finally, the claim that all the primitive cognitive acts that occur in perception as a result of which perceptual experiences have propositional content also occur in entertaining propositions in thought and understanding sentences of natural language entails a degree of cognitive unity and convergence in perception, thought and language that seems to me implausible. A second problem with Soames’s view is that it requires commitment to idiosyn- cratic metaphysical views that many, and probably most, would reject. Since Soames identifi es propositions with event types, in asking whether all the propositions exist on his view that intuitively should exist, we are asking whether all the relevant event types

10 Soames [2010] p. 81

King020513OUK.indd 129 11/23/2013 12:59:00 PM 130 JEFFREY C. KING

exist. Soames provides two principles regarding the conditions under which event types qua propositions exist: 1. If an event type E has instances that exist, then E exists.11 2. if (i) R is an n-place property for which there have been events in which an agent

predicates R of things, and (ii) o 1 . . .o n are objects for each of which there have been events in which an agent refers to it, then (iii) there exists a proposition p which is

the (minimal) event type of referring to o1 . . .on and predicating R of them— even if no one has ever performed that predication, and hence there exist no instances of p .12 Th ese principles seem plausible enough. However, there are lots of objects that have never been referred to and properties that have never been predicated of anything. Principles 1 and 2 don’t secure the existence of event types qua propositions involving such objects and properties. Hence, on Soames’s account such propositions don’t exist. But surely, many such propositions exist and are true (e.g. the proposition that some never-referred-to molecule o is a molecule, which Soames himself considers). Soames responds to this worry by saying that this would only be a problem if propositions had to exist to be true. But they don’t!13 Th ese propositions that have as constituents never-referred-to objects and never-predicated relations are merely possible and so don’t exist on Soames’s view. But he holds that many non-existent things possess properties. Soames gives as examples an individual having the property of being dead, being referred to by me and being admired by someone. He also gives the example of Plato and Aristotle instantiating the relation of non-identity. So, according to Soames, there is nothing to prevent non-existent, merely possible propositions—like the proposition that the never-referred-to molecule o is a molecule—from having the property of being true. I think it is fair to say that most philosophers would reject Soames’s alleged examples of non-existent things possessing properties (e.g. four dimensionalists would deny that Nixon possessing the property of being dead is an example of a nonexist- ing thing possessing a property; nihilists would deny that a sentence like “Nixon is dead” expresses (or seems to express) a truth because a nonexistent thing possesses a property). Relatedly, I think it is also fair to say that most philosophers reject the view that non-existents may possess properties. It is precisely because they wish to reject such a view that most actualists are so-called serious actualists and most presentists are serious presentists. (Th e former are actualists who hold (roughly) that if a thing o possesses a property at a world w, it exists at w; the latter are presentists who hold (roughly) that if a thing o possesses a property at a time t, it exists at t.) So adopting Soames’s view of propositions requires one to hold a metaphysical view that most philosophers reject.

11 C h a p t e r 6 p . 1 0 1 12 C h a p t e r 6 p . 1 0 2 13 Chapter 6 p. 103

King020513OUK.indd 130 11/23/2013 12:59:00 PM CRITICISMS OF SOAMES AND SPEAKS 131

A third problem with Soames’s account is that it appears to allow there to be too many propositions. More specifi cally, it distinguishes propositions that intuitively shouldn’t be distinguished; and it is forced to posit the existence of propositions that intuitively don’t exist. Taking these points in turn, Soames claims that propositions are event types in which an agent performs a sequence of cognitive acts such as predicating a property of an object , applying a function to an argument, and performing the operation of reduction on a relation and an argument, whereby an n-place relation and an argument are mapped to the relevant n-1 place relation.14 Obviously, on this view diff erent sequences of cognitive acts yield diff erent event types and so diff erent propositions. But this forces advocates of the view to claim that there are more propositions than there intuitively are. Consider this sequence of cognitive acts: an agent fi rst performs the cognitive action of reduction by combining the argument Juliet with the 2-place relation loves (on the second “argument place”), yielding the 1-place property ___ loves Juliet. Th e agent then predicates this (directly, in Soames’s sense) of Romeo. Th at’s the (or a) proposition that Romeo loves Juliet. Call this proposition Romeo 1 . But now consider this diff erent sequence of cognitive acts: an agent combines Romeo with the relation loves by reduction (on the fi rst “argument place”), yielding the property Romeo loves__ . Th e agent then predicates this of Juliet. Th at also looks like the proposition that Romeo loves Juliet. But it is a diff erent sequence of cognitive acts than the fi rst, and so is a diff erent proposition. Call this proposition Romeo 2 . Finally, consider this sequence of cognitive acts: an agent predicates loves of the pair Romeo, Juliet in that order. Th is again looks to be the proposition that Romeo loves Juliet. But it too is a diff erent sequence of cognitive acts than the fi rst two. Call this proposition Romeo 3 . So how does Romeo love Juliet? Let’s count the ways. It looks like there are three distinct propositions, because of three distinct event types of agents performing sequences of cognitive acts, that are all propositions to the eff ect that Romeo loves Juliet. Th is is surely implausible. Soames might try to block there being the proposition Romeo 2 by claiming that the cognitive act of reduction when applied to an n-place relation and an argument has to give us back the n-1 place relation that results from the argument “saturating” the nth argument place of the relation. So if I perform reduction on Romeo and loves, it can only yield the property loves Romeo. To this I can only respond that I fi nd that I can in thought combine Romeo with the relation loves yielding the property Romeo loves__ and that I can then predicate that property of Juliet.15 S o a m e s also might try to block there being a proposition Romeo 3 by denying that I can predicate a 2-place relation of Romeo and Juliet, in that order. Again, I can only report that I can do this in thought. So here Soames’s account delivers three propositions where it

14 Soames [2010] p. 115, 120; and Chapter 6 pp. 99–100 and especially 123–124. 15 Frege seems to agree. Concerning thoughts expressed by sentences containing multiple names, Frege [1979] writes “A thought that is put together in this way is just what traditional logic calls a singular judgement. We must notice, however, that one and the same thought can be split up in diff erent ways and so can be seen as put together out of parts in diff erent ways.” (pp. 201–202).

King020513OUK.indd 131 11/23/2013 12:59:00 PM 132 JEFFREY C. KING

should deliver one. Similar remarks apply to relations of higher adicity and arguments, where we will get an even greater proliferation of “the same” proposition. A second case of Soames’s theory delivering propositions where it shouldn’t concerns his account of the de se . On Soames’s view, a proposition can require to be entertained that some of its constituents be cognized in a particular way or it can fail to do so. So for example, what Soames calls the de re proposition that John Perry is the messy shopper is entertainable simply by predicating being the messy shopper of Perry, where there are no constraints on how Perry is thought of (though one must refer to him directly). What Soames calls the de se counterpart of this proposition is entertainable by predicating being the messy shopper of Perry while thinking of Perry in “the fi rst person way.” 16 Only Perry can entertain this proposition. So entertaining this proposition requires one to cognize Perry in a special way. Apparently worried that the claim that propositions can require in order to be entertained that things be cognized in a certain way could lead to there being all manner of strange propositions, Soames writes:

Surely, there are limits on what cognitive acts or operations propositions can encode. Th inking of a certain tune while predicating redness of an object is a (complex) cognitive activity of some sort, as is predicating a relation of a pair, while feeling aff ection toward its members. Presumably, the event types of doing these things are not propositions because one of their cognitive components is orthogonal to how the agent represents things to be.17 Later Soames says that whether a proposition can require that something be thought of in a given way in order to be entertained depends on whether the way of thinking of the thing is representational: whether it is a matter of the thing being represented in a certain way. In entertaining the de se proposition that Perry is a messy shopper, the required way of thinking about Perry is the fi rst person way. Since this is a matter of representing Perry in a certain way , this proposition is kosher. However, in predicating being smart of Annie while feeling aff ection for her, my feeling aff ection for Annie is apparently for Soames not representing her in a certain way. Hence, there is no proposition in which being smart is predicated of Annie, and which to be entertained requires me to feel aff ection for Annie in so doing. Th at all sounds good. Th e problem is that there are all kinds of ways of thinking of things that constitute representing them as being a certain way, so the way of thinking/representing meets Soames’s criterion for being allowably “encoded” in propositions, where it is just very implausible to suppose that there are propositions such that to be entertained requires thinking of things in those ways. Hence Soames’s account commits him to the existence of propositions that it is implausible to think exist. For example, Soames seems committed to a proposition the entertaining of which requires me to predicate redness

16 Chapter 6 pp. 110 17 Chapter 6 p. 118

King020513OUK.indd 132 11/23/2013 12:59:00 PM CRITICISMS OF SOAMES AND SPEAKS 133

18 of o, while thinking of it as inhabiting a world in which water is H 2 O . Similarly for the proposition whose entertaining requires me to think of o as self identical while predicating redness of it, the proposition whose entertaining requires me to think of o as identical to something while predicating redness of it, and so on. So here Soames’s account predicts the existence of propositions whose existence is implausible. A fourth problem with Soames’s view is that it appears to involve a circularity. As I’ve indicated, like me, Soames thinks that it is important to explain how/why it is that propositions have truth conditions, and so are representational .19 His explanation is that propositions qua event types are representational “ because of their intrinsic connection to inherently representational events [tokens] in which agents predicate some things of others.” 20 Soames’s idea apparently is that event tokens of e.g. agents predicating redness of o are inherently things with truth conditions. Th e event type of an agent predicating redness of o has truth conditions because of its “intrinsic” connection to event tokens that are inherently representational. I’ll consider later whether this explanation is successful. For now, note that Soames’s explanation here has to appeal to possible event tokens of agents predicating some things of others. For consider an event type/proposition P that has no tokens. Th en there are no actual event tokens with inherent truth conditions for P to be intrinsically connected to. However, P has truth conditions because any possible token of P inherently has truth conditions.21 H e n c e , the explanation of why/how P has truth conditions appeals to its intrinsic connection to possible tokens of P inherently having truth conditions. Now presumably possible tokens of P are just tokens of P in other possible worlds. Hence, Soames’s explanation of why P has truth conditions appeals to possible worlds (in which P has tokens), and hence presupposes that there are such things. But now how does Soames understand possible worlds? He writes:

Suppose, as I do, that a world-state is a property, attributable to the universe, of making true a set of basic propositions that tells a complete world story—e.g., the property making it true that the earth is the third planet from the sun in the solar system, that the earth is round, that the earth is largely covered with water, that the earth is inhabited by humans and other animals, etc., etc., etc. Imagine defi ning a world-state w by fi lling this out so as to provide such a complete

18 Don’t say that there is such a proposition: that o is red and inhabits a world in which water is H 2 O. On Soames’s view, just as we want to distinguish the proposition that I am the messy shopper and am thinking of myself in the fi rst person way from Soames’s (alleged) de se proposition that I am the messy shopper, so we want to distinguish the proposition the entertaining of which requires me to think of o as inhabiting a

world in which water is H 2O while I predicate redness of o from the proposition that o is red and inhabits a

world in which water is H 2O. On Soames’s view, the former does not predicate a conjunctive property of o; the latter does. 19 Soames [2010] pp. 6, 32, 55–56, 64, 106–107; Chapter 6 pp. 96–97 20 Soames [2010] p. 107. Similar remarks occur on pp. 96–97 in Chapter 6 21 See p. 104 for Soames’s [2010] explanation of why the proposition that snow is white is true at a world w where it has no tokens, where Soames explicitly appeals to possible tokens of the proposition qua event type that snow is white. See also Chapter 6 p. 96 where the same appeal is made.

King020513OUK.indd 133 11/23/2013 12:59:01 PM 134 JEFFREY C. KING

story—where by “complete” I mean one that answers all questions relevant to a contextually determined inquiry. 22 Soames’s analysis of possible worlds obviously presupposes that there are propositions with truth conditions. Indeed, Soames even claims that the basic propositions mentioned in the quotation above are constituents of the property that he identifi es with a possible world in that same quotation.23 But Soames’s explanation of how/why propositions have truth conditions presupposes that there are possible worlds. So for Soames, that there are propositions with truth conditions presupposes that there are possible worlds (since the explanation for why at least some propositions have truth conditions appeals to merely possible tokens of them). But that there are Soamesian possible worlds presupposes that there are propositions with truth conditions (since these are constituents of the property that is the possible world—it is the property of making these propositions true, and surely that presupposes that there are these propositions with truth conditions). Th at is circular. More generally, that Soames’s explanation of how/why propositions have truth conditions appeals to possible worlds precludes anyone who adopts Soames’s view from constructing possible worlds by using propositions in any way (e.g. as sets of propositions as in Adams [1981]). Th us, endorsing Soames’s view here precludes adopting types of views in another area that many fi nd attractive. As I’ve indicated several times, Soames agrees with me that some explanation must be given of the fact that propositions have truth conditions and so represent things as being a certain way. Th e fi nal diffi culty I’ll raise for Soames’s account of propositions is that his explanation of how/why propositions have truth conditions fails. Soames states his explanation of why propositions represent/have truth conditions as follows:

Th is suggests that the proposition that o is red is simply the minimal event type in which an arbitrary agent predicates being red of o. Th is event-type is representational because every conceivable instance of it is one in which an agent represents something as being a certain way. What it represents is what is representationally common to all such instances. Since every such instance is one in which o is represented as being red, we speak, derivatively, of the proposition itself representing o as red. 24 Later he puts the same point this way:

Otherwise put, to entertain the proposition that o is red is to predicate redness of o, which is to do something that results in an instance of the event type that the proposition is. Th e representationality, and hence truth conditions, of the proposition are due to the representational features of these possible instances. 25

22 Chapter 6 pp. 113–114 23 Chapter 6 p. 115 24 Chapter 6 p. 96 25 Chapter 6 p. 96

King020513OUK.indd 134 11/23/2013 12:59:01 PM CRITICISMS OF SOAMES AND SPEAKS 135

Soames doesn’t say so here, but he claims that instances/tokens of the event type of predicating redness of o are things that “inherently” have truth conditions and so represent.26 Th is is the fi rst step in his explanation of how/why propositions qua event types have truth conditions. For the second step, Soames claims that since every event token of the event type predicating redness of o has truth conditions, we say— derivatively (see the last sentence of the fi rst quotation above)—that the event type itself has truth conditions. I’ll be challenging both the steps of Soames’s explanation here. Th e fi rst step of Soames’s explanation is the claim that event tokens of agents predicating redness of o inherently have truth conditions. However, Soames gives no argument for this claim nor does he explain how/why such event tokens have truth conditions inherently. Th e most he says is that in predicating redness of o, an agent represents o as red.27 But how do we argue from that claim to the claim that the event token of the agent predicating redness of o itself represents o as red and so is true iff o is red? 28 Soames doesn’t say; and it just doesn’t seem to follow at all. In general, when an agent bears a relation R to something o at a time, the event token of the agent bearing R to o does not itself bear R to o. If I hug Annie, the event token of my hugging Annie doesn’t hug Annie. So why, from that fact that an agent represents o as red (by predicating redness of it), would it follow that the event token of the agent representing o as red itself represents o as red? In any case, Soames off ers no argument that event tokens of predicating redness of o have truth conditions. Further, the claim that the event tokens in question inherently have truth conditions is mysterious. I can see how event tokens could have truth conditions in virtue of agents interpreting them in certain ways. But how could an event token inherently have truth conditions? How could an event token have truth conditions by its very nature? Th at seems as mysterious as the claim that propositions are sui generis abstract entities that have truth conditions by their natures and independently of minds and languages, which I and Soames himself both reject as unintelligible. Given that it is a mystery how event tokens could have truth conditions inherently, in the absence of any argument for the claim that the event tokens of agents predicating redness of o inherently have truth conditions or any explanation of how this could be so, we should reject the claim.

26 See Soames [2010] p. 107. In the fi nal sentence of the above quotation Soames appeals to the “representational features” of possible event tokens of the event type of an agent predicating redness of o. Th ese representational features presumably include the having of truth conditions. 27 “. . .what it is to entertain it [the proposition that o is red] is simply to predicate redness of o and so represent o as red.” Chapter 6, p 96. Here it is an agent doing the entertaining, predicating and representing. 28 Looking at the quotations from Soames, it might be thought that all he wants to claim is that in an event token of an agent predicating redness of o the agent represents o as red and not that the event token of her predicating redness of o does so as well. But given that o is represented as being red, something must be such that it is true iff o is red (Soames oft en moves from the claim that something represents o as red to the claim that it is true iff o is red). When Soames talks about representational features of event tokens, he suggests that the thing that is true iff o is red is the event token of predicating redness of o. And in Soames [2010] he is quite explicit in claiming that event tokens in which agents predicate things of other things are representational (p. 107).

King020513OUK.indd 135 11/23/2013 12:59:01 PM 136 JEFFREY C. KING

Finally, what evidence there is suggests that the event tokens in question do not have truth conditions inherently. Suppose Vicky is perceiving o as red and that this event token is quite salient to us. If I say, nodding at Vicky “What is occurring is true (false)” or “Th e event of Vicky seeing o as red is true (false),” this just sounds like a category mistake. (Indeed, the second sounds like a misguided attempt to say that the event occurred.) And the same is true of any attempt to predicate truth or falsity of an event token of an agent predicating redness of o. 29 But if Soames is right that such event tokens are inherently things with truth conditions, why would predicating truth or falsity of them sound so anomalous as to seem like a category mistake? I conclude that what evidence there is suggests that event tokens in which agents predicate redness of o do not have truth conditions inherently. Hence the fi rst step in Soames’s explanation of how/why propositions have truth conditions fails. 30 Recall that the second step of Soames’s explanation as to how/why propositions have truth conditions was to say that the event type of an agent predicating redness of o, which Soames identifi es with the proposition that o is red, itself has truth conditions derivatively , because every possible event token of an agent predicating redness of o has truth conditions. Of course, we have seen reason to reject the claim that such event tokens do inherently have truth conditions. But suppose we were to grant that they did. Would it follow that the relevant event types derivatively represent and so have truth conditions as Soames claims? I think there is good reason to think not, at least in any substantive sense, even by Soames’s lights. Further, there is reason to think the claim that event types have truth conditions in any substantive sense (derivatively or otherwise) is false, again, even by Soames’s lights. Taking the former point fi rst, suppose that event tokens of predicating redness of o represent o as being red. Does it follow that event types of predicating redness of o derivatively represent o as being red? Only in a very stipulative and non-substantial way. Soames tells us that the sense in which the event type of predicating redness of o derivatively represents o as red is that in every instance of the event type o is represented as being red. Th at means that we could in an exactly similar sense say that the event type of an agent hitting Alan derivatively touches Alan because in every instance of it

29 E.g. Suppose I ask Vicky to think of o as red. As she is doing so, if I say “Th e event of Vicky thinking of o as red is true (false).” or “Th e event Vicky is now bringing about is true (false).” again this sounds like a category mistake. 30 Someone might attempt to use this sort of argument against my account of propositions. Doesn’t predicating truth or falsity of the facts that I claim are propositions sound anomalous too? Doesn’t that show that the facts I claim are propositions don’t have truth conditions? But I have an explanation of this not available to Soames. I can be acquainted with the fact that is the proposition that Rebecca swims qua fact or qua proposition. When I am acquainted with it as just another fact in the world, I have no reason to think it has truth conditions since facts generally do not have truth conditions. However, when I am acquainted with this fact qua proposition that Rebecca swims, I cannot fail to see that it has truth conditions, (this is discussed in King [2007] pp. 50–52). Note that this explanation works precisely because facts do not have truth conditions inherently, but rather are endowed with truth conditions by us. Since Soames holds that his event tokens inherently have truth conditions, this sort of explanation is not available to him.

King020513OUK.indd 136 11/23/2013 12:59:01 PM CRITICISMS OF SOAMES AND SPEAKS 137

Alan is touched. Clearly there is no substantive sense in which the event type of hitting Alan touches Alan. But then there is also no substantive sense in which the event type of predicating redness of o represents o as being red. Th is is closely related to another reason why, even by Soames’s lights, event types don’t represent, and so have truth conditions in anything but a nonsubstantial, stipulated sense even if we grant that event tokens do. Recall why Soames thinks that event tokens of agents predicating redness of o have truth conditions and so represent o as red. Th e reason is that by predicating redness of o, the agent represents o as red and thereby we have something that is true iff o is red.31 So that something is predicated of something “in” an event token is the key to its having truth conditions. Th e problem is that nothing is predicated of anything “in” the event type of an agent predicating redness of o. Th is is easy to see by supposing this event type exists at a world without agents and so has no instances. It will still be the case that nothing is predicated of anything at that world. But then these event types, unlike their tokens, lack exactly what is required for the tokens to represent and have truth conditions according to Soames: predicating something of something. Perhaps Soames would say that the event type of predicating redness of o derivatively predicates redness of o, and so has truth conditions, in the sense that in all its instances, redness is predicated of o. 32 But we have already seen how non-substantial and stipulative this sense of derivative predication is bound to be. Th e crucial point is that the sense in which the event type of predicating redness of o derivatively predicates redness of o is simply that in each of its instances redness is predicated of o. Th e fact remains that the event type itself predicates nothing of anything. But then it remains true that the event types lack what Soames thinks is the key to having truth conditions: predicating something of something. Th at in instances of the event type predication occurs, and in this sens e something is derivatively predicated of something in the event type, is quite beside the point. Hence, by Soames’s own lights, the event types he takes to be propositions lack what is essential for having truth conditions. Putting aside Soames’s argument for the claim that the event type of predicating redness of o has truth conditions, let’s turn to the second point mentioned above: there is reason to think that the claim that the relevant event types have truth conditions is false, even by Soames’s lights. Th e problem is that, as was the case with event tokens of predicating redness of o, what independent evidence there is suggests that the event

31 Soames makes clear that predicating is the key to these event tokens having truth conditions, and so representing, when he writes “. . .what it is to entertain it [the proposition that o is red] is simply to predicate redness of o, and so represent o as red .” Chapter 6, p. 96 my emphasis. Soames [2010] writes “When we see an object o as red, we predicate redness of o. It is in virtue of this that our perceptual experience represents o as being red. ” (p. 81, italics my emphasis). Soames is talking about the so-called defl ationary theory of propositions here and not his own “cognitive realist” theory. But the alleged insight that the presence of predicating is the key to an event token having truth conditions is preserved in his theory, as the previous quotation makes clear. 32 See Soames [2010] pp. 100–101 where he is discussing whether the act type of predicating redness of o predicates anything of anything.

King020513OUK.indd 137 11/23/2013 12:59:01 PM 138 JEFFREY C. KING

type of predicating redness of o does not have truth conditions. To see this, note fi rst that we can speak of event types as occurring or happening . Th us, if I see a pedestrian get hit by a car at the corner of Amsterdam and 87th, I can say “Th at happens every week.” and mean that the event type of a pedestrian getting hit at Amsterdam and 87th has instances every week. So the demonstrative “that” here picks out an event type. 33 Similarly, for locutions like “what just occurred” (e.g. my electricity goes off and I say “what just occurred happens every day at this time.”). So by indicating a token of an event type and using expressions like those just mentioned, I can talk about the event type. And by talking about an event type happening , I am talking about its having instances. Now according to Soames, entertaining a proposition is simply tokening an event of the type that is the proposition (by performing the acts of predication involved in tokening the type).34 So suppose I ask Vicky to entertain the proposition that arithmetic is incomplete, if consistent, and she complies. Hence she tokens an event that is of the type that is the proposition that arithmetic is incomplete, if consistent (roughly, by predicating being incomplete if consistent of arithmetic). I now say “What just occurred is true.” Th is is bizarre to the point of being incoherent. But if Soames is right, this should sound fi ne and be true. For the expression “what just occurred” should be capable of being used to talk about the event type that Soames claims is the proposition that arithmetic is incomplete if consistent. And of course since Godel proved this proposition, it is true.35 Finally, evidence similar to the above strongly suggests that the event types that Soames identifi es with propositions are not propositions. 36 As before, imagine that we ask Vicky to entertain the proposition that arithmetic is incomplete if consistent and she complies by tokening the event type that Soames claims is the proposition. Again, we should be able to talk about the event type using the expression “what just occurred.” I say “What just occurred was proved by Godel.” Again, this is incoherent and surely is a category mistake. But again, if Soames were right this should be true: again, Godel did prove the proposition that arithmetic is incomplete if consistent. Similar remarks apply to predicating of the relevant event types many things that can sensibly be predicated of the propositions Soames identifi es with these event types. 37 Further, if we predicate of propositions properties that are had by the event types that Soames claims are propositions, again the predications are bizarre to the point of incoherence: “What

33 If it picked out the event token, I would have asserted the absurdity that the relevant event token happens every week. 34 Soames [2010] p. 106 35 Note that even if “what just occurred” picked out the event token of Vicky entertaining the proposition that arithmetic is incomplete, if consistent (as it surely could), the sentence “What just occurred is true.” should still be true in this situation on Soames’s view since according to him, the event token in question is true. 36 Of course, the above evidence that the relevant event types don’t have truth conditions is also evidence that they aren’t propositions. My point here is that there is additional evidence that they are not propositions. 37 For example, “What just occurred entails that Hilbert’s program is impossible.”

King020513OUK.indd 138 11/23/2013 12:59:01 PM CRITICISMS OF SOAMES AND SPEAKS 139

Godel proved occurred twice today.”; “What Godel proved just happened.”; “What Godel proved occurred all over the world today.” All of this strongly suggests that the event types identifi ed by Soames are not propositions. Th e arguments just given that Soames’s event types don’t have truth conditions and are not propositions amount to noting the following. When the events types are picked out with language appropriate to picking out event types, it sounds incoherent to predicate truth or falsity of them. It is also incoherent to predicate of them other things that are coherently and truly predicated of propositions. Th e explanation of the incoherence is that the event types are neither true nor false, nor are they propositions. And when we pick out propositions using language appropriate to picking out propositions (“the proposition that. . .”), it is incoherent to predicate of them things that are coherently and truly predicated of event types. Th e explanation of this incoherence is that propositions are not event types. Now Soames uses arguments of exactly this sort to refute the view that propositions are act types. Soames [2010] writes

Whereas it is perfectly coherent to say that predicating brilliance of John is what I just did or that the proposition that John is brilliant is false, it would be incoherent to say “*Th e proposition that John is brilliant is what I just did.” or “* What I plan to do (when I plan to predicate brilliance of John) is false.” Moreover, although it is correct to say that I believe, and Godel proved, the proposition that arithmetic is, if consistent, incomplete, it would be absurd to say “ What I believe, and Godel proved, is something I just did.” Th e source of the absurdity is not hard to locate. Act types—like kissing Martha or predicating incompleteness of arithmetic—are either themselves a certain kind of property, or something closely akin to properties. As such, they are not the kinds of things that have truth conditions.38 Hence, if Soames thinks that his arguments refute the view that propositions are act types, he should think that my arguments refute the view that propositions are event types. 39 Having now explained and criticized Soames’s account of propositions, I want to reiterate the important points of convergence between his view and mine mentioned at the outset. Soames and I both claim that classical conceptions of propositions, according to which they represent things being a certain way, and so have truth conditions independently of minds and languages and by their very natures, must be rejected. Th e problem is we have no idea how anything could do this, and so such theories make the fact that propositions have truth conditions a complete mystery. It is important to see that if Soames and I are right about this, it takes virtually every theory of propositions except his and mine off the table (though see the last paragraph of my comments on Speaks’s view below). Soames and I also think that propositions have truth conditions in virtue of things agents do. Th is makes certain facts about agents explanatory prior

38 Pp. 101–102 39 Couldn’t an argument like this be given against my view of propositions as well? See my discussion in note 30 of why I think I can explain the anomalousness of predicating truth and falsity of the facts I claim are propositions in a way that is consistent with my claim that they are propositions.

King020513OUK.indd 139 11/23/2013 12:59:01 PM 140 JEFFREY C. KING

to propositions having truth conditions. Whether the details of my view or Soames’s is ultimately correct, we both believe that the correct account of propositions must respect these insights.

Speaks’s account of propositions Jeff Speaks defends the view that propositions are properties. Th e proposition expressed by “Amelia talks,” for example, is the property being such that Amelia instantiates talking .40 Speaks holds that this property is expressed by the sentence in virtue of the fact that the syntactic concatenation in the sentence contributes the relation ___is such that ___ instantiates___ to the proposition expressed, and the contents of “Amelia” and “talks” (Amelia and the property talking , respectively) fi ll the second and third slots in this relation (in that order), yielding the property being such that Amelia instantiates talking. Truth for propositions is just instantiation: a proposition is true iff it is instantiated. Further, if a proposition is instantiated, everything instantiates it. If the actual world is such that Amelia instantiates talking , than so am I, you and everyone we know. 41 Further, a proposition P is true at a world w iff if w were actual, P would be instantiated. 42 Since we want there to be false and even necessarily false propositions, Speaks’s account requires uninstantiated properties and even properties that couldn’t be instantiated. As Speaks notes, this is a cost of his account but the price doesn’t seem terribly high. Obviously, it does mean, though, that no one whose theory of properties eschews uninstantiated properties can adopt Speaks’s account. Turning to the propositional attitudes, according to Speaks to believe that Amelia talks is to bear an attitude towards the property being such that Amelia talks. In particular, Speaks claims it is to believe something is such that Amelia talks. But putting things this way obscures the fact that for Speaks belief is a relation between an individual and a property (what appears to be said to be believed here is that something is such that Amelia talks, and that doesn’t sound like a property). I think that to better make explicit that belief is a relation to a property for Speaks, he should say that to believe that Amelia talks is to bear the believes-instantiated relation to the property being such that Amelia talks (I’ll discuss why Speaks really needs to say this below). Speaks holds that one may believe-instantiated a property because there is some particular thing

40 Speaks indiff erently goes back and forth between saying the proposition that Amelia talks is the property being such that Amelia instantiates talking and the property being such that Amelia talks. I will as well. 41 To paraphrase Wang Chung. 42 Th is simple view really isn’t Speaks’s because of propositions like the proposition that Rick Santorum doesn’t exist. Pick a world w where Santorum doesn’t exist. If this world were actual, arguably the property being such that Rick Santorum doesn’t exist itself wouldn’t exist. But then arguably this property wouldn’t be instantiated and so the proposition that Rick Santorum doesn’t exist wouldn’t be true. However presumably we want the proposition that Rick Santorum doesn’t exist to be true at w. Hence a diff erent account is called for here.

King020513OUK.indd 140 11/23/2013 12:59:01 PM CRITICISMS OF SOAMES AND SPEAKS 141

that one believes instantiates the property. So according to Speaks, two people may believe Amelia talks, where one believes o instantiates being such that Amelia talks and the other believes that o’ instantiates being such that Amelia talks, where o≠o’ (and the fi rst has no beliefs involving o’ and the second has no beliefs involving o). Speaks seems to think that the normal case might involve believing-instantiated the property being such that Amelia talks because of believing that the actual world instantiates it. Below I’ll cast doubt on this claim. Speaks also holds that there are other cognitive relations we bear to properties that are not propositions. (Here he wants to give an account of “fi rst person” mental states that others have called belief, desire, etc. de se. ) Agents can bear the relation of self-attribution to properties like being on fi re. Th is gives us “fi rst personal” mental states of the sort discussed in the literature on de se belief. But note that Speaks’s account here is not one of de se belief because the properties self-attributed are not propositions and self-attribution is not belief. On Speaks’s account, I could believe Jeff King is on fi re, which for Speaks amounts to my believing-instantiated the property being such that JK is on fi re. Or I could take myself to be on fi re, which for Speaks amounts to self-attributing the property being on fi re . Th e former is my standing in a relation to a proposition; the latter is not. It is worth noting here that Speaks does not really capture all the “fi rst personal” mental states that one who goes in for this sort of thing might want. First person desires for Speaks amount to “self-desiring” properties. So my fi rst person desire to be a professional skier is my self desiring the property being a professional skier (and this latter property is not a proposition).43 However, such an account stumbles on the Nolan [2006] example of the “fi rst person desire” not to exist. For Speaks, this would be understood as self-desiring the property not existing . Since this has to be understood as something like (fi rst personally) wanting to possess a property, and since I could not possess this property (for Speaks as for me, I must exist to possess a property), in self-desiring not to exist I am doing something incoherent on Speaks’s account. But intuitively, I am not. 44 As to the semantics of verbs of propositional attitude, Speaks takes “believes” to express the following two-place relation between a person and a property: ___ takes to be instantiated the property___ (similar remarks apply to other verbs of attitude).45 I t i s important to see that for Speaks e.g. belief ascriptions do not by means of their semantics ever assert that one self-attributes a property (similar remarks hold for desire ascriptions etc.). Th us for Speaks, both “I believe JK is on fi re” and “I believe I am on fi re” (uttered by me) express the proposition that JK stands in the believes-instantiated relation to the property (proposition) being such that JK is on fi re . Speaks suggests that perhaps the proposition that JK self attributes being on fi re can be pragmatically

43 Chapter 5 p. 85 44 Th ere may be many things bad about desiring not to exist, but it being incoherent to do so is not one of them. 45 Chapter 5 p. 85

King020513OUK.indd 141 11/23/2013 12:59:01 PM 142 JEFFREY C. KING

conveyed by my uttering “I believe that I am on fi re.” But Speaks is upfront about the fact that he has no explanation of the mechanism by means of which this proposition is pragmatically conveyed by uttering the sentence in question, while acknowledging that an account of the mechanism is required in order to sustain the suggested pragmatic approach. Having sketched Speaks’s account, and before turning to critical remarks, let me say that I think Speaks’s view has a lot to recommend it. Indeed, it is the best version of the propositions-as-properties view known to me. Let me turn to critical remarks. My fi rst and second objections concern features of Speaks’s account of propositional attitudes. Recall that Speaks claims that to believe that Amelia talks is to believe that something is such that Amelia talks. I suggested above that Speaks should say instead that to believe Amelia talks is to believe-instantiated the property being such that Amelia talks. Here is why. If we say that believing that Amelia talks is believing something is such that Amelia talks, what do we say about this latter belief? Applying the account again, we get that believing something is such that Amelia talks is believing that something is such that something is such that Amelia talks. But now we must again apply the account to this latter belief, which will be believing something is such that something is such that something is such that Amelia talks. Once again, the account applies to the latter belief and so on. To avoid this regress, we must understand Speaks to be saying that to believe that Amelia talks is to believe-instantiated the property being such that Amelia talks. (As mentioned above, this is a matter of bearing the two-place believe-instantiated relation between individuals and (propositions) properties to the property being such that Amelia talks .) Th at is why in explicating his view above, I took him this way. Call a belief like believing-instantiated the property being such that Amelia talks a general belief (since you are merely “taking the property to be instantiated” without necessarily taking any particular thing to instantiate it). Now recall that Speaks claims that we oft en have such a general belief because we believe of some particular o that o instantiates being such that Amelia talks . Call this a specifi c belief (since you take a particular thing to instantiate the property being such that Amelia talks ). So Speaks says that sometimes we have the general belief that Amelia talks (by believing-instantiated the property being such that Amelia talks ) because we have the specifi c belief that o instantiates being such that Amelia talks . But believing that o instantiates being such that Amelia talks again on Speaks’s view amounts to believing-instantiated the property being such that o instantiates being such that Amelia talks . Th is means that the so-called specifi c belief that o instantiates being such that Amelia talks has really simply turned out to be the general belief of believing-instantiated the property being such that o instantiates the property of being such that Amelia talks. In short, contrary to what Speaks suggests, all belief is general on his view: it is always a matter of believing-instantiated some property (that is a proposition). It is important to appreciate how peculiar this view is. It would be the analogue of holding that for non-proposition properties, we only

King020513OUK.indd 142 11/23/2013 12:59:01 PM CRITICISMS OF SOAMES AND SPEAKS 143

ever believe that something instantiates them. We never believe of particular things, that they instantiate them. Th at is a quite strange view. At the very least, Speaks should explain how we come to e.g. believe instantiated the property being such that Amelia talks without believing that a particular thing instantiates the property. Of course, as we saw, Speaks did hope to explain how one comes to believe-instantiated the property being such that Amelia talks precisely by claiming that one oft en does so because one believes that some particular thing o instantiates the property. He hoped to explain having general beliefs as the result of having specifi c beliefs. But given what I have just argued, that specifi c belief collapses into general belief on Speaks’s view, the proposed explanation now looks unnatural and implausible. For given what I have just argued, Speaks’s claim that we sometimes believe that something instantiates being such that Amelia talks because we believe that some particular thing o does so ends up in more offi cial terminology as being the claim that we sometimes believe-instantiated being such that Amelia talks because we believe-instantiated being such that o instantiates being such that Amelia talks . But as Speaks himself admitted (p.c.), this order of explanation doesn’t seem very natural when stated in the more offi cial terminology as we just did. Th is leaves Speaks without an explanation of how we come to simply believe-instantiated the property being such that Amelia talks . A second problem again concerns the propositional attitudes. As we just saw, for Speaks believing must be understood as bearing the believes-instantiated relation to properties (that are propositions). Similarly, desiring will be understood as desires-instantiated; assuming will be assumes-instantiated, and so on. So propositional attitudes for Speaks will be, roughly speaking, a matter of bearing various attitudes towards the instantiation of certain properties.46 For lots of attitudes, that seems quite plausible. But for some, it does not. I can consider, explain and understand the claim that arithmetic reduces to logic and in none of these cases does it seem that my attitude has anything to do with whether the alleged property being such that arithmetic reduces to logic is instantiated. Take considering: when I consider the claim that arithmetic reduces to logic I can do so without any attitude at all about the instantiation of the property being such that arithmetic reduces to logic. I may simply have an interest in really thinking about what the claim comes to. So here Speaks’s account of the attitudes seems strained at the very least. My third criticism is that it just does not seem as though the property being such that Amelia talks is something that is true or false. To say that it is seems like some sort of category mistake. Perhaps someone would respond that despite this, the property in fact is (say) true. Th ey might add that when we consider the claim that Amelia talks, we take it to be true; and the “that” clause here designates the property in question. So we do take properties like being such that Amelia talks to be true and false. Th e problem with this is that properties are most transparently expressed by predicates. If

46 Chapter 5 p. 85

King020513OUK.indd 143 11/23/2013 12:59:01 PM 144 JEFFREY C. KING

propositions really are properties as Speaks claims, why when we consider the predicate “being such that Amelia talks” that allegedly expresses the proposition in question are we not inclined to say that it expresses something true or false? It looks like Speaks will have to claim that when we encounter a proposition qua property as the thing designated by a “that” clause, we treat it as something that is true or false. But when we encounter it as the thing expressed by a predicate, we don’t do so. Th is is made all the more peculiar by the fact that, as indicated above, it is predicates that canonically and transparently express properties. Hence, when we encounter a proposition qua property as the thing expressed by a predicate, we will be more aware that it is a property. If propositions are properties, why when we encounter them via the linguistic devices that most make clear that they are properties do we precisely not want to treat them as things that are true and false? Why must we encounter propositions via linguistic devices that disguise the fact that they are properties (e.g. “that” clauses) in order to treat them as things that are true and false? I don’t see what answer Speaks can give to these questions. Fourth, and related to the third worry, Speaks himself holds that not all properties are propositions—things that are true and false and possess modal properties. For example, Speaks holds that being red is a property but not a proposition. 47 Th is seems right, as being red does not seem to be something that is true or false. However, this raises the question of which properties are propositions on Speaks’s view. Speaks admits that he has no account of what distinguishes the properties that are propositions from other properties. I think this is problematic for two reasons. First, absent such an account we really haven’t been told what propositions are. We can’t identify propositions with properties, because some properties aren’t propositions. And we can’t identify propositions with properties of type X, since Speaks admits he can’t give any account of what type X is. So there is an important sense in which Speaks’s view leaves us without an account of what propositions are . Second, not only do properties like being red not seem to be propositions—things that are true and false and possess modal properties—but above I claimed that even the properties Speaks takes to be propositions, like the property being such that Amelia talks, don’t seem to be things that are true and false and possess modal properties. If in general it doesn’t seem like properties are the kinds of things that can be true or false and bear modal properties, then we need to be told why/how it is that certain properties can be true or false and bear modal properties. We need to be told what is “special” about them such that they can do these things other properties can’t do. But, again, this is exactly what Speaks is unable to do. It seems to me this is deeply unsatisfying. We have entities—properties—that don’t seem to be the kinds of things that are true or false and bear modal properties. We are then told that some of them in fact are true and false and do possess modal properties.

47 Chapter 5 p. 90.

King020513OUK.indd 144 11/23/2013 12:59:01 PM CRITICISMS OF SOAMES AND SPEAKS 145

But nothing can be said about which properties are like this; and how/why they have these features that other properties lack. Now Speaks has a clever reply to this type of objection. He correctly points out that those who think that mental properties are a subset of physical properties do not in general provide a criterion for distinguishing the mental physical properties from the non-mental physical properties. We don’t take this as a decisive objection to the view that the mental properties are a subset of the set of physical properties. Hence that Speaks provides no criterion for distinguishing the proposition properties from the non-proposition properties is not a decisive objection to his view that propositions comprise a subset of the set of properties. In response, let me say that I do think that no criterion has been given for distinguishing the two is an objection to the view that mental properties are a subset of the physical properties. And I am sure that those who hold the view agree that this must eventually be done if the theory is to be successful. It is however reasonable for those who hold this view to think that continued study of mental and (non-mental) physical properties of the brain will yield further understanding; and that, as a result, an account of what distinguishes mental physical properties from non-mental physical properties will be forthcoming. But I, at least, am much less sanguine about the idea that further study of properties and propositions will provide a deeper understanding of them that will yield an account of how to distinguish proposition properties from non-proposition properties. So I don’t think that failing to provide a criterion for distinguishing mental physical properties from non-mental physical properties is on a par with Speaks failing to provide a criterion for distinguishing proposition properties from non-proposition properties. Fift h, there are conjunctive propositions, negated propositions, and disjunctive propositions.48 If propositions are properties, as Speaks claims, there must be conjunctive, negated and disjunctive properties. 49 However, as is well known, many who believe in properties do not think there are conjunctive, and especially negated and disjunctive, properties.50 One reason for this is that it is widely thought that when two things both possess the same “real” property, they should resemble each other or have a common nature; but common possession of negated and disjunctive properties does not in general make for similarity or possession of a common nature. So because Speaks’s account commits one to negated and disjunctive properties, it can only be adopted by the most promiscuous in their views about what properties there are. I would have hoped that a theory of propositions could remain neutral on this question. A sixth and fi nal objection concerns Speaks’s identifi cation of truth and instantiation for propositions. 51 It is generally thought that in having truth conditions, a proposition

48 Less contentiously, there are conjunctive, disjunctive and negated sentences that express propositions. 49 Less contentiously, there are properties expressed by conjunctive, disjunctive and negated predicates. 50 Or properties expressed by conjunctive, disjunctive and negated predicates. 51 Th e qualifi cation here is due to the fact that on Speaks’s view, for properties like being red instantiation isn’t truth.

King020513OUK.indd 145 11/23/2013 12:59:01 PM 146 JEFFREY C. KING

in some sense specifi es conditions that have to be met by a world for the proposition to be true there. Consider a proposition P specifying such conditions and a world w that meets them. Surely, we want to say in such a case that P is true at w because w is a certain way. Indeed, this seems like a truism. However, one would also think that a thing’s possessing an intrinsic property generally explains why the thing is a certain way. Th at I possess the property of being 6 feet tall explains why I am a certain way. Possession of the property constitutes my being a certain way. Now surely the same should be true of worlds possessing intrinsic properties; that the world possesses an intrinsic property constitutes its being a certain way. Suppose a world w possesses the property being such that snow is white . Th is is an intrinsic property of w.52 Th en, just as in other cases, that should explain why w is a certain way. However, Speaks claims that w possessing or instantiating a property like being such that snow is white is just this property qua proposition being true at w. But then on the Speaks account, we should say w is a certain way, because the property/proposition being such that snow is white is true at w (i.e. is instantiated at w). Unfortunately, this precisely reverses what we said was the proper order of explanation mentioned above; the proposition that snow is white is true at w because w is a certain way. Surely, this is the right order of explanation. Hence, that Speaks’s account has it that w is a certain way because a proposition is true there, is a strong reason for rejecting the account. Let me close by saying why I think a view of the sort Speaks sketches, according to which propositions are some kind of world properties, is the best alternative to views of the sorts that Soames and I favor on which propositions have truth conditions because of some facts about agents. Th e advantage that Soames and I see with such views is that they promise to off er an explanation of how/why propositions have truth conditions; and we both think the latter is something that very much needs to be explained. Th ere is a sense in which a view like Speaks’s off ers an explanation for why propositions have truth conditions. Propositions are properties of worlds (and everything or nothing). As such, they are by their nature the sorts of things that are instantiated or not by worlds. Th ey are also the kinds of things that would be instantiated or not if this or that world were actual. But instantiation just is truth for propositions/properties; and that a proposition/property would be instantiated if w were actual just is truth at w. So propositions by their nature are true or false (i.e. instantiated or not) at worlds, and so have truth conditions, simply because they are the kinds of things that would or would not be instantiated if this or that world were actual. Th e explanation here takes as primitive that properties by their natures are and would be instantiated or not by things, but that does not seem to be such a bad thing to take as primitive.

52 E.g. a duplicate of w would have to possess it.

King020513OUK.indd 146 11/23/2013 12:59:01 PM 8 Representational Entities and

Representational Acts

J e ff S p e a k s

Here’s one thing that the three of us have in common: we all dislike the idea that propositions could be entities which are intrinsically representational—in the sense that they both are representational and would exist and be representational, even if there were no subjects around to do any representing. My response to this negative claim is pretty straightforward: I deny that propositions are representational entities and, instead, identify the proposition expressed by “Amelia talks” with the property of being such that Amelia talks. Th is property is not— any more than any other property—a representational entity; properties are not about anything. I then argue that we can use properties of this kind to give an account of the sorts of phenomena for which we might have thought we needed representational entities: the representational mental states of subjects, and the fact that sentences, like the contents of these states, have truth conditions. King and Soames both give a diff erent kind of response to the denial that there are intrinsically representational entities. Th ey think that even if there are no intrinsically representational propositions (in the above sense), we still need representational propositions. It’s just that we need to explain how these entities come to have representational properties; and each tries to explain this by saying how the relevant entities come to have the representational properties they do in terms of the mental acts (broadly construed) of thinking subjects. King’s and Soames’s theories diff er in many important ways; in particular they diff er on the question of which mental acts provide the relevant representational properties. (In King’s case, these are interpretations of sentential and propositional relations, and in Soames’s they are events of predication.) But they both satisfy the very abstract sketch of the preceding paragraph, and so both are theories which fi ll in the following diagram in diff erent ways:

King020513OUK.indd 147 11/23/2013 12:59:01 PM 148 JEFF SPEAKS

Representational properties of non-fundamental mental states

Fundamental mental Representational states (intrinsically properties of representational) propositions

Representational properties of sentences

I’ll turn to the details of King’s and Soames’s theories below. But the fact that they share the above structure is enough to raise two general questions about their views. Th e fi rst is a question of motivation. Given that both King and Soames want to sup- ply propositions with representational properties, there’s a pretty clear sense in which both theories say more than mine. But once we have intrinsically representational propositions off the table, why should we try to come up with derivatively representational surrogates? Th is question can be brought into sharper focus by pointing out the ways in which King and Soames accept the basic ontology of the theory I (adapting the views of Chisholm, Lewis, van Inwagen, and others) provide. Neither King nor Soames expresses any skepticism about properties; on the contrary, each makes free use of properties in his account of propositions. (Each, like me, thinks of properties as among the constituents of propositions.1 ) Neither King nor Soames expresses any skepticism about our ability to have cognitive access to properties; again, each presupposes this sort of access in giving his account of propositions. So it seems that each should agree that the properties with which I identify propositions exist, and that they are the sorts of things to which we have cognitive access. Th is last point bears emphasis. Both King and Soames emphasize, in diff erent ways, the naturalistic credentials of their theories. King thinks of his account as one of naturalized propositions because it avoids saying that entities are representational independently of thinking subjects—still more naturalistic, then, is a theory which avoids saying that any entities are representational, full stop. Both King and Soames wonder how our cognitive access to propositions as traditionally conceived might be understood, and take pains to explain how we might have access to the entities with which they identify propositions. But in giving this account, both presuppose rather than explain our cognitive access to properties—hence it’s hard to see how it could be easier to have cognitive access to the entities with which they identify propositions than to

1 Th ough see Ch. 11 for some further discussion of the meaning of “constituent” in this context.

King020513OUK.indd 148 11/23/2013 12:59:01 PM REPRESENTATIONAL ENTITIES AND ACTS 149

the properties which I take propositions to be. It may well be that our cognitive access to properties is something which needs explanation; but, if so, this is a challenge for my theory no more than for those of Soames and King. I suggest, then, that the ontology of the theory I prefer is a proper subset of the ontologies of the theories of King and Soames, and our cognitive access to the entities with which I identify propositions can be no harder to understand than our cognitive access to propositions as conceived by King and Soames. It then seems to me that if we are to prefer their theories to the property theory, we must think that their theories explain something which my admittedly sparser theory cannot. What I don’t quite see is what this crucial explanandum could be. A second sort of question for theories of the kind that Soames and King provide can be presented by way of a dilemma. If propositions have representational properties, then either they have them essentially, or they don’t. But there is trouble either way. If propositions have their representational properties essentially, then it is hard to see how any theory could explain why they have those properties. Th is is just because there seem to be limits on the sort of explanation we can give of why something has a property, if it has that property essentially. Suppose that my desk—call it “Fred”—is essentially wooden. Given this, can we explain why Fred is wooden? One might well think that we can’t. Or, at least, we can’t explain this in the same way that we can explain why a book is sitting on a particular table—aft er all, Fred simply couldn’t exist without being wooden, and so we can’t recount events in the history of Fred’s existence which led to its woodenness. Th at said, there is, of course, another sense in which we can explain how Fred came to be wooden—and that is just that we can explain how Fred came to be. But, given that Fred could not exist without being wooden, this is the only sort of explanation of Fred’s woodenness which we can give. Hence, on the view that propositions have their representational properties essentially, it seems that the only explanation we can give of their possession of those properties is an explanation of how the relevant proposition came to be. But one might think that giving an explanation of why a proposition exists is not the same thing as giving an explanation of its possession of certain representational properties. Th is is best brought out by thinking about an example. Consider a proponent of propositions as traditionally conceived: someone who thinks that propositions have their representational properties independently of anything that anyone thinks or does. Such a person could, it seems, think of propositions as structured and he could also think that some propositions—singular propositions— have contingently existing objects as constituents. Refl ecting on these commitments, our traditional proposition theorist might also come to believe that singular propositions can’t exist unless their constituents do. So far, it seems, none of these commitments force our imaginary theorist—who of course has a lot in common with many actual philosophers—to give up her traditional view of propositions, according to which “propositions have truth conditions by their very natures and independently

King020513OUK.indd 149 11/23/2013 12:59:02 PM 150 JEFF SPEAKS

of minds and languages” (63) and “propositional representation is primary and the agent’s representation is to be explained in terms of it” (136) But now consider a singular proposition about some artefact—say, the paper airplane I just made. How should our traditional proposition theorist think about this proposition? Given the commitments just sketched, she should say that by making the paper airplane I brought into existence a bunch of singular propositions with that airplane as a constituent—including, for example, the proposition that that paper airplane is white. Now — should our traditional proposition theorist say that her construction of the airplane explains the representational properties of the proposition that that airplane is white? It seems to me that she should not. She should say that her construction of the airplane explains the existence of that singular proposition, but that once it came to exist, it brought its representational properties with it; it had those properties, as it were, automatically. Nothing in the making of the airplane explained the proposition’s having those representational properties, and hence nothing in this story should force our traditional proposition theorist to stop being a traditional proposition theorist. Th is example can be used to formulate a challenge to those who want to explain the representational properties of propositions and who take the fi rst horn of our dilemma, saying that propositions have their representational properties essentially. Let E be something proposed as the explanation for the representational properties of the proposition that that airplane is white. We should agree that my making the paper airplane does not explain the representational properties of this proposition. So we should also agree that if E is to provide a genuine explanation of this proposition’s representational properties, it must bear some relation to that proposition’s having its representational properties which my act of making the paper airplane does not. (So it can’t just be the relation of bringing that proposition into existence.) Th e challenge is to say what this relation could be. If one thinks that it’s hard to see what this relation could be, this might push us toward the second horn of the dilemma, according to which propositions don’t have their representational properties essentially. Th is view fi ts much more easily with the idea that we can explain the representational properties of propositions: if they don’t have those properties essentially, then it is no harder, in principle, to understand how we could imbue them with these properties than it is to explain how we could paint a fence white. But this view faces other problems. If propositions don’t have their representational properties essentially, then it is possible for some proposition to exist, and either lack representational properties altogether, or to have diff erent representational properties than it actually has. But (if propositions have representational properties) there is a presumably necessary connection between a proposition’s representational properties and its truth conditions. Hence if the proposition that grass is green does not have its representational properties essentially, it could presumably exist without being true iff grass is green. But this leads to very odd results; in particular, it seems to make claims like the following true:

King020513OUK.indd 150 11/23/2013 12:59:02 PM REPRESENTATIONAL ENTITIES AND ACTS 151

Possibly, John believes that grass is green, and grass is green, but John’s belief is not true. Possibly, grass is green, and the proposition that grass is green is false (or exists and lacks a truth-value).

Th is seems unacceptable. 2 Both Soames and King opt for the fi rst horn of the dilemma: both think that we can explain the representational properties of propositions despite the fact that propositions have those properties essentially. Th e discussion of this horn of the dilemma above gives us an initial reason to doubt that such explanations can succeed. In what follows, I’ll defend this claim, by turning to the specifi cs of the explanations which King’s and Soames’s theories provide.

K i n g ’ s explanation of the representational properties of propositions On King’s theory, propositions are facts. To follow his example, let’s consider the proposition that Michael swims, which he names FAST . Th is fact includes Michael, the property of swimming, and the propositional relation corresponding to the open sentence

[PR] there is a context c such that x is the semantic value (relative to c and assignment g) of a lexical item e of some language L and y is the semantic value (relative to c and assignment g) of a lexical item e’ of L such that e occurs at the left terminal node of the sentential relation R that in L encodes ascription and e’ occurs at R’s right terminal node. 3

A fi rst pass at King’s view then identifi es the proposition that Michael swims with the fact obtained by assigning Michael as value to the free variable “x,” and the property of swimming as value to the free variable “y.” As noted above, King denies that propositions like this have representational properties on their own; rather, we do something to give them these representational properties. So what is it that we do? Th e central thing that we do is that we interpret the propositional relation [PR] as encoding ascription; as King puts it, “FAST has truth conditions because speakers interpret its propositional relation as ascribing the property of swimming to Michael.” But, one might wonder, how do we manage to interpret propositional relations?

2 One point worth noting here: above I’m using “essential property” as equivalent to “property a thing has necessarily.” But it is plausible, following Fine (1994) and others, that we can understand the essential properties of a thing as a proper subset of the properties that it has necessarily; and perhaps we can explain why a thing necessarily possesses certain properties in terms of its essential properties. Given all of this, one might try to escape the dilemma above by saying that propositions have their representational properties necessarily but not essentially, and that the essential properties of propositions can explain their representational properties. Th is seems to me like a promising general strategy; but I’m not quite sure how to make it work in the context of Soames’s or King’s theory. Th anks to Ben Caplan for helpful discussion of this point. 3 As King notes, we might think of propositional relations as either involving specifi c sentential relations, or as involving existential generalization over sentential relations. I prefer the second version of the theory, which is what I use in what follows—but nothing hangs on this choice here.

King020513OUK.indd 151 11/23/2013 12:59:02 PM 152 JEFF SPEAKS

King’s answer is that speakers interpret propositional relations by interpreting sentential relations like concatenation, which fi gure in those propositional relations; as he says, “my explanation of why speakers interpret the propositional relation of the proposition that Michael swims in the way they do appeals covertly to the way in which they interpret the sentential relation in the sentence ‘Michael swims.’ ” (King, p. 21; emphasis in original) Th is, in turn, leads to the question of what it is for a language user to interpret a sentential relation in a certain way. And the answer to this question, King says,

“is simply that we compose the semantic values at the terminal nodes of the proposition in the way we do” (13)

Here we’ve reached the “fundamental mental states” mentioned above: they are a matter of speakers composing the semantic values of simple expressions in a way which results in their assigning truth conditions to the relevant sentences. Note that for King’s view to be plausible, we must think of the interpretation of sentential relations—i.e., the composing of certain semantic values of expressions standing in certain syntactic relations—as not just explaining how the relevant propositional relation gets interpreted, but also as metaphysically suffi cient for the interpretation of that propositional relation. 4 For suppose that it were not. Th en a speaker could interpret the syntactic relation in “Michael swims” as encoding ascription—thereby ensur- ing that there is some language in which a term for Michael is concatenated with a term for swimming, and in which concatenation encodes ascription—without the proposition that Michael swims existing. (Since this proposition, on King’s theory, could not exist unless [PR] encodes ascription.) But this would be absurd. 5 In what follows, I want to discuss three sorts of worries about this theory: (1) that propositional relations necessarily have certain representational properties, so their possession of these properties can’t be explained by what speakers do; (2) that the best

4 Hence I think that we can’t say, as King (2007), 60 does, that we can explain the fact that “the propositional relation inherits its signifi cance from the sentential relation” in terms of “something that we and our linguistic ancestors did.” (Unless it was not just something that we’ve done, but something which it was metaphysically impossible for us not to do.) 5 Th is point is also relevant to a distinction between two diff erent ways in which King sometimes states his theory. King sometimes suggests that the facts with which he identifi es propositions include this fact about how the relevant propositional relations are interpreted. (See King Ch. 4 p, 9, and King (2009), 265.) Th is suggests that FAST is not the fact (a) [PR](Michael, swimming) but rather the more complex fact (b) [PR](Michael, swimming) & [PR] encodes ascription In what follows, I’ll ignore this complication in King’s view, and discuss the simpler version (a) of the theory. I don’t think that, in the end, the diff erence between these versions of the view matters much. Th is is because, given the way that King defi nes the relevant notions, it is impossible for (a) to obtain without (b), or for (b) to obtain without (a). Th e second direction is trivial; the fi rst holds because, by the argument just given, interpreting the relevant sentential relation as encoding ascription must be metaphysically suffi cient for also encoding the propositional relation involving that sentential relation. But then since [PR] could not hold between Michael and swimming unless R was interpreted as encoding ascription, it follows that, necessarily, if (a) is a fact, then (b) is as well.

King020513OUK.indd 152 11/23/2013 12:59:03 PM REPRESENTATIONAL ENTITIES AND ACTS 153

understanding of the nature of interpreting sentential relations leads to circularity; and (3) that there’s an obstacle to generalizing King’s theory to provide an account of the contents of perceptual states. 1. Consider the claim that [PR] encodes ascription. Let’s suppose that this is true. Is it a necessary truth, or a contingent one? [PR] is a “pure” relation, whose constituents include no contingently existing concrete objects, and hence it presumably exists necessarily. So for the above claim to be contingent, there must be some world at which either (i) [PR] exists and encodes something other than ascription, or (ii) [PR] exists and encodes nothing at all. We can rule out (i). If (i) were true, then speakers would presumably have to interpret [PR] as encoding something other than ascription, and their doing so would involve the interpretation of some sentential relation R. If they interpreted R as encoding ascription, they would also (by the argument above) so interpret [PR]. And if they interpreted R as encoding something other than ascription, then they would be interpreting some propositional relation other than [PR], since the latter is defi ned partly in terms of sentential relations which encode ascription. (ii) is less easily ruled out. So let’s leave it open. We can still conclude that the following is a necessary truth:

If [PR] encodes anything, then it encodes ascription. Hence it is also a necessary truth that

If [PR] relates anything, then the fact that it relates them is true iff the fi rst instantiates the second. So propositional relations are necessarily existing entities which necessarily have representational properties—like the property of being such that for any x and y, the fact that they relate x and y is true iff the fi rst instantiates the second, and is about x and is about y. Because these are necessary truths, they are not under the control of speakers, and speakers do not bring them about. But at this stage, one might worry that propositional relations are the sorts of things whose existence King is concerned to deny—entities which have their representational properties “by [their] very nature and independently of minds and languages.” King might reasonably, and correctly, reply that on his account there are no propositions, and there is no truth and falsity, without speakers doing something to bring it about that there is. I’m a bit skeptical that this reply is enough to allay the worry that propositional relations are a bit too much like the primitively representational propositions King wants to do without. One way to bring out the reasons for this is to focus on just what speakers do to bring truth and falsity into the world. Th e role played by speakers is simply to make it that case that [PR] relates some things; once they have brought into existence a fact of the form [PR](x,y) there’s no further work for them to do in giving this fact its representational properties. But, given this, it seems like the role played by speakers on this theory is uncomfortably close to the role played by our paper airplane maker in the example discussed above.

King020513OUK.indd 153 11/23/2013 12:59:03 PM 154 JEFF SPEAKS

We can focus this worry in the way mentioned above, by asking: what relation does the interpretation of [PR] bear to the representational properties of FAST which the making of the paper airplane does not bear to the proposition that that paper airplane is white? It seems to me that this is a diffi cult question for the proponent of King’s view to answer; and this, in turn, suggests that we are getting no more of an explanation of the representational properties of propositions than is provided by the making of the paper airplane. King might reply that the crucial diff erence here is a diff erence between the relevant action types—making a paper airplane, on the one hand, and interpreting a propositional relation, on the other. Making a paper airplane, he might object, simply isn’t the sort of action which can explain the representational properties of the many singular propositions involving the airplane—interpreting a sentential and a propositional relation, however, is. But this focus on exactly which actions of language-using subjects do the relevant explanatory work leads to another, related sort of worry about King’s view. In order to avoid appeal to merely possible contexts, King allows that it is suffi cient for an arbitrary singular proposition that o is swims to exist that there be some syntactic structure

x swims

in English, where the “x” is a variable which has a semantic value only relative to an assignment. For any object o, there will be, aft er all, some assignment which assigns o as the value of “x.” But that will be suffi cient for [PR] to hold between o and the property of swimming, which will be suffi cient for the relevant singular proposition to exist. But this means that the fi rst time someone used a variable expression, it was—metaphysically speaking—a much more momentous event than one might have thought. For this event brought a host of representational entities into the world: the infi nitely many singular propositions which predicate swimming of something. In one sense, this is nice for King’s view: it secures the existence of a great many propositions, and at very little cost. But this virtue has a corresponding vice, and rein- forces the worries raised in connection with the example of the paper airplane above. Th e worry is that we are getting too much for free from the nature of [PR]—to which no thinking subjects contribute anything—to get a satisfying explanation, in terms of thinking subjects, of the representational properties of propositions. We can push the point one step further. As King notes (note 11), natural languages plausibly contain expressions which are best treated as variables over properties. But then we could presumably have a sentence in a natural language, like “Something is some way” which involves the syntactic structure

King020513OUK.indd 154 11/23/2013 12:59:03 PM REPRESENTATIONAL ENTITIES AND ACTS 155

where “x” is a variable over objects and “y” a variable over properties. But could uttering “Something is some way” really suffi ce to explain the representational properties of every singular monadic predication? 2. Th ese have all been worries about the extent to which King’s theory explains rather than assumes the representational properties of propositions. Let’s turn our focus now to the nature of the fundamental mental state which is supposed to be doing the explaining: speakers’ interpretation of sentential relations. Our understanding of the interpretation of sentential relations is sharply constrained by King’s explanatory ambitions. Because the interpretation of sentential relations is supposed to explain the existence and representational properties of propositions, interpreting a sentential relation cannot be, or presuppose, any propositional attitudes. Th is constraint seems to be inconsistent with some things King says about the interpretation of sentential relations. Recall that the interpretation of sentential relations is simply a matter of speakers composing semantic values in the way that they do. Th is is supposed to explain how certain syntactic relations acquire semantic signifi cance, which in turn explains how propositional relations and propositions acquire representational properties. But sometimes King talks as though the explanatory order runs in the other direction, and that speakers compose semantic values in the way that they do because the relevant syntactic relations have a certain semantic signifi cance. He says, for instance, that

“Speakers work their way up the propositional relation composing semantic values in the way they do, thus interpreting the propositional relation, because it is the way the semantic signifi - cance of the syntax dictates that they compose these semantic values.” 6

But this can’t be right. If interpretations of sentential relations are supposed to be what brings the properties of sentential relations into the world, we can’t in general understand interpretations of sentential relations to be responses to those very properties. King might, without serious cost, simply take back those remarks about the interpretation of sentential relations. But there’s another circularity worry in the vicinity which, I think, can’t be as easily dismissed. Th e interpretation of sentential relations is supposed to be a matter of speakers composing the semantic values of the terminal nodes of a propositional relation. But that means that the interpretation of sentential relations can only get off the ground if those terminal nodes have a semantic value. But this looks, from the point of view of King’s theory, very odd. If the interpretation of sentential relations is supposed to be what brings representational properties into the world, how can we understand the interpretation of sentential relations in terms of an antecedent assignment of semantic values to expressions? If we’re reluctant to take as primitive the representational

6 King, p. 86. See also pp. 77–78: “Th at is, what is it that we do that amounts to our so interpreting it? It is simply that we compose the semantic values at the terminal nodes of the propositional relation in the way we do. In the end, this is just a refl ex of the sentential relation R having the semantic signifi cance it does.” Better, I think, for King to say that R’s having the semantic signifi cance that it does is a refl ex of the way we compose the relevant semantic values.

King020513OUK.indd 155 11/23/2013 12:59:04 PM 156 JEFF SPEAKS

properties of propositions, we can hardly be content with taking the representational properties of linguistic expressions as primitive. Th ere are a few diff erent ways to press this worry. One is the simple point that any theory which aims to provide a general explanation of the instantiation of representational properties must provide an explanation of facts about the possession of semantic values by simple natural language expressions. King’s theory cannot provide such an explanation, since the fundamental mental states which are doing the explaining are a matter of language users deriving the semantic values of complex natural language expressions on the basis of the semantic values of the simple ones. Th is shows, I think, that King’s theory is incomplete. Once we have an account of the semantic values of simple expressions we might try to use King’s theory to ascend from there to an account of the semantic values of complex expressions and the representational properties of propositions, but we need to look elsewhere for the fi rst, crucial step. Th is might lead us to think that even if King’s theory can’t do all of the work it sets out to do, it might at least do some. All we need to do, on this view, is to look for an explanation of the representational properties of simple linguistic expressions and conjoin that with King’s account. And surely, one might think, the fact that we need to do this is no objection to King’s account as such. Th e representational properties of propositions are one thing, and the representational properties of simple expressions another; and King’s aim, one might think, was just to explain the former. Th e problem, though, is that if we are motivated by King’s guiding thought that it is unacceptably mysterious for something to have representational properties “by their very natures,” we have to fi nd some way to explain the representational properties of linguistic expressions as well; and it is hard to see how we can do this without making use of propositions in a way which the proponent of King’s account cannot accept. Th e possession of semantic values by simple expressions is, on King’s view, explanatorily prior to speakers’ interpretations of sentential relations; hence it is also explanatorily prior to the propositions whose existence are explained by those interpretive acts. Th is places a very strong constraint on our account of the facts in virtue of which expressions like names and predicates come to have the semantic values that they do: our account won’t be able to make use of propositional attitudes like the beliefs and intentions of language users. But virtually all of the most well worked-out theories of the facts in virtue of which expressions have the contents they do make use of propositional attitudes of this sort.7

7 I’m thinking of accounts in the spirit of Lewis (1975) and Schiff er (1972). One way around these problems would be for King to take a diff erent view of the interpretation of sentential relations, and understand it in terms of language users taking certain sentences to have certain truth conditions. Th is has the advantage that it does not obviously make use of explanatorily prior facts about the assignments of semantic values to the terminal nodes of certain logical forms. Th e disadvantage, though, is that “taking sentences to have certain truth conditions” seems itself to be a propositional attitude.

King020513OUK.indd 156 11/23/2013 12:59:04 PM REPRESENTATIONAL ENTITIES AND ACTS 157

3. Th is worry leads into the question of how, on King’s view, we should understand the mental states of subjects who existed prior to the existence of natural languages, and hence (one might think, given the above story) prior to the existence of propositions. Do we really want to deny that such subjects had beliefs, desires, and perceptual experiences, each of which plausibly involves a relation to a proposition? King is well aware of this problem, and tries to make room for the existence of such subjects. In particular, he fi nds plausible the idea that, prior to the existence of languages like English, organisms enjoyed perceptual experiences which had propositions as their contents. King further suggests that “there will be an account of those contents in the spirit of the present account of the contents of natural language sentences.” I agree with King’s thought that an account of propositions should be consistent with pre-linguistic but genuinely propositional perceptual experiences, but am less sure that I can see how his account meets this constraint. Let’s consider how we might adapt the preceding story to the case of perceptual experiences; to fi x ideas, let’s think about a frog’s visual experience of a fl y sitting on a green leaf, and let’s simplify by supposing that the content of this visual experience is a singular proposition which predicates the property of being in a certain location (“L”) of the fl y. Presumably this, like any such singular proposition, could be expressed by a sentence. Hence, on King’s view, the content of the frog’s experience will then be the proposition expressed by such a sentence, i.e. the fact that the fl y and L stand in the propositional relation [PR]. Our question is: what did the frog do to bring this proposition into existence? I take it that the frog must have done something to interpret the propositional relation [PR]. But this seems to show that we need to change our view of what the relevant propositional relation is—aft er all, the frog isn’t interpreting any sentential relations. And it won’t do to say that [PR] works fi ne for the propositions expressed by sentences but that we need to invoke some other propositional relation for mental states, since it is quite plausible that some propositions can both be expressed by a sentence and be the content of a mental state, like a perceptual experience. I think that the best thing for King to say here is that even if the frog does not interpret any sentential relations, he does do something similar: he interprets his experience. Th is suggests that we should then revise our view of propositional relations like [PR] so that they will be as applicable to the case of interpretations of experiences as to the case of interpretations of sentences. Th is is obviously non-trivial. But let’s suppose that we’ve done it. A problem still remains, which has to do with a fundamental diff erence between sentences and perceptual experiences. Presumably, the phenomenal character of the frog’s experience is fi xed independently of the frog’s interpretation of his experience; were it not, it’s hard to see what the frog could be interpreting. But according to a plausible intentionalist thesis, phenomenal properties are identical to certain representational properties, which involve relations to propositions. Hence it seems that if the phenomenal character of the frog’s experience is fi xed independently of his interpretation of his experience, the content of his experience must be as well.

King020513OUK.indd 157 11/23/2013 12:59:04 PM 158 JEFF SPEAKS

But if this is right, and the analogue of interpretation of sentential relations plays no role here, we lose our explanation of how the fl y and L come to stand in whatever the relevant propositional relation turns out to be. And then we lose our explanation of the existence of the relevant proposition, as well as of its representational properties. To be sure, King does not pretend to a fully worked out theory of the contents of perceptual experience, and it may be that his story can be told in some way other than the way I’ve just argued to be objectionable. 8 But I do think that the disanalogies between perceptual experiences and sentences cast some doubt on the idea that King’s theory can be smoothly generalized—as it surely must be, if it is to be acceptable—to the case of perceptual experience.

S o a m e s ’ s explanation of the representational properties of propositions For Soames, the fundamental mental states are not interpretations of sentential relations but rather token mental events of predication. What are these, and how do they explain the representational properties of propositions? Here’s the way Soames introduces the notion:

“ Th ink again about the proposition that o is red. It is the content of an occurrent perceptual or cognitive state whenever the agent predicates being red of o. Whenever an agent does this, a concrete event occurs, at a specifi c time and place, in which the agent predicates this property of that object. Th is suggests that the proposition that o is red is simply the minimal event type in which an arbitrary agent predicates being red of o.” (136) Consider the token event of an agent judging that o is red. Th e idea is that token events of this sort are always underwritten by distinct token events of predicating redness of o. And what’s true of judgement also holds for a wide variety of mental events—including entertaining the proposition that o is red, accepting that o is red, asserting that o is red, denying that o is red, judging that o is red, and visually representing o as red. Each token of these event types is accompanied by a distinct token event of predicating redness of o. Th e primitive representational facts are facts about the representational properties of these token mental events of predication. Propositions are types of such events, and inherit their representational properties from the representational properties of their tokens. I’d like to raise four sorts of questions about this account: (1) questions about whether we have reason to believe that the relevant token events exist; (2) questions about their

8 One idea would be to say that the frog interprets his own brain state rather than the phenomenal character of the experience, and that the connection between that brain state and its content is contingent rather than necessary. Maybe something like this would work, but there are at least two worries. One is the sheer implausibility of the idea that frogs interpret their own brain states in the way that we interpret sentences. Th e second and more fundamental is that, even if the connection between the underlying brain state and its content is contingent, this does not imply that it is mediated by the interpretation of the frog. More plausibly, it is fi xed by facts about (counterfactual and actual) causal relations in which that brain state stands, which are fi xed independently of the frog’s interpretation.

King020513OUK.indd 158 11/23/2013 12:59:04 PM REPRESENTATIONAL ENTITIES AND ACTS 159

frequency; (3) questions about how, if they have these properties necessarily, these representational properties of propositions could be explained by anything we do; and (4) questions about why, given that types don’t invariably inherit the properties of their tokens, propositions should inherit the representational properties of theirs. 1. On reading Soames’s description of the fundamental events of predication, one might be tempted to reply Hume-style, that “when I enter most intimately into what I call my mental states, I always stumble on some particular propositional attitude or other, a judgement or an assertion or an experience. I never can catch an event of predication, and never can observe anything but the attitudes.” When I think about various propositional attitudes—events of visually representing that o is red, judging that o is red, and asserting that o is red—I really don’t notice an event of predication which accompanies each one. But if they are always accompanied by such an event, what explains this? One might reply that I do notice the event of predication—it’s just that I don’t notice it as such. Aft er all, I do notice the similarity in content between these diff erent mental states—and isn’t this, on his view, just to notice the events of predication? Not quite: noticing sameness of content, on Soames’s view, is noticing sameness in associated event types (contents being identical to event types). But to see that various mental states are all relations to the same event type is not to notice the occurrence of tokens of that type. A diff erent and more promising reply is is suggested by some remarks Soames makes when explaining predication: he might say that every occurrent propositional attitude is accompanied by a certain basic propositional attitude: roughly, the attitude of “calling the proposition to mind.” Soames calls this attitude entertaining. Perhaps we could simply identify the event of coming to have this propositional attitude—entertaining the proposition that o is red—with the token event of predicating redness of o. Soames says things which suggest this; for instance, he says that “to entertain the proposition that o is red is to predicate redness of o.” 9 If such an identity claim were true, this would make the existence of the token events of predication much less dubious—for, in each of the cases discussed above, we obviously do call the relevant proposition to mind. But it is hard for me to see how this identity claim could be true. For suppose that token events of predication were identical to token events of a subject coming to entertain a proposition. Th en, on Soames’s view, a certain token event e would be identical to a token event e* of a subject coming to stand in a relation to an event type E of which e is a token. But if token events x, y are identical, then any type of which x is a token must also be a type of which y is a token. But one type of which e is a token is the proposition that o is red. It then follows that e* must be a token of the proposition that o is red. But then propositions are event types with the surprising property that each of its token events is some subject coming to stand in a relation to that type.

9 136–137. See also Soames (2010), 81–82.

King020513OUK.indd 159 11/23/2013 12:59:04 PM 160 JEFF SPEAKS

Th is is worrying for a few reasons. First, it leads to troubles understanding what the view says. Soames claims that the proposition that o is red is a certain event type E. If we want to know what E is, we naturally look to its tokens. But when we look to its tokens we fi nd that they are events of subjects coming to bear a certain relation to E,which leaves our question about the nature of E unanswered. Th is regress is not obviously vicious. But it does, I think, make trouble for some of the explanatory claims that Soames seems inclined to make. Soames is inclined to say that the representational properties of propositions are explained by the representational properties of token events of predication, and is at least open to the idea that the existence of the relevant propositions is also explained by the existence of the relevant token events of predication. But it is hard to see how this is going to work if the token events of predication are identical to events of subjects coming to stand in relations to the propositions whose existence and representational properties the token acts of predication were supposed to explain. A further, related worry comes from refl ection on the propositional attitude of entertaining a proposition, on the supposition that events of coming to stand in this attitude are identical to events of predicating properties of objects. Th ese events of predicating properties of objects are supposed to have their representational properties intrinsically—in particular, on pain of circularity, they are not supposed to have their representational properties explained in terms of the representational properties of any proposition. But since these events of predication are identical to events of coming to be in a certain propositional attitude state—entertaining the relevant proposition— it follows that at least one propositional attitude state also must have its representational properties intrinsically, and to be such that its representational properties are not explained by the representational properties of any proposition. But this raises a question. If we are willing to say this about one propositional attitude state, why not say this for all of them? Why not—so far as the explanation of the representational properties of mental states are concerned—do away with the primitive acts of predication entirely, and let each of the familiar propositional attitudes—belief, assertion, etc.—be intrinsically representational states, whose status as representational is not explained by the representational properties of the propositions to which they are relations? 10 Now nothing in his theory forces Soames to identify predicating redness of o with entertaining the proposition that o is red; his statements to the eff ect that coming to be in the second state just is a matter of coming to be in the fi rst state are also consistent with holding that the two states (and events of coming to be in those states) are

10 One might say this while still giving the proposition that grass is green a role in determining exactly which representational property one has when one believes that grass is green. Th is is of course analogous to what Soames says about predication: the objects of the predication — in our example, o and redness — are not themselves representational entities, but the objects of a given event of predication obviously play a role in determining the representational properties of that event. More on this sort of account of the attitudes in Ch. 11 below.

King020513OUK.indd 160 11/23/2013 12:59:04 PM REPRESENTATIONAL ENTITIES AND ACTS 161

distinct, but closely related—perhaps each is metaphysically necessary and suffi cient for the other, and the fi rst in some sense explains the second. But this leads us back to the Hume-style worries sketched above. We’re then postulating a primitive mental state type which, of necessity, accompanies all of our occurrent propositional attitudes, but with which we are not acquainted. Our best hope for establishing our familiarity with predication, aft er all, was to identify it with the propositional attitude of entertaining a proposition. Indeed, the basic worry here needn’t be put in quite such a Humean way. Even if we’re not worried about the fact that we never notice these events of predication, one might legitimately wonder why is it better to posit a primitively representational mental event with which we have no immediate acquaintance than to say that belief, judgement, assertion, et. al., with which we are acquainted, are primitively representational in just the way that Soames thinks that predication is. (For that matter, a traditional proposition theorist might wonder at this point why it is better to posit a primitively representational mental state than to posit primitively representational propositions.) Th e problem here is not just that predication is taken as primitive; every theory will, aft er all, have some primitives. As Soames points out, negation is plausibly a primitive notion; this hardly shows that we don’t understand negation. 11 Th e problem is that predication is both a type of mental state with which we are not acquainted, and taken as primitive. Th e fi rst fact makes it legitimate to ask for more information about what predication is; the second fact raises the worry that we will not get a satisfactory answer. 2. Let’s set aside these questions about the nature of events of predication and instead ask a question about their frequency. Do they accompany every event of a subject coming to be in a propositional attitude state, or just some such events? I think either way we answer this question, we run into diffi culties. Soames’s answer to this question is that events of predication accompany only some such events; events of subjects coming to be in occurrent mental states. And there’s a good reason for taking this line. If we said that believing that o is red requires an event of predicating redness of o, this would make events of predication truly mysterious. Aft er all, I presumably have plenty of beliefs which I have never actively called to mind; surely I can simply passively believe that Indiana is not a grapefruit without ever having carried out the relevant predications. But Soames is also inclined to think that, at least for simple propositions like the proposition that grass is green, propositions do not exist unless there are token events of predication of the right sort—in particular, the proposition that grass is green does not exist in w unless at least one person in w has predicated greenness of something, and at least one person in w has predicated something of grass. Nothing in the view that propositions are event types requires us to adopt this claim about the existence conditions of event types. But it would be hard to give up this view

11 Soames (2010), 29.

King020513OUK.indd 161 11/23/2013 12:59:04 PM 162 JEFF SPEAKS

about the existence conditions of propositions, given Soames’s interest in explaining the representational properties of propositions. Aft er all, if event types did exist necessarily, then propositions would exist, and have all of their representational properties, whether or not there were any thinking subjects around to do any representing. And if this were true, it would be hard to see how Soames’s theory could live up to the guiding ideas

“(i) that the perceptual and cognitive activity of agents is the conceptual basis of all representation and (ii) that propositions are representational in virtue of the relations they bear to this representational activity.” (8)

What role could agents play, if all of the propositions existed and had all of their representational properties, no matter what they did? 12 Th e problem is that this pair of commitments—that belief does not require predication and that existence does require certain facts about predication—invalidates some intuitively valid inferences. Consider some property which could be such that no one has ever predicated of anything—say, the property of being 726-sided. Can someone in the relevant world still have beliefs involving this property? It seems that they can; in particular it seems that they can believe that no circles are 726-sided without having ever predicated the property of being 726-sided of anything. Now consider the inference:

J e ff believes that no circles are 726-sided. Scott believes that no circles are 726-sided. ——————————————————— Th ere’s something that Jeff and Scott both believe.

Th is certainly seems to be valid. Indeed, as noted in Chapters 1 and 2, it’s the sort of trivially valid inference oft en used to defend the existence of propositions. But on the present theory it comes out invalid—even though Jeff and Scott both bear the belief relation to the proposition that no circles are 726-sided, this proposition need not exist, and hence there is no guarantee that there is something that Jeff and Scott both believe.13 To my mind, this consequence of the theory is made more worrying by the fact that superfi cially quite similar inferences, like

J e ff asserts that no circles are 726-sided. Scott asserts that no circles are 726-sided. ——————————————————— Th ere’s something that Jeff and Scott both assert.

12 Th ough, as mentioned in note 2 above, one might try to answer this question via a theory according to which some sorts of necessary truths are explained in terms of others. 13 One might think that there is an interpretation on which this argument is, even on Soames’s view, valid: namely one in which we interpret the existential quantifi cation in the conclusion as possibilist. But if we force an actualist interpretation of the quantifi er, by changing the conclusion to “there is actually. . .,” and changing the premises from “believes” to “actually believes,” the argument seems no less valid.

King020513OUK.indd 162 11/23/2013 12:59:04 PM REPRESENTATIONAL ENTITIES AND ACTS 163

come out valid—aft er all, assertion, being an occurrent mental state, does require an event of predication, and hence entails the existence of the relevant propositions. Th e fact that we are forced to treat the “believes” inferences diff erently than the “asserts” inferences seems to me to be a cost of the view. 3. Setting aside these worries about the validity of certain inferences, there’s a more purely metaphysical worry arising from this discussion of belief and predication, which is just that Soames’s theory is committed to the claim that subjects can believe propositions which do not exist, and never did. Th is is a commitment of which Soames is well aware. But it does show that the view that propositions as event types is one which no philosopher committed to serious actualism—the thesis that nothing can have a property or stand in a relation in w without existing in w —should accept. But rather than focusing on the plausibility of denying serious actualism, I’d like to focus on its consequences for Soames’s explanatory claims. While Soames does not of course think that propositions can have just any property without existing, he does think that they can have quite a few: not just being believed, but also being true or false, and representing the world as being a certain way. But one might think that this reinstates the worry above—which echoes one of the objections to King’s view—that the role actually played by thinking subjects on Soames’s view is thinner than it might at fi rst seem. Consider a barren world w , with no subjects doing any thinking or talking. As far as I can tell, on Soames’s view, every proposition has all of its representational properties in w : the proposition that grass is green is, in w , about grass and about greenness; in w , it represents grass as green; it w , it is true iff grass is green. No subjects inw are to thank for the proposition having these representational properties, there being no such subjects to thank. How can this be squared with Soames’s claim that propositions are representational in virtue of the relations that they bear to representational activity? Th ere are really two problems here. Th e fi rst is the problem of explaining the fact that the proposition that grass is green has representational properties in the barren world w . Th e second is about whether the fact that propositions have representational properties in w impugns our explanation of the representational properties of propositions in the actual world. Let’s focus fi rst on what explains the representational properties of the propositions in w . I think that the best thing for Soames to say here is that propositions can have representational properties in a barren world like w because certain counterfactual claims about those propositions are true in w : in particular, the proposition that grass is green is about grass in w because it is true in w that were some subject to, for example, judge that grass is green, that judgment would involve predicating greenness of grass. But here it seems to me that we risk reversing Soames’s preferred order of explanation. Setting aside worries about the existence of acts of predication, I don’t doubt that this counterfactual is true in w . What I wonder is what makes it true. It seems to me

King020513OUK.indd 163 11/23/2013 12:59:04 PM 164 JEFF SPEAKS

that something about w must explain the truth of this counterfactual—but it’s hard (in the absence of any actual predications of the right sort) to see what could explain its truth other than facts about the representational properties of the proposition that grass is green. But we can’t both explain the representational properties of propositions in terms of certain counterfactuals, and explain the truth of those counterfactuals in terms of the representational properties of propositions. Now return to the second problem, about propositions in the actual world. What role did the activity of we, the thinking and talking subjects of the actual world, have to play in explaining the representational properties of propositions? Th e preceding remarks about w put some constraints on the way in which we answer this question: we can’t say that, were it not for our thought and talk, these propositions wouldn’t have those representational properties; aft er all, they have them all in our barren world w. Our contribution is only this: that we brought them into existence. But this seems to me a less signifi cant achievement when we know that they already had all of their representational properties before they came to be. Th is cluster of worries is obviously analogous to the the fi rst objection to King raised above. Th e situation for Soames’s view seems to me in one way better and in one way worse than King’s theory. Soames’s view avoids the problem that saying “Something is some way” brings into existence every singular monadic proposition; and, in this way, he manages to secure a closer, and potentially more explanatory, relation between the fundamental mental states and the propositions whose representational properties they are supposed to explain. But this virtue is connected to a corresponding vice. By making the connection between the fundamental mental states and propositions more demanding, Soames makes it harder for propositions to exist; which in turn makes it more tempting to say that propositions have all of their representational properties whether or not they exist, and no matter what we (actually) do—which casts doubt, from a diff erent angle, on the central explanatory claim. 4. Let’s set these questions aside, and grant for purposes of argument that there are token events of predication of the sort Soames describes, that they occur when the theory says they do, and that their having representational properties is explained by the nature of predication rather than by any built-in representational properties of its objects. Th ere are still residual worries about how these acts of predication could explain the representational properties of propositions. Soames explains the relation between the two succinctly:

“ Th is suggests that the proposition that o is red is simply the minimal event type in which an arbitrary agent predicates being red of o. Th is event-type is representational because every conceivable instance of it is one in which an agent represents something as being a certain way.” (136) Th is seems straightforward: there are event tokens which are inherently representational, and the event types of which these are tokens—the propositions—inherit their representational properties from those token events.

King020513OUK.indd 164 11/23/2013 12:59:04 PM REPRESENTATIONAL ENTITIES AND ACTS 165

But, as Ben Caplan has pointed out (in private communication), this is not quite as straightforward as it might at fi rst sound. Aft er all, it is not in general true that event types inherit the properties of tokens of that type. A token event of predication has the property of being a token event, whereas the corresponding type does not; a token event of eating dinner must have a certain duration, whereas the type does not. So if types do not in general inherit the properties of their tokens, why should we think, as Soames encourages us to, that the representational properties of token events of predication explain the properties of propositions? Soames might reply that representational properties are special: perhaps oft en event types don’t inherit the properties of their tokens, but they do when the relevant properties are representational ones. But this is doubly problematic. On the one hand, it seems ad hoc; why should representational properties be diff erent in this way? And, on the other hand, it seems false—even if we restrict ourselves to representational properties, event types don’t always inherit the representational properties of their tokens. To see this, consider again a token event of predicating redness of some particular object o. Suppose that here Bob performs the predication, and does so while eating an Oreo cookie. Th is token event is, it seems, an event of the following types: (i) the event type of Bob predicating redness of o while eating an Oreo. (ii) the event type of someone predicating redness of o while eating an Oreo. (iii) the event type of someone predicating redness of o. (iv) the event type of someone predicating something of o. (v) the event type of someone predicating something of something. (vi) the event type of someone doing some predicating. (vii) the event type of someone doing something. On Soames’s theory, I think, only (iii)-(v) are propositions, and hence inherit representational properties from the token event described above. But we want to know why just these get that privilege: (i), (ii), (vi) and (vii) are, aft er all, equally event types of which our token event is a token. No appeal to the specialness of representational properties, as opposed to properties like having a certain duration, will answer this question. As it stands, this is less an objection to the theory than a question to which the theory ultimately owes an answer which, at present, I don’t think it provides. Th e question is: given that types oft en do not invariably inherit the properties of their tokens, what explains the fact (if it is a fact) that propositions inherit the representational properties of their tokens? Th e foregoing examples are enough to show that this is not a question which has a trivial answer, and hence that the token/type relations to which Soames appeals in his theory are not as obviously explanatory as one might have thought. 14

14 Th anks to Ben Caplan for very helpful discussion of these issues, and for comments on a previous draft of this paper.

King020513OUK.indd 165 11/23/2013 12:59:04 PM 9

Propositions vs Properties and Facts

S c o t t S o a m e s

Propositions as properties I begin with a friendly amendment. According to Speaks (p.5) the proposition expressed by “Amelia talks” is the property: being such that Amelia instantiates talking, which is also the semantic content of the predicate “is such that Amelia talks.” Since the predicate is complex, its content should be a structured complex of the contents of its grammatical constituents. Th is suggests that the content of “Amelia instantiates talking”—namely, being such that instantiates Instantiation—should be included in the content of “Amelia talks.” Since we don’t want this, Speaks should drop instantiation , and identify the content of “Amelia talks” with being such that Amelia talks , while treating “is such that” as a syncategorematical element that turns sentences into predicates without adding an extra constituent to the content. Speaks motivates his view, in part, by noting:

“believing a proposition is taking the world to be a certain way. But if, as it seems, ‘ways things are’ are properties, this indicates that having a belief is taking a certain attitude toward a property.” (p. 6) Th ere is something right about this, but it doesn’t favor the view that propositions are properties (of the sort he has in mind). Th e claim that believing a proposition is taking the world to be a certain way approximates the more discriminating claim that believing a proposition is taking something (or things) to be a certain way (or ways). Sometimes the thing so taken is the entire universe, but oft en it’s not. To believe that o is red is to take o to be a certain way, which involves taking a certain stance toward a property. To believe the proposition is to (be disposed to) predicate redness of o, thereby representing o as red (while endorsing that predication). In this way, we can accommodate Speaks’s truism without taking the properties he identifi es to be the things believed. 1

1 My view of propositions as event types also accommodates Speaks’s truism. I here leave it open whether event types might themselves be properties of events, or of agents (if they are identifi ed with acts). Whatever

King020513OUK.indd 166 11/23/2013 12:59:04 PM PROPOSITIONS VS PROPERTIES AND FACTS 167

For Speaks, propositions are a certain kind of property, and truth is instantiation. Th is makes it diffi cult to capture the fact that truth is a kind of accuracy in representation. A map or portrait is accurate, or veridical, when it represents its subject matter as being how it really is; a proposition is true when it represents things as they really are. Th is parallel seems to be lost when propositions are identifi ed with properties that, as Speaks admits, aren’t intrinsically representational.(p. 7) Unless he can identify some sense in which they are representational, he will lose the pretheoretic connection between truth and accuracy. Just as believing that so-and-so involves taking (representing) things to be a certain way, so does doubting/denying/imagining that so-and-so. Since all are attitudes to the same proposition, there is a sense in which all involve taking things to be the same way. However, since agents who believe , deny , doubt, or imagine that so-and-so have diff erent takes on things, there is also a sense in which the ways they take things to be are diff erent. Although both senses are legitimate, only the former is at issue when one asks “Is what is believed/denied/doubted/ imagined true?” To ask this is to ask whether the way things are taken to be that is common to these cases is the way things really are. Capturing this common way things are represented to be requires either (i) taking propositions to be intrinsically representational independent of the attitudes of agents, or (ii) postulating an ur-attitude —like entertaining a proposition—that is both inherently representational and part of the characterization of the other attitudes. Th e former strategy, which Speaks correctly rejects, was that of the early Russell. Th e latter strategy is central to my theory of propositions as cognitive event types. On my view, for an agent A to entertain the proposition that Amelia talks is for A to cognitively represent her as talking. A’s act, and the event type of performing that act, represent Amelia as talking because agents who perform it do. Just as torturing someone is said to be a violent act because agents who do that are violent and events in which it is done are violent episodes, so predicating talking of Amelia, and the event type of so doing, may be said to represent Amelia as talking. From this we derive the truth conditions of the proposition. It is true at w iff at w Amelia is as she is represented to be by one who entertains it (at any world-state). If propositions are to be identifi ed with properties of the world, Speaks must provide a similar story connecting truth, representation, and the attitudes. De se attitudes are a special case. At fi rst blush, Speaks’s view that they are 2-place relations between an agent and a property might seem to hold promise in uniting the de se with the non de se . In the end, however, this seemingly promising idea founders on the evidently correct principle TB (true belief).

TB. A’s belief is true iff what A believes is true.

one decides about that, propositions in my sense are not the properties with which Speaks identifi es propositions.

King020513OUK.indd 167 11/23/2013 12:59:04 PM 168 SCOTT SOAMES

Although those properties that are objects of ordinary beliefs are, on Speaks’s account, true iff they are instantiated, this does not extend to putative objects of de se beliefs—like being Rudolf Lingens or being in danger —which aren’t truth bearers at all. Nor, as he recognizes, does the property view of propositions provide a good way of retaining the de se and the non de se as beliefs in the same sense. Instead, he is forced to posit another attitude distinct from belief in order to capture what Lewis would call belief de se. (Th e same proliferation is required for other attitudes.) As I argue in chapter 6, there is no such problem when propositions are taken to be cognitive event types. Not only is TB preserved, but the relation between the de re and the de se falls out automatically. On the cognitive account, concrete cognitive events in which an agent entertains a de se proposition are always simultaneously events in which the agent entertains the corresponding de re proposition (though the converse does not hold). From this it follows that both propositions are truth evaluable, and that believing the de se proposition guarantees believing its de re counterpart (but not vice versa). Far from facilitating a satisfying account of the de se, taking propositions to be the kind of properties that Speaks identifi es them with is an obstacle to giving such an account. At the end of chapter 5, he raises two interesting problems for his own view of propositions that extend to other theories as well. Th e demarcation problem for his theory is to specify which properties are propositions and which are not. Having identifi ed truth with instantiation of properties to which we bear certain cognitive attitudes, he needs to explain why other properties (to which we bear diff erent but related cognitive attitudes) aren’t true when they are instantiated. Aft er considering some inadequate ways of drawing the distinction, he takes the problem to remain unsolved. Th e situation is diff erent for propositions as cognitive event types. Although some open questions concerning demarcation remain, there is no fundamental problem explaining why certain event types have truth conditions and others do not. Th ose that have truth conditions are those that represent things as being one way or another, in virtue of the fact that they are event types in which agents perform acts that represent things, in part by predicating properties of predication targets. Th is is not the whole story because predication (which involves cognizing certain targets in one or another way) hasn’t been fully specifi ed, and because the range of other cognitive operations (including applying functions and performing certain function-like operations) hasn’t been exhaustively explored. I think that enough has been done to justify optimism about future progress—though, of course, the proof will be in the pudding. Speaks’s other unsolved problem—the substitution problem—also generalizes to my theory. If Bob believes that Amelia talks, then for some x, x = the proposition that Amelia talks and Bob believes x. Since, on my view, the proposition that Amelia talks = the cognitive event type in which an agent predicates talking of Amelia, it follows that for some x, x = that event type and Bob believes x. Since this is what is expressed by

King020513OUK.indd 168 11/23/2013 12:59:04 PM PROPOSITIONS VS PROPERTIES AND FACTS 169

1. Bob believes the event type of predicating talking of Amelia. this sentence is one of the truths we learn by doing philosophy. If Speaks continues to stand by his property theory, he should draw a similar conclusion about (2). 2. Bob believes the property of being such that Amelia talks. Th is line of argument does not apply to (3) and (4). 3. Jeff fears that the Trojans will beat the Irish this year. 4. Jeff fears the proposition that the Trojans will beat the Irish this year. Th e reason it doesn’t is that “fears” is ambiguous between a reading in which it combines with sentential clauses and one in which it takes a direct noun phrase object. Because of this (6) does not follow from (5).2 5. Jeff fears that so-and so. 6. Th ere is an x such that x = the proposition that so-and-so and Jeff fears x.

Propositions as facts I agree with two tenets of King’s view: that propositions represent things as being certain ways, and so have truth conditions, and that their intentionality is due to the intentionality of agents. But I don’t fully understand his explanation of the second. It begins with the claim that sentence (7a) is the fact given in (7b), which consists in “Michael” standing in a certain “sentential relation” to “swims.” (p. 11)

7a. Michael swims. b. the fact consisting in “Michael” occurring as the left terminal node that is the daughter of a node that also dominates the right terminal node at which “swims” o c c u r s .

Let us call the relation in which “Michael” stands to “swims” in (7a/b) “R.” Next King identifi es another fact, given by (7c), called “an interpreted sentence” (corresponding to (7a)).

7c. the fact consisting in there being a possible context of utterance c such that (i) Michael is the semantic value of “Michael” in c, (ii) the property swimming is the semantic value of “swims” in English, and (iii) “Michael” stands in R, which in English encodes ascription (predication), to “swims”

2 Speaks raises this possibility, which he attributes to King.

King020513OUK.indd 169 11/23/2013 12:59:05 PM 170 SCOTT SOAMES

(7c) includes quantifi cation over possible contexts, while also including the English language, the man Michael, the property swimming, the notion the semantic content of an expression relative to a context , R, and the notion encoding ascription . 3 Th is was the view of King (2007); here, he modifi es it, explaining how quantifi cation over possible contexts can be eliminated in favor of quantifi cation over assignments of objects to variables. Th us (7c’) replaces (7c). 7c. the fact consisting in there being a context of utterance c and an assignment f of values to (individual) variables such that (i) Michael is the semantic value of “Michael” relative to c,f, (ii) the property swimming is the semantic value of “swims” in English, and (iii) “Michael” stands in R, which in English encodes ascription (predication), to “swims” Th e fi rst crucial claim in King’s explanation of the intentionality of the proposition that Michael swims is the claim that speakers of English have “cognitive access” to the facts he calls interpreted sentences, including (7c’). (pp. 10–13). Next, he articulates a principle that allows him to conclude that speakers have cognitive access to propositions from the fact that they have cognitive access to interpreted sentences. Call a fact

that consists simply of objects o 1. ..o n instantiating an n-place property P a witness of the

related fact consisting in there being some x 1. . . x n Px 1.. . x n . King’s principle is that “having cognitive access to a witness for a fact is a way of having cognitive access to the fact witnessed.” (p13). His third claim is that (7c’) is a witness to the proposition (7d) that Michael swims. 7d. the fact consisting in there being some language L, some expressions e and e’ of L, some syntactic relation R of L, and some context c and assignment f of objects to variables such that (i) Michael is the semantic value of e relative to c,f, (ii) the property swimming is the semantic value of e’ in L, and (iii) e stands in R, which in L encodes ascription (predication), to e’ in some sentence of L Although there is one further step needed to explain the intentionality of propositions, there are already some matters to attend to. First, (7c’) is not a witness of (7d), since the move by which we reach the latter from the former is not bare existential generalization (as the defi nition requires); it is more complex, explicitly introducing the notions of an expression, a language, and a syntactic relation of the language. Of course, “Michael” and “swims” are expressions, English is a language, and R is a syntactic relation (in which some expressions stand to others in sentences). But (7c’) doesn’t say that they are, so the move to (7d) imports content not found in (7c’). Since I suspect King would be happy to add this content to (7c’), I will here take that to have been done. Second, although (7d) speaks of some language , more is needed to ensure the existence of propositions in which properties not expressed by a predicate of any existing

3 For R to encode ascription is, I take it, for R to be used by speakers to predicate the property expressed by the predicate expression that stands in R to the term or terms in question, of the referents of those terms.

King020513OUK.indd 170 11/23/2013 12:59:05 PM PROPOSITIONS VS PROPERTIES AND FACTS 171

language are predicated of objects. Th is may be done either by adding quantifi cation over all possible languages to (7c’) and (7d) (in a manner analogous to the treatment of contexts in King (2007)), or by invoking assignments of properties to predicate variables in actual languages (in a manner analogous to the use of assignments of objects to individual variables). Let this also be assumed. Th ese are matters of detail. Th e serious questions are What kind of cognitive access is King talking about? and Is the witness principle true for that kind of access? Suppose (i) that “cognitive access” to the fact consisting in such-and-such being so-and-so involves knowing or believing that such-and such is so-and-so . Th en the witness prin-

ciple will follow from the claim that when one knows or believes that Po 1 . .. o n , one also

knows or believes that that for some x 1. . . x n Px 1 . . . x n, plus the principle (ii) that one who

knows or believes the latter thereby has “cognitive access” to the fact that for some x 1.

.. x n Px 1.. . x n . Although this story might sound plausible, I doubt it is King’s view. For one thing, he is highly skeptical of the idea that that the proposition that S (which can be known or believed) is the same as the fact that S (which he seems to suggest cannot). But without this identity the story requires further explanation, which he doesn’t give, in order to be plausible. Worse, it is implausible to suppose that fl edgling language users mastering simple sentences like “Michael swims” are acquainted with the “interpreted sentence” (7c’) in a way that depends on knowing, believing, or even imagining:

that there is a context of utterance c and an assignment f of values to (individual) variables such that (i) Michael is the semantic value of “Michael” relative to c,f, (ii) the property swimming is the semantic value of “swims” in English, and (iii) “Michael” stands in R, which in English encodes ascription (predication), to “swims”. Th is is too complicated for neophyte language users; it also contains concepts—context, assignment, semantic value, R, ascription (i.e. predication ), English—that such language users cannot be assumed to possess. More importantly, the fact that belief and knowledge are relations to propositions disqualifi es them from playing a role in the explanation of how agents endow otherwise non-intentional facts with intentional properties by “interpreting” them. To take King’s “cognitive access” to presuppose propositional attitudes as prerequisites for such “interpretation” would be to destroy the explanation by assuming what is to be explained. Being aware of this, he doesn’t appeal to such attitudes, contenting himself with the unexplained phrase “cognitive access.” To my mind, this trades one problem for another. Although cognitive access is central to his explanation, the notion has been left too underspecifi ed to bear the load placed on it. Setting this aside, we have reached the stage of the explanation at which speakers have cognitive access to the general metalinguistic fact (7d). Somehow “cognitive access” to this extraordinarily complex fact is supposed to lead fl edgling speakers to “interpret” the relation R* expressed by the following formula.

λxy [there is some language L, some expressions e and e’ of L, some syntactic relation R of L, and some context c and assignment f of objects to variables such that (i) x is the semantic value of

King020513OUK.indd 171 11/23/2013 12:59:05 PM 172 SCOTT SOAMES

e relative to c,f, (ii) y is the semantic value of e’ in L, and (iii) e stands in R, which in L encodes ascription (predication), to e’ in some sentence of L] R* is a complex two-place relation that holds between Michael and swimming , if (7d) is a fact. Since King identifi es (7d) with the proposition that Michael swims, he calls R* “the propositional relation.” Th e idea is, in eff ect, to treat the fact (7d) as a kind of pseudo sentence made up of two pseudo words, the man Michael and the property swimming, standing in the pseudo grammatical relation R*. As always, “grammatical relations” carry semantic signifi cance. Just as the real grammatical relation R is used by English speakers to predicate the property expressed by a predicate expression P of the items designated by Ps argument expressions, so the pseudo grammatical relation R* is used by anyone who entertains the proposition to predicate swimming of Michael. For King, this is what it is to entertain the proposition that Michael swims. In short, in King’s view as in mine, the proposition represents Michael as being one who swims because agents who entertain it do so. Although I fi nd this high-level agreement between King and me to be satisfying, I worry about his account of entertaining a proposition. Th e account begins with the problematic claim that by virtue of understanding (7a) we have cognitive access , in some robust but unexplained sense, to what King calls “the interpreted sentence,” which is the fact (7c’). Since (7c’) specifi es which expression plays the referring role, and which the predicating role, while further indicating the property predicated and the object that is its predication target, this “cognitive access” presumably involves understanding the sentence (7a) as predicating swimming of Michael. Step 2 gets us from “the witness” (7c’) to the complex general fact (7d) to which it is claimed we also have the required “cognitive access.” Step 3 portrays us as picking out the highly complex relation R* and conceptualizing (7d) as consisting of Michael standing in R* to swimming. Step 3 sees agents as undertaking, for some unknown reason, to endow this fact with intentionality by using it (the pseudo-sentence/proposition) to represent something else—Michael—as being a certain way—a swimmer—in the way that parallels their use of the ordinary sentence (7a) to do the same thing. Not only is there no explanation of why agents do this, it is anomalous that anything of this sort should be required. Merely understanding (7a)— which, on King’s account, is analytically prior to any of his further steps—involves agents using the sentence to predicate being one who swims of Michael. One who does this represents him as a swimmer and hence thinks of him as one who swims—which surely is itself a propositional attitude. Th us, one who understands the sentence (in King’s sense) should be seen as already bearing a propositional attitude of the most basic kind to the proposition that Michael swims. To understand the situation in this way is to recognize that making sense of the very fi rst step in King’s putative explanation involves presupposing that agents bear propositional attitudes to the proposition whose intentional properties he sets out to explain. Since this presupposes some other, conceptually prior, way of entertaining the proposition that Michael swims, it seems to me that his explanation fails.

King020513OUK.indd 172 11/23/2013 12:59:05 PM PROPOSITIONS VS PROPERTIES AND FACTS 173

King must not see it this way. Perhaps he takes the fi rst step in the process—namely as understanding the sentence, in the sense of having “cognitive access” to (7c’)—as not involving one’s predicating anything of Michael, not representing him as one who swims, and hence as not entertaining, or bearing any attitude, to the proposition that he swims. How this can be so is a mystery to me. Since the agents in question use and understand the sentence which, in eff ect, tells them to predicate swimming of Michael, I would have thought that their understanding the sentence and “cognitive access” to (7c’) would already have them thinking of Michael as a swimmer—and hence as entertaining the proposition that he is, independent of any explanatorily downstream “interpretation” of the “propositional fact” (7d). Since, by contrast, King thinks such interpretation is required, the fi nal step in his journey is to argue that once agents have embarked on interpreting the propositional fact and relation, there is only one reasonable way for them to do so. He says:

“But even if we are now convinced that it is [7d]’s propositional relation that we interpret as ascribing the property of swimming to Michael, we need to say what constitutes interpreting

it . . . It is simply that we compose the semantic values at the terminal nodes of the propositional relation in the way that we do. [Note the treatment of the fact (7d) as a kind of pseudo sentence and the relation R* as a pseudo grammatical construction.] In the end this is just a refl ex of the sentential relation R having the semantic signifi cance that it does [i.e. ascription/predication]. When we entertain a proposition, we work our way up the propositional relation, combining semantic values to yield new semantic values for further combining. [Again propositions as

pseudo sentences.] . . . In the case of [7d], were we to do anything other than ascribe the property of swimming to Michael, we would not be combining semantic values in a manner that is consistent with the way we interpret the syntax of the sentence [7a] . It just isn’t coherent to interpret the sentential relation R as ascribing the semantic value of ‘swimming’ to the semantic value of ‘Michael,’ while composing the semantic values Michael and the property of swimming in some other way as one moves up the propositional relation of [7d]. Semantic values only get composed once in understanding sentence [7a], and hence entertaining the proposition [7d]. We either do so in a way dictated by the sentential relation R or not. To do so in the way dictated by the way we interpret the sentential relation R just is to interpret the propositional relation as encoding ascription. To summarize, [the proposition 7d] has truth conditions because speakers interpret its propositional relation as ascribing the property of swimming to Michael.” (13–14, my emphasis) Th e sentences emphasized in the passage make it sound as if there are two (simul- taneous?) interpretations going on here (running in parallel?)—one of the sentence, which involves the sentential relation R, and one of the proposition, which involves the propositional relation R*. Th e two are brought into harmony by the need for consistency, which dictates that the latter agrees with the former. My problem is that I don’t understand the need for the apparent duality in the fi rst place, in which a pair of interpretations must be brought into harmony. By King’s own account, his elaborate construction requires some conceptually antecedent understanding of the sentence to provide the facts needed to construct the proposition the interpretation of which must be made consistent with the conceptually prior understanding of the sentence. If, as

King020513OUK.indd 173 11/23/2013 12:59:05 PM 174 SCOTT SOAMES

I believe, this understanding of the sentence already requires one to bear an elementary attitude (entertaining) to a, or the, proposition that represents Michael as a swimmer (and nothing more), then, what is to be explained is presupposed at the fi rst step. Perhaps King will explain where precisely, and why, he disagrees. In addition to this worry, there is a further, elementary point to be emphasized. However “interpreting,” and hence entertaining, propositions is ultimately explained, King is committed to the idea that it always involves understanding sentences. For me, this is too logo-centric. Th ere are many actual and possible agents, including human beings, who bear propositional attitudes to propositions presented to them in perception and non-verbal thought that seem not to be presented to them by any spoken or written sentence they understand. Although some philosophers may be tempted to speculation about the “languages of thought and perception” of all possible agents with perceptions, beliefs, desires, and expectations, one’s theory of propositions shouldn’t force one to this extremity. 4 I will return to a related point below. Th e critique off ered here is, of course, directed at King’s claims that the use and understanding of language by agents is explanatorily prior to their attitudes to propositions, and to their endowment of propositions with intentionality. In an earlier exchange between us I took the presence of passages like the following, from King (2007), to pose a temporal problem as well.

“Consider the time at which sentences like ‘Rebecca swims’ fi rst came into existence . . . As should now be clear, the existence of sentences such as [this] brings into existence facts such as 4b’’ [that bear the same relation to ‘Rebecca swims’ as (7d) bears to (7a)] where let us suppose the propositional relation doesn’t yet encode the instantiation function [now called ‘ascription’] but the sentential relation of ‘Rebecca swims’ does. Since we now claim that the propositional relation encoding the instantiation function [ascription] is part of the fact that is the proposition that Rebecca swims, 4b’’ is not yet that proposition. Indeed neither the proposition that Rebecca swims, nor, we may suppose, any other proposition exists yet. Th us it must be that the language does not yet contain verbs of attitude, modal operators, or that-clauses. However, sentences have truth conditions . . . As verbs of attitude enter the language speakers begin to talk about structured contents . . . In short, when English came into existence and prior to it having the resources to talk about propositions, it brought into existence facts like 4b” . . . As speakers began to attempt to talk about structured contents by means of that-clauses, they implicitly took these contents to have the same truth conditions as the sentences with those contents . . . Perhaps it was indeterminate at fi rst which eligible facts are the structured contents of sentences.But the facts that in the end

4 Th e point is not to deny that predications that occur in perception or “non-verbal” thought are constituted by the agent’s use of an internal language-like representational system some elements of which designate the predication targets while other elements designate the properties predicated. Th e point is to remain neutral on such speculation. I certainly believe that some instances in which natural language users predicate properties of objects are constituted by the agents’ use and understanding of the natural language expressions they employ. However, I neither assert nor deny that all predications of properties, and instances of entertaining propositions, are similarly constituted by the agent’s employment of an internal representational system that mediates between the propositions entertained and their worldly subject matter.

King020513OUK.indd 174 11/23/2013 12:59:05 PM PROPOSITIONS VS PROPERTIES AND FACTS 175

are most eligible to be structured contents, propositions, must share the truth conditions of the sentences whose contents they are eligible to be . . . ” 5 On this basis, I interpreted King as implausibly claiming that there was a time before propositions existed when speakers used and understood sentences of primitive languages (without modal operators, attitude verbs, or that-clauses). Since this seems clearly in line with the words just quoted, I was surprised to read that he now claims my earlier interpretation was a misunderstanding. Nevertheless, I am pleased with what seems to me to be his change in view. However, as he notes in chapter 4 above, there is still a temporal problem to be faced. Languages are cognitively complex social institutions. To speak and understand them agents must have beliefs and intentions about expressions and what they are used to talk about. Th ey must also have beliefs and expectations about what other agents know and what they don’t, as well as what is of interest to them and what isn’t. Speakers need further beliefs and expectations about how their hearers will interpret their words, and how well they will read the speakers’ beliefs and intentions. Because of this, there is, I think, no speaking a language by agents who don’t fi rst possess a rich store of propositional attitudes. Th is was a problem for King (2007). Since the existential quantifi ers used in specifying the facts with which he identifi ed propositions did not range over the nonexistent, the existence of propositions at t depended on the existence of one or more languages at t. From his account of what it is to entertain a proposition it further followed that no one can entertain a proposition at a time when it doesn’t exist. Th us, he was forced to the implausible suggestion that there were no propositions or agents with propositional attitudes before there were languages. Recognizing the language-of-thought hypothesis as a possible way out, he rightly did not rest his case on it, and so admitted that we must take seriously the idea “that strictly speaking our proto-linguistic ancestors did not have propositional attitudes because propositions didn’t exist then.”6 Nevertheless, he suggested that “they had some sort of “ proto-intentional states ”: proto-beliefs and proto-intentions .” 7 Summing up, he put his tentative conclusion as follows:

“Propositions and real intentional states with propositional content come into existence together. Hence we need not suppose that our proto-linguistic ancestors literally had propositional attitudes prior to the existence of language and propositions. It is enough to suppose that they had proto-intentional states not too diff erent in kind from those had by many animals today.” 8 As I said in our earlier exchange, I don’t fi nd this convincing. Whatever these “proto-intentional states” are, they can’t be relations to representational bearers of

5 King (2007) pp. 60–61, my emphasis. See also the paragraph spanning pp. 66–67. 6 Ibid., p. 66 7 Ibid., p. 66, my emphasis. 8 Ibid., p. 67.

King020513OUK.indd 175 11/23/2013 12:59:05 PM 176 SCOTT SOAMES

truth conditions, lest they raise the same problems that genuine propositional attitudes do. If the postulated primitive states are not relational in this way, we need to be told how, if at all, they are representational, and how, if they are not, they provide the rich conceptual resources necessary to give birth to language. Whereas I don’t think this can be done, King argued that it must be possible since everyone faces a version of the same problem.

“[E}ven if propositions existed eternally, there was a time at which no creatures had mental states with propositional content. Hence, some account must be given of how creatures came to have propositional attitudes. If we consider creatures immediately prior to the time when creatures had propositional attitudes and creatures who fi rst had them, some explanation will have to be given of how the latter managed to get in cognitive contact with propositions. But in sketching such an account one is faced with the challenge of describing the minds of our ancestors without using verbs of propositional attitudes. Here again it seems one would have to invoke proto-intentional states and proto-intentional action as part of the explanation,just as I did above. Hence, on this score my account is not in any worse shape than an account of which propositions are eternal. ”9 I don’t think this is quite right. From my perspective, King’s focus on language as the loci of propositions led him to misconceive the problem. Th e cognitive requirements required to speak and understand even a very simple language are complex. Because of this, antecedent propositional attitudes are required to explain both the birth of language and the way children acquire it. It was because King denied this, while tying propositional attitudes to understanding sentences, that he was confronted with a puzzle. How can agents have the complex cognitive abilities needed to master a language, and thereby come to have propositional attitudes, without having the attitudes to begin with? His 2007 answer embraced the deus ex machina of “proto-beliefs” and “proto-intentions,” which somehow have just the power needed for agents to master language, and so to acquire propositional attitudes, without having whatever features of propositional attitudes that make understanding a language necessary for having them. Th e suggestion that we all face this sort of puzzle is simply not true. On my account, all thought and perception involves propositional attitudes. Consider vision. To see something is to see it as being a certain way (e.g. as red, round, etc.)—which on my story is to represent it as being so-and-so by virtue of predicating the property being so-and-so of it. Since this is one of the basic ways of entertaining a proposition, any creature that can see has propositional attitudes. Th e same can be said of nonverbal thinking. To think of something is to think of it as being a certain way, which is to represent it as being that way by virtue of predicating a property of it. Th is, too, is to bear an attitude to a proposition. Hence, nonlinguistic agents capable of perception and cognition may, depending on their capacities, bear cognitive attitudes to more or less conceptually rich sets of propositions. Th is, I maintain, is what makes it possible to explain the birth and acquisition of language.

9 P. 67, my emphasis.

King020513OUK.indd 176 11/23/2013 12:59:05 PM PROPOSITIONS VS PROPERTIES AND FACTS 177

For this reason, I am pleased to see the big step King takes in this direction with the following words from chapter 4.

“I believe that many things have content other than sentences of natural languages. Maps, diagrams, perhaps pictures and, most importantly for present purposes, perceptual experiences have contents. In the case of each sort of thing that has content, there will be an account of those contents in the spirit of the present account of the contents of natural language sentences . . . it is plausible to suppose that the contents of perceptual experiences have truth conditions. Finally it seems reasonable to suppose that the contents of perceptual experiences can be objects of attitudes like belief, desire, etc. But then our prelinguistic ancestors could have had beliefs and desires whose objects are the contents of perceptual experiences. Th ese attitudes could then fi g- ure in the account of how language, and the contents of natural language sentences came into existence.” (22–23) Because his new views concerning types of contents/propositions not tied to language haven’t yet been presented, no serious assessment of them can now be made. Th ere is, however, a worry to be registered. Since his account of propositions expressed in language ties them, and attitudes we bear to them, so completely to sentences and our cognitive relations to them, it seems likely that the class of propositions to which we are, on his approach, related by visual experience will be entirely disjoint from the class of linguistically expressed propositions. Th e same, I suspect, can be said about the class of linguistically expressed propositions and the class of propositions to which we are related by nonverbal thought (if he also recognizes these). Th is has the potential for creating problems for the account of how sentential clauses are used to report the contents of perceptual and cognitive experiences. It may also create problems for explaining the ways in which language users eff ortlessly integrate the propositional information brought to them through language, perception, and nonverbal cognition. For these reasons, I suspect it is a mistake to start with a thoroughly linguistic account of propositions expressed in language, with the hope of graft ing it on later accounts of those with which we are nonlinguistically acquainted. Instead, I prefer to start with a notion of propositions not tied to any single mode of presentation, and to work for further specifi cation from there. Time will tell which of these research programs is the more successful. King closes chapter 4 discussing an objection in my earlier article “Propositions” that took it for granted that facts in his sense are things that can be referents of ⌜ the fact that S⌝ . He says that this was a mistake, indicating that what he then meant (and now means) by “fact” is “n objects standing in an n-place relation, n properties standing in an n-place relation, and so on.” (King p.30) On this understanding—which indeed is what I took him to mean— Annie’s being smart qualifi es as a fact, as does Jeff ’s being dif- ferent from Scott . Th ese, I assumed, were regarded as complex entities the existence of which were taken to make the propositions that Annie is smart and that Jeff is diff erent from Scott true. Taking it to be obvious that ⌜the fact that S ⌝ designates a fact, if it designates anything at all, I assimilated Annie’s being smart and Jeff ’s being diff erent from

King020513OUK.indd 177 11/23/2013 12:59:05 PM 178 SCOTT SOAMES

Scott to the fact that Annie is smart and the fact that Jeff is diff erent from Scott —thereby reaching the familiar conclusion that the fact that Annie is smart and the fact that Jeff is diff erent from Scott are what philosophical defenders of facts take to make the propositions that Annie is smart and that Jeff is diff erent from Scott true. (King pp. 29–30) If I now understand him correctly, it is this last step that he disputes, when “fact” is taken as he understands it. Th ough he doesn’t deny it—either now or in King (2007)—he doesn’t affi rm it either. This being so, the argument to which he objects was based on a premise to which he was not committed.10 Nevertheless, in chapter 4 above he is, for the sake of argument, willing to assume that ⌜ the fact that S⌝ does designate a fact, while providing evidence that even so, what it designates is diff erent from what ⌜the proposition that S ⌝ designates. It is good that he does this, because the constructions he considers in attempting to establish this conclusion support, by and large, the substitutability of ⌜so-and-so’s being such-and-such⌝ and ⌜the fact that so-and-so is such-and-such⌝ for one another—which, on his meth- odology, supports the substantive view that he complains my misunderstanding forced on him. Viewed in this light, his response is an attempt to provide precisely the empirical evidence my objection requested. 11 His evidence consists in the problematic results of substituting one of ⌜the fact that S ⌝ and ⌜the proposition that S ⌝ for the other, or for substituting either for ⌜ that S⌝ , under various verbs. In many cases such substitution changes meaning, truth value, or grammaticality. Although King is cautious about interpreting these results, he is right to suggest that, taken at face value, they do make a prima facie case for distinguishing the referents of ⌜ the proposition that S ⌝ from those of ⌜ the fact that S⌝ , while taking ⌜that S⌝ to be capable of designating those of either. To that extent, his examples provide a reasonable response to my earlier objection. Nevertheless, I remain uncertain what facts are supposed to be, and what relation they bear to propositions. I am also troubled by the observation that his substitution tests can cut in directions diff erent from those he indicates; as they do with a venge- ance when one of the designators, K1—K4 (of what he takes to be the proposition that Michael swims) is substituted for “the proposition that Michael swims” or for “that Michael swims” in examples (8a,b,c) K1. there being some language L, some expressions e and e’ of L, some syntactic relation R of L, and some context c and assignment f of objects to variables such that (i) the property swimming is the semantic value of e in L, (ii) Michael is

10 In footnote 30, p. 149 of King (2007), to which he directs us in chapter 4, he says “I am not assuming that expressions of the form ‘the fact that p’ designate what I have called facts throughout the book. It is a substantive claim that they do so. I will remain neutral on that question here. But I shall call the things that they designate ‘facts’ in this chapter and assume that they are not propositions.” I am afraid I picked up his usage in the text while overlooking this footnote. 11 Th e reader is invited to make such substitutions in (4a-f) and (8b) of chapter 4.

King020513OUK.indd 178 11/23/2013 12:59:05 PM PROPOSITIONS VS PROPERTIES AND FACTS 179

the semantic value of e’ relative to c,f and (iii) e stands in R, which in L encodes ascription (predication), to e’ in some sentence of L K2. Michael’s standing in the relation there being some language L, some expressions e and e’ of L, some syntactic relation R of L, and some context c and assignment f of objects to variables such that x is the semantic value of e relative to c,f, y is the semantic value of e in L, and e stands in R, which in L encodes ascription (predication) to e’ in some sentence of L to the property swimming K3. the fact that there is a context c and assignment f of objects to variables such that Michael is the semantic value, relative to c and f, of some expression e of some language L, the property of swimming is the semantic value of some expression e’ of L, and e stands in R, which in L encodes ascription (predication), to e’ in some sentence of L K4. the fact that Michael stands in the relation there being some language L, some expressions e and e’ of L, some syntactic relation R of L, and some context c and assignment f of objects to variables such that x is the semantic value of e relative to c,f, y is the semantic value of e in L, and e stands in R, which in L encodes ascription (predication) to e’ in some sentence of L to the property swimming

8a. Fred believes (the proposition) that Michael swims. b. It is likely that Michael swims. c . Th at the moon causes the tides is true.

Th ese substitutions produce apparent absurdity. Does this show that the fact designated by K1—K4 really isn’t the proposition King takes it to be? If not, why do the changes rung by the substitutions he mentions in his examples (4–8) show that ⌜the proposition that S ⌝ and ⌜ the fact that S ⌝ designate diff erent things? As I said in connection with Speaks’s discussion of the substitution problem, a theorist like King will have to accept some of the seemingly absurd results of substitution as true (if the unprob- lematic sentences into which they are substituted are true). Th is contributes to my uncertainty about what the results of substitution in his examples (4–8) really show. I am similarly unconvinced by his rebuttal of my example (9), which appears to indicate that what is regretted—namely the fact that Pam is pregnant —is sometimes believed, and hence is nothing more than the proposition that Pam is pregnant (contra King).

9. Pam regrets that she is pregnant. Although her parents don’t realize it yet, in time they will come to believe it.

King mentions two ways of accommodating this data. Th e fi rst, derived from Parsons (1993), takes the antecedent of the occurrence of “it” in the fi nal clause to be an occurrence of “^[Pam is pregnant]” in the complex singular term “c^[Pam is pregnant]” that is the object of “regret.” Th e fi rst of these designates the proposition that Pam is

King020513OUK.indd 179 11/23/2013 12:59:05 PM 180 SCOTT SOAMES

pregnant, while the second designates the fact that Pam is pregnant—which the function designated by “c” assigns as value to the proposition as argument. Th us, King concludes, the truth of (9) can be made compatible with the distinctness of the fact from the proposition. Th is strikes me as too quick. Although the explanation requires there to be a complex term for the fact in a position in which its argument expression can serve as the antecedent of a later pronoun, this is an accidental feature of the example chosen. For example, consider (10)

10a. Pam regrets something that her parents don’t yet realize, but will soon come to believe. b. Something Pam regrets is now merely suspected by her parents, but will soon be believed by them.

Not only do these seem fi ne, they seem to entail that some one thing can be both regretted and believed, or both regretted and suspected. It also seems obvious that the same things that can be known can, and are, believed—despite the fact that King takes belief to require propositions as objects, while seeming to take examples like (11) (his (4c)) to show that the objects of knowledge are the non-propositional referents of clauses ⌜the fact that S⌝ .

11. Th e fact that the moon causes the tides is well known.

Th us, there is much more here to explain. King’s second strategy for accommodating (9) is based on (12).

11. Th e book I just stole from the library is on my desk. It was written in 1801 and has been translated into many languages.

According to King, the antecedent of “it” is the occurrence of “the book I just stole from the library” in the fi rst sentence, which he takes to designate a concrete object— a copy of said book—while “it” designates the abstract object itself. Th is is taken to show that antecedents and anaphors can refer to diff erent but related things, which, he thinks, is how (9) may be understood. Again, I am skeptical. I may truly remark “Th e book I stole from the library that is now on my desk is the same as the book Mary stole from her library that is now on her desk.” Here I am talking about a book type—e.g. War and Peace —which is both sitting on my desk, and sitting on Mary’s, having been stolen by each of us from our respective libraries. What this illustrates is also at work when one says that one wrote the same word on the board twice on separate days. Abstract objects can have properties—like being written on the board, being on a desk, being stolen from the library, and the like, by virtue of properties had by their tokens. Th us, the defi nite description in (12) can be understood as denoting the book itself, rather than a copy of the book, even though the book is truly said to be sitting on the desk (by virtue of the fact that the copy is).

King020513OUK.indd 180 11/23/2013 12:59:05 PM PROPOSITIONS VS PROPERTIES AND FACTS 181

Th is completes my critical remarks. Despite the inconclusive sparring over the referents of ⌜ the fact that S ⌝, ⌜the proposition that S ⌝ and ⌜ that S ⌝, the crucial points in my discussion of King’s chapter 4 are the challenges I pose for his explanation of the intentionality of propositions, and my related concern that he ties the propositions expressed by sentences, and the attitudes we bear to them, too closely to those sentences and our understanding of them. Despite these points of disagreement, I am pleased that King now recognizes propositions expressed by pictures and diagrams, and that he takes perceptual experience to have propositional content. Th ese additions expand the broad areas of agreement between our two views, both of which remain works in progress. 12

12 Th anks to Brian Bowman for helpful comments on this chapter.

King020513OUK.indd 181 11/23/2013 12:59:05 PM King020513OUK.indd 182 11/23/2013 12:59:05 PM P A R T I V Further Th oughts

King020513OUK.indd 183 11/23/2013 12:59:05 PM King020513OUK.indd 184 11/23/2013 12:59:05 PM 1 0

Responses to Speaks and Soames

J e ff rey C. King

R e s p o n s e t o S p e a k s Speaks begins by raising two objections to Soames’s and my views that he thinks apply to both of them in virtue of a common structure they share. I’ll only consider his objections as they apply to my view. Th e fi rst objection runs as follows. I make free use of properties. Th ey are constituents of my propositions. Hence, Speaks claims, I accept the ontology of his theory according to which the proposition that Amelia talks is the property being such that Amelia talks. Speaks, by contrast, does not accept the existence of the “representational surrogates” that I claim are propositions. So: (S1) Th e ontology of Speaks’s theory is included in my theory and not vice versa. Further, for me, properties are constituents of propositions and in trying to explain the cognitive access agents have to propositions, I simply assume that agents have cognitive access to the properties that are their constituents. So since Speaks thinks that propositions are just properties, it seems that I must admit that it is no easier to have cognitive access to the entities I claim are propositions than it is to have cognitive access to the entities Speaks claims are propositions. Hence: (S2) Our cognitive access to the entities Speaks claims are propositions cannot be harder to explain than our cognitive access to the entities I claim are propositions. From (S1) and (S2), Speaks reasonably concludes that the only reason for preferring my theory to his would be that my theory explains something not explained by his ((S1) and (S2) together seem to entail that Speaks’s sparser-than-my ontology yields a theory of propositions and explains how we have cognitive access to them at least as well as my theory). But Speaks does not see what that would be.

King020513OUK.indd 185 11/23/2013 12:59:05 PM 186 JEFFREY C. KING

In response, let me say that I am happy enough to accept (S2), but I reject (S1). Perhaps more importantly, I reject a presupposition of Speaks’s argument. Taking the former point fi rst, as my criticisms of Speaks in Chapter 7 suggested, when I get serious about thinking about properties, I ultimately think that sharing of properties should make for real, objective similarity. But this means that, fi rst, I fi nd it hard to swallow the idea that there are negated or disjunctive properties. 1 For sharing properties like not being a skier or being a skier or a planet obviously does not make for real similarity. Th at in turn means that I will want no part of properties like being such that George W. Bush is not a skier or being such that George W. Bush is a skier or there are eight planets in Earth’s solar system, which Speaks alleges are expressed by the sentences “George W. Bush is not a skier.” and “George W. Bush is a skier or there are eight planets in Earth’s solar system.” Second, even sharing of the properties Speaks alleges are expressed by simple sentences without negation or disjunction does not make for objective similarity. Everything possesses the property being such that George W. Bush was a poor student, which Speaks alleges to be expressed by the sentence “George W. Bush was a poor student.” But this does not make everything objectively similar to everything else. So I would want no part of the property being such that George W. Bush was a poor student either. Hence, I do not accept all of Speaks’s ontology, and so (S1) is false. 2 Perhaps more importantly, Speaks’s objection here presupposes that the properties he alleges are propositions are able to play the role propositions are intended to play as well as my candidates for propositions do. Aft er all, the appeal to Speaks’s sparser ontology only has bite if his ontology can do everything that mine can. However in Chapter 7 I gave reasons for thinking that Speaks’s properties are not suited to play the role of propositions. But then I have given reason for rejecting the presupposition Speaks’s argument requires here. Speaks gives a second objection to my theory and that of Soames that he thinks applies to both theories in virtue of their shared structure before turning to specifi c criticisms of each theory. However, in this case he gives a more specifi c version of the objection when he turns to specifi c objections to my view. Hence, I consider this more specifi c version below rather than the more general version stated at the outset. Th e fi rst objection Speaks raises specifi cally to my account of propositions must be stated with some care, since it raises some complex issues. Speaks correctly claims that I hold that propositions have their representational properties essentially. He then

1 More cautiously, properties expressed by negated or disjunctive predicates, like “is not a skier” and “is either a skier or a planet”. 2 One response to my concerns here would be to invoke something like a Lewisian theory of naturalness. One could then say that sharing of the perfectly natural properties makes for objective similarity, but not so when one gets down to non-natural properties like those Speaks alleges are expressed by the sentences mentioned in the text. I have tended to resist this picture because I tend to be attracted to a sparse account of properties whereby the only properties that exist are highly natural. Th e remarks in the text here refl ect this attraction.

King020513OUK.indd 186 11/23/2013 12:59:05 PM RESPONSES TO SPEAKS AND SOAMES 187

worries that as a result of this, I cannot explain why/how propositions have representational properties/truth conditions. For, he says, you cannot explain why a thing has any essential property. Speaks considers his essentially wooden desk—Fred—and asks whether we can explain why it is wooden. Speaks thinks that we cannot, since Fred couldn’t exist without being wooden and so we cannot explain how it came to be so. At most we can explain how Fred came into existence. Similarly, Speaks claims, given that I hold that propositions have their truth conditions essentially, at most I can explain how propositions came into existence. Further, Speaks adds, explaining how propositions came into existence is not the same thing as explaining how they have truth conditions. To show this, he provides the following example. Suppose there is a proponent of the view that propositions are structured, having objects, properties and relations as constituents, and who thinks of propositions as traditionally conceived: namely, they have their truth conditions by their very natures and independently of minds and languages. Now suppose I make a white paper airplane. Consider a singular proposition about the airplane, say the proposition that o is white, where o is the airplane in question. In explaining how this proposition came to exist, the theorist should say that by making the plane, the proposition was brought into existence. Let’s call this the white plane explanation . But, Speaks says, in explaining in this way how the proposition that o is white came into existence, have I thereby explained how/why the proposition has truth conditions? He correctly answers that I clearly have not. Nothing about the making of the plane explains how/why the proposition that o is white has truth conditions. Hence, in this case explaining how/why the proposition that o is white came into existence does not explain how/why that proposition has truth conditions. Now Speaks claims that my explanation of how/why the proposition that Michael swims came into existence is relevantly similar to the white plane explanation above and so, like the latter, does not explain how/why the proposition that Michael swims has truth conditions. If Speaks were right here, this would be a very damning consequence since I have insisted that a theory of propositions must explain how/why propositions have truth conditions and have rejected various theories for not being able to do this. Speaks’s argument for the claim that my explanation of how the proposition that Michael swims (came into existence and) has truth conditions is relevantly like the white plane explanation, and so doesn’t really explain why that proposition has truth conditions, begins with some claims about what I call the propositional relation of the proposition that Michael swims (PR). I claim that the proposition that Michael swims consists of Michael standing in a certain relation—PR—to the property of swimming. 3 I claim that the following two-place relation is PR—the propositional relation of the

3 I sometimes add that the proposition consists of Michael standing in the propositional relation to the property of swimming and the propositional relation possessing the property of encoding ascription . However, I agree with Speaks that which of these things I identify with the proposition makes little diff erence. See Chapter 8 note 5 of the present work.

King020513OUK.indd 187 11/23/2013 12:59:05 PM 188 JEFFREY C. KING

proposition that Michael swims: there is a context c, assignment g and language L such that for some lexical items a and b of L, ____ is the semantic value of a relative to g and c and ____ is the semantic value of b relative to g and c and a occurs at the left terminal node of syntactic relation R that in L encodes ascription and b occurs at R’s right terminal node. Th e proposition that Michael swims, then, consists of Michael standing in this relation to the property of swimming. Note for later that a particular sentential relation R is a component of PR. 4 Forgetting about the internal complexity of PR for a moment, I claim that one of the reasons the proposition that Michael swims is true iff Michael swims is that speakers interpret PR as ascribing the property of swimming to Michael. I sometimes put this by saying that PR encodes ascription . As I’ll discuss below, and as should already be apparent, the claim that PR encodes ascription—is interpreted by agents as ascribing the property of swimming to Michael—is an important part of my explanation as to why the proposition has truth conditions. Speaks claims that I am committed to the following, since PR does not have any contingently existing concrete objects as constituents:5 (K1) Necessarily, PR exists. Now, Speaks asks, consider the claim that PR encodes ascription. Is it necessary or contingent? Speaks settles on the following: (K2) Necessarily, if PR encodes anything, it encodes ascription.

Speaks seems to think that from K2 it follows that: (K3) Necessarily, if PR relates anything, then the fact that it relates them is true iff the fi rst instantiates the second. 6 Speaks thinks that K1-K3 show, in eff ect, that PR exists necessarily with its representational oomph. All speakers can ever do is get things to stand in PR. But then by K3, the resulting fact has truth conditions. But getting things to stand in PR provides no explanation of why propositions have truth conditions, just as creating a white paper

4 In Chapter 4 I suggested quantifying over syntactic relations instead of having a particular syntactic relation be a component of PR. I am going to ignore that here and I don’t think it aff ects anything I say (see note 9 below). If it does, that may be a reason for not going the route suggested in Chapter 4 and quantifying over syntactic relations. 5 Chapter 8 p. 153 of the present work. I follow Speaks here in talking about constituents of complex properties and relations. I prefer to reserve the word “constituents” for talking about (certain) elements of propositions. In what follows, I eventually lapse into my favored terminology and talk of the components of complex properties and relations. 6 I am not sure why Speaks thinks K3 follows from K2, since K2 would seem to allow PR to exist and encode nothing. But if that were the case and if PR related two things x and y, I don’t see why the resulting fact would be true iff x instantiates y. Perhaps Speaks thinks that for PR to relate anything, it must encode ascription.

King020513OUK.indd 188 11/23/2013 12:59:05 PM RESPONSES TO SPEAKS AND SOAMES 189

plane provides no explanation of why the proposition that o is white has truth conditions. Speaks makes these points in the following passage:

Th e role played by speakers is simply to make it that case that [PR] relates some things; once they have brought into existence a fact of the form [PR](x,y) there’s no further work for them to do in giving this fact its representational properties. But, given this, it seems like the role played by speakers on this theory is uncomfortably close to the role played by our paper airplane maker in the example discussed above. We can focus this worry in the way mentioned above, by asking: what relation does the interpretation of [PR] bear to the representational properties of FAST which the making of the paper airplane does not bear to the proposition that that paper airplane is white?7 Now I agree with Speaks that if I held K1-K3, I would not have explained how/why propositions have truth conditions. Details aside, that PR encodes ascription is a big part of the explanation of how/why propositions have truth conditions. But if PR exists necessarily and encodes ascription (at least whenever it relates things), then speakers could play no role in its so doing. We would in eff ect be positing a necessarily existing relation that gives propositions their representational oomph. Th at doesn’t look like much of an explanation of that oomph. But I don’t hold all of K1-K3. In particular, I reject K1. As to K2, I hold that PR encodes ascription essentially—in every world in which it exists, it encodes ascription. Finally, I accept K3. I am sure Speaks will fi nd my rejection of K1 puzzling. Recall that his reason for thinking me committed to K1 is that PR does not have any contingently existing concrete objects as constituents. Th is suggests that Speaks thinks that the only reason a property or relation could fail to exist is if it is something like the property of loving Annie, which contains the concrete, contingently existing Annie as a constituent. At deprived worlds where Annie sadly fails to exist, so too will properties that have her as a constituent. Now I agree that properties and relations can fail to exist for this reason. But I think that some properties and relations that don’t (or don’t obviously) have contingently existing concrete objects as constituents nonetheless exist only contingently. Before moving on, let me note that I am not alone in this. Robert Stalnaker [2010] writes:

Consider, for example, colour properties—paradigms of purely qualitative properties. If there are metaphysically possible worlds with radically diff erent physical laws, perhaps worlds without light or other kinds of electromagnetic radiation, then there will be worlds in which nothing is or could be colored, and I think it would also be reasonable to conclude that the color properties would not exist in such a world. 8

7 Chapter 8 p. 154 of the present work. FAST is the proposition that Michael swims. 8 Stalnaker [2010] p. 23

King020513OUK.indd 189 11/23/2013 12:59:06 PM 190 JEFFREY C. KING

I won’t argue that matter here, but I also think that if certain views about what it is to be a species are correct (e.g. if it involves reproductive isolation) then it is plausible that properties like being a member of species X exist only contingently. Fine and good, but why think relations like PR exist contingently? Because it is plausible to think that some of the properties and relations that are components of PR exist contingently. Here I’ll focus on the syntactic relation R that the words in the English sentence “Michael swims” stand in. But I think that a similar case could be made for other properties and relations that are components of PR.9 It seems pretty clear that at lifeless possible worlds, English doesn’t exist. At such worlds, there are no language users and no facts about language users; and it seems implausible that English could exist in the absence of such facts. Consider a lifeless world w. Does the fact that English doesn’t exist at w consist in there being no tokens of English expressions at w? It doesn’t seem so. Th ere are many English sentences (types) that have no tokens in the actual world. But it seems very plausible that these sentences exist in the actual world nonetheless. Aft er all, we do say that there are English sentences that have no tokens. But this suggests that for English to fail to exist at w, it can’t merely be a matter of tokens of English sentences failing to exist at w. Th e types must fail to exist there too. But now if the English sentence type “Michael swims” fails to exist at w, it seems plausible to suppose that this is because the word types “Michael,” “swims” and the syntactic relation R all fail to exist at w. But since R is a component of PR, this means that PR won’t exist at w either.10 So the picture is that agents brought language into existence and in so doing brought into existence expression types and the syntactic relation R interpreted as encoding ascription. In so doing, they brought into existence PR, also interpreted as encoding ascription. Having done this, they provided the proposition FAST with its representational properties. So here, unlike the case of the white plane explanation, the explanation of how speakers brought the proposition into existence does explain how/why the proposition has truth conditions. Th is is because in the present case, unlike the case of the white plane explanation, the explanation of how the proposition came into existence includes an explanation of how speakers brought PR into existence, where the latter gives the proposition FAST its representational oomph. 11 Speaks’s second objection concerns how expressions like “Michael” and “swims” get their semantic values. Speaks correctly points out that on my view, part of the explanation as to why/how propositions have truth conditions is that speakers interpret

9 E.g. I suspect the property of being a syntactic relation that encodes ascription exists contingently. See note 4. 10 Obviously, I presuppose here that if a component of a relation fails to exist at a world, so does the relation. Speaks seems to accept this too. 11 It should be clear, then, that I disagree with Speaks when he says “Th e worry is that we are getting too much for free from the nature of [PR]— to which no thinking subject contributes anything—to get a satisfying explanation, in terms of thinking subjects, of the representational properties of propositions.” (Chapter 8 p. 154 of the present work; my emphasis). Th inking subjects brought [PR] into existence. So we do get an explanation of the representational properties of propositions in terms of thinking subjects.

King020513OUK.indd 190 11/23/2013 12:59:06 PM RESPONSES TO SPEAKS AND SOAMES 191

syntactic relations, and so propositional relations, in the way that they do. But in order to explain how speakers interpret the syntactic relation in the sentence “Michael swims,” we must assume that “Michael” has Michael as its semantic value and “swims” has the property of swimming as its semantic value. In general, in order to explain speakers interpreting any syntactic relations we must assume that lexical items have semantic values. But then this means that (some) lexical items having semantic values is explanatorily prior to speakers interpreting syntactic and propositional relations, and so explanatorily prior to the existence of propositions. Th is in turn means that the account of how (some) lexical items come to have semantic values cannot appeal to agents having propositional attitudes whose objects are propositions expressed by natural language sentences. Speaks worries that plausible accounts of how expressions get their semantic values will have to appeal to such propositional attitudes, and so will be inconsistent with my account of propositions. Speaks is correct that on my view some lexical items having semantic values is explanatorily prior to the existence of propositions expressed by natural language sentences. However, as I said in Chapter 4, I think that things other than natural language sentences have truth evaluable content. As I indicated, maps, diagrams, perhaps pictures and, most importantly for present purposes, perceptual experiences all have contents.12 For each sort of thing that has content, there will be an account of those contents in the spirit of the present account of the contents of natural language sentences. In the case of perceptual experiences, there is still considerable controversy regarding what their contents are like. I believe it is plausible to suppose that the contents of perceptual experiences have truth conditions; and that these contents can be the objects of propositional attitudes. As I indicated in Chapter 4, this means that our prelinguistic ancestors could have had beliefs and desires whose objects are the contents of perceptual experiences. Th ese attitudes could then fi gure in the explanation of how lexical items acquired semantic values. So I am committed to explaining how some lexical items acquired semantic values without appealing to propositions expressed by sentences of natural language. Agents having propositional attitudes towards the contents of perceptual experiences can fi g- ure in the explanation of how agents secured semantic values for these lexical items. Note that once we have explained how agents secure semantic values for a few lexical items, form sentences, interpret syntactic and hence propositional relations, and so bring propositions expressed by natural language sentences into existence, attitudes towards these existing propositions expressed by natural language sentences can then play a role in the account of how new lexical items get semantic values. So the restric- tion that only attitudes towards the contents of perceptual experiences can be appealed to in explaining how lexical items get semantic values only applies to the fi rst lexical items that get semantic values. As I’ve said, there is still considerable controversy about

12 Of course some deny that perceptual experiences have contents.

King020513OUK.indd 191 11/23/2013 12:59:06 PM 192 JEFFREY C. KING

the nature of the contents of perceptual experiences but I am here assuming they are truth evaluable. However, many other questions remain including what sorts of properties fi gure in these contents.13 Still, the claim that one can give an account of how the fi rst lexical items got their semantic values by appealing only to propositional attitudes that have as their objects the contents of perceptual experiences does not strike me as implausible. Hence, I can reasonably hope that my theory of propositions is consistent with an account of how agents got language up and running and hence brought propositions expressed by natural language sentences into existence, contrary to the worry Speaks is raising here. As just indicated, I think that perceptual experiences have truth evaluable contents; and as I indicated both above and in Chapter 4, I would expect there to be a theory of their contents that is in the spirit of my theory of the contents of natural language sentences. Speaks’s fi nal objection purports to cast doubt on whether such an account of the contents of perceptual experience will be forthcoming. In considering how a view of the contents of perceptual experiences that is in the spirit of my view of the semantic contents of natural language sentences might look, Speaks begins with a frog’s visual experience of a fl y sitting on a green leaf. Call this experience FLY . Speaks proposes to simplify the example by supposing that the content of FLY is a singular proposition that predicates the location L of the fl y. Call this proposition fl y-at-L. Speaks just assumes that fl y-at-L can be expressed by some sentence S of natural language. (As we’ll see, I do not make this assumption.) On a view like mine, fl y-at-L consists of the fl y standing in some propositional relation [PR] to location L. Now, Speaks asks, what did the frog do to bring this proposition into existence? Since on my view, in the case of natural language sentences and the propositions they express, speakers endow propositional relations with truth conditions, and so bring propositions into existence, by interpreting syntactic relations and thereby interpreting propositional relations, Speaks assumes that in the present case I must say that the frog does something that constitutes interpreting the propositional relation of fl y-at-L , thereby endowing the proposition with truth conditions. Obviously it cannot be the frog interpreting sentential relations. And so we cannot view [PR] as being in part built out of sentential relations as we have been doing to this point. Given Speaks’s assumption that fl y-at-L can be expressed by a sentence S, [PR] is the propositional relation of the proposition expressed by sentence S. Th is means, Speaks thinks, that we must somehow alter our account of propositional relations so that [PR] can be the propositional relation of the proposition expressed by S and the proposition that is the content of FLY, since these are the same proposition (fl y-at-L). Suppose, Speaks says, that we have somehow done that.

13 See Siegel [2006] for a good discussion of the relevant issues and an argument to the eff ect that the class of properties that fi gure in the contents of perceptual experience is considerably wider than one might have thought. If Siegel is right, that is good news for me!

King020513OUK.indd 192 11/23/2013 12:59:06 PM RESPONSES TO SPEAKS AND SOAMES 193

But we still haven’t said how the frog brings fl y-at-L into existence by endowing [PR] with semantic signifi cance thereby endowing fl y-at-L with truth conditions. As I noted above, Speaks assumes that this will be a matter of the frog interpreting [PR]. But what constitutes his doing that? Speaks thinks the best answer I can give is that the frog interprets his perceptual experience, thereby endowing it with truth conditions/ content. And Speaks thinks that this claim is implausible. For, as Speaks notes, the phenomenal character of FLY is fi xed independently of the frog allegedly interpreting his experience (if not, Speaks rightly asks, what is the frog interpreting?). On one version of intentionalism Speaks fi nds plausible, for FLY to have a certain phenomenal character just is for it to have a certain content. 14 But then if phenomenal character is fi xed independently of the frog interpreting his experience, its content is too. Th is means that the explanation of how FLY gets its content by the frog interpreting his experience and thereby [PR] that Speaks attributes to me cannot be right. Th us, Speaks claims, I no longer have an explanation of the existence of representational properties of fl y-at-L. Before providing a detailed response to Speaks’s objection here, it will be useful to think about what it would be to provide an account of the content of some non-linguistic entity, say a map, that is in the spirit of my account of the semantic contents of natural language sentences. Suppose we are given a very simple map with points on it representing three small towns. Th ese dots are the analogues of words in a sentence and like them have semantic values; the towns they represent. Th e spatial relations between the dots in the map are the analogues of sentential relations and like the latter have semantic signifi cance; they represent spatial relations between the towns in question.15 Th e “propositional relation” of the proposition that we assign to the map will be built out of the spatial relations between the dots on the map. Just as in the case of the sentential relation and the propositional relation of a proposition expressed by a sentence, the propositional relation of the map is interpreted the way it is because the spatial relations among the dots on the map are interpreted the way they are. Crucial point: is there any guarantee that there is some notion of propositional relation such that the propositional relation of the proposition expressed by the map is the same as the propositional relation of a proposition expressed by a natural language sentence? No, and in the present case it seems to me doubtful that there is such a notion of propositional relation . If that is right, then no natural language sentence will express the same proposition as the map. Personally, I fi nd this claim quite plausible.

14 Speaks only appears to need the weaker claim here that some phenomenal properties of a perceptual experience are identical to representational properties of the perceptual experience. For if that is so, and the former are fi xed independently of the frog interpreting [PR], the latter must have been as well. And that means that at least some of the representational properties of a perceptual experience are fi xed independently of the frog interpreting his experience. 15 I don’t think this is actually quite right and things are a bit more complicated (I think the spatial relations between dots in the map are actually more like words too than like sentential relations; the analogue of a sentential relation for the map seems to me to be that the dots and spatial relations are confi gured in a certain way in the map). But this is good enough for present purposes.

King020513OUK.indd 193 11/23/2013 12:59:06 PM 194 JEFFREY C. KING

Finally, is the account of how the propositional relation of the proposition expressed by the map gets semantic signifi cance, so that the proposition represents something, going to be the same as the account of how the propositional relation of a proposition expressed by a natural language sentence gets its semantic signifi cance? Very likely not, for two reasons. First, there is no reason to think that the semantic signifi cance of the propositional relation of the proposition expressed by the map is the same as that of any propositional relation of a proposition expressed by a natural language sentence. Second, the respective mechanisms by means of which the propositional relation of the proposition expressed by the map acquires semantic signifi cance, and that by means of which the propositional relation of a proposition expressed by a natural language sentence does so, are diff erent. In the case of the propositional relation of the proposition expressed by the map, the explanation for how it got its semantic signifi cance will appeal to the means by which the spatial relations among dots in the map got their semantic signifi cance. Presumably, this will be due to a stipulation on the part of the map maker. In the case of the propositional relation of the proposition e.g. that Michael swims, the explanation for how it got its semantic signifi cance will appeal to how the syntactic relation R got its semantic signifi cance. I have suggested that the latter will be explained by our biologically endowed language faculty. So here diff erent mechanisms explain how the propositional relation of the proposition expressed by the map got its semantic signifi cance and how the propositional relation of the proposition that Michael swims got its semantic signifi cance. Th ese points carry over to the account of the contents of perceptual experiences that is in the spirit of my account of the semantic contents of natural language sentences, except that there is more uncertainty about the nature of the fi nal account in this case because of current uncertainty about various features of the nature of perceptual experience. But consider again the frog’s perceptual experience FLY that has the proposition fl y-at-L as its content. Is there a notion of propositional relation such that the propositional relation of fl y-at-L is the propositional relation of some proposition expressed by a natural language sentence? Just as was the case with the propositional relation of the proposition expressed by the map, there is no guarantee of this. Perhaps there is a general notion of propositional relation applicable to the contents of sentences and perceptual experiences. Perhaps there is not. If there is not, then no natural language sentences will express the contents of perceptual experiences and vice versa. Speaks thinks the claim that the same proposition can be expressed by a sentence and be the content of a perceptual experience is “quite plausible.” Others who work on perceptual experience disagree (I discuss this further below in responding to Soames). For myself, I have never found the claim tremendously compelling that the same proposition can be expressed by a sentence and be the content of perceptual experience. So if it turns out that on my account of content the claim comes out false, I don’t view this as a big cost. But whether or not there is a notion of propositional relation applicable to the contents of both perceptual experiences and natural language sentences, what

King020513OUK.indd 194 11/23/2013 12:59:06 PM RESPONSES TO SPEAKS AND SOAMES 195

about Speaks’s worry that the account I will be forced to give of how the frog brings the proposition fly-at-L into existence will be inconsistent with a plausible intentionalist thesis? Recall that Speaks assumed that I would have to claim that the propositional relation of the proposition that is the content of FLY—the frog’s perceptual experience—is endowed with semantic significance, thus bringing the proposition fly-at-L into existence, by the frog interpreting his perceptual experience. But, Speaks claims, the phenomenal character of FLY is fixed independently of his interpreting his experience. And it is plausible to hold the intentionalist claim that fixing phenomenal character of a perceptual experience just is fixing its content (at least in part). But then the claim cannot be right that the frog brings the proposition fly-at-L, that is the content of his perceptual experience, into existence by interpreting his experience. Hence, according to Speaks, I am left without an account of how the frog brings fly-at-L into existence and endows it with representational properties. Where Speaks’s objection goes wrong is in assuming that I will claim that the propositional relation of fl y-at-L gets its semantic signifi cance—thus bringing fl y-at- L into existence—by the frog interpreting his perceptual experience. As indicated above, I think Speaks assumed this because my account of how speakers endowed the propositional relation of the proposition that Michael swims with semantic signifi cance appeals to speakers interpreting syntactic relations. From this, I think Speaks inferred that my account of the contents of other things will always appeal to the same mechanism as to how propositional relations are endowed with semantic signifi cance, bringing propositions into existence: an agent interpreting something as a result of which the propositional relation is itself interpreted by the agent. Hence, the only plausible thing he saw for me to say in this case was that the frog interpreted his perceptual experience as a result of which the propositional relation of fl y-at-L is endowed with semantic signifi cance. However, as we saw from the map example discussed above, accounts of contents of non-linguistic things in the spirit of my account of the contents of natural language sentences need not, and in general could not, explain how the propositional relations of propositions expressed by non-linguistic entities acquire semantic signifi cance by appealing to the same mechanism whereby propositional relations of propositions expressed by natural language sentences acquire semantic signifi cance. In the present case, assuming the intentionalist claim that Speaks fi nds plausible to the eff ect that fi xing phenomenal character of a perceptual experience is fi xing its content, my account of the contents of perceptual experience can hold that whatever fi xes the phenomenal character of a perceptual experience endows its propositional relation with semantic signifi cance and brings the proposition that is its content into existence. We thereby explain how the proposition that is the content of the perceptual experience comes into existence with representational properties in a way that is consistent with the intentionalist thesis Speaks fi nds plausible.

King020513OUK.indd 195 11/23/2013 12:59:06 PM 196 JEFFREY C. KING

R e s p o n s e t o S o a m e s Soames’s fi rst objection concerns my notion of speakers having cognitive access to sentences of their languages, construed as facts in my sense; to what I called interpreted sentences in Chapter 4 of the present work and elsewhere; and to the facts that I claim are propositions. In the end my explanation for how/why propositions have truth conditions is that speakers interpret the propositional relation of the proposition that Michael swims as having a certain semantic signifi cance. As I’ve said, speakers interpret the complex two-place relation between Michael and the property of swimming as ascribing the property to Michael. Now I claimed that in order for it to be plausible that speakers interpret the propositional relation of the proposition that Michael swims, they must have some sort of cognitive connection or cognitive access to the fact that I claim is the proposition. Th e idea here seems to me simple and obviously correct. Consider some fact that I have no cognitive connection to: say some fact on the other side of the galaxy. Surely it would be very implausible to suppose that I interpret some relation that is a component of this fact in a certain way. I take it that it is implausible precisely because I have no cognitive connection to the fact in question. I also claimed that it appears that speakers have cognitive access to the fact that is the proposition that Michael swims in virtue of deploying sentences of their language, since by the time they are deploying such sentences they have propositional attitudes towards the contents of the sentences they are using. Hence, propositions must exist at that time, and that means that speakers must be interpreting their propositional relations. My thought was that speakers clearly have a cognitive connection to the sentences of the languages they employ, where I thought of sentences themselves as facts in my sense (since they are arguably properties (word types) standing in a (syntactic) relation). I then introduced the notion of an interpreted sentence, which is just the sentence taken together with the semantic relations the lexical items in the sentence bear to their semantic values, as well as the semantic values themselves. Th us, the interpreted sentence is just a slightly “larger” fact than the fact that is the sentence. I claimed that having cognitive access to a given sentence also gives one cognitive access to the interpreted sentence corresponding to it. Finally, I considered a fact consisting of

objects o1 ,. . .on standing in the n-place relation R and called it a witness for the fact of

there being x 1 ,. . .,x n such that Rx1 ,. . .,xn . In such a case, let’s say that fi rst fact is a witness

with respect to o 1 ,. . .,o n for the second. I claimed that having a cognitive connection to a witness fact is suffi cient for having cognitive access to the fact it witnesses. Now since interpreted sentences are witnesses to the facts that are the propositions expressed by the corresponding sentences (with respect to language, lexical items, context etc.), this means that having cognitive access to an interpreted sentence suffi ces for having cognitive access to the relevant proposition. We thereby have an explanation of how speakers have cognitive access to the facts I claim are propositions in virtue of deploying the relevant sentences—speakers’ cognitive access to sentences gives them cognitive access to interpreted sentences which in turn gives them cognitive access

King020513OUK.indd 196 11/23/2013 12:59:06 PM RESPONSES TO SPEAKS AND SOAMES 197

to the facts I claim are propositions. Hence we can see how speakers could interpret the propositional relations of these propositions, thereby endowing them with truth conditions. Soames objects that the notion of cognitive access/connection that is central to my explanation here of how propositions acquire truth conditions is unexplained and hence “too underspecifi ed to bear the load placed on it.”16 Further, given that I haven’t explained what sort of cognitive access I am talking about, Soames sees no reason to think that having cognitive access in the sense I require to a witness fact suffi ces for having cognitive access to the fact it witnesses. Th ere is a sense in which I agree with his complaint. In King [2009], where I fi rst introduced the notion of cognitive access/ connection in giving the explanation of how/why propositions have truth conditions, I wrote:

I’m going to continue to speak loosely about speakers having cognitive access or a cognitive connection to this or that fact. At some point I need to get more serious about the sort of cognitive connection required. But this is not that point.17 So I acknowledge that in using the notion of cognitive access/connection that I have not fully specifi ed in my explanation of how/why propositions have truth conditions, I am issuing a promissory note that at some point needs to be cashed in. Th ough I can’t fully cash it in presently, perhaps I can make a small down payment by way of saying a few things that may be of help. As we have seen, the cognitive access to the sentence, interpreted sentence and proposition that I appeal to is used to explain how we go about interpreting the sentential and propositional relations. Hence, the operative notion of cognitive access must be such that having cognitive access to a fact puts one in a position to interpret features of the fact in certain ways. You must in some sense have the fact in mind in such a way as to be in a position to interpret its features. But, as Soames noted, since the cognitive access in question is being used to explain how propositions expressed by natural language sentences are representational, and so have truth conditions, the cognitive access in question cannot be a matter of having any propositional attitudes towards propositions expressed by natural language sentences (though it could involve having attitudes towards the contents of perceptual experiences). Now I take it that it is clear that speakers have sentences, and even interpreted sentences, in mind in the relevant sense. Th ey are cognitively in a position to interpret the sentential relations of these facts. But now why think they are thereby in a position to interpret the propositional relations of the facts that I claim are propositions, which the relevant interpreted sentences witness?18 Th at is, why think that if one has a witness

16 Chapter 9 p. 171 17 p. 269 18 Soames pointed out that interpreted sentences are not quite witnesses for the relevant propositions. But he and I agree, I think, that that they are close enough to being so that the ways in which they fall short won’t aff ect my argument.

King020513OUK.indd 197 11/23/2013 12:59:06 PM 198 JEFFREY C. KING

fact in mind in the relevant sense, one is thereby in a position to have the fact witnessed in mind? Here is a tentative, rough suggestion.When one has a given fact in mind, one

is able to abstract from certain features f 1 ,. . .,fn of a fact one has in mind, thereby having

in mind the fact witnessed with respect to f1 ,. . .,f n by the original fact. Not only does it seem plausible that one is able to abstract from features of a witness fact in this way, thereby having the fact it witnesses in mind, but it also seems plausible that we do this in moving from having a sentence in mind to interpreting the propositional relation of the proposition expressed by the sentence and thus understanding the sentence. Consider the sentence “Michael swims”:

Michael swims

and suppose a speaker has the sentence in mind. Th e speaker then accesses the semantic values of the sentence—relative to the context of utterance—with the result that the speaker now has the interpreted sentence in mind:

1IE.

Michael swims

2007 2007 2007 2007 20072007

Melbourne 2007 Melbourne 20 Melbourne 200

Melb 20 Melbourne Melbourne 2007Melbourne 200

At this point, the speaker abstracts from all the now irrelevant features of the interpreted sentence: the words in it, what language it is in, the context in which the utterance was made, and so on. Th e speaker thereby has the proposition that Michael swims in mind, which the interpreted sentence witnessed with respect to a language, context, lexical items and so on. Hence, the speaker is in a position to interpret its propositional relation, and thereby entertain the proposition.

King020513OUK.indd 198 11/23/2013 12:59:06 PM RESPONSES TO SPEAKS AND SOAMES 199

As I said, this account, of why having cognitive access to the witness for a fact puts one in a position to have cognitive access to the fact it witnesses, is tentative and it may be that there is a better account to be found. However, it does seem to me plausible, along the right lines and hence promising so far. Let me close my response to Soames’s present objection by noting that he doesn’t seem well placed to make this sort of objection to my view. He objects that I use an unexplained notion, cognitive access/connection, as a central part of my explanation as to how/why propositions have truth conditions. As indicated, he complains that the notion “has been left too underspecifi ed to bear the load placed on it.”19 But central to Soames’s own explanation of how/why propositions have truth conditions is his notion of predication , which he explicitly takes to be primitive. Hence Soames’s complaint about a primitive/unexplained notion playing a central role in the explanation of how/why propositions have truth conditions applies to his own view. Soames’s second objection to my view concerns what it is to entertain e.g. the proposition that Michael swims. As we have just seen, one way to entertain the proposition starts with having cognitive access to the sentence “Michael swims,” which I have just cashed out in terms of having the sentence in mind in such a way as to be in a position to interpret its sentential relation. One then is in a position to have the interpreted sentence mentioned above in mind. Th is in turn places one in a position to have the proposition itself in mind and interpret its propositional relation as ascribing the property of swimming to Michael, thereby entertaining the proposition. Soames objects that the fi rst step here, having the sentence “Michael swims” in mind, already presupposes that one understands the sentence and so entertains the proposition. Hence he claims my account of entertaining the proposition that Michael swims already assumes at the fi rst step what it sets out to explain. As the remarks just made in response to Soames’s fi rst objection hopefully made clear, the fi rst step of my explanation of one way to entertain the proposition that Michael swims—acquiring cognitive access to the sentence “Michael swims”—does not presuppose that one understands the sentence. As I indicated, it just requires one to have the sentence in mind and so be in a position to interpret the sentential relation . It does not e.g. require one to actually do so. Th e idea is that understanding a sentence begins with having it in mind in such a way as to be in a position to interpret its sentential relation. So having the sentence in mind in this sense just does not require understanding it. Looking through Chapter 4, I couldn’t fi nd any passages in which I said or implied that having cognitive access to a sentence required understanding it, but perhaps it is somehow suggested by something I said. In any case, it isn’t my view. So Soames is wrong that the fi rst step of my explanation of this way of entertaining the proposition that Michael swims presupposes that one already entertains the proposition.

19 Chapter 9 p. 171

King020513OUK.indd 199 11/23/2013 12:59:11 PM 200 JEFFREY C. KING

In making the above objection Soames raises some other issues that are worth clear- ing up. He cites a passage from Chapter 4 in which I am trying to explain why the propositional relation of the proposition that Michael swims is interpreted in the way it is, endowing the proposition with truth conditions. 20 Soames seems puzzled by what I say there, so let me try again. I said there that interpreting the propositional relation of the proposition that Michael swims as ascribing the property of swimming to Michael is a refl ex of the fact that the sentential relation of the sentence “Michael swims” has the semantic signifi cance it does. Let me try to explain this idea. As we have seen, one way of entertaining the proposition that Michael swims is to fi rst have the sentence “Michael swims” in mind. (Recall that the sentential relation of this sentence has a certain semantic signifi cance: it ascribes the semantic value of “swims” to the semantic value of “Michael.”) One then accesses the semantic values of “Michael” and “swims.” One thereby has the interpreted sentence 1IE above in mind. Abstracting from the fact that “Michael swims” is an English sentence, containing certain lexical items, that was uttered in a certain context and so on, one thereby has the fact that is the proposition that Michael swims in mind. Only now are semantic values composed, and so the semantic signifi cance of the sentential relation of “Michael swims” is now cashed in. Th ereby, the property of swimming is ascribed to Michael. However, since it is the fact that is the proposition that Michael swims that we have in mind when semantic values are composed, we count as interpreting its propositional relation as ascribing the property of swimming to Michael. So interpreting the sentential relation in the way we do, by composing semantic values in a certain way at a certain point in the process just described, just is interpreting the propositional relation in the relevant way. Soames’s third objection concerns the contents of natural language sentences and the contents of perceptual experiences. Soames thinks that, on my view, the class of propositions that are contents of perceptual experiences and the class of propositions that are the contents of natural language sentences will likely be disjoint. As I indicated in responding to Speaks, this certainly could be a consequence of my view. Soames thinks there are two potential problems with this consequence. First, a view with this consequence may not be able to explain the fact that we appear to use sentential clauses in characterizing the contents of perceptual experiences. 21 Second, a view with this consequence may have problems explaining how we integrate propositional information we acquire through language, perception and thought. As for the fi rst worry, Soames seems to be thinking that on a view like his where the contents of perceptual experiences are also the contents of natural language sentences, it is easy to explain why we use sentential clauses to characterize the contents

20 Chapter 9 p. 173 21 Soames puts the point by saying that views with the consequence in question might have problems giving an account of “how sentential clauses are used to report the contents of perceptual . . . experiences.” But putting it this way seems to beg the question against views on which sentential clauses are not used to literally report the contents of perceptual experiences (e.g. because contents of sentential clauses are never the contents of perceptual experiences).

King020513OUK.indd 200 11/23/2013 12:59:12 PM RESPONSES TO SPEAKS AND SOAMES 201

of perceptual experiences: we do so because in many cases the content of a sentential clause is the (partial) content of a perceptual experience. Of course if, on my view, it turns out that the set of propositions expressed by natural language sentences and the set of propositions that are the contents of perceptual experiences are disjoint, I cannot appeal to this simple account. However, the problem with the simple account is that, as I indicated in Chapter 7, the literature on perceptual experience makes clear that there are all sorts of reasons why the contents of perceptual experiences might not be expressible by sentential clauses. First, perceptual experiences might not have contents; second, they might not have propositional contents; third, they might have propositional contents not expressible by natural language sentences. Th e third possibility seems likely if, as many now think, perceptual experience has so-called non-conceptual content. For it is plausible to suppose that the non-conceptual contents of perceptual experiences would not be expressible by natural language sentences. Since the simple account that Soames favors of why we use sentential clauses to characterize perceptual experience might not be correct for all sorts of reasons, the fact that my view of content might require me to reject it is not very worrisome. Regarding the second worry, Soames’s idea seems to be that if the contents of perceptual experiences are diff erent from the contents of natural language sentences, it might be hard to explain how we integrate information received from language and perception. I suspect this isn’t much of a worry either. As I indicated above in responding to Speaks, I fi nd it overwhelmingly plausible that the contents of maps are not expressible by means of natural language sentences. Yet we have no trouble integrating information derived from maps and from language. Similarly, I suspect at least certain kinds of pictures have semantic content and I am doubtful that these contents are expressible by natural language sentences. But here again, we have no trouble integrating information derived from pictures and language. Of course, it would be nice to explain how we do this. But we do it and this suggests to me that we have some way of easily and quickly integrating diff erent kinds of content. Soames’s fi nal critical remarks concern the semantic values of expressions like “the fact that Michael swims,” “the proposition that Michael swims” and “that Michael swims.” In Soames [2012], he based an argument against my view of propositions on the premise that I am committed to the claim that “the fact that Michael swims” and “the proposition that Michael swims” designate diff erent things, and that “that” clauses can designate either. I responded to this criticism in Chapter 4, in part by denying that I am committed to the above claims, and Soames now agrees that I am not. But also in Chapter 4, for the sake of argument, I supposed that I was committed to the claim and off ered evidence in its favor. I also tried to explain away evidence Soames gave for the claim that the expressions in question may designate the same thing. In Chapter 4, I gave three kinds of evidence that expressions of the form “the fact that. . .” and expressions of the form “the proposition that. . .” designate diff erent kinds of things.22 Th e fi rst is that while expressions of the form “the fact that. . .” can felicitously

22 See Chapter 4 pp. 66–68

King020513OUK.indd 201 11/23/2013 12:59:12 PM 202 JEFFREY C. KING

occur in so-called factive contexts and cannot so occur in non-factive contexts, the opposite is the case for expressions of the form “the proposition that. . . .”23 Th e second is that expressions of the form “the fact that. . .” can happily occur in subject position in “causal statements” (“Th e fact that the brick hit the window caused it to break”), whereas expressions of the form “the proposition that. . .” cannot. Th e third is that quantifying across a factive and non-factive context is usually terrible, (e.g. *’Everything Glenn regrets John believes.’); whereas quantifying across two factive or two non-factive contexts is virtually always fi ne (“Everything John says Glenn believes.”/“Everything John regrets Glenn discovers.”). 24 C a l l t h e fi rst two sorts of evidence the substitution data; and call the third the quantifi cational data . Th e most straightforward explanation of the substitution and quantifi cational data is that e.g. “the fact that the brick hit the window” and “the proposition that the brick hit the window” designate diff erent sorts of entities; that factive contexts predicate properties appropriate for what the former designate and not the latter, whereas the opposite is true of non-factive contexts; and that causal statements express relations between what the former designate and something else and not what the latter designate and something else. Now curiously, Soames didn’t comment on the quantifi cational data. But he agrees that the substitution data provides prima facie reason for thinking the relevant expressions designate diff erent things. However, he worries that what he calls “substitution tests” can “cut in diff erent directions,” as when you take the true sentence “Fred believes that Michael swims” and substitute for the “that” clause a defi nite description (or some other expression) designating the fact that I claim is the proposition that Michael swims. Th e result appears absurd or false. But then, if I think the substitution data shows that expressions of the form “the fact that. . .” and “the proposition that. . .” designate diff erent things, shouldn’t I take this to show that the “that” clause and the defi nite description designate diff erent things, so that the “that” clause does not designate the fact I claim is the proposition that Michael swims? Call this the substitution problem. Of course, as Soames notes, he, Speaks, and I all have the same problem here. When you substitute a certain defi nite description designating the thing any of us claims is the proposition that Michael swims for “that Michael swims” in the sentence above, the result is seemingly absurd or false. Th e substitution problem leaves Soames uncertain about whether the substitution data supports the conclusion I claim it does. If you think, as I do, that the substitution data is strong support for the conclusion that expressions of the form “the fact that. . .” and “the proposition that. . .” designate diff erent kinds of things, what is wanted is to fi nd some diff erence between that data

23 Since bare “that” clauses can occur in both, this provides evidence that they can designate either the things designated by expressions of the form “the fact that. . .” or the things designated by expressions of the form “the proposition that. . . .” 24 I say “virtually always fi ne,” because you can always create some sort of pragmatic anomaly, as with “Everything John believes Glenn says.” I take it this sounds a bit odd out of context because of the strange- ness of one person saying everything another person believes. But even here you can generally rig the context to make it okay. Imagine the sentence preceded by “I know why John likes Glenn so much.”

King020513OUK.indd 202 11/23/2013 12:59:12 PM RESPONSES TO SPEAKS AND SOAMES 203

and the data cited in the substitution problem that suggests diff erent things are going on in the two cases. I think there is such a diff erence. In the case of the substitution data, we have expressions of ordinary English that someone claims designate the same thing such that when we substitute them for each other in the contexts I described we get the various anomalies discussed in Chapter 4 and above. Th e explanation I have given of this data, which again involves expressions of ordinary English, is surely the most natural and straightforward. To repeat, the reason expressions of the form “the fact that. . .” can occur in factive contexts and not in non-factive contexts and the reverse is true of expressions of the form “the proposition that. . .” is that expressions of these two forms designate entities of diff erent sorts; and factive contexts predicate properties of the one sort of thing, whereas non-factive contexts predicate properties of the other sort of thing. Finally, the properties felicitously predicated of the one sort of thing cannot be felicitously predicated of the other sort of thing. Th is will be discussed further below, but this explanation is supported by the quantifi cational data as well; if factive contexts predicate properties of one kind of entity and non-factive contexts predicate properties of another kind of entity, and the properties appropriately predicated of the one kind of entity cannot be appropriately predicated of the other kind of entity, then we wouldn’t expect to be able to quantify into both a factive and a non-factive context with a single quantifi er. Th e examples given above and others show that this is indeed what we fi nd in ever so many cases. However, again, Soames did not comment on the quantifi cational data in responding to me. Finally, the causal statement data suggests the same thing. Th e reason expressions of the form “the fact. . .” can occur in subject position in causal statements and expressions of the form “the proposition that. . .” cannot is that these expressions designate diff erent sorts of things and only things designated by the former can cause things. Th e data in the substitution problem , on the other hand, does not involve two expressions of ordinary English that someone claims designate the same thing. It involves one expression of ordinary English (a “that” clause) and another expression in “philosophiclese” that someone claims designates the same thing as the ordinary language expression and that purports to provide something like an analysis or philosophical characterization of the thing designated by the ordinary language expression. When we take a true sentence containing the expression of ordinary English and substitute the expression of philosophiclese, we get something seemingly absurd. Th e reason that this seems like a very diff erent phenomenon from the substitution data is that the present phenomenon is ubiquitous in attempts to provide philosophical analyses or characterizations of things. But then you would expect the explanation of the data here to involve the nature of philosophical analysis or characterization. It would seem crazy to think the explanation of the substitution data would involve such notions. To see that the phenomenon is ubiquitous in attempts to give philosophical analyses, consider the following two examples:

King020513OUK.indd 203 11/23/2013 12:59:12 PM 204 JEFFREY C. KING

1. Some have been attracted to the view that people are material-fi lled regions of space-time. But consider the following two sentences: a. I had lunch with Annie b. I had lunch with material-fi lled region x of space-time. Since a is true and b is absurd, “Annie” and “material-fi lled region x of space-time” designate diff erent things. So Annie is not material-fi lled region x of space-time. 2. Some have thought that events are properties of time intervals. But consider the following two sentences: a . Th e event of Caesar’s dying took place in Rome in 44 BC. b . Th e property of being a time interval during which Caesar died took place in Rome in 44 BC. Since a is true and b is absurd, “the event of Caesar’s death” and “the property of being a time interval during which Caesar died” do not designate the same thing. So the event of Caesar’s death is not the property of being a time interval during which Caesar died. It is oft en the case that when one can give arguments like this, one can give the argument against virtually any proposed analysis. We noted above that the substitution problem affl icts the theories put forward by Speaks, Soames and me. It would also affl ict any other theory of propositions. But then if such arguments worked across the board, they would show the impossibility of any philosophical analysis of the relevant notion. Th at seems absurd. So a resolution of the present issues, including the substitution problem, would require reconciling the possibility of a correct philosophical analysis with data like 1 and 2 above (I reject the analysis in 2 at any rate, but not because of the above arguments). To repeat, to think that the substitution data has anything to do with philosophical analysis seems to me on its face to be ludicrous. Th is strongly suggests that the substitution problem and the substitution data involve quite diff erent mechanisms. In turn, this means that the substitution problem should not lead us to question the natural, straightforward explanations off ered above of the substitution data, contra Soames.25 Th e other critical remarks Soames makes concerning expressions of the form “the proposition that. . .” and “the fact that. . .” concerns the evidence he gave in Soames [2012] for the claim that they designate the same sort of entity, and sometimes even

25 On the one hand, I am expressing skepticism here that the substitution problem really is a problem for any of my, Speaks’s and Soames’s theories of propositions, and more generally that arguments like 1 and 2 above should be taken seriously. On the other hand, in Chapter 7 I gave arguments against Soames’s view that event tokens and types have truth conditions, where these arguments might seem to resemble 1 and 2 above. But they really don’t. Th e important feature of the substitution problem and arguments 1 and 2 above is that they involve substituting an expression of “philosophiclese” for an ordinary language expression that is claimed to designate the same thing. By contrast, my arguments against Soames in Chapter 7 involved only expressions of ordinary language (“the proposition that o is red”; “what Godel proved”; “what just occurred”). Hence they are importantly unlike the substitution problem and arguments 1 and 2 above.

King020513OUK.indd 204 11/23/2013 12:59:12 PM RESPONSES TO SPEAKS AND SOAMES 205

the same entity, which I discussed in Chapter 4. Soames [2012] off ered the following sentence 9. Pam regrets that she is pregnant. Although her parents don’t realize it yet, in time they will come to believe it . Since the fi nal “it” seems to have “that she is pregnant” as its antecedent, it presumably designates the same thing as the latter. But this seems to show that the thing regretted and the thing believed are the same, contrary to what I have claimed. In Chapter 4, I described two strategies for explaining the felicity of 9 in a way that is consistent with the view that things regretted are diff erent kinds of things from things believed. In Soames’s critical remarks in Chapter 9, he criticizes those strategies. Th ough I continue to think both strategies are plausible despite Soames’s criticisms, I won’t re- litigate that case except for one point. Soames criticizes my fi rst strategy by saying that it isn’t general enough to handle the following example in which there is quantifi cation into a factive and non-factive context: 10.b. Something Pam regrets is now merely suspected by her parents but will soon be believed by them. Th is appears to show that one and the same thing is regretted and believed. But Soames apparently overlooked the fact that in Chapter 4 I explicitly discussed how the fi rst strategy could be extended to handle examples like 10b. 26 Since this is the only objection Soames makes to my fi rst strategy, this leaves him with no objection to it. In any case, rather than trying to provide more detail as to how to explain 9 and 10b in a way that is consistent with the view that factive and non-factive contexts predicate properties of diff erent kinds of things, let me focus on why we should explain 9 and 10b in this way. Th at is, I want to explain why we should not take 9 and 10 to show that factive and non-factive contexts sometimes predicate properties of the same thing, but instead consider it to be anomalous data that should be explained by maintaining the view that factive and non-factive contexts predicate properties of diff erent kinds of things. Th ough I discussed this in Chapter 4, I wish to be a bit more explicit here. Th e fi rst point to make, which I made in Chapter 4, is that the substitution data and quantifi cational data provide lots of quite good evidence that expressions of the form “the fact that. . .” and “the proposition that. . .” designate diff erent kinds of things and that factive contexts predicate properties appropriate to only what the former designate, while non-factive contexts predicate properties appropriate to only what the latter designate. Hence, this gives us good reason to think that examples like 9 and 10b should be explained in a way that is consistent with these claims. But the second point to make is that even if we confi ne our attention to data of the sort Soames’s 9 and 10b represent, careful consideration of that data actually supports

26 See my discussion in Chapter 4 of example 10 on p. 70.

King020513OUK.indd 205 11/23/2013 12:59:12 PM 206 JEFFREY C. KING

the claims that expressions of the form “the fact that. . .” and “the proposition that. . .” designate diff erent kinds of things and that factive contexts predicate properties appropriate to only what the former designate, whereas non-factive contexts predicate properties appropriate to only what the latter designate. To see this, let me fi rst note that we really should be considering data simpler than 9 and 10b. In the case of 9, we want to see whether a pronoun in a non-factive context can felicitously have as its antecedent a “that” clause in a factive context. Inserting a clause containing negation and an anaphoric pronoun in a factive context (“her parents don’t realize it yet”) whose antecedent is the “that” clause in the fi rst factive context in 9 (and a similar clause in 10b) between the two clauses we are interested in can only produce noise. Aft er all, the pronoun in “her parents don’t yet realize it” should be felicitous (the pronoun and its antecedent are both in factive contexts), and this may improve the felicity of the fi nal pronoun (I’ll add that a fair number of informants I checked with, already found 9 “not so good”). So to avoid noise produced by unnecessary and irrelevant complexity, let’s try the most minimal example relevant: namely an example with just one factive context, one non-factive context and no negation or anything else. Consider the following, focusing on the reading where “it” is anaphoric on “that she is pregnant” (and not on “Pam regrets that she is pregnant”—see note 27): 9’. *Pam regrets that she is pregnant. Her parents believe it. I fi nd this quite awkward. When we try other such examples, with a very few excep- tions (discussed below), they all sound bad to varying degrees: 9’a. *Pam resents that she is pregnant. Her parents doubt it. 9’b. *Pam hates that she is pregnant. Her parents think it. 9’c. *Pam comprehends that she is pregnant. Her parents assume it. 9’d. *Pam forgot that she is pregnant. Her parents denied it. 27 9’e. *Pam made clear that she is pregnant. Her parents said it. On the other hand, it is easy to produce examples that are immaculate with two factives or two non-factives: 9’f. Pam likes that she is pregnant. Her parents hate it. 28 9’g. Pam remembered that she is pregnant. Her parents forgot it. 9’h. Pam resents that she is pregnant. Her parents like it. 9’i. Pam is aware that she is pregnant. Her parents resent it. 9’j. Pam hopes that she is pregnant. Her parents doubt it.

27 Th is, like some (or maybe all?) of the other asterisked sentences, is okay when “it” takes as its antecedent the entire fi rst conjunct (“Pam forgot that she is pregnant.”). But this reading is predicted to be fi ne and so irrelevant. Recall that we are focused on the reading where the antecedent of the pronoun is the “that” clause (“that she is pregnant”). Here I have gone to past tense because the present tense (“Pam forgets she is pregnant.”) has a habitual-like reading and we are trying to avoid noise. I do the same thing in at least one other case for similar reasons. 28 Th anks to Annie King for the excellent example.

King020513OUK.indd 206 11/23/2013 12:59:12 PM RESPONSES TO SPEAKS AND SOAMES 207

9’k. Pam thinks that she is pregnant. Her parents deny it. 9’l. Pam says that she is pregnant. Her parents believe it. 9’m. Pam denies that she is pregnant. Her parents assert it. Surely, the weight of the data here is on the side of the claim that pronouns in non-factive contexts in general cannot have as their antecedents “that” clauses in factive contexts. Th e ease with which we can get immaculate examples with two factives or two non-factives and the diffi culty of getting immaculate examples with mixes of factives and nonfactives strongly suggests that such mixed cases are in general not licensed. Th is, of course, is just what the view that factive and non-factive contexts predicate properties of diff erent sorts of things predicts. But then this means that when we fi nd some small number of examples where a pronoun in a non-factive clause is felicitously anaphoric on a “that” clause in a factive context, we should consider it anomalous data and try to explain it in a manner that is consistent with the claim that this generally cannot happen. Th at is, we should explain the data away and not take it to show that factive and non-factive clauses predicate properties of the same sorts of things. Th is includes Soames’s 9 to the extent that one does fi nd it felicitous.29 It also includes felicitous examples like the following, which are few and far between: 9’o. Pam knows that she is pregnant. Her parents believe it. 30 Th is is the best example of a mixed case I could come up with and it is hard to come up with any other case (not involving “knows”) that is reasonably good. It is interesting that it involves “know” and “believe” in light of the quantifi cational data discussed in Chapter 4 and below (see note 31 below). Virtually everything I have said about Soames’s 9 applies to his 10b. Again, to avoid noise, we should instead consider simpler examples. I fi nd the most natural examples of quantifying across multiple factive or non-factive contexts are those involving uni- versal quantifi ers, as in “Everything John says Frank believes.” Now let’s try the simple analogue of Soames’s example of this form: 10’ *Everything Pam regrets her parents believe. Here, I think there is a clear judgment that this is bad. Again, when we consider other examples involving a factive and non-factive context, they are bad to varying degrees (I’ve followed each with an example of two factive contexts for contrast; I’m also being a bit briefer here than I was in the case of 9):

29 I’ve already said that many informants fi nd Soames’s original example at least a bit awkward; and the simpler, less noisy 9’ is quite awkward. Th is suggests that to the extent Soames’s 9 is better, this needs to be explained. I suggested above that perhaps having an intervening clause with a factive context containing a pronoun anaphoric on the “that” clause, which should be felicitous, helps explain why the fi nal pronoun in 9 is not too bad. 30 Even here, I fi nd a bit of awkwardness as compared to “Pam knows that she is pregnant. Her parents believe that she is.”

King020513OUK.indd 207 11/23/2013 12:59:12 PM 208 JEFFREY C. KING

10’a. *Everything Pam resents her parents claim. 10’af. Everything Pam resents her parents forget. 10’b. *Everything Pam realizes her parents think. 10’bf. Everything Pam realizes her parents remember. 10’c. *Everything Pam discovered her parents said. 10’cf. Everything Pam discovered her parents regret. 10’d. *Everything Pam learned her parents asserted. 10’df. Everything Pam learned her parents remembered. Here again the data comes down heavily on the side of the claim that you cannot quantify across a factive and a non-factive context and so, again, on the side of the claim that there is no one thing of which both a factive and a non-factive context predicates a property. So, again, when we fi nd some small number of examples in which we do seem to be able to felicitously quantify across a factive and non-factive context, we should explain these examples while retaining the view that factive and non-factive contexts predicate properties of diff erent things. 31 Th is includes Soames’s example 10b above. Turning now to rather diff erent issues, I’ll close the present chapter by discussing the notion of propositional constituency that drops out of my account of propositions and what I view as some desirable features of that notion. It will be useful to proceed by considering a paper by Joshua Armstrong and Jason Stanley (henceforth J&J) [2011] in which they argue against standard Russellian accounts of singular thought. In so doing, J&J bring up some issues concerning propositional constituency that will help illuminate my account of the latter and some of its attractive features. J&J claim that the notion of constituency is taken as primitive by Russellians.32 Th ey also claim that there is pressure on the Russellian to hold that constituency is transitive: if x is a constituent of proposition P and P is a constituent of proposition Q, then x is a constituent of Q. Th e pressure for endorsing this, J&J claim, comes from the fact that if the proposition that John runs contains John as a constituent, so do the propositions that it is not the case that John runs and that John runs or Sue walks. In other words, if o is a constituent of proposition P and P is a constituent of a complex proposition Q built up out of P, truth functions and other propositions, then o is a constituent of Q. Th ough I agree with this latter claim, we’ll see that it does not force one to hold that propositional constituency is transitive. Indeed, we’ll see that neither of the claims J&J make about propositional constituency hold given my characterization of the notion—I will not take propositional constituency as primitive and it is not transitive. Assuming that constituency is transitive for the reasons given, J&J argue that one can grasp a proposition containing an object as a constituent without being acquainted with that object, as follows:

31 Again, the examples that seem most felicitous involve “knows” and a non-factive. Th e best seem to involve “believes” as in “Everything John knows Frank believes.” 32 J&J [2011] p. 220

King020513OUK.indd 208 11/23/2013 12:59:12 PM RESPONSES TO SPEAKS AND SOAMES 209

Let “Harry” name the proposition that Hannah is a philosopher. Harry contains Hannah as a constituent. Suppose that John tells Bill that he is happy that Harry is true. John knows that “Harry” refers to a proposition, but is not himself acquainted with Hannah. If Bill trusts John, it seems that he can come to know by testimony that John is happy that Harry is true, and a fortiori grasps that proposition. But Harry contains Hannah as a constituent, and by transitivity, so does the proposition that John is happy that Harry is true. So Bill’s thought that John is happy that Harry is true is not a singular thought about Hannah, though it has as its content a Russellian singular proposition containing Hannah as a constituent. Bill can therefore grasp a proposition containing an object as a constituent, without having acquaintance with that constituent.33 Th is argument fails given my account of propositions and the resulting account of propositional constituency. Seeing how and why will illuminate these accounts of mine. For simplicity, let’s make one change in J&J’s example and let “Harry” name the proposition that Hannah swims (instead of the proposition that Hannah is a philosopher—things are simpler if we consider a proposition with only two constituents). Consider the sentence “Hannah swims” represented in tree form:

11.

Hannah swims

Call the syntactic relation between subject and predicate here R . As we have seen, on my view R will be a component of the propositional relation of the proposition expressed by 11, which can be represented as follows:

11P.

33 J&J [2011] p. 220

King020513OUK.indd 209 11/23/2013 12:59:12 PM 210 JEFFREY C. KING

where Hannah is at the left terminal node of the propositional relation and the property of swimming is at the right terminal node. As should be familiar, the syntactic relation R provides all of the signifi cant structure to the proposition 11P. Th e vertical lines here represent the semantic relation being the semantic value of ___ relative to f and c . Now it is pretty clear that given our theory of propositions and propositional relations, we don’t have to take constituency as primitive as J&J claim Russellians must do. We can characterize rigorously what it is to be a constituent of a proposition. Let’s begin with what I call the simple constituents of a proposition: all entities at terminal nodes of the propositional relation of the proposition P are simple constituents of P. Th us, 11P’s simple constituents are Hannah and the property of swimming, just as one would want and expect. For more complex propositions, it is probably useful to have a notion of complex constituents of propositions. In order to characterize these, we need a bit of terminology. Propositional relations, like the syntactic relations that give them their structure, have nodes. A node (immediately) dominates its daughter nodes. A node dominates the daughters of every node it dominates. We can now characterize complex constituents of a proposition as follows: for every non-terminal node n in the tree of the propositional relation of the proposition P, the subtree rooted in n that includes all simple constituents at terminal nodes of P domi- nated by n is a complex constituent of P. Given a proposition P, whose propositional relation looks like this:

the sub-trees rooted in nodes 1 and 2 are complex constituents of P. Th ere would be four simple constituents of P occupying the terminal nodes of the propositional relation. Let’s return to J&J’s example. “Harry” is the name of the proposition that Hannah

swims. On the present account, Harry has Hannah as a simple constituent. Where P H,s is the proposition that Hannah swims, the proposition that John is happy that Harry is true, expressed by the sentence “John is happy that Harry is true.”, looks roughly as follows (some detail suppressed, nodes numbered and italicized expressions representing the semantic values of those expression (relative to parameters)):

12P. 1 2 3 4 John happy that

PH,s True

King020513OUK.indd 210 11/23/2013 12:59:14 PM RESPONSES TO SPEAKS AND SOAMES 211

As you would expect, given that the name “Harry” has the proposition that Hannah swims as its semantic value, the sentence “John is happy that Harry is true” expresses a proposition that has the proposition that Hannah swims as a simple constituent, along with John, happy, that, and true. But Hannah is not a simple constituent of 12P, since she does not occupy a terminal node of the propositional relation of 12P. Nor of course is she a complex constituent: those are the sub-trees of 12P rooted in nodes 2, 3 and 4 (and 1, if we don’t rule out a proposition being a complex constituent of itself). So if, in J&J’s example, Bill knows that John is happy that Harry is true without being acquainted with Hannah, it just doesn’t follow that Bill grasps a proposition (12P) that has an object as a constituent (Hannah) that Bill is not acquainted with. Th us, J&J’s argument that one can grasp singular propositions with respect to o without being acquainted with o fails on the present view of propositions and constituency.

On the present view, in cases like 12P, a proposition, P H,s, is a (simple) constituent of another proposition, 12P, where the former’s (simple) constituents aren’t (simple) constituents of the latter. (So constituency is not transitive on the present view. I’ll return to this below.) As a result, the following sentences express diff erent propositions, since the proposition expressed by 13b has a (simple) constituent (Hannah) not had by the proposition expressed by 13a:

13a. John is happy that Harry is true. 13b. John is happy that the proposition that Hannah swims is true.

And importantly, there is a point to distinguishing these propositions. It seems plausible to claim that to grasp a proposition one must be acquainted with its constituents.34 If we adopt this view, it follows that in J&J’s story, Bill does not grasp or know the proposition expressed by 13b, which does have Hannah as a simple constituent on the present view (since Bill fails to be acquainted with Hannah). But he may grasp and know the proposition expressed by 13b (it depends on the details of the story and the particular account of acquaintance one adopts). Surely these results seem quite plausible. Th us not only does the present view of propositions distinguish between the propositions expressed by 13a and 13b (as well as between the propositions expressed by “Harry is true” and “Th at Hannah swims is true.”), but, coupled with the plausible view that to grasp a proposition one must be acquainted with its constituents, the present view of propositions and constituency has it that one can grasp the proposition expressed by 13a without grasping the proposition expressed by 13b. And this seems intuitively correct. Consider a related example discussed in Richard [1993]:

14a. Russell defended logicism. 14b. Russell defended the claim that arithmetic reduces to logic.

34 Of course J&J attempted to argue against this view, but as we saw, their argument fails on my view of propositions and constituency. See also Hawthorne and Manley [2012] for a dissenting view.

King020513OUK.indd 211 11/23/2013 12:59:15 PM 212 JEFFREY C. KING

On the present view, 14a and 14b express diff erent propositions in just the way that 13a and 13b do (the proposition expressed by 14b has (simple) constituents not had by the proposition expressed by 14a: e.g. the reducing relation). Suppose Glenn knows that logicism is a doctrine to the eff ect that there is some intimate relationship, he isn’t sure what, between arithmetic and logic. Suppose he would have no idea what it means to say that arithmetic reduces to logic. Finally suppose he has heard that Russell was a defender of logicism. Th en it seems to me quite plausible that Glenn knows the proposition expressed by 14a but not that expressed by 14b. If we were to say this (and hold on to the view that to grasp a proposition, one must be acquainted with its constituents), we would have to say that in knowing the proposition expressed by 14a, Glenn is acquainted with the proposition that arithmetic reduces to logic but not with at least some of its constituents (e.g. the reduction relation). Again, this strikes me as a quite plausible thing to say. Finally, note that for these reasons, we should say that 13a and 13b can diverge in truth value. If John can be acquainted with Harry without being acquainted with some of its constituents (e.g. Hannah), then John could grasp the proposition that Harry is true without grasping the proposition that the proposition that Hannah swims is true. But then in such a situation 13a would be true and 13b would be false (assuming that to be happy that P requires grasping P, as seems undeniable). On the present view, that one can grasp the proposition expressed by 14a and not grasp the proposition expressed by 14b is an instance of the more general phenomenon whereby in order to grasp a proposition one is required to have a robust cognitive connection to its constituents, but not to things that are in some sense parts of those constituents.35 To take the simplest case, to grasp 13a I must be acquainted with John but I need have no acquaintance whatsoever with the metal screws in his ankle. A more interesting case, given certain assumptions, concerns properties and relations. Suppose that some properties and relations are complex and have other properties as components. Suppose, for example, that the property of being an instance of knowledge is the property of being a belief that is true and justifi ed. On this way of thinking the property of being justifi ed is a component of the property of being an instance of knowledge. Of course, the property of being justifi ed may itself be complex and have components. Now suppose that e.g. Michael Smith [1997] is right in claiming that the relation x has a reason to Ψ is the complex relation of x being such that in nearby possible worlds in which she has a maximally informed, consistent and unifi ed set of desires, she desires to Ψ. Th is complex property has as components the (complex) property of being a maximally informed, consistent and unifi ed set of desire, that of being a possible world and so on. Now the proposition expressed by: 15. Glenn has a reason to ski.

35 See King [2002b] and King [2007] Chapter 7 for discussion.

King020513OUK.indd 212 11/23/2013 12:59:15 PM RESPONSES TO SPEAKS AND SOAMES 213

has as a constituent the relation x has a reason to Ψ, but it does not have the properties of being a maximally informed, consistent and unifi ed set of desires or of being a possible world as constituents , even though the latter are components/parts of the former. Th us it is that John can grasp the proposition expressed 15 while having no notion of a maximally informed, consistent and unifi ed set of desires nor of a possible world. He simply has a much more robust epistemic connection to the complex property x has a reason to Ψ than he does to some of its components. Further, the account of constituency being discussed resolves at least some of the issues involved in the paradox of analysis, as is argued in King [2007]. One issue is how, if knowledge is justifi ed true belief, the following can express diff erent propositions: 16a. Knowledge is knowledge. 16b. Knowledge is justifi ed true belief. On the present account it is easy to see how this is so. Th e proposition expressed by 16a does not contain the property of being justifi ed as a constituent (it is rather a component of a constituent of that proposition, but it is itself neither a simple nor complex constituent of the proposition); that expressed by 16b does. Th ey thereby are diff erent propositions and have diff erent requirements on being grasped. Of course, this does not resolve all the issues surrounding the paradox of analysis, but it is a good start. See King [2007] Chapter 7 for a fuller discussion of those additional issues. One fi nal point. I mentioned that the on the present account, propositional constituency isn’t transitive (see the discussion of 12P above). But what about the motivation J&J off ered for it being transitive? Consider the proposition that Rebecca swims. On the present view, Rebecca is a simple constituent of it. But it is easy to see that on the present view, Rebecca is also a simple constituent of both the proposition that it is not the case that Rebecca swims and the proposition that Rebecca swims and Shane skis. In both cases she will be at a terminal node of the relevant propositional relation. In general, if P is any proposition and Q is a proposition built up of P, other propositions and truth functions, all simple constituents of P will be simple constituents of Q on the present view of constituency. Further, P will be a complex constituent of Q. Now we can see the diff erence between this case and 12P above. Hannah is a simple constituent

of a simple constituent of 12P: the proposition PH,s . Th ereby Hannah can’t be a simple constituent of 12P: she can’t occur at a terminal node of 12P’s propositional relation, since she is a proper part of something that occurs at one. Nor can she be a complex constituent. However, in the present case not only is Rebecca a simple constituent of the proposition that Rebecca swims, P, but in virtue of P being a complex constituent of the proposition that Rebecca swims and Shane skis, Q, Rebecca is a simple constituent of Q as well. So we both eat our cake with P and Q and have it with 12P too! Th ereby, we deny the transitivity of propositional constituency, while capturing the motivation J&J gave for it. In summary, on the present account of propositions we can rigorously characterize the notion of (simple and complex) constituents of a proposition and so do not have to

King020513OUK.indd 213 11/23/2013 12:59:15 PM 214 JEFFREY C. KING

take it as primitive as J&J claim Russellians must. Th e resulting account distinguishes between propositions that intuitively should be distinguished (those expressed by 13a and 13b; 14a and 14b; and 16a and 16b). Further, when coupled with the plausible view that grasping a proposition requires being acquainted with its constituents, our account makes plausible claims about the diff erent conditions under which the propositions distinguished can be grasped (with the result that one can grasp 13a and 14a without grasping 13b and 14b). In the case of the diff erence between the propositions expressed by 16a and 16b (assuming knowledge is justifi ed true belief) and the diff erent requirements on grasping them (assuming that to grasp a proposition one must be acquainted with its constituents), the account of constituency contributes to a resolution of the paradox of analysis. Th at this account of propositional constituency, which drops out of the present account of propositions, has these various desirable features can only provide further support for the theory of propositions that yields it.

King020513OUK.indd 214 11/23/2013 12:59:15 PM 1 1 Representation and Structure

in the Th eory of Propositions

J e ff S p e a k s

My aim in this essay will be to critically examine two aspects of current orthodoxy about propositions: that they are representational and that they are structured. Along the way I’ll also pause to discuss a few of the objections raised by Soames and King to the view of propositions I defend.

A r e propositions representational? To say that propositions are representational is, to a fi rst approximation, to say that they are about things. I think that there are three main arguments in favor of the view that propositions have, in this sense, representational properties: (i) Mental states, such as beliefs, are about things, and this fact about mental states is best explained in terms of the representational properties of the propositions which are their contents; (ii) Sentences are about things, and this fact about sentences is best explained in terms of the representational properties of the propositions which (relative to contexts) they express; (iii) Propositions are the sorts of things that can be true or false, and this shows that they must have representational properties. Each of these arguments has much to be said for it. But I think that none is as strong as it at fi rst appears. A fi rst point to make about argument (i) is that, as noted in chapter 5 above, it is weaker in the hands of King and Soames than in the hands of a traditional proposition theorist, who thinks of all representational mental states as analyzable as relations to intrinsically representational entities. Once we admit, as King and Soames do, the existence of fundamental mental states which are intrinsically representational, and don’t borrow their representational properties from the representational properties of propositions,

King020513OUK.indd 215 11/23/2013 12:59:15 PM 216 JEFF SPEAKS

it’s not obvious why we shouldn’t just say this about all mental states—that is, it’s hard to see why we wouldn’t treat belief, desire, etc. as intrinsically representational attitudes to propositions which are not, in themselves, representational. But let’s set this polemical point to the side, and ask: would the representational properties of beliefs be objectionably mysterious if not explained in terms of the representational properties of propositions? First, let’s get clear on what we’re saying has representational properties when we say that, for example, my belief that South Bend is lovely is about South Bend. Th ere are, I think, two sorts of things that we can be saying when we make claims like this. On the one hand, we might be saying that I stand in a certain “aboutness” relation to South Bend—which we might call the “believes-about” relation. On the other hand, we might be saying that there’s a certain entity—perhaps a mental sentence or something of the sort—which has the dual properties of being a belief of mine and of being about South Bend. I’ll consider these interpretations in turn. Consider fi rst the fact that I stand in the “believes-about” relation to South Bend. To stand in the believes-about relation to South Bend is, on my view, to stand in the belief relation to a proposition one of whose constituents is South Bend.1 Hence the fact that I stand in the believes-about relation to South Bend is not to be explained solely in terms of the properties of the proposition that South Bend is lovely—part of the explanation is my standing in the belief relation to this proposition. On my view these relations, rather than the entities to which the relations are borne, are the source of the representational properties of subjects associated with the attitudes. Hence, to explain these representational properties of subjects, we don’t need to appeal to the representational properties of propositions, but only to representational ways of being related to those propositions. Th is does not mean, of course, that the proposition has no role to play in explaining the representational properties of a subject: the representational properties that I instantiate when I believe that grass is green are diff erent than the representational properties that I instantiate when I believe that snow is white in virtue of diff erences between the propositions that grass is green and that snow is white. But this does not mean that these propositions are themselves representational, or about anything. An analogy might help. Consider the view that all theft is morally wrong. On this view, the wrongness of an act can be explained simply by its being an act of theft . But this is consistent with the moral properties of the act also being partly explained by the object of the theft —one might think that stealing a car is morally worse than stealing a piece of bubble gum. But the fact that the moral properties of an act of theft is partly a function of the moral properties of the object of the theft does not imply that

1 Th ough see below for some second thoughts about “constituent” talk. Others, of course, will specify the relevant property of propositions diff erently—in terms of, for example, singular Fregean modes of presentation of South Bend. I’m ignoring such diff erences for simplicity here—they are independent of the question at issue.

King020513OUK.indd 216 11/23/2013 12:59:15 PM REPRESENTATION, STRUCTURE, AND PROPOSITIONS 217

those objects—cars, and pieces of bubble gum—themselves instantiate moral properties. Just so, the fact that which propositions a subject believes determines which representational properties the subject instantiates does not imply that the propositions themselves must instantiate representational properties. Of course, even if our talk about the beliefs of subjects does not directly force us to attribute representational properties to propositions, one might worry that if we do identify propositions with non-representational entities, like properties, we’ll be unable to give a satisfactory account of the attitudes. King, for example, objects to the account I proposed in chapter 5 that

“I can consider, explain and understand the claim that arithmetic reduces to logic and in none of these cases does it seem that my attitude has anything to do with whether the alleged property being such that arithmetic reduces to logic is instantiated. Take considering: when I consider the claim that arithmetic reduces to logic I can do so without any attitude at all about the instantiation of the property being such that arithmetic reduces to logic. I may simply have an interest in really thinking about what the claim comes to. So here Speaks’s account of the attitudes seems strained at the very least.”

Th ere are two lines of objection here worth separating. A fi rst line of objection focuses simply on the oddness of saying that beliefs are relations to esoteric properties, like the property of being such that arithmetic reduces to logic. In a way, this is a fair complaint; it does sound odd to talk about properties as the objects of belief. But this has little to do with the choice of properties as the category to which propositions are to be assimilated, and more with the very idea that propositions are to be assimilated to some category or other. To see this, consider King’s view. On that view, to consider the proposition that Amelia talks, one must consider the fact that there is a context and assignment relative to which Amelia is the semantic value of an expression e of L and talking is the value of an expression e’ of L such that e occurs at the left terminal node of a relation R that in L encodes ascription and e’ occurs at R’s right terminal node. Th is is surely no less odd than the view that one must consider the property of being such that arithmetic reduces to logic in order to consider the proposition that arithmetic reduces to logic. Th e problem here is no more particular to King’s theory than to mine; parallel remarks could be made about Soames’s theory and, I suggest, about any theory which makes informative and surprising claims about what propositions are. King is right that these odd-sounding claims about belief are the costs of such views. But I think (as I think that King and Soames do as well) that these costs are outweighed by the benefi ts of reducing propositions to another category of entity in which we have independent reason to believe. Th e second line of objection (which, I think, is what King had in mind in the above passage) focuses not on the oddness of thinking that attitudes are relations to properties, but that propositional attitudes in general must be attitudes involving the instantiation of properties. Perhaps belief does involve believing a property to be

King020513OUK.indd 217 11/23/2013 12:59:15 PM 218 JEFF SPEAKS

instantiated—but surely, as King says, one can consider a proposition without taking any attitude at all toward the question of whether the relevant property is instantiated. But here I am inclined to think that King exaggerates the distance between belief and consideration. Of course, considering a proposition does not involve believing that the relevant property is instantiated; but that is just because belief and considering are diff erent propositional attitudes, and hence involve diff erent attitudes toward the instantiation of the relevant property. It seems to me plausible that when we consider a proposition, we are considering what would be the case were that proposition true; and it is natural, on my sort of view, to take this to be an attitude toward the instantiation of the relevant property. 2 One might think, though, that there is a diff erent sort of attitude toward a proposition, which we might also express using “consider,” which is diff erent than considering whether a certain proposition is true. Th is is the attitude we might have toward a proposition when we’re simply examining its properties—for example, we might consider its structure, or consider whether this or that is a constituent of the proposition. We might, as King puts it, simply be interested in “really thinking about what the claim comes to.” But I think that examining, in this sense, the proposition that arithmetic reduces to logic is best understood as a propositional attitude to some proposition other than the proposition that arithmetic reduces to logic. It might, for example, be a matter of judging that (or hypothesizing that, or wondering whether) the proposition that arithmetic reduces to logic has a certain structure. And these propositional attitudes can, without strain, be understood as involving the instantiation of the relevant property (which will, in this case, be the property of being such that the proposition that arithmetic reduces to logic has the relevant structure). Th e case is analogous to one in which I consider the properties of some object, like my desk. Th is need not involve me being in some mental state whose content is my desk; rather, the contents of such mental states are propositions which attribute one or another property to my desk. Let’s consider the other interpretation of argument (i), according to which “my belief” refers to something like a mental sentence which qualifi es as one of my beliefs and has the representational property of being about South Bend.3 Do we need to explain what it is for such a mental sentence to be about South Bend in terms of the representational properties of the proposition which that mental sentence has as its content?

2 It’s worth adding that the “believes-to-be-instantiated” bit of Ch. 5 is a dispensable add-on to the theory of propositions sketched there; it is an attempt to explain how, if propositions are properties, we might understand propositional attitude relations to propositions. But one might well give a theory of propositions without also giving a theory of the attitudes. Th at, I think, is what King does; he gives an account of propositions, and of how we can have cognitive access to them, but never explains exactly what it is to believe a fact of the relevant sort. 3 Nothing much is built into “mental sentence” here—the following remarks would apply just as well to theories which want to avoid commitment to anything like a language of thought, and so make use of syntactically unstructured “belief states” instead.

King020513OUK.indd 218 11/23/2013 12:59:15 PM REPRESENTATION, STRUCTURE, AND PROPOSITIONS 219

On my view, for one of my mental sentences to be about South Bend is for that sentence to have as its content a proposition one of whose constituents is South Bend. Hence— as with the representational properties of subjects—the representational properties of mental sentences are not explained solely by a proposition, but rather by those sentences standing in a certain relation—which we express by phrases like “has the content that”— to that proposition. But then, as in the case of subject-level representational properties, it is open to us to trace the source of the representational properties of mental sentences not to the proposition to which it is so related, but to the relation itself. Th e above remarks about mental sentences generalize to give us a response to argument (ii) above: the argument that we need to appeal to the representational properties of propositions to explain the representational properties of sentences of public languages like English. Such sentences also have their representational properties in virtue of standing in certain relations—like that expressed by “_ is the semantic content of _ in C”—to propositions. Hence, as above, the representational properties of sentences can be explained not in terms of the representational properties of the propositions they semantically express, but by the representational properties of the relation of semantically expressing. Th is style of explanation can, it seems to me, be further generalized to bearers of contents other than sentences. In chapter 9, Soames objects to my view of propositions that it

“makes it diffi cult to capture the fact that truth is a kind of accuracy in representation. A map or portrait is accurate, or veridical, when it represents its subject matter as being how it really is; a proposition is true when it represents things as they really are. Th is parallel seems to be lost when propositions are identifi ed with properties that, as Speaks admits, aren’t intrinsically representational. Unless he can identify some sense in which they are representational, he will lose the pre-theoretic connection between truth and accuracy.” On my view, maps and portraits, like sentences, have contents, and I’m inclined to think that, despite the fundamental diff erences in the way these contents are encoded, maps and portraits have the same kinds of contents as sentences—namely, the sorts of properties discussed in chapter 5. But, given this, the parallel between maps and sentences seems pretty direct; each are representational in virtue of standing in certain relations to propositions which are not themselves representational. For a map to be a map of South Bend, for example, just is for that map to have as its content some proposition one of whose constituents is South Bend. And just as a sentence is true iff the property which is its content is instantiated, a map is accurate iff the property which is its content is instantiated. 4

4 Th ere remains the awkward question of why, if maps are this closely related to sentences, it sounds a bit odd to say that a map is “true.” Parallel worries arise, as is well-known, about the construal of perceptual experiences as a propositional attitude. I’m inclined to think that this oddness does not refl ect anything deep about the subject matter—and in any case Soames’s worry here is not that my view makes maps and sentences too similar, but that it does not make them similar enough.

King020513OUK.indd 219 11/23/2013 12:59:16 PM 220 JEFF SPEAKS

In response to arguments (i) and (ii), then, I have recommended a two-part strategy. When presented with an example of something which has a representational property, fi rst note that this representational property is never explained solely in terms of a proposition, rather in terms of the thing standing in a certain relation to that proposition. Second, note that we can plausibly take relations of this sort, rather than the propositions to which the relations are borne, to be the source of the representational properties to be explained. One might object that this strategy just pushes our problem—the problem of explaining representational properties—back a step, so that we are still left with the problem of explaining the representational properties of relations like belief or “has the content that.” Th is is correct. But there are two things to be said in response. First, the fact that we located the source of representational properties in relations like these does not mean that we are forced into taking these relations as primitive. Perhaps some such relations can be analyzed in terms of others—as on theories (like those of Grice and Lewis) which explain natural language meaning in terms of propositional attitudes, or on theories (like Fodor’s) which explain the attitudes in terms of the representational properties of mental sentences.5 And perhaps we can explain some such relations in non-representational terms, as those pursuing the project of “naturalizing intentionality” have tried to do. Second, if we do take one or more of these relations in question as primitive, the best candidates here will likely be relations in which thinking subjects stand to propositions. But then it seems that we’re locating our primitives—by Soames’s and King’s lights as well as by mine—in exactly the right place. Th is leaves us only argument (iii), which might well seem the most fundamental. Even if sentences and mental states don’t require the introduction of representational properties of propositions, one might think that the fact that propositions can be true or false just shows that they obviously do have representational properties. Isn’t having truth conditions a representational property par excellence ? It is important to disentangle the substantive issue here from a merely verbal one. I agree that propositions have truth conditions. (Th ough, unlike Soames and King, I don’t think that their possession of these truth conditions is explainable in terms of anything about us.) What I deny is that propositions are about anything. Th e interesting debate is not about whether we should call the possession of truth conditions a representational property; the interesting question is whether it is coherent to hold that propositions can have truth conditions despite not being about anything. As noted in chapter 5, this sort of view is less of a departure from the way that we ordinarily think about these matters than it might at fi rst seem. Even those who don’t identify propositions with properties think of properties as true of objects but not about anything. But if this is coherent, why not the view that propositions are true, but not about anything?

5 See, respectively, Grice (1968), Grice (1969), Lewis (1975), and the essays in Fodor (1990).

King020513OUK.indd 220 11/23/2013 12:59:16 PM REPRESENTATION, STRUCTURE, AND PROPOSITIONS 221

One might worry, though, that saying that propositions can have truth conditions without being about anything will force us to deny some platitudes about truth. For example, it seems plausible that something is true just in case it represents the world as being some way, and the world is that way—but that can’t be right if, as I think, propositions are true or false but don’t represent the world as being any way at all. I think that this is the strongest argument in favor of ascribing representational properties to propositions. But I don’t think that it is decisive. Th e right thing to say, it seems to me, is that these seeming platitudes are really platitudes—but only when we’re talking about the notion of truth applicable to sentences or beliefs. On my view, it is a platitude that a sentence is true iff it represents the world as being some way, and the world is that way—what it is for a sentence to represent the world as being some way is for that sentence to have a certain property—a way things could be—as its content, and what it is for the world to be that way is for the property to be instantiated. But no such claim holds about the truth of propositions. Like many, I think that propositions are the fundamental bearers of truth and falsity, and that the truth and falsity of sentences, mental states, maps, etc. is to be analyzed in terms of the truth and falsity of propositions to which these entities stand in the relevant relations.6 On a view like this, it will be no surprise if the right account of truth and falsity for propositions is diff erent from the right account of truth and falsity for the non-fundamental bearers of truth and falsity. Hence, I think, it should not be surprising if platitudes which are correct about the truth and falsity of sentences don’t hold when we’re talking about the truth and falsity of propositions. 7

A r e propositions structured? Let’s now turn away from the question of whether propositions are representational to the question of whether they are structured. One might think that despite our disagreement on the fi rst question, Soames, King, and I at least all agree on the answer to the second. Each of us, aft er all, talks freely about the constituents of propositions—and if propositions have constituents, don’t they have to be structured? However, on closer inspection, I think that matters are a bit less clear. A way to bring this out is by distinguishing between what we might call metaphysically “lightweight” and “heavyweight” senses of the claim that propositions are structured. I gave an example of a metaphysically lightweight interpretation in §2 of chapter 5 above. Th ere I suggested that we think of the constituents of the proposition expressed by a

6 For defense, see Soames (1999). 7 It might also be worth noting that it seems not to be a platitude that propositions are about anything. Th e following hardly counts as commonsense: “Th at South Bend is lovely is about South Bend.” By contrast, it is hard to deny either of “ ‘South Bend is lovely’ is about South Bend” or “Th e belief that South Bend is lovely is about South Bend.”

King020513OUK.indd 221 11/23/2013 12:59:16 PM 222 JEFF SPEAKS

sentence S as the contents of the subsentential expressions in S. Now, this sort of use of the word “constituent” does not make wholly trivial the claim that propositions have constituents—given that propositions are identical only if they have the same constituents, the claim that propositions have constituents in this sense entails that if two sentences S , S * express the same proposition, then S contains an expression with a certain content iff S * does. And this will be denied by proponents of “coarse-grained” conceptions of propositions, like the view that propositions are sets of worlds. For on that view necessarily equivalent sentences will always express the same proposition—but two sentences can be necessarily equivalent even if one contains terms with a content which no expression in the other sentences has. Th e obvious examples here are math- ematical truths—but there are plenty of others. 8 So there is a substantive debate to be had about whether propositions have constituents in the lightweight sense. But one might reasonably object that this is not really a debate about whether propositions really have constituents at all. For surely to say that propositions have constituents must be to say about propositions something at least analogous to what we say about tables and chairs when we say that they have constituents. In this metaphysically heavyweight sense, to say that propositions have constituents is to say that there are entities—paradigmatically the contents of subsentential expressions fi guring in the sentence which expresses the proposition—to which propositions stand in some relation which either is, or is closely analogous to, the relation between parts and wholes. Th ere’s no obvious contradiction in accepting the claim that propositions have constituents in the metaphysically lightweight sense, but denying that they have constituents in the metaphysically heavyweight sense. One might, aft er all, adopt the following view about propositions:

Primitivism. Propositions are a sui generis category of entity, and hence are not identical to sets of possible worlds. Further, a pair of sentences can be necessarily equivalent despite expressing distinct propositions. Indeed, propositions are much more fi ne-grained than sets of possible worlds; two sentences can express the same proposition only if those sentences are synonymous, and two sentences can be synonymous only if they are composed of synonymous subsentential expressions. But propositions, like other abstract objects, are simple, and have no parts; hence propositions have (in the metaphysically heavyweight sense) no constituents.9 Intuitively, this sort of primitivist agrees with the traditional doctrine of structured propositions about the individuation of propositions—about when, for example, the proposition expressed by one sentence is identical to the proposition expressed by

8 One can adopt the well-known recipe for constructing such cases from Soames (1987)—just pick any sentence expressing a necessary truth, and conjoin it with an arbitrary sentence S. Th e conjunction and S will be equivalent, and hence according to the view that propositions are sets of worlds, will express the same proposition. But in the standard case our necessary truth will contain some expression which is not synonymous with any expression in S. 9 Th is is the view of, among others, Plantinga (1974), and is defended at length in Keller (2012).

King020513OUK.indd 222 11/23/2013 12:59:16 PM REPRESENTATION, STRUCTURE, AND PROPOSITIONS 223

another. So the primitivist can agree that propositions have constituents in the lightweight sense. But they disagree about whether these claims about the individuation of propositions are best explained by attributing genuine complexity—i.e., the having of parts—to propositions. Do the theories of propositions defended in Part II by Soames, King and me entail that propositions have constituents in the metaphysically heavyweight sense? I don’t see that they do. We defend, respectively, the views that propositions are event-types, facts, and properties. But I think that each of our theories leaves open the question of whether these event-types, facts, and properties have constituents in the metaphysically heavyweight sense—that is, we leave it open whether event-types, facts, and properties stand in part/whole relations (or some closely analogous relation) to anything. 10 To see this, consider the following thought-experiment. Imagine being presented with a compelling argument for the conclusion that no abstract objects, and in particular no properties, facts, or cognitive event types, have parts. Now ask yourself: would this aff ect any of the claims made in the chapters in Part II of this book, or help us to decide which of those theories was most likely to be true? I am not sure that it would. Of course, we’d now have to understand the use of “constituent” in those chapters in a metaphysically less-than-serious way, as a useful term for specifying just which properties, facts, or cognitive event-types propositions are supposed to be, but not as imply- ing that there’s any genuine complexity in the entities so specifi ed. But, other than that, it doesn’t seem as though anything would be diff erent. I think that this openness of our views on this score is in one sense a virtue, and in another, a vice. It is a strategic virtue because it makes our views consistent with (respectively) various views of the nature of event-types, facts, and properties. And, relatedly, it means that we don’t have to confront diffi cult questions which confront Millians who are also believers in the claim that propositions have constituents in the heavyweight sense. For example: on standard views, part/whole relations are transitive, so that if x is a constituent of y and y is a constituent of z, then x is a constituent of z. But it looks as though Millians will have to deny this principle, since I can be a constituent of a singular proposition about me without every part of me also being a constituent of that proposition. One can, fortunately, entertain singular thoughts about me without thereby entertaining singular thoughts about every part of me.11 Th e vice corresponding to this virtue, though, is pretty obvious, and that is just that the above virtue is obtained only by failing to explain the relationship between

10 Th ough I think that the theories leave this question open, I’m not saying that Soames and King themselves have no views on the matter. For example, King expresses sympathy with the idea that the constituents of propositions are parts of those propositions in King (2007), p. 120 note 42. 11 See for discussion Gilmore (2011). For a more wide-ranging discussion of the problems with assimilating talk about the constituents of propositions to standard theories of part/whole relations, see Keller (2012). It’s also worth noting that some views of propositions as in some sense structured—like the view on which they are ordered pairs—don’t run into this problem. For criticism of “ordered pair” views, see King (2007).

King020513OUK.indd 223 11/23/2013 12:59:16 PM 224 JEFF SPEAKS

propositions and the entities which the three of us all refer to as that proposition’s constituents. If it isn’t parthood, what is it? Let’s return for a moment to the primitivist view of propositions sketched above. Th e primitivist agrees with Soames, King, and I that propositions have constituents in the metaphysically lightweight sense, but denies that propositions have constituents in the heavyweight sense. But I’ve just said that the theories which Soames, King, and I give are consistent with the denial of the claim that propositions have constituents in the heavyweight sense. Does that mean that our theories are consistent with this primitivist view? Not quite. Th e primitivist says something which each of us denies, namely that propositions are a sui generis category of entities, and hence rules out the sorts of theories that each of us try to provide by, respectively, assimilating propositions to cognitive event-types, facts, and properties. Th is brings out the fact that we should distinguish three diff erent questions on which a theory of propositions should take a stand. Th e fi rst is a question about whether propositions have constituents in a lightweight sense, which boils down to questions about the individuation of propositions—e.g. questions about the conditions under which a pair of sentences express the same proposition, or a pair of subjects believe the same proposition. Th e second is a question about whether propositions have constituents in the heavyweight sense, which is just the question of whether propositions are simple (lack parts) or complex (have parts). Th e third is a question about what sorts of things propositions are. Are propositions a basic category of our ontology, or are they a subclass of some other more fundamental category of thing? Th e independence of these three questions deserves some emphasis. Occasionally the fi rst two are implicitly confl ated, as when one infers from the falsity of the possible worlds view of propositions (and hence the falsity of the claim that propositions lack constituents in the lightweight sense) that propositions must be structured (and hence have constituents in the heavyweight sense). 12 And occasionally the second two are confl ated, as when the view that propositions are simple (and hence lack constituents in the heavyweight sense) is taken to imply the claim that propositions are irreducible to any other category of entity. Th e focus of this book has been overwhelmingly on the third of the above questions. (Th ough, in each case, our answers to this third question plausibly entail answers to the fi rst.) We’ve managed, by contrast, mainly to ignore the second. In a way, this emphasis makes sense. Just as we can ask whether propositions are complex or simple, we can ask whether properties, facts, and other categories of abstracta are complex or simple; and, plausibly, whatever considerations lead us to say that propositions are or are not complex will lead us to the same conclusion about properties and facts.13 Hence one

12 See, for discussion, Keller (2012). 13 Of course, I think that propositions are properties, and King thinks that they are facts. But both of us think that there are properties and facts which are not propositions, and questions about complexity will arise for these as well.

King020513OUK.indd 224 11/23/2013 12:59:16 PM REPRESENTATION, STRUCTURE, AND PROPOSITIONS 225

might well think that our second question is not really a question about propositions in particular, but rather a question about abstract objects more generally. Th is doesn’t change the fact that the second question is one to which any complete theory of propositions, especially ones which are expressed using talk about structure and constituents, owes an answer. But I wonder whether the pervasive use of structure-talk has made this second question seem more fundamental than it really is. What hangs on the question of whether propositions are simple or complex, unstructured or structured? If anything hangs on this question, it must be that the claim that propositions have constituents in the metaphysically heavyweight sense explains something. So our question reduces to: What might the claim that propositions have constituents in this sense explain? A natural, and popular, suggestion is that structure explains representational properties; we can explain what it is for a proposition to be about Socrates by saying that the proposition contains Socrates as a constituent.14 But, as Keller (2012) convincingly argues, this sort of explanatory claim is problematic. One might put her point like this: either constituency is parthood, or it isn’t. If it is, then we can’t explain aboutness in terms of constituency, for many things have parts without being in any sense “about” those parts. And if it isn’t, then, absent a theory of the constituency relation, we’ve just replaced one primitive—aboutness—with another. Where’s the explanatory gain there? Th is dilemma does not, of course, show that there is no good answer to the question of what the claim that propositions have constituents in the metaphysically heavyweight sense explains. But I think that it does strongly suggest that the explanatory value of the attribution of structure to propositions has been a bit under-explained. While I still fi nd the view that propositions are structured attractive, I’m less and less sure exactly what is explained by adding to one’s theory of propositions the claim that they have parts.15

14 For versions of this thought, see, among many other places, King (2007), 6, and Braun (1993), 461. We can set aside in the present context the fact that I don’t think that propositions have representational properties to be explained; we might still think that we can explain the representational properties of beliefs and sentences in part in terms of the constituents of the propositions that they have as their content 15 Th anks to Lorraine Juliano Keller for helping to engender the skeptical worries just voiced. Discussion of these topics with Lorraine, along with a reading of her excellent dissertation, Whence Structured Propositions? has prevented me from making at least some of the metaphysically irresponsible claims which I would otherwise have made in this book.

King020513OUK.indd 225 11/23/2013 12:59:16 PM 1 2 Clarifying and Improving the

Cognitive Th eory

S c o t t S o a m e s

In this chapter I reply to the probing and provocative critiques of Speaks and King, which I use (i) to clarify aspects of my position that had previously been underde- veloped and insuffi ciently clear, (ii) to introduce improvements needed to properly understand the sense in which propositions are genuinely representational, and (iii) to discuss the explanatory burden in terms of which any theory of propositions must be judged.

I n r e p l y t o S p e a k s Th e real explanatory burden Speaks notes that I agree with him that being such that Amelia talks and other properties of this sort exist, and that we have cognitive access to them. Since both of us must explain this access, while I must also explain our access to the cognitive event types I have identifi ed as propositions, he concludes that his explanatory debt is less than mine. I disagree; the explanation I provide of our access to (degenerate) properties like being such that Amelia talks is a trivial extension of the explanation already provided of our access to the propositions from which they are derived. Whereas I explain our access to these properties, he doesn’t. Although we both presuppose agents’ access to simple properties , he is silent, whereas I am not, about how complex properties are generated, how they are individuated, and how we access them. In my system negating a property being so-and-so , predication of which represents its target as being so-and-so , generates the property not being so-and-so, predication of which represents its target as not being so-and-so . Similar stories can be told about conjoining and disjoining properties, generating n-1 place properties from n place properties, and forming complex properties in other ways. In each case, agents’ access to complex properties is explained by operations they perform on more fundamental

King020513OUK.indd 226 11/23/2013 12:59:16 PM CLARIFYING AND IMPROVING THE COGNITIVE THEORY 227

properties, while the complex properties themselves are individuated by the ways in which predication of them represents their targets, from which their contributions to the truth conditions of propositions can be read off . Th is idea is extended by operations generating properties whose representational contents, individuation conditions, and cognitive accessibility are parasitic on those of already generated propositions. Operating on the proposition that John loves Mary or Bill hates Mary, agents can generate the property—λx [John loves x or Bill hates x]—predication of which represents its target as being one whom John loves or Bill hates. Taking the degenerate case of this operation in which no constituent is abstracted from the original proposition, we generate the property being such that John loves Mary or Bill hates Mary predication of which of any target represents precisely what the proposition represents. When S expresses p, ⌜is such that S ⌝ stands for the property predication of which of represents what p represents (and nothing further). Since the individuation of, and our cognitive access to, the property are parasitic on the already explained individuation of, and access to, the proposition, these properties don’t add any further explanatory burden. It is Speaks who faces the problem of individuating and explaining our cognitive access to the properties expressed by ⌜is such that S ⌝ and ⌜is such that R ⌝ for arbitrary S and R. How can he do so without invoking propositional intermediaries? Well, how are Amelia and being a talker related to being such that Amelia talks? Th is Speaksian property/proposition is not the property that x instantiates iff Amelia instantiates being a talker. Th ere is no unique property satisfying that condition, since being such that Amelia talks and 1st-order arithmetic is incomplete does too. Nor is it enough to say that the proposition is a property satisfying the condition, since Speaks needs diff erent properties satisfying the condition to be objects of diff erent attitudes. One could build formal structures—trees, tuples, sets of sets—out of simple properties and objects, and then stipulate their instantiation conditions . But unless the structure and the conditions assigned can be shown to be nonarbitrary, this will, at most, model propositions, not identify them.1 One could, of course, go representational by maintaining (i) that being such that Amelia talks is the property that represents Amelia as a talker (without representing anything further) because one who predicates it of anything represents Amelia that way, (ii) that being such that the earth is round is the property that represents the earth as round (and nothing further) for a similar reason, (iii) that being such that Amelia talks or the earth is round is the property that represents Amelia as a talker or the earth as round (and nothing further) because one who predicates it of anything does, and so on. But this would take Speaks down my road, which he doesn’t wish to travel—in part, I suspect, because it leads to the question, “Since predicating being a talker of Amelia represents her as a talker, shouldn’t we already have the proposition that she

1 Th e problem of identifying propositions (be they properties or not) with essentially arbitrary structures intended to individuate them, which are then assigned truth conditions, is discussed on pp. 52–55 of What is Meaning? (Soames 2010).

King020513OUK.indd 227 11/23/2013 12:59:16 PM 228 SCOTT SOAMES

talks , before reaching the property being such that she does?” Eschewing this representational route, he needs his own compositional theory of the structures and instantiation conditions of such-that properties that nonarbitrarily distinguishes them from one another and explains how agents with limited cognitive resources can access indefi nitely many. Until he provides one, his explanatory burden is more, not less, daunting than mine.

Acts, events, and the chimera of bare predication I now turn to the substantive questions he raises about the view of propositions I articulated in chapter 6, and the clarifi cations of that view that his questions demand. I begin with his concern about what may be described as events of bare predication. He doubts that instances of the event type predicating redness of a certain coff ee mug o exist. He says, “When I think about. . .events of visually representing that o is red, judging that o is red, and asserting that o is red. . . I really don’t notice an event of predication which accompanies each one.” (12) What he doesn’t notice, and thinks can’t be noticed, isn’t the visual experience of seeing something as red. One can certainly be aware of concrete events in which one sees something as red, as well as those in which one ima- gines it as red. We are also aware of moments in which we judge something to be red, or assert it to be. What Speaks doesn’t notice are other events of merely predicating redness of o that also occur whenever some seeing, imagining, judging, or asserting takes place. Not noticing them, he doubts that there are any. One might worry, as I do, that this objection puts too much weight on what can be established about our mental lives by mere introspection. However, when it comes to predication there is more to be said. I fear that what Speaks is aft er are events of bare predication in a sense that goes well beyond what I am committed to, and which I fi nd to be unrealistic. What I am committed to is the claim that all instances of the event types seeing o as red and imagining o as red—as well as instances of the types judging and asserting that o is red—are instances of the event type cognizing o as red , a.k.a. predicating redness of o . Th is does not mean that every instance of these types is made up of a smaller constituent event of merely predicating redness of o plus some further accompanying event involving one or another cognitive doing. On the contrary, some of the event types I have mentioned (seeing and imagining that o is red) do not have instances that are composite in this way. Th ough the point is not as transparent for the other event types I have mentioned (judging and asserting that o is red), it is not obvious that their instances are composite either. However, the matter is complicated by the need to carefully distinguish cognitive acts from instances of the event types that consist in the performance of those acts by agents. Let’s start with cognitive acts. Some of these—judging and asserting that o is red— involve further cognitive acts in addition to predicating redness of o, whereas others— seeing and imagining—do not. To judge or assert that o is red is to think of o as red and to do something else. In the case of judging, this something else is endorsing, in

King020513OUK.indd 228 11/23/2013 12:59:16 PM CLARIFYING AND IMPROVING THE COGNITIVE THEORY 229

the sense of adopting that way of thinking—of o as red—as a potential basis for further thought or action. In the case of assertion, the something else is an act of publicly committing oneself to o’s being as one represents it to be. In each case, we (i) cognitively represent o to be red (which we also do when we merely imagine o as being red), and (ii) take a further stance toward that representation (which we do not take when we merely imagine o to be red). So, all events of judging or asserting that o is red involve a distinctive kind of cognizing—“predicating redness of o”—accompanied by other cognitive doings analytically distinguishable from it. Th is point generalizes to many related attitudes including questioning, denying, and so on. Of course, it does not follow that any of these cognitive events involve an initial event of predication succeeded in time by another event of endorsing, questioning, or what have you. How these cognitive acts are performed—simultaneously or in sequence—is not for philosophy to decide. Th us the theory generates no expectation that even the most powerful intro- spector should be able to notice events consisting of an agent’s performance of an act of predication in splendid isolation (in the absence of the performance of any further acts) in cases in which an agent judges, asserts, questions or denies something. So the fact that Speaks doesn’t notice them tells us nothing about the theory. 2 Th e point is strengthened when one considers simpler cognitions—seeing, visualizing, or imagining o as red. Any event consisting of an agent’s doing one of these things is an instance of the agent’s cognizing o as red—a.k.a. predicating redness of o . Of course, not all instances of predicating are instances of seeing, not all instances of predicating are instances of visualizing, etc. How then do these diff erent instances of predicating diff er from one another? Does seeing o as red consist of predicating redness of o plus doing something else (the doing of which is no part of the predicating), while visualizing o as red consists of that same predicating plus doing a diff erent something else , and similarly for imagining o as red? I think not. To see o as red is to predicate redness of o (i.e. to cognize o as red) in a certain way , while to visualize o as red is to predicate redness of o in a diff erent way. But these diff erent ways no more involve the performance of diff erent acts (the doing of which is no part of the predicating), than the diff erence between punching a bag with one’s right hand and punching it with one’s left hand involves the performance of diff erent additional acts (the doing of which is no part of either punching). When it comes to events the lesson is clear. Just as there is no bare event of my punching the bag that is not identical with an event of my punching it with my right hand or identical with an event of my punching it with my left , so there is no bare event of my predicating redness of o (i.e. of cognizing o as red) that is not identical with my seeing o as red, my visualizing o as red, imagining o as red, or my

2 Events of predication occur when an agent does something representational—like judge, assert, and the like. Since not all propositional attitudes—e.g. believing and assuming—require occurrent cognitive events, but may sometimes be fully dispositional, agents may sometimes hold propositional attitudes without performing any acts of predication, or experiencing any predicational events constitutive of the propositions to which the attitudes are born.

King020513OUK.indd 229 11/23/2013 12:59:16 PM 230 SCOTT SOAMES

cognizing o as red in some other way.3 In short, there are no events of bare predication (i.e. of cognizing but not of cognizing in any particular way) of the sort Speaks is seek- ing. To seek them is to misunderstand the theory. Th e special role of entertaining among the propositional attitudes Speaks also worries that my identifi cation of the event of my predicating redness of o at t with the event of my entertaining, at t, the proposition that o is red, leads me to claim that the entraining event is itself an instance of the proposition entertained. Although he seems to realize that there is no circularity, regress, or absurdity here, he fi nds the claim dubious because it doesn’t help him identify which events, if any, are events of predicating redness of o. Unable to fi nd instances of bare predication, he remains in the dark about the event type predicating redness of o, and so learns nothing by being told that it is the proposition that o is red. Th e problem is his ill-advised search for bare predications, not the identity claim. Although entertaining is a genuine relation between agents and propositions, and hence a propositional attitude, it is not, in my view, a relation we bear to special metaphysical objects independent of us, patiently waiting for us to cognize them. On the contrary, it is just that conception that has led to so much trouble. Since for me propositions are kinds of cognitive doings, entertaining them is not a matter of thinking about them in some special way, but of embodying them in one’s cognitive life. If propositions are event types, p and the event type entertaining p are identical. To entertain p is not to have p in mind or to cognize it in any way; it is to perform a cognitive act resulting in an instance of p. Since for there to be such an instance is for an agent to represent something as being so-and-so, the act performed by the agent is similarly representational, as, we may suppose, are the concrete event and the abstract event type. Th is is how the intentionality of propositions is related to the intentionality of possible agents who entertain them. It is against this backdrop that we must understand Speaks’s comment, “ at least one propositional attitude state [entertaining p] also must have its representational properties intrinsically. . .[rather than being] explained by the representational properties of any proposition.” (13) Th e “also” in this remark refers to concrete events of predication, which are taken to have their properties “intrinsically.” Here, Speaks tracks my previous remarks that propositions inherit their representational properties from their possible instances, which unfortunately leaves the impression that, for me, the latter are intrinsically intentional while the former are only derivatively so. Translating his talk of the state of entertaining p into my terminology

3 To pursue the analogy, just as I can, on refl ection, determine that I have punched with my right hand (which is diff erent from punching with my left ), so, I claim, I can determine, on refl ection, that I have visually represented o as red (which is diff erent from conceptually representing o as red). Hence I am aware of representing in diff erent ways, even though I notice no bare acts of representing, because there are none. Above and throughout, I use “predicate” to underline that representing is something we do.

King020513OUK.indd 230 11/23/2013 12:59:16 PM CLARIFYING AND IMPROVING THE COGNITIVE THEORY 231

of the event type of entertaining p (which, in chapter 6 is identifi ed with p), I take him to be characterizing its intentionality as intrinsic. I now fi nd this way of putting things unfortunate. Th e proper way to proceed is, I think, (i) to identify the act of predicating being so-and-so of o as representational because for an agent to perform it is for the agent to represent o as so-and-so, and (ii) to explain whatever intentionality is possessed by the event type of performing this act and the individual instances of that type in terms of (i). Next Speaks asks: “why not say this [that they are intrinsically representational] for all... [such cognitive acts/ event-types]? Why not... let [not just the attitude of entertaining, but also] each of the familiar attitudes—belief, assertion, etc.—be intrinsically representational states [acts/event types], whose status as representational is not explained by the representational properties of the proposition to which they are relations?” (13–14)

Consider the attitudes affi rming, denying , and occurently doubting that o is red. Each involves taking a cognitive stance toward the proposition that o is red. Although the stance diff ers from case to case, what it is a stance toward is the same. Because the attitude entertaining abstracts away from any stance, it is perfectly suited to capturing what we focus on when we ask whether what is affi rmed, denied, or doubted is true. Th is is the fundamental sense in which affi rming, denying, and occurently doubting that o is red all represent o as being the same way. Th is representational commonality is captured by taking those attitudes to involve entertaining the proposition that o is red plus a further cognitive ingredient that varies from one attitude to the next. Th e presence of this further ingredient is why we shouldn’t treat the richer attitudes Speaks has in mind as we treat entertaining. Th ere may be a further sense in which the richer attitudes are representationally diff erent from one another. Aft er all, an agent’s take on the world will diff er markedly depending on whether the agent affi rms, denies, or doubts that o is red. But this further representationality results from how the cognitive stances involved in these attitudes interact with their propositional object. What these cognitive stances amount to is the least developed aspect of my view. So far, I have said that to judge that o is red is to predicate redness of o while affi rming or endorsing that predication, to believe that o is red is to judge, or be disposed to judge, that it is , and so on. To this, I here add a further cautionary note. To endorse or affi rm a proposition p that one has entertained is not to predicate a property or relation of p, or to perform any representation-modifying operation on p. It is to entertain p in a certain way, which results in that cognitive event playing a certain committing role in one’s cognitive life. Th e same is true of other stances, such as wondering. Th is must be so, since even cognitively unsophisticated creatures that are unable to identify and target the types of which the cognitive events of their own experience are instances can bear the attitudes of judging, believing and wondering to propositions.

King020513OUK.indd 231 11/23/2013 12:59:16 PM 232 SCOTT SOAMES

Existence and belief Speaks’s next objection focuses on the conjunction of my views (i) that an agent can believe a proposition p without ever entertaining p and (ii) that this can happen even in cases in which the p doesn’t exist. Th e objection is that if this were so, then certain arguments that are clearly valid wouldn’t be. For example, Speaks says, Argument A would not be valid.

Argument A A1. Jeff believes that no circles are 726-sided. A2. Scott believes that no circles are 726-sided. AC. So, there is something that Jeff and Scott both believe. Since A is valid, he concludes that the conjunction of (i) and (ii) is false. Th e example is not well chosen. Suppose that A1 and A2 are true because Jeff and Scott are disposed to affi rm the relevant proposition, even though neither they, nor anyone else, has ever entertained it, or ever predicated being 726 sided (having 726 sides) of anything. Th is is not suffi cient to show that the proposition doesn’t exist. Since the property is complex, agents can cognize it by applying a certain functional operation to the arguments 726 and the property having sides. Hence, the proposition fi ts the existence conditions sketched in chapter 6. What is required for it to exist is not that anyone has cognized being 726 sided or predicated it of anything, but that each argument of the cognitive operation has been cognized and the operation itself has been applied. Since these conditions have been fulfi lled in the actual world-state, the proposition said to be believed by Jeff and Scott does exist. Might there be world-states at which the conclusion is false and the premises are true, even though no one had ever cognized the number 726 or the property having sides ? As for the latter, I doubt that agents could implicitly believe the proposition about circles and sides if no one had the concept having sides because no one had ever cognized it. Could they believe it if no one had cognized the number 726? Perhaps. But now there is a diff erent worry. Why we should think that Jeff and Scott have implicit de re beliefs about the uncognized number, when for many uncognized objects—e.g. particular stones buried in Antarctica—we don’t think that agents have implicit de re beliefs about them? Th e answer, I think, is that we have a systematic linguistic means— the numeral system—mastery of which allows us to directly designate each number. Appealing to this systematicity, we may plausibly extend the existence conditions given in chapter 6 to allow for the existence of propositions entertainable by those who have mastered this or related systems. Systematicity—in which cognitive acquaintance with both simples and the operations that build complexity from them—was at the heart of the existence conditions proposed in chapter 6 for propositions. While the extension suggested here is new, it is well within the spirit of those conditions. Since adopting it undermines the case for falsity of AC, we still have no convincing objection to the conjunction of (i) and (ii) above.

King020513OUK.indd 232 11/23/2013 12:59:16 PM CLARIFYING AND IMPROVING THE COGNITIVE THEORY 233

Coming up with a better example isn’t easy. On the account I have off ered, there are real but nonexistent propositions, some of the simple constituents of which neither have been cognized already nor are cognitively accessible by any systematic means mastered by agents. Th ere are also propositions that agents believe without having entertained them. But it isn’t easy to show that some propositions are members of both classes. Displaying them in an argument is out of the question, since to do so is to guarantee their existence. Th ere is, of course, no such bar to displaying an existing proposition that could have been believed without existing, or was in fact believed before it existed. However, as the discussion of Argument A illustrates, even this is daunting. Suppose, for the sake of argument, we fi nd such a proposition, expressed by some sentence S. Switching the premises to the past tense gives us Argument A*. A1*. At t, Jeff believed that S. A2*. At t, Scott believed that S

Next consider the following conclusions. a * Th ere is something that Jeff and Scott both believed at t. b * Th ere exists something that Jeff and Scott both believed at t. c* At t, some proposition was believed by both Jeff and Scott. d* At t, there was a proposition that both Jeff and Scott believed. e* At t, there existed some proposition that both Jeff and Scott believed. Suppose that A1* and A2* are true, and that the proposition expressed by S exists now, but didn’t exist at t. Th en a* and b* will be true, and we won’t have a counterexample. Whether or not c* is a counterexample depends on whether we can at t quantify over things not existing at t. Since I have argued we can, I can recognize the truth of c*. I also believe that we can, and sometimes do, use “there is/are” to range over domains that include nonexistent things. Th us, d* has a reading in which it is true. What Speaks needs is e*. If the right sort of sentence S can be produced, A1* and A2* will be true and e* will be false. Such a result, though mildly surprising, wouldn’t be a weighty objection. 4 What we are engaged in is theory construction, not ordinary language analysis. Sometimes philosophical theory leads to correct, but surprising and even mildly counterintuitive results about which pretheoretic opinion isn’t determinative. So long as wholesale rejections of commonsense convictions are avoided, as they are here, philosophical explanation may sometimes prevail.

4 As Brian Bowman has pointed out to me, one can probably fi nd the right sort of sentence S if one con- structs an argument in which one concludes that there (now) exists a proposition that Jeff and Scott didn’t believe at t, from ⌜~ Scott believed S at t ⌝ and ⌜~ Jeff believed S at t⌝ . However, this doesn’t add much to the weight of the objection.

King020513OUK.indd 233 11/23/2013 12:59:16 PM 234 SCOTT SOAMES

Representation without cognition Speaks considers a barren world-state bw with no cognitive agents and (according to the view I have outlined) no existing propositions. Still, I hold that the proposition gg that grass is green is true at bw; it is true at any world-state w iff (i) gg represents grass as being green, and (ii) at w grass is that way. I have argued that (i) is true because gg is the event type of one’s predicating greenness of grass, which just is for one to represent grass as green (and nothing further). So understood, the intentionality of gg expressed by (i) isn’t relativized to world-states. But since what gg represents doesn’t vary from state to state, there is no harm in speaking of each state as being one at which it represents grass as being green. For Speaks, this means that on my account gg represents grass as being green at bw “because it is true in [b]w that were some subject to, for example, judge that grass is green, that judgment would involve predicating greenness of grass.”(17). But, he asks, what makes this counterfactual true ? He answers that the representational properties of gg make it true. But now, he thinks, we have gone in a circle, since we have illegiti- mately explained the truth of the counterfactual in terms of the representational properties of the proposition while also explaining the representational properties of the proposition in terms of the truth of the counterfactual. Th ere is no circle. Th e counterfactual is, of course, true at bw because were some agents to perform the cognitive act of predicating greenness of grass, they would thereby represent it as green. But the representational properties of the proposition are not explained by a conceptually prior appeal to the truth of the counterfactual. On the contrary, according to the view for which I have argued, the proposition represents grass as green because it is the event type in which an agent performs the representational act of predicating greenness of grass. Th e act is representational because to perform it is to represent grass in this way. From this it follows that all possible events of entertaining gg are instances in which agents represent grass as green, which in turn entails (but is not entailed by) the truth of Speaks’s counterfactual. Th us, there is no circle. Rather, one and the same thing—the inherently representational act—explains both the intentionality of the proposition and the truth of the counterfactual. 5 Types, tokens, and representation Toward the end of his critique, Speaks takes up a closely related point, objecting to my previous, all too familiar claim that cognitive event types are representational because their (possible) tokens are. He worries that since types don’t inherit all properties of their tokens, we have no reason to think that they inherit the representational properties of their tokens. Th ere is a legitimate concern here, but it has already been

5 Th is explanation supersedes my careless comments (from 2010 on) about propositions inheriting intentionality from their possible instances—comments that, doubtless, contributed to Speaks’s impression of circularity.

King020513OUK.indd 234 11/23/2013 12:59:16 PM CLARIFYING AND IMPROVING THE COGNITIVE THEORY 235

addressed. By distinguishing the representational act of predicating redness of o both from the abstract event type in which an agent does so and from concrete events that are instances of that type, I can eliminate any suggestion that the tokens transfer their representational properties to the type. It is all right to think of the event type and its tokens as representing o as red because the act does. However, it is not the act itself that most fundamentally represents o as red, but the agents who perform it who do. Of course, the properties of agents—of doing this or that—are not literally transferred to the acts they perform, or to the event types or instances in which someone performs them. For agents to predicate redness of o and thereby to represent o as red is for them to do something. Since acts don’t do anything, but rather are the things done, this is not precisely the sense in which the act predicating redness of o represents o as red; nor is it the sense in which events or event types do. Rather, there is an extended sense of representing o as red, attributable to acts, the function of which is to allow us to use the intimate relation these entities bear to the cognitive experience of agents to track their mental states and to assess their veridicality. Th e extended sense in which acts (and perhaps events) are said to represent is related to the more basic sense in which agents represent in a manner analogous to the way in which the extended sense in which some acts are commonly said to be intelligent, stupid, or thoughtless is related to the more fundamental sense in which it is agents who are intelligent, stupid, or thoughtless. Although we already have a notion of an agent’s overall accuracy on this or that subject, we also need to assess the veridicality of the agent’s, or our own, individual sayings, doings, and cognizings—which requires discrete entities that can be assigned truth conditions one by one . Th is leads us to speak derivatively of certain aspects of agents’ cognitive activity as representing what they, the agents , represent, and so as being true or false in so far as those agents represent things accurately or inaccurately when they perform the acts in question. For this we need notions of truth and falsity, plus a notion of cognitive doings that represent things as being various ways, where the sense in which these doings represent is not identical to, but rather is a natural extension of, the sense in which agents represent .6 Th ese cognitive products are what philosophers call “propositions,” and which, up to now, I have identifi ed with cognitive event types, but which I now think might better be identifi ed with the cognitive acts themselves. 7

6 I will say more about this extended sense below. 7 Th e reason that Speaks’s event types (i), (ii), (vi), and (vii) on page 18 are not good candidates for propositions, and are not naturally assigned representational content that makes them bearers of truth conditions, is that they are not connected closely enough to the cognitive lives of agents to serve the function that is the raison d’être of this extended sense of representing.

King020513OUK.indd 235 11/23/2013 12:59:16 PM 236 SCOTT SOAMES

I n r e p l y t o K i n g Expressible perceptual content King is surprised that I assume, without extended argument, that the content of perceptual experience is expressible in natural language. I think the worry is exaggerated. Th ough complications exist, and questions can be raised, it is, I think, overwhelmingly plausible that much perceptual content is linguistically expressible, provided one observes certain niceties. Suppose I truthfully report “Th is looks red” on the basis of seeing an object o, which is red. When one sees something as red, one typically, perhaps always, sees it as some fi nely individuated shade of red. Since not all these shades are (nonindexically) named, there is no guarantee that the particular shade I visually predicate of o is nonindexically expressed by any English term. But surely, it is expressible. If we can see it, attend to it, and discriminate it from other shades, we can name it, if the need arises. Th at is the sense in which it is nonindexically expressible—as well as being indexically expressible as “that shade (of red).” In addition to visually predicating this shade of o, do I also visually predicate the redness of o? I think so. For any particular red-shade, predicating it of o also counts as predicating being red of o. Hence, visually entertaining the proposition that o is that shade of red also counts as visually entertaining the proposition that o is red. 8 Other questions about visual content are more challenging. One of these, which I raised in chapter 6, is whether the representational content of a visual perception of a complex scene can be encompassed by a complex web of related propositions. Th ough I am not certain that it can, I do think that much of that content is propositional. Other important and far-reaching questions arise concerning how the vague color terms of natural language come to encode the diff erent properties they do (in diff erent contexts as used by diff erent speakers), and how we should think of the not-fully-determinate clouds of propositions asserted by utterances of sentences containing them in diff erent contexts. However that is work for another time. 9 How many propositions? King also worries about how many propositions I allow. He points out that philosophers who hold presentist or actualist views, and so require true propositions to exist, will fault my account for providing too few propositions. Since King off ers no arguments for those views, I won’t reply, other than to say that I believe them to be mis- taken on independent grounds. Because I am trying to track the truth, my view of propositions is embedded in a framework that incorporates what I take to be other

8 As for the vexed but much discussed question about the general relation between perceptual and conceptual content, I largely align with the position outlined by Jeff Speaks in “Is Th ere a Problem about Nonconceptual Content?,” Th e Philosophical Review, 114 , 2005 , 359–398 . 9 S e e m y Rethinking Language, Mind, and Meaning , Princeton University Press, forthcoming.

King020513OUK.indd 236 11/23/2013 12:59:16 PM CLARIFYING AND IMPROVING THE COGNITIVE THEORY 237

philosophical truths. Th at said, the resulting account remains a package deal, some aspects of which are detachable from others. For example, my student Justin Dallmann has shown that one can recapitulate my account of propositions in a “serious actualist” framework by trading my distinction between existent and non-existent truth bearers for a distinction between propositions the existence of which (at the actual world) is grounded in what is actually concrete and those the existence of which (at the actual world) is grounded in what is merely possibly concrete.10 Th ough I don’t favor this reconstrual, I commend it to serious actualists who fi nd themselves unable to recant their metaphysical error. In addition to raising worries that I make room for too few propositions, King gives two reasons for thinking that I countenance too many. He claims (i) that my account of the de se and related cases opens the fl oodgates to too many proposition-building operations which allow too many proposition-candidates to be constructed, and (ii) that sometimes performing legitimate proposition-building acts in diff erent orders produces diff erent cognitive event types where there is only one proposition. I will take up these points separately. First the de se. What makes the fi rst-person way of thinking of oneself a proposition-building operation is its direct, non-descriptive role in inference and action. Because of this, admitting it (and related de se ways of cognizing) doesn’t commit me to other garden-variety “ways of thinking” as analogous proposition builders. De se ways of identifying predication targets diff er from King’s examples—thinking of

o as inhabiting a world in which water is H2 O and thinking of o as self-identical—in not introducing any new predications or functional applications. One can, of course,

think of o as the x: x = o & x inhabits a world in which water is H 2O, or as the x: x = o & x is self-identical. To do so in the service of predicating redness of o is to entertain a

proposition that requires one to apply fthe to the propositional function corresponding to the extra descriptive condition with the intention of predicating being red of the result. 11 Th ough these descriptive propositions exist, they are of a diff erent kind than the de se propositions to which King assimilates them. Th us, he is wrong to imagine that I am committed to propositions the constituents of which are simply o and redness, the entertainment of which requires o to be cognized in one of his ways. Comparing (1) and (2) provides further perspective. 1a. Scott Soames is the messy shopper. b. I am the messy shopper. (used de se by SS) 2a. Russell sought to prove logicism. b. Russell sought to prove that arithmetic is reducible to logic.

10 Justin Dallmann, “Existence and the Cognitive Event Type Th eory of Propositions,” (unpublished manuscript, USC). 11 I here employ the Fregean defi nite description operator fthe . Other choices are possible.

King020513OUK.indd 237 11/23/2013 12:59:17 PM 238 SCOTT SOAMES

Th e (immediate) constituents of the (a) and (b) propositions are the same in both cases, as is the form of predication (direct). Th e propositions diff er only in that the (b) propositions impose an extra requirement on the way in which a predication target must be cognized by one who entertains the proposition. Proposition (1b) requires SS to be cognized in the fi rst-person way; (2b) requires its propositional constituent to be entertained. Although in (1b) this extra requirement doesn’t involve any further predications (functional applications, etc.), in (2b) it does. However, in neither case does the extra requirement involve predicating anything further of (or operating in any further way on) the relevant predication target . Th is, I suspect, is what King missed in wrongly concluding that, for me, extra requirements on how predication targets are cognized can introduce new predications of, or operations on, them. Th ey can’t. Nor is it arbitrary that proposition-building acts can involve targeting propositions in a way that involves entertaining them (as in (2b)). Th is is simply the combination of two cognitive acts, both of which we know independently to be proposition building—entertaining propositions and directly targeting them (as we can do with anything with which we are acquainted or for which we have a name). Once all this is clear, King’s contention that proposition-building acts employed in my analyses of (1b) and (2b) overgenerate propositions, and so leave us with too many, can be seen to be groundless. Th is is not true of his second worry, which raises a real issue, albeit a minor one. Is the proposition that Romeo loves Juliet the cognitive act (or event type) in which loving is predicated of the pair consisting of Romeo followed by Juliet? Is it the act (or event type) in which one fi rst operates on loving and Juliet to form the property loving Juliet, which is predicated of Romeo? Or is it the act (or event type) in which this order is reversed? Since there are three slightly diff erent acts (or event types), it might seem that I am saddled with three diff erent propositions where there should be only one. Th ough puzzling, this issue is not, I think, very serious. One response would be to allow three diff erent but related Romeo-loves-Juliet propositions, while characterizing attitudes like judging, believing, and asserting in a way that guarantees that an agent who bears them to one of the three propositions bears them to all three. A diff erent response would be to identify the proposition that Romeo loves Juliet with the act (or event type) in which one either predicates loving of , or combines loving with Juliet and predicates the resulting complex property of Romeo, or combines loving with Romeo and predicates being one whom Romeo loves of Juliet. Short of investigating how to extend these (and perhaps other) strategies generalize, I will not here attempt to adjudicate between them. 12 Still, I see no reason to think that the issue (which arises for most accounts of structured propositions including King’s) can’t be resolved.

12 As Brian Bowman has reminded me, particular languages, like English, which recognize verb phrases as sentential constituents but not subject+transitive verb combinations, might constrain the proposition candidates expressible by their sentences. Even so, using such languages to report the attitudes of others would itself raise the issue.

King020513OUK.indd 238 11/23/2013 12:59:17 PM CLARIFYING AND IMPROVING THE COGNITIVE THEORY 239

Circularity? King’s next contention—that my account of propositions is circular because the explanation of their representational properties presupposes possibility, while my notion of a possible world-state presupposes propositions—is misguided on two counts. First, and foremost, his critique tacitly assumes that ordinary modal notions like what could be and what could not possibly be are conceptually dependent upon, and so to be analyzed in terms of, the conceptually prior notion of a possible world-state. As I have repeatedly argued, this Lewisian assumption couldn’t be further from the truth. 13 Although there are both epistemically and metaphysically possible world-states—ways the world could be (or have been)—they are defi ned in terms of our ordinary modal notions, rather than the other way around. As I explain in chapter 3, on my analysis, the notion of a proposition conceptually depends on objects, properties and cognitive acts of agents, while the notion of a possible world-state—i.e. a maximal property of a certain sort that the universe could have instantiated—conceptually depends on truth, propositions, and our ordinary modal notions. Since propositions don’t conceptually depend on possible world-states, an explanation of how they manage to be representational can make use of ordinary modal notions, including the possibility of a cognitive act being performed and an event type having instances, without circularity. Second, as I made clear in my reply to Speaks, a proposition represents things as being a certain way because it is either the cognitive act of representing things as being that way , or the event type of performing that act. Th ough it follows from this explanation that any possible performance of the act is one in which an agent represents things as being a certain way, it is not obvious that the explanation conceptually presupposes any modal notions at all (though even if it did, there would be no circularity). Agents, Acts, and Events King’s fi nal objection targets the substance of my (old) explanation of the representationality of propositions. Since my view of how this explanation should go has under- gone a subtle change in the last nine months, I have more sympathy with his critique than I once did. As he notes, in the past I have oft en given the explanation by claiming (i) that certain concrete cognitive events (e.g. of predicating redness of o) are inherently representational, and (ii) that event types inherit their representational properties from those of their instances. As indicated in the fi nal section of my reply to Speaks above, I now see the matter diff erently, and I hope more clearly. Th e explanation begins with agents. First, we observe that when an agent sees or thinks of o as red, the agent represents o as red. Next, we consider what the agent

13 S e e S o a m e s , “ Th e Place of David Lewis in Analytic Philosophy,” forthcoming in “Th e Place of David Lewis in Analytic Philosophy,” David Lewis, eds. Barry Loewer and Jonathan Schaff er, Oxford : Wiley-Blackwell ; chapter 5 of Soames , Th e Philosophy of Language , Princeton and Oxford : Princeton University Press , 2010 , and Soames , “Actually,” Aristotelian Society Supplementary Volume 81 , 2007 , 251–277 , reprinted in Soames, Philosophical Essays , Volume 2, Princeton and Oxford: Princeton University Press, 2009.

King020513OUK.indd 239 11/23/2013 12:59:17 PM 240 SCOTT SOAMES

does—namely represent o as red, which I call “predicating redness of o.” At this point, we appeal to a derivative sense of “represent” in which this act itself represents o as red . Th ough distinct from the primary sense in which an agent represents o as red, this extended sense is related to that primary sense in a way analogous to the way in which the sense in which some acts are intelligent, stupid, thoughtful, or kind is related to the primary sense in which agents who perform those acts are intelligent, stupid, thoughtful, or kind. Very roughly, (i) for an act to be intelligent or thoughtful is for it to be one the performance of which marks one as behaving intelligently or thoughtfully, and (ii) for a cognitive act to represent o as red is for it to be one the performance of which marks one as representing o as red. As indicated in my reply to Speaks, we, as agents, need this extended sense of representation in part because we wish to isolate individual aspects of the thought and perception of ourselves and others in order to assess them for accuracy. When o is such that to perceive or think of o as red is to represent it accurately, it is both enormously useful and very natural to seek an entity—a particular sort of perceiving or thinking— plus a property that entity has when this sort of perceiving or thinking is accurate. Th e entity is a proposition, which is either the cognitive act of representing o as red or the cognitive event type of so doing. Th e property is truth, which the act (or event type) has iff to perform it (or to bring about an instance of the event type) is to represent o as o really is. In What is Meaning? I ruled out acts as propositions on the basis of a short-sighted ordinary-language argument about what is or isn’t an absurd “category mistake” of the sort that fi lls the last few pages of King’s current critique of my view. 14 As I said in my response to Mark Richard at the session on What is Meaning? at the Eastern Division Meetings of the APA in December of 2011, I now see the error of those ways. Because our task is theory construction—which in philosophy as well as empirical science can, when successful, usher in new, surprising, and sometimes counterintuitive truths— ordinary-language style arguments that deny this have no more force against the act view of propositions than they do against the event-type view. 15 Since I no longer see a compelling reason to analyze propositions as event types as opposed to acts, I no

14 What is Meaning?, pp. 101–102. 15 It is, for example, common in the philosophy of language for propositions to be said to be the meanings of non-indexical sentences, despite the fact that this goes strongly against the grain of some of our ordinary ways of speaking about meaning. Th is is noted in Richard Cartwright’s classic article, “Propositions,” in R. J. Butler , ed., Analytical Philosophy , First Series, Oxford : Basil Blackwell , 1962 ; reprinted in his Philosophical Papers, Cambridge , MIT Press, 1987 . On pp. 49–50 of the latter he says, “If what someone asserts, on some occasion [namely a proposition] is itself the meaning which the words he utters have, on that occasion of their utterance, then anything predicable of what he asserts must also be predicable of the meaning of his words. But it is obvious on very little refl ection that ever so many things predicable of what is asserted cannot (on pain of nonsense) be predicated of the meaning of a sentence. And the fundamental point to be noticed in this connection is that although we may predicate of something asserted that it is (or was) asserted, this cannot be predicated of the meaning of a sentence. It simply makes no sense to say that someone asserted the meaning of a sentence [my emphasis]...Just as the meanings of sentences cannot be asserted, neither can they be affi rmed, denied, contradicted, questioned, challenged, discounted, confi rmed, supported, verifi ed, withdrawn, or repudiated; and whereas what is asserted can be said to be

King020513OUK.indd 240 11/23/2013 12:59:17 PM CLARIFYING AND IMPROVING THE COGNITIVE THEORY 241

longer see a serious objection to propositions as a species of purely representational cognitive acts.16 Th is revised way of looking at things circumvents King’s fi nal objection(s). Crucially, it debunks the idea that, on my view, the representational properties of acts or events (types or tokens) are simply transferred to them on the basis of the absurd supposition that every property of an agent who performs an act must also be a property of the act, or of events in which an agent performs it.17 Although my talk, from 2010 onward, of the derivative sense in which we speak of propositions as being representational was meant to signal that such a view was never in play, it is clear that more explicit discussion was needed, of the sort I have now provided. Despite his protest, it seems to me that King is in pretty much the same boat. For him, a proposition F (which he takes to be a very complex linguistically based fact) represents o as red, and so is true iff o is red, because agents use F to represent o as being that way. Of course, it is not the case that whenever agents use x to do y that x itself does y. I use a spoon to eat my soup, but my spoon doesn’t eat my soup. Since what I use something to do is not, in general, what it does, we may ask King a version of his own question. Why should we conclude from the alleged fact that an agent uses F to represent o as red that F itself represents o as red? To answer this question, King must, I suspect, appeal to what he must recognize to be an extended sense of “represents”—not of course the extended sense I invoke, but a diff erent one—that makes one of his facts representational dependent on what agents allegedly represent when using it. 18

accurate, exaggerated, unfounded, overdrawn, probable, improbable, plausible, true, or false, none of these can be said of the meaning of a [i.e. any] sentence.” Try it. Bill asserted/proved/contradicted/supported/ questioned/withdrew the proposition that mathematics is reducible to logic vs. *Bill asserted/proved/contradicted/supported/questioned/withdrew the meaning of the sentence “Mathematics is reducible to logic.” Whereas the former sound fi ne, the latter sound like category mistakes—incoherent or without sense (when they are not taken as suggesting some entirely diff erent content). Similarly for *Th e meaning of the sentence “Mathematics is reducible to logic” is plausible, probable, or untrue. But these are not incoherent or without sense, as Cartwright himself came to realize between 1967 and 1986. (See the addenda on pp. 52–53 for per- suasive argument.) But then, what reason is there to deny that some meanings may be propositions even though certain things truly attributable to propositions initially sound as if they couldn’t be true of meanings (and conversely)? Th ere is no good reason; the results of fruitful and systematic theorizing justify the revision of some of our ordinary, pretheoretic thought and talk. Th is is just as true in the case of successful theories that identify propositions with cognitive acts or event types as it is in the case of theories that identify the meanings of some sentences with propositions. 16 Chapters 3 and 6 were submitted in September of 2011 to Speaks and King for their criticism, before my change of mind on this point. Since those chapters could not be altered to refl ect this change of mind aft er my co-authors had begun working on their critiques, my restatement had to wait for this chapter on “further thoughts.” In fact, I now am now more inclined to identify propositions with cognitive acts than with event types. 17 King says, “In general, when an agent bears R to something o at a time [and so has the property bearing R to o ], the event token of the agent bearing R to o does not itself bear R to o. If I hug Annie, the event token of my hugging Annie doesn’t hug Annie. So why, from the fact that an agent represents o as red (by predicating redness of it), would it follow [my emphasis] that the event token of the agent representing o as red itself represents o as red?” Later he makes the same argumentative move concerning event types. 18 One might construe King in a slightly diff erent way—not as holding that agents use F in order that they may represent o as red, but rather that, already being able to represent o as red themselves, they use this ability to stipulate that the otherwise brute fact F is henceforth to be understood as representing o as red. However,

King020513OUK.indd 241 11/23/2013 12:59:17 PM 242 SCOTT SOAMES

What is to be explained? As I see it, the issue between King and me is which (if either) of these imagined ways of extending the primary sense in which agents represent is, or should be, in play when we think of, and theorize about, propositions as representational. Th is is a matter not of arbitrary stipulation, but of theoretical insight. More generally, the three theories of propositions sketched in this book are attempts to outline sound and fruitful conceptions capable of playing the roles for which propositions are needed in both philosophy and empirical science. It is true that each of us holds views that are at least mildly revisionary. Up to now, propositions— what is said, believed, etc.—have not ordinarily been thought to be either the cognitive acts/events that are central to my account, the complex linguistic facts that are central to King’s, or the complex properties central to Speaks’s conception. But, since the task is theory construction, this is of no great consequence. To assess our theoretical accounts, one must determine which best accommodates the most important features of our uncontentious pretheoretic talk of propositions, while providing us with entities that can play the theoretical roles for which we need propositions in philosophy, psychology, biology, linguistics, and philosophical logic. Th e following are a few of the facts that I think need to be explained by any successful theory of propositions: (i) that one who judges or affi rms that that o is red, himself represents o as red, and cognitively commits himself to o’s being so; (ii) that such an agent may thereby stand in the judging, affi rming, and believing relations to the proposition that o is red without having any conception of propositions, and without having the ability to represent them as bearing properties and standing in relations to anything; (iii) that agents with suffi cient cognitive sophistication can acquire the ability to represent propositions as having properties and standing in relations by focusing on their own cognitive acts and experiences of representing things as being one way or another, by grouping these acts and experiences into similarity types (on the basis of what things in the diff erent cases have been taken to be what ways), and by treating the diff erent types as units, thereby implicitly identifying propositions as what similar types have in common without forming any worked out positive conception of what these unities are; (iv) that judging, affi rming, or believing that that o is red does not require an agent to have mastered any language; agents could stand in these attitude relations to the proposition even if there were no sentences or languages at all;

this is not a plausible story for him to tell, in part because it presupposes that agents already bear propositional attitudes to the propositions his account is supposed to explain, and in part because agents do not have his enormously complex linguistic facts in mind as things to be endowed with representational properties by their stipulations.

King020513OUK.indd 242 11/23/2013 12:59:17 PM CLARIFYING AND IMPROVING THE COGNITIVE THEORY 243

(v) that all propositions represent things as being certain ways and so are true iff the things in question are as they are represented to be; (vi) that the proposition that o is red would represent o as being red, and could be true, even if there were no agents; (vii) that it is possible for one and the same proposition to be the content of a perceptual experience, a nonlinguistic thought, and an assertive utterance of a sentence; (viii) that the proposition that some past philosophers, including Socrates and Plato, don’t exist itself both exists and is true, even though Socrates and Plato no longer exist; (ix) that the proposition (a) that o is red is distinct from the propositions (b) that o is red and o is self-identical and (c) that o is red and 1st-order arithmetic is incomplete, and that one may stand in the affi rmation, judgment, or belief relations to (a) without standing in those relations to (b) and (c), but one cannot stand in those relations to (b) or (c) without standing in them to (a). (x) that the proposition that Cicero shaved himself, represents Cicero as having the property being a self-shaver, and so is distinct from the proposition that Cicero shaved Cicero, even though the same are true in the same metaphysically and epistemically possible world-states. (xi) Th at the propositions that Russell sought to prove that arithmetic is reducible to logic and that Russell sought to prove logicism are diff erent—since the latter can be asserted or believed by someone who doesn’t believe or assert the former—even though they represent precisely the same things as being precisely the same ways, and hence have identical truth conditions. (xii) that the points just made in (xi) also hold for the propositions that I wrote this chapter and that Scott Soames wrote this chapter. In taking (i—xii) to be facts, I am not claiming that they are the contents of privileged intuitions that must, if at all possible, be preserved by theories of propositions. As far as I can see, few, if any, of our strongly held pretheoretic convictions are so privileged. Rather, what are oft en called intuitions are things we strongly believe, frequently but not always with good reason, prior to conscious theorizing of the sort found in logic, philosophy, linguistics, or psychology. For this reason, it makes considerable sense that we should seek to preserve and explain most of them, while being ready to revise some of them when necessary Th is is the background against which one should judge pretheoretically surprising identifi cations of propositions with one or another class of entities. Since the explanations provided by the theory I have sketched seem superior to those provided by other theories, I judge it to be more likely to be correct than they are. According to it, propositions are, very roughly, ways of thinking, conceiving, or perceiving things to be. Although this doesn’t sound terribly surprising or counterintuitive, it becomes so when one gets more specifi c—identifying propositions with cognitive acts (in which

King020513OUK.indd 243 11/23/2013 12:59:17 PM 244 SCOTT SOAMES

case they become a species of things done) or with the event types in which one performs those acts (in which case they become a species of things that happen). It is true that both of these claims sound jarring at fi rst. However, the unrefl ective opinion that propositions can be neither things we do nor things that happen is not sacrosanct and may itself be due either to a failure to theorize, or to a tendency to do so incorrectly. It may also be true that any theory of propositions that leads to plausible explanations of facts like those illustrated by (i-xii) will lead to jarring surprises of its own. Since we are not in a position to rule this out in advance, we must not hobble ourselves by pro- hibiting surprises of the sort that I am willing to accept, or that my co-authors are. As I see it, success in our common enterprise will be success in identifying what agents have been referring to all along when speaking of propositions, and what properties they have ascribed to these entities when characterizing them as having been asserted or believed, or as having truth conditions—even if little of the theoretical detail about what these entities are, or how precisely we or they manage to represent the world, is something we are in a position to know without careful theory construction.19

19 Th is chapter has been greatly improved by Brian Bowman’s many helpful comments.

King020513OUK.indd 244 11/23/2013 12:59:17 PM Bibliography

Adams, R.M ( 1974 ). “ Th eories of actuality”. Noûs 8 ( 3 ): 211–231 . ______( 1981 ) . “Actualism and thisness”. Synthese 49 ( 1 ): 3–41 . Armstrong, J. and Stanley , J. (2011 ), Singular Th oughts and Singular Propositions, Philosophical Studies 154: 204–220 . Asher, N. (2000 ) . Events, Facts, Propositions and Evolutive Anaphora. In Speaking of Events , Higginbotham, Pianesi and Varzi eds., New York: Oxford University Press . Balaguer, M. (1998 ) . “Attitudes without propositions”. Philosophy and Phenomenological Research 58 ( 4 ): 805–826 . B a r w i s e , J . a n d P e r r y , J . ( 1 9 8 3 ) . Situations and Attitudes . Cambridge, MA: MIT Press . Braun, D. ( 1993 ) . “Empty names”. Noûs 27 ( 4 ): 449–469 . Burgess, J.P. ( 1983 ) . “Why I am not a nominalist”. Notre Dame Journal of Formal Logic 24 ( 1 ): 93–105 . Cartwright, R. ( 1962 ) . Propositions. In R. J. Butler (ed.), Analytical Philosophy, First Series. Oxford: Basil Blackwell. Reprinted in Cartwright ( 1987 ). ______( 1987a) . On the Origins of Russell’s Th eory of Descriptions. In Philosophical Essays , Cambridge MA : MIT Press , 1987 , 95–133 . ______( 1987b ). Philosophical Essays . Cambridge MA : MIT Press. Chalmers, D.J. (1996 ). Th e Conscious Mind: In Search of a Fundamental Th eory . Oxford: Oxford University Press. Chisholm, R.M. (1976 ). Person and Object: A Metaphysical Study. Chicago: Open Court. ______( 1981 ). Th e First Person: An Essay on Reference and Intentionality . Minneapolis, MN: University of Minnesota Press. Chomsky, N. (2000 ). New Horizons in the Study of Language and Mind . Cambridge: Cambridge University Press. Church, A. ( 1950 ) . “On Carnap’s Analysis of Statements of Assertion and Belief”. Analysis 10 ( 5 ): 97–99 . Collins, J. ( 2007 ) . “Syntax, More or Less”. Mind 116 ( 464 ): 805–850 . ______( 2011 ). Th e Unity of Linguistic Meaning . Oxford: Oxford University Press. Crimmins, M. (1998 ) . “Hesperus and Phosphorus: Sense, pretense, and reference”. Philosophical Review 107 ( 1 ): 1–47 . Dallmann, J. (unpublished) . Existence and the Cognitive Event Type Th eory of Propositions. Davidson, D. (1967 ) . “Truth and meaning”. Synthese 17 ( 1 ): 304–323 . Reprinted in Davidson (1984). ______( 1968 ) . “On saying that”. Synthese 19 ( 1–2 ): 130–146 . Reprinted in Davidson (1984). ______( 197 6 ) . Reply to Foster. In Truth and Meaning: Essays in Semantics , ed. G. Evans and J. McDowell. Oxford: Oxford University Press.______(1984 ). Inquiries Into Truth And Interpretation . Oxford: Oxford University Press. ______( 2005 ). Truth and Predication . Cambridge MA: Harvard University Press. Feit, N. ( 2008 ). Belief About the Self: A Defense of the Property Th eory of Content . Oxford: Oxford University Press.

King020513OUK.indd 245 11/23/2013 12:59:17 PM 246 BIBLIOGRAPHY

______( 2010 ) . “S elfl ess Desires and the Property Th eory of Content”. Australasian Journal of Philosophy 88 (3 ): 489–503 . Fine, K. (1985 ) . Plantinga on the reduction of possibilist discourse. In J. Tomberlin and P. van Inwagen (eds.), Alvin Plantinga ( Dordrecht : D. Reidel ): 145–186 . ______( 1994 ) . “Ess ence and mo da lity”. Philosophical Perspectives 8: 1–16 . ______( 2007 ). Semantic Relationism . Oxford: Blackwell. Fodor, J.A. ( 1990 ). A Th eory of Content and Other Essays . Cambridge, MA: MIT Press. Foster, J. (1976 ) . Meaning and Truth Th eory. In Truth and Meaning: Essays in Semantics . Oxford: Oxford University Press. Frege, G. ( 1918 ). Der Gedanke , Beitrage zur Philosophie des deutschen Idealismus , 1 , 1918 , 58–77 ; trans. “Th e Th ought” by A. and M. Quinton, in Mind 65, 1956 , 289–311; trans. “Th ought,” in Beaney, Th e Frege Reader , Oxford: Blackwell, 1997, 325–345, quoted at p. 333. ______( 1979 ) . A br ief Survey of my logical Doctrines. In Gottlob Frege: Posthumous Writings . Chicago: Th e University of Chicago Press. . Gaskin, R. ( 2008 ). Th e Unity of the Proposition . Oxford: Oxford University Press. Gilmore, C. (forthcoming) . Parts of Propositions. In Shieva Kleinschmidt (ed.), Mereology and Location . Oxford: Oxford University Press. Greenberg, G. ( 2011 ) . Semantics for Pictures. Unpublished ms. Grice, P. (1968 ) . “Utterer’s Meaning, Sentence Meaning, and Word Meaning”. Foundations of Language 4: 225–242 . ______( 1969 ) . “Utterer’s meaning and intention”. Philosophical Review 78 (2 ): 147–177 . Hanks, P. (2009 ) . “Recent work on propositions”. Philosophy Compass 4 ( 3 ): 469–486 . ______( 2011 ) . “St r uc ture d Prop osit ions as Typ es”. Mind 120 ( 477 ): 11–52 . Harman, G. ( 1974 ) . Meaning and semantics. In Milton K. Munitz & Peter K. Unger (eds.), Semantics and Philosophy . NY: New York University Press. Reprinted in Harman (1999). ______( 1999 ). Reasoning, Meaning, and Mind . Oxford: Oxford University Press. ______( 2003 ) . “C ategor y mist a kes in met aphysics and epistemology”. Philosophical Perspectives 17 ( 1 ): 165–180 . Hawthorne, J. and Manley , D. (2012 ), Th e Reference Book . Oxford: Oxford University Press. Jackson, F. ( 1998 ). From Metaphysics to Ethics . Oxford: Oxford University Press. J ohnston, M. (2004 ) . “Th e obscure object of hallucination”. Philosophical Studies 120 ( 1–3 ): 113–183 . Kaplan, D. ( 1977 / 1989 ) . Demonstratives. In Joseph Almog , John Perry & Howard Wettstein (eds.), Th emes From Kaplan . Oxford: Oxford University Press , 1989 Keller, L.J. . ( 2012 ). Whence Structured Propositions? Dissertation, University of Notre Dame. King, J.C. (1995 ) . “Structured propositions and complex predicates”. Noûs 29 ( 4 ): 516–535 . ______( 2002a ) . “D esignating propositions”. Philosophical Review 111 ( 3 ): 341–371 . ______(2002b ), Two S or ts of C laims ab out L og ica l For m. In Logical Form and Language , G. Preyer and G. Peter (ed.), New York : Oxford University Press , ( 2002 ) pp. 118–131 . ______( 2007 ). Th e Nature and Structure of Content . Oxford: Oxford University Press. ______( 2009 ) . “Questions of Unity”. Proceedings of the Aristotelian Society , V o l . CIX , Part 3 ______( 2011 ) . “On Fineness of Grain”. Philosophical Studies DOI: 10.1007/s11098-011-9844-9. ______( 2012 ). “ Prop osit iona l unity : w hat’s t he problem, who has it and who solves it?” Philosophical Studies DOI: 10.1007/s11098-012-9920-9 ______(ms.), “Acquaintance, Singular Th ought and Singular Propositions,” unpublished ms.

King020513OUK.indd 246 11/23/2013 12:59:17 PM BIBLIOGRAPHY 247

Kölbel, M. (2001 ) . “Two dogmas of Davidsonian semantics”. Journal of Philosophy 98 ( 12 ): 613–635 . Kripke, S.A. ( 2008 ) . “Frege’s theory of sense and reference: Some exegetical notes”. Th eoria 74 ( 3 ): 181–218 . Kroon, F. ( 2004 ) . “Descriptivism, pretense, and the Frege-Russell problems”. Philosophical Review 113 ( 1 ): 1–30 . Larson, R. and Segal, G. ( 1995 ). Knowledge of Meaning . Cambridge, MA: MIT Press. Lange, M. ( 2009 ). Laws and Lawmakers: Science, Metaphysics, and the Laws of Nature . Oxford: Oxford University Press. LePore, E. and Loewer, B. ( 1989 ) . “You Can Say Th at Again”. Midwest Studies in Philosophy 14 ( 1 ): 338–356 . ______( 2011 ). Meaning, Mind, and Matter: Philosophical Essays . Oxford: Oxford University Press. Lewis, D. ( 1970 ) . “General semantics”. Synthese 22 ( 1–2 ): 18–67 . ______( 197 5 ) . L anguages and language. In Keit h Gunders on (e d.), Minnesota Studies in the Philosophy of Science . University of Minnesota Press. ______( 1979 ) . “Att itudes de dicto and de se”. Philosophical Review 88 ( 4 ): 513–543 . Loewer, B. and LePore, E. (1981 ) . “Translational semantics”. Synthese 48 ( 1 ): 121–133 . McGlone, M. (2012 ) . “Propositional structure and truth conditions”. Philosophical Studies 157 ( 2 ): 211–225 . Moltmann, F. ( 2003 ) . “Propositional attitudes without propositions”. Synthese 135 (1): 77–118 . ______(for t hcoming). Abstract Objects and the Semantics of Natural Language. New York: Oxford University Press . Montague, R. (1973 ) . Th e proper treatment of quantifi cation in ordinary English. In Patrick Suppes , Julius Moravcsik & Jaakko Hintikka (eds.), Approaches to Natural Language. Dordrecht. Reprinted in Richmond Th omason, ed., Formal Philosophy: Selected Papers of Richard Montague , New Haven: Yale University Press. Nolan, D. ( 2006 ) . “Selfl ess desires”. Philosophy and Phenomenological Research 73 ( 3 ): 665–679 . Parsons, T. ( 1993 ) . “On denoting propositions and facts”. Philosophical Perspectives 7: 441–460 . Perry, J. (1977 ) . “Frege on demonstratives”. Philosophical Review 86 ( 4 ): 474–497 . ______( 1979 ) . “ Th e problem of the essential indexical”. Noûs 13 (December):3–21 . Plantinga, A. (1974 ). Th e Nature of Necessity . Oxford : Clarendon Press. Richard, M. ( 1993 ) . “Articulated terms”. Philosophical Perspectives 7:207–230 . ______( 2000) . Semantic pretense. In Anthony Everett and Th omas Hofweber (eds.), Empty Names, Fiction and the Puzzles of Non-existence , 205–232 . CSLI. R u s s e l l , B . ( 1 9 0 3 ) . Principles of Mathematics . London: Routledge. ______( 1904 ), “ Meinong’s Th eory of Complexes and Assumptions”. Mind , 13: three installments: 204–219 , 336–354 , 509–524 ; reprinted in Russell, Essays in Analysis (New York: George Braziller), 1973. ______( 1905 ) . “On D enot ing”. Mind 114 ( 456 ): 873–887 . ______( 1910 ) . On the nature of truth and falsehood. In Bertrand Russell (ed.), Philosophical Essays. London: Longmans , Green. Salmon, N.U. ( 1986 ). Frege’s Puzzle . Atascadero, CA: Ridgeview. ______( 1987). “Existence”. Philosophical Perspectives 1: 49–108 . ______( 1995) . Relational Belief. In On Quine: New Essays , ed. Paolo Leonardi and Marco Santambrogio , 206–228 . Cambridge: Cambridge University Press.

King020513OUK.indd 247 11/23/2013 12:59:17 PM 248 BIBLIOGRAPHY

______( 1998 ). “ Is Belief De Re Reducible to Belief De Dicto?” Canadian Journal of Philosophy , Supplementary volume 23: 85–110 . ______( 2005 ) . “On designating”. Mind 114 (456 ): 1069–1133 . Salmon, N.U. and Soames, S. (eds.) ( 1988 ). Propositions and Attitudes . Oxford: Oxford University Press. S c h i ff er, S.R. ( 1972 ). Meaning . Oxford: Clarendon Press. ______( 1987 ). Remnants of Meaning . Cambridge, MA: MIT Press. Siegel, S. ( 2006 ). Which properties are represented in perception? In Tamar S. Gendler & John Hawthorne (eds.), Perceptual Experience . Oxford: Oxford University Press. Soames, S. (1987 ). “Direct reference, propositional attitudes, and semantic content”. Philosophical Topics 15 ( 1 ): 47–87 . Reprinted in Salmon & Soames (1988) and in Soames (2009), v. 2. ______( 1992 ) . “ Tr ut h, me aning, and underst anding”. Philosophical Studies 65 (1–2 ): 17–35 . ______( 1994 ) . “Att itudes and anaphora”. Philosophical Perspectives 8: 251–272 . ______( 1999 ). Understanding Truth . Oxford: Oxford University Press. ______( 2002 ). Beyond Rigidity: Th e Unfi nished Semantic Agenda of Naming and Necessity . Oxford: Oxford University Press. ______( 2003 ) . “Underst anding defl ationism”. Philosophical Perspectives 17 (1 ): 369–383 . Reprinted in Soames (2009), v. 2. ______( 2005 ). Reference and Description: Th e Case Against Two-Dimensionalism . Princeton, NJ: Princeton University Press. ______( 2006) . Understanding assertion. In Judith Jarvis Th omson & Alex Byrne (eds.), Content and Modality: Th emes From the Philosophy of Robert Stalnaker. Oxford: Oxford University Press. Reprinted in Soames (2009), v. 2. ______( 2007a ) . “Ac tua l ly”. Aristotelian Society Supplementary Volume 81 ( 1 ): 251–277 . Reprinted in Soames (2009), v. 1. ______( 2007b ). “ What Are Natura l Kinds? ” Philosophical Topics 35 ( 1–2 ): 329–342 . ______( 2009 ). Philosophical Essays: Natural Language: What It Means and How We Use It , vol- umes 1 and 2. Princeton, NJ: Princeton University Press. ______( 2010 ). What Is Meaning? Princeton, NJ: Princeton University Press ______( 2011 ). Prop osit ions. In D elia Graff Fara & Gillian Russell (eds.), Th e Routledge Companion to Philosophy of Language. Routledge . ______(for t hcoming-a) Language, Mind, and Meaning: Th e Hempel Lectures . Princeton University Press , forthcoming. ______(for t hcoming-b) Th e Analytic Tradition in Philosophy, volume 1. Princeton University Press, forthcoming. ______(for t hcoming-c). David Lewis’s Place in Analytic Philosophy. In Barry Loewer & Jonathan Schaff er (eds.), David Lewis. Chichester: Wiley-Blackwell. Speaks, J. (2005 ). “Is there a problem about nonconceptual content?” Philosophical Review 114 ( 3 ): 359–398 . ______( 2006) . “Truth theories, translation manuals, and theories of meaning”. Linguistics and Philosophy 29 ( 4 ): 487–505 . ______(2012 ) . “On Possibly Nonexistent Prop osit ions”. Philosophy and Phenomenological Research 85 ( 3 ): 528–562 . Stalnaker, R. ( 1978 ) . “Assertion”. Syntax and Semantics 9:315–332 . Reprinted in Stalnaker (1999). ______( 1984 ). Inquiry . Cambridge: Cambridge University Press.

King020513OUK.indd 248 11/23/2013 12:59:17 PM BIBLIOGRAPHY 249

______( 1990 ) . “Mental content and linguistic form”. Philosophical Studies 58 ( 1–2 ): 129–146 . ______( 1999 ). Context and Content: Essays on Intentionality in Speech and Th ought . Oxford: Oxford University Press. ______( 2003 ). Ways a World Might Be: Metaphysical and Anti-Metaphysical Essays . Oxford: Oxford University Press. ______(2010 ) . Merely Possible Prop osit ions. In Modality: metaphysics, logic and epistemology , ed. by B. Hale and A. Hoff mann. Oxford: Oxford University Press Stanley, J. ( 2001 ) . “Hermeneutic factionalism”. Midwest Studies in Philosophy 25 ( 1 ): 36–71 . Turner, J. ( 2005 ) . “Strong and weak possibility”. Philosophical Studies 125 (2 ): 191–217 . ______( 2010 ) . “Fitt ing att itudes de dicto and de se”. Noûs 44 ( 1 ): 1–9 . van Inwagen, P. ( 1990 ). Material Beings . Ithaca, NY : Cornell University Press. ______( 2004 ) . A Th eory of Properties. In Dean W. Zimmerman (ed.), Oxford Studies in Metaphysics , Volume 1. Oxford: Clarendon Press. Walton, K.L. ( 1990 ). Mimesis as Make-Believe: On the Foundations of the Representational Arts . Harvard University Press. Williamson, T. ( 2002 ) . Necessary existents. In A. O’Hear (ed.), Logic, Th ought, and Language . Cambridge: Cambridge University Press. Yablo, S. ( 2000 ) . A Paradox of Existence. In Anthony Everett and Th omas Hofweber (eds.), Empty Names, Fiction and the Puzzles of Non-existence , 275–312 . CSLI.

King020513OUK.indd 249 11/23/2013 12:59:17 PM King020513OUK.indd 250 11/23/2013 12:59:17 PM Index

actualism 77 , 163 Fodor, Jerry 220 Adams, R. M. 6 , 57 , 134 Foster, J. 20 anaphora 69 , 100 , 111–2 , 180 , 206–7 Frege, Gottlob 1 , 2 , 25–6 , 29–3 , 37 , 47 , 61 , 72–3 , Armstrong, Joshua 208–14 78 , 97 , 109 , 117 , 131 , 216 articulated terms 113 A s h e r , N . 6 7 Gaskin, R. 1 Gilmore, C. 223 Balaguer, M. 17 Greenberg, G. 60 B a r w i s e , J o n 3 8 Grice, H.P. 220 Bowman, Brian 124 , 181 , 233 , 238 , 244 Braun, D. 245 Hanks, P. 1 Burgess, Alexis 84 , 89 , 90 Harman, Gilbert 22 , 24 , 67 Burgess, John 17 Hawthorne, J. 211 Higginbotham, Jim 50 Caplan, Ben 90 , 151 , 165 Hume, David 159 , 161 Cartwright, R. 119 , 240–1 Chalmers, David 42 intentionalism 157–8 , 193 , 195 Chisholm, Roderick 72 , 78–83 , 85 , 89 , 148 C h o m s k y , N 6 9 Jackson, Frank 42 Church, A. 12 Collins, John 1 , 48 , 60–4 Kaplan, David 42 , 99 , 109 components Keller, Lorraine 90 , 222–5 of facts 50 , 58–9 King, Annie 206 of event-types 118 Kolbel, Max 21 of properties and relations 190 , 212–3 Kripke, Saul 15 , 42 , 99 , 109 constituents K r o o n , F. 1 7 of facts. See components of properties 75 , 188 Lange, M. 89 of propositions 28–32 , 48 , 50 , 56–7 , 72–5 , 94 , language of thought 175 , 218 99 , 101 , 109 , 110 , 113–5 , 119–21 , 124 , 130–2 , Larson, R. 20 134 , 148–9 , 153 , 185 , 188–9 , 209–13 , 216 , Lee, Matthew 90 219–25 , 233 , 237–8 LePore, Ernest 13–16 , 22 of sentences 29 Lewis, David 72 , 78–83 , 85 , 86 , 108 , 110 , 148 , 168 , Crimmins, M. 17–18 186 , 220 , 239 Loewer, Barry 13–16 , 22 Dallmann, J. 237 Davidson, Donald 12–13 , 19–24 , 89 Manley, D. 211 defi nite descriptions 100–2 , 109 , 119–20 , 123 , M c G l o n e , M 1 180 , 202 , 237 Meinongianism 101–2 Millianism 101 , 105 , 117 , 119 , 123 , 223 event types 87 , 97 , 100–24 , 128–33 , 135–9 , 158–68 , Moltmann, M 1 223 , 226 , 228 , 230–1 , 234–41 Montague, R. 23

facts 2 , 7–8 , 20 , 44 , 51–9 , 65–72 , 75 , 87 , 95 , 136 , negation 98–9 , 103 , 161 , 186 , 206 139 , 146 , 151–2 , 156–8 , 162–81 , 190 , 196–7 , Nolan, Daniel 81–3 , 141 223–4 , 241 non-existents 101 , 130 , 237 Feit, N. 78 , 81 fi ctionalism 17–18 paratactic analysis 13–16 Fine, Kit 81 , 116–8 , 151 Parsons, T. 66–9 , 179

King_Index.indd 251 11/19/2013 3:12:23 PM 252 INDEX

perceptual experience 26 Russell, Bertrand 25–30 , 32–3 , 47 , 48 , 61 , 71–4 , content of 6 , 8 , 47 , 60 , 93–8 , 128–9 , 137 , 153 , 78 , 97 , 119–24 , 167 , 208 , 237–43 157–8 , 177 , 181 , 191–7 , 200–1 , 219 , 236 Perry, John 38 , 107–113 , 132 Salmon, N. 84 , 86 , 101 , 119 Plantinga, A. 222 Schiff er, S. 14 , 156 possible worlds 6 , 8 , 57–8 , 77 , 96–7 , 133–4 , Segal, Gabriel 20 189–90 , 212–3 , 222–4 , 239 semantic relationism 116–18 conception of propositions as sets of 2 , 16 , sentential relations 50–5 , 59–60 , 63 , 151–8 , 169 , 33–44 , 47 173–4 , 188 , 192–3 , 197 , 199–200 predication 27–34 , 37 , 72 , 75 , 86 , 95 , 97–100 , 102 , Siegel, S. 192 104 , 110–124 , 128–30 , 137–8 , 147 , 155 , 158–74 , Stalnaker, Robert 6 , 14 , 34 , 37 , 40–1 , 189 179 , 199 , 226–31 , 237 , 238 Stanley, J. 17 , 208–14 proposition syntax -al functions 99–100 , 111–2 , 237 relations of 29 , 48–9 , 54–64 , 74–5 , 93–4 , -al relations 50–63 , 147 , 152–60 , 172–7 , 187 , 152–5 , 170–9 , 188–96 , 209–10 191–200 , 209–13 semantic signifi cance of 49 , 55 , 57 , 59 , 75 , 111, entertaining a 55 , 97–9 , 104–8 , 110–1 , 115–8 , 155 , 172–3 , 193–6 122 , 124 , 129 , 132–3 , 135 , 138 , 158–61 , 167 , 172–6 , 199–200 , 223 , 230–2 , 234 , 236 , 238 “that” clauses 7–8 , 9–19 , 64–70 , 87–8 , 93 , 105–6 , unity of 26–33 , 97 174–5 , 178–80 , 201–8 representational properties of 29–31 , 33–5 , 48 , t r u t h 77–8 , 92–6 , 105–9 , 113 , 118 , 127 , 132–5 , 147–9 , conditions 17 , 27 , 31–7 , 47–8 , 51–5 , 58–61 , 64 , 150–67 , 174–6 , 185–6 , 189–90 , 193 , 195 , 197 , 80 , 91–7 , 106 , 108 , 110 , 116–8 , 120 , 124 , 127–8 , 215–21 , 225–31 , 234–6 , 239 133–40 , 145–52 , 156 , 167–9 , 173–7 , 187–93 , singular 41 , 76 , 101 , 107 , 112 , 114–5 , 120 , 196–200 , 204 , 220–1 , 227 , 235 149–50 , 154 , 157 , 187 , 192 , 209 , 211 , 223 defl ationism about 103–4 propositional attitudes 2 , 6 , 39–40 , 44 , 53 , functions 98–9 , 119 , 208 , 213 59–60 , 77–86 , 117 , 140–3 , 155–61 , 171–6 , monadic vs. dyadic 35–8 191–7 , 217–9 , 229–30 T-sentences 19–24 de se 106–15 , 132–3 , 141 , 167–8 , 237 . See also Turner, Jason 78 , 81 propositional attitudes, fi rst personal Two-dimensional semantics, 38 , 42–4 fi rst personal 80–1 , 85–6 , 141 ascriptions of 7 , 12–15 , 39–40 , 79–88 , 99 , van Inwagen, Peter 12 , 72 , 77 , 90 , 148 112 , 141 variables 39–40 , 44 , 56 , 112 , 154 , 170–1 , 178–9 assignment of values to 44 , 56–8 , 63 , Q u i n e , W. V. 8 6 100 , 110 , 112 , 151 , 154 , 170–1 , 178–9 , 188 , 217 R e c a n a t i , F r a n c o i s 9 4 representational properties Walton, K. 18 as intrinsic 33 , 78 , 133 , 146–8 , 160 , 167 , 215–6 Williamson, T. 81 of maps 47 , 60 , 167 , 191 , 201 , 219 , 221 See also perceptual experience, proposition Ya b l o , S . 1 7 Richard, Mark 17–18 , 105 , 211 , 240 Yli-Vakkuri, Juhani 90

King_Index.indd 252 11/19/2013 3:12:24 PM