The Indexical Point of View: On Cognitive Significance and Cognitive Dynamics [1 ed.] 0367554801, 9780367554804

Table of contents :
Cover
Half Title
Series Page
Title Page
Copyright Page
Table of Contents
Preface
Introduction
Overview
Preliminaries
An Outline of the Book
1 Character, Content, and Cognitive Significance
1.1 Introduction
1.2 Character and Cognitive Significance
1.3 A Problem with Characters: Demonstratives
1.4 A Problem with Characters: Pure Indexicals
2 Other Kinds of Content and Cognitive Significance
2.1 Introduction
2.2 Official and Reflexive Content
2.3 Utterances, Beliefs, and Cognitive Significance
2.4 Notion-Networks
2.5 Conclusion
3 Anti-individualism and Cognitive Perspective
3.1 Introduction
3.2 Sameness and Difference of Thought-Contents
3.3 Opacity of Thought-Contents
3.4 Transparency of Thought-Contents
3.4.1 Transparency of Difference
3.4.2 Transparency of Sameness
3.4.3 The Enterprise Case
3.4.4 Internal Facts
3.4.5 Concepts
3.4.6 Concluding Remarks
3.5 The Neo-Fregean View
3.6 Transparency Again
3.7 Senses
3.8 Conclusion
4 Cognitive Dynamics, Belief Retention, and Cognitive Significance
4.1 Introduction
4.2 The Problem of Cognitive Dynamics
4.3 The Fregean View
4.4 Kaplan’s View
4.5 Linguistic Meaning and Thinking about Days
4.6 Thoughts and Transparency
4.7 Sourceless Beliefs
4.8 The Content of Thought
5 Beliefs and Characters
5.1 Introduction
5.2 Doxastic Characters
5.3 The Mismatch between Characters and Belief States
5.3.1 The Existence of Character Is Not Tied to the Existence of Belief State
5.3.2 The Existence of Belief State Is Not Tied to the Existence of Character
5.3.3 Identity of Character Is Not Tied to Identity of Belief State
5.4 The Retention of Thought
5.5 Modes of Presentation and Meaning
6 Slicing Thoughts
6.1 Introduction
6.2 Individuating Thoughts
6.3 Blocking the Intrapersonal Proliferation of Thoughts
6.4 Stretching Thoughts
6.5 The Sameness of Thoughts
6.6 Blocking the Interpersonal Proliferation of Thoughts
6.7 Conclusion
7 How Many Modes of Presentation Do We Need?
7.1 Introduction
7.2 Re-Identification of Objects
7.3 The Modal and the Non-modal Criteria of Difference
7.4 Against the Modal Criterion of Difference
7.5 A Case for the Modal Criterion of Difference
7.6 Name-Like Modes of Presentation
7.7 Demonstrative Modes of Presentation
7.8 Temporal Modes of Presentation
7.9 Nominal Terms and Indexicals
8 Tracking and Reporting
8.1 Introduction
8.2 Tracking Indexicals
8.3 Personal Pronouns
8.4 ‘You’, ‘I’, and Cognitive Significance
8.5 Token Meanings
8.6 Roles Strike Back: The Interpersonal Case
8.7 Roles Strike Back: The Intrapersonal Case
8.8 Correspondence Relations
8.9 Correct Speech Reports
9 Conclusion
9.1 Introduction
9.2 Singular Thoughts and Acquaintance
9.3 The Distinctness of Indexical Beliefs
9.4 Mental Files
Notes
References
Index

Citation preview

The Indexical Point of View

This book argues that there is a common cognitive mechanism underlying all indexical thoughts, in spite of their seeming diversity. Indexical thoughts are mental representations, such as beliefs and desires. They represent items from a thinker’s point of view or her/his cognitive perspective. We typically express them by means of sentences containing linguistic expressions such as ‘this (F)’ or ‘that (F)’; adverbs like ‘here’, ‘now’, and ‘today’; and the personal pronoun ‘I’. While generally agreeing that representing the world from a thinker’s cognitive perspective is a key feature of indexical thoughts, philosophers disagree as to whether a thinker’s cognitive perspective can be captured and rationalized by semantic content and, if so, what kind of content this is. This book surveys competing views and then advances its own positive account. Ultimately, it argues that a thinker’s cognitive perspective – or her indexical point of view – is to be explained in terms of the content that is believed and asserted as the only kind of content that there is which thereby serves as the bearer of cognitive significance. The Indexical Point of View will be of interest to philosophers of mind and language, linguists, and cognitive scientists. Vojislav Bozickovic is Professor of Philosophy at the University of Belgrade, Serbia. Previously, he was lecturer at Macquarie University and the University of Tasmania. His published work has appeared in American Philosophical Quarterly, Erkenntnis, Linguistics and Philosophy, Philosophical Psychology, and Synthese.

Routledge Studies in Contemporary Philosophy

Perception and Reality in Kant, Husserl, and McDowell Corijn van Mazijk Social Functions in Philosophy Metaphysical, Normative, and Methodological Perspectives Edited by Rebekka Hufendiek, Daniel James, and Raphael van Riel Microaggressions and Philosophy Edited by Lauren Freeman and Jeanine Weekes Schroer Cross-Tradition Engagement in Philosophy A Constructive-Engagement Account Bo Mou Perception and the Inhuman Gaze Perspectives from Philosophy, Phenomenology, and the Sciences Edited by Anya Daly, Fred Cummins, James Jardine, and Dermot Moran Logics of Genocide The Structures of Violence and the Contemporary World Edited by Anne O’Byrne and Martin Shuster Revising Fiction, Fact, and Faith A Philosophical Account Nathaniel Goldberg and Chris Gavaler The Indexical Point of View On Cognitive Significance and Cognitive Dynamics Vojislav Bozickovic

For more information about this series, please visit: https://www.routledge.com

The Indexical Point of View On Cognitive Significance and Cognitive Dynamics

Vojislav Bozickovic

First published 2021 by Routledge 52 Vanderbilt Avenue, New York, NY 10017 and by Routledge 2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN Routledge is an imprint of the Taylor & Francis Group, an informa business © 2021 Vojislav Bozickovic The right of Vojislav Bozickovic to be identified as author of this work has been asserted by him in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging-in-Publication Data A catalog record for this title has been requested ISBN: 978-0-367-55480-4 (hbk) ISBN: 978-1-003-09426-5 (ebk) Typeset in Sabon by codeMantra

Contents

Preface Introduction Overview 1 Preliminaries 7 An Outline of the Book 17

ix 1

1

Character, Content, and Cognitive Significance 1.1 Introduction 21 1.2 Character and Cognitive Significance 21 1.3 A Problem with Characters: Demonstratives 23 1.4 A Problem with Characters: Pure Indexicals 25

21

2

Other Kinds of Content and Cognitive Significance 2.1 Introduction 27 2.2 Official and Reflexive Content 27 2.3 Utterances, Beliefs, and Cognitive Significance 30 2.4 Notion-Networks 35 2.5 Conclusion 37

27

3

Anti-individualism and Cognitive Perspective 3.1 Introduction 39 3.2 Sameness and Difference of Thought-Contents 39 3.3 Opacity of Thought-Contents 43 3.4 Transparency of Thought-Contents 45 3.4.1 Transparency of Difference 46 3.4.2 Transparency of Sameness 49 3.4.3 The Enterprise Case 51 3.4.4 Internal Facts 52 3.4.5 Concepts 53 3.4.6 Concluding Remarks 55

39

vi

Contents 3.5 3.6 3.7 3.8

4

The Neo-Fregean View 56 Transparency Again 58 Senses 60 Conclusion 62

Cognitive Dynamics, Belief Retention, and Cognitive Significance 4.1 Introduction 63 4.2 The Problem of Cognitive Dynamics 64 4.3 The Fregean View 64 4.4 Kaplan’s View 66 4.5 Linguistic Meaning and Thinking about Days 68 4.6 Thoughts and Transparency 71 4.7 Sourceless Beliefs 73 4.8 The Content of Thought 74

63

5

Beliefs and Characters 5.1 Introduction 77 5.2 Doxastic Characters 77 5.3 The Mismatch between Characters and Belief States 82 5.3.1 The Existence of Character Is Not Tied to the Existence of Belief State 82 5.3.2 The Existence of Belief State Is Not Tied to the Existence of Character 82 5.3.3 Identity of Character Is Not Tied to Identity of Belief State 83 5.4 The Retention of Thought 84 5.5 Modes of Presentation and Meaning 87

77

6

Slicing Thoughts 6.1 Introduction 90 6.2 Individuating Thoughts 90 6.3 Blocking the Intrapersonal Proliferation of Thoughts 93 6.4 Stretching Thoughts 95 6.5 The Sameness of Thoughts 96 6.6 Blocking the Interpersonal Proliferation of Thoughts 99 6.7 Conclusion 100

90

7

How Many Modes of Presentation Do We Need? 102 7.1 Introduction 102 7.2 Re-identification of Objects 102 7.3 The Modal and the Non-modal Criteria of Difference 103

Contents  vii 7.4 7.5 7.6 7.7 7.8 7.9

Against the Modal Criterion of Difference 107 A Case for the Modal Criterion of Difference 109 Name-Like Modes of Presentation 110 Demonstrative Modes of Presentation 111 Temporal Modes of Presentation 114 Nominal Terms and Indexicals 116

8

Tracking and Reporting 8.1 Introduction 122 8.2 Tracking Indexicals 122 8.3 Personal Pronouns 123 8.4 ‘You’, ‘I’, and Cognitive Significance 123 8.5 Token Meanings 125 8.6 Roles Strike Back: The Interpersonal Case 127 8.7 Roles Strike Back: The Intrapersonal Case 128 8.8 Correspondence Relations 130 8.9 Correct Speech Reports 131

122

9

Conclusion 9.1 Introduction 133 9.2 Singular Thoughts and Acquaintance 133 9.3 The Distinctness of Indexical Beliefs 135 9.4 Mental Files 136

133

Notes References Index

141 157 163

Preface

This is not yet another work that dwells on the issue of the relationship between singular and descriptive thoughts. The way it handles indexical thoughts enables us to bypass this time-worn issue altogether thanks to securing that indexical thoughts are thoughts of a distinct type no matter whether they are based on acquaintance or not, which way we also bypass the charge that there is no good reason to impose acquaintance constraints of any sort on reference or singular thought. It also enables us to establish that there is a common cognitive mechanism underlying all indexical thoughts, in spite of their diversity on the surface level. This book draws upon years of delving into specific flaws that I found with the prominent representative views about indexical beliefs. It took me a while to articulate these flaws, let alone to decipher those flaws that these views have in common. Once I achieved this, an alternative picture of indexical beliefs dawned on me which enabled me to steer through these prominent views in the process of developing it. During this process, many of the previous versions of the book have been but totally discarded until I was left with nothing to hang on to except for a few published articles that reluctantly prompted me to kick a new start. The new manuscript emerged, but it left much to be desired. Yet another attempt followed, and somewhat surprisingly, things started falling into place. As expected, the output of the new literature on the subject continued to pour out that was in need of being accounted for every time I took a repeated look at each of the book chapters, but there came a time when this process had to end lest I go back to square one. I wish to thank the Central European University in Budapest for granting me a six-month research fellowship that allowed me to muse on the central issues that this book revolves around and the Budapest colleagues for their hospitality. My thanks also go to Aleksandra Sojic, Miljana Milojevic, Una Stojnic, Masan Bogdanovski, and most of all Timothy Williamson for their help and encouragement. Last but not least, I am indebted to the readers for this book and their valuable comments as well as Andrew Weckenmann and Allie Simmons of Routledge Press working with whom has been a great pleasure.

x Preface Chapter 2 is based on my ‘Cognitive significance and reflexive content’, Linguistics and Philosophy 31 (2008). I am grateful to Springer Nature for allowing me to use this material. Chapter 3 includes my ‘Anti-individualism and transparency’, Synthese 197 (2020). I am grateful to Springer Nature for allowing me to use this material. Chapter 4 is based on my ‘Belief retention: a Fregean account’, Erkenntnis 80 (2015). I am grateful to Springer Nature for allowing me to use this material. Chapter 5 includes parts of my ‘Do characters play a cognitive role?’, Philosophical Psychology 18 (2005). I am grateful to Taylor & Francis Publishing for allowing me to use this material. (https://www.tandfonline.com/) Chapter 6 is based on my ‘Slicing thoughts’, American Philosophical Quarterly 54 (2017). I am grateful to The University of Illinois Press for allowing me to use this material.



Overview Indexical thoughts – mental representations such as beliefs and desires – that we recognize by those linguistic expressions that contain demonstratives such as ‘this (F)’ or ‘that (F)’, indexicals such as ‘here’, ‘now’, ‘today’, the first-person singular pronoun ‘I’, and the like, represent the items they are about from a thinker’s point of view or her cognitive perspective. We think that this [observed] ship is an aircraft carrier, that it is dark here [in the room that we are in], that it is late for tea now [at the present time], that it is sunny today [the present day], and the like. While generally agreeing that representing the world from a thinker’s point of view is a key feature of indexical thoughts, philosophers disagree as to whether her point of view is to be captured and rationalized by semantic content, and if so, whether the content in question is the content that is believed and asserted or some other kind of content that some philosophers have come up with for this purpose. The disagreement comes to the fore when it comes to cases in which a thinker takes a single thing for two different ones or vice versa. By examining relevant attempts that have been made to account for such cases, I will argue that a thinker’s cognitive perspective – her indexical point of view – is to be explained in terms of the content that is believed and asserted as I take it to be, as the only kind of content that I admit of, which thereby serves as the bearer of cognitive significance in the sense to be specified below. Frege, who started the ball rolling by dealing both with indexical thoughts and cases in which a thinker takes a single thing for two different ones, holds that only a sentence complete in every respect expresses a thought as its informational content (Frege 1918/1977, 28), which also serves as the content of one’s belief. Unlike an indexical thought, a thought that is expressed by such a sentence does not represent the world from a thinker’s point of view. The sentence ‘The evening star is a body illuminated by the Sun’ is one such sentence (see Frege 1892/1980, 62). The thought that this sentence expresses contains as its constituent the sense of the expression ‘The evening star’ which is supplied by this expression’s linguistic or conventional meaning. In understanding the meaning

2 Introduction of ‘The evening star’, one grasps its sense via which it designates Venus. The expression ‘The evening star’ designates the same heavenly body as the expression ‘The morning star’ but they have different meanings and thereby senses. The two senses represent Venus from two different meaning-based points of view rather than from a thinker’s point of view which is something like that [observed] heavenly body, etc. The different cognitive perspectives over Venus that the two senses provide are a matter of semantics in making it the case that a certain object is being designated by them. As Frege put it, a = a and a = b are obviously statements of differing cognitive value (Frege 1892/1980, 56). This makes the semantic accounts of ‘a’ and ‘b’ different. In Frege’s own words: If the sign ‘a’ is distinguished from the sign ‘b’ only as an object (here by means of its shape), not as a sign (i.e. not by the manner in which it designates something), the cognitive value of a = a becomes essentially equal to that of a = b, provided a = b is true. A difference can arise only if the difference between the signs corresponds to a difference in the mode of presentation of the thing designated. (Frege 1892/1980, 57) Hence, the mode of presentation associated with ‘a’ differs from the mode of presentation associated with ‘b’. It is the expression’s sense ‘wherein the mode of presentation is contained’ (Frege 1892/1980, 57). Although for Frege sense serves various theoretical purposes, whenever it is safe to do so, I will speak of sense and mode of presentation interchangeably as do most of the authors that I discuss in the book. (In so doing, I will deploy one or the other of these two expressions depending on the way of speaking favoured by a particular author.) Unlike the sentence ‘The evening star is a body illuminated by the Sun’, which as “a sentence complete in every respect expresses a thought”, the sentence ‘Today is beautiful’ does not express a thought since the linguistic meaning of the indexical ‘today’ that this sentence contains as its constituent does not supply its (complete) sense. This is evidenced by Frege’s well-known claim that if someone wants to say today what he expressed yesterday using the word ‘today’, he will replace this word with ‘yesterday’ since the change of sense would otherwise be effected by the differing times of utterance (Frege 1918/1977, 10). This is to say that the same thought can be expressed by utterances of sentences that contain indexicals that belong to different meaning types. The belief that day d is beautiful that the subject holds and expresses by ‘Today is beautiful’ on day d she will re-express and update on day d + 1 by ‘Yesterday was beautiful’ in the normal course of circumstances. Conversely, different thoughts will be expressed by different utterances of the same sentence that belongs to the same meaning type, such as

Introduction  3 ‘Today is beautiful’ as uttered on d and d + 1, respectively, by the same or different persons. In view of this, Frege comes up with two notions of sense: one accounting for the cognitive value that a sign conveys, and the other representing the truth conditions of a sentence, between which there is a tension (see the next section). By examining representative attempts to resolve this tension, I will extract and defend a view that is encouraged by their inability to resolve it. I start off by taking a close look at Kaplan’s and Perry’s seminal attempts to resolve this tension. They replace Frege’s one level (thought-) content by two levels of content. As referentialists, they hold that the content that is believed and asserted is Russellian. The content that (an utterance of) an indexical sentence such as ‘Today is beautiful’ expresses consists of the day being referred to and the property of being beautiful. But there is another level of content. It accounts for cognitive significance conveyed by ‘today’, i.e. for how the content that is expressed is believed or thought of. For now, we can think of cognitive significance of ‘today’ as its linguistic meaning: the present day. (I shall be more specific about this in due course.) Similar to the linguistic meaning of ‘The evening star’, this kind of content determines the referent of (an utterance of) ‘today’ by providing a cognitive perspective over the referent. But, unlike in ‘The evening star’ case, in the ‘today’ case, this is achieved in a context-dependent way in that on different days of utterance, the linguistic meaning of ‘today’ determines different referents, i.e. different days, by providing the same type of cognitive perspective (the present day) over different referents. However, there are problems with this view. One problematic kind of case is one in which a subject who, say, in pointing to Venus in the morning assents to an utterance of the sentence ‘This body is illuminated by the Sun’, and dissents from another utterance of the same sentence in pointing to Venus in the evening. The Russellian content expressed and believed is in the two cases the same as is the linguistic meaning of the two utterances. Yet, these utterances convey different cognitive values and differ in cognitive significance in that one may rationally accept one of them as true but not the other. This outcome has led some referentialists (such as Wettstein 1986, 2004, and Almog 2008) to hold that it is thankfully not part of the business of semantics to deal with cognitive significance, its business being to get the truth conditions right and tell us what content (proposition) various sentences express. Yet, I follow Kaplan and Perry in holding that it is part of the business of semantics to explain cognitive significance. A correct semantic theory needs to provide us with an appropriate interface between what sentences mean (express) and how we use them to communicate beliefs in order to motivate and explain action

4 Introduction (see Perry 2012, 9; see also Perry 1988/1993, as well as Korta and Perry 2011, 56).1 I also follow them (in what they follow Frege) in holding that in being explanatory of the subject’s reasoning and actions, bearers of cognitive significance need to account for the subject’s cognitive perspective over the referent. If, for example, the subject takes it that this body [pointing to Venus in the morning] is not the same as this body [pointing to Venus in the evening], her cognitive perspective over Venus will be different in these two cases which in turn needs to be explained in terms of the difference in the cognitive significance that the two utterances convey. This calls for an examination of those belief states that we associate with uses of perception-based demonstratives such as ‘this’ and ‘that’ (‘this body’, in the foregoing example, and the like). This is the focus of the first three chapters of the book. First, I examine Kaplan’s and Perry’s seminal views which serve as the starting point of my inquiry (Chapter 1). Following this, I turn to Perry’s more recent and elaborate view of cognitive significance that postulates new levels of content in addition to Russellian content (Chapter 2). I argue that neither of these views is satisfactory in accounting for the noted differences in cognitive value by discussing cases such as the one just noted in which the subject respectively refers to Venus by uttering ‘This body is illuminated by the Sun’ in the morning and in the evening, but mistakenly takes it for two different heavenly bodies. In so doing, I suggest a better alternative, not least because Perry’s attempt to account for these differences in terms of second-order belief-contents is at odds with the acknowledged fact that the subject’s reasoning and actions are a matter of her first-order beliefs about the world. I then discuss yet another level of content that Perry postulates in terms of which Lawlor (2007) attempts to account for the cognitive perspective of the subject in the reverse kind of case in which, say, she first points and refers to the evening star and then (at the appropriate time of day) to the north star and mistakes them for the same heavenly body. I argue that, its intentions notwithstanding, this attempt fails to account for the subject’s cognitive perspective. The same holds for those anti-individualist views that try to account for the subject’s cognitive perspective in both such cases which I discuss in Chapter 3. To be sure, according to the referentialist view, Russellian content is not an internal property of the subject. Similarly, the guiding idea of anti-individualism is that the content believed and asserted, as this view takes it to be, is not an internal property of the subject. But, unlike the referentialists, a number of anti-individualist philosophers hold that content that is believed and asserted serves to explain the subject’s cognitive perspective. My focus in Chapter 3 is on the representative varieties of anti-individualism where, in relation to perception-based demonstrative thoughts, I argue that, their intentions notwithstanding they fail to

Introduction  5 provide a theory of content that explains the subject’s cognitive perspective. Then I propose an alternative theory of content that explains the subject’s cognitive perspective. Another problematic kind of case for the ‘Kaplan–Perry’ view concerns the so-called temporal indexicals ‘today’ and ‘yesterday’ as well as ‘tomorrow’, which I tackle in Chapter 4. The discussion in this chapter shifts from perception-based demonstrative thoughts to thoughts that we express by utterances of sentences that contain temporal indexicals and the problem of cognitive dynamics – the question of what standard adjustment the subject needs to make in the expression of her indexical thought in order to express the same belief over time. Namely, Kaplan considers seriously the strategy of adjusting verbal expression in order for the same belief to be expressed as the context changes, in line with Frege’s aforementioned suggestion concerning the ‘today’/‘yesterday’ case. But Kaplan finds that this strategy falls short of supplying us with some obvious standard adjustment to enable us to account for those cases in which a subject such as Rip Van Winkle has lost track of time. Suppose that on day d, before falling asleep, Rip never forms any explicit belief other than the one that he expressed by ‘Today is beautiful’. When he awakes 20 years later, the belief is updated, given Rip’s view of how the context has changed, in terms of ‘Yesterday was beautiful’. What is left of the original belief, according to the suggested strategy, is a belief about the day before he woke up – the day that ‘yesterday’ designates in virtue of its linguistic meaning (in the context of utterance). This Kaplan finds strange since Rip seems to remember (that) d (was a beautiful day). Having noticed this problem, Kaplan does not try to solve it. But he makes a clear point that retaining a belief with which a subject such as Rip Van Winkle began requires an internal continuity of belief, as does Frege on the evidence of his foregoing remarks on the ‘today’/‘yesterday’ case. Endorsing this, I argue that the internal continuity in the subject’s belief needs to be accounted for in terms of Frege’s claim that representing a certain day as the same from one occasion to the next is to think of it via the same sense, i.e. under the same mode of presentation – which makes the thought of which it is a constituent the bearer of cognitive significance. Cognitive significance is of a piece with the internal continuity of the subject’s belief, and we need to deal with it in accounting for belief retention in deploying the temporal indexicals ‘today’ and ‘yesterday’ (even if it turns out that cognitive significance is not part of the business of semantics). In Chapter 5, I compare the proposed account of the link between cognitive significance and the internal continuity of the subject’s belief with an alternative view proposed by Perry and show that this view is implausible. Following this, I provide a positive account of belief retention in relation to the ‘today’/‘yesterday’ case. To get a further grip on the notion of cognitive significance and the related content, I introduce

6 Introduction a criterion of difference for thoughts derived from Frege both in its neo-Fregean and its referentialist formulation. In the light of this criterion, Chapter 6 looks at how this criterion ties in with the issue of belief retention. I move forward with my discussion of keeping track of time by means of deploying temporal indexicals to a discussion of keeping track of perceived objects. By taking a close look at the standard version of this criterion, I argue that applying it to perception-based demonstrative thoughts slices thoughts too finely which leads to their proliferation and makes the criterion implausible. I block this proliferation by transforming this criterion into a related one that is shown to be essential in individuating thoughts as they are conceived of in this book. The key notion here turns out to be our taking for granted the identity of the individual thought about (synchronically or diachronically), consisting in the fact that the subject cannot sensibly raise the question of its identity. This will show that our unreflective assumptions about the sameness or difference of such individuals, which play a foundational role in our reflective employment of concepts, supply us with a common cognitive mechanism that underlies all indexical thoughts. While I argue that representing an object or day from one occasion to the next amounts to thinking about it under a single mode of presentation, some philosophers have argued that there are two different kinds of mode of presentation here. In Chapter 7, I discuss their views and show them to be unjustified. This lends further support to my aforementioned contention that a thinker’s cognitive perspective or her indexical point of view is to be explained in terms of a single kind of content. By broadening our discussion to also cover cases of re-identification of objects, I show how we need to individuate modes of presentation as thoughtconstituents in capturing the subject’s cognitive perspective. The cognitive perspectives of the utterer of ‘I was wounded’ and the utterer of ‘You said you were wounded’ addressed to the utterer of the former sentence differ. The former involves the first person point of view while the latter does not. This suggests that they cannot think of the addressee in the same way. In spite of this, and in keeping with Frege’s demand that in a correct speech report, the thought expressed by ‘I was wounded’ needs to be the thought that is being referred to in ‘You said you were wounded’, I show how this demand can be met. In so doing, I  pay due respect to Frege’s underlying assumption that the semantic content of ‘I’ and ‘you’ needs to be objective and interpersonally stable and shareable. This is the subject matter of Chapter 8. In Chapter 9, I strengthen my position by arguing that it is not committed to some views that have recently been disputed. In this context, I show that indexical beliefs are singular beliefs in a way that does not call for an examination of the relationship between indexical beliefs and those beliefs that we typically express by sentences containing definite descriptions. This enables us to bypass a time-worn debate about the relationship between singular and descriptive beliefs.

Introduction  7 Of the vast literature on the issues with which I deal in this book, I survey those prominent and influential rival views which I find to be most relevant for the goals that this book seeks to achieve. One goal is to establish the nature of the link between cognitive significance and cognitive dynamics (i.e. belief retention); another, related to the former, is to show that there is a common cognitive mechanism underlying all indexical thoughts, in spite of their seeming diversity. On each of the key issues with which I deal in the book, I advance and defend my own position by confronting it with its relevant representative rivals. This is established through the stages that I outline in the last section of this Introduction. That there is a common cognitive mechanism underlying all indexical thoughts in outline means that what makes an indexical thought the same is representing the individual thought about as the same by way of unreflectively taking it for granted that it is the same individual. If I take it for granted that two different parts of the same individual that I synchronically perceive belong to the same individual, I will think about it under the same mode of presentation which serves as a thought-constituent. And the same holds for diachronic thoughts and belief retention in general. If in thinking about an individual that I perceive I take it for granted that it is the same individual from one occasion to the next, I will think about it under the same mode of presentation throughout (even if I do not have a continuous perceptual experience of it). If in thinking about day d I take it for granted from d through to d + 1 that it is the same day I will think about d under the same mode of presentation throughout. If in thinking about a certain location I take it for granted that it is the location that I was at a little while ago, I will think about it under the same mode of presentation throughout, and the like. Conversely, if I entertain doubts as to whether two different parts of the same individual that I synchronically perceive belong to the same individual, I will think about it under two different modes of presentation and hence entertain two different thoughts about it. The same applies to diachronic cases featuring individuals, days, and locations. To give an example, suppose that on day d I first take it that it is January 10, and then, due to my mistaken belief that midnight has passed, a few minutes later, I no longer take d to be January 10, I will respectively think of d under two different modes of presentation. Since modes of presentation and thoughts of which they are constituents serve to capture the subject’s cognitive perspective and thereby account for cognitive significance, it follows that her cognitive perspective is accounted for in the same kind of way with respect to all indexical representations. I hope this helps the reader navigate the book.

Preliminaries To have an indexical belief is to be in an appropriate mental state that bears cognitive significance in the sense described above. To retain an

8 Introduction indexical belief over time is to keep being in such a state or go through a series of such states that are internally aligned. 2 Issues of cognitive significance and of belief retention as studied by cognitive dynamics are thus interrelated. To recall, a theory of cognitive dynamics is concerned with the question of what standard adjustment the subject needs to make in the expression of her indexical thought in order to express the same belief over time. Hence, the explication of what it is to retain an appropriate belief (i.e. attitude) provided in this book calls for an explication and discussion of what it is to be in an appropriate mental state. This, in turn, calls for deciding when two of the subject’s mental states are, or are not, the same, which becomes particularly tricky when the subject’s thoughts stretch through time. I hope this book gives us some clues as to how to settle these issues. An example of the interrelation between thoughts as bearers of cognitive significance and belief retention can be found in Frege’s “retention claim” (henceforth RC): (RC) If someone wants to say today what he expressed yesterday using the word ‘today’, he will replace this word with ‘yesterday’. Although the thought is the same its verbal expression must be different in order that the change of sense which would otherwise be effected by the differing times of utterance may be cancelled out. (Frege 1918/1977, 10)3 Senses are for Frege bearers of cognitive significance and so are the thoughts of which they are constituents, in conformity with the compositionality principle according to which the sense of any complex expression needs to be built up out of the senses as its parts. (RC) states that to think of a certain day as the same from one occasion to the next is to think of it via the same sense. Cognitive significance is thus of a piece with the internal continuity of the subject’s belief as studied by cognitive dynamics and we need to deal with it in accounting for belief retention and vice versa. (RC) also reveals Frege’s departure from the doctrines that he generally holds. In the case of a non-indexical designator, understanding its linguistic or conventional meaning amounts to grasping its sense as the mode of presentation of the item it designates. Understanding the meaning of the expression ‘The evening star’ thus supplies the subject with thinking of Venus via one particular sense, while understanding the meaning of the expression ‘The Morning star’ supplies the subject with thinking of Venus via a different sense, hence via a different cognitive path. But this link between meaning and sense cannot be preserved in the case of indexicals. Frege’s foregoing remark shows that the same sense (and thought) can be expressed and retained by means of deploying indexicals with different meanings, i.e. ‘today’ and ‘yesterday’. This

Introduction  9 example also shows that different senses can be expressed by means of the indexical ‘today’ on differing times of utterance. This is to say that the following two claims that are vital to Frege’s philosophy of language and thought are not true of indexical terms and sentences: 1 2

The regular connection between a sign, its sense, and [its referent] is of such a kind that [A] to the sign there corresponds a definite sense and [B] to that in turn a definite [referent]… (Frege 1892, 58).4 A declarative sentence (having a truth value) is true or false absolutely (e.g. in Frege 1918/1977, 27–28).

(A) and (B) cannot be inter-contextually jointly satisfied by indexical terms because these terms are context-sensitive. As noted, the case of indexicals creates a tension between two Fregean notions of sense that Frege himself also felt. As Kaplan put it, sense was first introduced to represent the cognitive significance of a sign and thus to solve Frege’s problem: how can a = b, if true, differ in cognitive significance from a = a. But, sense is also taken to represent the truth conditions (see Kaplan 1989a, 501, n. 26). If there should be such a thing as a definite sense of an indexical such as ‘today’, supplied by its linguistic meaning that accounts for cognitive significance, it cannot represent the truth conditions of a sentence containing the given indexical since its truth conditions vary with reference. While the linguistic meaning of the indexical ‘today’, i.e. the present day, remains the same in different contexts of utterance, utterances of ‘Today is beautiful’ have different truth conditions on different days. Conversely, the sense that represents the truth conditions of such a sentence is not constituted by the indexical’s linguistic (conventional) meaning since it can vary from occasion to occasion under the same linguistic meaning as when ‘Today is beautiful’ is uttered on different days. Hence, one who wishes to retain the sense-reference framework in the case of indexicals needs to abandon (1) and adopt either (A) or (B). Either way (2) is false. Frege’s suggestion that uttering a sentence such as ‘Today is beautiful’ on different days effects a change of sense conforms with (B) rather than (A). Hence, (2) – the claim that a declarative sentence (having a truth value) is true or false absolutely – does not hold. A sentence such as ‘Today is beautiful’ can change its truth value with the day of utterance and is sometimes true, sometimes false. Yet, a thought that is expressed by such a sentence on a particular occasion, e.g. that d is beautiful, is true or false absolutely. Together with Frege on the evidence of (RC), neo-Fregean philosophers such as Evans (1985), McDowell (1984), and Campbell (1987) hold that indexical senses as bearers of cognitive significance do not line up with the linguistic meanings of indexical expressions; hence indexical senses are not meaning-governed. In accordance with this, Evans claims that

10 Introduction the way of thinking of an object to which the general Fregean conception of sense directs us is a way of keeping track of an object. This permits us to hold that a subject on d2 is thinking of d1 in the same way as on d1, despite lower-level differences, because the thought-episodes on the two days both depend upon the same exercise of a capacity to keep track of time (Evans 1985, 311). Since these senses furnish the thoughts that both account for cognitive significance and represent the truth conditions of the utterances of indexical sentences such as ‘Today is beautiful’, there is no tension between the two Fregean notions of sense. However, there are seemingly other problems with the Fregean and neo-Fregean views. According to these views, thoughts are contents (objects) of belief. Frege and Evans also allow different people to think of the same item via different senses. Frege is focused on proper names. 5 Evans finds this to be the case with indexicals. He suggests that the sense via which the speaker thinks about a perceived object in uttering, say, a demonstrative sentence such as ‘This is F’ will be different from the sense via which the hearer of this utterance thinks about the same object if they identify it via two relevantly different parts (see Evans 1982, 333). This makes thoughts as contents (objects) of belief interpersonally shifty and unstable and, inter alia, hinders us in predicting correctly the conditions under which two people have said the same thing. It also creates an unnecessary proliferation of senses (as we shall see below). Furthermore, it has been claimed that (neo-)Fregean senses fail in fulfilling one of their key roles – that of being bearers of cognitive significance. As Kaplan put it: If one says ‘Today is beautiful’ on Tuesday and ‘Yesterday was beautiful’ on Wednesday, one expresses the same thought according to [RC]. Yet one can clearly lose track of the days and not realize one is expressing the same thought. It seems then that thoughts are not appropriate bearers of cognitive significance. (Kaplan 1989a, 501, n. 26) The claim is that due to losing track of the days, the subject will misidentify the relevant thought over time. This makes thoughts psychologically inert, contrary to what they are meant to be. Frege’s solution to the aforementioned problem of how can a = b, if true, differ in cognitive significance from a = a, relies on a difference in senses and thoughts as bearers of cognitive significance. This link is captured by a criterion of difference for thoughts that is derived from Frege (see, e.g., Evans 1982, 18f; Frege 1892/1980, 62). It states: (CD) Two thoughts are different if it is at the same time possible for a rational subject to take conflicting epistemic attitudes towards

Introduction  11 them. Hence, if the thought is the same, the subject cannot at the same time rationally take conflicting epistemic attitudes towards it. But, as just noted, according to Kaplan, a rational subject can take conflicting epistemic attitudes towards the same thought as expressed by ‘Today is beautiful’ and ‘Yesterday was beautiful’ on two consecutive days, which entails that Fregean thoughts are not epistemically transparent as they are meant to be for both Frege and Kaplan. (In the passage quoted, Kaplan in effect suggests that bearers of cognitive significance need to be transparent). This outcome is avoided if we adopt (A), as Kaplan and Perry do. They take the linguistic (conventional) meaning of an indexical such as ‘today’ and ‘yesterday’ to play the role of sense as a bearer of cognitive significance. It has the feature that a rational subject cannot take two different utterances of ‘Today is beautiful’ to have different meanings, i.e. to supply different, if incomplete, senses, or take ‘today’ and ‘yesterday’ to have the same meaning, which makes the sameness or difference in meanings (senses) epistemically transparent for the subject. More specifically, Kaplan and Perry defuse the aforementioned tension between the two notions of sense by postulating two different kinds of meaning: one accounting for the cognitive significance of a sign and the other representing the truth conditions of a sentence. The former kind of meaning is obtained by transforming the Fregean sense of a non-indexical expression, such as a (non-indexical) definite description (‘The evening star’ in the foregoing example), which determines the individual that it designates by specification in a context-free way, into Kaplan’s character and Perry’s role of an indexical such as ‘today’, to which no definite context-free reference corresponds. Similar to the Fregean sense of a non-indexical expression, which both determines reference by specification (though in a context-free way) and provides a cognitive path that leads to it (which is also a matter of semantics), it is in terms of character that cognitively significant mental states of the subject are accounted for and reference, i.e. content (given the context), as the other kind of meaning, is determined. Being a property of expression types, character is the kind of meaning of an expression, which is set by linguistic conventions, and determines the content of the expression, i.e. what is said, in every context and commonly called its semantic value. ‘Because character is what is set by linguistic conventions, it is natural to think of it as meaning in the sense of what is known by the competent language user’ (Kaplan 1989a, 505).6 On the other hand, content is that which accounts for the truth conditions of an utterance of an indexical sentence. The relationship of character to content Kaplan (1989a, 524) likens to that traditionally regarded as the relationship of sense to denotation; character is a way of presenting content. In Perry’s initial view:

12 Introduction When we understand a word like ‘today’, what we seem to know is a rule taking us from an occasion of utterance to a certain object. ‘Today’ takes us to the very day of utterance, ‘yesterday’ to the day before the day of utterance, ‘I’ to the speaker, and so forth. I shall call this the role of the demonstrative. (Perry 1977/1993, 8; see also Perry 2015) As one can think of a single object via different Fregean senses, one can also think of the same content via different characters (roles). One can think of a single day (as a content constituent) first under the character of ‘today’ (the present day) and then under the character of ‘yesterday’ (the previous day). But unlike the Fregean sense of a non-indexical description, it is possible to think of different objects (days, etc.) under the same character. While in Frege’s theory, a given manner of presentation presents the same object to all mankind, character will in general present different objects of thought to different persons or even different objects of thought to the same person at different times (Kaplan 1989a, 530). Unlike Fregean senses, character and role are not part of the content believed and asserted which, in a true referentialist fashion, Kaplan and Perry take to be Russellian. It is a singular proposition that consists solely of objects and properties or relations. Character and role rather specify the manner in which the content that is believed is believed and, in this sense, present us with the (part of the) object of belief.7 Unlike those senses that line up with (B), characters and roles are interpersonally (and intrapersonally) stable in having fixed meanings. ‘Today’ will mean (something like) the present day for every competent language user as well as for the same language user on different occasions of utterance and similarly for other indexicals. In consisting solely of objects and properties or relations, the singular propositional content that can be thought under different characters or roles is also kept stable. That is: One reason we need singular propositions is to get at what we seek to preserve when we communicate with those who are in different contexts. Fregean thoughts will not do, and neither will mere truth-values. (Perry 1988/1993, 231) This enables us to predict correctly the conditions in which two people have said the same thing which is not so with Fregean senses which, as noted, may vary interpersonally (and intrapersonally over time, as we shall see). Hence, (2) – the claim that a declarative sentence (having a truth value) is true or false absolutely – does not hold. This is in line with the fact that (2) is false when it comes to a context-sensitive language. A sentence such as ‘Today is beautiful’ can change its truth value with the day of utterance and is sometimes true, sometimes false. Yet a content

Introduction  13 that such a sentence expresses on a particular occasion, e.g. that d is beautiful, can be seen as true or false absolutely (see Perry 1979/1993). The characters (roles) of ‘today’ and ‘yesterday’ amount to more than two different manners in which the subject can think of the same day, i.e. of the same Russellian content. They also enable the subject to reexpress the belief that she once formed, as time goes by. Suppose that the subject accepts ‘Today is beautiful’ on Tuesday as a result of believing that Tuesday is beautiful. In order to re-express this belief on Wednesday, the subject needs to replace ‘today (the present day)’ with ‘yesterday (the previous day)’ and accept ‘Yesterday was beautiful’. But, as Kaplan suggests in relation to (RC), for such a belief to be re-expressed and retained, it is not enough that the subject believes the same singular content from Tuesday through to Wednesday. An internal continuity in the subject’s belief is also required. Cognitive significance is thus, once again, of a piece with the internal continuity of the subject’s belief. But, while for Frege and neo-Fregeans retaining a belief such as that Tuesday is beautiful from Tuesday through to Wednesday involves thinking of Tuesday via the same sense, for Kaplan and Perry, this involves thinking about Tuesday first as the present day and then as the previous day. In contrast with this, Lewis’s view of cognitive significance concerning these matters is incapable of accounting for belief retention. Lewis argues that Kaplan’s notion of same-saying and a fortiori the notion of same-believing is insignificant (Lewis 1980). He claims that beliefs figure in common sense psychology. They serve to explain and predict behaviour. So, Lewis individuates beliefs by cognitive significance. But, according to him, the subject who keeps track of Tuesday through to Wednesday and on Tuesday accepts ‘Today is Tuesday’ and on Wednesday accepts ‘Yesterday was Tuesday’ does not hold the same belief throughout. Lewis claims that to have an appropriate indexical belief is to self-ascribe the corresponding property (Lewis 1979, 518). To have the former belief is to self-ascribe the property of being temporally located on Tuesday, while to have the latter belief is to self-ascribe the property of being temporarily located on Wednesday. But, if so, we are committed to holding rather implausibly that the subject cannot continue to believe on Wednesday what she believed on Tuesday in assenting to ‘Today is beautiful’ (see Bradley 2013). This conclusion will be supported by my arguments in Chapter 4. On a different note, Lewis argues that indexical beliefs are beliefs of a special kind in that they are not equivalent to any beliefs expressed without indexicals. The subject can have all the non-indexical information without having the indexical information. He illustrates this by a well-known scenario about two gods who inhabit a certain possible world and know exactly which world it is. They know every proposition that is true at their world. Insofar as knowledge is a propositional attitude (in that when an agent believes something, she is standing in a certain relation to a proposition – that of believing it), they are

14 Introduction omniscient. Still, we can imagine them to suffer ignorance: neither one knows which of the two he is. They are not exactly alike. One lives on the top of the tallest mountain, and the other lives on top of the coldest mountain. Neither one knows whether he lives on the tallest mountain or on the coldest mountain. They know all the information about their world, except that information that they would express by ‘I am the god on the tallest mountain’ or ‘I am the god on the coldest mountain’. For Lewis, this shows that some belief and some knowledge cannot be understood as propositional, but can rather be understood as self-ascription of properties (Lewis 1979, 520–521). This shows that the information that the two gods would express by using the first-person singular pronoun is not equivalent to any information that is expressed without indexicals.8 In claiming this, Lewis is influenced by Perry (1977/1993) who provides a number of similar scenarios, not necessarily involving the first person. These scenarios are designed to show that relevant indexical beliefs are not equivalent to any beliefs expressed without indexicals, i.e. that the subject can have all the non-indexical information without having the indexical information, and that beliefs that are naturally expressed with appropriate indexicals can have different behavioural consequences from truth-conditionally equivalent beliefs expressed without indexicals. Here is one of Perry’s scenarios, featuring the indexical ‘now’: [A] professor who desires to attend the department meeting on time and believes correctly that it begins at noon, sits motionless in his office at that time. Suddenly he begins to move. What explains his action? A change of belief. He believed all along that the department meeting starts at noon; he came to believe, as he would have put it, that it starts now. (Perry 1979/1993, 34) A similar point can be made about other indexicals, with a view to showing that indexical beliefs about oneself, locations and the like are beliefs of a special kind in that they are relevantly different from analogous non-indexical beliefs. In the case just quoted, the professor who came to believe, as it were, that the meeting starts now in explaining his action would use the indexical ‘now’ to characterize the belief he came to have. In Perry’s view, this is an essential indexical in that replacing it by other terms destroys the force of the explanation or at least requires certain assumptions to be made to preserve it (Perry 1979/1993, 35). Perhaps less obviously, this also applies to demonstrative pronouns. Here is a handy example from Perry who brings it up in relation to locating beliefs: [The subject S] stands in the wilderness beside Gilmore Lake, looking at the Mt. Tallac trail as it leaves the lake and climbs the mountain.

Introduction  15 He desires to leave the wilderness. He believes that the best way out from Gilmore Lake is to follow the Mt. Tallac trail up to the mountain to Cathedral Peaks trail, on to the Floating Island trail, emerging at Spring Creek Track Road. But he does not move. He is lost. He is not sure whether he is standing beside Gilmore Lake, looking at Mt. Tallac, or beside Clyde Lake looking at Jack’s Peak, or beside Eagle Lake looking at one of the Maggie peaks. Then he begins to move along the Mt. Tallac trail. If asked, he would have explained the crucial change in his beliefs this way: “I came to believe that this is the Mt. Tallac trail and that is Gilmore Lake”. (Perry 1979/1993, 34–35; see also Perry 2013 for a related discussion) However, Cappelen and Dever (2013) provide us with a number of examples that are designed to show that perspectival representations – those that represent the items they are about from a thinker’s point of view – are not representationally essential, i.e. that there are no essential indexicals in Perry’s sense. They think that the examples that Lewis and Perry provide are just instances of Frege’s co-reference problem – how can a = b, if true, differ in cognitive significance from a = a – and give us no reason to think that indexicality is distinctive in this respect. In relation to the last example, they support this claim by assuming that Byrde Lake is identical to Gilmore Lake, but the hiker does not know that. They say: Our hiker is lost. He knows he’s standing next to Byrde Lake, but doesn’t know whether Byrde Lake is Gilmore Lake. As a result, he isn’t sure whether he is standing beside Gilmore Lake looking at Mt. Tallac, or beside Clyde Lake looking at Jack’s peak, or beside Eagle Lake looking at one of the Maggie peaks. Then he begins to move along the Mt. Tallac trail. If asked, he would have explained the crucial change in his belief this way: “I came to believe that Gilmore Lake is Byrde Lake”. (Cappelen and Dever 2013, 62) This is supposed to show that in the initial Gilmore Lake example, the utterances of ‘this’ and ‘that’ do not play the special role that Perry assigns to them, i.e. that they are not essential. And the same holds for other indexicals.9 However, this does not entail that indexical beliefs can be assimilated to other kinds of belief. This is evidenced by the fact that a belief (or thought-content) that one expresses by ‘Today is beautiful’ on d will not be the same as the belief (or thought-content) that one would express by the same form of words on d + 1 which is not the case with the sentence ‘The evening star is a body illuminated by the Sun’ or with any sentence that is complete in every respect, to use Frege’s

16 Introduction turn of phrase. The sense of the expression ‘The evening star’ represents the same individual regardless of the context in which this sentence is uttered and regardless of one’s point of view. It represents Venus from a meaning-based point of view rather than from a thinker’s point of view which is something like that [observed] heavenly body. (In Chapter 7, we shall see that indexical beliefs are special in yet another respect.) Leaving aside the issue as to whether there are essential indexicals, I want to note that the examples that Lewis and Perry provide are indeed instances of Frege’s co-reference problem. As a consequence, these examples conform to (CD) which states that two thoughts (belief states) are different if it is at the same time possible for a rational subject to take conflicting epistemic attitudes towards them. The professor from Perry’s example could have easily and without irrationality continued to believe that the department meeting starts at noon without believing, as Perry puts it, that it starts now, and missed it. Hence, his ‘now’-belief is different from his ‘noon’-belief. (Kaplan also relies on (CD) in talking about bearers of cognitive significance.) Acknowledging that indexical beliefs cannot be assimilated to other kinds of belief, we have also learned that a satisfactory theory of indexical beliefs needs to be capable of accounting for belief retention. Belief retention, in turn, requires an internal continuity in the subject’s belief which is of a piece with cognitive significance which itself conforms to (CD) which explains how can a = b if true differ in cognitive significance from a = a. In accordance with this, I will argue that accounting for cognitive significance in terms of those devices that conform with (A), such as those proposed by Kaplan and Perry does not work. My first task is to show that none of these devices can serve to account for the differences in cognitive significance that arise in relation to Frege’s co-reference problem – how can a = b, if true, differ in cognitive significance from a = a. I start off with the case in which two different utterances of the same demonstrative expression type refer to the same object yet the subject disbelieves that ‘That1 = that 2’, as when for example, she views the stern and the bow of a single ship in the harbour whose middle is obscured by a large building and take it that That1 [pointing to the ship’s stern] is not that 2 [pointing to the ship’s bow] (see Perry 1977/1993, 12–13). Then, I show that similar problems beset cases that involve temporal indexicals ‘today’ and yesterday’. The positive view that I offer conforms with (B) rather than (A). In this respect, this view is similar to the kind of view adopted by Frege, on the evidence of (RC), and neo-Fregeans. However, unlike these views, the view that I advance is not committed to the noted interpersonal instability of content and a proliferation of senses. It also avoids an intrapersonal proliferation of senses that Evans and Campbell do not seek to avoid. In a true Fregean spirit, neo-Fregeans hold that a theory of content should provide an explanation of the

Introduction  17 subject’s cognitive perspective. But, in a true anti-individualist spirit, they also hold that perception-based demonstrative thought-contents depend for being the contents they are on relations to objects beyond the body of the subject. In agreement with the former requirement, I argue that their anti-individualist view of content faces problems when it comes to explaining the subject’s cognitive perspective. I also discuss related non-Fregean anti-individualist views (those that reject Fregean sense) and dismiss them on similar grounds. I propose an alternative view which meets this requirement. In so doing, I also depart from some of Frege’s own views. We shall see, though, that Fregeans are right in holding that the subject’s cognitive perspective is captured by sense. If the subject harbours doubts as to whether an object (or two different objects) that she perceives synchronically or diachronically is the same she will think about it (or each them, respectively) via two different senses. If the subject represents an object as the same synchronically or diachronically (without initially taking it for two different objects), she thinks of it via the same sense.10 To claim that she is instead to think of it via different senses leads, we shall see, to an unnecessary and implausible proliferation of senses in the kind of case in which it seems intuitively plausible that only one sense is in play. If the subject is also unaware that she is thinking (synchronically or diachronically) of the same object via different (let alone many different senses), then senses and thoughts play no role in capturing the subject’s cognitive perspective and thoughts fail to be the bearers of cognitive significance, contrary to what they are supposed to be. A similar point is made about thinking of days and locations. One of the rewarding outcomes of the position that I defend in this book is that it provides compelling evidence in favour of the view that there is a common cognitive mechanism underlying all indexical thoughts, in spite of their diversity on the surface level. This is to say that our ‘today/ yesterday’-thoughts, ‘this/that’-thoughts (either as perception-based or memory-based), ‘here/there’-thoughts, or ‘I’-thoughts turn out to be more alike than they appear to be.

An Outline of the Book As noted, one goal that this book seeks to achieve is to establish the nature of the link between cognitive significance and cognitive dynamics (i.e. belief retention); another, related to the former, is to show that there is a common cognitive mechanism underlying all indexical thoughts, in spite of their seeming diversity. It is part of this picture that a thinker’s cognitive perspective is captured by the content that is believed and asserted which thereby serves as the bearer of cognitive significance. This is established through the following stages, each corresponding to a

18 Introduction chapter in this book, by way of showing that the relevant rival attempts to account for cognitive significance fail to achieve this on their own terms. Chapter 1 traces Kaplan’s successive attempts to account for cognitive differences that arise in relation to the co-reference problem: how can a = b, if true, differ in cognitive significance from a = a. The focus is on perception-based demonstratives. Contrary to what they are supposed to be, characters as the kind of meaning of expressions that are set by linguistic conventions and other related devices that Kaplan deploys in trying to solve this problem do not capture the subject’s cognitive perspective. I then also show that characters of temporal indexicals such as ‘today’ are also unfit for this role. Chapter 2 deals with Perry’s more recent and elaborate attempt to solve the co-reference problem in relation to perception-based demonstratives that builds on his own earlier view that was discussed in the Introduction which resembles Kaplan’s original view dealt with in Chapter 1. The discussion in this chapter shows that, Perry’s intentions notwithstanding, this attempt rests on a misrepresentation of the subject’s cognitive perspective and, a fortiori, that the kind of content that he appeals to fails to capture her cognitive perspective. The same is then shown to hold for a related kind of content that Perry and Lawlor deploy in order to capture the subject’s cognitive perspective in those cases in which she mistakes two different objects for a single one. The discussion in Chapter 3 shows that representative anti-individualist attempts fail to account for the subject’s cognitive perspective, again in relation to perception-based demonstratives. I argue that Brown’s view fails in that it rather implausibly disregards this perspective altogether. Stalnaker fails to square his anti-individualist thought-contents with the subject’s cognitive perspective, despite his claims to the contrary. Neo-Fregeans, on the whole, manage to explain the subject’s cognitive perspective in terms of thought-contents but curiously and implausibly do not subscribe to this view when it comes to some special cases. Then, I propose an alternative theory of content that explains the subject’s cognitive perspective. While the first three chapters focus on those thought-contents that we express by means of demonstratives such as ‘this’ and ‘that’, Chapter 4 deals with those thought-contents that we express by means of temporal indexicals such as ‘today’ and ‘yesterday’ as studied by cognitive dynamics. In considering the issue of belief retention concerning days and in focusing on the Rip Van Winkle problem pointed out by Kaplan, I come up with the view that keeping track of days in accordance with the subject’s cognitive perspective, or her point of view, is captured by the content of her diachronic thought about days. This way, it is established that cognitive significance is of a piece with the internal continuity of the subject’s belief.

Introduction  19 In Chapter 5, I further advance the view advocated in Chapter 4. I argue that the subject’s cognitive perspective fails to be captured by socalled doxastic characters that Perry invokes for this purpose. The focus is again on thoughts that we express by means of temporal indexicals such as ‘today’ and ‘yesterday’ as studied by cognitive dynamics. Following this, I propose an alternative account of belief retention. According to it, our ways of thinking of days are not governed by the linguistic meanings or characters of the indexical expressions that we associate with them. This helps me meet my objective, showing that the subject’s cognitive perspective is captured by the content that is believed and asserted as I take it to be, which thereby serves as the bearer of cognitive significance. Chapter 6 deals with belief retention regarding perception-based demonstrative thoughts as studied by cognitive dynamics. In this context, I propose and defend a version of the criterion of difference for thoughts that is derived from Frege which, unlike the standard version of this criterion, enables thoughts (thought-contents) to capture the subject’s cognitive perspective. This lends support to my view that cognitive significance is of a piece with the internal continuity of the subject’s belief by way of showing how thought-content captures the subject’s cognitive perspective. In Chapter 7, I further develop the proposed account of content individuation and belief retention in relation to keeping track of both objects and days. The position concerning the re-identification of objects that I develop here lends further support to the version of the criterion of difference for thoughts proposed and dealt with in Chapter 6 by confronting it again with the standard version of this criterion. In so doing, I also address the issue of whether perception-based demonstrative thoughts are thoughts of a special kind in relation to thoughts that are not perception-based in the context of an ongoing debate about these matters and conclude that demonstrative thoughts are in a certain way indeed special. Chapter 8 looks at speech reports. In keeping with Frege’s demand that semantic content needs to be objective and interpersonally stable and shareable, and hence, that the thought expressed by subject S needs to amount to the thought that is being referred to in a correct speech report, I suggest how this demand can be met in the face of the fact that S’s ‘I’-thought cannot be interpersonally shareable. In Chapter 9, I strengthen my view of thought-content individuation by arguing that it is not committed to the problems that some accounts of singular beliefs have been charged with. I show that indexical beliefs as I take them to be are singular beliefs no matter whether they are subject to acquaintance constraints or not. This enables us to bypass a timeworn debate about the relation between singular and descriptive beliefs as well as make it manifest that in a relevant respect indexical beliefs are

20 Introduction distinct from other beliefs. It is also forthcoming that there is a common cognitive mechanism underlying all indexical thoughts, in spite of their diversity on the surface level. Following this, I argue that in having a persisting belief about an individual, the subject is not thinking about it via a mental file, eschewing thus the charge that has been levelled at mental file accounts.

1

Character, Content, and Cognitive Significance

1.1 Introduction We have seen that Kaplan tries to account for the cognitive differences between the true a = b and a = a in terms of the difference in the characters when ‘a’ and ‘b’ are indexical expressions of different meaning types, while Perry speaks of their different roles. But Kaplan has noticed that in the case of perception-based demonstratives such as ‘this’ and ‘that’, characters are unfit for this role. Having noticed this, he makes a couple of attempts to account for the cognitive differences between an informative and uninformative use of ‘That = that’, where the two occurrences of the demonstrative ‘that’ refer to the same object, in terms of various accompanying features. In this chapter, I argue that none of these attempts are successful. I also point out that in the case of indexicals such as ‘today’, character is unfit to account for relevant cognitive differences. This should help pave the way for the subsequent discussion that is aimed at discerning the bearers of cognitive significance that capture the subject’s cognitive perspective.

1.2 Character and Cognitive Significance As the bearer of cognitive significance, each Kaplanian character or Perry’s role is supposed to account for the common nature that different belief states might have. Grasping the character or role of, for example, the indexical ‘today’ by different people or by the same person at different times encapsulates the same way of thinking about the day(s) being referred to. While contents are represented by functions from possible circumstances to extensions (Carnap’s intensions), characters are represented by functions from possible contexts to contents. Indexicals have contextsensitive (but stable) character: their content varies with context. The notion of cognitive value is then accounted for in terms of the notion of character. When on different days I say, ‘Today is beautiful’, I am said to be in the same cognitive state while believing different things: the two utterances have the same cognitive value or character but different contents (see Kaplan 1989a, XVII).

22 Character, Content, Cognitive Significance Like for Frege, cognitive value is of a piece with the semantic determinant of reference. For Frege, a (non-indexical) definite description designates an object in virtue of the descriptive condition that it encapsulates. In understanding the meaning of ‘The evening star’, one grasps its sense that contains a mode of presentation of Venus whereby it designates Venus via a cognitive path that leads to it. Similarly, an indexical such as ‘today’ designates its referent in virtue of the descriptive condition it encapsulates which amounts to its character the present day. In both these cases, the encapsulated descriptive condition is the bearer of cognitive value in that it presents the object in such-and-such a way. But while in the case of such a definite description the object is supposed to be picked out in a context-free way, the referent of an indexical is determined in virtue of its character given the context. Now, Frege’s sense first introduced to represent the cognitive significance of a sign is also taken by him to represent truth conditions (Kaplan 1989a, 501, n. 26). Since there is a pull between these two notions that even Frege felt, Kaplan, as noted, suggests that we should tease out of Frege’s notion of sense two different kinds of meaning: the character or linguistic meaning of an expression and the content or the proposition expressed which is (in the indexical case) Russellian in that it consists solely of objects, properties, or relations. And it is the character of an indexical expression, i.e. a rule that takes us from context to content, that accounts for cognitive as well as semantic significance (given the context). However, the character is not part of the content which represents the truth conditions of an utterance of an indexical sentence. Perry’s original view is very similar. In addition to the relational mode of presentation (or identifying condition for reference, in his later writings), he also speaks of the relative mode of presentation which in the case of the indexical ‘tomorrow’ he glosses by stating that ‘tomorrow’ denotes a day only given a day, i.e. relatively not absolutely. The relative mode of presentation corresponds to what Perry calls the ‘role’ of such a term, i.e. a rule taking us from an occasion of utterance to a certain object (day), that is akin to Kaplan’s character. It is a rule for determining reference by its relation to an element of the context of utterance. What language associates with the indexical word is such a relative mode of presentation (Perry, Postscript to 1977/1993, 27–28). In this way, what is believed (in the context) is determined by how it is believed. Thereby, in the relevant cases, the Fregean requirement that sense determines reference is met by splitting up the sense, i.e. thought, of the classical Fregean doctrine, which takes it to be the object or content of belief, into two components. Reference becomes the object or content of belief (i.e. Thought in Kaplan’s (1989a, 530) and Perry’s (1977/1993) terminology) and is determined (in the context) as well as presented in a particular way by Sense (Kaplan’s character, Perry’s role). As noted, Perry adopts (A), i.e. that to the sign there corresponds a definite (though incomplete) sense, and abandons (B), the claim that

Character, Content, Cognitive Significance 23 to the sense belongs a definite referent. So does Kaplan who, to recall, claims that while in Frege’s theory, a given manner of presentation presents the same object to all mankind, character will in general present different objects of thought to different persons or even different objects of thought to the same person at different times (Kaplan 1989a, 530). What is left of the Fregean sense does not fully determine reference by specification as does the sense of a (non-indexical) definite description. Sense amounts to the descriptive content of an indexical expression that is encapsulated by its character. It plays a part in the determination of reference, i.e. of the proposition expressed, but does not amount to a full-fledged descriptive content. This is what enables Perry to urge that references of singular terms do not depend on Fregean senses or identifying descriptions in the mind of the speaker and that expressions used do not have such senses attached to them by the conventions of language.1 ‘The beliefs of the speaker need not supply conditions that single out a unique individual. Even if the speaker has such beliefs, the reference is not determined by those beliefs’ (Perry 1988/1993, 227).

1.3 A Problem with Characters: Demonstratives While urging that characters are the bearers of cognitive significance, Kaplan has noticed that they are too coarse grained to capture the differences in cognitive significance (i.e. psychological states) for all indexicals (and so has Perry in his 1977/1993, 13). He wonders how the differences in cognitive significance are to be explained in the case in which different utterances of the same demonstrative expression (type) refer to the same object, but one does not believe that, as it were, ‘That1 = that2’. 2 The issue concerns perception-based (deictical) demonstratives which are in Kaplan’s view unlike pure indexicals such as ‘today’ and ‘yesterday’ or ‘here’ and ‘now’ (at least in some of their uses) in that they are in need of supplementation which both accomplishes the referring job as well as accounts for the cognitive value.3 Kaplan’s explanation of the informative case of ‘That1 = that 2’ fails on the same grounds as does tying cognitive significance to character itself. Although the two demonstrative utterances have the same character, Kaplan’s original claim (Kaplan 1989a) was that there is a difference in the speaker’s demonstrations supplementing, respectively, the two utterances of ‘that’, where demonstrations are understood as types which can be defined as narrow psychological states. Their form is “The individual that has appearance A from here now”, where an appearance is something like a picture with a little arrow pointing to the relevant subject (Kaplan 1989a, 526), and it does not seem to be essential to a demonstration that it be mounted by any agent at all (Kaplan 1989a, 525). The notion of demonstration is a theoretical concept, meaning that there are no cases in which an utterance of a

24 Character, Content, Cognitive Significance demonstrative is not accompanied by a demonstration since “a demonstration may also be opportune and require no special action on the speaker’s part, as when someone shouts ‘Stop that man’ while only one man is rushing toward the door” (Kaplan 1989a, 490). However, our problem is not solved by this move since there are still cases in which the identity of the linguistic character and that of the demonstration are kept fixed, while the cognitive value varies. An illusionist may come up with a trick creating the impression that an object sitting in one spot has been replaced with another one qualitatively identical with it, whereas this is not so. In the process, he may say, ‘That1 is that 2’, where both utterances of ‘that’ are consecutively taken to refer to the same object, while relying on demonstrations that are of the same type. Similarly, under the same circumstances, one object may unnoticeably be replaced with a similar one in a flash such that it will be informative to the audience to be told ‘That1 is not that 2’ and so on.4 Later on, Kaplan (1989b) takes the demonstration associated with an utterance of a demonstrative as a mere externalization of the speaker’s directing or perceptual intention – the intention aimed at a perceived object which may or may not be the object that the speaker has in mind (Kaplan 1989b, 583; see also Perry 2012, 70f, for the role that this kind of intention plays, as well as Perry 2009, for a discussion of directing intentions in relation to Kaplan; see also Korta and Perry 2011, 41f). It is this intention rather than the demonstration that Kaplan now takes to be the semantically significant supplement to an utterance of a demonstrative. And, in keeping with the view that the semantic path to reference is supplied by the subject’s way of thinking of the reference, this intention is meant to also account for cognitive value. Should directing intentions be taken as types defined as narrow psychological states, it follows that a directing intention may stay fixed while the cognitive value varies and the foregoing argument concerning the speaker’s demonstration can be re-adjusted to show that the directing intention cannot account for cognitive significance. But Kaplan is somewhat reluctant to see directing intentions as types. He says that ‘[t]he same demonstrative can be repeated, with a distinct directing intention for each repetition of the demonstrative’ (Kaplan 1989b, 586). This is to say that directing intentions are not separable from particular contexts in which they arise. In view of this, Kaplan admits that the ‘cognitive uncertainties of “that1 is that2” may no longer be an aspect of meaning’ (Kaplan 1989b, 588; see also Lalor 1997).5 While agreeing with Kaplan that cognitive significance is not an aspect of meaning as a property of linguistic types, I disagree with his claim that the speaker’s directing intentions account for the cognitive uncertainties of ‘that1 is that 2’. In disbelieving that ‘that1 is that 2’, the subject will be in two different cognitive (psychological) states, as evidenced by the fact that she may at the same time rationally assent to

Character, Content, Cognitive Significance 25 ‘That1 is F’ and dissent from ‘That 2 is F’, although they both designate the same object. Since each of these cognitive states is coupled with a distinct directing intention, there is no mismatch between these states and the directing intentions as there is between these states and the linguistic meaning or character of the demonstrative ‘that’ in the foregoing utterances which is the same in both cases (as well as between the cognitive states and demonstrations in the sense described). But, in spite of the fact that each of these utterances is accompanied by a distinct directing intention, these intentions do not account for the fact that the subject is in two different cognitive states. As a result of her false belief that there are two different objects in play, and disbelieving that ‘that1 is that 2’, the subject may aim these intentions at what she falsely believes to be two different objects. But, being just the intentions that are aimed at the perceived object, these intentions do not account for the difference in the subject’s cognitive states. A difference in her cognitive states is rather the result of her false belief that there are two different objects in play.6

1.4 A Problem with Characters: Pure Indexicals Unlike demonstratives, the reference of pure indexicals such as ‘today’ and ‘yesterday’ is fixed without the aid of supplementary features such as demonstrations or directing intentions. The context of utterance takes care of this. If I utter ‘Today is beautiful’ on d, I refer to d. As a function from context to content, the character of ‘today’ ensures this. And unlike demonstratives such as ‘this’ and ‘that’, pure indexicals such as ‘today’ and ‘yesterday’ have fixed characters; they are the present day and the previous day. So, when I assent to ‘Today is a beautiful day’ on d, I am thinking of d as the present day and when I assent to ‘Yesterday was a beautiful day’ on d + 1, I am thinking of d as the previous day and my ways of thinking of d are adjusted with the change of context. But I may get the days wrong in the sense pointed out by Perry (1980/1993, 80). Suppose that Smith, whose watch is an hour fast, accepts ‘Today is my husband’s birthday’. But just before 11, she realizes she got it wrong. It is March 1 and not March 2. She glances at her watch, at 11, and it shows midnight – she thinks to herself ‘so today is my husband’s birthday’. Smith’s respective assenting to and dissenting from the two consecutive utterances of ‘Today is my husband’s birthday’ show her as thinking of d under two different modes of presentation, although the character – the present day – is in both cases the same. Hence, once again, characters are too coarse grained to capture the differences in cognitive significance. Once again, cognitive significance is not an aspect of meaning. And the same holds for Perry’s roles. In the former case, the subject’s refusal to accept the true ‘That1 = that2’ is a result of her taking the same object for two different ones. Similarly, the difference in the

26

Character, Content, Cognitive Significance

cognitive states that Smith is in is a result of her taking the same day for two different ones. In Chapter 4, I shall return to the questions concerning cognitive significance in relation to the pure indexicals such as ‘today’ and ‘yesterday’. For now, I want to examine yet another attempt to account for the cognitive uncertainties of ‘that1 is that1’ in terms of the “Kaplan–Perry” framework.

2

Other Kinds of Content and Cognitive Significance

2.1 Introduction The inability of Kaplan’s and Perry’s foregoing attempts to capture the differences in cognitive significance that arise in relation to the co-reference problem has led Wettstein (1986, 2004), who is also a referentialist, to declare that it is thankfully not part of the business of semantics to deal with cognitive significance. The business of semantics is to get the truth conditions right and tell us what propositions various sentences express. But Perry has rightly argued that it is part of the business of semantics to explain cognitive significance (although it is not an aspect of meaning as a property of linguistic types). A correct semantic theory needs to provide us with an appropriate interface between what sentences mean (express) and how we use them to communicate beliefs in order to motivate and explain action (Perry 1988/1993, 2012, 9). In view of this, Perry has made a new attempt to account for the co-reference problem by postulating new levels of content in addition to Russellian content. He accounts for the cognitive uncertainties of ‘That1 is that2’ by means of one of these contents. Yet another such content he deploys in order to account for the cognitive perspective of a subject who sincerely employs an empty proper name. In terms of this content, Lawlor (2007) has tried to explain the cognitive perspective of the subject in the reverse case to the one that has puzzled Frege, Kaplan, and Perry, in which the subject takes a false ‘That1 is that2’ as true as a result of not realizing that the two occurrences of the demonstrative ‘that’ in this identity statement refer to two different objects. I argue that neither of these attempts is plausible, making it clear that the subject’s cognitive perspective needs to be explained in terms of her first-order perception-based beliefs about the world and not in terms of their spin-offs in the shape of additional contents that Perry and Lawlor take to capture the subject’s point of view.

2.2 Official and Reflexive Content Perry believes that a semantic theory for natural languages ought to be concerned with the issue of cognitive significance – of how true identity

28  Other Kinds of Content statements containing different (utterances of) indexicals and proper names can be informative, held to be unaccountable by the referentialist view. The informativeness that he has in mind – one that has puzzled Frege, Kaplan, and Wettstein – concerns knowledge about the world. In an attempt at solving this puzzle on referentialist terms, he comes up with the notion of cognitive significance as a special kind of a secondorder content that should account for cognitive significance in the former sense (Perry 2012). Focusing on his treatment of perception-based demonstratives, I will argue that he fails to do so both on the level of second-order contents containing demonstrative utterances and on the level of second-order contents containing perceptual buffers as new notions associated with the perceptions and used to temporarily store the ideas we gain from perceptions, which he holds to be causally connected to each other. Perry speaks of cognitive significance in terms of cognitive states, in particular the (first-order) beliefs that are about the same individual that might motivate the speaker to make one statement about it but not another. This happens when pointing to a dog’s head sticking out from one side of a pillar she accepts an utterance of ‘This dog is hungry’ but not another utterance of the same sentence that is associated with her pointing to a dog’s tail sticking out from the other side of the pillar. A similar point applies to the listener whose beliefs ‘would show understanding of the one [statement] but not of the other’ (Perry 2001a, 6). In an interesting twist, Perry then introduces the concept of cognitive significance as a special kind of content created by an utterance. I will argue that in its own right, such a content is incapable of explaining differences in behaviour-predicting psychological states concerning informative identities, i.e. that cognitive significance in Perry’s sense has no bearing on cognitive significance in the former sense. This makes Perry’s semantics unexplanatory of the subject’s relevant action and behaviour, in spite of his insistence to the contrary. Perry distinguishes what is asserted or believed, i.e. the referential or official (Russellian) content (consisting solely of objects and properties or relations), from other kinds of content of our statements (i.e. utterances of declarative sentences) and beliefs. One of these contents, which he calls reflexive, accounts for cognitive significance. This content is a proposition associated with an utterance of an indexical sentence in virtue of the meaning of the type of the sentence it is an utterance of. It, in turn, supplies an identifying condition for reference, the condition an object must satisfy in order to be the referent of an indexical expression, derived from its meaning, i.e. from the rules of language. The meaning of an indexical amounts to a rule associating it with such an utterance-reflexive identifying condition which is itself the reflexive content of an utterance of the indexical (Perry 2001a, 77f, 2012, 89f).1 The reflexive content of an utterance of ‘I’ is the speaker of this utterance of

Other Kinds of Content   29 ‘I’, while the reflexive content of an utterance of ‘you’ is the addressee of this utterance of ‘you’. Similarly, the reflexive content of an utterance of ‘that dog’ is the dog that the speaker of this utterance of ‘that dog’ is attending to. The subject who believes a certain referential content expressed by an utterance of an indexical sentence will, on this view, also grasp (though not believe as the subject matter of the given utterance) its corresponding reflexive content supplying the identifying condition for reference as well as the reflexive truth conditions on the utterance of the sentence. The reflexive content of an utterance is the proposition that the utterance meets the conditions that the Kaplanian character of the sentence used establishes for its truth (Perry, Postscript to 1988/1993, 247). That is, the reflexive truth conditions of an utterance derive directly from the (linguistic) meaning of the sentence that it is an utterance of, whereas which proposition is expressed depends also on the agent, time, and the circumstances of utterance (Perry 1988/1993, 236). They differ from what Perry calls the traditional truth conditions which are incremental in that they are conditions on the subject matter: they are what else has to be true, given the linguistic and contextual facts about the utterance (Perry 2001a, 81, 2012, 94). Perry is careful to distinguish between reflexive contents of statements (i.e. utterances of declarative sentences) and those of beliefs expressed by them, but insists that there is a close link between the two. Hence, ‘[i]f [a] statement expresses a belief, the [reflexive] content of the statement, given its architectural connections to the belief, will be the same as the reflexive content of the belief’ (Perry 2001a, 96, 2012, 109). This he calls the connected reflexive content of a statement. He finds that statements containing utterances of perception-based demonstratives such as ‘this’ and ‘that’ have this kind of content in virtue of the underlying perception-based beliefs that these statements are causally linked to. The material that is predicatively associated with the uses of demonstratives depends on the ideas that are associated with the corresponding perceptual buffers. These are new notions associated with the perceptions and used to temporarily store ideas we gain from the perceptions until we can identify the object we are looking at (Perry 2001b, 121). They are tied to ways of perceiving and ways of acting (Perry 2001b, 158). The reason why Perry takes reflexive contents as explanatory of the subject’s relevant action and behaviour lies in his following Frege (concerning the non-indexical case) in tying the notion of cognitive significance to that of meaning in virtue of which it plays a dual role of both determining reference (in the context) and accounting for the cognitive state of the subject. The meaning-based reflexive content of an utterance of ‘that dog’ – the dog that the speaker of this utterance of ‘that dog’ is attending to – both, as it were, denotes the object attended to and accounts for the cognitively significant psychological state of the subject accepting the utterance. This link is, however, gratuitous. In virtue of identifying

30  Other Kinds of Content conditions that the conventions of language associate with them, indexical terms do indeed denote (in the context), but these conditions do not account for the (relevant) cognitive state that the subject having an indexical belief (i.e. attitude) is in, as I will now turn to show. 2

2.3 Utterances, Beliefs, and Cognitive Significance We have seen that meaning-based Kaplanian characters and other supplementary cues cannot account for the difference in cognitive significance arising in relation to the co-reference problem concerning the cognitive uncertainties of ‘That1 = that2’. But, Perry thinks that once we tie cognitive significance to an utterance of a demonstrative (and redefine the concept in the process) rather than to its linguistic meaning-type (or a demonstration, etc.) in the way he suggests, this problem can be readily accounted for. Perry does acknowledge that the key role is played here by the difference in the subject’s perceptual buffers but thinks that it can be assigned to appropriate reflexive contents. I will show first that the reflexive contents of utterances are unfit for this role and then that the same holds for the reflexive contents of corresponding beliefs. From this, it follows that Perry’s semantic apparatus cannot account for the subject’s relevant action and behaviour. To see this, consider one of Perry’s own examples. It concerns a picture in which a dog’s head emerges from one side of a pillar and a dog’s behind from the other side. Suppose Perry points first at its head and then at its tail and asks his audience ‘Do you think that dog is that dog?’ He happens to be referring to the same dog, Stretch, twice and to be using the same demonstrative phrase twice. Yet, he reasonably expects these two acts of reference to have different cognitive effects on the listener who is in a position to see his utterance. In Perry’s own words: That is, I plan on him hearing two tokens of ‘that dog’ seeing two different utterances, having his attention directed towards two different parts of the scene or picture, having two different perceptions, and establishing two different buffers, at least temporarily, to keep track of what am I going to tell him and what he is going to observe about however many dogs there turn out to be. (Perry 2001a, 63–64, 2012, 74) This is in line with Perry’s foregoing claim that the speaker’s beliefs that are about the same individual might motivate him to make one statement about it but not another. They also might motivate him to, for example, try to convince the listener that, as it were, ‘This dog is (not) this dog’, and be the ultimate source of the desired cognitive effect on the listener. They are (first-order) perception-based beliefs that are furnished, respectively, by different perceptual buffers which are in Perry’s view causally

Other Kinds of Content   31 connected with the speaker’s utterances and are also responsible for the fact that different cognitive effects might be exerted on the listener. For: This is a quite different effect than saying something like “Do you think that dog is not self-identical”, with only one demonstration, or “Do you think that dog is not the same as that one” with two exactly similar demonstrations to the head. (Perry 2001a, 64, 2012, 74) True, the listener’s grasp of the given reflexive content (containing two different utterance-reflexive identifying conditions) will, in the former, unlike in the latter case, direct his attention towards the two parts of the same dog, the perception of which will enable him to judge whether they do or do not belong to the same dog. But this does not show that this content should account for the cognitive differences arising from his perceptions of the given dog via its two parts. All it shows is that the role of the reflexive content of an utterance is analogous to what Perry thinks is the role of the speaker’s demonstration or of his directing intention which Perry describes as the intention to refer to an object X simply in virtue of the meanings of one’s words and the context, both pre-existing and supplied by the speaker (Perry 2001a, 60, 2012, 70). In Perry’s view, the role of the speaker’s demonstration and of his directing intention is just to create a cognitive path for the listener (Perry 2001a, 63–65, 2012, 73–75), which he defines as follows: A skilled communicator will have some conception of a cognitive path that can lead from the way he thinks of the object about which he has information, to the way in which the listener must think about it, to act in the way that the speaker wants to bring about. (Perry 2001a, 54, 2012, 64) Two different demonstrations will thus create two different cognitive paths. ‘The demonstrations serve to start the paths off in different directions, to associate them with different perceptions, on the hearer’s part, the perceptions of the [dog’s] head and of the [dog’s] tail’ (Perry 2001a, 54, 2012, 102–103). Although Perry keeps the notion of a cognitive path separate from that of cognitive significance as the reflexive content of an utterance of a demonstrative (sentence), they both contain an identifying condition uniquely satisfied by the object being referred to. Perry’s shifting the bearers of cognitive significance from characters and demonstrations taken as types (and/or directing intentions) to reflexive contents of utterances is meant to enable him to account for the co-reference problem and to keep cognitive significance within semantics by keeping it linked with meaning. By being utterance-reflexive, the identifying condition that

32  Other Kinds of Content such a content contains is supposed to be relational rather than descriptive: for something to be its object, it must stand in a certain relation to the utterance itself (Perry 1990/1993, 281). It is thus unlike the identifying condition associated with a demonstration which Kaplan (1989a, 514) likens to the (non-relational) Fregean sense of a definite description. Still, Perry thinks that in going from world to world what we are ‘taking with us’ is not the referent but the utterance of a demonstrative (Perry 2001a, 90, 2012, 103). This is to say that the same utterance of a demonstrative (phrase) may refer to different objects in different possible worlds. He also thinks that our perceptions of objects can be similarly taken from world to world (Perry 2001a, 92, 2012, 105) in a way that allows the same utterance to (counterfactually) correspond to different perceptions and hence to different perception-based psychological states of the subject, as in the case of Kaplan’s characters. In reply, it might be said that each utterance of such a demonstrative phrase that plays a part in expressing a belief is associated with one buffer only. For, every statement (i.e. utterance of a declarative sentence) expressing a belief will, in Perry’s view, have the same reflexive content as the (buffer-involving) belief it expresses. This seems to make the identifying condition supplied by such a statement fit to explain the difference in psychological states that we are after. The reflexive contents of the two utterances of ‘that dog’ in ‘That dog is (not) that dog’ are: That dog 1: the dog that the speaker of that dog 1 is attending to That dog 2: the dog that the speaker of that dog 2 is attending to Perry thinks that ‘this difference is enough to explain the difference in cognitive significance’ (Perry 2001a, 89, 2012, 102) not only in terms of the difference in the relevant reflexive contents but also in terms of the difference in the subject’s psychological states, i.e. beliefs, related to the subject’s different perceptions of the same dog and critical to his action and behaviour that Perry takes to be a matter of semantics. Unlike with the linguistic meanings of demonstratives, characters, and accompanying demonstrations (and other cues) taken as types, there is no mismatch between these reflexive contents and the psychological states of the subject. True, they are second-order contents having these utterances as their subject matter but this does not rule out that ‘what the dog that the speaker of that dog 1 is attending to’ is known by the listener to be a particular perceptually salient dog that he is thinking of, and likewise for the utterance of that dog 2 . This is to say that the respective reflexive contents supply entry points into the buffers, the difference in which will explain the difference in the subject’s psychological states (i.e. cognitive significance) arising in relation to the co-reference problem. A competent listener will effortlessly grasp the reflexive contents of the two utterances of ‘that dog’ that the speaker is

Other Kinds of Content   33 using to refer to Stretch and the respective identifying conditions they contain even if he is unable to perceptually single out Stretch. And, once he has singled it out, these contents will line up with his perception-based psychological states. Still, what the subject is judging is whether the two dog parts that he perceives are parts of the same dog, not whether the referent of the utterance of that dog 1 and the referent of the utterance of that dog 2 are the same. The difference in perceptual buffers that the subject has formed about Stretch is a matter of his first-order beliefs about the world. In falsely believing that that dog1 is not the same as that dog 2 the subject is making an error of fact in thinking about the states of affairs in the world. This suggests that this case ought to be explained in terms of the first-order perception-based beliefs about Stretch and not by their second-order spin-offs. If so, Perry has failed to solve the co-reference problem which he sets out to solve, which is to say that his semantic apparatus is incapable of solving it, as is Kaplan’s or Wettstein’s.3 We can also reach this conclusion by a different route. If b1 and b2 as notions associated, respectively, with two different perceptions of Stretch are two different buffers, we obtain: That dog 1: the dog that the speaker of that dog 1 is attending to using the buffer b1 That dog 2: the dog that the speaker of that dog 2 is attending to using the buffer b2 This way we get buffers fixed to utterances such that to the same utterance the same buffer corresponds across possible worlds. We also do justice to the fact that it is the difference in buffers that motivates the utterer to start the cognitive paths off in different directions and to associate them with different perceptions, on the hearer’s part, the perceptions of the dog’s head and of the dog’s tail. Yet, second-order reflexive contents once again cannot account for the relevant psychological states of the subject. Perry is right in claiming that buffers are tied to ways of acting. But, the ways of acting of the subject who has, say, gained new information about Stretch by being told ‘That dog is the same as that dog’ cannot be accounted for in terms of the second-order reflexive content – that, as it were, buffers b1 and b2 are of the same dog. The subject’s ways of acting are world-directed. In gaining this piece of information about the world, he has corrected an error of fact and abandoned his false belief that he would express by ‘That dog is not the same as that dog’. This suggests that these buffers should rather be associated with first-order contents capturing the subject matter of our beliefs about the world, which is something that Perry seeks to avoid. For him, this kind of content which he calls referential or official content should consist solely of objects and properties or relations.

34  Other Kinds of Content Perry supports this claim by two arguments. The first is the argument from counterfactual truth conditions. He thinks that at least for a wide variety of cases, referentialism gives the right predictions about the possible worlds or situations in which we would count what we say as true. The second argument is the argument from same-saying. In a wide variety of cases, we are in a position to predict correctly the conditions in which two people have said the same thing. ‘We need to say things whose truth or falsity turns on the same objects having the same properties, or standing in the same relations’ (Perry 2001a, 5, 2012, 5–6). This would not be the case if we were to incorporate identifying conditions – buffers included – into these contents, since they may vary from person to person. That buffers ought to be associated with first-order rather than second-order contents does not mean that buffers need to be a part of first-order beliefs. This is made possible by the fact that as new notions associated with the perceptions and used to temporarily store ideas we gain from the perceptions, they are relational rather than descriptive: their objects stand in a certain relation to them. But, we can still assign a semantic role to them in that the correctness or incorrectness of propositional-attitude ascriptions needs to be sensitive to their identity and existence. (For the sake of clarity and simplicity, I grant here that the first-order contents are Russellian; later on, I will propose my own view of content.) This kind of a semantic role differs from the one that Perry assigns to the bearers of cognitive significance in the sense of utterance-reflexive identifying conditions, which are, as noted, tied to meaning and play a dual role of both determining reference (in the context) and of accounting for the cognitive state of the subject, in spite of being (taken to be) relational rather than descriptive. On a par with definite descriptions, Perry also wishes to keep these identifying conditions within the contents of statements. The content of a statement is a proposition that incorporates the conditions under which the statement is true (Perry 2001a, 17, 2012, 23). In the case of (non-indexical) definite descriptions, the truth conditions are provided within the first-order content of a statement by the identifying condition that an object must uniquely satisfy if it is to be the object specified by the description. This kind of identifying condition, supplied by the meaning of the contained description, plays the given dual role and is part of the statement’s subject matter (Perry 2001a, 26, 2012, 33). In the indexical case, the truth conditions are, in Perry’s view, also provided by the relevant identifying condition playing such a dual role, but within the second-order reflexive content, i.e. as the reflexive truth conditions on the utterance of the sentence. Reflexive contents of utterances then belong with demonstrations and the speaker’s directing intentions in forming cognitive paths supplied by the speaker even if a difference in them always matches a difference

Other Kinds of Content   35 in the subject’s psychological states. What about the reflexive contents of perception-based beliefs? Unlike the reflexive contents of utterances, they are buffer-dependent. They are tied to the same buffers for their identity and existence that the underlying first-order beliefs are tied to. Without having a perception-based belief that might be expressed by the statement ‘That dog is the same as that dog’, the subject would not be in a position to grasp its reflexive content – that buffers b1 and b2 are of the same object, i.e. Stretch, because there would be none such. The speaker’s first-order beliefs that are about the same individual also motivate him to start a cognitive path off for the hearer, rather than their second-order spin-offs, which is something that even Perry has acknowledged. Like Kaplan, Perry has then failed to show that the referentialist view of appropriate demonstrative expressions can account for differences in cognitive significance concerning our first-order beliefs about the world. Perry’s notion of cognitive significance as a second-order reflexive content of utterances and beliefs is unfit for this role.

2.4 Notion-Networks As noted, Perry distinguishes the referential or official contents from other kinds of truth-conditional contents of our statements and beliefs. One of these contents is reflexive content that we have just dealt with. Another is intentional content which is invoked to account for the psychological state of the subject in the case in which the subject sincerely assents to an utterance of a sentence containing an empty proper name (see Perry 2001a, 13–14, 123ff, 2012, 14, 165ff). I wish to explore here how this strategy works in the kind of case in which the subject confuses two different objects that she perceives for a single object. I wish to do so in order to support my claim that the object-directed psychological states of the subject need to be explained in terms of the subject’s first-order perception-based beliefs and not along the lines suggested by Perry. In the next chapter, which deals with cases of confusion more fully, I will re-enforce this claim. In dealing with the case of confused belief, Lawlor (2007) makes use of Perry’s notion-networks (as in Perry 2001a). The idea is this. Upon perceiving an object, the subject forms a notion of it. As she gathers information about it, ideas of its properties are attached to the notion, and the information is stored in a temporary perceptual buffer which may give rise to a steadier mental file. A notion-network is the total inter-subjectively maintained set of notions and files about the given object. If there is no single object responsible for the notion, the notion has no origin, and a statement such as ‘That is F’ (used in such a way that the demonstrative is not specifically about any particular object) expresses no Russellian content. The fact that the subject’s belief is not

36  Other Kinds of Content empty is accounted for by the intentional content of the subject’s statements and beliefs which Lawlor calls notional content. It is determined ‘when we specify truth conditions for an utterance by making reference to the origin on the notion that supports the utterance’ (Lawlor 2007, 163). It performs a traditional job of belief-content – that of capturing the subject’s cognitive perspective. Since in the kind of case with which Lawlor is concerned, the object-notion the subject has created has no origin (as we shall see below), it ends in a block having to do with the events ‘that preclude any referent from being identified’ (Perry, 2001a, 123, 2012, 169). To illustrate the suggested strategy, consider a case in which the subject has confused two different dogs looking alike for a single one (as a variant case to that mentioned above in which the subject mistakes the morning star and the north star for the same heavenly body). When she sees one of the dogs, she dubs him ‘Fido’. She subsequently observes Fido and his look-alike a number times, though never together. She decides to feed Fido and at the end of the day makes the following inference (with ‘that dog’ as a memory-based demonstrative phrase that is not specifically about either of the dogs): That dog ate breakfast That dog ate lunch That dog ate dinner On the basis of this, the subject judges that That dog had three square meals today According to Perry’s strategy, the subject reasons well here, regardless of the emptiness of her confused statements. For, each of the subject’s claims in the argument has a related notional content, taking N to be the relevant notion-network: That N has an origin, and he ate breakfast That N has an origin, and he ate lunch That N has an origin, and he ate dinner So, That N has an origin, and he has eaten three square meals today The claim is that due to the subject’s confusion, each of the premises is false as is the conclusion. Yet, she is reasoning well and her reasoning is truth-preserving, hence her beliefs, although referentially empty, are not empty after all (see Lawlor 2007, 167).

Other Kinds of Content   37 True, the subject’s commitments can be captured in this way, and her reasoning can be described as rational. The subject extends her commitments about the world by means of inferential moves that are recognizably rational (Lawlor 2007, 166). Yet, this is not the way to capture her cognitive perspective. If she is to really represent a single dog, we are not going to capture her own point of view in terms of the given notion-networks and notional contents but rather in terms of her taking her incoming information to stem from a single dog. The fact that the subject’s reasoning can be represented in the specified way in the twodog case, and hence that she is being rational, does not make notional contents capture her own point of view. Her point of view is one of representing a single dog which is a matter of her first-order belief and not a matter of taking her notion-network to have an origin. She is not thinking about the origins of her notion-networks. She need not even have the concept of a notion-network. Although she may be committed to this kind of reasoning, she is not going to undertake it in the normal course of circumstances. As remarked, upon perceiving a certain object, the subject forms a notion of it. The fact that an object-notion has no origin in the case when the subject has confused two different objects for a single one does not affect her cognitive perspective which is the same as it would be if her object-notion had an origin. To this I shall return in the next chapter.

2.5 Conclusion The foregoing discussion has shown that none of the referentialist views that I have considered are capable of providing a satisfactory account of the co-reference problem concerning informative identities consisting in the fact that statements containing co-referential perception-based demonstratives may differ in cognitive significance, i.e. have different cognitive values. In arguing against Perry’s recent attempt to solve this problem, I have shown that a viable solution to this problem needs to be supplied in terms of the subject’s first-order perception-based beliefs and not in terms of her second-order reflexive contents of statements or beliefs. I have also argued that object-directed psychological states of the subject need to be explained in this way in the case in which the subject has confused two different objects for a single one rather than in terms of the intentional contents of her statements and beliefs introduced by Perry and utilized by Lawlor. In contrast with this, neo-Fregeans account for the co-reference problem in terms of the subject’s first-order perception-based beliefs. In conformity with this, they appeal to two different senses as two different ways in which the same object is presented to the subject. Yet, senses seem to fail to capture the subject’s cognitive perspective by failing to be the bearers of cognitive significance in relation to the “yesterday/

38  Other Kinds of Content today”  case, as Kaplan claims. And, unlike referentialist contents (i.e. Perry’s official contents), senses may vary interpersonally which makes them shifty and unstable. This, in turn, makes senses unsuitable for being parts of attitude contents, in view of the noted fact, that attitude contents need to enable us to predict correctly the conditions in which two people have said the same thing. This requirement is not met if we incorporate identifying conditions – buffers included – into these contents, since they may vary from person to person. In addition, the anti-individualist view of thought-content that the neo-Fregean philosophers embrace fails to explain the subject’s cognitive perspective in yet another respect, together with other anti-individualist views of thought-content. This is the topic of the next chapter.

3

Anti-individualism and Cognitive Perspective

3.1 Introduction We have seen that according to Kaplan’s and Perry’s referentialist views, Russellian content is not an internal property of the subject. Antiindividualist philosophers adopt a similar view. But, unlike the referentialists, a number of anti-individualist philosophers believe that content believed and asserted serves to explain the subject’s cognitive perspective. In this chapter, I examine representative anti-individualist views. In relation to perception-based demonstrative thoughts, I will argue that, their intentions notwithstanding, Fregean and non-Fregean antiindividualists alike fail to provide a theory of content that explains the subject’s cognitive perspective and propose an alternative theory of content that explains the subject’s cognitive perspective. Anti-individualists such as Burge (2010) hold that thought-content serves to explain the subject’s cognitive perspective thanks to the transparency of sameness and difference of thought-contents in spite of their not being an internal property of the subject. A case in point is Stalnaker’s (2008), whose view merits a full-length discussion. In agreement with his claim that cases in which the subject falsely identifies different things do not threaten a violation of the transparency of difference of thought-contents, I will argue that, his expectations to the contrary, this view of transparency of thought-contents does not serve to explain the subject’s cognitive perspective on Stalnaker’s own terms and that the intricacies involved in his argumentation for saving his anti-individualist project are indirectly supportive of an individualist account of content.

3.2 Sameness and Difference of Thought-Contents Individualism about content (content-internalism) maintains that all or most thought-contents do not depend for being the contents they are on any relations to entities beyond the body of the subject while antiindividualism about content (content-externalism) maintains that there are thought-contents that depend for being the contents they are on relations to such entities. Individualism maintains that having a thought with a

40  Anti-individualism & Cognitive Perspective certain intentional content is an intrinsic property of the subject while anti-individualism maintains that having such a thought is a relational rather than an intrinsic property of the subject (see, e.g., Burge 2010, 25f). As a result, it seems that individualism is compatible with the transparency of thought-content while anti-individualism is not compatible with it (see, e.g., Brown 2004, chap. 5). The transparency thesis is the thesis that a subject can realize a priori, i.e. without using empirical information, whether two thoughts or thought-constituents have the same or different contents. Boghossian states it as follows: (a) If two of a thinker’s token thoughts possess the same content, then the thinker must be able to know a priori that they do; and (b) If two of a thinker’s token thoughts possess distinct contents, then the thinker must be able to know a priori that they do. (Boghossian 1994, 36, 2011, 457) Boghossian (1994, 36) uses “a priori” to mean “independent of outer experience” in a way that is consistent with knowledge being a priori if it is based on inner experience or introspection (see also Brown, 2004, chap. 5). When mental contents violate either (a) or (b), we get cases in which the subject who intuitively looks fully rational is made to look as if she is committing a simple logical fallacy in her reasoning. Violations of transparency blur the distinction between errors of reasoning and errors of fact. On the face of it, anti-individualism violates (a) and (b) and should, therefore, be rejected (see Boghossian 2011, 457–458). To this I shall return shortly. Yet, both Fregean anti-individualists who combine anti-individualism with Fregean sense and (relevant) non-Fregean anti-individualists who reject Fregean sense (see Brown 2004, 20) hold that a theory of content should provide an explanation of the subject’s cognitive perspective. In agreement with this, and in relation to perception-based demonstrative thoughts, I will argue that relevant anti-individualist views of content fail to explain the subject’s cognitive perspective and propose an alternative view that meets this requirement. To start with, consider the neo-Fregean view about these matters as proposed by Evans (1982), Peacocke (1983), McDowell (1986), and Campbell (1994). As Fregeans, these philosophers hold that if a rational subject simultaneously entertains two appropriate thoughts or thought constituents with the same content (i.e. sense), she can grasp a priori that they have the same content. As anti-individualists, they hold that singular thoughts are object-dependent in that the identity (and existence) of their contents is tied to the identity (and existence) of the objects they are about. Insofar as this entails that the subject who is perceiving two different objects but takes them for a single object can think two thoughts with different contents without realizing a priori that they have different

Anti-individualism & Cognitive Perspective  41 contents, these philosophers face the following asymmetries pointed out by Brown (who traces them to the Fregean conception of rationality which she finds ill-motivated as we shall see below): Grasp of Sameness, not Difference: A rational subject can always simultaneously grasp a priori that two thoughts or thought constituents with the same content have the same content while she cannot always grasp a priori that two thoughts or thought constituents with different contents have different contents Grasp of Validity, not Invalidity: A rational subject can always grasp a priori simple instances of validity and contradiction, but cannot always grasp a priori simple instances of invalidity This is to say that the transparency of thought-content will lead a rational subject to put two thoughts with the same content together in various simple valid inferences, while the opaqueness of difference of thought-content will lead a rational subject to put thoughts with different contents together in various invalid inferences. To illustrate this, consider the following example derived from Perry (1977/1993) and Stalnaker (2008). Suppose a rational subject who perceives a certain aircraft carrier with the British flag flying from it reasons as follows: I This ship (pointing to the ship’s bow) is an aircraft carrier This ship (pointing again to the ship’s bow) is British Therefore, there is a British aircraft carrier in the harbour1 According to Fregean anti-individualists, the subject will think of the given ship via the same sense in both premises. She will grasp a priori that it is the same sense and that the given inference is valid. For, in making such an inference, she presumes the identity of the ship designated by the two utterances of ‘this ship’. It would be useless to add this as a premise. The notion of sense is designed for this very purpose: one gets to presume rather than assert identity when sense is the same and the inference may trade directly upon the fact of co-reference of two singular terms or of two different utterances of the same or different indexicals (see, e.g. Campbell 1987, 2002, chap. 5; see also Recanati 2012, chap. 4). The attempt to reduce the identity presumption to an implicit identity judgement, and have the subject think of the given ship via two different senses, embarks us on an infinite regress. For, we would also need to make sure that the ship referred to in the suppressed premise is

42  Anti-individualism & Cognitive Perspective the same as the ship referred to in the explicit premises, and we would need further suppressed premises to secure this. ‘The problem recurs, and we are embarked on a regress’ (see Campbell 1987, 75). Now, consider a case in which the subject takes two different ships that she perceives in a harbour via their respective parts for a single ship. Suppose that behind a large building, the bow of an aircraft carrier is visible on the left, while behind the building on the right, the stern of a different shorter ship with the British flag on it blocks the view of the stern of the first ship such that the subject believes that this ship (looking at the bow of the first ship) is the same as that ship (looking at the stern of the second ship). As a result, the subject will be disposed to reason as follows: II This ship (pointing to the bow of the first ship) is an aircraft carrier This ship (pointing to the stern of the second ship) is British Therefore, there is a British aircraft carrier in the harbour As a result, Fregean anti-individualists seem committed to the view that the subject will grasp two senses that are different as a result of observing two different ships, yet fail to realize a priori that they are different due to her failure to realize that the ships are different. As a result, she will fail to grasp a priori that the inference is invalid. In an attempt to avoid these asymmetries, anti-individualists can adopt one of the following two theses: Opacity Thesis: Sameness of thought-contents can be opaque, as can their difference Transparency Thesis: Difference in thought-content is always transparent, as is their sameness While Fregean anti-individualists adopt the Transparency Thesis, non-Fregean anti-individualists are free to choose between the Opacity and the Transparency theses. Those non-Fregean anti-individualists who take it that the subject will grasp the thought constituent with the same content as long as the object that she is thinking about is the same have the Opacity Thesis at their disposal. The guiding assumption in urging that one of these theses needs to be adopted is that if thought-content is to serve to explain and rationalize the subject’s cognitive perspective, this needs to be done in the same kind of way whether the content is the same or different. It would be odd to explain in the one case the subject’s cognitive perspective in terms of content that lines up with the subject’s cognitive perspective while not in the other. As Brown has aptly put it regarding Fregean anti-individualists, ‘[it] seems curiously asymmetric

Anti-individualism & Cognitive Perspective  43 for a single position to affirm transparency of sameness of content but deny transparency of difference of content’ (Brown 2004, 226). I shall turn now to discussing the relevant anti-individualist views that the adoption of the Opacity Thesis or the Transparency Thesis gives rise to. I will show that all of these views are deficient and argue for an alternative that conforms with the Transparency Thesis. This alternative is committed to a conception of transparency of thought-content that is encapsulated in a criterion of difference for thought-contents derived from Frege mentioned above, which depicts thought-content capturing the subject’s cognitive perspective: (CD) Two thoughts have different contents if it is at the same time possible for a rational subject to take conflicting epistemic attitudes towards them. Hence, if two thoughts have the same content, the subject cannot at the same time rationally take conflicting epistemic attitudes towards them. 2

3.3 Opacity of Thought-Contents Consider the Opacity Thesis – the view that sameness of thought-contents can be opaque, as can their difference. This is in line with the anti-individualist view as defined above. It shuns Grasp of Sameness, not Difference by claiming that a rational subject need not always grasp a priori that two thoughts or thought constituents with the same content have the same content, just as she need not always grasp a priori that two thoughts or thought constituents with different contents have different contents. And it shuns Grasp of Validity, not Invalidity by claiming that the subject need not always grasp a priori simple instances of validity and contradiction, just as she need not always grasp a priori simple instances of invalidity. Since Brown’s position is representative of this kind of anti-individualism, I shall confine my discussion to this position (see Brown 2004, chap. 5). In Brown’s view, commitment to Grasp of Sameness, not Difference and Grasp of Validity, not Invalidity on the part of Fregean anti-individualists shows that the underlying Fregean conception of rationality encapsulated in (CD) is lopsided and hence ill-motivated. To abandon this conception, and with it (CD), is to adopt the Opacity Thesis. To illustrate this, suppose that the subject takes the two ship-ends that she sees on the left and the right of a large building, which in fact belong to the same ship, to belong to two different ships. As a result, she will refrain from making the following inference: I* This ship (pointing to the ship’s bow) is an aircraft carrier This ship (pointing to the ship’s stern) is British Therefore, there is a British aircraft carrier in the harbour

44  Anti-individualism & Cognitive Perspective To shun Grasp of Sameness, not Difference and Grasp of Validity, not Invalidity in conformity with the Opacity Thesis is to hold that the thought constituents expressed by the two respective utterances of ‘this ship’ have the same content but the subject does not grasp this a priori. Hence, the given inference is valid, yet the subject does not grasp a priori that it is valid (see Brown 2004, 162, 200). Still, the subject’s failure to make such a valid inferential move is justified, i.e. rational, given her view of the situation. Brown believes that this provides non-Fregean anti-individualists with an alternative explanation of the phenomena that Fregeans want to explain by appealing to sense. She also believes that non-Fregean anti-individualism is incompatible with transparency of sameness of thought-content just as it is incompatible with transparency of difference of thought-content, i.e. that it is committed to the Opacity Thesis (although, as we shall see below, Stalnaker does not think so). ‘As a result, anti-individualism undermines a subject’s ability to grasp a priori the logical properties of her thoughts. Thus, if anti-individualism is true then, even in simple cases, a subject may have contradictory beliefs, make invalid inferences, or fail to make valid inferences’ (Brown 2004, 191–192). This is to say that thought-contents as Brown individuates them do not serve to explain the subject’s cognitive perspective. Still, Brown wants to explain and rationalize this perspective and in so doing she trades on the kind of thought-content that conforms with (CD) and the Fregean conception of rationality. In taking the same ship for two different ones in the foregoing sense, the subject has (as Brown would put it) the contradictory belief that This ship (pointing to the ship’s bow) is not This ship (pointing to the ship’s stern). To explain how she can hold such a belief and her cognitive perspective is to say that she is thinking of the same ship in two different ways, i.e. via two different senses, as a result of her conception of the situation which is one of ‘being presented with two different ships’. Alternatively, we may leave thought-content out when accounting for the subject’s cognitive perspective, reasoning, and actions and thus deprive it of one of its key roles. 3 As will become clear in due course, the complementary view that the rational subject need not always grasp a priori that two thoughts or thought constituents with different contents have different contents, also faces problems when it comes to capturing the subject’s cognitive perspective. I conclude that Brown has failed to show how the phenomena that Fregeans want to explain by appealing to sense can be explained in terms of the Opacity Thesis. She has failed to show how perception-based demonstrative thought-contents as she individuates them serve to explain the subject’s cognitive perspective. This calls for individuating perception-based demonstrative thought-contents in terms of the Transparency Thesis.

Anti-individualism & Cognitive Perspective  45

3.4 Transparency of Thought-Contents According to the Transparency Thesis, difference in thought-contents is always transparent, as is their sameness. While Stalnaker adopts this thesis, he contends, along the lines of non-Fregean anti-individualism, that the property of having a thought with a certain intentional content is a relational rather than an intrinsic property of the thinker (Stalnaker 2008, 114, n. 5). This means that if the relevant information of which the subject is in possession causally derives from two different objects, she will in the relevant circumstances entertain two different thought- contents (propositions), even if she takes this information to derive from a single object. This seems to make a difference of thought-contents (propositions) opaque to the subject. However, in relation to perception-based demonstrative thoughts, and against Boghossian’s violation of transparency charge, Stalnaker (2008, 115) argues that an anti-individualist account of the facts that determine content can be reconciled with a suitably qualified version of a principle of epistemic transparency. He argues that cases in which the subject falsely identifies different things do not threaten a violation of the transparency of difference of thought-contents (propositions) (Stalnaker 2008, 126), in a bid to meet his requirement, shared by other anti-individualists, that thought-content serves to explain the subject’s cognitive perspective. He does not defend the principles of the transparency of sameness and difference exactly as Boghossian states them (in the foregoing formulation) nor does he provide an alternative formulation. His aim is just to show ‘that an anti-individualist thesis about content attribution is compatible with an account of reasoning and errors of fact, as Boghossian rightly says that any adequate account must be’ (Stalnaker 2008, 115). For: [i]f we could be wrong, on empirical grounds, about the contents of our own thoughts, then we could be wrong, on empirical grounds, about the validity of our reasoning, and this seems incompatible with the idea that we can separate the assessment of reasoning from the assessment of the truth of the premises on which the reasoning is based. (Stalnaker 2008, 114–115) Stalnaker’s position is representative of this kind of anti-individualism, and I shall confine my discussion to this position in what follows. Granting to Stalnaker that cases in which the subject falsely identifies different things do not threaten a violation of the transparency of difference of thought-contents, and in agreement with the view that thought-content should serve to explain the subject’s cognitive perspective, I argue that, his intentions notwithstanding, this view of transparency of thoughtcontents does not serve to explain the subject’s cognitive perspective on

46  Anti-individualism & Cognitive Perspective Stalnaker’s own terms and that the intricacies involved in his argumentation for saving his anti-individualist project are indirectly supportive of an individualist account of the subject’s cognitive perspective. In so doing, I leave intact some of his key claims that are plausible in their own right and propose an alternative that serves them better. This will help advance further the view of how we should account for the subject’s cognitive perspective that is advocated in this book. 3.4.1 Transparency of Difference As noted, Stalnaker holds the anti-individualist view that the property of having a thought with a certain intentional content is a relational rather than an intrinsic property of the thinker (Stalnaker 2008, 114, n. 5).4 It is part of this view that if the relevant information of which the subject is in possession causally derives from two different objects, thought-contents (propositions) based on the causal relations with these objects will be different even though she takes this information to derive from a single object. This seems to make thought-contents (propositions) opaque to the subject. But Stalnaker does not think so. As we have seen, he thinks that ‘an anti-individualist account of the facts that determine content can be reconciled with a suitably qualified version of a principle of epistemic transparency’ (Stalnaker 2008, 115). In accordance with this, Stalnaker claims that cases in which the subject falsely identifies different things do not threaten a violation of the transparency of difference of thought-contents (propositions) (Stalnaker 2008, 126). To show this, Stalnaker considers the aforementioned case in which the subject takes two different ships that she perceives in a harbour via their respective parts for a single ship. Behind a large building, the bow of an aircraft carrier is visible on the left, while behind the building on the right, the stern of a different shorter ship with the British flag in it blocks the view of the stern of the first ship such that the subject believes that this ship (looking at the bow of the first ship) is the same as that ship (looking at the stern of the second ship). As a result, the subject will be disposed to reason as follows: II This ship (pointing to the bow of the first ship) is an aircraft carrier This ship (pointing to the stern of the second ship) is British Therefore, there is a British aircraft carrier in the harbour It is clear that the two premises are about different ships and both are true, but the conclusion is false. Since the premises are about different ships, the propositions that this ship is an aircraft carrier and that that ship is in aircraft carrier are distinct by the lights of anti-individualism but the subject seemingly fails to grasp that they are different. In failing to grasp that the given inference is invalid, she will fail to grasp that that

Anti-individualism & Cognitive Perspective  47 these propositions are different. We get a case in which the subject who intuitively looks fully rational, and is merely missing some empirical information, is made to look as if she is committing a simple logical fallacy in her reasoning. However, Stalnaker thinks that while the two premises are respectively about different ships and true, and the conclusion is false, in thinking that there is only one ship in play, the subject is not making an unwitting logical error. She is not confusing the proposition that this ship is an aircraft carrier with the distinct proposition that that one is. Her reasoning is best represented as involving a false tacit presupposition, a suppressed premise, that this ship is that one, rather than a false belief that the two thoughts have the same content. Stalnaker continues: This way of representing the reasoning does not assume that [the subject] has entertained the possibility that the two ships are different – the possibility that distinguishes the two propositions – or that the proposition that excludes this possibility is in any way encoded at some perhaps subpersonal level in [the subject’s] cognitive apparatus. Most of what we presuppose is presupposed simply by not recognizing the possibilities in which the presuppositions are false. The explicit statement of the tacit presupposition is part of the theorist’s representation of the situation. (Stalnaker 2008, 127) Stalnaker suggests that the reason it is appropriate to credit the given subject with the presupposition in question has to do with the fact that she would have no difficulty understanding the claim that this ship is an aircraft carrier, while that one is not, and would not take the claim to be a simple contradiction.5 He thinks that it is reasonable to say that, while the subject has the correct belief that the two statements (that this is an aircraft carrier and that that one is) distinguish between the possibilities compatible with what she is presupposing in the same way, we need not say that she believes that they express the same proposition, relative to a wider range of possibilities (Stalnaker 2008, 128). Since inference (II) is valid, the subject is not confusing the proposition that this ship is an aircraft carrier with the distinct proposition that that one is although she errs as to how many ships there are. There is no threat of a violation of the transparency of difference of these propositions, and the distinction between errors of reasoning and errors of fact is preserved. While the subject is confusing the two ships for one ship, she is not confusing the two propositions for one proposition. Insofar Stalnaker is right in claiming that an anti-individualist thesis about content attribution is compatible with an account of reasoning that is clear about the difference between errors of reasoning and errors of fact, that any adequate account must be (Stalnaker 2008, 115).

48  Anti-individualism & Cognitive Perspective The claim that the explicit statement of the subject’s presupposition that this ship is that one is part of the theorist’s representation of the situation is itself part of Stalnaker’s view that it is us interpreters who attribute partial propositional contents to pick out contextually relevant aspects of the subject’s total state of understanding at a time. As a result, there is no single interest-independent objective propositional content of the subject’s thoughts. There is no invariant, context-independent way of characterizing her thought-contents (Stalnaker 2008, 102–105, 131). In those cases in which the subject confuses two different objects for a single one, Stalnaker attributes partial intensions defined on the set of possibilities compatible with the subject’s empirical presuppositions about how many objects there are. These partial intensions are not exhaustive or definitive characterizations of the subject’s state of mind but interest-relative samplings of complex cognitive states. Stalnaker’s reconciliation of an anti-individualist account of the facts that determine content with transparency is rooted in a basic interpretative principle to assign anti-individualist contents in such a way as to be reflected in the subject’s own reasoning and dispositions. This suggests that in the two-ship case that we are discussing the theorist attributes to the given subject the presupposition that this ship is that one in accordance with the subject’s empirical presupposition that there is a single ship in play, which is neither articulated in the subject’s reasoning nor subpersonally encoded. Compliance with the subject’s empirical presupposition about how many ships there are is required because we ascribe thought in order to explain action and to assess the reasoning of thinkers, and such explanations and assessments cannot turn on facts that are inaccessible to the subject (see Stalnaker 2008, 114). Stalnaker continues: Thinkers are things with a capacity to make their actions depend on the way the world is, and with dispositions to make their actions depend on the way they take the world to be. Theorists and attributors of thought characterize these capacities and dispositions by locating the world as the thinker takes it to be in a space of relevant alternative possibilities. The theorist uses actual things and properties to describe these possibilities, and that is why content depends on facts about the actual world … [W]hen [the thinker] conflates distinct things, or thinks of one thing as two,…, we may find it difficult to characterize a world according to the thinker that is apt for describing that person’s cognitive capacities and dispositions. But our descriptive resources are rich and flexible, and in context, we can usually find a way. What counts as a correct description of the world according to the thinker may depend on the attributor’s context. A principle of epistemic transparency is satisfied, according to this picture, not because the thinker is directly acquainted with

Anti-individualism & Cognitive Perspective  49 an inner object that has an inner content essentially, but because an apt description of a thinker’s cognitive state, if it is to explain the rational capacities and dispositions it is intended to explain, must represent the way the world is according to the thinker in a way that satisfies it. (Stalnaker 2008, 131) We have seen that in the discussed two-ship case, a principle of transparency is satisfied in that in giving an apt description of the subject’s cognitive state, the theorist attributes to her the presupposition that this ship is that one which makes inference (II) valid. Hence there is no threat of a violation of the transparency of difference between the proposition that this ship is an aircraft carrier, and the proposition that that ship is an aircraft carrier. But since such a description must represent the way the world is according to the subject, in view of the fact that there is incontrovertible sameness of the ship from the subject’s point of view, the theorist attributes to her this presupposition in accordance with her belief that the two statements (that this is an aircraft carrier and that that one is) distinguish between the possibilities compatible with what she is presupposing in the same way, and not because ‘we need not say that [s] he believes that they express the same proposition, relative to a wider range of possibilities’ as Stalnaker suggests (Stalnaker 2008, 128).6 As a result, Stalnaker’s attempt to reconcile an anti-individualist account of the facts that determine content with transparency that is rooted in a basic interpretative principle to assign anti-individualist contents in such a way as to be reflected in the subject’s own reasoning and dispositions fails. The anti-individualist contents in question, i.e. the propositions that this ship is an aircraft carrier and that that ship is an aircraft carrier, are not reflected in the subject’s own reasoning and dispositions since they do not serve to represent the way the world is according to the subject. This is so in spite of the acknowledged fact that the subject is not confusing these two propositions thanks to the validity of (II), which way the distinction between errors of reasoning and errors of fact is preserved. 3.4.2 Transparency of Sameness To recall, Stalnaker claims that an anti-individualist thesis about content attribution is compatible with an account of reasoning that is clear about the difference between errors of reasoning and errors of fact that any adequate account must be. We have seen that this difference is preserved in the case in which the subject confuses two different ships for a single one. The issue that I want to address now is how the difference between errors of reasoning and errors of fact is supposed to be preserved in the reverse case (which Stalnaker is not committed to but will help further

50  Anti-individualism & Cognitive Perspective our discussion). Suppose that behind a large building, the bow of an aircraft carrier is visible on the left, while on the right of the building, the stern of the same ship with the British flag on it is visible but the subject takes them to belong to two different ships. As a result, she will refrain from making the following inference: I* This ship (pointing to the ship’s bow) is an aircraft carrier This ship (pointing to the ship’s stern) is British Therefore, there is a British aircraft carrier in the harbour We have seen that, in conformity with anti-individualism, Stalnaker claims that if the relevant information of which the subject is in possession causally derives from two different objects, thought-contents (propositions) based on the causal relations with these objects will be different even though she takes this information to derive from a single object. In a parallel fashion, suppose that the fact that the relevant information of which the subject is in possession causally derives from the same ship makes it the case that the proposition that this ship is an aircraft carrier is the same as the proposition that that ship is an aircraft carrier, such that (I*) is valid. Can the strategy that Stalnaker employs in the case of (II) work here in accounting for the error the subject is making? In the case of (II), the subject is not making an error of reasoning in spite of making an error of fact thanks to the premise ‘this ship is that one’, which turns an invalid argument into a valid one. By contrast, adding the premise ‘this ship is not that one’ to (I*) does not affect its validity. With or without this premise, (I*) is valid. This leaves us with a puzzle of how to explain the fact that the subject fails to grasp its validity without committing any logical error.7 The distinction between errors of reasoning and errors of fact seems blurred. The subject who intuitively looks fully rational, and is merely missing some empirical information, is made to look as if she is committing a simple logical fallacy in her reasoning. In reply, it can be said that this is simply an available inference which the subject fails to make; it is not an error. In aiming to avoid epistemic errors, Stalnaker need not to be concerned about failures to make available inferences. He could also claim that here it is a different erroneous implicit thought which explains the “error” (again, not an error, merely a missed opportunity), namely, the tacit presupposition that this ship is not that one. By default, each of these moves could preserve the difference between errors of reasoning and errors of fact. For, in failing to make an inference, the subject is not committing an error of reasoning. However, failing to make an inference in accordance with either of these replies makes the proposed anti-individualist content irrelevant in describing the subject’s cognitive state. That this kind of content does not figure in describing the subject’s cognitive state by Stalnaker’s own lights will be shown in what follows.

Anti-individualism & Cognitive Perspective  51 3.4.3 The Enterprise Case While Stalnaker does not consider an argument such as (I*) he does consider the underlying case in which the subject takes the same ship as perceived via its two different parts for two different ships. He quotes Perry: Suppose I am viewing the harbor from downtown Oakland; the bow and the stern of the aircraft carrier Enterprise are visible, though a large building obscures the middle. The name “Enterprise” is clearly visible on the bow, so when I tell the visitor “This is the Enterprise”, pointing toward the bow, this is readily accepted. When I say, pointing to the stern clearly several city blocks from the bow, “That is the Enterprise”, however, she refuses to believe me. (Perry 1977/1993, 12–13; Stalnaker 2008, 81–82) Stalnaker claims that in this kind of case, the content of what is expressed or believed in a context is not detachable from the context in which it is expressed or believed (Stalnaker 2008, 81–82). This is in line with Stalnaker’s (1978) basic two-dimensional semantics which assigns truth conditions relative to pairs of worlds. Possible-worlds content can then be assigned relative to subsets of worlds when for a given subset, the two-dimensional matrix restricted to that subset produces constant rows. ‘This is F’ and ‘That is F’ always produce different full twodimensional matrices since there are worlds in which ‘this’ and ‘that’ are used targeting different objects. However, restricted to certain sets of worlds, they may produce the same possible-worlds content. So, both in the case in which the subject takes two different ships for a single one and in the case in which she takes a single ship for two different ones, transparency can let the subject know that the ‘this’ and ‘that’ thoughts are different (relative to a sufficiently broad set of worlds, i.e. relative to a wider range of possibilities, as stated above). As a result of the transparent difference of the ‘this’ and ‘that’ thoughts (relative to a sufficiently broad set of worlds), the subject who takes a single ship for two different ones will refrain from reasoning in accordance with (I*) which turns out invalid, which protects her from making an error of reasoning. Hence, the aforementioned problem concerning the transparency of the sameness of thought-content does not arise. If, on the other hand, the subject is presupposing a restricted set of worlds (which is the effect of Stalnaker’s identity presupposition), the ‘this’ and ‘that’ thoughts will coincide in contextual content, and the subject can also transparently know this. However, as noted, Stalnaker also holds that it is us interpreters who attribute restricted sets of worlds, i.e. partial propositional contents, to pick out contextually relevant aspects of the subject’s total state of understanding at a time (as evidenced by Stalnaker’s aforementioned

52  Anti-individualism & Cognitive Perspective claim that the explicit statement of the identity presupposition is part of the theorist’s representation of the situation). There is no single interest-independent objective propositional content of the subject’s thoughts. When the subject confuses two different objects for a single one, partial intensions attributed to her are not exhaustive or definitive characterizations of her state of mind but interest-relative samplings of complex cognitive states. For, in addition, there are also different antiindividualist contents relative to a wider range of possibilities which are for Stalnaker (as we have seen, mistakenly) assigned in such a way as to be reflected in the subject’s own reasoning and dispositions. In contrast with this, in the Enterprise case no anti-individualist contents are invoked. The ‘this’ and ‘that’ thoughts do not have the same content (i.e. the proposition that this ship is an aircraft carrier is not the same as the proposition that that ship is an aircraft carrier), although the relevant information of which the subject is in possession causally derives from the same ship whether she is observing it via its bow or via its stern. Hence, contents, i.e. intensions, attributed to the subject compatible with her empirical presuppositions about how many objects there are, turn out to be exhaustive or definitive characterizations of the subject’s state of mind in the relevant sense. What by Stalnaker’s own lights turns out to be reflected in the subject’s reasoning and dispositions in the Enterprise case is her taking it that there are two different ships in play. 3.4.4 Internal Facts To say that thought-contents (intensions) attributed to the subject compatible with her empirical presuppositions about how many ships there are exhaustively characterize the subject’s state of mind is to appeal to internal facts about the subject. In accounting for the Enterprise case, Stalnaker takes internal cognitive facts to determine which intensions represent the subject’s state of mind. This is in line with his aforementioned claim that a principle of epistemic transparency is satisfied because an apt description of a thinker’s cognitive state, if it is to explain the rational capacities and dispositions it is intended to explain, must represent the way the world is according to the thinker in a way that satisfies it. In representing the world as it is according to the thinker – according to how many ships she takes there to be – the theorist needs to appeal to internal facts in order to explain the subject’s cognitive state.8 Recall that in the foregoing case in which the subject confuses two different ships for a single one, Stalnaker attributes partial intensions defined on the set of possibilities compatible with the subject’s empirical presuppositions about how many ships there are which are not exhaustive or definitive characterizations of the subject’s state of mind but interestrelative samplings of complex cognitive states. Once again, Stalnaker

Anti-individualism & Cognitive Perspective  53 takes internal cognitive facts to determine which intensions represent the subject’s actual cognitive state. In addition to these partial intensions, in the two-ship case, Stalnaker invokes anti-individualist contents, i.e. contents that are different relative to a wider range of possibilities, that account for the subject’s cognitive state. But, as I have argued, these contents are not reflected in the subject’s own reasoning and dispositions since they do not serve to represent the way the world is according to the subject. So, we are left with partial intensions to account for this. As argued, the subject’s being disposed to reason along the lines of (II) is represented as involving a false tacit presupposition, a suppressed premise, that this ship is that one because of her empirical presupposition that one ship is in play. 3.4.5 Concepts So how are we then to spell out these internal facts? If an apt description of the subject’s cognitive state must represent the way the world is according to the subject, the subject’s taking the Enterprise for two different ships sees her as thinking of it via two different concepts or modes of presentation. It is because of this that in the kind of case in which the subject takes the same ship for two different ones she is not disposed to reason as follows: I* This ship (pointing to the ship’s bow) is an aircraft carrier This ship (pointing to the ship’s stern) is British Therefore, there is a British aircraft carrier in the harbour This explains why she is not making an unwitting logical error in taking an invalid argument for a valid one in spite of making an error of fact. How about the reverse case in which the subject has confused two different ships for a single one? We can think of the subject’s cognitive state as involving a single concept or mode of presentation which informs partial intensions defined on the set of possibilities compatible with the subject’s empirical presuppositions about how many ships there are. In relation to this kind of case, Recanati correctly suggests: [D]oes this mean that there is one [mode of presentation] with two objects of reference…? Or that there are two [modes of presentation] and that I am mistakenly operating on the assumption that there is one? If the latter, in what sense can [modes of presentation] be said to play the role … of capturing the subject’s point of view? But there is a third option: there is a single [mode of presentation],… but that [mode of presentation] rests on a false presupposition of identity, so it fails to refer (rather than referring to two objects simultaneously, as per the first horn of the dilemma). (Recanati 2013, 1851, n. 7)9

54  Anti-individualism & Cognitive Perspective Now, recall that thanks to the false presupposition of identity the subject is, according to Stalnaker, not making an unwitting logical error in being disposed to reason as follows in the foregoing two-ship case: II This ship (pointing to the bow of the first ship) is an aircraft carrier This ship (pointing to the stern of the second ship) is British Therefore, there is a British aircraft carrier in the harbour The false presupposition that this ship is that one ensures that (II) is valid. The conclusion would follow if this presupposition that plays the role of a suppressed premise were true. To be sure, the fact that the given concept or mode of presentation fails to refer does not stand in the way of the fact that the two original premises in (II) are about two distinct ships and both true, as Stalnaker claims. They are about distinct ships due to the fact that the subject is having an appropriate contextual referential relation with each of the ships in the process of producing them. It should be noted that the concepts or modes of presentation that inform partial intensions are not creations of the problematic Cartesian picture of the mind and hence creatures of darkness, as Stalnaker (2008, 105) takes concepts to be. Since the subject’s realizing in the Enterprise case that different concepts are involved amounts to her taking the Enterprise for two different ships, her doing so does not commit her to having (direct) access to her own mental states. The same applies to the foregoing two-ship case. Since the subject’s realizing that a single concept is involved amounts to her taking the two ships for a single ship, her doing so does not commit her to having (direct) access to her own mental states. To paraphrase Stalnaker’s foregoing remark, the subject is not (committed to being) directly acquainted with an inner object that has an inner content essentially (Stalnaker 2008, 131). This way we bypass the issue of whether there is such a thing as an inner realm of a kind that Stalnaker alludes to. This also enables us to bypass the issue as to whether the subject who realizes that she is deploying a single concept or two different concepts can be introspectively aware of this, as Boghossian claims, or whether their views are compatible. (To be sure, Boghossian’s definitions of the transparency of sameness and difference are spelt out in terms of thought-contents which contain concepts as their constituents (Boghossian 2011, 461).) A fortiori, the kind of access that the subject is having to thoughtcontents both in the Enterprise case and in the two-ship case does not involve her having meta-beliefs about them. And she is surely not having (direct) access to the facts concerning reference since in the two-ship case, she does not realize that the concept (mode of presentation) fails to refer.10 Still, concepts do not depend for being the concepts they are on relations beyond the body of the subject and are insofar in line with

Anti-individualism & Cognitive Perspective  55 individualism, as evidenced by the fact that in the two-ship case, the subject is deploying a single concept or mode of presentation (that fails to refer) although the relevant information of which she is in possession causally derives from two different ships. Since concepts are not part of the propositions (thought-contents) which are for Stalnaker unstructured, i.e. sets of possible worlds, this internalist feature of concepts is not transferred to propositional content. (To be sure, individualism and anti-individualism are theses about thought-content and, short of being content constituents, concepts cannot transfer this feature to contents.) But this comes at the cost that propositional contents are not reflected in the subject’s reasoning and dispositions. This comes at the cost that the intricacies involved in the argumentation for saving Stalnaker antiindividualist project are indirectly supportive of an individualist account of the subject’s cognitive perspective. This does not mean that we should abandon Stalnaker’s anti-individualism or that we should opt for structured propositions instead of unstructured ones (as Boghossian 2011 and Schroeter 2013 are inclined to do). All we need to do is recognize and acknowledge the role of concepts (modes of presentation) in representing the subject’s cognitive perspective (and in attributing to her the presupposition that this ship is that one in the relevant case) without fearing that they are creations of the problematic Cartesian picture of the mind. 3.4.6 Concluding Remarks In relation to perception-based demonstrative thoughts, Stalnaker tries to reconcile an anti-individualist account of the facts that determine thought-content with a suitably qualified version of a principle of epistemic transparency. He convincingly argues that cases in which the subject falsely identifies different things do not threaten a violation of the transparency of difference of thought-contents. He is right in claiming that in being disposed to reason in accordance with (II) the subject is not making an unwitting logical error, whereby the distinction between errors of reasoning and errors of fact is preserved. He rightly claims that the subject’s reasoning is best represented to involve a false tacit presupposition, a suppressed premise, that this ship is that one. But in fulfilling his own requirement that we ascribe thought in order to explain action, and to assess the reasoning of thinkers, and that such explanations and assessments cannot turn on facts that are inaccessible to the subject, he relies on a broadly individualist account of the subject’s state of mind.11 He relies on the subject’s conception of the situation (which we saw is also something that Brown trades on). This does not mean that we should abandon Stalnaker’s anti-individualism or that we should opt for structured propositions. The aim of my discussion was rather to show how the subject’s cognitive perspective needs to be explained in the context of Stalnaker’s anti-individualism and in accordance with some of

56  Anti-individualism & Cognitive Perspective his key claims that are plausible in their own right. I want to note that belief-contents and their constituents that I aim to individuate in this book are not incompatible with Stalnaker’s anti-individualism. They are just those internal cognitive items that determine which intensions represent the subject’s actual cognitive state. It would take us far afield to even try to outline what implications this outcome might have for the broader debate over anti-individualism. But it is safe to say that much depends on the status of a suppressed premise in those cases in which the subject’s reasoning is best represented as involving such a premise.12

3.5 The Neo-Fregean View Unlike Stalnaker, Evans rightly observes that the subject who mistakes two different objects for a single one needs to entertain the claim ‘This1 ≠ this2’ in order to entertain thoughts with different contents, not just have no potential difficulty understanding it. In discussing a case in which the subject mistakes a previously seen object for an object that she is presently seeing, Evans urges that she needs to be able to understand the claim ‘This1 ≠ this2’ in order to have a thought at all, but if she does not also entertain this claim, she will think a single mixed thought (Evans 1982, 296–297). Only when she recovers from the confusion will she think two thought constituents with different contents. For Evans, these are the original senses that the subject has formed by observing each of the objects separately, but this is of no consequence for the present discussion. What is of consequence is that this way of individuating senses makes them explanatory of the subject’s cognitive perspective. This is to say that the subject will deploy a single sense as a matter of taking it that a single object is in play. If she entertains the claim ‘This1 ≠ this2’, she will deploy two different senses as a matter of taking herself to be (possibly) representing two different objects. Consequently, a difference in thought-content is always transparent, as is their sameness, in accordance with the Transparency Thesis. On the other hand, when the information that causally derives from two different objects is combined in an indecomposable way such that the subject is unable to understand the claim ‘This1 ≠ this2’, there is not just one object which the sense identifies such that no coherent thought-content can be found (Evans 1982, 297).13 The subject rather suffers an illusion of thought and, in attempting an inference along the lines of (II), she does not make a simple invalid inference, but rather suffers an illusion of making an inference. This enables Evans to shun the foregoing asymmetries. He is not committed to the view that the subject will grasp two senses that are different as a result of observing two different ships, yet fail to realize a priori that they are different due to her failure to realize that the ships are different. And she will not fail

Anti-individualism & Cognitive Perspective  57 to grasp a priori that an invalid inference is invalid. For, a case in which the subject thinks two thoughts with different contents without realizing a priori that they have different contents does not arise. Evans is wrong in drawing this conclusion, Millikan argues – a point that she substantiates with reference to a variant case, which is in her view for Evans again a case of “no thought at all”: …imagine a person losing track and apparently, but wrongly, perceiving the same squirrel eating first six and then seven more Brazil nuts. The result is an indelible memory of the squirrel who ate thirteen whole Brazil nuts at a sitting. Surely this is not a case of no thought at all, but a case where two contents have been blended, a case where thought is equivocal. (Millikan 2000, 153) While it is plausible to claim that the subject is thinking here a thought of a relevant kind, one may wonder whether Millikan is right in claiming that this thought is equivocal, i.e. that there is a single sense with two objects of reference, or a single sense, playing the mode of presentation role that rests on a false presupposition of identity which fails to refer and does not refer to two objects simultaneously, as Recanati claims. For reasons stated above, I favour Recanati’s proposal. If it is correct, it should follow that in the case in which Evans takes thought to be decomposable sense is not mixed as he holds but rather fails to refer since it also rests on a false presupposition of identity. Alternatively, if resting on a false presupposition of identity does not prevent sense from being mixed, the same should hold for the indecomposable case. What is of significance here is not which of these two alternatives is correct but the fact that either way the subject is credited with thinking a (single) thought of the appropriate kind rather than no thought at all, which makes thought fulfil its key role of capturing the subject’s point of view which is one of taking it that a single object is in play.14 To be sure, on either alternative, the subject has reached the conclusion that there is a squirrel who ate 13 whole Brazil nuts by presupposing the identity of the two squirrels rather than asserting it, which as we have seen is the mark of the sameness of sense. The subject’s cognitive perspective is one of (unreflectively) taking it (for granted) that a single squirrel is in play, which makes her reason and act in conformity with it. It is simply that in the present kind of case, unlike one in which a single squirrel would be in play, the identity presupposition is false. If we take it with Stalnaker that this identity presupposition is an extra premise in the subject’s reasoning that is neither articulated nor subpersonally encoded, then it is this premise that is false. These considerations suggest that in the indecomposable case, the subject is thinking a single thought. As a result, this is not a case in which

58  Anti-individualism & Cognitive Perspective the subject thinks two thoughts with different contents without realizing a priori that they have different contents. Such a case does not arise. The claim that the subject is suffering an illusion of thought here also blocks this outcome, but we have seen that this claim is implausible. In discussing cases in which the perceptual information that the subject is receiving causally derives from two different objects but where the majority of this information does not derive from just one of the objects, such as when one watches someone manipulating two matchboxes even though it looks as if there is just one matchbox, Campbell remarks ‘that there is no saying which one you are identifying demonstratively’ (Campbell 2002, 98–99). But, once again, this is because the subject is thinking an equivocal thought or a thought that fails to refer (as the case may be), not because she is suffering an illusion of thought. And, the same, mutatis mutandis, holds for synchronic cases such as the foregoing two-ship case. I conclude that while Fregean anti-individualism is compatible with the Transparency Thesis, its constituent “illusion of thought” claim makes it riddled with difficulties.15

3.6 Transparency Again In relation to Evans and Campbell, I have argued that the cognitive perspective of a subject who has confused two different objects for a single one cannot be plausibly explained by taking her to be suffering an illusion of thought, and in relation to Stalnaker, I have argued that the subject’s perspective cannot be plausibly explained by taking her to be thinking two thoughts with different contents. My suggestion was that the subject is rather thinking a single thought that is either equivocal or fails to refer (as the case may be). If this is right, then the foregoing claim that in presuming identity, the subject deploys the same sense and in so doing grasps a priori that it is the same sense, holds whether or not the information that she is receiving causally derives from a single object or from two different ones. Conversely, in taking herself to be (possibly) representing two different objects, the subject deploys two different senses by the lights of (CD) and in so doing grasps a priori that they differ, whether or not the information that she is receiving causally derives from a single object or from two different ones. This is to say that a rational subject can always grasp a priori that two thoughts with different contents have different contents just as she can always grasp a priori that two thoughts with the same content have the same content, i.e. that the Transparency Thesis is met. So, we do not face Grasp of Sameness, not Difference. To be sure, the subject need not deploy a single sense every time she takes it that a single object is in play. In such a case, her cognitive perspective is a matter not just of how she takes the world to be but also of

Anti-individualism & Cognitive Perspective  59 the senses that she deploys. Consider again a case in which the subject takes the two ship-ends that she sees on the left and the right of a large building, which in fact belong to the same ship, to belong to two different ships. Suppose she learns in the meantime that she was wrong and that the two ship-ends belong to the same ship. At this point, she will start to take it that the ship as presented to her via its bow and via its stern is a single ship. But she may continue to deploy the two senses in thinking about it (at least for some time). In keeping with the requirement that senses need to be transparent in order to serve to explain the subject’s cognitive perspective, in continuing to deploy the two senses, she will grasp a priori that they are different by associating them with the two different perspectives that she initially had on the given ship.16 True, her realization that the same ship is in play will be reflected in her behaviour. Before she came to realize this, she could have believed that the ship whose bow she sees is sailing to Mombasa and wish to embark on it while at the same time believing that the ship whose stern it sees is not sailing to Mombasa and not wish to embark on it. In finding out that one ship is in play, she will adjust her behaviour. She will, for example, come to believe that it does not matter for her goals at hand whether she embarks the given ship from its bow entrance or from its stern entrance. That this change of attitude does affect the fact that the original senses that she still deploys are transparent to her is shown by comparing this case with the case featuring ‘sentences complete in every respect’, to use Frege’s turn of phrase. The subject who knows that the evening star is the morning star will not take the sentences ‘The evening star is a body illuminated by the Sun’ and ‘The morning star is a body illuminated by the Sun’ to have the same sense (meaning). She will think different thoughts here by the lights of (CD) because it is in principle rationally possible for her to take them to be different heavenly bodies. (To this I shall return in Chapter 6.) What about Grasp of Validity, not Invalidity? To answer this question, consider again: This ship is an aircraft carrier This ship is British Therefore, there is a British aircraft carrier in the harbour If there is one ship in play, as in (I), the argument is valid, and the subject grasps a priori that it is valid. If there are two ships in play, as in (II), and the premises are about two different ships (and true) while the subject takes them to be about a single ship, the argument is invalid although the subject fails to grasp a priori that it is invalid. Of course, the argument is valid if we take it with Stalnaker to involve a false suppressed premise that this ship is that one. (As noted, the fact that the concept or mode of presentation fails to refer here does not stand in the way of the fact that

60  Anti-individualism & Cognitive Perspective the two original premises in (II) are about two distinct ships and both true, as Stalnaker claims. They are about distinct ships due to the fact that the subject is having an appropriate contextual referential relation with each of the ships in the process of producing them.) This way Grasp of Validity, not Invalidity, is avoided, and with it, Brown’s charge that the Fregean conception of rationality is ill-motivated. Brown says: ‘What could motivate the view that a rational subject can grasp a priori simple instances of some logical properties but not others?’ (Brown 2004, 227). But, even if we do not help ourselves to Stalnaker’s suppressed premise and take the given argument to be invalid, we can still give a sensible answer to this question by saying that the reason why a rational subject can grasp a priori simple instances of some logical properties but not others lies in her ignorance of the world and not of the thoughts that she is thinking. For, we do not face Grasp of Sameness, not Difference. In so doing, we comply with our guiding assumption (stated in Section 3.2) that if thought-content is to serve to explain the subject’s cognitive perspective, this needs to be done in the same kind of way whether it is the same or different. I conclude that perception-based demonstrative thought-contents need to conform with the Transparency Thesis if they should serve to explain the subject’s cognitive perspective, and that equivocal thought-contents or thought-contents that fail to refer (as the case may be) meet this requirement. In conformity with the Transparency Thesis, the thoughtcontents of the present account are not subject to Grasp of Sameness, not Difference, which I show to be independent from Grasp of Validity, not Invalidity.

3.7 Senses The foregoing discussion makes it clear that we need to individuate thought-contents in terms of (CD) in order to explain the subject’s cognitive perspective in terms of them. (CD) does justice to Stalnaker’s claim that we ascribe thought in order to explain action, and to assess the reasoning of thinkers, and that such explanations and assessments cannot turn on facts that are inaccessible to the subject (Stalnaker 2008, 114). Hence, in the case in which the subject who believes of the same ship (or of two different ships) that This ship (pointing to the ship’s bow) is not This ship (pointing to the ship’s stern), the subject’s cognitive perspective is explained by claiming that she is thinking of the same ship (or of two different ships) in two different ways, i.e. via two different senses, because she takes it that two different ships are in play. Similarly, in the case in which the subject believes of the same ship that This ship (pointing to the ship’s bow) is This ship (pointing to the ship’s stern), her cognitive perspective is explained in terms of deploying a single sense or concept because she presupposes that the same ship is in play.17 I have

Anti-individualism & Cognitive Perspective  61 argued that her cognitive perspective should also be explained in terms of deploying the same sense or concept when she believes of two different ships that This ship (pointing to the ship’s bow) is This ship (pointing to the ship’s stern) due to presupposing that the same ship is in play. It is just that in the latter case, the sense or concept that she deploys is either equivocal or fails to refer (as the case may be). This position is the result of overcoming the weaknesses of the antiindividualist views that I have discussed. To recall, we have seen that in adopting the Opacity Thesis makes perception-based demonstrative thought-contents as Brown individuates them unable to explain the subject’s cognitive perspective. Still, Brown wants to explain this perspective and in so doing she trades on the kind of thought-content that conforms with (CD) and the Fregean conception of rationality. Unlike Brown, Stalnaker adopts the Transparency Thesis. But, as argued, when it comes to those cases in which the subject confuses two different objects for a single one, Stalnaker fails to show that difference in thought-contents is transparent so as to serve to explain her cognitive perspective. Unlike Stalnaker’s anti-individualism, Fregean anti-individualism is compatible with the Transparency Thesis (as well as with (CD)). However, it implausibly holds that the subject who (in the relevant cases) confuses two different objects for a single one is thinking no thought at all and thus gives the wrong account of the subject’s cognitive perspective. Returning to the claim that a case in which the subject thinks two thoughts with different contents without realizing a priori that they have different contents does not arise, I want to bring out a couple of points that follow from the foregoing discussion. If the subject mistakes the previously seen object a for an object b that she is presently seeing and presupposes that a = b, she thinks of a and b via the same sense whether she is able to understand the claim ‘This1 ≠ this2’ or not and in so doing grasps a priori that it is the same sense. And, the same, mutatis mutandis, holds for synchronic cases. Alternatively, if the subject does not presuppose that a = b, but rather judges that a = b, she deploys two different senses. The fact that she judges that a = b makes it the case that she can a priori distinguish the two senses, so we do not face Grasp of Sameness, not Difference. Similarly, if the subject comes to entertain the claim ‘This1 ≠ this2’ as a result of recovering from confusion, she can a priori distinguish the two senses. Once the subject decides that a = b, she may end up thinking about both objects via the same sense (as a result of merging the two senses18), and in so doing grasps a priori that it is the same sense. Alternatively, she will deploy the two original senses (at least for some time) as in the foregoing case in which the subject learns that the two ship-ends that she took to belong to two different ships in fact belong to the same ship and in so doing grasp a priori that the senses are different.

62  Anti-individualism & Cognitive Perspective The case in which the subject falsely presupposes that a = b is similar to the case in which the subject re-identifies the same object by immediately recognizing that the seen object is the remembered object which involves a single sense that rests upon a true presupposition of identity. The case in which the subject judges that a = b is similar to the case in which the subject re-identifies the same object by way of judging that the seen object is the remembered object which involves two different senses.19 While such an identity judgement involving recognition of the same object is true, such a judgement is false when two different objects are in play. The subject’s falsely judging that, say, This ship (the ship that she is perceiving) is that ship (the ship that she remembers) involves, respectively, two different senses that she can distinguish a priori in spite of taking herself to be thinking of a single ship via these senses. She thinks that the two senses co-refer. Since she is wrong, her identity judgement is false. And, the same, mutatis mutandis, holds for synchronic cases. As a result, a case in which the subject fails to distinguish a priori two thoughts with different contents (i.e. senses) does not arise.

3.8 Conclusion In relation to perception-based demonstrative thoughts, I have argued that difference in thought-contents needs to be transparent just as their sameness if thought-content is to serve to explain the subject’s cognitive perspective, which is believed to be one of its key roles (as far as the views that I have been discussing are concerned). In establishing this conclusion, I have shown that their intentions notwithstanding, representative anti-individualist views fail to give us a theory of thought- content that provides such an explanation, whether they hold that sameness of thought-contents can be opaque, as can their difference, or that difference in thought-contents is always transparent, as is their sameness.

4

Cognitive Dynamics, Belief Retention, and Cognitive Significance

4.1 Introduction The discussion in this chapter shifts from perception-based demonstrative thoughts to thoughts that we recognize by those linguistic expressions that contain temporal indexicals such as ‘today’ and ‘yesterday’, the problem of cognitive dynamics, and the question of what standard adjustment the subject needs to make in the expression of her indexical thought in order to express the same belief over time. We shall see shortly that retaining a belief about a certain day that the subject has formed on that very day requires an internal continuity in her belief, as Frege suggests on the evidence of his foregoing remarks about indexicals. Frege also claims that to think of a certain day as the same from day d through to d + 1 is to think of it via the same sense as the bearer of cognitive significance. However, Kaplan has argued that Fregean thoughts cannot be the bearers of cognitive significance since one can think the same Fregean thought from one occasion to the next without realizing this, thus linking the issue of cognitive significance to that of belief retention as does Frege. Kaplan comes up with his own version of the Fregean strategy for accounting for belief retention that does not face this kind of a problem. But he finds it deficient because it leads us to implausibly deny that one who is lost in time retains the beliefs one had before this occurred. In this chapter, I take issue with Kaplan although in conformity with his plausible demands about belief retention and argue that a situation does not arise in which one can fail to realize that one is thinking the same thought from one occasion to the next. This lends further support to the view that thoughts are the bearers of cognitive significance as well as explanatory of belief retention. It is established that cognitive significance is of a piece with the internal continuity of the subject’s belief, making it clear that we need to deal with it in accounting for belief retention.

64  Belief Retention

4.2 The Problem of Cognitive Dynamics Kaplan formulates the problem of cognitive dynamics in the following way: [W]hat does it mean to say of an individual who at one time sincerely asserted a sentence containing indexicals that at some later time he has (or has not) changed his mind with respect to his assertion? What sentence or sentences must be he willing to assert at the later time? (Kaplan 1989a, 537–538, n. 64) In other words, a theory of cognitive dynamics is concerned with the question of what standard adjustment the subject needs to make in the expression of her indexical thought in order to express the same belief over time. To recall, Frege’s suggestion is: (RC) If someone wants to say today what he expressed yesterday using the word ‘today’, he will replace this word with ‘yesterday’. Although the thought is the same its verbal expression must be different in order that the change of sense which would otherwise be effected by the differing times of utterance may be cancelled out. (Frege 1918/1977, 10) Frege claims here that to retain a belief about a certain day d, say, that d is beautiful, from one day to the next amounts to thinking the same thought. He also claims that the subject needs to be disposed to make contextually appropriate adjustments in the expression of this thought in order to retain it. I will argue that the first claim is right, i.e. that retaining such a belief amounts to thinking a thought of a broadly Fregean kind. To have and retain such a thought, it is required that it is about d as its sole causal source and that the subject represents d as the same day from one day to the next. But, if this is right, then Frege’s claim that the subject can retain the thought with which she began just in case she is disposed to adopt a contextually appropriate indexical expression cannot be right. The goal of the ensuing discussion is to resolve this issue.

4.3 The Fregean View According to (RC), retaining on d + 1 a belief that the subject has formed on day d about d that it was beautiful amounts to her thinking about d via the same sense over time. This sense can be expressed by different (appropriately aligned) indexicals, i.e. ‘today’ and ‘yesterday’ in Frege’s example. This is supported by the following considerations. Suppose

Belief Retention  65 that on d, the subject assents to ‘Today is sunny’. Then, remembering (the weather on) d and keeping track of it by taking all the relevant information to have d as its single source, on d + 1, she assents to ‘Yesterday was warm’. From this, she is entitled to validly infer ‘Something is (was) sunny and warm’. In making this inference, she is presuming rather than asserting the identity of the day designated by the utterances of the two indexicals, which makes the sense the same, as pointed out above in relation to spatio-temporal objects (i.e. ships). Her inference does not rest on a suppressed identity premise, and the attempt to reduce the identity presumption to an implicit identity judgement, and have the subject think of d via two different senses, embarks us on an infinite regress. For, we would also need to make sure that the day referred to in the suppressed premise is the same as the day referred to in the explicit premises, and we would need further suppressed premises to secure this which embarks us on a regress. The fact that the subject needs to resort to her memory when a day that she is thinking about is no longer directly accessible to her seems to suggest that a perception-based thought has given way to a memorybased one, hence that belief retention involves a succession of different thoughts and senses. The belief that the subject expresses by ‘Today is sunny’ is based on her being acquainted with d, while the belief that she expresses on d + 1 by ‘Yesterday was warm’ is based on her memory of it. This seems to entail that one and the same thought cannot be perception-based at one time and memory-based at another. Yet, she is entitled to validly infer ‘Something is (was) sunny and warm’ from ‘Today is sunny’ and ‘Yesterday was warm’. In making such an inference, she trades on identity. This then seems to suggest that trading on the identity of d need not involve the same sense. To this issue, I shall return in Chapter 7 with a focus on perceived objects. Here, it is the view that to retain at a later time, a belief about d that one formed on d that can be expressed by ‘Today is beautiful’ amounts to retaining the same sense (thought) over time that is called into question.1 For, as noted, Kaplan thinks that a Fregean thought cannot account for belief retention since it fails to be the bearer of cognitive significance: If one says ‘Today is beautiful’ on Tuesday and ‘Yesterday was beautiful’ on Wednesday, one expresses the same thought according to [RC]. Yet one can clearly lose track of the days and not realize one is expressing the same thought. It seems then that thoughts are not appropriate bearers of cognitive significance. (Kaplan 1989a, 501, n. 26) There is also a problem with Frege’s strategy of adjusting verbal expression in order for the same belief to be expressed as the context changes that Kaplan points out. Adopting this strategy prevents us from

66  Belief Retention acknowledging that one who has lost track of the days has retained a belief with which one began such as a belief that one expressed by ‘Today is beautiful’. In acknowledging this, and in defending the view that retaining a belief amounts to retaining a single thought, I will argue that Kaplan is wrong in claiming that one can fail to realize that one is expressing the same thought.

4.4 Kaplan’s View Kaplan comes up with a conceptual apparatus that avoids the noted problem he finds with (RC). As a referentialist, he holds that a propositional content expressed by an utterance of an indexical sentence such as ‘Today is beautiful’ is Russellian in that it consists solely of the day being referred to and the property of being beautiful. But this kind of content cannot be the bearer of cognitive significance for the same reason that in Kaplan’s view, Fregean thoughts cannot be the bearers of cognitive significance: one can lose track of the days and not realize that one is expressing the same content twice. Content does not serve to rationalize the subject’s cognitive perspective. Kaplan illustrates this by citing a case featuring personal pronouns: I first think, “His pants are on fire”. I later realize “I am he” and thus come back to think “My pants are on fire”. Still later, I decide that I was wrong in thinking “I am he” and conclude “His pants were on fire”. If in fact, I am he, have I retained my belief that my pants are on fire simply because I believe the same content, though under a different character?… [T]his does not capture my sense of retaining a belief…. (Kaplan 1989a, 537, n. 64) Yet, Kaplan considers seriously Frege’s strategy of adjusting verbal expression in order to enable the same belief to be expressed as context changes, making his account of belief retention part of his account of cognitive significance – in a manner similar to Frege on the evidence of (RC). In terms of Kaplan’s own apparatus (1989a), this amounts to adjusting character as the kind of meaning of an indexical expression which is set by linguistic conventions and which both determines the Russellian content of the expression (what is said) in every context and is the bearer of cognitive significance of the expression. The character presents different objects (of thought) to different persons and to the same person at different times (Kaplan 1989a, 530). It accounts for the common element that different belief states have irrespective of context. 2 The character of ‘today’ (the present day) presents the day the subject is thinking of in one particular way, while the character of ‘yesterday’ (the previous day) presents it in a different particular way

Belief Retention  67 which makes character play the role of a mode (manner) of presentation. This makes the same Russellian content, e.g. that day d is beautiful, expressible by (and graspable under) the characters of ‘Today is beautiful’ and ‘Yesterday was beautiful’ as the context changes, which suggested to Kaplan a Frege-inspired strategy of accounting for belief-retention hinted at in (RC). But Kaplan finds this strategy falls short of supplying us with some obvious standard adjustment to make to the character as illustrated by the case of Rip Van Winkle. Rip Van Winkle, who slept for 20 years, woke up thinking that he had slept for just one day (see Kaplan 1989a, 537–538, n. 64). Suppose, to gloss this case along the lines of Perry (1997, 36), that on day d, before falling asleep, Rip never forms any explicit belief other than ‘Today is beautiful’. When he awakes 20 years later, the belief is updated, given Rip’s view of how the context has changed, to ‘Yesterday was beautiful’. What is left of the original belief, according to the suggested account, is a belief about the day before he woke up – the day that the character of ‘yesterday’ determines (in the context of utterance) and exhibits as its referent via a cognitively significant mode of presentation. This Kaplan finds strange. For Rip seems to remember (that) d (was a beautiful day). Kaplan’s foregoing comment on (RC) as well as on the “pants on fire” case makes it clear that, for a belief to be retained, it is not enough that the subject thinks of the same day or person and believes the same Russellian content from one occasion to the next. An internal continuity in the subject’s belief is also required that, Kaplan observes, eludes being accounted for in terms of some obvious standard adjustment in the character. In relation to this, more recently, Kaplan has claimed: I may be tracking the passing days very carefully. I became acquainted with the day yesterday and expressed that way of being acquainted in my use ‘today’. Assuming no recognition or tracking failures and no memory failures, I should be able to continue to have the day in mind in the same way today, though of course I will refer to it as ‘yesterday’. (Kaplan 2012, 138) In agreement with these claims, I will argue that the internal continuity in the subject’s belief needs to be accounted for in terms of Frege’s claim contained in (RC): that representing a certain day as the same from one occasion to the next is to think of it via the same sense, i.e. under the same mode of presentation, which makes the thought of which it is a constituent the bearer of cognitive significance. Cognitive significance is of a piece with the internal continuity of the subject’s belief, and we need to deal with it in accounting for belief retention – even if it turns out that it is not part of the business of semantics, as Wettstein (1986, 2004) holds.

68  Belief Retention

4.5 Linguistic Meaning and Thinking about Days Acknowledging that Rip has retained the belief about d (that it was a beautiful day) conflicts with the following consequence of the Fregeinspired strategy for accounting for belief retention: (P) One’s sincere acceptance of an utterance of an appropriate temporal indexical commits one to thinking of the day it designates in virtue of its linguistic meaning, whichever day it is that one intends to think about. (P) is part of Kaplan’s view that indexicals shape the subject’s ways of thinking of Russellian content in virtue of their meanings. On the evidence of (RC), Frege also seems to embrace (P) – in spite of the fact that in his view, the meanings of ‘today’ and ‘yesterday’ do not shape the subject’s ways of thinking of the days, i.e. that ‘today’ and ‘yesterday’ are just means of expressing and retaining the same thought as the context changes. Evans takes Frege’s line further and claims, contra Kaplan, that Rip has not retained the belief with which he began. He argues that after waking up, Rip is instead suffering an illusion of having retained such a belief because he has lost track of time (similarly to the case featuring perceived objects dealt with above). He says: I see no more strangeness in the idea that a man who loses track of time cannot retain the beliefs than in the idea that a man who loses track of an object cannot retain the beliefs about it with which he began. If one has in fact lost track of time without knowing it, then one could think that one had retained one’s belief when one has not. (Evans 1985, 311, n. 21) I take it, though, that Kaplan is right here in suggesting that Rip has retained the belief with which he began.3 For otherwise, any similar mis-tracking of days would need to deprive one of the belief with which one began. Suppose that on Tuesday, I form an indexical belief that I express by ‘Today is beautiful’. Then, being unaware that midnight has passed, I utter the same sentence again with the intention of re- expressing that very belief. But this surely does not deprive me of continuing to believe what I believed before midnight; nor does it make me think of Wednesday that it is beautiful. Although I have mis-tracked Tuesday in this way, I have kept track of it in a sense that enables me to retain the given belief: Tuesday is the sole causal source of the belief I am having, and I am representing it as the same day from one occasion to the next. And the same should hold for Rip. The relevant belief that he is having after waking up has d as its sole causal source and he is representing d as the same as the day that his original belief was about. This makes

Belief Retention  69 (P) gratuitous. In spite of the fact that, upon waking up, Rip will reach for ‘Yesterday was beautiful’, given his view of how the context has changed, the belief he intends to express is not aimed at the day before he woke up – which in the given context, ‘yesterday’ denotes in virtue of its linguistic meaning. For, that day is neither the causal source of his belief nor is it the day upon which he is cognitively fixed, given that he has slept through it. Yet, notwithstanding Rip’s intentions, ‘yesterday’ can be taken to denote here that very day in virtue of the rules governing its use (and he is saying something false about it if it was not beautiful, if we take its denotation to affect truth-conditional content). Accordingly, as will become clear, thoughts and senses are the bearers of cognitive significance, and they allow us to account for belief retention even though the referent of at least some utterances of indexicals – e.g. Rip’s ‘yesterday’ – is not determined by the sense that the utterer attaches to them. This is a departure from Frege, who took sense to play both these roles. To be sure, Donnellan’s (1966) distinction between denotation and reference which Donnellan applied to definite descriptions can be readily applied in the present context. (Recall also Perry’s claim that indexicals are like definite descriptions in that they denote and like names in that they refer.) A definite description denotes whatever (if anything) satisfies it uniquely, which need not be the same individual that it refers to as a matter of the speaker’s standing in an appropriate contextual relation with the referent. In Donnellan’s own words: … denoting and referring should not be confused. If one tried to maintain that they are the same notion, one result would be that a speaker might be referring to something without knowing it. If someone said, for example, in 1960 before he had any idea that Mr. Goldwater would be the Republican nominee in 1964, “The Republican candidate for president in 1964 will be a conservative” (perhaps on the basis of an analysis of the views of party leaders) the definite description here would denote Mr. Goldwater. But would we wish to say that the speaker had referred to, mentioned, or talked about Mr. Goldwater? I feel these terms would be out of place. (Donnellan 1966, 293) Returning to the Rip Van Winkle case, this is to say that the denotation of Rip’s use of ‘yesterday’ is the day before he woke up while the reference, the day that Rip refers to and believes it to be beautiful, is d. This resembles the case concerning perception-based demonstratives whose referent is, in Kaplan’s view (in relation to Donnellan), properly determined by the speaker’s intention to refer to a perceived object (Kaplan 1989b, 582–584) in that it is Rip’s intention to refer to d rather than the character the previous day that is critical in fixing the reference of ‘yesterday’. Similarly, Recanati claims that singular reference is fixed by the

70  Belief Retention subject’s acquaintance relations since the reference need not satisfy the singular predicate (character). For, in the present case, the day referred to d – the day that Rip’s belief is both aimed at and derived from on the basis of his formerly being acquainted with it – does not satisfy the singular predicate (character) ‘previous day’. In line with this, Recanati claims that after waking up, Rip thinks of d as ‘yesterday’. Then he goes on to say: Now which day does he refer to when he says ‘Yesterday was [beautiful]’? This is a tricky question. In virtue of the reference rule associated with the word ‘yesterday’, it seems that it must refer to the day preceding the day of utterance. But that is not the day Rip is referring to and characterizing as [beautiful]. Rip refers to the day he remembers, namely d, of which he wrongly believes that it is the previous day… Donnellan’s distinction between denotation and reference comes in handy here: we can say that the denotation of Rip’s use of ‘yesterday’ is the day before his utterance, while the reference – what Rip himself refers to and describes as [beautiful] – is d, the last waking day he remembers. (Recanati 2012, 179–180)4 On the other hand, when the subject keeps track of the days along the lines of (RC), there is no mismatch between the day that ‘today’ and ‘yesterday’ denote in virtue of the rules governing their use and the day the subject is thinking about. In this kind of case, it is inappropriate, indeed irrational, for the subject not to be disposed to accept an indexical that, in the context of utterance, denotes the day upon which she is cognitively fixed, which creates a certain dependence between them. But this dependence does not entail (P). Short of relying on (P), we can acknowledge the fact that Rip has retained the belief with which he began. I will argue that retaining the belief about day d with which one began consists in one’s keeping track of d in compliance with the two aforementioned conditions: that it stems from d as its sole causal source, and that one is representing d as the same day from one occasion to the next. Such a belief amounts to a thought of a broadly Fregean kind that persists through a change of context. It is unconstrained by (P), and it serves to explain and rationalize the subject’s cognitive perspective. In acknowledging that the linguistic meanings of indexicals such as ‘today’ and ‘yesterday’ do not shape our ways of thinking of days, i.e. that ‘today’ and ‘yesterday’ are just means of expressing the same thought as the context changes, I take Frege’s point of departure but do not follow it up with (P), which stands in the way of acknowledging that Rip has retained the belief with which he began. In trying to substantiate this view of belief retention, my first task will be to show that such a thought is the bearer of cognitive significance,

Belief Retention  71 contrary to what Kaplan would have us believe. I shall then discuss its other features.

4.6 Thoughts and Transparency As remarked, when commenting on (RC), Kaplan claims that one can lose track of the days and not realize that one is expressing the same thought, which in his view shows that thoughts are not appropriate bearers of cognitive significance. But this is not right. For the situation does not arise in which the subject can mistake one thought for two. For unlike Rip, a subject from Kaplan’s foregoing scenario will lose track of the days, and this will deprive her of thinking the thought with which she began, such that the possibility of mistaking one thought for two different ones does not arise. If she acquired a belief on Tuesday that she expressed by ‘Today is beautiful’ and then lost track of it such that she does not realize that the sentence that she assents to the following day – ‘Yesterday was beautiful’ is about the same day, the latter thought will not be the same as the former. For, she is not representing the given Tuesday as the same day from one occasion to the next, and it will be informative for her to be told that both thought episodes concern the same day. Similarly, when the subject does not have problems keeping track of a single day and keeps thinking a single thought from one day to the next, she will not fail to re-identify the thought that she is thinking. The fact that her belief has the required internal continuity, as well as that it has this particular day as its sole causal source, ensures this. This is in line with (CD) which states that it is impossible for a rational thinker to take conflicting epistemic attitudes to a single thought.5 To make this clear, consider a parallel case concerning perceptionbased demonstrative thoughts about objects. Campbell has claimed that to think of a perceived object via the same sense from one occasion to the next amounts to keeping track of it (Campbell 1987; see also Evans 1982, 196). Keeping track of an object is done unreflectively and may involve collecting new information about it. It is part and parcel of the unreflective use of perception-based demonstratives that the subject takes the perceptual information that she is receiving about an object from moment to moment to have a single object as its causal source, i.e. that she represents it as the same. Raising the question about the identity of the perceived object is a reflective project which requires that the subject be thinking about the character of her perceptions. It takes us away from the ground-floor, unreflective use of perception-based demonstratives (Campbell 1987, 284). This includes cognitive skills that belong to a sub-personal level that are non-conceptual (Campbell 1987, 283). If the subject actually does make what Campbell calls a division in the perceptual information that she is receiving, i.e. if she suspects that one

72  Belief Retention object has been substituted for another, so that she can raise the question whether it is the same thing that is in question, then we have two modes of presentation (Campbell 1987, 284–285), i.e.: We can acknowledge this, while respecting the Fregean principle [CD] that it is impossible for a rational thinker to simultaneously take conflicting attitudes to a single thought. For ‘rationally taking conflicting attitudes’ here will require the thinker to make a division in his input information: and once he has actually made the division, we will indeed have different senses. (Campbell 1987, 285)6 A similar point can be made about tracking days, although this involves different cognitive skills from those that are involved in tracking objects. In keeping track of an object, the subject maintains or can maintain a (more or less) continuous experiential link with the object. (I shall discuss this in Chapter 6.) By contrast, keeping track of a certain day cannot by its nature involve this kind of link once the day has passed (and nor can the subject re-encounter the day), so it needs to rely on the subject’s memory of the day. (To this I shall return below.) Yet the two kinds of case are similar in the following important respect. As long as the subject represents an object or day as the same from one occasion to the next, she will both think of it under the same mode of presentation and be aware of her doing so. Representing an object or day as the same makes her ‘know immediately that it is the same thing that is in question’ (Campbell 1987, 285), so that she knows that the mode of presentation is the same, i.e. she cannot fail to re-identify it from one occasion to the next. Hence, she cannot fail to re-identify the thought with which she began – the thought containing the given mode of presentation as its constituent. In so doing, she is not thinking second-order thoughts about her first-order beliefs, and nor is she comparing her later thought tokens with earlier ones that are no longer available to be thought about. Thinking the same thought from one occasion to the next is just a matter of the subject’s taking herself to be thinking of the same day throughout. This makes the sameness in thoughts transparent to the subject. If two thoughts are the same (i.e. have the same content), the subject can realize a priori that they are the same (see Boghossian 2004, 2011, Brown 2004, chap. 6). This is in line with (CD) according to which it is impossible for a rational thinker to take conflicting epistemic attitudes to a single thought. (I shall deal with this in Chapter 6.) Yet this does not entail the Cartesian view that the subject is omniscient about contents of her own mind (as we have seen above in relation to Stalnaker’s claim that concepts are creatures of darkness). Since the subject’s realizing that she is thinking the same thought from one occasion to the next (inter alia) amounts to representing the day that she is thinking about as the same

Belief Retention  73 from one occasion to the next, her doing so does not involve her having (direct) access to the contents of her own mind. In fact, the issue of whether the subject is having (direct) access to the contents of her own mind does not arise.7 That a rational thinker cannot take conflicting epistemic attitudes to a single thought makes it the case that thoughts are the bearers of cognitive significance, i.e. it makes them explanatory of the subject’s actions and behaviour. They serve to rationalize the subject’s cognitive perspective. Suppose that on Tuesday, I acquire a belief that I express by ‘Today is beautiful’. I keep track of the given Tuesday through to Wednesday and update the belief I am holding by ‘Yesterday was beautiful’. This meets the internal continuity requirement by making the thought that I think on Tuesday and through into Wednesday the same. This gives rise to one kind of action. But if I were to lose track of the given Tuesday and fail to realize that the sentence that I assent to the following day ‘Yesterday was beautiful’ is about the same day, the latter thought would not be the same as the one with which I began, as argued. This gives rise to another kind of action.8

4.7 Sourceless Beliefs Keeping track of a day and thinking of it under the same mode of presentation from one occasion to the next may involve thoughts whose existence does not require that the subject has epistemic contact with it, i.e. that she be acquainted with it. On day d − 1, the subject may form a belief about d which she expresses by ‘Tomorrow will be beautiful’ although she has not been in epistemic contact with d. This is what Perry (1997) aptly calls a sourceless belief, i.e. a belief typically held about future days that cannot be part of the cause of our beliefs. When d comes, and turns out to be beautiful, the subject may try to update the belief she has on the basis of her coming into epistemic contact with d by accepting ‘Today is beautiful’. Her sourceless belief about d’s being beautiful has given way to a belief that involves her having epistemic contact with d. Or alternatively, she may form a belief about d that she expresses by ‘Tomorrow will be beautiful’ on d − 1, after which she sleeps through d, and then wakes up on d + 1, knowing which day it is. Still convinced about the weather on d – which she was in no position to experience – she may update her belief by accepting ‘Yesterday was beautiful’, in which case one sourceless belief gives way to another. Unlike those beliefs that have their causal source in days with which the subject has been in epistemic contact, sourceless beliefs about days seem to line up with (P) – the claim that one’s sincere acceptance of an utterance of an appropriate temporal indexical commits one to thinking of the day it designates in virtue of its linguistic meaning, whichever day it is that one intends to think about. For the subject who is having a

74  Belief Retention sourceless belief – such as one that she may express by ‘Tomorrow will be beautiful’ – will be guided by character, i.e. the linguistic meaning, of ‘tomorrow’, in thinking about the forthcoming day with which she has not yet been in epistemic contact. I have argued that (P) needs to be abandoned so that we can acknowledge the fact that Rip has retained the belief with which he began. But, in view of the possibility of having a sourceless belief of the aforementioned kind, it may seem that this fact can be acknowledged without abandoning (P). It may be claimed that when Rip awakes, he has two different beliefs that line up, respectively, with two different characters, which, in turn, makes characters play the modes (manners) of presentation role. One of these beliefs lines up with ‘Yesterday was beautiful’, which is about the day before he woke up – let us call it day d* – while the other lines up with ‘That day was beautiful’ which is about d. For otherwise, we cannot account for the fact that, in the given context, he is to assent to all of these statements: 1 2 3

Yesterday was beautiful That day was beautiful That day = Yesterday

For after waking up, Rip is disposed to assent to the false (3); and given (3), the true (2) is inter-derivable with (1). Rip might think that (1) and (2) are about the same day, but that would be another mistake. Appearances aside, Rip does not hold both beliefs. For, this would require that in addition to the belief about d that it was a beautiful day that Rip is credited with he also has a belief about d*, which, due to his sleeping through d*, cannot be a belief that has its causal source in d*.9 But it will not do to credit him with a sourceless belief about d* either. For, in the described circumstances, it is not a sourceless belief about d* that Rip aims at in accepting ‘Yesterday was beautiful’, but rather he aims at the belief that has d as its source. Even if such a sourceless belief were available to Rip, it would be irrelevant in the present context in which Rip is updating his belief about d that has its causal source in d. So, it is not the case that Rip believes the false (3), i.e. that d = d*. He rather takes ‘that day’ and ‘yesterday’ to refer to d in the given context in virtue of their linguistic meanings and similarly for (1). Rip believes falsely that ‘yesterday’ refers to d as a result of having a false (first-order) belief that d has the property of being the previous day. Hence, (P) does not hold.10

4.8 The Content of Thought In believing falsely that ‘yesterday’ refers to d, Rip does in a sense think of d as the previous day, making it the case that the linguistic meaning

Belief Retention  75 of ‘yesterday’ plays a role in shaping his way of thinking of d. To put this into the right perspective, recall the claim that thinking of d under the same mode of presentation from one occasion to the next requires (inter alia) that the subject represents d as the same which has been argued to be the case with Rip. Recall also that, unlike keeping track of an object, keeping track of a certain day cannot by its nature involve a continuous experiential link with the day once it has passed (nor can the subject re-encounter it), so it needs to rely on the subject’s memory of the day. In view of this, to think of d under the same mode of presentation from one occasion to the next and to retain the belief that it is (was) beautiful amounts to updating it by way of attributing to d various features, not all of which need to be true of it. The subject will typically associate with d predicates that (she thinks) are true of it. The property of being the previous day which amounts to the character of ‘yesterday’ belongs here too (although in Rip’s case it does not pertain to d). Sometimes, this may be the key property the subject associates with the day she thinks about (as when sleeping through it), in which case she may be said to be thinking of it under the character of ‘yesterday’. (We have seen above that a similar thing occurs with the indexical ‘tomorrow’.) In those cases, in which the subject needs to resort to ‘that day’ in the role of a memory-based demonstrative that does not have a fixed character, in order to update her belief, her mode of presentation of d needs to include various supplementary features she associates with d. Since these features supplement the character, it may be claimed that they are part of it which is alright as long as we take character so conceived to be part of the thought expressed, i.e. of the propositional content, and not something extraneous to it in the way character is to Russellian content within Kaplan’s framework. All the features the subject takes to pertain to d and d alone (together with the fact that d is the sole causal source of the subject’s belief) constitute a single mode of presentation of d. Sometimes, it is exhausted by the (fixed) character of the indexical used (as, e.g., in the ‘tomorrow’ case), sometimes it is not. In other words, in continuing to think of d on d + 1 that it was beautiful, the subject will replace the property of being the present day with the property of being the previous day. What makes it the same belief that has been retained throughout is the presence of the internal continuity of the subject’s belief. The properties of being the present day and the property of being the previous day are just amongst the properties that the subject attaches to and detaches from d as time goes by without affecting the identity of the mode of presentation under which she is thinking of d from d to d + 1. In this respect, this case is not different from a case in which in thinking of d, the subject first attaches to it the property of being sunny and detaches is from d as time goes by due to a change in weather and replaces it with some other property. Similarly, in thinking of d, the subject may collect new information about d and

76  Belief Retention think of it under the same mode of presentation throughout. To allow factors such as these to affect the identity of the mode of presentation of d would either make us let each such variation affect the identity of the mode of presentation or create a cut-off line, neither of which is plausible. The former option embraces an unnecessary proliferation of modes of presentation (akin to the one that will be discussed in Chapter 6), while the latter is unlikely to provide us with a principled reason for showing why it is that some variations in properties will affect the identity of the mode of presentation of d and others will not.11 Modes of presentation in the present account play the role that Fregean modes of presentations are intended to play. To illustrate this, suppose again that the subject assents on d to ‘Today is sunny’. Then, remembering (the weather on) d and keeping track of it by taking all the relevant information to have d as its single source, on a later occasion, she assents to ‘That day was warm’. From this, she is entitled to validly infer ‘Something is (was) sunny and warm’. In making such an inference, she presumes the identity of the day designated by the utterances of the two indexicals. It would be useless to add this as a premise. The notion of sense (as the mode of presentation) is designed for this very purpose: one gets to presume rather than assert identity when sense is the same, and the inference may trade directly upon the fact of co-reference of two singular terms or of two different utterances of the same or different indexicals (see, e.g., Campbell 1987, 2002, chap. 5). This ties in with Campbell’s foregoing claim that if one does succeed in keeping track of an object over time, then one must know immediately that it is the same thing that is in question (Campbell 1987, 285). The broadly Fregean account of belief retention that has emerged is a natural consequence of accommodating the plausible view embraced by both Frege and Kaplan that belief retention requires an internal continuity in the subject’s belief. In contrast with this, Lewis’s view outlined in the Introduction is incapable of accounting for belief retention since on that view, there is no internal continuity in the subject’s belief (in spite of Lewis’s individuating beliefs by cognitive significance in that beliefs for him serve to explain and predict behaviour). We saw that according to Lewis, the subject who keeps track of Tuesday through to Wednesday and on Tuesday accepts ‘Today is Tuesday’, and on Wednesday accepts ‘Yesterday was Tuesday’ does not hold the same belief throughout. Lewis rather claims that to have an appropriate indexical belief is to self-ascribe the corresponding property (Lewis 1979, 518). To have the Tuesday-belief is to self-ascribe the property of being temporally located on Tuesday, while to have the Wednesday-belief is to self-ascribe the property of being temporarily located on Wednesday. But, if so, we are committed to holding rather implausibly that the subject cannot continue to believe on Wednesday what she believed on Tuesday in assenting to ‘Today is beautiful’.12

5

Beliefs and Characters

5.1 Introduction In continuing my discussion of the Rip Van Winkle case, in this chapter, I discuss Perry’s attempt to account for the belief state that Rip is in when he awakes and adopts and believes it that, as it were, ‘yesterday was beautiful’. In agreeing with Kaplan that upon waking up, Rip has retained the belief with which he began, Perry suggests that Rip has retained this belief under various backup doxastic characters that are, in his view, analogous to but not derived from Kaplanian linguistic characters. I will argue that, similarly to Kaplan’s linguistic characters, these characters are unsuitable for the role that they are supposed to play. Following this, I will propose an alternative account of belief retention according to which our ways of thinking of days are not governed by the linguistic meanings or characters of the indexical expressions that we associate with them. This will help us meet our objective, showing that the subject’s cognitive perspective is captured by the content that is believed and asserted as I take it to be.

5.2 Doxastic Characters In agreeing with Kaplan’s suggestion that upon waking up, Rip has just one belief about d which is memory-based, Perry sticks to the character/ content framework and suggests that Rip has retained this belief under various backup doxastic characters. He postulates a level of meaning for indexical beliefs that is analogous to, but not derived from, that of character for sentences – characters that are quite independent of language (Perry 1997, 21). Since characters correspond to narrow psychological states, human behaviour should be explained and predicted in terms of characters rather than in terms of (Russellian) contents. It is characters, not contents, that line up with causal roles of beliefs, where the causal role of a state is taken to be ‘various combinations of factors that bring the state about, and the various combinations of factors it brings about in turn’ (Perry 1997, 20). The same proposition/content can correspond to different causal roles/characters and the other way around. Following

78  Beliefs and Characters Kaplan’s foregoing “pants on fire” example (Kaplan 1989a, 537–538, n. 64), suppose that Kaplan accepts sincerely the character ‘My[self] pants are on fire’, where the bracketed material identifies the underlying cognitive role involved, based on relations that an object can have to a given episode of thought or a particular belief. Suppose next that he sees himself in a mirror but believes himself to be seeing someone else with his pants on fire and, pointing at the man in the mirror, accepts ‘His [the man I am looking at] pants are on fire’. The different characters that are involved here account for two different kinds of action that Kaplan is disposed to take. In the former case, he would, for example, try to take his pants off and dispose of them or to douse himself with water, while in the latter case, he would try to douse with water “the man in the mirror”. However, the content corresponding to these two characters is the same – that David Kaplan’s pants are on fire – and it is psychologically inert. It ‘does not contain the mode of presentation or cognitive fix that the believer has on the individual’ (Perry 1980/1993, 71). Alternatively, the character ‘My[self] pants are on fire’ accounts for ‘the common nature that different beliefs with different contents belonging to different people at different times might have, and by virtue of which these different situations might instantiate the same psychology of content’ (Perry 1997, 23). The notion of doxastic character is a successor of Perry’s notion of role considered above. Perry used to call these roles senses or modes of presentation but insisted that they differed from Frege’s senses in that Frege’s senses, being ideal procedures for determining reference (Perry 1977, 4), are meant to be complete, which is not the case with roles and characters. (To recall, Kaplan claims that while in Frege’s theory, a given manner of presentation presents the same object to all mankind, character will in general present different objects of thought to different persons or even different objects of thought to the same person at different times (Kaplan 1989a, 530).) And, as Fregean thought is ‘first and foremost that for which the question of truth arises’, it lines up with content rather than role/character. In keeping sense and thought distinct, Perry spoke of entertaining a sense and of apprehending a thought (Perry 1977/1993, 22).1 Turning to the ‘today’/‘yesterday’ case as our present concern, the difference between Frege and Perry is this. When, in a normal course of circumstances, the subject thinks of a certain day d, first as today and then on d + 1 as yesterday, that it is (was), say, a beautiful day, she is, on Frege’s view, thinking the same sense/thought throughout. What is being adjusted is just the means of expressing this sense/thought. By contrast, ‘today’ and ‘yesterday’ are for Perry more than this. They are different (doxastic) characters/senses and so are ‘today [the day of this thought] is beautiful’ and ‘Yesterday [the day before the day of this thought] was beautiful’, although the Russellian content/thought corresponding to them in the given context – that d is beautiful – is the same. The

Beliefs and Characters  79 difference in the (doxastic) characters that the subject accepts is, in turn, meant to account for the difference in her belief states and behaviour. This might seem right, for, she will apparently think of d in two different ways: first, for example, as the present day and then as the previous day. Her thinking about d as the present day can make her pragmatically attached to it because she can make changes in what happens on it, whereas her thinking about d as the previous day cannot make her pragmatically attached to it because she can no longer make any changes in what happens on it. Since she is no longer in a position to actively explore d, she is not epistemically attached (see Perry 1997, 33). However, considering some non-standard cases will reveal that it is not characters that account for indexical belief states and behaviour but rather senses or modes of presentation of a broadly Fregean kind which are wholly independent of characters, and that Perry is wrong in conflating the two. Like Kaplan, Perry (1997) holds that when Rip awakes, he retains the belief that he acquired on d that d was beautiful. When he awakes, he updates it according to his view of how the context has changed. Rip’s view about the change of context is mistaken, and the new character ‘Yesterday [the day before the day of this thought] was beautiful’ is not a way of believing the original content, where, to recall, the bracketed material identifies the underlying cognitive role involved, based on relations that an object can have to a given episode of thought or a particular belief. But that is no reason to say that Rip has lost his original belief, claims Perry, since he retains it under various other backup characters. When Rip believes, towards evening, as it grows dark, of d, that it was beautiful, he has certain memories of that day. So, the character ‘That day [the day I remember] is or was beautiful’ is, in Perry’s view, available to sustain Rip’s belief when the attempt at the updating goes awry. Even if these memories fade, there is, claims Perry, the character ‘That day [the day this belief was acquired] is or was beautiful’ (see Perry 1997, 36). What Perry suggests is that Rip has retained and updated his belief about d, that d was beautiful, although the character that he accepts is, according to our (ex hypothesi correct) view of how the context has changed, not suitable for him as a means of expressing it. In the given context, this character does not denote d but rather d* – the day before he woke up, and is thereby not a way of believing the original content. On the other hand, the suggested backup characters are, according to our (correct) view of how the context has changed, suitable for Rip as a means for expressing his belief of d, that d was beautiful, because, in the given context, they are ways of believing the original content whereby they also denote d. This suggests that it is too strong to claim that the bracketed material constitutes Rip’s way of thinking of d. For the belief that Rip is having throughout does not seem to involve a detailed specification of d as the

80  Beliefs and Characters day of this thought, the day I remember or the day this belief was acquired, and the like. Rip’s thinking of d, that it was beautiful, in the situation in which he is when he awakes surely does not seem to rely on such a detailed specification of d. The natural alternative is that the belief that Rip formed on d is a plain indexical belief to the effect that it is beautiful that he later updates on the basis of his memory, while the bracketed material serves only as a pointer to us in theorizing about Rip. But, if this is correct, this belief is not governed by characters as it is supposed to. For, all Perry has shown is that such characters are available to Rip according to our, not his, view of how the context has changed. Characters are epistemically transparent in that one can tell them apart. As a result, Rip will know that ‘Yesterday [the day before the day of this thought] was beautiful’ and ‘That day [the day I remember] was beautiful’ are different characters (types). If so, it is unclear why he is to accept such a backup character (which is awkward for him at that) if he does not know that the character ‘Yesterday [the day before the day of this thought] was beautiful’ in the given context is not suitable for him as a means of expressing his belief about d, that d was beautiful. On the other hand, granting Perry that Rip has retained his belief about d under various backup characters as well as under the mistakenly accepted ‘Yesterday [the day before the day of this thought] was beautiful’ means that the identity of the belief state that he is in is not affected by whichever of these characters he accepts. This suggests that (doxastic) characters (as Perry conceives of them) are not ways of believing the content. Rip has retained and updated his belief about d in spite of his acceptance of the unsuitable character, and it involves a particular way of believing the content which will be the same if he is to accept any of the suitable characters. According to Rip’s view of how the context has changed, we have a situation in which he chooses (between) two or more equally suitable characters (types) as (between) two or more different means of updating the same belief. The underlying cognitive roles that accompany these characters presuppose that having memories of d is necessary for Rip’s retaining his belief about d. This is so even if his memories have faded, and the only backup character left for him to hang on to is ‘That day [the day this belief was acquired] was beautiful’. For, Rip is updating his belief about d because he remembers it. Otherwise, there would be nothing to update. This is to say that it is remembering d that furnishes Rip’s disposition towards d, which underlies his belief about it. Characters are there only to express an independently generated belief (state). In general, retaining and updating a belief about a certain day require having a memory of it. This, in turn, makes it possible for us to be in the same indexical belief state about that day under different characters (types). It is then not true that the identity of such belief states that explain and predict human behaviour is tied to the identity of characters.

Beliefs and Characters  81 On finding out that he slept for 20 years, Rip will need to update his belief yet again as a result of taking a different temporal perspective towards d which requires additional identifying information. (We shall see in Chapter 7 that this does not affect the sameness of sense over time.) Rip will typically express this belief by means of the character ‘That day [the day I remember] was beautiful’, no longer finding the character ‘Yesterday [the day before the day of this thought] was beautiful’ suitable. For, according to his (this time correct) view of how the context has changed, it does not denote d, the day that his belief state is about. As a necessary condition for Rip’s being in the belief state that he is in about d when he awakes, his remembering d makes d, together with the time at which the belief is held, the relevant contextual feature on which his acceptance of the appropriate character is based.2 On d, when Rip acquired his belief, the object of his belief (i.e. d) and the time at which this belief was held (d again) coincided, making the character ‘Today [the day of this thought] is beautiful’ appropriate for Rip to express his belief by. When he woke up and learned that he slept for 20 years, the object of his belief (i.e. d) and the time at which his belief was now held were (taken to be) wide apart, making the character ‘That day [the day I remember] was beautiful’ appropriate for this purpose, and similarly for other characters. The time at which the belief is held determines which character needs to be used to match d in respect of its position in the sequence of days. This is to say that the cognitive fix that Rip is having on d directs the character’s semantic fix, i.e. that the character is playing only the semantic, not the cognitive role. As a rule that takes us from context to reference, i.e. as a function from context to content, a character – claims Perry (1997, 15) in relation to Kaplan’s linguistic characters – takes a context as input (as its argument) and provides a reference (content) as output (as its value). In terms of the result we have reached, this means that d and the time at which Rip’s belief about it is held are taken to be the input (argument) of the appropriately chosen character, whereas d is also its output (value). By virtue of his cognitive fix on d and of the time of holding his belief, Rip will be led to accept the character that will have a semantic fix on it. Perry’s view that behaviour is to be explained in terms of character (Perry’s sense, causal role) which takes us from an occasion of utterance (or its doxastic analogue) to a certain object involves the view that the sense of an indexical is a criterion for recognizing, i.e. determining, the referent, adjusted to the indexical case. The cognitive and the semantic roles that character is taken to be playing are seen as tied together such that there is a pre-established harmony between them (Perry 1980/1993, 76–77). There is indeed a pre-established harmony between the cognitive and the semantic features of one’s indexical belief, but not, as we have seen, a harmony that has the cognitive feature lining up with the character.

82  Beliefs and Characters

5.3 The Mismatch between Characters and Belief States In discussing Perry’s treatment of the Rip Van Winkle case, I have shown that identity of belief state that accounts for cognitive significance is not tied to the identity of character. Perry is concerned with doxastic characters but the same considerations apply to Kaplan’s linguistic characters as do the following remarks. 5.3.1 The Existence of Character Is Not Tied to the Existence of Belief State When Rip awakes, he accepts the character ‘Yesterday [the day before the day of this thought] was beautiful’ which, in the given context denotes d*, the day before he woke up. Yet, in Perry’s view, he cannot be in a belief state that is to be accounted for in terms of this character because he does not remember this day. Since he slept through it, there is nothing to remember. Similarly, suppose somebody else, say, Brown, on d also acquires a belief about d that it is beautiful and on d accepts the character ‘Today [the day of this thought] is beautiful’. During the following night, she suffers a loss of memory and no longer remembers d, although she takes herself to remember it and on d + 1 she sincerely accepts the character ‘Yesterday [the day before the day of this thought] was beautiful’. As a result of this, a memory-based belief state that this character is meant to account for is not available to her. The character’s playing a semantic role it is playing is thereby not sufficient for it to play a cognitive role as it is supposed to. Accepting the character ‘Yesterday [the day before the day of this thought] was beautiful’ does not make either Rip or Brown be in an indexical belief state that they would be in if they were to remember respectively d* and d. Hence, the existence (acceptance) of character is not tied to the existence of a corresponding belief state. 5.3.2 The Existence of Belief State Is Not Tied to the Existence of Character Consider the case in which the subject is unsure if midnight has passed, although it has passed. Suppose, to follow an example from Branquinho (1999), that at 11:58 p.m., on a certain day, make it d again, a certain man, Jones, accepts ‘Today [the day of this thought] is a nice day’. Three minutes later, Jones is unsure if midnight has passed and has no means of checking it and neither accepts nor rejects ‘Yesterday [the day before the day of this thought] was beautiful’ at 00:01 a.m., on day d + 1 (as well as ‘That day [the day I remember] was beautiful’, ‘Today [the day of this thought] is or was beautiful’, or ‘This day [the day that I am experiencing] is or was beautiful’) because he has lost track of time. For Evans,

Beliefs and Characters  83 this means that Jones is prevented from retaining at 00:01 a.m. on d + 1 the belief he held on 11:58 p.m. on d. Still, there seems to be no doubt that Jones continues to believe what he believed three minutes earlier, even if he is ex hypothesi not able to think of the day in question as, for example, yesterday or the day which immediately precedes today, or to identify it by means of knowledge of his own position in time. Evans’s requirement that keeping track of time in this sense is necessary for retaining appropriate beliefs is thus shown to be unjustified. For, Jones has retained his belief at 00:01 a.m. on d + 1, on the basis of remembering d. He has kept track of time in the relevant sense in that the belief state he is in is aimed at d and d alone. One can thus have a belief about a certain day without accepting any character at all, making it the case that the existence of belief is not tied to the existence (acceptance) of character. The same result can be established by pointing out that the subject who mistakenly takes herself not to remember a day that she in fact remembers will not be inclined to accept a character that would be of any significance. Neither needs the subject to be aware that she is in a memory-based belief state the whole time that she is in it. As a result, while being unaware of this she will not be inclined to accept any of the relevant characters. She can perhaps be said to be still disposed to accept such a character, but then she will be equally disposed to accept any characters that are suitable for expressing the belief state that she is in which would just reinstate the foregoing claim that the identity of her belief state is not tied to the identity of character. 5.3.3 Identity of Character Is Not Tied to Identity of Belief State We have seen that, on Perry’s view, the same character accounts for the common nature that different belief states have, and by virtue of which different situations instantiate the same psychology of content, irrespective of content and context. The character ‘Today [the day of this thought]’ will always instantiate the same psychology of content, no matter when, where, and by whom it is accepted, and which day is being thought about. With this in mind, recall Perry’s own example concerning linguistic characters which can be equally spelt out in terms of doxastic characters. Smith, whose watch is an hour fast, accepts ‘Today is my husband’s birthday’. But just before 11, she realizes she got it wrong. It is March 1 and not March 2. She glances at her watch, at 11, and it shows midnight – she thinks to herself ‘so today is my husband’s birthday’. Smith’s respective assenting to and dissenting from two consecutive utterances of ‘Today is my husband’s birthday’ shows her as being in two different belief states although the character is in both cases the same. It is then not true that the identity of character is tied to

84  Beliefs and Characters the identity of belief states that explain and predict human behaviour. (Perry’s subsequent views on similar issues suggest that he would handle the difference in Smith’s belief states in terms of a difference in reflexive contents of utterances and beliefs discussed and dismissed in Chapter 2.)

5.4 The Retention of Thought The preceding discussion suggests the following view of belief retention. ‘Today is beautiful’, uttered on Tuesday, and ‘Yesterday was beautiful’, uttered on Wednesday’, will convey the same cognitive value, i.e. involve the same mode of presentation of d, just in case the subject takes d as the same from Tuesday through to Wednesday. This ensures that the subject thinks of d under the same mode of presentation from one occasion to the next which is, in turn, required for the belief with which she began to be retained (which has been claimed to be the case with Rip). In so doing, she will associate with d a cluster of features and properties she takes d to possess. These may include the properties of being the present day or of being the previous day. Although they respectively amount to the characters of ‘today’ and ‘yesterday’, these properties do not shape the subject’s way of thinking of d in the way in which they are supposed to on Kaplan’s and Perry’s views that were examined above. They are just properties the subject associates with d alongside other properties and features. The case of the subject such as Jones from the foregoing example who is unsure whether midnight has passed and who refrains from accepting either ‘today’ or ‘yesterday’ in order to express his belief about d shows that such a property need not even be involved in the mode of presentation of d, while Rip’s picking up ‘yesterday’ to express and update his belief about d upon waking up shows that such a property, in this case the property of being the previous day, can mistakenly be taken to pertain to d. True, in those cases in which everything goes smoothly and the subject keeps track of d through to d + 1, the character of ‘yesterday’ will typically play the key role in updating the subject’s belief about d on d + 1. But as time goes by, in updating her belief, the subject will need to resort to ‘that day’ in the role of a memory-based demonstrative, which does not have a fixed character such that her mode of presentation of d needs to include various supplementary features she associates with d – making it the case that all of them together shape her mode of presentation of d. In thinking of d under the same mode of presentation from one occasion to the next, the subject cannot mistake it for two different modes of presentation. As a result, it once again follows that how the subject takes the world to be matters when it comes to individuating her modes of presentation and explaining her reasoning and actions. If the subject makes an error of fact and fails to realize that ‘Today is beautiful’ uttered on Tuesday and ‘Yesterday was beautiful’ uttered on Wednesday are both

Beliefs and Characters  85 about d, she will think of d under two different modes of presentation (as argued in Chapter 4). The fact that the subject will think of d under the same mode of presentation as long as (inter alia) she represents d as the same ensures that the identity of a mode of presentation is not affected by whether the subject assigns new features to d in the process of thinking about it over time, or withdraws those she once took it to have, or by whether these features really pertain to d. Similarly, a particular mode of presentation of d need not involve properties that d satisfies (or is believed to satisfy) uniquely. To illustrate this, Perry’s Smith example comes in handy once again. Smith’s respective assenting to and dissenting from two consecutive utterances of ‘Today is my husband’s birthday’ shows her as thinking of d under two different modes of presentation in spite of the fact that each of these modes of presentation contains as its constituent the property of being my [Smith’s] husband’s birthday as well as that of being the present day (although not at the same time). This would be so even if these modes of presentation were, respectively, to incorporate all and only the same features and properties. This way the tricky issues do not arise, such as whether every variation in the features and properties the subject takes d to possess should change the identity of the mode of presentation of d or whether all those features that the subject takes d to possess ought to be true of d in order for her to have a mode of presentation of d. What matters is that the fluctuation of these features does not make the object (content) of belief shifty and unstable as long as it is individuated in terms of the subject’s representing d as the same. It may be objected that this precludes modes of presentation from playing a cognitive role. It seems that what one believes about d is going to play a part in explaining one’s actions such that if mode of presentation stays constant regardless of what one believes, then mode of presentation cannot explain action. I will discuss this issue in Chapter 7 and show that this kind of objection is not warranted. For now, I want to note that what I have just claimed is in line with the fact that an influx of information about d does not change the mode of presentation of d from one occasion to the next in the process of the subject’s keeping track of it, as pointed out in relation to keeping track of a perceived object (see, e.g., Evans 1982, 174–176,194 ff; Campbell 1987). We shall deal with keeping track of objects in subsequent chapters. Here, I want to note that it is not plausible to claim that an influx of information about d, or a passage of time, changes the subject’s mode of presentation of d. As Sainsbury put it (in relation to perceived objects): ‘That would make it impossible to sustain a thought over time or over informational enrichment’ (Sainsbury 2002, 157). This way of individuating modes of presentation also makes them interpersonally stable. Suppose two people are having a communication about d on d + 1 both of which know which day it is but have spent d

86  Beliefs and Characters in different places being involved in different activities such that a good many properties that they respectively associate with d do not match. On d + 1, the speaker utters ‘Yesterday was beautiful’ and the hearer who, say, spent d indoors disagrees. No matter how vast the disagreement between them is as to what properties d is taken to have and no matter by which features each of them remembers d, they are both thinking of d under the same mode of presentation as long as they both represent d as the same day from d through to d + 1. That this is the virtue of the view that I am advancing will become clear in the light of the argument of the next chapter concerning the individuation of thoughts. As noted, in those cases in which the subject keeps track of d from one day to the next, the character of ‘yesterday’ will typically play the key role in updating her belief about d on d + 1. The belief that d is beautiful that the subject has formed on d on the basis of her direct encounter with d and expressed by means of ‘Today is beautiful’ is updated on the basis of her memory of d and expressed by means of ‘Yesterday was beautiful’. This suggests that only certain characters can be employed to account for the change of context as one (who is in this respect unlike Rip) keeps thinking about d from d to d + 1. Only certain (indexical) expressions are fit for playing the part in updating the subject’s belief and not just any expressions that are about d: ‘today’ can be replaced with ‘yesterday’ but not with ‘December 17, 2019’ even if d is December 17, 2019. It is also the case that when the subject thinks of d on d + 1, accepting ‘Yesterday was beautiful’, her mode of presentation of d is going to be the same whether her belief about weather on d is true or false. She might have been mistaken about the weather from the start or misremembered it later on; yet in both these cases, as well as in the case of having a true belief about it, her mode of presentation of d will remain the same. It is the same mode of presentation of d that constitutes her belief state on d and d + 1, making it (in addition to the same character used to express it) the key common element shared by the true and the false belief. This is to say that the subject’s mode of presentation of d may involve all, and only the same properties whether the subject’s belief about weather on d is true or false. On d + 1, the subject who has tracked d through to d + 1 may attach the same properties to d whether her belief is true or false because its truth and falsity depend on factors that are external to her conception of the situation. If, on the other hand, the subject changes her mind about weather on d and, at one point of time, starts believing that she was wrong in thinking that it was beautiful, she will drop the property of being beautiful from the body of information forming her mode of presentation of d. If so, does this make a split in her mode of presentation of d? Not at all! As noted above, the subject’s mode of presentation of d will remain the same as long as her relevant belief has d as its sole causal source and she represents d as the same day from d through to d + 1. This ensures that the identity of the mode of

Beliefs and Characters  87 presentation is not affected by whether the subject is attaching new features to d in the process of thinking about it over time or withdrawing those she once took it to have, or whether these features really pertain to d. If on d + 1, the subject’s belief about (weather on) d draws upon her memory of d, and d alone, and is aimed at d, and d alone, as its causal source, as the same day that her original belief was about, her mode of presentation of d will be the same no matter which properties the subject ascribes to d. Memory-based and perception-based belief states of the subject are states explanatory of action and behaviour. Rip’s behaviour in the foregoing scenario is the same no matter which of the suggested characters he accepts. As a result of his confusion, Rip will be prone to behaving in the same way whether he accepts ‘Yesterday [the day before the day of this thought] or ‘That day [the day I remember] was beautiful’. Our discussion of Perry’s treatment of the Rip Van Winkle case has shown that identity of belief state is not tied to identity of character. Similarly, the discussed case featuring Smith who is in two different belief states under two different tokens of the same character shows that it is these belief states and not the character that account for her behaviour. Acquiring such a belief concerning d requires having at some point cognitive contact (encounter) with it.

5.5 Modes of Presentation and Meaning The ‘today’/‘yesterday’ case should make it clear that the subject’s mode of presentation of day d that is captured, respectively, on the two consecutive days by means of ‘Today’ and ‘Yesterday’ does not line up with the linguistic meaning of these indexicals. The fact that the subject may think of d under the same mode of presentation while adopting different temporal indexicals to account for the change of context once again shows that the identity of a belief state that she is in is not tied to the identity of character, i.e. linguistic meaning. The subject may first express her belief on d by means of ‘Today is beautiful’ and on the following day by means of ‘Yesterday was beautiful’ while thinking of it under the same mode of presentation. Think also of Jones, who is unsure if midnight has passed and refrains from accepting any indexicals (characters) while continuing to believe what he believed a couple of seconds ago before he became unsure if midnight has passed, which shows that the existence of the subject’s relevant belief state is not tied to the existence of character, i.e. meaning. In spite of this, he will continue to think of d under the same mode of presentation. Similar remarks apply to demonstratively identified spatio-temporal objects. As long as I represent an object from one occasion to the next as the same, I will think of it under the same mode of presentation. While, for example, I keep track of a man I am looking at, I will be in the

88  Beliefs and Characters same perception-based belief state over a period of time, although I may successively express my belief by means of expressions having different linguistic meanings such as ‘This one is a spy’ and ‘That person is a spy’. I may even use a descriptive sentence such as ‘The man over there is a spy’ or ‘The individual I spoke to a little while ago is a spy’, in which the contained descriptions function as perception-based demonstratives. My choice of expression is not governed by any rules of linguistic meaning which is to say that linguistic meaning is not constitutive of the belief state I am in. The featured descriptive sentences serve to express this dynamic thought although their linguistic meanings are not dynamic.3 Alternatively, I can choose any of these expressions to express my belief state at a given time. Furthermore, no particular expression needs to spring to mind in the process of being in such a belief state. So, we cannot go by linguistic meaning in individuating the belief state that I am in. We have also seen that there is a reverse problem with tying cognitive significance to meaning/character, consisting in the fact that the subject can be in two different cognitively significant mental states while accepting, respectively, utterances with the same linguistic meaning/character. The Smith case attests to this. Smith’s respective assenting to and dissenting from two consecutive utterances of ‘Today is my husband’s birthday’ shows that she is in two different belief states while the linguistic meaning remains the same. The same happens when one disbelieves that ‘That1 = that2’ in the case in which two utterances of the same demonstrative are co-referential. That the subject’s mental states are different here follows from the fact that she has made a division in the input information that she is receiving which will lead her to rationally accept ‘That1 is F’ without accepting ‘That2 is F’, where F is a property which she, respectively, believes and disbelieves the given object to have. This verdict is the outcome of the aforementioned criterion of difference for thoughts (CD) which is derived from Frege and applied in the present case to those thoughts that are associated with two different utterances of the same sentence type. As noted above, this criterion is accepted by the Fregeans as well as by Kaplan and Perry in terms of their own frameworks. It has to do with the fact that modes of presentation of objects thought about need to obey a cognitive constraint stating that a rational subject cannot at the same time believe and not believe of a certain object that it has a certain property while thinking of it under the same mode of presentation. For a Fregean, in Perry’s words, this criterion looks like this: If a person who understands the meaning of sentences S and S’ of language L can consistently accept S and not accept S’, then S and S’ must express different propositions [thoughts].

Beliefs and Characters  89 However, Perry adopts the following version of this criterion in order to enable reflexive contents of utterances and beliefs that we looked at in Chapter 2 to be cognitively significant in his sense: If there is some aspect of meaning by which an utterance u of S and an utterance u’ of S’ differ, so that a rational person who understands both S and S’ might accept u but not u’, then a fully adequate semantics should say what it is. (Perry 2001a/2012, 8–9) However, we saw that Perry’s notion of cognitive significance as a second-order reflexive content of utterances and beliefs is unfit for the role it is supposed to play. The discussion in this book shows that bearers of cognitive significance are to be individuated in terms of the former version of this criterion. Yet, as noted, Perry is right in claiming that it is part of the business of semantics to explain cognitive significance, i.e. that a correct semantic theory needs to provide us with an appropriate interface between what sentences mean (express) and how we use them to communicate beliefs in order to motivate and explain action. Since this criterion is an important tool in individuating the relevant mental states of the subject, or simply thoughts as I take them to be, we need to be clear about what it amounts to. This is the subject matter of the next chapter.

6

Slicing Thoughts

6.1 Introduction According to the aforementioned criterion of difference for thoughts derived from Frege (CD), two thoughts are different if it is at the same time possible for a rational subject to take conflicting epistemic attitudes towards them. But applying this criterion to perception-based demonstrative thoughts seems to slice thoughts too finely and lead to their proliferation which makes the criterion implausible. I will argue that such a proliferation of thoughts is blocked by transforming this criterion into a related one that is shown to be essential in individuating thoughts (or thought-contents) as they are conceived of in this book. This has to do with the fact that what makes demonstrative sense synchronically or diachronically the same is the subject’s unreflective taking for granted that the object that she is perceiving is a single object that does not require her to keep track of it in the sense supposed by Evans and Campbell. The virtue of the view of thought individuation that I propose is that in addition to blocking an intrapersonal proliferation of thoughts, it also blocks an interpersonal proliferation of thoughts in relevant cases.

6.2 Individuating Thoughts To be able to play the role that they are expected to play, Fregean senses need to obey a cognitive constraint which states that a rational subject cannot at the same time believe and not believe of a certain object that it has a certain property while thinking of it via the same sense, i.e. under the same mode of presentation. When stated in terms of thoughts containing relevant senses as their constituents, this constraint becomes the following criterion of difference for thoughts derived from Frege (see, e.g., Frege 1892/1980, 62) considered in Chapter 3: (CD) Two thoughts are different (have different contents) if it is at the same time possible for a rational subject to take conflicting epistemic attitudes towards them. Hence, if the thought is the same (has the same content), the subject cannot at the same time rationally take conflicting epistemic attitudes towards it.

Slicing Thoughts  91 That it is possible for a rational subject to take conflicting epistemic attitudes towards two thoughts does not entail that she needs to do this in order for them to be different. The subject who knows that the evening star is the morning star will not at the same time take conflicting epistemic attitudes towards the thoughts expressed respectively by the sentences ‘The evening star is a body illuminated by the Sun’ and ‘The morning star is a body illuminated by the Sun’. Yet she will think different thoughts here because it is in principle rationally possible for her to take them to be different heavenly bodies and hence to take such conflicting epistemic attitudes. Many philosophers, including neo-Fregeans such as Evans (1982, 18–22), McDowell (1986, 142, 2005, 49), and Peacocke (1986, 5), take this criterion (or some version of it) to be crucial for individuating relevant beliefs (and other attitudes) as featured in propositional attitude psychology.1 Others think that we should keep away from it because the possibilities used to individuate the objects of belief (and other attitudes) must be those that are relevant to naturalistic relations, and these must be genuine rather than epistemic possibilities (see Stalnaker 1984, 24–25). Whatever the merits of this objection, the issue I want to deal with here is that of this criterion’s plausibility when it comes to individuating perception-based demonstrative thoughts. For it threatens to generate a proliferation of thoughts and discredit itself not least because it is bound to deprive thoughts of their status as the bearers of cognitive significance. I will argue that such a proliferation of thoughts is blocked by transforming this criterion into a related one, which I show to be essential in individuating thoughts as they are conceived of here. This has to do with the fact that what makes demonstrative sense synchronically or diachronically the same is the subject’s unreflectively taking for granted that the object that she is perceiving is a single object, as we shall see below. Since here my focus is on perception-based demonstrative thoughts, consider once again the Enterprise case. Suppose a subject who is perceiving two different ends of the Enterprise, whose middle is obscured by a large building, takes them to belong to two different ships, and assents to a thought expressed by an utterance of ‘This ship is the Enterprise’ accompanied by a pointing to its bow and dissents from a thought expressed by another utterance of the same sentence which is accompanied by a pointing to its stern. Since by hypothesis she acts rationally, (CD) says that she is thinking two different thoughts rather than taking conflicting epistemic attitudes towards the same thought. This is so because she has made an error of identification. It is always possible for the subject to make an error of identification by taking two different parts of the same object to be parts of two different objects. 2 To be sure, (CD) is meant to hold only synchronically, since it is possible for a rational subject to take conflicting epistemic attitudes towards

92  Slicing Thoughts the same thought at different times. It is rationally possible for a subject observing a certain ship to assent at a particular time to a thought expressed by ‘This ship is the Enterprise’, then change her mind (for whatever reason) about it being the Enterprise and dissent from the same thought as expressed by a subsequent utterance of the same sentence. But, should such a change of epistemic attitude be the result of the subject’s mistaken belief that one ship has been replaced with a different one, she will be in two different mental states and think different thoughts by the lights of (CD). This is so because the subject may take herself to be perceiving two different objects from one occasion to the next even if she is receiving a unified stream of information from a single object. The subject may never move her eyes from a certain object yet think (falsely) that different objects keep materializing in the place at which she is looking (cf. Richard 1990, 227). What makes the thoughts different in these cases is the subject’s (mis) taking the object (ship) that she is (synchronically or diachronically) perceiving for two different ones. But, suppose the subject correctly takes herself to be perceiving a single ship and hence assents to two respective utterances of ‘This ship is the Enterprise’ that are used to refer to the same ship via its bow and stern, respectively, or on two consecutive occasions. It is still in principle rationally possible for her to take the given ship for two different ones (either synchronically or diachronically) as it is rationally possible for the subject who believes that the evening star is the same as the morning star to take them for two different heavenly bodies. We have seen that the latter possibility makes it the case that the sentences ‘The evening star is a body illuminated by the Sun’ and ‘The morning star is a body illuminated by the Sun’ express different thoughts even for a subject who takes them to be about the same heavenly body. This is because the expressions ‘The evening star’ and ‘The morning star’ will still have different senses for her as two different identifying conditions of Venus supplied by their linguistic meanings. Now, to treat the perceptual case that we are concerned with analogously is to say that because it is in principle rationally possible for the subject to take the given ship for two different ships by taking the perceived bow and stern to respectively belong to two different ships or by taking it that one ship has been replaced with another she is thinking of it via different senses whether she takes them to belong to the same ship or not. But, this does not ring true. The subject who takes the perceived bow and stern to be parts of the same ship does not grasp here senses that she can tell apart as a result of the difference in the linguistic meanings of the relevant expressions. The linguistic meaning of the two respective utterances of ‘This ship is the Enterprise’ that she assents to in the foregoing case is the same. So, there is nothing akin to linguistic meaning that would prevent her from thinking of the given ship via as many senses as the possibility of taking it for two or more different ones in principle allows. Not

Slicing Thoughts  93 only does this lead to an implausible proliferation of senses in the kind of case in which it seems intuitively plausible that only one sense is in play; the fact that the subject is unaware that she is thinking (synchronically or diachronically) of the same ship via different, let alone many different senses, makes senses and thoughts play no role in capturing the subject’s cognitive perspective and thoughts fail to be the bearers of cognitive significance.

6.3 Blocking the Intrapersonal Proliferation of Thoughts Hence, if perception-based demonstrative thoughts are to capture the subject’s cognitive perspective and be explanatory of her actions and behaviour along the lines of (CD), we need to find a way to block such a proliferation of senses. To show how this can be done, consider first a diachronic case in which the subject perceptually tracks a stationary or moving object, discussed by neo-Fregean philosophers such Evans (1982, chap. 6, 1985) and Campbell (1987). Their key claim is that to think of a perceived object via the same sense from one occasion to the next amounts to keeping track of it. Demonstrative sense is, on this view, individuated in terms of keeping track of an object (Evans 1982, 196; Campbell 1987, 285). It is allowed to persist over small interruptions in the subject’s perception such as when a perceived object disappears momentarily behind an obstacle (see Evans 1982, 176; Campbell 1987, 280 f).3 As we have seen, keeping track of an object is performed unreflectively (Campbell 1987, 284). It may involve collecting new information about the object and includes cognitive skills that belong to a sub-personal level and that are not conceptual (Campbell 1987, 283). It is part and parcel of the unreflective use of perception-based demonstratives that the subject does not make a division in the perceptual information she is receiving. The mere possibility of such a division does not, in Campbell’s view, show that one who is unreflectively tracking an object is actually in a position to ask whether this object (at t1) is this object (at t2). If the subject actually does make such a division, so that she can raise the question whether it is the same thing that is in question, then we have two senses. As noted in Chapter 4, Campbell claims that we can acknowledge this, while respecting the Fregean principle (CD), that it is impossible for a rational thinker to simultaneously take conflicting attitudes to a single thought. For ‘rationally taking conflicting attitudes’ here will require the thinker to make a division in his input information, and once he has actually made the division, we will indeed have different senses (Campbell 1987, 285). The point that Campbell is trying to make is that as long as the subject is tracking an object in the way suggested, she will think of it via the same sense and not be in a position to take it for two different objects, which blocks a proliferation of senses and thoughts. The mere possibility

94  Slicing Thoughts of making a division in the perceptual information she is receiving does not count in individuating the thought(s) she is thinking. She may in principle come to make or could have made as many such divisions as possible, but this surely does not make her think of the given object via as many senses as this possibility in principle allows (of whose shift she is unaware, at that). To make this clear, let us compare this case with the foregoing evening star/morning star case. In a clear sense, it is not rational for the subject who takes the evening star and the morning star for the same heavenly body to (simultaneously) take conflicting epistemic attitudes towards (the thoughts expressed by) the sentences ‘The evening star is a body illuminated by the Sun’ and ‘The morning star is a body illuminated by the Sun’. Still, it is in principle rationally possible for her to do so because she can come to take them (or could have taken them) for two different heavenly bodies. As a result, she will think two different thoughts for the reasons stated above. Similarly, as long as the subject takes the object that she is perceiving to be the same from one occasion to the next, in our example, a ship, in a clear sense, it is not rational for her to take conflicting epistemic attitudes towards the thoughts expressed by two or more consecutive utterances of ‘This ship is the Enterprise’ (provided she has not changed her mind about it being the Enterprise). Still, it is in principle rationally possible for her to do so insofar as it is in principle rationally possible for her to make a division in the perceptual information she is receiving. But, unlike with the morning star/evening star case, this does not make her think two different thoughts for reasons stated above (concerning the analogous synchronic case). So, what is of relevance when it comes to individuating perception-based thoughts along the lines of (CD) is not whether the subject can in principle rationally take conflicting epistemic attitudes towards her thoughts, but whether she is in a position to do so, i.e. whether she entertains doubts as to whether the object is the same. (To this I shall return in Chapter 7.) If so, (CD) needs to be transformed into: (CD’) Two thoughts are different (have different contents) if a rational subject is at the same time in a position to take conflicting epistemic attitudes towards them. If, by contrast, the subject is not in a position to take such conflicting epistemic attitudes and, so to speak, subconsciously presumes that her different perceptual inputs come from a single object, the thought she is thinking is the same according to Campbell (although (CD) and (CD’) as such do not entail this).4 That its constituent demonstrative sense is the same from one occasion to the next is shown by the fact that such a subject is entitled to make valid inferences from, say, ‘This cup is F’ and ‘This cup is G’, to ‘Something is F and G’, in a situation in which she is observing a certain cup. As remarked above, in making such an

Slicing Thoughts  95 inference, the subject must presume the identity of the object designated by each utterance of the demonstrative. It would be useless to add this as a premise (see Campbell 1987). One presumes rather than asserts identity when the sense is the same. This blocks a proliferation of thoughts as far as the kind of case that Campbell considers is concerned and makes thoughts explanatory of the subject’s actions and behaviour in making them line up with the subject’s cognitive perspective – which is a matter of how she takes the world to be in that she takes it that the same object (cup) is in play from one occasion to the next. Sometimes we take an object that we perceive over time for two (or more) different ones or are unsure if the object that we perceive now is the same as the object that we perceived a little while ago. To adapt an example from Evans (1982, 172) and Noonan (1986, 76–77), I may confront an array of tightly packed qualitatively indistinguishable pills and try to concentrate upon a single pill. Even if I keep looking, I may not be confident after some time that I am still looking at the same pill that I began with. But such situations are not very common which is to say that we are reliable but not infallible. We rather track objects unreflectively as part of the way in which we normally represent objects and their causal powers in that we unreflectively take a continuity of our perceptual experience of the object we perceive over time to be the result of its continuing causal power.

6.4 Stretching Thoughts So far, I have been concerned with those perception-based thoughts that are in line with Campbell’s (and Evans’s) claim that to think of an object via a single demonstrative sense over time is to keep track of it in the sense of having a continuous perceptual experience of it (allowing for small interruptions in the subject’s perception such as when the object momentarily disappears behind an obstacle). Yet, we regularly have perception-based tracking thoughts that go beyond this. Some of them involve only occasional perceptual links with their objects. If this was to make us think a succession of demonstrative thoughts rather than a single one, a proliferation of thoughts reinstates itself, diminishing the importance of Campbell’s account of demonstrative thoughts. The unreflective taking for granted of identity does not, however, require one to have a continuous perceptual experience of an object in order to be able to think of it via a single demonstrative sense from one occasion to the next. As Strawson has stressed in a somewhat different context, our methods, or criteria, of identification must allow for facts such as the following: that the field of our observation is limited, that we go to sleep, and that we move (Strawson 1959, 32). In view of this, what makes sense the same can be readily generalized to cover these cases

96  Slicing Thoughts too. The key point is that the subject will think of an object via the same sense as long as she unreflectively represents it as the same. In the kind of case with which Campbell is concerned, the continuity of the subject’s perceptual experience is an essential part of the unreflective presupposition that the perceived object is one and the same and that therefore there is only one perception-based demonstrative sense involved. But the continuity of perceptual experience is not necessary for this. As a rule, when it exists, we have the unreflective assumption that a single object is involved. But in addition, when I put my cup on the desk, turn aside to turn on the light, and then turn back to the desk, I take it for granted in the same way that it is the same desk and cup, and there is no proliferation of senses. Similarly, if I place my cup on a bedside table at night, I take it for granted that it is the same cup in the morning. So, the unreflective taking for granted of identity seems to be much more prevalent and blocks the proliferation of senses even further. This is to say that our unreflective assumptions, which play a foundational role in our reflective employment of concepts, are widespread. The unreflective taking for granted of identity makes the subject think of the given object via the same sense and hence blocks the proliferation of senses. This need to cast the net wider also has repercussions for the role of contingent empirical facts and findings from cognitive science on the project: Campbell’s claim that ‘if one does succeed in keeping track of an object over time, then one must know immediately that it is the same thing that is in question’ (Campbell 1987, 285), or findings in contemporary vision theory according to which visual tracking is done nonconceptually (see Pylyshyn 1999; see also Jeshion 2010) bear diminished importance.5 For the pervasiveness of these unreflective assumptions is something with which we are all familiar, as exemplified by the description of the desk and cup case, above. In fact, in a world in which objects often change their identity, or if a specific subject has become sceptical of her own perceptions for whatever reason, keeping one’s eyes attentively on an object does not guarantee knowing immediately that it is the same thing that one perceives. So, this reliance on a contingent, empirical fact might actually spoil the project. What matters is that any reflection presupposes some unreflective taking for granted that blocks the proliferation of senses.

6.5 The Sameness of Thoughts The fact that the unreflective taking for granted of identity does not require the continuity of a subject’s perceptual experience also means that the proposed criterion of sameness for thoughts applies to a wider range of cases than those for which Campbell accounts. In providing a rationale for (CD’), this criterion also makes (CD’) conform to this wider

Slicing Thoughts  97 range of cases. To illustrate this, consider the diachronic case first. If the subject takes for granted the identity of an object that she is perceiving over time (whether she is continuously keeping her eyes on it or not), the sense will be the same throughout. The fact that she could have wondered (or may come to wonder) whether a single object is in play will not make a shift in sense. The fact that she could have made (or may come to make) as many divisions in her input information as possible, but in fact has not, surely does not show that because of this sheer possibility, she will think of the given object via as many senses as this possibility in principle allows (of whose shift she is unaware, at that). We can tackle the synchronic case analogously. If the subject observes an object at a particular time and takes it for granted that it is a single object, i.e. does not make a division in her input information as in the foregoing ship example, the sense will be the same. The fact that it could have been informative for her to be told that this ship (pointing to its bow) is this ship (pointing to its stern) will not make a shift in senses in the actual case. It follows from this that, unlike in the evening star/morning star case, such a subject cannot think of the given ship via the same sense(s) as the subject who believes that (or wonders whether) two different ships are in play (and similarly for a diachronic case). This is a consequence of the foregoing analysis showing that senses do not line up with the divisions in input information the subject could have made (or can come to make), or – what amounts to the same thing – with the divisions in input information that a similarly situated subject with different background beliefs is in a position to make. This makes senses capture the subject’s cognitive perspective, in that the subject’s taking herself to be thinking of a single (different) object(s) makes the sense(s) the same (different). Sense thus serves to rationalize the subject’s cognitive perspective. Short of this, the thought-content of which it is part would no longer serve one of its key purposes. It is part of this view that the subject is not in a position to take conflicting epistemic attitudes towards the same thought (at the same time). For, wondering whether there are in fact two (or more) different objects rather than one, which gets her in a position to rationally take such conflicting epistemic attitudes, will make the thoughts different. This is to say that thoughts are individuated in terms of how the subject takes the world to be. But this view then seems to face the objection that thoughts do not always line up with how the subject takes the world to be and thus fail to rationalize the subject’s cognitive perspective, that is, fail to be the bearers of cognitive significance. Suppose that in a qualitatively similar situation to that outlined above, there really are two ships in play instead of one, while the subject thinks that one ship is in play (as in Chapter 3) and takes one ship’s bow and the other ship’s stern to belong to the same ship and assents repeatedly to ‘This ship is the Enterprise’ while shifting her

98  Slicing Thoughts attention from one ship to the other in the meantime. Should the subject’s cognitive perspective be rationalized in terms of thoughts, we need to claim that she is thinking a single thought in order to explain her cognitive perspective which is one of representing the two ships as a single ship. (In the diachronic case, this will occur when the subject loses the object with which she began, without realizing this.) This claim seems to be supported by the fact that in taking for granted that, as it were, This ship1 = This ship2 , she is not judging (falsely) that the two ships are the same, which would admittedly make the senses different. But, we have seen that in order to think of something via the same sense, the subject needs to be entitled to make valid inferences of the form ‘This is F’ and ‘This is G’: ‘Hence something is both F and G’, without relying on a further premise, which is not the case here. ‘Something is F and G’ does not follow from ‘This ship is F’ and ‘This ship is G’, in which the respective occurrences of ‘This ship’ are about different ships (as in Chapter 3). It seems then that in assenting repeatedly to ‘This ship is the Enterprise’, the subject is unknowingly thinking two different thoughts due to the fact that she is, respectively, confronted with two different objects, i.e. two different ships. But, since thoughts then play no role in accounting for the subject’s cognitive perspective which is one of representing the two ships as a single ship, it is not clear what role they are supposed to play. This problem is distinct from the slicing problem that I am concerned with here. It concerns not (CD) but a logically independent criterion of sameness for thoughts never employed by Frege (though philosophers like Carruthers (1987) and Luntley (1997) endorse it by adopting a biconditional version of (CD)): (CS) If two thoughts are different, a rational subject can at the same time take conflicting epistemic attitudes towards them. I have argued that if the subject takes it for granted that the object that she is perceiving is a single object, the thought that she is thinking is the same because she is not in a position to take such conflicting epistemic attitudes. This is in line with (CS) (via its contrapositive). But (CS) generally does not hold. It does not hold for analytic truths (or falsities) since a rational subject cannot take conflicting epistemic attitudes towards two different (independent) analytic truths (or falsities). More importantly, in the present case, (CS) does not hold if we allow the subject to be thinking two different thoughts without realizing, in the sense noted above. Here, we do not have the slicing problem but the equivocation problem which is a version of the ‘Twin Earth problem’ involving the subjectively same presentation of different objects which was dealt with and resolved in Chapter 3. I want to note, though, that Evans (1982, 175–176, 1985, 311, n. 21) and Campbell (1987, 285) block the possibility of the subject’s thinking

Slicing Thoughts  99 two different thoughts when she has confused two different objects for a single one (they are concerned with diachronic cases). As we have seen, they hold that to think of an object over time via the same demonstrative sense is to keep track of it (in the sense of having a continuous perceptual experience of it). In discussing a case in which the subject mistakes a previously seen object for an object that she is presently seeing, Evans, as remarked in Chapter 3, urges that she needs to be able to understand the claim ‘This1 ≠ this2’ in order to have a thought at all, but if she does not also entertain this claim, she will think a single mixed thought (Evans 1982, 296–297). On the other hand, when the information that causally derives from two different objects is combined in an indecomposable way such that the subject is unable to understand the claim ‘This1 ≠ this2’, there is not just one object which the sense identifies such that no coherent thought-content can be found (Evans 1982, 297). The subject ends up thinking no thought at all and is only having an illusion of thought.6 But, we have seen that this is not plausible. As noted, the subject is rather thinking here a thought that fails to refer (Recanati) or an equivocal thought (Millikan), as the case may be. This way the subject’s cognitive perspective is accounted for in terms of thinking a particular demonstrative thought.7 The fact that ‘Something is F and G’ follows from ‘This ship is F’ and ‘This ship is G’ without relying on the identity presupposition is the mark of the sameness of sense. It is just that in this case, unlike in that in which a single ship would be in play, the identity presupposition is false. In the vast majority of cases, however, we do not commit such errors in identification of objects – which is to say that in representing objects as the same either with respect to their spatial parts or over time we are reliable but not infallible. Hence, it holds that, in the vast majority of cases, to a single sense, there corresponds a single object, which entitles us to make valid inferences of the aforementioned kind.

6.6 Blocking the Interpersonal Proliferation of Thoughts Returning to the proliferation of thoughts issue, the virtue of the view of thought individuation that I have proposed is that in addition to blocking an intrapersonal proliferation of thoughts, it also blocks an interpersonal proliferation of thoughts in the kind of case discussed by Strawson (1959, chap. 1) and Evans (1982). Suppose the hearer perceptually identifies object o, referred to by the speaker. Suppose they are having a communication about o and both assent to an utterance of ‘This is F’ that is about o. Evans suggests that the sense via which the speaker thinks of o will be different from the sense via which the hearer thinks about o if they identify o via two relevantly different parts. What makes it communication, rather than misunderstanding, is the fact that there is a single inclusive object encompassing both the part perceived  by  the speaker

100  Slicing Thoughts and the part perceived by the hearer (see Evans 1982, 333). This makes senses as constituents of thoughts as contents (objects) of belief interpersonally shifty and unstable. As we have seen, this, in turn, makes them unsuitable for being parts of attitude contents that contain them, in view of the noted fact that attitude contents need to enable us to predict correctly the conditions in which two people have said the same thing. It also opens the tricky question as to how far we should go in slicing senses along these lines and what should be our guiding principle in so doing. Luckily, individuating senses and thoughts in the way suggested enables us to bypass these issues, yet still have thoughts shape the subject’s cognitive perspective. In accordance with this, we may claim that as long as in this kind of case, the speaker and the hearer (unreflectively) represent such an object, in our example a ship, as the same, they will think of it via the same shared sense and think the same thought about it in assenting to ‘This ship is F’, regardless of whether they are observing it via the same or different parts and regardless of whether they are attaching the same properties to it and collecting the same information about it. As noted, individuating thoughts in the way suggested also makes it impossible for a rational subject to take conflicting epistemic attitudes towards the same thought (at the same time) on which (CD) and (CD’) hinge. True, Mates (1950) has doubts about this when it comes to Carnap’s (1947) treatment of synonymity in terms of intensional isomorphism which trades on this. He thinks that it is possible for a linguistically competent subject to take conflicting epistemic attitudes towards (thoughts expressed by) two sentences that differ only in that one of them contains the expression ‘period of fourteen days’, while the other contains the expression ‘fortnight’, which are obvious synonyms, i.e. have the same sense. But, whatever the merits of Mates’s contention when it comes to the cases with which he is concerned, it is inappropriate when it comes to demonstrative thoughts. As argued, the subject’s thinking the same demonstrative thought is a matter of her unreflectively taking for granted the identity of the perceived object. Wondering whether a single object is in play will make the thoughts different which rules out the possibility of taking conflicting epistemic attitudes towards the same thought.8

6.7 Conclusion I have argued that what makes demonstrative sense the same is the subject’s unreflective taking for granted that the object that she is synchronically or diachronically perceiving is a single object. In conformity with this, to think of an object via the same demonstrative sense from one occasion to the next does not require the subject to keep track of it in the sense of having a continuous perceptual experience of it as Evans

Slicing Thoughts  101 and Campbell think. Individuating demonstrative senses in this way provides a rationale for (CD’) rather than (CD) which is generally taken to be a criterion of difference for thoughts. Unlike (CD), (CD’) takes two thoughts to be different if a rational subject is at the same time in a position to take conflicting epistemic attitudes towards them and not because it is in principle possible for her to do so. It thereby captures the subject’s cognitive perspective and blocks a proliferation of senses and thoughts. This goes some way towards meeting the aforementioned requirement pointed out by Stalnaker that the possibilities used to individuate the objects of belief (and other attitudes) must be those that are relevant to naturalistic relations rather than epistemic possibilities, but I shall not pursue this issue here.9

7 How Many Modes of Presentation Do We Need?

7.1 Introduction I have argued that in representing an object or day from one occasion to the next as the same the subject thinks about it under a single mode of presentation. However, some philosophers have argued that two different kinds of mode of presentation are involved here. In this chapter, I want to discuss their views and show that they are unjustified. This will lend further support to my contention that a thinker’s cognitive perspective or her indexical point of view is to be explained in terms of a single kind of content. By broadening our discussion to also cover cases of re-identification and recognition of objects, it will once again be forthcoming that we need to account for the subject’s cognitive perspective in terms of modes of presentation as they were individuated in the last chapter.

7.2  Re-identification of Objects Consider first the issue of the subject’s perceptual re-identification of a certain object that she has encountered before. The question that I want to deal with now is in what consists the continuity of the subject’s belief in this kind of case. Suppose the subject sees a cup on her desk and, as it were, thinks ‘This cup is green’. Later on, she sees the same cup on the same desk but takes it to be a different cup and thinks ‘This cup is chipped’. Since she does not realize that the same cup is in play, in making a division in the input information that she is receiving, she will think of it under two different modes of presentation: one based on her memory of it stemming from her former perception of it and the other based on her current perception of it. There is no internal continuity in the subject’s belief. But suppose that taking a closer look at the cup she comes to notice a familiar mark on it and realizes that the cup that she is seeing now on her desk is the cup that she saw there before. Since her recognition of the cup is not immediate but rather involves her making an identity judgement that that cup = this cup, the two modes of presentation become linked (see

How Many Modes of Presentation Are There?  103 Recanati 2013a). This way the internal continuity in the subject’s belief that was initiated before the error was committed is restored. When, on the other hand, the subject immediately recognizes the cup that she is seeing now on her desk as the cup that she saw there before, there is just one mode of presentation of the given cup in play throughout. This mode was initialized during her first encounter with it, stored into memory, and re-activated during the encounter in which she recognized it. This is supported by the following discussion.

7.3  The Modal and the Non-modal Criteria of Difference As noted, in individuating modes of presentation, Evans (1982, 176) and Campbell (1987, 280ff) are concerned with the case in which the subject keeps her eyes attentively on an object, allowing a single mode of presentation to persist over small interruptions in the subject’s perception such as when an object that the subject perceives disappears momentarily behind an obstacle. They also hold that while thinking about an object over a period of time under the same mode of presentation, the subject may collect new information about it from perception by noting that it is, say, moving, that it looks such-and-such from the back, that its shape is such-and-such, and so on. In addition, Campbell (1987) claims that the same mode of presentation can be used both in visual and tactual judgements as when, for example, one judges ‘This cup is green’ by looking at the cup and then judging ‘This cup is chipped’ by touching it a few moments later, from which one is entitled to immediately infer ‘Something is green and chipped’. In relation to this, Campbell contends: Someone who simply never took himself to be keeping track of objects from instant to instant, or from sense-modality to sense modality, would be someone for whom perception had ceased to have, as part of its intrinsic character, the representation of objects as independently existing things. Ways of thinking of objects are thus intrinsically coarse-grained with respect to the underlying perceptual information. (Campbell 1987, 291) I have argued that a single mode of presentation can persist in those cases in which the subject no longer perceives the object she once perceived. This involves cases in which she remembers the given object or re-encounters it on a later occasion. Suppose that at a later time, the subject attends to the cup that she initially perceived, presuming that it is the same cup as before. The recognition is immediate. She is entitled to immediately infer that something is green and chipped even though her first thought episode, associated with the premise ‘This cup is green’,

104  How Many Modes of Presentation Are There? is based on her original encounter with the given cup, whereas her later thought episode, associated with the premise ‘This cup is chipped’ is based on her perception of the cup during her recognition of it which relies on her memory of it. In accordance with this, Papineau (2006, 3) has claimed that in such cases, the subject thinks of an entity under the same perceptual concept during the entire period. Perceptual concepts allow subjects to think about perceptible entities. Such concepts are formed when the subjects initially perceive relevant entities, get stored into memory, and are re-­ activated by later perceptual encounters. This is to say that information accumulates and gets carried through various encounters with the same entity (Papineau 2006, 4). Papineau takes perceptual concepts to be mental files associated with ‘sensory templates’ which will be set up on initial encounters with the relevant objects. They will be then re-­ activated on later perceptual encounters, via matches between incoming stimuli and stored template (Papineau 2006, 114).1 Individuating mental files in this way has come under attack from Recanati (2012, 2013b, 2016). He acknowledges the existence of coarsegrained mental files which garner the accumulating body of information about an individual through time. Yet, he claims that this way of individuating mental files nonetheless disregards different finer-grained mental files that are based on different ways in which the subject may be related to the individual. Drawing an analogy with indexicals, Recanati suggests that there are mental concepts – mental files which inter alia play the role of the modes of presentation – which the subject opens when she stands in an epistemically rewarding relation (ER relation) with a certain object (the referent of the file). The demonstrative file THAT CUP is based on certain contextual relations to the given cup in virtue of which the subject can not only perceive it but can also focus her attention on it in a dynamic and discriminating manner. When the subject can no longer perceive the cup or focus her attention on it, another epistemically rewarding relation takes hold and with it another mental file – the relation in virtue of which the subject remembers the cup. When the subject recognizes the cup, a recognitional file takes hold in virtue of an epistemically rewarding relation of familiarity. As long as she is capable of recognizing the cup in question, she can think of it via a recognitional file. In agreement with Papineau, Recanati holds that the same body of information persists throughout this process but thinks that it is too coarse grained to play the role of modes of presentation, for which role the suggested mental files are more suitable. He says: I keep the fine-grained ER relations, used to account for indexical thought and Frege cases. But, in the spirit of Evans and Papineau, I introduce dynamic files, which are sequences of files in the static sense. (Recanati 2016, xiii)

How Many Modes of Presentation Are There?  105 As a result, Recanati argues that in immediate recognition, the subject does not think of the cup under the mode of presentation or mental file that was initialized during the original encounter with it. She rather thinks of it via a recognitional-demonstrative file. This file is of a distinctive type based on two ER relationships with the cup: one based on memory and the other based on the current perception of the cup (see Recanati 2013a, 2013b). Thus, this file contains both the information that the given cup is green and that it is chipped. That the same file, the same mode of presentation, is involved ensures that the subject is entitled to validly infer ‘Something is green and chipped’ from ‘This cup is green’ and ‘This cup is chipped’ without appealing to the identity premise ‘That cup is this cup’. The fact that a single file should include here two different ER relationships suggests that trading on identity requires coarse-grained modes of presentation. In discussing ‘now’-thoughts and ‘then’-thoughts, Recanati says: …to account for cross temporal cases we have to make room for coarse-grained modes of presentation enabling the mixing of temporal perspectives (e.g. the simultaneous exploitation of memory and perception). Coarse-grained modes of presentation correspond, in my framework, to composite files based on multiple ER relations…. (Recanati 2016, 80) While trading on identity requires coarse-grained files or modes of presentation, Recanati insists that we still need fine-grained files. They conform with (CD) which Recanati formulates as follows (2016, 80): Two modes of presentation m and m’ are distinct if it is possible for the subject to raise doubts as to whether, through their deployment, she is thinking about the same thing. (Recanati 2016, xv) Consider now the foregoing case brought up by Campbell in which the subject perceptually tracks an object via different ER relations, through perceiving it and through touching it. Unlike Campbell, Recanati takes it that due to the fact that two different ER relations are involved – the visual one and the tactual one – two different modes of presentation are involved although the same body of (accumulating) information is also involved. Two different modes of presentation are involved by the lights of (CD) because it is possible for the subject to come to doubt that the object seen is the object touched, even if she actually harbours no doubt at all. But Recanati thinks that there is no unnecessary multiplication of mental files here. For, ‘potential Frege cases’ are used only to individuate modes of presentation construed as types. Two modes of presentation

106  How Many Modes of Presentation Are There? types, m and m’, are distinct just in case it is possible for a rational subject to entertain doubts as to whether through their deployment, she is thinking about the same thing. But when it comes to token modes of presentation, what counts is doubt actually harboured, not doubt that is merely possible. When the subject trades upon the identity of the object seen and the object touched, there is a single mental file based on several ER relations. Similarly, to the foregoing case featuring memory and recognition, the file is of a distinctive type based on a composite ER relation. Identity of the object seen and the object touched is presumed. ‘The tokens that are deployed are individuated by the subject’s actual dispositions at the time of deployment’ (Recanati 2016, xvi). While in the case of mode of presentation types, Recanati applies the foregoing modal version of (CD), in the case of mode of presentation tokens, he applies a non-modal version of (CD): The modes of presentation m and m’ are distinct if the subject entertains doubts as to whether, through their deployment, she is thinking about the same thing. (Recanati 2016, xv) It is the non-modal version of (CD) that Campbell employs to individuate perception-based modes of presentation. However, Recanati relies on the modal version of (CD) in order to establish that the two ER relations in question give rise to two distinct (fine-grained) modes of presentation types. Short of this, there seems to be no rationale for taking these ER relations to give rise to distinct mode of presentation types. Recanati takes it as constitutive of modes of presentation that they should comply with (CD). He takes it that mental files play the mode of presentation role so they can account for Frege cases in accordance with (CD). But, mental files also play the semantic coordination role in line with the criterion of sameness for modes of presentation as in Evans and Campbell so as to enable trading on identity or co-reference de jure as Recanati (2016, xi) calls it.2 Since the latter role is of a piece with a non-modal version of (CD), he holds on to the modal version of (CD) so as to ensure that in the foregoing situation we are dealing with modes of presentation at all. In Recanati’s own words: We need fine-grained files to play the [CD] role because the subject can always wonder whether, e.g., the object he sees is the object he encountered in the past and remembers. In such a case, there has to be two distinct modes of presentation (by [CD]). Now to play the second role (the [trading on identity] role), files must be coarsegrained, given that cross-modal and cross-temporal integration licenses [trading on identity]. (Recanati 2016, 72)

How Many Modes of Presentation Are There?  107

7.4  Against the Modal Criterion of Difference This is to say that if we stick to the non-modal version of (CD), as Campbell does, we do not seem to need fine-grained modes of presentation. That we should proceed this way is shown as follows. Recanati claims that in trading on the identity of the object seen and the object touched, what is deployed is a single mental file, based on several ER relations. The file is of distinctive type, based on a composite ER relation. Identity of the object seen and the object touched is presupposed through the compounding of ER relations. That is obviously compatible with the possibility for the subject to come to doubt that the object seen is the object touched, i.e. to stop presupposing identity. Coming to doubt is a dynamic operation which amounts to splitting the composite file. This can always happen, but the sheer possibility of doubt is not sufficient to entail that two distinct files are actually deployed (before the split). (Recanati 2016, xvi) The claim is that the single mental file that is deployed is composite. It contains two finer-grained files that are distinct by the lights of the modal version of (CD), i.e. because of the sheer possibility of coming to doubt that the object seen is the object touched even though the subject harbours no such doubt at all. This is because: At any point in the [dynamic] sequence, the subject can wonder whether or not the tracked individual is the same; this is only possible if the various ER relations (hearing the voice, seeing the face, etc.) determine distinct modes of presentation. (Recanati 2016, xiv) What Recanati suggests is that the subject can wonder whether the tracked individual is the same from one occasion to the next only when different ER relations are involved. Unlike dynamic modes of presentation that span successive ER relations to the reference, these static modes of presentation are as fine-grained as the ER relations they are based on (Recanati 2016, xiv–xv). However, it is in the same sense merely possible for the subject at any point in time to wonder whether the individual that she is tracking via a single ER relation is the same. The subject may take herself to be perceiving two different individuals from one occasion to the next even if she is receiving a unified stream of information from a single individual. As noted, she may never move her eyes from an object yet think (falsely) that different objects keep materializing in the place at which she is looking (see Richard 1990, 227). In the process of receiving a unified stream of information, the information may be enriched by

108  How Many Modes of Presentation Are There? way of seeing what the individual looks from below, above, behind, and the like. We have seen that it is not plausible to claim that an influx of information about a perceived individual, or a passage of time, changes the mode of presentation since this would make it impossible to sustain a thought over time or over informational enrichment (Sainsbury 2002, 157). In line with this, Recanati takes his fine-grained modes of presentation as not decomposable into more fine-grained ones, while he takes coarse-grained modes of presentation to be so decomposable. This shows that there does not seem to be a principled reason to take the modal version of (CD) to apply only to those cases that involve various ER relations such as the visual one and the tactual one which, according to Recanati, give rise to distinct mode of presentation types. For, it is in the same sense possible for a subject to start harbouring doubts as to whether the object that she is uninterruptedly perceiving is the same as the object that she perceived a little while ago due to, say, (unexpected) informational enrichment. The cup that the subject is looking at may unexpectedly change its colour (for whatever reason), making the subject wonder whether it is the same cup as before. Short of providing such a reason, a curious multiplication of modes of presentation is forthcoming. While keeping her eyes on an object, the subject may come to think falsely that many different objects keep materializing in the place at which she is looking. To recall, should we individuate modes of presentation in terms of this sheer possibility of doubt, not only do we have an unnecessary multiplication of modes of presentation where it seems intuitively plausible that only one mode of presentation is in play; the fact that the subject is actually unaware that she is thinking of the same object under different, let alone many different modes of presentation, makes modes of presentation opaque rather than transparent as they are supposed to be. That modes of presentation need to be transparent is shown by the fact that we ascribe them to a person to enable assessments of her rationality and to explain her behaviour.3 As a consequence, postulating to the subject thinking of the given cup under (putative) modes of presentation that are not transparent defeats the purpose. A way to block this implausible outcome is to stick to the non-modal version of (CD), i.e. (CD’), and discount of fine-grained modes of presentation and composite files in accounting for the subject’s way(s) of thinking of the object seen and touched. As Campbell contends, modes of presentation will be different just in case the subject makes a division in the input information that she is receiving, i.e. just in case she comes to harbour doubts as to the identity of the object seen or the object seen and touched in accordance with the non-modal version of (CD), i.e. (CD’). This may seem to have some potentially problematic consequences. Consider two subjects who are equally rational but who have different evidence concerning the relevant identity – one of them has strong evidence against the identity of the cup she has seen and the cup she is touching, the other one does not. Clearly, the first subject will come to

How Many Modes of Presentation Are There?  109 actually doubt that the identity holds, while the second one will not. On the basis of the non-modal version of (CD), we need to conclude that the first subject has two different modes of presentation, while the second one does not. It then follows that the identity relations between a subject’s modes of presentation depend on what evidence she has. Some people might want to reject this claim and take the identity of one’s mental representations to depend on other factors, such as the way one thinks of the object in deploying the mental representation in question. In response, note at any rate it follows from the non-modal version of (CD) that the identity relations between a subject’s modes of presentation depend on what evidence she has whether we also accept the modal version of (CD) as Recanati does, or not. True, the modal version of (CD) allows us to individuate the type of mental representation in terms of the type of the underlying ER relation. But, the non-modal version of (CD) has this consequence even if we ignore the foregoing contention that there does not seem to be a principled reason to take the modal version of (CD) to apply only to those cases that involve various ER relations. This is to say that a difference in ER relations is neither sufficient nor necessary for a difference in the modes of presentation. The subject can think of a given object (cup in the foregoing example) under the same mode of presentation across different ER relations. Alternatively, she can think of the same object under different modes of presentation across the same ER relation. To be sure, acknowledging that two different ER relations (types) are involved in the coarse-grained mode of presentation that persists from one sensory modality to the other does not support either version of (CD). This is to say that even if two different ER relations are involved, it does not follow that either version of (CD) applies to the case in question. It does not follow that the antecedent of either version of the principle is satisfied in that case. These ER relations are accessible to the subject in that she knows whether she is standing in a visual or a tactual relation with the relevant object, and similarly for the ER relations underlying perception, memory, and recognition. This is so whether or not the subject entertains doubts as to whether she is thinking about the same thing or whether or not it is merely possible for her to entertain such doubts. Hence, we lack convincing motivation for holding that this coarse-grained mode of presentation corresponds to a composite file based on multiple ER relations that is composed of two distinct fine-grained modes of presentation (types), as Recanati would have us believe.

7.5  A Case for the Modal Criterion of Difference In claiming that we should abandon the modal version of (CD) in the foregoing cases, I do not claim that Frege and other philosophers are wrong in applying their modal version(s) of (CD) to the cases with which they are concerned. Frege says that the thought in the sentence ‘The

110  How Many Modes of Presentation Are There? morning star is a body illuminated by the Sun’ differs from the thought in the sentence ‘The evening star is a body illuminated by the Sun’ because anybody who did not know that the evening star is the morning star might hold the one thought to be true, the other false (Frege 1892/1980, 62). Now, competent language users who know that the morning star is the evening star will be aware of the difference in the modes of presentation of Venus expressed by ‘The morning star’ and ‘The evening star’. For the linguistic meanings of these two expressions which belong to two distinct linguistic types yield two distinct modes of presentation of Venus. Being of distinct types the linguistic meanings of ‘The evening star’ and ‘The morning star’, will be distinct for the competent language user even in those cases in which she is deploying their tokens in, for example, assuming that the evening star is the morning star (in the sense of our remarks in Chapter 6). This being so, the subject does not trade on the identity of Venus. In contrast with this, in the foregoing case featuring sight and touch as well as in the case of immediate recognition, the subject is trading on the identity of the object which we have seen can accommodate the modal version of (CD) only at the cost of having an unnecessary multiplication of modes of presentation, which is avoided by discounting of fine-grained and composite modes of presentation in favour of coarse-grained ones.

7.6  Name-Like Modes of Presentation To recall, in a similar way, Papineau claims that when the subject keeps thinking of an object through perception, memory, and (immediate) recognition, she is thinking of it under the same perceptual concept during the entire period. Papineau believes that we do not need to postulate demonstrative files either for having putative distinguished epistemic inputs or special behavioural outputs. In support of the former claim, Papineau (2013) argues that although we often use indexical words, as when we want to say today what we expressed yesterday using the word ‘today’ by replacing it with ‘yesterday’, there is nothing correspondingly indexical in our thoughts. For example, in immediately recognizing the given cup, the subject is thinking about it under the same file as when she formed the perceptual concept of it. Her thinking of it involves, not different files based on different ER relations, but a single permanent file which is name-like. Name-like files are designed to be permanent repositories of information about the item in question and are not dependent on any particular sources of information about that item. In support of the latter claim, Papineau says: I take it that the direct control of fine-tuned motor behaviour is managed by an automatic sub-personal system, analogous to the system that determines which words we use to voice our thoughts.

How Many Modes of Presentation Are There?  111 This motor control system will respond to directives like grab that thing there [that I can see/feel/hear], and to this extent will indeed deploy tokens of a type mental term that thing there. But this automatic motor control system is not in the business of storing information about the things it refers to, and so will not have any information accumulating files associated with its tokens of that thing there. (Papineau 2013, 172) Papineau believes that there are only name-like files. If the subject has a lasting name-like file of John, when she addresses him by means of ‘you’, there is no additional mental file in play. This is just another case in which the subject’s selection of words is generated automatically, courtesy of an automatic and unconscious system that figures out what grammatical string of words will best serve to express her thought in the current context (Papineau 2013, 5). In contrast with this, Recanati urges that there are also demonstrative files by showing that modes of presentation can be as fine-grained as one wishes. Suppose the subject suddenly comes to entertain a doubt about the identity of the man in front of her and think: ‘Is that man really John?’ In Papineau’s view, she would open another potentially lasting name-like file which will disappear once the subject realizes that it is John in front of her. But Recanati thinks that it is not enough to posit two files to account for this case. We need to account for two types of mode of presentation, corresponding to the singular terms ‘that man’ and ‘John’, respectively, the former of which is based on the subject’s current perceptual relation to John while the latter is based on multiple information sources. Each of these modes of presentation is in the business of storing information about John. While Papineau acknowledges the existence of indexical modes of presentation in thought (at the interface with the action-guiding system), he thinks that such modes of presentation are not in the business of storing information about referents. The information is rather stored by name-like coarse-grained files (see Recanati 2013b, 18).

7.7  Demonstrative Modes of Presentation It is true that there are two modes of presentation here by the lights of the non-modal version of (CD), i.e. (CD’). But this does not settle the question as to whether we have here two modes of presentation of the same type a la Papineau or two modes of presentation that belong to two distinct types a la Recanati. The fact that in perceiving an object or person, the subject can also focus her attention on it in a dynamic and discriminating manner suggests that the subject is standing in a special (perceptual) relation with it which is in the business of storing information about it in a special way. For Evans and Campbell in keeping track

112  How Many Modes of Presentation Are There? of an object from one occasion to the next, the subject is thinking of it under a distinctive mode of presentation as a result of standing in such a distinctive perceptual relationship with it.4 Hence, this mode of presentation is of a distinct type from the namelike one that is based on multiple information sources. Its distinctness is the result of the subject’s standing in a perceptual ER relation with the object which gives rise to a distinguished epistemic input as a result of which the subject is in a position to explore the object.5 This is something that the subject is not in a position to do when she is no longer in the perceptual presence of the object and stands in the memory-­ based ER relation with it. The perceptual ER relation also gives rise to a special behavioural output. While perceiving the object in her vicinity, the subject is in a position to act upon it. When she is no longer in the perceptual presence of it and stands in the memory-based ER relation with it, she is no longer in a position act upon it. This suggests that Papineau is wrong in denying that there are distinguished epistemic inputs and special behavioural outputs which give rise to demonstrative files (modes of presentation) and that Recanati is right about this.6 But, this does not entail that we need fine-grained modes of presentation (types) based on different ER relations on top of coarse-grained ones to account for the case in which the subject keeps thinking of an object via perception, memory, and immediate recognition. One can argue with Cappelen and Dever (2013, 48) that ‘nothing general follows about the connection between indexicality and agency. All kinds of difference in beliefs can make a difference to what it is rational to do’. At any rate, to have finegrained modes of presentation, we also need the modal version of (CD), accepting which leads to the problems pointed out above. It is thus the coarse-grained mode of presentation that accounts for the subject’s way of thinking of an object through distinct ER relations with it. But, being based on distinguished epistemic inputs and special behavioural outputs, each characteristic of the perceptual ER relation, it is not a name-like file (mode of presentation) as Papineau holds. It may seem that the position that I have reached is unstable – that it either collapses into Papineau’s model or into Recanati’s model. The distinguished epistemic inputs and special behavioural outputs that are characteristic of the perceptual ER relation are not instantiated by the memory relation. These inputs/outputs depend on the possibility of perceiving and acting directly upon the object. This made us argue against Papineau that these considerations concerning distinctive inputs/outputs should lead us to conclude that demonstrative files (modes of presentation) are of a distinct type from name-like files. So, by the same reasoning, the memory file should be of a distinct type from the perceptual-­demonstrative file. Hence the memory file and the ­perceptual-demonstrative file are of distinct types after all, which is ­precisely what I wish to deny.

How Many Modes of Presentation Are There?  113 In response, I will repeat that for the different ER relations to give rise to distinct mental files, i.e. to different mode of presentation types in the sense of Recanati, we need the modal version of (CD) that I have discarded for reasons stated above. (As we saw, Recanati holds on to the modal version of (CD) just to ensure that in this kind of situation we are dealing with modes of presentation at all.) The fact that the relevant mode of presentation is of a distinct type from the name-like one in having the distinguished epistemic inputs and special behavioural outputs, resulting from the subject’s standing in a perceptual ER relation with the object, does not entail that there are fine-grained modes of presentation on top of coarse-grained ones. Unlike the modal version of (CD), the non-modal version of (CD), i.e. (CD’), does not entail this. (As we saw, the non-modal version of (CD) does not even settle the question as to whether we are dealing here with two modes of presentation of the same type a la Papineau, or with two modes of presentation that belong to two distinct types a la Recanati.) A coarse-grained mode of presentation can take distinct ER relations in its stride without having them give rise to finer-grained modes of presentation. This view is supported by the following considerations. As noted, Campbell (1987) has urged that keeping track of an object is done unreflectively. It is part and parcel of the unreflective use of perception-based demonstratives that the subject does not make a division in the perceptual information that she is receiving. The sheer possibility of such a division does not show that in unreflectively tracking an object, the subject is actually in a position to ask whether this object (at t 1) is this object (at t 2). This is to say that modes of presentation are not to be individuated in terms of the sheer possibility of doubt, which way we block an unnecessary multiplication of modes of presentation, as argued. However, as I have argued, the unreflective taking for granted of identity does not require the subject to be in constant perceptual contact with the object. For example, when I put my cup on the desk, turn aside to turn on the light, and then turn back to the desk, I take it for granted that it is the same desk and cup. Similarly, if I place my cup on a bedside table at night, I take it for granted that it is the same cup in the morning. In both these cases, my taking it for granted that it is the same cup involves an immediate recognition of the cup. My thinking of the cup (or the desk) every time I immediately recognize it involves two ER relations with it – the memory-based relation and the current perceptual relation, as Recanati claims. But since unreflective taking for granted plays the key role here no matter in which ER relation I stand with the given cup (or desk), it is natural to extend Campbell’s insight beyond the perceptual case and take it that I am thinking of the same cup (or desk) under the same (coarsegrained) mode of presentation throughout, as I have done above. The fact that my unreflective taking for granted of identity makes it the case that I am thinking of the same cup (or desk) from one occasion

114  How Many Modes of Presentation Are There? to the next under the same mode of presentation supports the view that there is no need to take the information that I have stored about the given cup (or desk) during my initial encounter with it via a perceptual ER relation with it to constitute a fine-grained mode of presentation in addition to a coarse-grained one, as Recanati claims. Not only is this consistent with the foregoing claim that while I am standing in a perceptual ER relation with the cup I am standing in a special relation with it; it is also supported by Campbell’s foregoing insight that someone who never took to be keeping track of an object over different ER relations (in Campbell’s case from sensory modality to another) would have no grip on objects as independently existing things. To individuate modes of presentation in this way is to say that the subject’s mode of presentation of a given object such as the cup in the foregoing case may involve all and only the same properties whether the subject’s belief about the cup’s properties is true or false. If, on the other hand, the subject changes her mind about the cup’s properties and at one point in time starts believing that she was wrong in thinking that it was green, she will drop the property of being green from the body of information forming her mode of presentation of the given cup. Still, as noted in relation to days, this does not make a split in her mode of presentation of it. The identity of the mode of presentation is not affected by whether the subject is attaching new features to the given cup in the process of thinking about it over time or withdrawing those she once took it to have, or whether these features really pertain to it. If, on a later occasion, the subject’s thinking of the cup draws upon her memory of it, and it alone, and is aimed at it, and it alone, as its causal source, as the same cup that her original belief was about, her mode of presentation of it will be the same no matter which properties the subject ascribes to it.

7.8  Temporal Modes of Presentation Consider again the ‘today’/‘yesterday’ case. In Chapter 5, it was noted that ‘today’ and ‘yesterday’ are for Perry different roles/characters/senses and so are ‘today [the day of this thought] is beautiful’ and ‘Yesterday [the day before the day of this thought] was beautiful’, although the content/ thought corresponding to them in the given context – that d is ­beautiful – is the same. The difference in the characters the subject accepts is, in turn, taken to account for the difference in her belief states and behaviour. This might seem right, for she will apparently think of this day in two different ways, first, say, as the present day and then as the previous day. Her thinking about it as the present day can make her pragmatically attached to it because she can make changes in what happens on it, whereas her thinking about it as the previous day cannot make her pragmatically attached to it because she can no longer make any changes

How Many Modes of Presentation Are There?  115 in what happens on it. Since she is no longer in a position to actively explore d, she is not epistemically attached (see Perry 1997, 33). Now, I have shown via some non-standard cases that (doxastic) characters do not account for indexical belief states and behaviour but rather senses or modes of presentation which are wholly independent of characters and that Perry is wrong in running the two together. However, Perry’s point about the pragmatic as well as epistemic attachment reinstates itself in the present context at the level of Recanati’s ER relations. If the subject is having a “demonstrative” ER relation with d while assenting to ‘Today is beautiful’, she is both pragmatically and epistemically attached to d, while she is no longer attached to it once d has passed and she continues to think of d on the basis of her remembering it and assents to ‘Yesterday was beautiful’. This seems to call for individuating modes of presentation in terms of ER relations if not in terms of characters. For, if the subject thinks of d as ‘Today is beautiful’, she is ascribing to it the property of being the present day and is pragmatically and epistemically attached to d. In continuing to think of d on d + 1 that it was beautiful, she will replace the property of being the present day with the property of being the previous day and is not pragmatically and epistemically attached to it. However, given the problems that this way of individuating modes of presentation is facing, pointed out above in relation to Recanati, and in the light of my suggestion that what matters in retaining the relevant belief about an item from one occasion to the next is the internal continuity of the subject’s belief, Perry’s point is best accounted for in taking these properties as something that the subject ascribes to and withdraws from d as time goes by without affecting the identity of the mode of presentation under which she is thinking of d from d to d + 1. This claim is supported by the foregoing considerations concerning the individuation of modes of presentation in the case of perceived objects. As noted, our unreflective assumptions, which play a foundational role in our reflective employment of concepts, are widespread. The unreflective taking for granted of identity makes the subject think of the given object via the same sense from one occasion to the next. This is so even if in the process the subject collects new information about it. The object may be moving such that the subject may note what it looks like from the back, how big it is and the like. The object may also, say, turn from blue to yellow, and the subject will adjust her information about it accordingly while thinking about it under the same mode of presentation. What I just said about keeping track of days is in line with this. If on d, I acquire a belief that d is beautiful, and then in moving from d to d + 1, I take it for granted that it is the same day that I am thinking of (that it is beautiful), then I am thinking of d under the same mode of presentation throughout. The properties of being the present day and of being the previous day are just amongst those features that succeed each other as I continue to believe on d + 1 that d was beautiful.

116  How Many Modes of Presentation Are There?

7.9  Nominal Terms and Indexicals I have argued that representing an object or day as the same from one occasion to the next makes the thinker think about it under the same mode of presentation. This is so even if in the process she updates her belief and changes her mind as to what features the item thought about has. But cases in which the subject needs to adjust indexicals as the context changes in order to re-express her belief seem to challenge this view since each indexical term seems to have a special significance in individuating our psychological states, i.e. for action and behaviour. According to Perry’s (1977/1993) early view mentioned earlier in this book, each indexical has a role which specifies the manner in which the content that is believed is believed. “When you and I entertain the [role] of ‘A bear is about to attack me’, we behave similarly. We both roll up in a ball and try to be as still as possible” (Perry 1977/1993, 23). The claim is that entertaining the same indexical by the same person at different times or by different people is sufficient to give rise to one particular kind of human action. The reverse claim is that a difference in indexicals is sufficient to give rise to different kinds of action. I will now examine these two claims. When I assent to ‘It is rainy now’ at a particular time and later on I assent to ‘It was rainy earlier’ in order to re-express the same belief, I  seem to be in two different psychological states as a result of the difference in indexicals. The same applies to ‘It is rainy here’ and ‘It is rainy there’. For, in believing the first sentence in each pair, I have a reason to carry an umbrella, while in believing the second, I do not. On the other hand, in keeping track of the time at which it was rainy and of the place at which it is rainy, I am entitled to validly infer ‘Something is F and G’ from ‘It is F now’ and ‘It was G earlier, i.e. from ‘It is F here’ and ‘It was G there’, without relying on an identity premise. This makes the sense (mode of presentation) in the respective premises the same. Does this mean that two different modes of presentation (types) are involved? According to Prosser (2005), there are two different modes of presentation here: one that persists through a change of indexicals and individuated in terms of trading on identity as just suggested and the other lining up with different indexicals. He holds that indexical terms such as ‘here’ are best thought of as complex demonstratives, i.e. noun phrases of the form ‘that Φ’, masquerading as unstructured terms. The nominal term – the ‘Φ’ component – picks out a relational property that has the thinking subject as one of the relata though the subject need not always conceive of the property as relational. In Prosser’s view, complex demonstratives can be plausibly thought of as involving two different modes of presentation: one singular, individuated by trading on identity, and the other picking out the property associated

How Many Modes of Presentation Are There?  117 with the nominal term. Suppose I see a sheep running on a hillside at t1 and form a belief: 1 That black sheep can run fast While I keep track of it perceptually, it slowly changes colour from black to white (it does not matter why) such that at t2 I come to be disposed to believe 2 That white sheep can run fast In trading on identity, I have retained the singular mode of presentation of the given sheep. Now, suppose I am frightened of white sheep but not of black sheep. Given a suitable set of background beliefs and desires, (2) would result in my running away while (1) would not. Hence, (1) and (2) have different psychological roles, which implies a difference in the modes of presentation. Prosser thinks that this can be easily explained in terms of the mode of presentation expressed by the nominal component without any need to deny that a singular mode of presentation of the sheep is also retained. Prosser believes that this lends support to his claim that a difference in the indexical term is sufficient to give rise to a difference in action (Prosser 2005, 376). But I think that this claim is too strong. It is safe to say that being frightened of white but not of black sheep, or vice versa, is uncommon. The difference between believing (1) or (2) will leave many people indifferent. Whether a certain sheep is black or white can, of course, prompt people to think or do different things but there is no uniform pattern. What course of action one is going to take is circumstantial and cannot be systematically linked to the meanings of indexicals.7 The same holds for the opposite claim that Prosser does not consider but is worth investigating in order to reinforce the point that I have just made. The claim is that the sameness of the indexical term is sufficient to give rise to the same kind of action. Suppose that in keeping track of the given sheep while it slowly changes colour from black to white, I consecutively come up with the following pair of beliefs: 1* That sheep can run fast 2* That sheep can run fast Given the same set of background beliefs and desires, (2*) would again result in my running away while (1*) would not, although the nominal term is in the two cases the same. Following Prosser’s lead, one can claim that there are two different modes of presentation here since each of the two utterances of the nominal term ‘sheep’ picks out a different relational property having me as one of the relata. But then, the difference in these modes of presentation is not the result of a difference in

118  How Many Modes of Presentation Are There? (the linguistics meanings of) indexicals. To allow modes of presentation to vary while the indexical stays the same is to leave indexicals out of accounting for cognitive significance, as we have seen above in relation to Kaplan’s and Perry’s views. My suggestion regarding Prosser’s example is that the newly acquired belief about the colour of the sheep makes me react the way I do is best accounted for without postulating another mode of presentation on top of the singular one. To see this, note that I would react in the same way as before if in the same situation I were first to believe 3 That sheep is black and then 4 That sheep is white This is in line with what I said above in discussing the cup case. The subject may form a belief that a certain cup that she perceives is green and then change her mind about its colour and drop the property of being green from the body of information forming her mode of presentation of the given cup individuated in terms of trading on identity. She comes to think, for example, that the cup is red. Suppose that she is differentially disposed to different colours such that she is frightened to drink from a red cup (it does not matter why) but not from a green cup. Do we need to introduce another kind of mode of presentation here on top of the singular one? My answer is: no! For, this view implausibly entails that there may be a shift in the subject’s relevant mode of presentation of the cup every time she comes to believe that it has or does not have a certain property. There are also two additional problems for such a view. First, there is no principled way of individuating such putative modes of presentation and linking them to human action. For, different people do not react to different colours in the same way, if at all. Someone else may be frightened to drink from a green cup but nor from a red cup or be indifferent about this. After all, it is not very common that people are frightened of some specimens of a certain artefact or natural kind and not of others as a result of their colours. Second, this would let all kinds of good and bad habits, idiosyncrasies, or irrationalities affect our modes of presentations. As noted, Prosser holds that indexical terms such as ‘here’ are best thought of as complex demonstratives, i.e. noun phrases of the form ‘that Φ’, masquerading as unstructured terms. In the case of the complex demonstrative ‘that black sheep’, the nominal term ‘black sheep’ can be detached from the demonstrative ‘that’ and placed in the predicate position.8 This enabled us to argue that in the discussed sheep case, only a singular mode of presentation is involved. On the other hand, no such detachment seems possible in the case of the indexical ‘here’. This suggests that in the latter case, we might need another mode of presentation

How Many Modes of Presentation Are There?  119 on top of the singular one. To resolve this, consider first the foregoing case in which I assent to ‘It is rainy now’ at a particular time and later on to ‘It was rainy earlier’ in order to re-express the same belief. I am in two different psychological states but this does not call for postulating two different modes of presentation here. Being in the same place, I first notice that it is raining and later on that it is no longer raining.9 My belief that the given place has a certain property has given way to a belief that it no longer has this property. The case is rather similar to the foregoing cup case in which there is no need to postulate two different modes of presentation due to a change in the subject’s belief about the cup’s colour. The only difference is that in the one case, we are dealing with a spatio-temporal object while in the other with a particular location. This is in line with (CD) and (CD’) which allow a subject to take conflicting epistemic attitudes towards the same thought at different times. We have seen that it is rationally possible for a subject who observes a certain ship to assent at a particular time to a thought expressed by ‘This ship is the Enterprise’, then change her mind (it does not matter why) about it being the Enterprise and dissent from the same thought as expressed by a subsequent utterance of the same sentence as long as she takes the ship to be the same. A similar thing happens with believing first that a given cup is green and then that is not green as a result of changing my mind about its colour as well as with believing that it is raining at a certain place which I occupy and then that it is not raining as a result of a change in weather. In both cases, I take conflicting epistemic attitudes towards the same thoughts at different times. A difference in my beliefs that leads to this is a result of the incoming information about the colour of the cup or weather at the given place, and not, as the foregoing discussion shows, of some putative additional modes of presentation of the given cup or place. According to Prosser, in replacing ‘It is rainy here’ with ‘It is rainy there’, I have moved on from one place to another, and in believing the first sentence, I have a reason to carry an umbrella, while in believing the second, I do not. The assumption is that in believing the second sentence, I have moved away from a place in which it is raining to a place in which it is not raining. But, if I have moved instead to a place in which it is also raining, I will believe about it ‘It is rainy here’ and still have a reason to carry an umbrella although I believe ‘It is rainy there’ of the previous place of which I believed ‘It is rainy here’ while I was there. A difference in indexicals in this case is not sufficient to give rise to a difference in action. What course of action I am going to take is circumstantial and cannot be systematically linked with the meanings of indexicals. I conclude that urging that there is a mode of presentation expressed by the nominal component of a demonstrative or indexical in addition to a singular mode of presentation of the item thought about is gratuitous. This conclusion rests on the denial of the claim that a difference

120  How Many Modes of Presentation Are There? in the indexical term is sufficient to give rise to a difference in action. In relation to complex demonstratives, I have also denied that sameness of the indexical term is sufficient to give rise to the same kind of action. However, with some indexicals, sameness of the indexical seems to give rise to the same kind of action: when you and I entertain ‘A bear is about to attack me’, we behave similarly. We both roll up in a ball and try to be as still as possible! When Kaplan and I entertain ‘My pants are on fire’, we also behave similarly. As we have seen in relation to Perry’s doxastic characters, characters line up with causal roles of beliefs, where the causal role of a state is meant to be ‘various combinations of factors that bring the state about, and the various combinations of factors it brings about in turn’ (Perry 1997, 20). The causal role of a state is thought of as its typical causes, the things we expect might cause an instance of that state, in more or less ordinary circumstances, and similarly with its typical effects. The typical causes of a state one is in when one exclaims ‘My pants are on fire’ are the following. One stood next to a fire, felt some unusual warmth in one’s neither regions, and smelt something like wool or cotton burning. The typical effects include strong emotions like fear, attempts to take one pants off and dispose of them or to douse oneself with water. It is in this sense that the sameness of the indexical ‘I’ can be seen as sufficient to give rise to the same kind of action. In a similar way, the indexical ‘now’ can be seen as giving rise to one kind of action each time it is entertained. In believing on different occasions that, as it were, the meeting starts now I am inclined to act in the same kind of way. The same holds for ‘here’. In believing on different occasions that, as it were, the meeting is held here I am also inclined to act in the same kind of way. But granting this does not entail that a difference in the indexical terms is sufficient to give rise to a difference in action. For, as noted, what course of action one is going to take is circumstantial and cannot be systematically linked with the meanings of indexicals. Hence, the view that in representing an object or day as the same from one occasion to the next makes the thinker think about it under the same mode of presentation even if she keeps updating her belief and changes her mind as to what features the item thought about has is left intact.10 The same applies to cases featuring personal indexicals which will be discussed in the next chapter, in relation to which Prosser aptly states: Indexicals that refer to persons… do not lend themselves to dynamic transitions between different terms in the same way that the spatial or temporal indexicals do. In particular, since one cannot become someone other than oneself, there are no circumstances in which one can keep track of oneself while exchanging indexical terms between ‘I’ and, for instance, ‘you’. (Prosser 2005, 375)

How Many Modes of Presentation Are There?  121 Prosser thinks that this does not show that the foregoing conclusions that he draws for spatial and temporal indexicals do not apply to personal indexicals. It just shows that further arguments are required in order to settle the matter. But, on the face of it, it is not clear how further arguments would enable Prosser to postulate here modes of presentation other than singular ones, but I shall not pursue this issue here.

8

Tracking and Reporting

8.1 Introduction Frege has claimed that a speech report can be correct just in case the thought referred to in the report is the thought expressed by the embedded sentence uttered on its own. Since this demand cannot be met by indexical thoughts, some philosophers have abandoned Frege’s sense/ reference framework altogether while others have tried to account for speech reporting within this framework. In finding none of these attempts satisfactory in the context that calls for Frege’s demand on speech reporting to be met, in this chapter, I provide an account of speech reporting that in effect meets this demand. In this context, I trace the role of thoughts and senses as they have been individuated in this book.

8.2 Tracking Indexicals To start with, suppose that on a given Tuesday, Jill expresses and believes it that 1

Today is beautiful

Suppose next that on the following day, Jack reports Jill by 2

Jill said that yesterday was beautiful

According to Frege, (2) is a correct report of (1) just in case the thought referred to in (2) is the thought expressed by (1). For, a sentence embedded in a non-extensional sentence does not refer to its customary referent but to the thought that it expresses when unembedded. In being (a) an objective property of the enclosed utterance, and in being (b) interpersonally stable and shareable, the thought expressed by (1) and referred to in (2) thus serves as its semantic content. This is to say that Jack’s report (2) of Jill’s (1) is correct just in case Jill’s utterance (1) expresses the same thought as Jack’s utterance of 3

Yesterday was beautiful

Tracking and Reporting  123 As we saw, Frege holds that one’s utterance of ‘today’ yesterday and one’s utterance of ‘yesterday’ today may have the very same sense (see Frege 1918/1977, 10). Hence, if on Tuesday, Jack had the same thought about it as Jill did and has retained it through to Wednesday, it is forthcoming that (2) will be a correct report of (1) which way Frege’s demand on speech reporting is met.

8.3 Personal Pronouns To draw upon an example from Frege, consider now a variant case in which Dr Lauben expresses the thought that he was wounded by uttering 4

I was wounded

and is being reported by Leo Peter by 5

You said you were wounded

In relation to this, the following passage from Frege has attracted much attention: … 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, 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 specified in this way. (Frege 1917/1977, 12–13) If we take this to entail that only Dr Lauben can grasp the thought that he is having about himself, then in reporting him by means of (5) Leo Peter cannot refer to the thought that Dr Lauben has expressed by (4). Hence, by Frege’s own lights, this report is not correct. The thought that Dr Lauben is thinking about himself is individuated in terms of the way he has himself in mind which is in line with how thoughts have been individuated in this book. But, in being neither objective nor shareable, this thought meets neither (a) nor (b) and cannot meet Frege’s demand on correct speech reporting and cannot serve as semantic content. Alternatively, we can disregard Frege’s claim that everyone is presented to himself in a special and primitive way (whether it is true or not) and opt for a different apparatus that meets both (a) and (b) as we shall see now.

8.4 ‘You’, ‘I’, and Cognitive Significance To be sure, thought is for Frege both the bearer of cognitive significance and the object (subject matter) that is believed and asserted. In view of

124  Tracking and Reporting this, and in keeping with (a) and (b), Kaplan, as noted, has teased out of Frege’s single notion of thought two different notions. One is character which corresponds to the Fregean mode of presentation of reference. To recall, character is the kind of meaning of an expression which is set by linguistic conventions, and determines the Russellian content of the expression, i.e. what is said, in every context. Similarly to Frege’s sense, character provides us both with a cognitive and semantic path to reference and is the bearer of cognitive significance. But, unlike Frege’s sense of an expression such as ‘The morning star’, character determines reference in a context-dependent way. The character of ‘I’ Kaplan describes as follows: ‘I’ refers to the speaker or writer … The phrase ‘the speaker or writer’ is not supposed to be a complete description, nor it is supposed to refer to the speaker or writer of the word ‘I’. … It refers to the speaker or writer of the relevant occurrence of the word ‘I’, that is, the agent of the context. (Kaplan 1989a, 505) The other kind of content is akin to reference and, as far as the foregoing utterances of (4) and (5) are concerned, it is the Russellian proposition, i.e. the ordered pair . In being the same in (4) and (5), it meets both (a) and (b). However, more seems to be required for a correct report in addition to expressing and believing the same Russellian content and adjusting the characters appropriately. In relation to this, Kaplan claims that an utterance of ‘I’ can have the same cognitive significance as an utterance of ‘you’. What he means is that in hearing and understanding (5), Dr Lauben takes the cognitive significance of Leo Peter’s utterance of ‘you’ in (5) to have the same cognitive significance as does his own utterance of ‘I’ in (4). The cognitive significance of the two utterances is, in Kaplan’s view, also likely to be the same for Leo Peter. But, in spite of acknowledging this intrapersonal stability of cognitive significance, Kaplan stops short of taking cognitive significance to be interpersonally the same. Granting that the utterances of ‘I’ and ‘you’ can respectively have the same cognitive significance for the hearer and the speaker, Kaplan addresses the question of whether the cognitive significance of the two utterances for the hearer is the same as the cognitive significance of the two utterances for the speaker. In reply, he says: Not if Frege is right about the first person. In a situation like this, where we are addressing one another, Frege does not expect [the hearer] to have himself in mind the way that [the speaker] has him in

Tracking and Reporting  125 mind. That sounds plausible to me, but all matters of interpersonal identification of cognitive state are difficult. (Kaplan 2012, 137) If this is right, then (5) cannot be a correct report of (4) in that (a) and (b) are not met. For (a) and (b) to be met, the cognitive significance of the utterances of ‘you’ and ‘I’ needs to be the same not just intrapersonally but also interpersonally. So, we are back to square one in spite of the fact that the Russellian content in (4) and (5) meets both (a) and (b). That a correct report should require (4) and (5) to have the same cognitive significance (as required by Frege in terms of his own apparatus) is something to which Kaplan also accedes. Kaplan notices that the fact that expressing the same Russellian content and adjusting characters in terms of ‘I’ and ‘you’ does not suffice for a communication between the speaker and the hearer to be successful and, a fortiori, for a speech report to be correct. This is suggested by Dr Lauben’s problem which in Kaplan’s view consists in his trying to communicate the thought that he expresses with ‘I have been wounded’. Under what character, asks Kaplan, must his auditor believe Dr Lauben’s thought in order for Dr Lauben’s communication to have been successful? It is important to note, claims Kaplan, that if Dr Lauben said ‘I am wounded’ in the usual meaning of ‘I’, no one else can report what he said, using indirect discourse, and convey the cognitive significance (to Dr Lauben) of what he said (Kaplan 1989a, 537–538; see also Kaplan 1979). This is to say that the sameness in cognitive significance is required for a speech report to be correct. This, in turn, requires (a) and (b) to be met. How we should account for this is the subject matter of the ensuing discussion.

8.5 Token Meanings In an attempt to come up with an account of correct speech reporting that meets (a) and (b), Sainsbury has proposed a related view. To understand Sainsbury, note that he uses ‘sense’ for Fregean senses as individuated by modes of presentation which for him cannot be individuated independently of the needs of semantics (Sainsbury 2002, 155–156). He uses ‘meaning’ for something similar to Fregean senses and takes it that correctness of reported speech is a guide to meaning. He does not argue for this natural hypothesis but explores its consequences (Sainsbury 2002, 137) Sainsbury claims that in stating that Dr Lauben has the thought that he was wounded in the passage quoted Frege implicitly suggests that in uttering ‘I’, Dr Lauben is thinking about himself under the mode of

126  Tracking and Reporting presentation which others can have, much as he does in the continuation of the same passage: But now he [Dr Lauben] may want to communicate with others. He cannot communicate a thought he alone can grasp. Therefore, if he now says ‘I was wounded’, he must use ‘I’ in a sense which can be grasped by others, perhaps in the sense of ‘he who is speaking to you at this moment’; by doing this he makes the conditions accompanying his utterance serve towards the expression of a thought. (Frege 1918/1977, 13) If this is right, then, holds Sainsbury (2002, 156, n. 3), this mode of presentation constitutes the sense of Dr Lauben’s tokens of the firstperson pronoun and is shared by appropriate tokens of the second and third-person pronouns in the mouth of others. (Note that Sainsbury (2002, 125) speaks of tokens and utterances interchangeably.) This is to say that tokens of different types such as ‘I’ and ‘you’ can have the same sense. Alternatively, and in keeping with the prevalent view that only Dr Lauben can grasp the mode of presentation of himself, Dr Lauben and Leo Peter do not share the same sense because the sense that Dr Lauben deploys cannot be shared by Leo Peter. In that case, claims Sainsbury, they rather share the same meaning. Dr Lauben’s token of ‘I’ in (4) has the same meaning as Leo Peter’s token of ‘you’ in (5). That they have the same meaning follows from Sainsbury’s natural hypothesis that correctness of reported speech is a guide to meaning, and the underlying assumption that (5) is a correct report of (4). It is understood that Frege’s demand on correct speech reporting can strictly be met just in case in (5) Leo Peter is referring to the thought that Dr Lauben is expressing by (4). But, if the token of ‘I’ in (4) has the same meaning in Sainsbury’s sense as the token of ‘you’ in (5) in conformity with (a) and (b), Frege’s demand on speech reporting is in effect met. In support of this, suppose that Frege’s demand is strictly met such that tokens of ‘I’ and ‘you’ have the same sense as is the case with the tokens of ‘today’ and ‘yesterday’ in the Jill and Jack case. Now, recall that senses as modes of presentation are in the business of storing information about the individual they are modes of presentation of. Modes of presentation are also coarse grained in that Dr Lauben and Leo Peter might think of Dr Lauben under the same mode of presentation while collecting different pieces of information about him. In reporting Dr Lauben’s (4) by means of (5), Leo Peter is likely to ascribe to Dr Lauben some features that Dr Lauben will not self-ascribe, while Dr Lauben is likely to selfascribe some features that Leo Peter will not ascribe to him. This will affect the ways in which each of them has Dr Lauben in mind without affecting the identity of the mode of presentation of him that they both deploy.1 In this respect, the situation is similar with token meanings.

Tracking and Reporting  127 Being coarse-grained, the meanings of Dr Lauben’s token of ‘I’ and Leo Peter’s token of ‘you’ are the same in spite of the difference in the ways in which each of them has Dr Lauben in mind. This meets the requirement that it is not enough for a speech report to be correct that the speaker and the reporter think about the same item. They also need to share the same sense or meaning as the case may be, which is, broadly speaking, in line with Frege’s demand on speech reporting.

8.6 Roles Strike Back: The Interpersonal Case It may be felt, though, that Sainsbury’s notion of meaning is arbitrary in that it does not fit in with how meaning is usually thought of. Meaning has been traditionally thought of as a feature of types, not tokens or utterances. The dominant tradition takes the meanings of different indexical tokens of an indexical of the same type as the same. In sticking to this tradition, Perry individuates the relevant mental states in terms of meaning types, i.e. in terms of the roles of indexicals where, to recall, role is defined as follows: When we understand a word like ‘today’, what we seem to know is a rule taking us from an occasion of utterance to a certain object. ‘Today’ takes us to the very day of utterance, ‘yesterday’ to the day before the day of utterance, ‘I’ to the speaker, and so forth. I shall call this the role of the demonstrative [indexical]. (Perry 1977/1993, 8) Roles are for Perry not part of the content expressed which is for him Russellian in that it consists of objects and properties and is the same for Dr Lauben and Leo Peter, i.e. . The role of an indexical term is the relative mode of presentation which is for Perry a rule for determining reference by its relation to an element of the context of utterance. What language associates with the indexical word is such a relative mode of presentation (Perry, Postscript to 1977/1993, 27ff). In view of this, Perry acknowledges the difference between ‘I-thoughts’ and ‘you-thoughts’, yet sees no reason to believe that ‘everyone is presented to himself in a particular and primitive way’, or at least to accept this with such a reading that it leads to incommunicable senses (Perry 1977/1993, 19). He considers a variant case to that discussed above in which Betty says to Max, ‘You are foolish’, and Max reports her correctly as ‘Betty said I am foolish’. Perry takes Betty to be thinking of Max under the relative mode of presentation corresponding to the role of ‘you’ and Max to be thinking of himself under a different relative mode of presentation corresponding to the role of ‘I’.

128  Tracking and Reporting The role of ‘I’ accounts for human action in that “when you and I entertain the sense [role] of ‘A bear is about to attack me’, we behave similarly. We both roll up in a ball and try to be as still as possible” (Perry 1977/1993, 23). 2 Conversely, Betty and Max, respectively, think of Max under two different relative modes of presentation, one corresponding to the role of ‘you’ and capturing Betty’s cognitive perspective over Max, and the other corresponding to the role of ‘I’, and capturing Max’s cognitive perspective over himself as a matter of the way each of them has Max in mind. The sense (role) that Betty associates with her utterance of ‘you’ is different from the sense that Max associates with his utterance of ‘I’. This is to say that, unlike Betty’s perspective over Max, Max’s perspective over himself is privileged although the sense via which he is presented to himself is public and communicable. Max has himself in mind from a first-person point of view while Betty has him in mind from a third-person point of view. In reporting Betty’s saying ‘You are foolish’ by means of ‘Betty said I am foolish’, Max is deploying the kind of sense that lines up with his privileged perspective over himself. However, this takes us back to square one and Dr Lauben’s problem which, as noted, consists in Dr Lauben’s trying to communicate the thought he expresses with ‘I have been wounded’. The question, in Kaplan’s words, is under what character must Leo Peter believe Dr Lauben’s thought in order for Dr Lauben’s communication to have been successful? For, if Dr Lauben said ‘I am wounded’ in the usual meaning of ‘I’, there is no one else who can report what he said, using indirect discourse, and convey the cognitive significance (to Dr Lauben) of what he said (Kaplan 1989a, 537–538). As a result, we seem to be committed to looking for a suitable correspondence relation between the way Dr Lauben has himself in mind and the way the auditor or reporter (Leo Peter) has Dr Lauben in mind.3

8.7 Roles Strike Back: The Intrapersonal Case Before I return to this issue, I want to note here that that there is also a problem with trying to account for belief retention in terms of roles. On Perry’s view, in moving from Tuesday to Wednesday, the content that Jill has expressed and believed, consisting of the given Tuesday and the property of being beautiful, stays constant. But, she no longer expresses and believes this content by means of (1) but rather by means of (3). What has changed is the role with which content is believed. On Tuesday, the content in question is expressed and believed in terms of the role of ‘today’. But the roles must change as time passes in order to express the same content. This suggests that belief retention requires a suitable correspondence relation between the role of ‘today’ in (1) and the role of ‘yesterday’ in (3). In relation to this, Bradley (2013) claims

Tracking and Reporting  129 that this is how we should account for belief retention and calls this process mutation. Mutation applies only when nothing that is uncertain is learnt. He offers the following sufficient condition on belief identity: If sentence 1 uttered at t1 and sentence 2 uttered at t2 express the same content, and the role expressed by sentence 1 has correctly mutated into the role expressed by [2], then both sentences express the same belief. (Bradley 2013, 301) In a note accompanying this Bradley says: By ‘correctly’ I mean to rule out cases in which the agent has lost track of time. For it is not obvious that Rip Van Winkle, who has unknowingly slept for 20 years, retains the belief he expressed on going to bed with ‘today is Monday’ after waking up and holding the belief expressed by ‘yesterday was Monday’. (Bradley 2013, 301–302, n. 14) Denying that a person such as Rip Van Winkle has retained the belief with which he began enables belief retention to be accounted for in terms of Perry’s roles. But we have seen that it is implausible to deny that a person who has lost track of time in this way has retained her belief. Granting that such a person has retained the relevant belief tells against accounting for belief retention in terms of Perry’s roles. To recall, in view of the Rip Van Winkle case, Kaplan finds this kind of strategy for accounting for belief retention falls short of supplying us with some obvious standard adjustment to make to the character. Given the relevant similarity between characters and roles that is clear by now, the same applies to roles. As a result, Bradley’s proposal, which is supposed to work only for the cases such as that in which the role expressed by sentence 1 has correctly mutated into the role expressed by sentence 2, bears diminished importance. The mismatch between the role of ‘yesterday’ and the belief that Rip has retained about d, the day he went to sleep, shows that the role of ‘yesterday’ is not a relative mode of presentation of d. For, Rip’s being disposed to accept ‘Yesterday was beautiful’ upon waking up is supposed to make him think of the day before he woke up as the previous day because the role of ‘yesterday’ is having a semantic fix on (i.e. denotes) that day. This is in line with the foregoing claim (P) that one’s sincere acceptance of an utterance of an appropriate indexical commits one to thinking of the day it designates (denotes) in virtue of its linguistic meaning, which was shown to be gratuitous. But, as we have seen in relation to Kaplan’s characters, after waking up, Rip thinks of d as the previous day in a looser sense. As argued, to think of d under the same

130  Tracking and Reporting mode of presentation from one occasion to the next, and to retain the belief that it is (was) a beautiful day, amounts to updating it by way of attributing to d various features, not all of which need to be true of it. The subject will typically associate with d predicates that (she thinks) are true of it. The property of being the previous day which amounts to the character, or in this case the relative mode of presentation or the role of ‘yesterday’, belongs here too though in Rip’s case it does not pertain to d. All the features the subject takes to pertain to d and d alone (together with the fact that d is the sole causal source of the subject’s belief) constitute a single mode of presentation of d. Sometimes, it is exhausted by the (fixed) character of the indexical used as, for example, in the ‘tomorrow’ case, sometimes it is not. In the cases on which Bradley is focused such as the case featuring Jill’s updating her belief expressed by (1) by means of (3), the property of being the previous day pertains to Tuesday when Jill utters and believes (3) on Wednesday. Still, the view of updating one’s belief that I have been defending leaves no room for taking it that in moving from Tuesday to Wednesday the mode of presentation under which she was thinking of Tuesday on that very day has given way to a different mode of presentation of it because it will include the property of being the previous day while no longer including the property of being the present day. The fact that on Tuesday, Jill thinks about that very day by being acquainted with it, while on Wednesday, her thinking about it draws upon her memory of it does not make a shift in the way in which she thinks about it, i.e. in the modes of presentation of it, as our discussion in the preceding chapters makes clear.

8.8 Correspondence Relations Returning to the Dr Lauben case, I have claimed that Perry’s view dealt with in Section 8.6 seems to commit us to looking for a suitable correspondence relation between the way in which Dr Lauben has himself in mind and the way the auditor or reporter (Leo Peter) has Dr Lauben in mind. If this is the case, then the Fregean demand that the thought referred to in a correct speech report is the thought expressed in the speech of which it is a report cannot be met. True, (a) is met in that the thought (i.e. the Russellian content) being referred to in the report is objective and communicable. But this is not the case with (b) which requires thought to be interpersonally stable and shareable. As a result, we need to depart from Frege’s model in those cases in which this model cannot be met and settle with a suitable correspondence relation between the speaker’s and the reporter’s thoughts. But, in looking for such a correspondence relation, we seem to be in the same position that prompted Sainsbury to introduce the notion of shared meaning. This is evidenced by Perry’s contention that a successful communication between two persons requires that their relevant utterances and beliefs be ‘internally’ about the

Tracking and Reporting  131 same thing. In the intrapersonal case, Perry helps himself to the notion of a mental file and claims that a person whose beliefs are ‘internally’ about the same thing can be seen as creating a single mental file about it (see Perry 1988/1993, 242–243). In a variant case, Recanati (2012, chap. 16) speaks of a suitable correspondence relation between the speaker’s first-person mental file and the hearer’s third-person mental file, where files play the modes of presentation role in the context of Frege’s sense/reference framework rather than Perry’s role/content framework. He takes it that the identificatory fact that Dr Lauben appeals to in order to secure reference to himself in communication belongs to a special category of identificatory facts: the category of communication-specific identificatory facts. This fact is appealed to in virtue of the linguistic meaning (type) of the indexical ‘I’ and, in being mutually known by both parties, belongs to their respective mental files. The fact that Dr Lauben and Leo Peter think of Dr Lauben differently – in terms of the first-person and the third-person mental files – is, claims Recanati, compatible with Kaplan’s foregoing claim that an utterance of ‘I’ can have the same cognitive significance as an utterance of ‘you’. What Recanati means by this is that when Dr Lauben hears ‘you’ that is addressed to him, he evokes his SELF file as the storage of information about himself, as he does when he says ‘I’. This, in Recanati’s view, shows that there is something that is shared that goes beyond reference. But Recanati contends that Dr Lauben’s original thought cannot be shared which is to concede that Fregean demands on speech reporting cannot be met both in respect of (a) and (b).

8.9 Correct Speech Reports It may be felt, though, that Sainsbury’s notion of meaning is not satisfactory in that it leads us to adopt a position that is strangely lopsided. Namely, recall that Jack’s report (2) of Jill’s (1) can be correct by Frege’s own lights provided Jack’s thought that he would express by (1) on Tuesday is the same as the thought that Jill expressed by (1) and that his utterance of (3) expresses the same thought as (1), in view of Frege’s contention that one’s utterance of ‘today’ yesterday and one’s utterance of ‘yesterday’ today may have the very same sense. On the other hand, Dr Lauben and his interlocutor or reporter can never deploy the same sense according to each of the foregoing views, including Frege’s own view. This creates an asymmetry in speech reporting. In the envisaged case, Jack’s report (2) of Jill’s (1) meets (a) and (b) in the strict Fregean sense in that the thought that Jacks refers to in (2) is the thought that Jill has expressed by (1). On the other hand, what Dr Lauben has in common with Leo Peter is not the thought that he is having about himself which is incommunicable but rather the meaning that the utterances of ‘you’ and ‘I’ have.

132  Tracking and Reporting However, the fact that this is the best we can do when reporting Dr Lauben’s (4) by means of (5) approves this. In support of this, note that the thought that Jill thinks about the given Tuesday is shareable in that it can be grasped by Jack and expressed by both of them by (1). In contrast with this, the first-person thought that Dr Lauben is thinking cannot be shared by anyone else. True, (5) could be a correct report of (4) by Frege’s own lights provided we could extend Frege’s contention that one’s utterance of ‘today’ yesterday and one’s utterance of ‘yesterday’ today may have the very same sense to the interpersonal case which way the noted asymmetry would be dispelled. Appearances aside, neither Frege, nor Evans (1985) who famously follows Frege in the intrapersonal case, claims this or gives us any resources that would enable this move.4 Evans overtly agrees with Frege that the thought that Dr Lauben expresses by means of (4) is unshareable. Still, Evans thinks that this thought is perfectly objective in that it can exist and have a truth value independently of anyone’s entertaining it, such that ‘there is no clash between what Frege says about ‘I’thoughts, and this, undeniably central, aspect of his philosophy’ (Evans 1985, 313). Nonetheless, in being unshareable by Evans’s own lights, the thought that Dr Lauben expresses by (4) cannot be the thought that is being referred to in (5). As we saw, the same holds for Kaplan in spite of his Frege-like pronouncements such as: ‘my utterance of “today” yesterday and my utterance of “yesterday” today may have the very same cognitive significance, provided I have kept track of the days correctly’ (Kaplan 2012, 137). What all these cases have in common, though, is that utterances or tokens of different indexical types can convey the same cognitive significance. The ‘today’/‘yesterday’ case attests to this as does Kaplan’s case in which utterances of ‘you’ and ‘I’ can have the same cognitive significance for a single person such as Dr Lauben. The interpersonal case is in this respect different from the intrapersonal case only in that different utterances of ‘you’ and ‘I’ can have here the same cognitive significance for two different people, which is in the foregoing discussion spelt out in terms of Sainsbury’s notion of meaning and extended to also hold interpersonally. In conclusion, a speech report can be correct just in case Dr Lauben’s utterance of (4) and Leo Peter’s report (5) involve the same sense. But Dr Lauben’s first-person thought is based on the way he has himself in mind, hence it is admittedly not shareable. What is shareable, though, in addition to reference is the meaning as a kind of cognitive significance associated with Dr Lauben’s utterance of ‘I’ and Leo Peter’s utterance of ‘you’, without denying that pronouns ‘I’ and ‘you’ belong to different meaning types. This way, Frege’s demand on speech reporting is met in that, being the same, the meaning of ‘I’ in (4) and ‘you’ in (5) is (a) is objective and (b) interpersonally stable and shareable.

9

Conclusion

9.1 Introduction In this chapter, I wish to round off my position by arguing that it is not committed to certain problems that some accounts of singular beliefs have been charged with. In so doing, I first show that indexical beliefs are singular beliefs no matter whether they are subject to acquaintance constraints or not, eschewing thus the charge that there is no good reason to impose acquaintance constraints of any sort on singular thought. Since at any rate, these beliefs differ from those beliefs that we typically express by sentences containing definite descriptions, we also bypass a time-worn debate about the relation between singular and descriptive beliefs. This also makes manifest that indexical beliefs are distinct from other kinds of belief. We also see that there is a common cognitive mechanism underlying all indexical thoughts in spite of their diversity on the surface level. Following this, I argue that in having a persisting belief about an individual, the subject is not thinking about it via a mental file, eschewing thus the charge that has been levelled at mental file accounts.

9.2 Singular Thoughts and Acquaintance It has generally been held that indexical beliefs are singular beliefs and thereby based on acquaintance. However, Hawthorn and Manley challenge this view. They claim that there is no good reason to impose acquaintance constraints of any sort on reference or singular thought. Acquaintance is for them, ‘a dispensable relic of a bygone era in the philosophy of language and mind’ (Hawthorne and Manley 2012, 25). In accordance with this, they individuate singular propositions and thoughts in terms of the following two principles: HARMONY: Any belief report whose complement clause contains either a singular term or a variable bound from outside by an existential quantifier requires for its truth that the subject believe a singular proposition.

134 Conclusion SUFFICIENCY: Believing a singular proposition about an object is sufficient for having a singular thought about it. (Hawthorne and Manley 2012, 38) These principles rule that indexical beliefs are singular beliefs whether we take them to be based on acquaintance or not. For it is not a necessary requirement on a singular belief that it be based on acquaintance. In not being subject to acquaintance constraints, and in not entailing that singular thoughts need to be Russellian propositions, the Harmony and Sufficiency Principles allow a demonstrative belief to be singular whether we take it to involve a name-like mode of presentation (as Papineau claims) or a mode of presentation that is of a distinct type. This needs to be established on an independent basis and I have tried to do this in Chapter 7. There I argued that the fact that in perceiving an individual the subject can also focus her attention on it in a dynamic and discriminating manner suggests that the subject is standing in a special (perceptual) relation with it which is in the business of storing information about it in a special way. This mode of presentation is of a distinct type from the name-like one that is based on multiple information sources. Its distinctness is the result of the subject’s standing in a perceptual epistemically rewarding (ER) relation with the object of her thought which gives rise to a distinguished epistemic input as a result of which the subject is in a position to explore the object. The perceptual ER relation in which the subject stands with the object involves her acquaintance with it. But this does not entail that all singular thoughts are based on acquaintance. A case in point is ‘tomorrow’-thoughts. To recall, on day d − 1, the subject may form a belief about d which she expresses by ‘Tomorrow will be beautiful’, although she has not been in epistemic contact with d. As we saw, this is for Perry (1997), a sourceless belief, i.e. a belief typically held about future days that cannot be part of the cause of our beliefs. Securing that demonstrative thoughts and indexical thoughts in general are thoughts of a distinct type whether they are based on acquaintance or not should also make clear that demonstrative thoughts are distinct from descriptive thoughts such as those that are expressed by means of sentences such as ‘The evening star is a body illuminated by the Sun’, taking ‘The evening star’ to be a definite description. This spares us from entering a time-worn debate about the relation between singular and descriptive thoughts that I have been steering clear of in this book.1 This also follows from the fact that indexical beliefs (mental states) cannot be assimilated to other kinds of thought. We saw that a belief that one expresses by ‘Today is beautiful’ on d will not be the same as the belief that one would express by the same form of words on d + 1

Conclusion  135 which is not the case with the sentence ‘The evening star is a body illuminated by the Sun’ or with any a sentence that is, in Frege’s words, complete in every respect. The sense of the contained expression ‘The evening star’ represents the same individual regardless of the context in which this sentence is uttered and regardless of one’s point of view. It represents Venus from a meaning-based point of view rather than from a thinker’s point of view. And the same holds for sourceless indexical beliefs. The belief that one expresses on d by ‘Tomorrow will be beautiful’ will not be the same as the belief that one would express by the same form of words on d + 1, and the like.

9.3 The Distinctness of Indexical Beliefs The foregoing discussion shows that indexical beliefs are distinct from other beliefs. Leaving aside the issue of whether there are essential indexicals in Perry’s sense in that replacing such an indexical by other terms destroys the force of the explanation, or at least requires certain assumptions to be made to preserve it, indexical thoughts cannot be assimilated to other kinds of thought. This holds for all indexicals in those uses that I have been concerned with throughout this book, ranging from the first-person singular pronoun ‘I’ to perception-based demonstratives as confirmed by our discussion of perception-based demonstratives in this chapter. As argued, the subject’s unreflective taking for granted of identity makes it the case that she is thinking about the same item from one occasion to the next under the same mode of presentation, be that a day, place, time instant, spatio-temporal object, or her own self. This, in turn, provides compelling evidence in favour of the view that there is a common cognitive mechanism underlying all indexical thoughts, in spite of their diversity on the surface level. This is to say that our ‘I’-thoughts, ‘today/yesterday’-thoughts, ‘here/there’-thoughts, ‘now/then’-thoughts, or ‘this/that’-thoughts (either as perception-based or as memory-based) turn out to be more alike than they appear to be. Similar to Campbell, these modes of presentation are coarse-grained. But, whereas Campbell only so much as acknowledges that a single demonstrative perception-based mode of presentation persists as long as the subject keeps track of an object by the same or different sensory modalities, I have argued that there is no reason to stop here and that this diminishes the importance of Campbell’s account of demonstrative thoughts. As long as the subject keeps a perceptual track of an object, the continuity of her perceptual experience is an essential part of the unreflective presupposition that the perceived object is one and the same, hence that there is only one perception-based demonstrative sense involved. But, as argued, the continuity of perceptual experience is not necessary for this. What counts is the subject’s taking for granted

136 Conclusion the identity of the object (synchronically or diachronically). Taking for granted of identity also underlies the sameness of the way of thinking of other items about which we have indexical thoughts. As we have seen, in representing a certain day or place as the same from one occasion to the next in taking it for granted that it is the same day or place the subject is thinking about it under a single (coarse-grained) mode of presentation. Similarly, if I think of myself from one occasion to the next, a thought that I can express by means of the first-person singular pronoun, I think about myself under the same mode of presentation. This is to say that our unreflective assumptions, which play a foundational role in our reflective employment of concepts, supply us with a common cognitive mechanism underlying all indexical thoughts. Unlike Campbell, Recanati and Papineau also go beyond the perceptual case. They take mental files or templates to play the role of persisting modes of presentation under which we can think of an object when it is no longer in our perceptual field as well as during our subsequent encounters with it. In following this line, I have argued that we do not need fine-grained modes of presentation on top of coarse-grained ones, as Recanati claims, and that Papineau is right in finding no room for them, but that Recanati is right in claiming that there is something distinctive about demonstrative modes of presentation which Papineau denies.

9.4 Mental Files This may seem to commit us to accounting for relevant belief states or content constituents in our sense in terms of mental files. On the face of it, thinking of an object from one occasion to the next under a persisting mode of presentation seems to amount to creating and maintaining a coarse-grained mental file about it. This seems to be supported by the fact that in thinking about an object over a period of time under the same mode of presentation, the subject may collect new information (or mis-information) about it from perception by noting that it is, say, moving, that it looks such-and-such from the back, that its shape is suchand-such, and so on, that gets stored into a mental file. Together with the rival views that I have been discussing, I take modes of presentation to have this feature. Campbell, Recanati, and Papineau take modes of presentation to have the feature of collecting information although each of them gives a different account of them (see also Kaplan 2012, 138, 169, n. 37). Perry and Lawlor hold a similar view when they speak of notion-networks. To recall, upon perceiving an object, the subject forms a notion of it. As she gathers information about it, ideas of its properties are attached to the notion, and the information is stored in a temporary perceptual buffer which may give rise to a steadier mental file. Each of these philosophers (except Campbell) holds that such information about the object gets stored in a mental file.

Conclusion  137 This ties in with the claim that the identity of the mode of presentation under which the subject is thinking of an individual that is formed during her original perceptual encounter with it is not affected by whether she is attaching new features to it in the process of thinking about it over time or withdrawing those she once took it to have, or whether these features really pertain to it. If on a later occasion, the subject’s thinking about it draws upon her memory of it, and it alone, and is aimed at it, and it alone, as its causal source, as the same individual that her original belief was about, her mode of presentation of it will be the same throughout no matter which properties she ascribes to it. Ascribing different properties to the object over time helps us take for granted the identity of the object from one occasion to the next as part of the way in which we normally represent objects and their causal powers in that we unreflectively take a continuity of our perceptual experience of the object that we perceive over time to be the result of its continuing causal power. In so doing, we typically rely on the relative stability of the object’s features such as, for example, its shape and colour, as well as location, which makes us think of it as being-thus-and-so. This kind of reliance also underlies our unreflective taking for granted of identity even without keeping our eyes continuously on the object. (Similar remarks hold for thinking about days, locations, and the like.) Appearances aside, thinking of an object under a coarse-grained mode of presentation from one occasion to the next is not to think of it via a (coarse-grained) mental file as mental files are thought of. To show this, suppose that mental files play the role of modes of presentation. Consider now Fine’s distinction between representing two objects as the same and representing them as being the same, i.e. with having the thought that they are the same. Fine says: Confusion about this distinction often arises in discussion of mental files. It is often supposed, for example, that when I learn that Cicero is Tully, the two mental files associated with “Cicero” and “Tully” merge into a single mental file. But there is a double error here. I take it that the view must be that, in representing the individuals as being the same, I thereby represent them as the same, otherwise, there is no reason to think that a merger would automatically take place. But this is not so. I can still recognize that I have learned something significant, viz. that Cicero is Tully, after being told that Cicero is Tully – which would not be so if I now represented the individuals as the same. Second, talk of merger is out of place. It is not that the merged file represents the individual as the same as the earlier files, since that would require that the earlier files represent the individual as the same. Rather, the new file, if I choose to create it, will represent the individual as being the same as the earlier files. Thus, what happens, in effect, is that I copy the contents of the earlier files to a

138 Conclusion new file and perhaps even throw away the earlier files. But what we then have is a supplementation or replacement rather than merger. (Fine 2007, 68–69) Fine is concerned here with proper names and does not deal with similar issues concerning demonstratives and indexicals, but the same kind of charge can be levelled at taking mental files to play the role of modes of presentation in the demonstrative case. However, the way I have individuated modes of presentation is not subject to Fine’s charge. I can still recognize that I have learned something significant, viz. that this ship [pointing to the bow of the Enterprise] is that ship [pointing to the stern of the Enterprise]. I may (at least for some time) invoke some or many of the features that I have previously attached to the Enterprise separately as I keep observing it from the same point of view as before such that its middle is still obscured by the large building. Admittedly, when I am no longer in the perceptual presence of the Enterprise and during subsequent encounters with it I may forget about, or no longer pay any notice to how I came to think that this ship [pointing to the bow of the Enterprise] is that ship [pointing to the stern of the Enterprise]. But I may still continue to rely on some of the properties that I ascribed to the ship separately before this realization in taking it to be the same ship from one occasion to the next whenever I fully or partially come to observe it or keep thinking about it on the basis of my memory. Relying on such features helps me keep track of it and unreflectively take it to be the same ship from one occasion to the next. That modes of presentation of the present account are not mental files is also shown as follows. It is claimed that we create a mental file of every object (or day) of which we are capable of having singular thoughts. To represent an object (day) as the same from one occasion to the next is to associate with it a single mental file that the subject can retain when she no longer perceives it and during subsequent encounters with it, as is the case with Papineau’s perceptual concepts and Recanati’s coarse-grained mental files. For Perry, the internal sameness of attribution required for a certain mental file to represent (in his example) a person as the same from one occasion to the next is provided by a cluster of predicates the subject has grouped together. They do not need a singular term to be knitted together. A quantifier phrase such as ‘some man’ or an ersatz name such as ‘a certain man’ serves the purpose (Perry 1980/1993, 84f; see also Perry 2001a, 128f, 2012, 200f).2 While modes of presentation of the present account contain related items, they do not play the role assigned to mental files as dossiers of information stored in a single “location” and do not face the status problem which Fine succinctly states as follows: … in virtue of what will information be stored in the same location or in a different location? After all, there is nothing intrinsic to the

Conclusion  139 idea of co-location which requires that co-located items should be related in any particular way. And surely, the answer to the question is that the location will be the same when the information represents its object as the same. Thus mental files should be seen as a device for keeping track of when objects are coordinated (represented as-the-same) and, rather than understand coordination in terms of mental files, we should understand the workings of mental files in terms of coordination. (Fine 2007, 68) It is in terms of such coordination that the workings of a mode of presentation are to be understood, rather than the other way around. The subject’s representing an object (day) as the same makes the associated mode of presentation keep track of this in the process of her having a continuous belief about the given object (day). As for coordination or internal sameness of attribution, it is natural to suppose that unreflective taking for granted of identity underlies it. Thinking of an item as the same (synchronically or diachronically) requires that the subject unreflectively takes its sameness for granted.3 This reveals that representing an item as the same in the kind of case with which I am concerned involves cognitive skills that are different from those cognitive skills that are involved in representing as the same an object that we refer to by utterances of proper names. To show this, note that for Fine, the sentences ‘Cicero = Tully’ and ‘Cicero = Cicero’, that contain proper names, both represent the objects as being the same, but only the second represents them as the same (Fine 2007, 40). The first sentence contains two co-referring proper names that do not sound and look the same, while the second sentence contains two co-referring proper names that sound and look the same. In other words, the sentence ‘Cicero is Cicero’ is uninformative because, as Devitt and Sterelny (1999, 70) put it, it hinges on “the law of identity.” To understand ‘is (the same as)’ is to master “the law of identity”: for any x, x is x. For, any instance of this law will be an uninformative consequence of that understanding. An instance of this law contains occurrences of the same name of the same object. Being an instance of this law, this sentence is uninformative. In contrast, the sentence ‘Cicero is Tully’ is not an instance of this law and does not seem to be an instance of it because ‘Cicero’ and ‘Tully’ sound and look different. In contrast with this, a true identity statement containing perceptionbased demonstratives that are of the same type, such as ‘this1 = this2 , may represent the object that is referred to as being the same as well as the same, as the case may be. Similarly, a true identity statement such as ‘this ship = the ship over there’, in which the given definite description functions as a demonstrative, may represent the object as the same, as well as being the same, as the case may be. Likewise, each

140 Conclusion kind of statement may be informative or uninformative, as the case may be. This suggests that representing an object as the same in the perceptual case is not constrained by the shape and the sound of the expressions by means of which we express our beliefs. It is neither sufficient nor necessary for representing an object as the same that the expressions flanking the identity sign be the same (i.e. of the same type). This reveals that our cognitive skills that are involved are more basic than those relating to the use of proper names in that representing an object as the same is not dependent on the sameness of expression types as is the case with ‘Cicero is Cicero’. In relation to this, Fine observes that it is possible to think of thoughts rather than names as intentionally representing objects as the same. Suppose I try to recall what I have learned about Cicero and think he is a Roman, he is an orator, without thinking these thoughts in those words or in words at all. Still, my two thoughts represent Cicero as the same. Fine continues: Coordination of this sort is, I believe, a pervasive aspect of our mental life. Suppose, for example, that I continuously observe an object – say a snake. I first see it coiled and later see it uncoil. The various momentary observations that make up the continuous observation then all represent the snake as the same. It is not like seeing a snake on two separate occasions and judging that it is the same. Here the series of observations actually represents the snake as the same from one moment to the next; and if the snake is not in fact the same (through some clever substitution, say), then one has suffered from a peculiar form of perceptual illusion. (Fine 2007, 67) These remarks bear striking resemblance to various claims that have been considered and defended throughout this book, and I shall not labour the point here lest I start recycling the topics that I have already dealt with. However, I conclude by pointing out, as Fine does, that coordination is a pervasive aspect of our mental life which is to acknowledge that our ground-floor level coordination concerns our perception-based thoughts, in terms of which Fine illustrates this claim. This suggests that the underlying cognitive skills are more basic than those that underlie coordination of names which is in line with the fact that the accompanying modes of presentation are a matter of our unreflective taking for granted of identity.4

Notes

Introduction 1 In relation to this, Perry says: The cognitive significance is “significance”, that is, a semantic property, having to do with meaning, reference, truth, and so on. And it is “cognitive”, that is, that aspect of meaning which is cognized by those who understand the sentence. (Perry 1988/1993, 234) 2 In relation to this, some philosophers have argued that to think of, for example, spatio-temporal objects over time is based on an ability to keep track of objects over time (see, e.g., Evans 1982, 174–175, 235–236, 311; Campbell 1987, 287f; Burge 2010, 198–199). In relation to this, Evans claims: Nobody can be ascribed at t a belief with the content ‘It is now Ψ’, who does not have the propensity as time goes on to from beliefs with the content ‘It was Ψ just a moment ago’, ‘It was Ψ earlier this morning’, ‘It was Ψ this morning’, ‘It was Ψ yesterday morning’, etc…. (Evans 1985, 309) But, Bradley counters this with a real-life case: Clive Wearing has a memory of less than 5 minutes, due to a virus that damaged his brain in 1985. For a few minutes at a time, he is perfectly normal, except for his lack of memories. If you tell him it is raining outside, he will believe you, and repeat it back if asked what the weather’s like. But he has no capacity later on to form the belief that it was raining this morning, as by then, he will have forgotten it. Presumably Evans has to say that Wearing does not really believe that it is now raining. This seems implausible. (Bradley 2013, 295–296) 3 Naturally, this is supposed to work for only one way of using the word ‘today’. If, for example, I say: ‘Never postpone till tomorrow, what you can do today’, I would not attempt to retain this thought tomorrow by the words ‘Never postpone till today what you can (could) do yesterday’. There seems to be contexts in which ‘today’ works as a simple temporal indexical (the present day) and others in which it means something like ‘whichever is your current day’. Having noticed this, I want to make clear that my concern here is only with those contexts in which ‘today’ and other related indexicals work as temporal indexicals in the sense of (RC).

142 Notes

Notes  143

144 Notes hungry’ involves a subutterance of ‘this dog’ (2012, 45–47). For simplicity sake, I ignore this distinction since it is irrelevant in the context of the present discussion. 2 Here, I follow Perry’s distinction having to do with the ‘mechanisms’ of designation. According to it, a term denotes if the conventions of language associate it with an identifying condition, and the term designates whatever object meets that condition. By contrast, a term names if the conventions of language associate it with the object(s) it designates. Another distinction that Perry draws is one between describing and referring. Although indexicals are like definite descriptions in that they denote, they refer while descriptions describe. In so doing, just like names, indexicals in Perry’s view contribute to (official) propositional content the individual they designate while descriptions contribute to such a content the identifying condition their meaning associates with them (see Perry 2001a, 30–32, 185–191, 2012, 36–39, 150–157). But we shall see that an alternative kind of content is more suitable for this role. 3 In discussing Perry’s view of cognitive significance concerning co-referring proper names, Reimer (2002) aptly remarks that in the opening paragraphs of his 1892/1980, Frege makes it clear that genuine knowledge cannot be mere meta-linguistic knowledge. We have seen that Perry wants to take differences in cognitive significance for what Frege, Kaplan, and Wettstein take them to be, but thinks that they can be taken care of by appropriate reflexive contents. Reimer might be right in claiming, contra Perry, that the co-reference problem concerning proper names can only be solved if we individuate names by their modes of presentation which she takes to be descriptive rather than relational (at least in the examples that she provides). I cannot deal with this issue here, leaving it open whether metalinguistic reflexive contents of utterances of sentences containing proper names (including the empty ones) are fit to account for full-fledged cognitive significance. 3 Anti-individualism and Cognitive Perspective 1 It is understood that the inference in this as well as in the related cases that follow involves a background presupposition that the relevant ship or ships demonstrated are in the harbour (see Stalnaker 2008, 126, n. 16). 2 See, e.g., Frege (1892/1980, 62), Evans (1982, 18–22), McDowell (1986, 142, 2005, 49), Peacocke (1986, 5), and Campbell (1987, 284, 2005, 205–206). Perry (2001a/2012, 8–9) also accepts a version of this criterion that conforms with his referentialism. I shall thoroughly deal with this criterion in Chapter 6. 3 See Wikforss (2006) who criticizes Brown along these lines in relation to co-referring natural kind terms and co-referring proper names. When it comes to demonstrative terms, there are, of course, referentialist philosophers such as Kaplan (1989) and Perry (2001a/2012) who adopt the Opacity Thesis but do not want the content believed or asserted to account for the subject’s cognitive perspective. Anti-individualists, by contrast, hold that this perspective should be accounted for by such content, and my foregoing criticism of Brown is based on this assumption. In developing a detailed non-Fregean conception of the subject’s rationality, Brown (2004, chap. 5) seems to abandon this assumption, and for this reason, this conception of rationality which merits discussion is of no consequence in the context of the present chapter.

Notes  145

146 Notes

9

10

11

12

13

14

worlds is representing about the subject’s mental states. Schroeter is concerned with standing rather than occurrent beliefs and raises different issues from those that I raise here. She thinks that the cases of Pierre and Peter should be accounted for in terms of mental files and believes that the mental file account is not hostile to Stalnaker’s externalism. Recanati speaks of mental files rather than of modes of presentation (and so does Schroeter in relation to the cases mentioned above). For the purposes of the present discussion, it suffices to speak of modes of presentation or concepts in a broad sense without being committed to any substantial theory of concepts. This way we are not exposed to objections levelled at mental file accounts such as those raised by Fine (2007, 67–69, 72) that I shall deal with in Chapter 9. Concepts as I take them to be are consonant with the way Fine accounts for “cognitive significance.” Schroeter (2013, 286) claims that depicting Boghossian’s Peter as thinking of H 2O and XYZ via the same mental file does not commit him to having transparent access to semantic facts about reference or co-reference which way Stalnaker’s anti-individualism is not threatened. I shall not pursue this issue here nor will I pursue the issue of whether invoking mental files is compatible with other varieties of anti-individualism such as Burge’s, in view of the fact that Burge wants to reconcile anti-individualism with a Fregean account of thought-content (see Burge 1979, 1986, 1993). For a discussion of Burge’s and related views, see Brown 2004, chap. 5, Wikforss 2006, and Schroeter 2008. As to whether having access to thought-contents involves having meta-beliefs (see Schroeter 2007, Recanati 2012, chap. 10, Wikforss 2015). I do not make any claims as to whether Stalnaker’s view of epistemic transparency is capable of meeting the requirements that it is supposed to meet in Stalnaker (1978) in relation to conversational situations. For a discussion of this and related issues, see Hawthorne and Magidor 2009, 2011, Stalnaker 2009, Almotahari and Glick 2010. A ‘water’/‘twater’ case is (in Stalnaker’s view) one such case in which a subject such as Boghossian’s Peter has been (slowly and) unknowingly switched from a world in which the substance called ‘water’ is H 2O to a world in which the substance called ‘water’ is XYZ. This is not to say that, according to Evans, no cognition takes place here. In the case in which the subject is having an illusion of an object, Evans says that it is not part of his proposal that the subject’s mind is wholly vacant; images and words may clearly pass through it, and various ancillary thought may even occur to her (Evans 1982, 45–46). It may be argued that there is also an additional difficulty for the view that the subject is here thinking no thought at all. If the subject believes of the same object (or of two different objects) that this1 ≠ this2 , she is credited with deploying two different senses as a matter of taking herself to be representing two different objects. Similarly, if the subject believes of a single object that this1 = this2 , she is credited with deploying a single sense as a matter of taking herself to be representing a single object. By contrast, in the case in which the subject confuses two different objects and takes it that this1 = this2 , her taking herself to be representing a single object is in the relevant circumstances not accounted for by any senses and thoughts (thought-contents) at all but is taken to be a matter of her suffering an illusion of thought. This deprives thoughts of their role in accounting for the subject’s actions and behaviour in the latter kind of case. But, Evans might reply that thought is not deprived here of its role, when it has a role; and

Notes  147

15

16

17

18 19

moreover, in accounting for the relevant kind of case, there is the “illusion of thought” that serves to account for the subject’s behaviour. It can be also objected that resorting to the “illusion of thought” takes away this position’s motivation. According to this position, thoughts generally account for the subject’s cognitive perspective, yet they are left out when it comes to some specific cases. And in the one case, the subject’s taking herself to be representing an object involves her thinking a thought of a particular kind, whereas in the other case, it does not. I will not pursue these issues here. Contesting Evans’s and Campbell’s “illusion of thought” claim and saddling them with the view that the subject can think two thoughts with different contents without realizing this a priori, Brown (2004, 215) has claimed that keeping track of each of two objects for a substantial period of time enables the subject to have a thought apiece about each of the objects at the time of her latter thought episode – one memory-based, the other perception-based – without noticing that they are different. But this seems to be the case in which, according to Evans, the original senses are recoverable, such that when the subject re-deploys them, she will be able to grasp a priori that they differ, contrary to what Brown would have us believe. It is to be noted that the point that I want to make is not affected whether we accept it that the subject may deploy the two original senses after she realizes that a single ship is in play or not, and I will not pursue this issue here. A related issue concerning mental files will be discussed in Chapter 9. In keeping with the conceptual apparatus deployed by Evans and Campbell, I spell out co-reference de jure in terms of deploying a single sense but it can also be spelt out in terms of, e.g., Fine’s (2007) semantic coordination (see also Pryor 2016). I shall return to this issue in Chapter 9. De jure coreference has recently been subject to various debates in relation to utterances that contain occurrences of noun phrases that co-refer in a guaranteed and evident way (see, e.g., Pinillos 2011, Goodsell 2014). For a recent discussion of the relation between Fregeanism and semantic relationism as advocated respectively by Fine (2007) and Pryor (2016), see Gray (forthcoming). Since Pryor and Gray are not concerned with demonstratives or indexicals in general, I cannot deal with this issue here. In Chapter 9, I thoroughly deal with problems that beset the notion of a merger in relation to Fine’s criticism of mental file accounts. See Recanati (2012, chap. 7; 2013), who distinguishes immediate recognition from slow recognition. In immediate recognition, identity is presumed without being conceptually articulated in the form of an identity judgement, such that there is a single mode of presentation (based on the composite relation of re-acquaintance), while in slow recognition, the subject judges that the seen object is the remembered object, making it the case that there are two different modes of presentation that are linked. I shall discuss Recanati’s view on recognition in Chapter 7.

4 Cognitive Dynamics, Belief Retention, and Cognitive Significance 1 Evans finds implausible the conception according to which belief retention amounts to having a succession of different beliefs. According to this conception, which Evans calls atomistic, the thoughts associated with sentences containing temporal indexicals cannot be grasped at later times. A persistence of a belief is held to be a succession of different but related beliefs concerning the same time. Evans finds this objectionable. He does acknowledge that these sequences of beliefs are very common – that human beings do

148 Notes have a general propensity, on forming one belief in this series, later to have the other beliefs in the series. Yet, it is unclear how one belief is supposed to give rise to another. One belief cannot give rise to another by any inference, since the belief that would be required to underwrite the inference is not a thinkable one. No sooner does one arrive in a position to grasp the one side of the belief than one has lost the capacity to grasp the other (Evans 1985, 309). For questioning some of Evans’s key claims here, see Bradley 2013. I want to note here that while being unsympathetic to the atomistic conception in relation to temporal indexicals, Evans accepts it by allowing two different subjects to successfully communicate by sharing not the same, but different, suitably related modes of presentation. Furthermore, somebody who would temporarily lose track of time just before midnight on d and be unsure whether midnight has passed but retrieve it shortly after midnight on d + 1, would in Evans’s view, first lose and then retrieve his belief. This is to say that he would either retrieve the same belief token or acquire a different one suitably related to it. The former option is implausible because belief tokens, once they have ceased to exist, cannot be retrieved but only replaced with other suitably related ones, while the latter is inconsistent with Evans’s view that belief-retention is the retention of the same belief token. 2 Kaplan (1989a) equates the notion of the character of an indexical expression with that of its linguistic meaning. But, as noted above, Braun (1996) has argued that as far as demonstratives such as ‘that’ are concerned, linguistic meaning does not amount to character, since various occurrences of ‘that’ can have different characters while retaining the same linguistic meaning. In ‘Afterthoughts’ (1989b, 598), Kaplan admits that in the case of proper names, character is not an adequate substitute for cognitive role. For, names such as ‘Hesperus’ and ‘Phosphorus’ have the same character in spite of their having different cognitive roles. These issues, however, have no bearing on the present discussion. 3 More recently, Kaplan has claimed that ‘[a] tracking error – misrecognition of the referent of “yesterday” as a different day, as Rip Van Winkle might do – is a more serious disorder…’ (Kaplan 2012, 169, n. 37). Surely, various things might conspire to deprive Rip from having the belief with which he began, but the kind of case that I am considering is one in which Rip has not confused the day he went to sleep with any other day by way of having been acquainted with it, as will become clear shortly. 4 While Recanati (2012) claims that for definite descriptions we need a threefold distinction between denotation, semantic reference, and speaker reference, what we need for indexicals, which are bound to be used referentially, is just a distinction between the cases in which an indexical is used to refer to its denotation, and the cases like Rip Van Winkle, in which an indexical is used to refer to something other than its denotation (speaker’s reference). What was said above makes it clear that other possible options that have been suggested to me regarding Rip and belief retention are implausible. They can be summarized as follows: (i) thanks to retaining his belief about d, Rip’s latter belief is about the day before he woke up although he takes it to be about d; (ii) in addition to having a belief about d, Rip has a belief about the day before he woke up that is based solely on the belief with which he began plus malfunctioning updating. As a result, Rip’s updating does not amount to retaining the belief with which he began but to replacing it with a new belief about the day before he woke up. Claiming that updating amounts to replacing, based on malfunction, does not rule out Rip’s retaining the belief with which he began; (iii) Rip is sufficiently confused that he

Notes  149

5

6

7

8

fails to retain any beliefs about d if we assume that there are conflicting reasons to say that the belief he has after waking up is about d (since it is a descendant of the belief with which he began) and reasons to say that it is about the day before he woke up (since it is a belief about the day before he woke up). See, e.g., Frege (1892/1980, 62, 1906/1979, 197). Many philosophers, including neo-Fregeans such as Evans (1982, 18–22), McDowell (1986, 142; 2005, 49), and Campbell (1987; 2005, 205–206), take this criterion (or some version of it) to be crucial for individuating relevant beliefs (and other attitudes) as featured in propositional attitude psychology. In a similar vein, Schiffer (1978, 180) speaks of Frege’s Constraint on modes of presentation, claiming that any candidate must satisfy this constraint in order to qualify as a mode of presentation; while Perry (2001a/2012, 8–9) speaks of a cognitive constraint on semantics and accepts a version of it that conforms with his referentialism. As noted, Perry has rightly urged against Wettstein that it is part of the business of semantics to deal with cognitive significance. But, we have seen that Perry’s framework is ill-suited to account for cognitive significance in relation to the co-reference problem. In the next chapter, I shall also pinpoint the problems that Perry faces in accounting for belief retention. While thinking a single thought about d over a period of time, the subject may first think that d was beautiful and later on change her mind about this due to, say, her misremembering the weather on d. It is in this sense possible for her to take the thought she is thinking first to be true and then to be false, but in doing so, she does not fail to re-identify it. This possibility does not arise, though, when the subject’s different thought episodes take place at the same time. This is a requirement that Evans (1982) builds into Frege’s principle. To this, I shall return in Chapter 6. It is to be noted that by Campbell’s own lights, this is not a simultaneous but rather a diachronic matter when tracking thoughts are involved. For he claims that if one does succeed in keeping track of an object over time, then one must know immediately that it is the same thing that is in question (Campbell 1987, 285). Dummett approvingly ascribes to Frege the view that we are omniscient about senses (Dummett 1981b, 51). But Brandom (1994, 571) wonders whether Frege was ever committed to this view (which Brandom is himself unsympathetic with). Almog saddles Frege with this view when he argues against Frege’s (alleged) contention that identity and distinctness judgements vis-à-vis senses we are omniscient about are to explain the informativeness of identity judgements vis-à-vis worldly objects about which we are not omniscient (Almog 2008, 567). On a different note, Almog believes that the notion of cognitive significance, although pre-theoretically available, is not of theoretical use. But, as we have seen above in relation to Wettstein – who makes a similar claim but within a different “conceptual framework” – cognitive significance is of a piece with the internal continuity of the subject’s belief, and it needs to be dealt with in accounting for belief retention – even if it turns out that it is not part of the business of semantic theory. To illustrate this, suppose that I am in the habit of watering my flowers with the same amount of water each day, but I lose track of the days in the way suggested. Not realizing on Wednesday that the thought I am thinking and expressing by ‘Yesterday was beautiful’ is about Tuesday, I end up thinking that I have watered my flowers on one of the two days that I respectively take Tuesday for but not on the other. To redress the balance, I decide to water

150 Notes

9 10

11

12

my flowers with double the amount of water – something I would not do if I were to keep track of the days and think a single thought from Tuesday through to Wednesday. This case significantly differs from the one on which the subject falsely identifies different ships, discussed in the last chapter. For, unlike Rip, such a subject has a confused belief that has its causal source in each of the ships. A similar response can be given to a similar objection that someone might raise concerning the foregoing case in which on Tuesday, I form an indexical belief that I express by ‘Today is beautiful’, and being unaware that midnight has passed, I utter the same sentence again with the intention of re-expressing that very belief. Someone might claim that, in conformity with (P), in the given circumstances, I believe the false Today1 = Today2 , i.e. that Tuesday = Wednesday. The response is that this is not right. I rather take the two respective utterances of ‘today’ to refer to Tuesday in virtue of their linguistic meaning. For I intend to re-express the belief I formed on Tuesday about Tuesday in spite of the fact that I come to be in epistemic contact with Wednesday. The emerging picture of persisting modes of presentation bears some resemblance to a number of views holding that having a singular thought of an object (in our case of a day) consists in creating a mental file about it by way of lumping together the information the subject collects about it (see, e.g., Perry 1980/1993, 84f; Recanati 2012). I shall discuss the relationship between these views and the view I have been advancing in Chapter 9. See Bradley (2013) for a more detailed discussion of Lewis’s view and the problems it faces.

5 Beliefs and Characters 1 In taking Frege’s sense to be an ideal procedure for determining reference, Perry follows Dummett (1981a, 589ff), but one can wonder whether there is anything in Frege’s writings to suggest such a verificationist leaning on his part. Taking Fregean sense to be such an ideal procedure may only obscure the fact that Frege’s theory of meaning and thought can account for the cases of belief-retention that we shall be looking at in the next chapter. 2 Cf. Kaplan ( 1989b, 588), who claims that in the case of indexicals such as ‘today’, the relevant contextual feature is always the referent. 3 See Luntley (1997) who argues that dynamic perception-based thoughts are not definable by rules of linguistic meaning even when these are rules of applied linguistic meaning, but rather in terms of the subject’s egocentric point of view. These rules alone, that is, do not suffice for the construction of such thoughts. 6 Slicing Thoughts 1 As noted, on a similar note, Schiffer speaks of Frege’s Constraint on modes of presentation, claiming that any candidate must satisfy this if it is to qualify as a mode of presentation. It says that necessarily, if m is a mode of presentation under which a minimally rational person x believes a thing y to be F, then it is not the case that x also believes y not to be F under m (Schiffer 1978, 180). However, Salmon (1989, 205) has claimed that for this principle to be adequate for the purposes of dealing with belief ascriptions concerning beliefs involving (the bearers of) proper names, it is better to replace it with a pair of principles which together entail the cited principle. The first, which

Notes  151 Salmon calls ‘Frege’s Thesis’, may be stated (using Schiffer’s theoretical apparatus and terminology) as: if x believes y to be F, then there is an object m that is a mode of presentation of y and x believes y under m to be F. The second principle, which Salmon calls ‘Schiffer’s Constraint’, is: if a fully rational person x believes a thing y under a mode of presentation m to be F and also disbelieves y under a mode of presentation m’ to be F, then m ≠ m’ and x construes m and m’ as modes of presenting distinct individuals. However, this has no bearing on the subject matter of my foregoing discussion. Later on, Schiffer offers a somewhat different formulation of Frege’s constraint and divides it in two parts. First, it says that a rational person x may both believe and disbelieve that a certain thing or property y is such and such only if there are distinct modes of presentation m and m’ such that x believes y to be such and such under m and disbelieves it to be such and such under m’. Then it says that there are distinct modes of presentation m and m’ such that a rational person x believes y to be such and such under m and disbelieves y to be such and such under m’ only if x fails to realize that m and m’ are modes of presentation of one and the same thing. This is to say that one cannot rationally believe and disbelieve something under one and the same mode of presentation, or under modes of presentation that one realizes are modes of presentation of the same thing (see Schiffer 1992, 502–503). A case in which the subject realizes that m and m’ are modes of presentation of the same thing is a case discussed above in which the subject deploys two different senses (at least for some time) after learning that a = b and in so doing does not disbelieve that a = b. 2 Whether the subject will actually take two such object-parts to be parts of the same object or not will inter alia depend on her discriminatory abilities (including perceptual errors) as well as on her background beliefs and expectations. 3 In 2002, however, Campbell claimed that demonstrative sense was given by conscious attention, focused by experienced location of the object thought about and Gestalt organization (90 ff). This is his alternative to Evans (1982), who claims that knowing the objective location of the object is both necessary and sufficient to provide the subject with demonstrative sense by way of furnishing the subject with discriminating knowledge of the object. (See Campbell 2002, 109–113 for the differences between the two views regarding this issue.) On the face of it, Campbell’s binding sense to location requires sense to change with location, which leads to a proliferation of senses. If this is correct, it is a departure from the view he held in his 1987 where he individuates senses in terms of keeping track of things whether their location changes or not. Whether Campbell is committed to this view depends on what we make of his taking the complex parameter used in binding together all the information related to the object to involve higher-order linkage between locations as the object the subject is keeping track of moves over time (Campbell 2002, 90). For a discussion of the role played by perceptual attention in forming singular beliefs about the perceived objects (see also Dickie 2010, 2015). 4 This is not to say that two thoughts are the same if a rational subject does not at the same time take conflicting epistemic attitudes towards them. To say that there is but one thought due to the fact that the rational subject fails to take such conflicting epistemic attitudes is problematic. The subject may be unfocused, or distracted, or cognitively overburdened in some way, or it may be simply too costly to bother discriminating items at some moment. On reflection, she would perhaps discriminate, but does not in fact do so.

152 Notes

5

6

7

8

9

Allowing conditions of distraction, over-burdened condition, or simple cost savings to affect the individuation of thought is to make thought an inappropriate category for normative theorizing about, say, psychological matters or inference. This suggests that we should instead hold that if a rational subject is not (on reflection) in a position to take conflicting epistemic attitudes towards two thoughts at the same time, then they are the same, but as we shall see below (in relation to (CS)), this is also problematic. We have seen that Campbell’s account of demonstrative thoughts is also diminished in importance if a proliferation of senses is allowed to occur in those cases in which our demonstrative thinking of objects goes beyond keeping track of them in Campbell’s sense. Brown claims that qua Fregeans neo-Fregeans such as Evans and Campbell assume transparency of sameness of content while qua anti-individualists, that is, by claiming that the identity (and existence) of appropriate thoughts are tied to the identity (and existence) of objects they are about, they deny transparency of difference of content. As noted, in Brown’s view (2004, chap. 6), this asymmetry has to do with the Fregean rationality constraint contained in (CD) that one cannot at the same time rationally take conflicting epistemic attitudes towards the same thought-content, which she finds ill-motivated. However, her charge is bypassed if in the kind of case in question, the subject is taken to be suffering an illusion of thought or to be thinking an equivocal thought or a thought that fails to refer. To be sure, confused beliefs are also recognized as beliefs of a special kind by the referentialists, who hold that singular thought-contents are Russellian propositions containing only objects and properties as their constituents. For when it comes to confused beliefs, no such proposition is expressed by a sentence containing a singular term that fails to refer (uniquely). But since there is still something going on in the cognitive life of a subject who assents to such a sentence, efforts have been made to account for this. As noted in Chapter 2, Perry (2001a/2012) thus explains the cognitive perspective of a subject using an empty proper name in terms of notion-networks, which Lawlor (2007) extends to cases involving confused beliefs. To recall, such a network is based on an idea or a notion the subject forms of an object she takes herself to be perceiving, which has no origin in cases in which there is no such object or if she has confused two or more different objects for a single one. An alternative suggested by some anti-individualists is to allow a subject to rationally take conflicting epistemic attitudes towards the same thoughtcontent by taking sameness of thought-content to be non-transparent (see, e.g., Owens 1990; Brown 2004). But then thought-content no longer serves to rationalize the subject’s cognitive perspective (see Kimbrough 1998; Wikforss 2006). The fact that thoughts conform to (CD’) does not entail that their identification needs to be elucidated by mentioning this criterion, as Peacocke has pointed out in relation to (CD). He claims that it is open for a Fregean to hold these two theses simultaneously: (i) it is a condition of adequacy on a theory of content (thought) that the contents it recognizes conform to (CD) and (ii) that a substantive theory of content which explains what it is for a thinker to be capable of judging a given content should not appeal directly to this criterion, but should rather entail it (Peacocke 1986, 5).

7 How Many Modes of Presentation Do We Need? 1 The point that he makes can be rephrased without committing ourselves to the existence of such templates, in terms of the subject’s representing an

Notes  153

2 3

4

5

6 7

8

9

10

object as the same from one occasion to the next. The modes of presentation that persist in this way play the same role that Papineau’s concepts are supposed to play. To this I shall return in Chapter 9. See Onofri (2015) for objecting (in relation to Recanati 2012) that mental files cannot simultaneously play both these roles. See also Recanati’s (2016, chap. 5) discussion of Onofri’s view and his reply to Onofri. See Recanati (2012, chap. 10) and Boghossian (1994, 39–40) whom Recanati follows here. As noted, Boghossian defines transparency as follows: (a) if two of a thinker’s token thoughts possess the same content, then the thinker must be able to know a priori that they do and (b) if two of a thinker’s token thoughts possess distinct contents, then the thinker must be able to know a priori that they do (Boghossian 1994, 36, 2011, 457). In line with this, Evans (1982, 226–227) has claimed that the subject can think demonstratively about an object without invoking the concepts of perception and of the self. Furthermore, in order to acquire the concepts of perception and of the self, the subject already needs to have demonstrative concepts. Thus, grasping the concept of perceptual experience requires of the subject to have a capacity to self-ascribe perceptual experiences. This draws upon the same concepts, including demonstrative ones, that she uses in forming beliefs about the external world. This shows that indexical thoughts that comprise such a mode of presentation are special which is in line with the noted fact that they cannot be assimilated to other kinds of thought. This does not entail, though, that they are representationally essential in Perry’s sense. Hence they are not subject to the charge that Cappelen and Dever (2013) level at Lewis and Perry discussed in Chapter 1 above. This may be taken to render groundless Ball’s (2015) claim that there are no indexical thoughts. See Recanati (2016, 5.3) for his response to Ball. As mentioned above, Cappelen and Dever note in a variant case, ‘… nothing general follows about the connection between indexicality and agency. All kinds of difference in beliefs can make a difference to what it is rational to do’ (Cappelen and Dever 2013, 48). In effect, this has been done by accounts that see (1) and (2), respectively, as a combination of two beliefs. According to these accounts, propositions associated with (1) are: (1a) That is a black sheep and (1b) That can run fast; and similar for (2). See, for example, Dever (2001) and Corazza (2002) as well as Prosser (2005, 378). Note that no part of the utterance ‘It is rainy now’ or ‘It was rainy earlier’ designates the place that it is about. This is in line with Perry’s (1986/1993) suggestion that Prosser (2005, 382) makes use of in a slightly different context that an utterance at time t in place p of ‘it is raining’ expresses the proposition rain (t, p) even though no part of the utterance designates p which makes p an unarticulated constituent of the utterance. In his 2019 work, Prosser argues that in the process of indexical communication, modes of presentation can be shared by two thinkers in all cases (see also Dickie and Rattan 2010 for a similar point). When perceptually based demonstratives are deployed, the sameness of the mode of presentation is guaranteed by the fact that thinkers can interpersonally trade on identity just as in the intrapersonal case because their tokens of ‘that’ in ‘That is a wasp’ in the presence of a mutually identified wasp are “locked together.” As in the intrapersonal case in which the subject may change her mind over time as to which properties the object does or does not possess, two thinkers may have different beliefs about the object (wasp), yet think about it under the same mode of presentation. On the face of it, this is in line with the way I have outlined a way in which the speaker and the hearer communicate in

154 Notes Sections 5.4 and 6.6 as well as with the points made above about collecting new information about the perceived object (cup). However, in the interpersonal case concerning indexicals such as ‘here’ and ‘now’, Prosser once again needs to resort to two modes of presentation discussed above. Since it is the singular mode of presentation that, in Prosser’s view, accounts for the interpersonal trading on identity, this view can be seen as supplementary to my view which does not so much as develop a general account of interpersonal cases, as long as we steer clear of the other mode of presentation that Prosser postulates. In the next chapter, I will discuss the issue as to whether successful communication requires the thinkers to share the same mode of presentation. Here, I want to note that, in addition to indexicals, this issue also concerns other terms. For a recent discussion, see Valente (2019). 8 Tracking and Reporting 1 The same applies to the case in which Jill and Jack express the same thought by means of (1). Similar remarks have been made in Section 5.4. As for the case featuring perception-based demonstratives, in Section 6.6, it was noted that as long as the speaker and the hearer (unreflectively) represent an object as the same, they will think of it via the same shared sense regardless of whether they are observing it via the same or different parts and regardless of whether they are attaching the same properties to it and collecting the same information about it. 2 We saw that Perry (1979/1993) holds that “perspectival representations” are representationally essential, i.e. that there are essential indexicals that as such differ from all other referring terms. To recall, Cappelen and Dever (2013) contest this, but this does not affect my arguments above nor does their claim that nothing general follows about the connection between indexicality and agency and that all kinds of difference in beliefs can make a difference to what it is rational to do (Cappelen and Dever 2013, 48). 3 This is in line with the frequently stated condition that a description or ascription of the subject’s belief states needs to be sensitive to the fact that her ways of thinking of an individual need not be the same as those of the ascriber. Richard (1990) tackles this kind of issue in terms of belief-contents which contain both objects and properties (as in a Russellian proposition) and linguistic expressions in a language of thought which represents them. He calls them Russellian annotated matrixes (RAMs), claiming that the ascriber’s RAM needs to be related in a certain context-restricted way to one of the believer’s RAMs for an ascription to capture the believer’s way(s) of representing the object’s and properties that her belief is about. Similarly, in dealing with what figures in the content of perceptual beliefs, Loar suggests a way in which an ascription of such a belief can, without mentioning a particular mode of presentation, only indicate its general type. In this, he follows Schiffer (whose theory he adjusts to include as modes of presentation non-descriptive ones in addition to the descriptive ones that Schiffer [1978] considers): Take a sentence S of the form ┌x believes that …y…┐, where the singular term y occurs de re (i.e. refers to a real object the belief is about). S implicitly refers to types of modes of presentation: ‘there is a mode of presentation m, of the contextually relevant type, which determines y, and x believes that …m….’ (Loar 1981, 98)

Notes  155

156 Notes it is subject to when we try to apply it to other kinds of thought, as noted above in relation to Mates. Mates’s criticism is levelled at general terms such as ‘fortnight’ in the foregoing example. In line with Mates, Fine (2007, 131) claimed that the Fregean has been hoisted by his own petard (see also Sainsbury and Tye 2012, 78). Mates’s criticism of (CD) does not seem to apply to ordinary proper names since it is not clear if there are (co-referring) proper names that are synonymous in the foregoing sense for it seems that, unlike general terms, proper names cannot be given a definition. (For a definition of general terms, as well as for a criticism of a Fregean explanation of the co-reference problem concerning general terms, see Fine 2007, 130.) The foregoing considerations make it clear that demonstrative thoughts are more basic than the thoughts we express by means of sentences containing proper names. Furthermore, Mates’s (and Fine’s) criticism levelled at the Fregean explanations does not affect my account of demonstrative (indexical) thoughts, and I shall not pursue these issues here.

References

Almog, J. 2008. Frege puzzles? Journal of Philosophical Logic 37, 549–574. Almotahari, M. and Glick, E. 2010. Context, content, and epistemic transparency. Mind 119, 1067–1086. Bach, K. 1992. Intentions and demonstrations. Analysis 52, 140–146. Ball, D. 2015. Indexicality, transparency, and mental files. Inquiry 58, 353–367. Bermudez, J. L. 2017. Yes, essential indexicals are really essential. Analysis 77, 690–694. Boghossian, P. 1994. The transparency of mental content. Philosophical Perspectives 8, 73–122. Boghossian, P. 2011. The transparency of mental content revisited. Philosophical Studies 155, 457–465. Bradley, D. 2013. Dynamic beliefs and the passage of time. In Capone, A., Feit, N. (eds.), Attitudes De Se. Chicago, IL: University of Chicago Press, 291–306 Brandom, R.1994. Making it Explicit. Boston, MA: Harvard University Press. Branquinho, J. 1999. The problem of cognitive dynamics. Grazer Philosophische Studien 56, 29–56. Braun, D.1996. Demonstratives and their linguistic meaning. Noûs 30, 145–173. Brown, J.2004. Anti-individualism and Knowledge. Cambridge, MA: MIT Press. Burge, T. 1979. Individualism and the mental. Midwest Studies in Philosophy 4, 73–122. Burge, T. 1986. Intellectual norms and the foundations of mind. Journal of Philosophy 83, 697–720. Burge, T. 1993. Concepts, definitions, and meaning. Metaphilosophy 24, 309–325. Burge, T. 2010. Origins of Objectivity. Oxford: Clarendon Press. Campbell. J. 1987. Is sense transparent? Proceedings of the Aristotelian Society 88, 273–292. Campbell, J. 1994. Past, Space, and Self. Cambridge, MA: MIT Press. Campbell, J. 2002. Reference and Consciousness. Oxford: Oxford University Press. Campbell, J. 2005. Information processing, phenomenal consciousness, and Molyneaux’s question. In Bermudez, J. L. (ed.), Thought, Reference and Experience, Themes from the Philosophy of Gareth Evans. Oxford: Clarendon Press, 195–219 Cappelen, H. and Dever, J. 2013. The Inessential Indexical; on the Philosophical Insignificance of Perspective and the First Person. Oxford: University Press.

158  References Carnap, R. 1947. Meaning and Necessity, A Study in Semantics and Modal Logic. Chicago, IL: The University of Chicago Press. Carruthers, P. 1987. Russellian thoughts. Mind 96, 18–35. Corazza, E. 2002. ‘She’ and ‘he’: politically correct pronouns. Philosophical Studies 111, 173–196. Corazza, E. 2004. Reflecting the Mind. Oxford: Oxford University Press. Dever, J. 2001. Complex demonstratives. Linguistics and Philosophy 24, 271–330. Devitt, M. and Sterelny, K. 1999. Language & Reality. 2nd edition. Oxford: Blackwell. Dickie, I. 2010. We are acquainted with ordinary things. In Jeshion, R. (ed.), New Essays on Singular Thought. Oxford: Oxford University Press, 213–245 Dickie, I. 2015. Fixing Reference. Oxford: Oxford University Press. Dickie, I. and Rattan, G. 2010. Sense, communication and rational engagement. Dialectica 64, 131–151. Dokic, J. (ed.) 1997. Cognitive Dynamics, European Review of Philosophy 2. Stanford, CA: CSLI Publications. Donnellan, K. 1966. Reference and definite descriptions. The Philosophical Review 75, 281–304. Dummett, M. 1981a. Frege, Philosophy of Language. London: Duckworth. Dummett, M. 1981b. The Interpretation of Frege’s Philosophy. London: Duckworth. Evans, G. 1982. The Varieties of Reference. Oxford: Clarendon Press. Evans, G. 1985. Understanding demonstratives. In Evans, G. (ed.), Collected Papers. Oxford: Clarendon Press, 291–321 Fine, K. 2007. Semantic Relationism. Oxford: Blackwell. Frege, G. 1892/1980. On sense and meaning. In Geach, P., Black, M. (eds.), Translations from the Philosophical Writings of Gottlob Frege. Oxford: Blackwell, 1980, 56–78. Frege, G. 1906/1979. A brief survey of my logical doctrines. . In Hermes, H., Kambartel, F., Kaulbach, F. (eds.), Posthumous Writings. Oxford: Blackwell, 1979, 197–202. Frege, G. 1918/1977. Thoughts. In Geach, P. T. (ed.), Logical Investigations. Oxford: Blackwell, 1977, 1–30. Goldberg, S. C. (ed.) 2015. Externalism, Self-knowledge, and Skepticism. Cambridge: Cambridge University Press. Goodman, R. 2016. Against the mental files account of singular thoughts. Review of Philosophy and Psychology 7, 437–461. Goodsell, T. 2014. Is de jure coreference non-transitive? Philosophical Studies 167, 291–312. Gray, A. (forthcoming). Indistinguishable senses. Noûs. Hawthorne, J. and Magidor, O. 2009. Assertion, context, and epistemic accessibility. Mind 118, 377–397. Hawthorne, J. and Magidor, O. 2011. Assertion and epistemic opacity. Mind 119, 1087–1105. Hawthorne, J. and Manley, D. 2012. The Reference Book. Oxford: Oxford University Press.

References  159 Jeshion, R. 2010. Singular thought: acquaintance, semantic instrumentalism, and cognitivism. In Jeshion, R. (ed.), New Essays on Singular Thought. Oxford: Oxford University Press, 105–140. Kaplan, D. 1968. Quantifying in. Synthese 19, 178–214. Reprinted in Davidson, D., Hintikka, J. (eds.) Words and Objections, Essays on the Work of W.V. Quine. Dordrecht: D. Reidel, 1969. Kaplan, D. 1979. On the logic of demonstratives. In French, P., Uehling Jr. T. E., Wettstein, H. K. (eds.) Contemporary Perspectives on the Philosophy of Language. Minneapolis: University of Minnesota Press, 401–412.Kaplan, D. 1989a. Demonstratives. In Almog, J., Perry, J., Wettstein, H. (eds.), Themes from Kaplan. Oxford: Clarendon Press, 481–563. Kaplan, D. 1989b. Afterthoughts. In Almog, J., Perry, J., Wettstein, H. (eds.), Themes from Kaplan. Oxford: Clarendon Press, 564–614. Kaplan, D. 2012. An idea of Donnellan. In Almog J., Leonardi, P., (eds.) Having in Mind, The Philosophy of Keith Donnellan. Oxford: Oxford University Press, 122–175. Kimbrough, S. 1998. Anti-individualism and fregeanism. The Philosophical Quarterly 48, 470–482. Korta, K. and Perry, J. 2011. Critical Pragmatics. Cambridge: Cambridge University Press. Kripke, S. 1979. A puzzle about belief. In Margalit, A. (ed.), Meaning and Use. Dordrecht: Reidel, 239–283. Lalor, B. 1997. Rethinking Kaplan’s “Afterthoughts” about ‘that’: an exorcism of semantic demons. Erkenntnis 47, 67–88. Lawlor, K. 2005. Confused thoughts and modes of presentation. The Philosophical Quarterly 55, 21–36. Lawlor, K. 2007. A notional worlds approach to confusion. Mind & Language 22, 150–172. Lewis, D. 1979. Attitudes de dicto and de se. The Philosophical Review 88, 513–543. Lewis, D. 1980. Index, context and content. In Kanger, S., Ohman, S. (eds.), Philosophy and Grammar. Dordrecht: Reidel, 79–100. Loar, B. 1981. Mind and Meaning. Cambridge: Cambridge University Press. Luntley, M. 1997. Dynamic thoughts and empty minds. In Dokic, J. (ed.), Cognitive Dynamics, European Review of Philosophy 2. Stanford, CA: CSLI Publications. McDowell, J. 1984. De re senses. The Philosophical Quarterly 34, 283–294. McDowell, J. 1986. Singular thought and the extent of inner space. In Pettit, P., McDowell, J. (eds.), Subject, Thought, and Context. Oxford: Clarendon Press, 137–168 McDowell, J. 2005. Evans’s Frege. In Bermudez, J. L. (ed.), Thought, Reference and Experience, Themes From the Philosophy of Gareth Evans. Oxford: Clarendon Press, 42-–65 Mates, B.1950. Synonymity. University of California Publications in Philosophy 25, 201–226. Millikan, R. G. 2000. On Clear and Confused Ideas; An Essay about Substance Concepts. Cambridge: Cambridge University Press.

160  References Noonan, H. W. 1986. Russellian thoughts and methodological solipsism. In Butterfield, J. (ed.), Language, Mind & Logic. Cambridge: Cambridge University Press, 67–90. Onofri, A. 2015. Mental files and rational inferences. Inquiry 58, 378–392. Owens, J. 1990. Cognitive access and semantic puzzles. In Anderson, A. C., Owens, J. (eds.), Propositional Attitudes: The Role of Content in Logic, Language and Mind. Stanford, CA: CSLI Publications, 147–173. Papineau, D. 2006. Phenomenal and perceptual concepts. In Alter, T., Walter, S. (eds.), Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism. New York: Oxford University Press, 111–144. Papineau, D. 2013. Comments on François Recanati’s Mental Files. Disputatio 36, 159–175. Peacocke, C. 1983. Sense and Content. Oxford: Oxford University Press. Peacocke, C. 1986. Thoughts, an Essay on Content. Oxford: Blackwell. Perry, J. 1977. Frege on demonstratives. In Perry 1993, 3–32. Perry, J. 1979. The problem of the essential indexical. In Perry 1993, 33–52. Perry, J. 1980. A problem about continued belief. In Perry 1993, 69–90. Perry, J. 1986. Thought without representation. In Perry 1993, 205–225. Perry, J. 1988. Cognitive significance and new theories of reference. In Perry 1993, 227–247. Perry, J. 1990. Individuals in informational and intentional content. In Perry 1993, 279–300. Perry, J. 1993. The Problem of the Essential Indexical and Other Essays. Oxford: Oxford University Press. Perry, J. 1997. Rip van Winkle and other characters. In Dokic, J. (ed.), 13–39. Perry, J. 2001a. Reference and Reflexivity. Stanford, CA: CSLI Publications. Perry, J. 2001b. Knowledge, Possibility, and Consciousness. Cambridge, MA: MIT Press. Perry, J. 2009. Directing intentions. In Almog, J., Leonardi, P. (eds), The Philosophy of David Kaplan. Oxford: Oxford University Press, 187–201. Perry, J. 2012. Reference and Reflexivity. 2nd edition. Stanford, CA: CSLI Publications. Perry, J. 2013. Self-locating beliefs. In Feit, N., Capone, A. (eds.), Attitudes De Se: Linguistics, Epistemology, Metaphysics. Stanford, CA: CSLI Publications, 385–408. Perry, J. 2015. Indexicality. In Segal, R., von Stuckrad, K. (eds.), Vocabulary for the Study of Religion, vol. 2 Leiden: Brill, 288–289 Pinillos, A. 2011. Coreference and meaning. Philosophical Studies 154, 301–324. Prosser, S. 2005. Cognitive dynamics and indexicals. Mind & Language 20, 369–391. Prosser, S. 2015. Why are indexicals essential? Proceedings of the Aristotelian Society 115, 211–233. Prosser, S. 2019. Shared modes of presentation. Mind & Language 34, 465–482. Pryor, J. 2016. Mental graphs. Review of Philosophy and Psychology 7, 309–341 Pylyshyn, Z. 1999. Is vision continuous with cognition? The case for cognitive impenetrability of visual perception. Behavioral and Brain Sciences 22, 341–365. Recanati, F. 1993. Direct Reference, From Language to Thought. Oxford: Blackwell.

References  161 Recanati, F. 2012. Mental Files. Oxford: Oxford University Press. Recanati, F. 2013a. Perceptual concepts: in defence of the indexical model. Synthese 190, 1841–1855. Recanati, F. 2013b. Mental files: replies to my critics. Disputatio 36, 207–242. Recanati, F. 2016. Mental Files in Flux. Oxford: Oxford University Press. Reimer, M. 1991. Do demonstrations have semantic significance? Analysis 51, 177–183. Reimer, M. 2002. Review of Perry, John Reference and Reflexivity. Notre Dame Philosophy Reviews. Retrieved June 23, 2019 from https://ndpr.nd.edu/news/ reference-and-reflexivity/. Richard, M. 1990. Propositional Attitudes: An Essay on Thoughts and How We Ascribe Them. Cambridge: Cambridge University Press. Sainsbury, R. M. 2002. Departing from Frege: Essays in the Philosophy of Language. London: Routlegde. Sainsbury, R. M. and Tye, M. 2012. Seven Puzzles of Thought and How to Solve Them: An Originalist Theory of Concepts. Oxford: Oxford University Press. Sainsbury, R. M. and Tye, M. 2015. Counting concepts: response to Paul Boghossian. In Goldberg, S.C. (ed.) 2015. Salmon, N. 1989. Illogical belief. In Salmon, N. (ed.), Content, Cognition, and Communication. Oxford: Oxford University Press. Originally published in Philosophical Perspectives 3, 243–285. Schiffer, S. 1978. The basis of reference. Erkenntnis 13, 171–206. Schiffer, S. 1992. Belief ascription. The Journal of Philosophy 89, 499–521. Schroeter, L. 2007. The illusion of transparency. Australasian Journal of Philosophy 85, 597–618. Schroeter, L. 2008. Why be an anti-individualist? Philosophy and Phenomenological Research 77, 105–141. Schroeter, L. 2013. Are concepts creatures of darkness? Analytic Philosophy 54, 277–292. Stalnaker, R. 1978. Assertion. In Stalnaker, R. (ed.), Context and Content. Oxford: Oxford University Press, 1999, 78–95. Originally published in Cole, P. (ed.), Syntax and Semantics 9: Pragmatics, 315–332. Stalnaker, R. 1984. Inquiry. Cambridge: Cambridge University Press. Stalnaker. R. 2008. Our Knowledge of the Internal World. Oxford: Oxford University Press. Stalnaker, R. 2009. On Hawthorne and Magidor On assertion, context, and epistemic accessibility. Mind 118, 400–409. Strawson, P. F. 1959. Individuals. London: Methuen. Valente, M. 2019. Communicating and disagreeing with distinct concepts: a defense of semantic internalism. Theoria 85, 312–336. Wettstein, H. 1986. Has semantics rested on a mistake?’ The Journal of Philosophy 83, 185–209. Wettstein, H. 2004. The Magic Prism, An Essay in the Philosophy of Language. Oxford: Oxford University Press. Wikforss, A. 2006. Content externalism and Fregean sense. In Marvan, P. (ed.), What Determines Content? The Internalism/Externalism Dispute. Newcastle: Cambridge Scholars Press, 163–179 Wikforss, A. 2015. The insignificance of transparency. In Goldberg, S.C. (ed.) 2015.

Index

Note: Page numbers followed by “n” denote endnotes. acquaintance 19, 70, 133–134, 147n19 action 3–4, 14, 27–30, 32, 44, 48, 55, 60, 73, 78, 84–85, 87, 89, 93, 95, 116–120, 128 Almog, J. 3, 149n7 Almotahari, M. 146n11 anti-individualism 4, 39–40, 43–46, 50, 55–56, 58, 61, 144n3, 144n4, 144n7 atomistic conception of temporal thinking 148n1 Bach, K. 143n6 Ball, D. 153n6 beliefs: indexical 6–8, 13–16, 19, 30, 68, 76–78, 80–82, 115, 133–135; retention 5–8, 13, 16–19, 63, 65–70, 76–77, 84, 128–129; sourceless 73–74, 134 (see also content) Bermudez, J. L. 142n9 Boghossian, P. 40, 45, 54–55, 72, 145n5, 145n7, 145n8, 146n10, 146n12, 153n3 Bradley, D. 13, 128–130, 141n2, 147n1, 150n12, 155n4 Brandom, R. 149n7 Branquinho, J. 82 Braun, D. 143n4, 148n2 Brown, J. 18, 40–44, 55, 60–61, 72, 82, 144n3, 145n7, 146n10, 147n15, 152n6 Burge, T. 39–40, 141n2, 145n4, 146n10 Campbell, J. 9, 16, 40–42, 58, 71–72, 76, 85, 90, 93–96, 98, 101, 103, 105–108, 110, 113–114, 135–136,

141n2, 144n2, 145n7, 147n15, 147n17, 149n5, 149n6, 151n3, 152n5, 152n6 Cappelen, H. 15, 112, 153n7, 154n2 Carnap, R. 21, 100 Carruthers, D. 98 Cartesianism 54–55, 72 causal roles 77, 80, 120 characters 11–13, 18–19, 21–25, 29–32, 66–67, 69–71, 74–75, 114–115, 120, 124–125, 128–130; doxastic 77–84, 86–88, 120 cognitive: dynamics 5, 7–8, 17–19, 63–64; significance 1, 3–5, 7–11, 13, 15–19, 21–26, 27–32, 34–35, 37, 63, 65–67, 69–71, 73, 76, 82, 88–89, 91, 93, 97, 118, 123–125, 128, 131–132; value 2–4, 21–24, 37, 84 concepts 53–55, 59–61, 72 content: notional 36–37; official 28, 33, 35, 38, 144; reflexive 28–35, 37, 84, 89, 144; Russellian 3–4, 12–13, 22, 27–28, 34–35, 39, 66–68, 75, 77–78, 124–125, 127, 130, 134; semantic 1, 6, 19, 122–123 coordination 106, 139–140, 147n17 Corazza, E. 142n9, 153n8 co-reference: de jure 106, 147n17 (see also trading on identity); problem 16, 18, 27, 30–33, 37, 144n3, 149n5, 155n4 criterion of difference for thoughts 10–11, 16, 43–44, 58–61, 71–72, 88, 90–94, 96, 98, 100–101; modal 105–113; non-modal 106–9, 111, 113 criterion of sameness for thoughts 94, 96, 98

164 Index demonstrations 23–25, 30–32, 34 demonstratives vs. pure indexicals 23; see also indexicals definite descriptions 6. 11–12, 22–23, 32, 34, 69, 88, 124, 133–134, 139 Dever, J. 15, 112, 153n5, 153n6, 153n7, 153n8, 154n2 Devitt, M. 139 Dickie, I. 151n3, 153n10, 155n3 Donnellan, K. 69–70 Dummett, M. 149n7, 150n1 Evans. G. 9–10, 16, 40, 56–58, 68, 71, 82–83, 85, 90–91, 93, 95, 98–101, 103–104, 106, 111, 132, 141n2, 144n2, 145n5, 146n13, 146n14, 147n15, 147n1 Fine, K. 137–140, 142n10, 146n9, 147n17, 147n18, 155n4 Frege, G. 1–6, 8–13, 15–17, 19, 22–23, 27–29, 43, 59, 63–70, 76, 78, 88, 90, 98, 104–106, 109–110, 122–127, 130–132, 135 Fregean 6, 9–13, 16–18, 22–23, 32, 37–45, 56, 58, 60–61, 63–66, 70, 72, 76, 78–79, 88, 90–91, 93, 124–125, 130–131 Frege’s “retention claim” (RC) 8–9, 13, 16, 64, 66–68, 70–71 Glick, E. 146n11 Goodman, R. 155n1 Goodsell, T. 147n17 Gray, A. 147n17 Hawthorne, J. 133–134, 146n11, 155n1 indexicals: essential 14–16, 135, 142n9, 153n5, 154n2 individualism see anti-individualism intentions, directing 24–25, 31, 34, 69 Jeshion. R. 96 Kaplan D. 3–5, 9–13, 16, 18, 21–24, 26–30, 32–33, 35, 38–39, 63–69, 71, 75–79, 81–82, 84, 88, 118, 120, 124–125, 128–129, 131–132, 136, 142n6, 142n7, 142n2, 143n3, 143n4, 143n5, 143n6, 148n2, 148n3, 155n2 Kimbrough, S. 152n8

Korta, K. 4, 24 Kripke, S. 145n8 Lalor, B. 24 Lawlor, K. 4, 18, 27, 35–37, 136 Lewis, D. 13–16, 76 Loar, B. 154n3 Luntley, M. 98 McDowell, J. 9, 40, 91, 144n2, 149n5, 154n3 Magidor, O. 146n11 Manley, D. 133–134, 155n1 Mates, B. 100, 155n4 matrices, two-dimensional 51 memory 65, 72, 75, 80, 102–106, 109–110, 112–114, 137–138 mental files 20, 35, 104–109, 110– 113, 131, 133, 136–139 Millikan, R. G. 57, 99 modes of presentation 2, 5–8, 22, 26, 53–55, 58–59, 67, 72–76, 78–79, 84–88, 90, 102–110, 111–114. 115–121, 124–131, 134–140 narrow psychological states 23–24, 77 naturalism 91, 101 nominal terms 116–119 Noonan, H. W. 95 notion-networks 35–37, 136 Onofri, A. 153n2 Owens. J. 152n8 Papineau, D. 104, 110–113, 134, 136, 138 Peacocke, C. 40, 91, 144n2, 152n9 perceptual buffers 28–30, 32–35, 38, 136 Perry, J. 3–5, 11–16, 18–19, 21–26, 27–35, 36–38, 39, 41, 51, 67, 69, 73, 77–85, 87–89, 114–116, 118, 120, 127–131, 134–136, 138, 141n1, 142n7, 144n1, 144n2, 144n3, 149n5, 150n1, 152n9, 154n2 personal pronouns 66, 120–121, 123–128, 130–132 Pinillos, A. 147n17 places, thinking of 116, 119, 153n9 possible worlds 13–14, 32–34, 51, 55 proper names 10, 27–28, 36, 69, 138–140, 142n4, 142n8, 144n2, 144n3, 148n2

Index  165 propositional attitudes 13, 34, 38, 91, 100 Prosser, S. 116–121, 142n9, 153n8, 153n9, 153n10 Pryor, J. 147n17 Pylyshyn, Z. 96 Rattan, G. 153n10 Recanati, F. 41, 53, 57, 69–70, 99, 103–109, 111–115, 131, 136, 138, 148n4, 153n2, 153n6, 155n1 recognition 62, 102–106, 109–110, 112–113 referentialism 3–4, 6, 12, 27–28, 34–35, 37–39, 66 re-identification of objects 6, 19, 62, 72, 102; see also recognition relationism, semantic 147n17 Reimer, M. 143n6, 144n3 representations, indexical 1, 7, 15, 153n5, 154n2 Richard, M. 92, 107, 143n4, 154n3 roles 11–13, 21–22, 25, 78, 114, 116, 127–131 Sainsbury, R. M. 85, 108, 125–127, 130–132, 145n5, 155n4 Salmon, N. 150–151 Schiffer, S. 149n5, 150n1, 154n3 Schroeter, L. 55, 145n8, 146n9, 146n10 semantics 2–3, 5, 11, 27–28, 31–32, 67, 89, 125, 149n5; twodimensional 51 senses 1–3, 5, 8–13, 16–17, 22–23, 32, 37–38, 40–42, 44, 56–62, 63–65, 67, 69, 71–72, 76, 78–79, 81, 90–97, 98–101, 114–116, 122–128,

131–132, 135; see also modes of presentation singular: terms 23, 41, 76, 111, 133, 138; thoughts 6, 19, 40, 133–134, 138 speech reports 6, 19, 122–128, 130–132, 154n3 Stalnaker, R. 18, 39, 41, 44–50, 51–56, 57–61, 72, 91, 101, 142, 145n5, 145n6, 145n8, 146n10, 146n11, 146n12 Sterelny, K. 139 Strawson, P. F. 95, 99 synonymity 100, 156n4 taking for granted the identity 6–7, 57, 90–91, 95–98, 100, 113, 115, 135–137, 139–140 thoughts see beliefs times, thinking of 116, 119–120 trading on identity 76, 100, 105–107, 110, 116–118 transparency of thought 11, 39–43, 44–49, 51–52, 54–56, 58–62, 71–72, 80 Tye, M. 145n5, 155n4 Valente, M. 153n10 validity 41–47, 49–51, 53–54, 56–57, 59–60, 145n5 Van Winkle, Rip 5, 67–71, 74–75, 77, 79–82, 84, 86–87, 129–130, 148n3, 148n4, 150n9 verificationism 150n1 Wettstein, H. 3, 27–28, 33, 67, 144n3, 149n5, 149n7 Wikforss, A. 144n3, 146n10, 152n8