UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The linguistic u-turn in the philosophy of thought Fleming, Michael Neil 1999

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1999-388867.pdf [ 12.64MB ]
Metadata
JSON: 831-1.0089231.json
JSON-LD: 831-1.0089231-ld.json
RDF/XML (Pretty): 831-1.0089231-rdf.xml
RDF/JSON: 831-1.0089231-rdf.json
Turtle: 831-1.0089231-turtle.txt
N-Triples: 831-1.0089231-rdf-ntriples.txt
Original Record: 831-1.0089231-source.json
Full Text
831-1.0089231-fulltext.txt
Citation
831-1.0089231.ris

Full Text

THE LINGUISTIC U-TURN IN THE PHILOSOPHY OF THOUGHT by Michael Neil Fleming M . Soc. Sci., University of Waikato, 1994 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE F A C U L T Y OF G R A D U A T E STUDIES Department of Philosophy We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH C O L U M B I A April 1999 © Michael Neil Fleming, 1999 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my writ ten permission. Department of fL^ y f o s o ^Ly The University of British Columbia Vancouver, Canada Date 2 « IXp^i f 1W DE-6 (2/88) 11 Abstract A central task of contemporary analytic philosophy is to develop an understanding of how our minds are connected to the external (or mind-independent) world. Arising from this task is the need to explain how thoughts represent things in the world. Giving such an explanation is the central endeavor of this dissertation—the aim being to contribute to our understanding of what it is for a subject to be thinking of a particular object. The structure of the dissertation is set, in part, by responding to the commonly held view that a satisfactory explanation of what it is to think of a particular object can be drawn out of, or extended from, an explanation of what it is to be referring to that particular object. Typically, in investigating these matters, it is accepted that there is an explanatory priority of language over thought. This is the Priority Thesis. Some take the Priority Thesis to reflect an appropriate methodological strategy. In this form, it implies the methodological point that the best way to describe thoughts is by describing them as they are expressed in language. Most, however, seem to take the Priority Thesis to be symptomatic of a substantive, metaphysical truth. This, to put it one way, is that the content of a thought is paralleled by the content of the associated linguistic expression. I call this the Assumption of Parallelism. This characterizes what we call Linguistic Turn philosophy (i.e., analytic philosophy). The body of the dissertation arises out of questioning the extent of the application ofthe Priority Thesis in developing theories of reference and thought. I call the move of the partial overturning ofthe Priority Thesis the Linguistic U-Turn. The overall conclusion is that we cannot explain what it is to think of a particular object by extending explanations of what it is to be referring to that particular object. In particular, I reject what I call the Causal Theory of Thought—the view that the representational properties of a thought are explained by the referential properties of the appropriate singular term. My aim, then, is to show that a popular conviction concerning the representational properties of thoughts about things in the world is not warranted. I l l Acknowledgments I gratefully acknowledge the support and invaluable feedback of my supervising committee: Professor Catherine Talmage, Professor Paul Bartha, and Professor Tom Patton—with special thanks to Professor Patton for maintaining his supervising role after his retirement. Each, in their own way, made me produce a much better dissertation than I would have without their contributions. iv T A B L E O F C O N T E N T S Abstract >< Acknowledgments ' i i I N T R O D U C T I O N 1 C H A P T E R 1: The Linguistic T u r n and The Linguistic U - T u r n 9 1.1 The Linguistic Turn and the Genesis of Analytic Philosophy 10 1.2 Frege 12 1.3 Reservations 15 1.4 The Priority Thesis 18 1.5 Linguistic Turn Philosophy: Some Contemporary Examples 19 1.6 Evans and The Linguistic U-Turn 26 1.7 Peacocke 34 1.8 Partial Commitments to the Linguistic U-Turn 37 C H A P T E R 2: Thoughts and Experience: Some Preliminaries 43 2.1 Subjective and Objective Thoughts 44 2.2 Experiences • 49 2.3 Object Independent Thoughts 51 2.4 Particular Thoughts 53 2.5 Object Independent Thoughts vs. Particular Thoughts 55 C H A P T E R 3: Reference 59 3.1 The Indexical Character of Singular Reference 62 3.2 Direct Reference 69 3.3 Evans on Reference 78 3.4 Referential Communication: Some Preliminaries 87 3.5 Producers and Consumers 89 3.6 Reference and Identifying Knowledge 93 C H A P T E R 4: Referential Communicat ion 98 4.1 The Nature of Communication 99 4.2 The Fregean Model of Communication 102 4.3 Grice 103 4.4 Evans on Communication 109 4.5 Connections L13 V CHAPTER 5: Particular Thought Outlined 116 5.1 The Requirements for a Theory of Thought 118 5.2 Russell's Principle 121 5.3 Explaining Russell's Principle 127 5.4 The Challenge to Russell's Principle: The Photograph Model 134 5.5 The Defence of Russell's Principle: Rejection of the Photograph Model 140 5.6 Information-based Thought 147 CHAPTER 6: Particular Thought Defended 151 6.1 Blackburn on Evans 155 6.2 Carruthers on Evans 171 6.3 Psychological Explanation in the Philosophy of Thought 180 6.4 The Representational Quality of Thoughts 185 CHAPTER 7: Two Forms of Mental Representation 193 7.1 Mental Representation: The Standard Picture 196 7.2 A Nonstandard Picture 199 7.3 Egocentric Mental States 207 7.4 The Ontological Project 214 7.5 Back to the Photograph Model 220 CHAPTER 8: The Taxonomy of Thoughts 223 8.1 The Standard for the Taxonomy of Thoughts 224 8.2 Internalism about Thought Individuation 226 8.3 Externalism about Thought Individuation 229 8.4 Evaluation of Externalism: What Twin Earth Arguments Do Not Show 232 8.5 Substitutivity Failures in Belief Contexts 238 SUMMARY AND CONCLUSION 255 Reference List 259 1 INTRODUCTION A widely held view is that at least some of our thoughts are genuine representations of, or are directed at, individual objects, properties, and/or relations in a mind-independent world. A dominant concern of contemporary analytic philosophy is the development of an understanding of how this can be so. This has generated many interesting tasks. What is of immediate concern is explaining how our thoughts come to. have this representational character. (How does something that I have in mind stand for something outside my mindl) Within this task we have the special problem of explaining how a thought can be a representation of a particular thing. That is, what makes a thought represent one thing and not some other superficially similar thing? For example, what makes the thought 'That tree is big' represent, or be about, a particular tree and not some other tree that appears the same? This final problem is the central concern of this dissertation. The objective of this dissertation, then, is to address some important issues in the philosophical investigation of the nature of the representational properties of thoughts—in particular, thoughts that are about, or concern, individual objects, properties or relations in the world. With respect to this, the aim is to contribute to our understanding of what it is for a subject to be thinking of a particular thing. In other words, to explain what it is to possess a Particular Thought. A further idea, however, helps to establish the structure of the dissertation. This comes from the commonly held view that thinking of an object is intimately connected to referring to that object. The idea is that a satisfactory explanation of what it is to think of a particular object can be drawn out of, or extended from, an explanation of what it is to be referring to a particular object when using an appropriate piece of a language. It is because of this that I begin the substantive discussion by looking, first, at theories of reference, and, for reasons that should become clear, referential communication. I will argue, however, that for at least some thoughts, reference to an object does not explain thought about the object. 2 In narrowing the focus as I do to thoughts about (and reference to) individual things, it is important to note here at the beginning that I am not claiming that this somehow exhausts all kinds of investigations of things that might be legitimately labeled "thoughts." For instance, most would readily agree that there are thoughts about sets of things (e.g., All cats walk softly), and thoughts arising from inductive inferences (e.g., The litmus paper will turn red in acid), which we might call General Thoughts. Also, that there are thoughts about logical relations and mathematical relations and entities (e.g., 8 is greater than 4). It is contentious what these thoughts are 'about', if anything. Given this, their status as one kind of thought or other may change accordingly (for instance, depending on whether one is a mathematical Platonist). However, the issue of the status of these thoughts is not important here. What is important is that it be noted that this investigation concerns only thoughts that are about, or concern, individual things in the world. This dissertation is founded in a certain type of approach to the investigation of thoughts—namely, that which has been an important focus this century of what has come to be known as analytic philosophy. It is important to see, then, what this amounts to for our inquiry. Most agree that a grounding assumption of analytic philosophy is that we are to explain the content and structure of thoughts in terms of the content and structure of language representations of these thoughts. Michael Dummett calls this the "Priority Thesis."1 Some take this to reflect an appropriate methodological strategy. In this form, the Priority Thesis implies the methodological point that the best way to describe thoughts is by describing them as they are represented in language. Most, however, seem to take the Priority Thesis to be symptomatic of a substantive, metaphysical truth. This, to put it one way, is that the content of a thought is paralleledby the content of the associated linguistic expression. I call this The Assumption of Parallelism. In general, then, primacy is given to the philosophy of language over other branches of philosophy, in particular, here, epistemology and the philosophy of thought. The primacy of the 1 Dummett (1991), p. 315ff. See Pataut (1996) for a discussion of this point. 3-philosophy of language is often characterized by the expression 'Linguistic Turn philosophy' (i.e.. analytic philosophy). Linguistic Turn philosophy is described, briefly, in the first part of Chapter 1. This claim about the order of explanation—explaining thought in terms of reference—can be made only under the methodological and metaphysical assumptions of the Priority Thesis. One ofthe general claims of the dissertation, then, is that once we look closely at what is involved in referring to a particular object (Chapter 3), and especially at referential communication (Chapter 4), we see that a satisfactory explanation of reference to a particular object can be made only by first explaining what it is to be thinking of that particular object. This reverses the methodology suggested by the Priority Thesis. It implies also a rejection of the Assumption of Parallelism, though as it turns out, the reversal leads to only a partial rejection. The claim, keeping it very general for the moment, is that only some thoughts are not paralleled by associated sentences. I am not rejecting everything that has been born out of taking the Linguistic Turn. In fact, much is retained. Readers familiar with contemporary writings in this area may see that my claim of questioning the methodological and metaphysical assumptions ofthe Priority Thesis is not so bold, or so radical, as initially might be thought. It seems to me that many working in this area have already taken some steps that suggest at least a partial, perhaps tacit, questioning ofthe assumptions of the Priority Thesis.2 All of these ideas began to take shape when I first encountered some recent ideas that questioned the Priority Thesis, both as it is reflected in an acceptance of a methodological strategy, and as it results in the substantive, metaphysical claim of the Assumption of Parallelism. The body of the dissertation arises, then, out of questioning the extent of the applicability of the Priority Thesis in developing theories 2 I'm reluctant to give a list of names because most readers will probably want to know why such and such is a good example, and I can't give that here. However, if I don't give some names, some readers may be suspicious of the claim. So, at the risk of creating some hard questions, I give the following short list. In different ways, 1 take Tyler Burge (1977) (regarding his account of belief de re); Brian Loar (1981), Stephen Schiffer (1987), John Barwise and John Perry (1983), and Nathan Salmon (1986) (regarding their employment of Pragmatics); and Kent Bach (1987) (regarding his notion of 'non-descriptive modes of presentation') as being good examples. 4 of reference and thought. Once we do this, we can begin to see how we fare in taking this move with respect to, for example, explaining the individuation of thoughts. I call the move of partially overturning the Priority Thesis The Linguistic U-Turn. It is characterized, briefly, in the second part of Chapter 1. In adopting the strategies of the Linguistic U-Turn I follow the work of Gareth Evans. There are, however, points at which I appear to deviate from the position that Evans held. I cannot be certain of this for two reasons. First, and obviously, it might be that I have failed to appreciate the ideas that Evans advocated. Second, Evans's major work, The Varieties of Reference, was published posthumously, after his early death, with substantial parts of the book being put together by its editor, John McDowell. Many agree, McDowell included, that the work is, in some important respects, incomplete. It may be that the ideas of a completed version either would adequately address concerns I raise, or convince me that I was wrong to deviate from what Evans held. In any case, it will become increasing clear, as the dissertation takes shape, that I owe a great debt to Evans's work. Taking the Linguistic U-Turn with respect to the philosophy of thought, along with noting that standard analyses (i.e., those grounded upon acceptance of the Priority Thesis) get something crucially right, forces, I think, a distinction between kinds of thoughts, or perhaps more accurately, kinds (or forms) of mental events. In Chapter 2,1 distinguish between three kinds of mental events that play important roles in the cognitive aspect of our lives. Two of these have been, or might otherwise be, labelled 'thoughts'. I give a precis here to help with the introductory outlines of the arguments of latter chapters. First, there are Experiences, which result from bringing into consciousness, perceptual informational content, i.e., the informational content of a perceptual state. Experiences are relevant to the investigation because we do have thoughts about experiences—that is, thoughts produced by directing our attention inward, so to speak, at what is presented to our minds in perceiving something.3 It is important to keep Experiences in mind because they play crucial roles in many discussions. Second, there are 3 See section 2.2 for more on the difference between experiences and thoughts about experiences. 5 Object-Independent Thoughts. These are thoughts that concern, or purport to be about something, but have no grounding in a particular object or context. These thoughts are best understood by contrasting them with Particular Thoughts. Particular Thoughts are thoughts about particular things (objects, properties or relations) in the world. Particular Thoughts are genuinely representational because they are grounded directly (i.e., non-derivatively) in the thing in the world which they are about. If the subject of a thought has no appropriate connection (e.g., in perception or memory) to the thing thought about, there is no Particular Thought; but there may be an Object-Independent Thought. An important claim of this dissertation is that there is significant benefit in distinguishing between thoughts that are about particular things, and thoughts that are not. Also, even though Object-Independent Thoughts are not genuinely representational, we retain the label 'thoughts' (contrary to Evans) principally for the reason that this satisfies the intuitions that we have which are based upon our phenomenological experiences of the 'goings on' on our own minds. In other words, given the kind of thing that Object-Independent Thoughts are, it would seem to be inappropriate not to call them thoughts. Within this three-way distinction, most important is the distinction between Object-Independent Thoughts and Particular Thoughts. The crucial difference between them is that Particular Thoughts are genuinely representational, whereas Object-Independent Thoughts are not. My suggestion is that this distinction allows a clearer understanding of the nature of the representational properties of thoughts. In other words, when we understand the difference between Object-Independent Thoughts and Particular Thoughts we have a clearer idea about what makes a thought genuinely representational. The conclusion is that what makes it the case that a subject's thought is about a particular object (or person, or state of affairs) is not some causal relation, but rather that the subject possesses a certain kind of knowledge. This is discriminating knowledge, to use Evans's phrase. It is that which connects the subject's thought with the particular object thought about, and allows the subject to individuate the object. 6 The theory of thought advocated is outlined and discussed in Chapters 5, 6 and 7. In Chapter 5, I provide a sketch of Particular Thought, drawing on Evans's work. These ideas have not received wide approval, or, it seems, wide agreement about their importance. However, there are several important works that discuss Evans's ideas. In Chapter 6,1 introduce some of these and offer partial responses to criticisms. I do note, however, that some of what these critics say is worthy of our attention. That is, some modification to Evans's overall position is suggested. This modification is set up in Chapter 7, along with some recent ideas that I take to be supportive of Evans's position. My suggestion is that these, together, allow us to take on board some of the what the critics suggest while accepting the major components of Evans's position. In giving a theory of thought we must remember that we are giving at least part of what would amount to a theory of mental representation. In general, a theory of mental representation must give the conditions under which it can be said that some mental event (i.e., a thought) represents a particular object or state of affairs. This aside, though, it is quite well known that, more specifically, there are two crucial things which a theory of mental representation must do. First, we know that sometimes we get things wrong. Sometimes, perhaps more often than we are aware, we /wwrepresent the world. An adequate theory of mental representation (and so, thought) must be able to account for this. Second, we know, also, that a subject's thoughts (or beliefs or desires) can be interpreted in different ways depending upon how the thoughts are individuated (or, in other words, what is interpreted to be represented by the thoughts). For example, Ryan might think that Cyrano de Bergerac is a great poet; but think, also, that the man who won the duel is no poet at all, even though the man who won the duel is Cyrano de Bergerac.4 The crux is that a subject's thoughts can be more specific than we might otherwise judge. A theory of thought, then, must allow us to make the right assessments of how 'fine-grained' are a thinker's thoughts. But there is an important caveat here. The theory must not allow the 4 Cyrano de Bergerac, Savinien (1619-1655). 7 assessments to be too fine-grained. For instance, we would not want to say that there is a difference in thought (content) between Ryan thinking that the man who won the duel was brave, and Justine thinking that the man who was the victor (in the duel) was brave, when they are thinking of the same man and the same duel. It seems that we do not want to carve up thoughts so finely that given understanding of the meaning of both 'the man who won the duel' and 'the man who was the victor (in the duel)', a subject can have different epistemic attitudes towards these thoughts, assuming, to make the point, that they are distinct. I set aside the task of accounting for misrepresentation for two reasons. First, it is an issue in its own right. It would take too much to set up exactly what the problem is, and how it has been, and might otherwise be, addressed. Aside from that, it is not commonly dealt with in the literature. However, discussion of the appropriate assessments of individuation are central to discussions of theories of thought. For this reason, I propose to conclude by addressing (in Chapter 8) the issue of the individuation, or taxonomy, of thoughts (i.e., the appropriate 'fine-grainedness' of thoughts) by looking at what (1) Twin Earth thought experiment considerations, and (2) puzzles of apparent failures of substitutivity of codesignative terms in propositional attitude contexts, tell us about how thoughts are individuated. In doing this, we will assess also the merits of what I call the Causal Theory of Thought. This is a theory of thought that has been developed out of an extension of what has become known as the Causal Theory of Reference—a theory with very close associations with so-called Twin Earth arguments. Excluding descriptive thoughts There is an important qualification to make concerning the topic of this dissertation. When talking about and investigating thoughts and instances of referential communication, my focus is, to generalize, particular kinds of thoughts. These thoughts, for want of a positive way of putting it, are non-descriptive. There are sentences containing referring terms, and thoughts associated with these, that are purely descriptive. For instance, consider: 8 (1) The hairiest spider in the world is called Boris. When (1) is purely descriptive, it, indeed, refers to a (singular) spider. However, "Boris" operates more like a title than a proper name. If we were to find a hairier spider than the one currently called Boris, then that (hairier) one would now be called Boris. I have nothing to say about these purely descriptive thoughts in this dissertation. Any reference to 'reference', 'thought' or 'referential communication' is to non-descriptive instances of these things, unless otherwise noted. 9 CHAPTER 1 The Linguistic Turn and The Linguistic U-Turn As the title suggests, this chapter has two parts. In general, it will amount to a small amount of history of philosophy, which will help set the scene, as well as introduce some important expressions and concepts that will appear throughout this dissertation. However, in doing this small amount of history, I do not propose that I am giving a significant piece of the history of contemporary analytic philosophy. Even though I think that investigations into these matters would be interesting and fruitful, it is not part of my purpose to provide an in-depth historical review in this present work. I should add that I do not adopt this approach just to side-step difficult historical or interpretive inquiry. All that is required for my purposes, here, is that we have a clear understanding of what I take to be important underlying assumptions of much of contemporary analytic philosophy. Outlining such assumptions is best achieved, I feel, first by tracing the strand of ideas that led to it. Given my purposes, then, it is not directly important whether, for example, Frege believed the Context Principle throughout his career. Nor, for that matter, whether Russell, Moore, Wittgenstein, Carnap or Quine were correct in their interpretations and applications of, say, Frege's ideas—assuming that it is correct, in each case, to speak of interpretations and applications. What is important is the outcome— that is, what more recent philosophers believe and the way in which they proceed in doing philosophy of language and mind given certain assumptions. To put it another way, and in the context of the topic of this dissertation, we need to have at least some sort of understanding of the nature of the move of the Linguistic Turn before we can understand what is (claimed to be) overturned in taking the Linguistic U-Turn. I describe the move of the Linguistic Turn the first part (sections 1.1-1.5). The second part (sections 1.6- 1.8) gives a brief account of how the assumptions of Linguistic Turn Philosophy have been questioned, and in some cases overturned, by some philosophers. The second part 10 also is not an exhaustive account. For the most part, I follow just one path among several that might appropriately be labelled the Linguistic U-Turn. 1.1 The Linguistic Turn and the Genesis of Analytic Philosophy At the heart of the move of the Linguistic Turn is the idea, which has been around for a very long time (probably since Socrates), that philosophy is a verbal or linguistic enterprise. When we investigate something, we look at what terms which refer to that thing mean. If we want to know what bravery is, we look at what we mean by "bravery." This is a methodological point that is more basic than the one associated with taking the Linguistic Turn. However, philosophers, throughout history, have had varying degrees of commitment to it. Socrates and perhaps a few others aside, it was not until the twentieth century that we see a deep commitment to this idea that philosophy is a verbal or linguistic enterprise. For one thing, we can see how this deep commitment moved one particular (though not amorphous) group to adopt a particular methodological strategy. This strategy was motivated by thinking that, at bottom, philosophical problems were linguistic problems. That is, it was thought, and commonly still is thought, that if you solve the linguistic problem, the philosophical problem disappears—or rather, we discover that really there was no problem at all. In fact, some went as far as to say that there were no genuine philosophical problems, only linguistic ones. Mid-century, this gave rise to two opposing schools: so-called Ideal Language philosophy, and Oxford or Natural Language philosophy. It might be easy, nowadays, to write-off these approaches because they have not succeeded in doing what they proposed.1 But we must recognise the context in which the move was made.2 Roughly, this context is born out ofthe idea that even though significant philosophical progress seemed to be being made by following certain 1 This is not to say that they had no success. Indeed, in some sense they both had great success. The point is that there still seem to be genuine philosophical problems that show no clear sign of being eliminated by linguistic^  based conceptual analysis. It is not clear that we are out of this context today. 11 'scientific' procedures (i.e., Mathematical procedures, producing the formal logic of Frege, Russell and A. N. Whitehead), philosophy was not like science. Specifically, no philosophical truth was going to be obtained by looking at experimental evidence. The problem is that if no philosophical truth is going to be obtained in this manner, just what is philosophical investigation seeking? What is the nature of philosophical inquiry? The answer came that philosophical inquiry must be essentially linguistic, which is reflected in, for instance, Moritz Schlick saying "I believe Science should be defined as the "pursuit of truth" and Philosophy as the "pursuit of meaning."3 By "meaning" he means linguistic meaning. But of course, he and his associates were strongly committed to the methodology of the Linguistic Turn. The basic idea seemed to be that if you cannot have experimental evidence, then there is nothing left but linguistic 'evidence'. Philosophical inquiry, then, in some sense precedes scientific inquiry. The job of philosophy was seen to be that of clearing the conceptual (linguistic) ground upon which (good) science could be built. In this light we can see the approach of linguistic philosophy as being motivated by the thought that logic would fix all problems. This was the approach that by most accounts began with Frege (the Ideal Language stream), and even though late in his career Frege had what seemed to be serious doubts about the project, the move had been made. By then, linguistic philosophy was going at full-steam. As an example, consider the way in which Russell thought he could account for the function of names and definite descriptions through his theory of descriptions. The basic approach seemed to be that if we could translate the problem expressions into some appropriate logical notation (Ideal language), then all would be solved. Lest you think that this basic approach has died out, a similar approach is taken by a number of contemporary writers attempting to deal with demonstratives. That is, they treat the problem as a purely 3 Moritz Schlick, "The Future of Philosophy", in Rorty (1968), p. 48. 12 semantic one, then apply the strategy of Ideal Language. Implicit in this methodology is a widening of the scope of Semantics. From this point of view, perhaps it is taken that all in philosophy is semantics (plus syntax). Of course, there are those who explicitly reject the purely semantic approach to demonstratives, utilizing instead pragmatic explanations. Perhaps we see here that these writers seem to be, to some degree, stepping out of the bounds of Linguistic Turn philosophy. This may be why the debates seem intractable. And this might be exacerbated by those employing Pragmatics failing to recognize, or failing to make explicit, the true nature and extent of their methodological shift. That is, they think that, or give the impression that, they are still well within the bounds of linguistic philosophy. 1.2 Frege According to some commentators, analytic philosophy has its roots in Frege's expression ofthe Context Principle in the Grundlagen. Apparently giving us a piece of advice, Frege warned us, "never to ask for the meaning of a word in isolation, but only in the context of a [sentence]."4 This would seem to place Frege as perhaps the earliest of the founders of contemporary analytic philosophy. However, there are some concerns about this as a bald statement of the history of analytic philosophy. There are some specific concerns regarding Frege's work and his use of the Context Principle. One thing is that it is not clear that Frege accepted the Context Principle, in general, after the Grundlagen.5 Another thing is that there is quite significant uncertainty about just what the principle means. Specifically, there is a problem concerning the meaning of "meaning" given that the Grundlagen was written before Frege 'discovered', as they say, the distinction between Sinn and Bedeutung. And of course, even if we are to consider the 4 Frege (1959), p. x. Although it is not without controversy, I follow the current trend of translating "Satz" as "sentence." 5 For an explicit example of this view, see Michael Resnik's two papers on the Context Principle in Sluga (1993). Michael Dummett is less explicit with respect to this point but in several places (e.g., Dummett (1993), pp. 4-6) expresses doubts about Frege's continued adherence to what he calls the Priority Thesis (see below), especially towards the end of his life. 13 possible interpretations o f the principle by plugging in either Sinn or Bedeutung, there is the problem of the correct translation of Bedeutung. There has been a strong tradition of translating it as "reference," but nowadays there is a strong movement to translate it as "meaning." There are more general concerns as wel l . For instance, some claim that this idea that analytic philosophy began with Frege's expression o f the Context Principle undermines the contributions o f others, especially Russell and Moore, to the initial development of analytic philosophy. Others go further, claiming that even Russell and Moore were not wholly committed to the doctrines currently associated ( with analytic philosophy. We see some evidence of Moore's reservations in some passages quoted below. I have no desire to get bogged down in a discussion of a set of necessary and sufficient conditions for analytic philosophy. Apart from it probably being impossible to come up with relatively uncontentious necessary and sufficient conditions, I do not see that it would be particularly helpful to my enterprise. Even though analytic philosophy, as a discipline, may be constituted by an amorphous group, we have the term, and it serves us well enough. However, what I think cannot be doubted is Frege's contribution in the development o f analytic philosophy. If nothing else, then, it must be accepted that definitions o f 'analytic philosophy' aside, Frege's work has had an enormous impact on those who would claim to be working in contemporary analytic philosophy. 6 It w i l l be useful, then, to map out some o f Frege's ideas that may have contributed to the development o f analytic philosophy. Frege is well known for his rejection o f Psychologism. He rejected outright the idea that introspection would give correct axioms and principles of mathematics (or probably any other science for that matter). What follows is a rejection of the theory o f meaning commonly held before Frege, which states that the meaning of a word is the idea with which it is associated, and that the meaning of a sentence just is the combination o f the meanings of the words that make up the sentence. Frege thought that what this theory of meaning did was confuse the objective with the subjective. Meanings (as thoughts, roughly) 6 Of course, the same can be said of Russell and Moore—perhaps, especially Russell. 14 are communicable, hence must be objective. Whereas ideas (as mental items) are by their nature not communicable, hence must be subjective. We can see, now, how Frege gets to his particular expression of the Context Principle (given above). If you reject Psychologism, then meanings have to be discovered in some other way. And given the objectivity of thoughts, Frege thought that the only alternative was to discover meanings through their expression in language.7 The Context Principle comes by seeing that words must be placed in a context to mean anything. As Frege stated the Context Principle, the context in question is a sentence. However, I think that it is fairly widely agreed, nowadays, that this is not going to cover all inquiries into meaning. There is the consideration from Ludwig Wittgenstein that sentences per se are not bearers of meanings; rather, it is particular uses of sentences by speakers that express meanings. That is, sentences must be placed in a particular context of utterance. As something of an extension of this general observation, there is the consideration from W.V. Quine (pre-empted, to some extent, by Pierre Duhem) that sentences must be viewed holistically, that is, they need to be placed in the context of theories, or even whole languages.8 Moreover, there is the possibility that the extension of context should not stop here. That is, I do not see any reason why we should not consider context as incorporating not just linguistic context, but physical context. Characterizing such an extension would get very complicated, but just to sketch out the idea, utterances are given by speakers to hearers who have background knowledge and are placed indexically in the world. Probably, this kind of holism applies only in certain cases, for instance, the search for meanings within the context of a theory of communication. The notion of linguistic meaning, however, would seem not to require taking account of this widening context. In fact, the very idea of linguistic meaning seems to come just from ignoring this fully extended context. It still must make some sense to This is precisely what leads to the Priority Thesis, introduced below. Quine thought that this undermined the very idea of 'meaning', but we need not go into this. 15 ask for the meaning of a word, even in isolation, say in developing a lexicon, but we just have to be aware about what we mean by this. It probably requires a highly qualified notion of meaning.9 1.3 Reservations As I suggested earlier, it is not clear that the instigators of the Linguistic Turn (Frege, Russell and Moore) really meant for the basic ideas to be taken as expressing the priority of language over thought. Frege did indeed say that the way to gain an understanding of our thoughts about the world was through the sentences of the language in which they were expressed. This is a crucial aspect of the methodology of the Linguistic Turn, giving rise to analytic philosophy. We see a rather explicit statement of this idea even in Frege's later writings. In the opening paragraph of "Compound Thoughts," he writes: It is astonishing what language can do. With a few syllables it can express an incalculable number of thoughts, so that even a thought grasped by a terrestrial being for the very first time can be put into a form of words which will be understood by someone to whom the thought is entirely new. This would be impossible, were we not able to distinguish parts in the thought corresponding to parts in the sentence so that the structure of the sentence serves as an image of the structure of the thought.10 This reflects adherence to certain crucial presumptions. From my point of view, the passage is in need of significant qualification (i.e., what kind of thought is 'imaged' by an associated sentence). Nevertheless, it is a forceful point, and as will become clear, not one that I reject out of hand. However, there is textual evidence that later in his career Frege had doubts about how far this basic idea of sentences 'imaging' thoughts should be taken. It is clear that he did use it as a 9 I am aware that I have moved very quickly over these claims about the various ideas of the importance of considerations of context. But I give them just to show the kind of subject matter of the investigations that seem to have led to the Linguistic Turn. 1 0 Frege (1977), p. 55. Directly following is an important, and revealing qualification. Frege continues, "To be sure, we really talk figuratively when we transfer the relation of whole and part to thoughts; yet the analogy is so ready to hand and so generally appropriate that we are hardly ever bothered by the hitches which occur from time to time." This seems to suggest that Frege is never to be taken to be wholly committed to the metaphysical claim of the Assumption of Parallelism. 16 methodological strategy. However, it is one thing to do this, and quite another to draw a potentially related, metaphysical conclusion—namely, here, that thoughts just are (in content and structure) as they are expressed in sentences. In "Sources of Knowledge of Mathematical and Natural Science," Frege is deeply troubled by the feature of language that allows us to construct phrases that contain (apparent) "proper names to which no object corresponds."" The example he gives is, the extension of the concept a. This contains two such proper names. The first is the whole phrase, and from this we get the paradoxes of set theory—Russell's Paradox is a specific example. The second is the embedded phrase "the concept a." It is from noting this puzzling feature of language that Frege says, "a great part of the work of a philosopher consists—or at least ought to consist—in a struggle against language." And further, that the "difficulties which this idiosyncrasy of language entangles us in are incalculable."12 These passages, in themselves, do not necessarily imply the reservation I mention above, but a following speculation about the possible state of affairs, given these troubles, does. Just below these passages, Frege writes: If we disregard how thinking occurs in the consciousness of an individual, and attend instead to the true nature of thinking, we shall not be able to equate it with speaking. In that case we shall not derive thinking from speaking; thinking will then emerge as that which has priority and we shall not be able to blame thinking for the logical defects we have noted in language.13 Effectively, this is a statement of the underlying assumption of what 1 will call the Linguistic [/-Turn (explained below). And even though it might be that he was, here, being somewhat speculative, I think we should take it that Frege, at the end of his career, at least took what I call the Linguistic fT-Turn to be a live possibility. 1 1 Frege (1979), p. 269. 1 2 Frege (1979), p. 270. 1 3 Frege (1979), p. 270. 17 It seems that Moore also had similar sorts of reservations, i.e., placing limits on the scope ofthe claims of the Linguistic Turn. We see from the passage below that not only did he have these reservations, but that he struggled with misinterpretations of what he was saying—misinterpretations that led some readers to think that he was making a stronger claim than he really was. In response to some criticisms from C. H. Langford about his notion of 'analysis'14, Moore writes: Now I think I can say quite definitely that I never intended to use the word ["analysis"] in such a way that the analysandum [that which is to be analysed] would be a verbal expression. When I have talked of analysing anything, what I have talked of analysing has always been an idea or concept or proposition, and not a verbal expression; that is to say, if I talked of analysing a "proposition," I was always using "proposition" in such a sense that no verbal expression (no sentence, for instance) can be a "proposition" in that sense.15 Now, as it stands it is difficult to know what exactly to take from this passage because of the potential vagueness of "concept" or "proposition." But from some following discussion I think we can see that Moore had the reservations that I mention above. In trying to work out what he means by the phrase "giving an analysis," Moore introduces a version of Frege's puzzle (though he makes no mention of Frege). We do not need to go into the details of what he says here, but we can get enough from noting that he thinks it a genuine puzzle—one to which he sees no solution. That it is a genuine puzzle implies that he thinks that the concepts (or propositions, in his sense) associated with a sentence (verbal expression) are not captured by the sentence. That propositions are not captured by sentences is what gives rise to the puzzle, which indicates, to me anyway, that even though Moore adopts the methodology ofthe Linguistic Turn—which he clearly does, given his employment of "analysis"—he does not seem to take the further step of drawing the related metaphysical conclusion. This conclusion is the one that leads to the Assumption of Parallelism, i.e., that the content of a concept (proposition or thought) is paralleled, in all respects, by the content of an associated linguistic expression. 1 4 Schilpp (1942). pp. 319-342. 1 5 Schilpp (1942), p. 661. 18 1.4 The Priority Thesis In his book, Origins of Analytic Philosophy, Michael Dummett says that a grounding assumption of contemporary analytic philosophy is what he calls "The Priority Thesis."16 This can be characterized in a number of different ways, that have important corollaries, but in principle, it is the idea that in explaining meanings, for example, there is a priority of language over thought. Dummett says this: What distinguishes analytic philosophy, in its diverse manifestations, from other schools is the belief, first, that a philosophical account of thought can be attained through a philosophical account of language, and, second, that a comprehensive account can only be so attained.17 Some, perhaps Frege and Moore, for example, seem to have held that something like what Dummett expresses merely reflects an appropriate methodological strategy. In this form, the Priority Thesis implies that the best way, or perhaps the only way, to describe thoughts is through describing them as they are expressed in language. Most, however, seem to hold a stronger version of the Priority Thesis (that incorporates the methodological point) that is grounded upon a substantive, metaphysical claim. This claim, to put it one way, is that the content of a thought is paralleled by the content of an associated linguistic expression. Or, in another way, that the content of a thought is exhausted by the content of an associated linguistic expression. I call this the Assumption of Parallelism.™ There are important associated claims that come from understanding what implications the Priority Thesis (or the Assumption of Parallelism) has for closely related investigations. Acceptance of Frege's Context Principle reveals acceptance of the Priority Thesis with respect to investigating the structure of thoughts and sentences. However, it should be noted that a version of the Context Principle could be formulated in terms of thoughts and components of thoughts (or thoughts and compound thoughts), rather 1 0 Dummett (1991), p. 315ff. 1 7 Dummett (1993), p. 4. 18 This use of "parallelism" comes from Dummett (1993), p. 127. 19 than sentences and words. But importantly, Frege did not formulate it this way. He said that only in the context of a sentence do words have meaning. So, the Context Principle formulated in this way is an expression of the Priority Thesis. However, it is the decision to formulate the Context Principle in this way, rather than acceptance of the idea underlying the general principle itself, which shows Frege's adherence to the Priority Thesis. Acceptance of the Assumption of Parallelism—that the content of a thought is matched (paralleled) in 'richness' by the content of the sentence that is taken to be the linguistic representation of the thought—reveals acceptance of the Priority Thesis with respect to investigating the content of thoughts and sentences. The idea that the 'richness' of the content of a thought is matched by the richness of the content of the associated sentence, implies that the sentence expresses the thought (without any loss of content). It is the Priority Thesis seen in this way that is found to be a grounding assumption of much of contemporary philosophy of language and mind. In general, then, in analytic philosophy, primacy is given to Philosophy of Language over other branches of philosophy—in particular, epistemology, the Philosophy of Mind, and the Theory of Thought. Even given the aforementioned reservations, we can characterize this as Linguistic Turn Philosophy. There is no need to list contemporary philosophers who in some way accept the Priority Thesis, and so accept the methodology of the Linguistic Turn. However, it will be helpful to see some contemporary, explicit examples of acceptance of the Priority Thesis, just to see exactly what is purportedly overturned in taking the Linguistic U-Turn. 1.5 Linguistic Turn Philosophy: Some contemporary examples Michael Dummett, as we have already seen, with his investigations of what he takes to be (at least some of) the origins of analytic philosophy, is explicit in his recognition of the Priority Thesis. Mark Sachs is another who directly addresses the issue of the defining characteristics of analytic philosophy. In a paper 20 in which he discusses the metaphysical assumptions behind the idea that sentences express a determinate sense, Sachs looks at what is required by analytic philosophy for the programme of providing a suitably robust language for science. This is something he, and many others, take to be an important role of analytic philosophy. He writes: The broad conception of the programme rests upon two presuppositions which constitute a metaphysical backdrop: (i) that there is a tenable dichotomy between language and the empirical world, such that it makes sense to think of the one as a description ofthe other, and (ii) that some such language can in principle be rendered as determinate as the world that is to be described.19 I take this to be reflective of the same thinking that underlies the Priority Thesis in the form of the Assumption of Parallelism. That is, it is held that a language—perhaps an artificial one—will have the appropriate structure and content-producing potential to represent the world accurately and fully. If this is so, language parallels the world. Notice, though, that as it is expressed in the above (quoted) metaphysical presuppositions, this parallelism is differently placed, i.e., not between language and thought, but language and the empirical world. This difference is not trivial, but I think we are justified in thinking that Sachs's understanding of the grounding assumptions of analytic philosophy is very closely related to the idea behind the Priority Thesis. We see this close relation in noting that even though his focus is the relationship between language and the world it is meant to represent, he also notes that a language must be rich enough to express "complex ideas," so that it may perform its role in communication. Thoughts are not out of the picture, just put to one side. Notice also that Sachs states that the presupposition "constitutes a metaphysical backdrop," as opposed to the weaker claim that we accept the presuppositions merely as part of a methodological strategy. Some writers find it necessary, or are careful enough, to acknowledge grounding assumptions, or to explain a methodological strategy. Tyler Burge is one. In "Individualism and the Mental" he writes: 1 9 Mark Sachs (1990), p. 176. 21 In an ordinary sense, the noun phrases that embed sentential expressions in mentalistic idioms provide the content of the mental state or event.... Thus the expression 'that sofas are more comfortable than pews' provides the content of Alfred's belief that sofas are more comfortable than pews.20 This reveals an acceptance of the Priority Thesis in the form of acceptance of the Assumption of Parallelism. It is not entirely explicit because of the potentially equivocal character of "belief." However, further discussion, and his subsequent overall argument, show that he accepts the Priority Thesis. Also, it is worth pointing out that Burge recognizes an important related assumption. In some terminological preliminaries he writes: Roughly speaking, grammatical contexts involving oblique occurrences have been fixed upon as being specially relevant to the representational character (sometimes called "intentionality") of mental states and events. Clearly, oblique occurrences in mentalistic discourse have something to do with characterizing a person's epistemic perspective—how things seem to him, or in an informal sense, how they are represented to him.2 1 From this he notes that the crucial point in the preceding discussion is the assumption that obliquely occurring expressions in content clauses are a primary means of identifying a person's intentional mental states or events.22 Two things are packed into this assumption. The first is that "believes that," "thinks that," etc., clauses clearly direct us to some aspect of an agent's cognitive make-up. This has no direct, necessary connection to the Priority Thesis, and so is carried through into Linguistic U-Turn philosophy. The second is that the imbedded clauses express the relevant content. This is connected to the Priority Thesis. It reflects the Assumption of Parallelism. We should note that the phrase "are a primary means o f might indicate that Burge may be committed only to acceptance of the Priority Thesis as a methodological strategy. However, z u Burge (1979), p. 74. 2 1 Burge (1979), p. 76. 22 Burge (1979), p. 76. The emphasis is mine. 22 he may just be alluding to the identification of the intentional character, not the content of the mental states. It is unclear, then, how strongly, if at all, Burge is committed to the stronger metaphysical claim.2 3 In general, implicit acceptance of the Priority Thesis is revealed in either the metaphysical assumptions or the methodological strategy employed by most writers in contemporary analytic philosophy. If we were to accept Dummett's criterion,24 this is what makes them Analytic Philosophers. Sachs, in his paper "Through a Glass Darkly," gives a very good illustration of what position such philosophers adopt. In setting up a "Backdrop" to his paper, he goes to the trouble of identifying where the Analytic Philosopher is placed with respect to the methodology of metaphysics. That is, in investigating the world, Analytic Philosophers do not move behind language, to the mind (thoughts), let alone to the world itself. He takes it that this reluctance is motivated by a move that is parallel to the move of the traditional sceptics. Just as they were worried about the reliability of our access to the world, Analytic Philosophers are worried about the reliability of our access to our minds. In Sachs's words: It should be noted that what we have here is a parallel to the old sceptical problem about the reliability of our access to the world. The question is unchanged: whether that access is adequate to the job of capturing the way the world is. The difference is that the source of possible trouble is now thought to be not the mind, located behind the lens [of perception], but rather language itself, a point beyond which the analytic philosopher was loath to retreat.25 This 'loathing' was felt by Frege with his rejection of Psychologism, and is, nowadays, well entrenched. However, I take it that Sachs takes an important step here, by asking the question "What assurance is there That he is not committed to the metaphysical claim may be supported by some of his other work, especially "Belief De Re" Burge (1977). Not being so committed is shown, in part, by his characterization- of belief de re as "a belief whose correct ascription places the believer in an appropriate nonconceptual, contextual relation to objects the belief is about." This is in contrast to beliefs de dicto that are "fully conceptualized" (pp. 345-6). 2 4 I want to make it clear that I am not suggesting that we should. 2 5 See Sachs, in Bell and Cooper (1990), p. 174. You can see the extent of this 'reluctance to retreat' especially in, say, Putnam-style externalist arguments. 23 that language displays contents of the sort required to represent the world accurately?"26 This is an extension of my questioning of the Priority Thesis. Even P. F. Strawson, probably an ancestral friend of the Linguistic U-Turners, shows his commitment to certain assumptions. In discussing several understandings of 'analysis', the defining characteristic of the methodological approach of analytic philosophy, Strawson concludes that for the analytic philosopher, "the actual use of linguistic expressions remains his sole and essential point of contact with the reality he wishes to understand"27 Admittedly, this reality is what Strawson calls "conceptual reality," but nonetheless, we are left with the impression that there is nothing accessible beyond this, or rather, nothing of any interest. Nathan Salmon admits to acceptance of a thesis that seems to be what I call the Assumption of Parallelism. In setting up his argument for a defence of a Russellian view of singular terms, Salmon says that he maintains a thesis "which both Frege and Russell more or less accepted." This is that the proposition that is the information content of a declarative sentence (with respect to a given context) is structured in a certain way, and that its structure and constituents mirror, and are in some way readable from, the structure and constituents of the sentence containing that proposition.28 As might be expected, and as might already be thought, it is rare to find clear examples of explicit acceptance of the Priority Thesis that are expressed in such a way that the examples stand, relatively resistant to interpretational problems. In other words, many passages which at first glance strike me as good examples would require lengthy discussions about the interpretations of key terms to establish the passage as a genuine example. The example from Burge was such a case. Some readers may have concerns about one or more of the other examples. Also, plucking quotations out of books and papers Sachs, in Bell and Cooper (1990), p. 174. Strawson, "Analysis, Science, and Metaphysics", in Rorty (1968), p. 320. Salmon (1990), in Anderson and Owens (1990), p. 216. The emphasis is mine. 24 without elaborating on the context in which the passage appears will always lead careful readers to be suspicious of the value of the quotation for showing the desired point. Engaging in these discussions would be interesting, and probably valuable, but it would seriously delay progress to what I take to be the more important parts of the dissertation. In any case, I do not think that it is necessary to give such examples, and to be so delayed. There are three reasons for this. First, I do not think that it is necessary to convince readers that the Linguistic Turn did take place some time early in the twentieth century, and that this move is intimately connected to what has become known as Analytic, or sometimes, Anglo-American Philosophy. Moreover, most are aware of at least the general nature of the Linguistic Turn. Second, I do provide what seem to be some relatively clear, interpretation-resistant examples. Third, many of the discussions in the main body ofthe dissertation show, implicitly, at least, the style of argument employed by those who accept the Priority Thesis. I will say a little more about this before we go on. The general idea behind the third reason is that the Assumption of Parallelism is reflected in the methodological approaches of many contemporary philosophers of mind and language. One fairly clear example is found in certain kinds of arguments that purport to show how mental states get their representational properties. These arguments come from thinking that mental 'aboutness', or mental representation, is like reference. That is, there is an assumption that there is a parallel between how we refer to an object and how we think about that object, and, importantly, that there is this order of explanatory priority, i.e., explaining thought about an object in terms of reference to the object. As will be shown, it is a crucial methodological assumption of these writers that thoughts are paralleled by sentences. Admittedly, this section provides a fairly rough characterization of analytic philosophy. It does, however, highlight certain important features that are shown to amount to underlying assumptions that lead either to guiding writers to adopt a particular methodological strategy, or to accept certain metaphysical assumptions, or both. 25 Of course, we always have to start from some position, magically suspended by links to these 'sky-hook' assumptions. Richard Rorty is one who is prepared to take on the charge of warning us of this. He begins his introduction to The Linguistic Turn by noting that periodically throughout the history of philosophy, methodological revolutions have taken place, usually instigated by just one philosopher. He cites Descartes, Kant, Hegel, Husserl and Wittgenstein (twice, of course).29 What is common to each revolt is the idea that the previous methodology is grounded somehow on faulty metaphysical and epistemological presuppositions. Further, that this is to be replaced by a new, "presuppositionless" methodology, making investigations 'scientific', and so providing knowledge rather than mere opinion, or rather faulty opinion. However, as it has turned out, each revolutionary has been usurped by yet another by showing that what was thought to be "presuppositionless" actually was not. As Rorty notes, this is not at all surprising, for, as he says, "To know what method to adopt, one must already have arrived at some metaphysical and some epistemological conclusions."30 Moreover, any attempt to defend such conclusions within the bounds of the new methodology will be exposed to the attack of circularity. Further still, if not defended, the proposed method, and resulting, substantive philosophical conclusions, will be seen to be inadequately justified. The situation seems hopeless. But it is not hopeless. These are just the methodological limitations we must live with, inescapably. And it is healthy to remember the position we are in, but perhaps not too frequently. These metaphilosophical issues aside, there is a recent revolution whose merits, I consider, are in need of evaluation. This revolution is the one I call the Linguistic U-Turn. Rorty (1968), p. 1. Rorty (1968), p. 1. 26 1.6 Evans and The Linguistic U-Turn In taking the Linguistic U-Turn, I follow, primarily, the work of Gareth Evans, though he was never, explicit about it, and some passages indicate that he may not have been fully aware of its implications and connections.31 In any case, many agree that Evans initiated a move in a direction away from the standard methodological practices of analytic philosophy—that is, in the direction of what I call the Linguistic U-Turn. Quite a number have either followed Evans or moved in this direction on their own. Christopher Peacocke, Martin Davies, John Campbell and Adrian Cussins have been important contributors carrying the movement forward, each in their own directions, after Evans's early death.32 Evans drew heavily from the work of Frege, and even though Evans is not explicit about it, the roots of the idea of the Linguistic U-Turn can be found, I think, in Frege's essay "Thoughts."33 At one point, Frege notes that sentences often contain "conversational suggestions" and can suffer through surface, transformations without affecting the underlying thought.34 From this he concludes that "the content of a sentence often goes beyond the thought expressed by it."35 But, he notes, the reverse is also possible. -He writes, "the mere wording . . . does not suffice for the expression of the thought."36 He has in mind, here, sentences containing indexicals, demonstratives, personal pronouns and (it seems) proper names. 3 1 For example, consider this passage: "a sentence S has a different cognitive value from a sentence S' just in case it is possible to understand S and S' while taking different attitudes towards them" (Evans (1982), p. 19.). The thing to note is that this might reflect that in explaining Frege's point, he thinks that it is sentences that are of primary concern, not thoughts. I am inclined to think that Evans should have held that we have attitudes to thoughts, not sentences. 3 2 See for example, Peacocke (1986); Davies (1995) and (1991); Campbell (1994); and Cussins (1992) and (1993). 3 3 Frege (1977), pp. 1-30. 3 4 Frege (1977), pp. 8-10. 3 5 Frege (1977), p. 10. 3 6 Frege (1977), p. 10. 27 In these cases, different thoughts can be associated with the same sentence. He says, "Different thoughts thus obtained from the same sentences correspond in truth value . . . Nevertheless the difference must be recognized."37 It is this difference that makes the Linguistic U-Turn possible. There are a number of ways of characterizing the Linguistic U-Turn. In general, it can be characterized as the reversal of the Priority Thesis. However, this hides a number of underlying reasons for taking the Linguistic U-Turn, as well as making too strong a claim. I am not making the claim, and I am sure that Evans was not implying, that all of what has been done and is being done under the assumptions of the Priority Thesis is incorrect. On the other hand, it is important to note that I am not backtracking here. I am still maintaining the importance of the reversal of the Priority Thesis in specific instances. For Evans, the reversal of the Priority Thesis comes about in this way: cognitive content is, at bottom, made up out of information, produced through the workings of an information system. He had as one of his goals to give a theory of thought based on the idea of an Information System. This is characterized in this way: When a person perceives something, he receives (or better, gathers) information about the world. By communicating, he may transmit this information to others. Any piece of information in his possession at a given time may be retained by him until some later time. People are, in short, and among other things, gatherers, transmitters and storers of information. These platitudes locate perception, communication and memory in a system—the information system—which constitutes the substratum of our cognitive lives.38 It is from this that Evans thinks it is preferable to take the notion of being in an information state with such and such a content as a primitive notion for philosophy, rather than to characterize it in terms of belief.39 Frege (1977), p.12. Evans (1982). p. 122. Evans (1982), p. 123. 28 And further, that it "is as well to reserve 'belief for the notion of a far more sophisticated cognitive state."40 That is, information is taken to be primitive to belief and so expression of belief. This view is in opposition to the standard view that knowledge is basic. Further, at the level of thought, knowledge is made up out of beliefs, where the notion of belief is connected with the notions of judgement and reasons. We might say, then, that beliefs are states of the conceptual or linguistic representational system. From Evans's point of view, then, with this reversal of the priority of knowledge and information, items of the information system are belief-independent. We have them independent of judgements made about their truth. This is the crucial.step, and it seems to me that if you take it you step outside the bounds of an inquiry that is purely linguistic. These are not merely methodological points, but are epistemological and metaphysical claims about the causal structure of the mind, and how it interacts with the world. They reflect underlying reasons for the reversal of a causal version of the Priority Thesis.4' The diagram below represents my take on what Evans has in mind. Evans (1982), p. 124. That is, not only is there to be an explanatory priority of thought over language, but also an explanatory priority of experience over thought. 2 9 Perception ""^Non-conceptual Content I Sensory-motor 1 Conceptual System Sys tern Conceptual Content (beliefs, reasons, judgements, etc) Memory Communication From this position, I hope to show the possibility of the priority of thought over language, in certain cases. Acceptance of this possibility should allow the development of alternative explanations of certain mental and linguistic phenomena, as well as alternative treatments of certain recalcitrant puzzles.42 Further, once we see the way the treatments are effected, we should see that certain thoughts are richer in content than the sentences of a language that are believed to express them. If the alternative explanations and treatments are plausible, they will help to show the priority of thought over language, and so will add to the justification for the Linguistic U -Turn. Before proceeding it will be useful to have a clearer idea of what is meant by certain kinds of thoughts being richer in content than associated sentences. Here is what Evans says: Information-based thoughts are distinct from descriptive thoughts, simply in virtue of being information-based. There can be no description '(p'—no matter how rich and detailed—such that entertaining the thought that the cp is F is ipso facto entertaining an information-based Actually, it may turn out that what I present may not amount to different, or perhaps better ways of dealing with the puzzles. Rather, the proposed structure of the mind/language/world complex is such that the puzzles do not arise. [Perhaps all solutions to puzzles are just reasons for thinking that, in each case, that there is no puzzle.] 30 thought. For an information-based thought requires the use of the subject's information system, and no description '(p'—no matter how rich and detailed—can guarantee an appropriate link with the subject's informational system. For instance . . . there is necessarily a gap between grasping 'The blonde girl who looks thus and so is F and 'That blonde girl is F, because there is always room for an intelligible realization: 'Ah! So that's the blonde girl you mean!'.43 A good example of the Linguistic U-Turn mode of inquiry is Evans's investigation of Reference through his theory of information and the notion of the information system. He notes that on some occasions reference is fixed by description only, but he sets these cases aside because, as he says, they fail to function as "Russellian singular terms," i.e., as genuine proper names. For instance, recall the example, "The hairiest spider in the world is (called) Boris," but where "Boris" acts more like a title, not a name, so that if a hairier spider than (this) Boris is found, then it would be called Boris. Of all the other ways of fixing reference, Evans claims that they have a common characteristic. This is: In order to understand an utterance containing a referring expression used in this way [i.e., non-descriptive], the hearer must link up the utterance with some information in his possession. Thus, if a speaker utters the sentence 'This man is F', making a demonstrative reference to a man in the environment he shares with the hearer, the hearer can understand the remark only if he perceives the man concerned, and, bringing his perceptual information to bear upon his interpretation of the remark, judges 'This man is F: that's what the speaker , 44 is saying . Here, we have the idea that it is information-based thoughts that are involved in fixing reference. Evans continues: Understanding the kind of use of a referring expression I am considering is not a matter of having beliefs with the right sort of content, but rather a matter of having, and using, information from the right source.45 Evans (1982), p. 308-9 Evans (1982), p. 305. Evans (1982), p. 307. 31 He gives an example of a speaker referring to a bird seen on a hunting trip taken with the hearer some time ago. Various things can be said to 'jog' the hearer's memory, but it is not until the right information is retrieved (presumably from her memory) by the hearer, and then appropriately linked to the use of the term, that reference is secured. The important point is that it is this information that leads to the hearer identifying the referent, not, by itself, the bare content of the linguistic expression containing the referring term (referring to the bird), nor the content of the descriptions used to 'jog' the hearer's memory. Further on the requirement of information for reference, here is a point at which it is claimed that a purely linguistic analysis is deficient. Evans reiterates an important point about reference made by Strawson in "Identifying Reference and Truth Values," in Logico-linguistic Papers. Roughly, the important point is that when a speaker refers, there is an assumption of shared (identifying) knowledge, and this knowledge is associated with what is thought to be adequately identifying descriptions. Evans thinks that this account suffers from the "general neglect of the concept of information."46 For Strawson, there is the association of knowledge (which is a species of belief) with the descriptions, rather than the idea that uses of referring expressions invoke information about an object (i.e., the one to which the speaker is referring). There is a problem here, however, for when it is recognized that the information is incorrect, 'knowledge' must be replaced by 'mere belief. So instead of being able to claim that uses of referring expressions are made with the assumption of shared information (Evans) we are left with claiming that referring expressions presuppose existential propositions, i.e., supposed identifying beliefs about the object. Evans thinks that this is a less interesting claim. This judgement aside, it seems we do more than presuppose that a hearer has some possibly right identifying belief about an object when we use a referring expression. Continuing the bird example, when a speaker says "Remember that magnificent bird we saw?" we do not think that the speaker presupposes that the hearer has something like "There is a 4 6 Evans (1982), p. 309. 32 magnificent bird we both saw," say. This is not sufficient. The hearer must recall the information he has about the bird, and the speaker must be assuming that he can recall it.47 Now, leaving aside for the moment the issue of whether this view about reference is to be accepted, it seems quite clear that this mode of inquiry is representative of Linguistic U-Turn philosophy. This point is perhaps further explained by looking a little more closely at Evans's concepts. The information state with which Evans is concerned in reference is the state produced by the operation of the information system, and as said before, this is 'belief-independent'. This yields a distinction between conceptual information (that is embodied by language), and nonconceptual information (that has its grounding in perception).48 I do not wish to preempt discussions that follow, but it will be helpful to get some idea of the background to this distinction. Evans divides the mind's capacities into: (i) an information gathering system (that is connected to the sensory/motor system), and (ii) a concept applying, reasoning and judging system, (i) is phylogenetically more basic. Humans, and perhaps a few other 'higher' species possess (ii). Evans says that "the information states which a subject acquires through perception are nonconceptual, or nonconceptualized."*9 These states, then, are the nonconceptual content of the information system. Processing of these states by the concept applying, reasoning and judging system, gives conceptual content. It might be objected that noting this distinction, in itself, does nothing to support the claim that Evans's mode of inquiry is representative of Linguistic U-Turn philosophy. Those doing Linguistic Turn philosophy might also accept the distinction. However, there is a fundamental difference between the two views. In admitting that there is nonconceptual information, the Linguistic Turn philosopher seems to be Probably, something analogous to the whole set of Gricean intentions is required. 4 8 We see, here, some sort of connection with the de dicto/de re distinction—especially if understood in the way Tyler Burge characterizes it. More on this in section 7.2. 4 9 Evans (1982), p. 227. 33 committed to the view that the distinction is of little or no interest because in investigations of thoughts in general, the important objects of analysis are the contents of conscious experience. These experiences are those got by a subject ascribing experience to herself—i.e., the operation characterized by "I am having such and such an experience."50 This Evans calls our "Intuitive concept of conscious experience." But, he says that "although it is true that our intuitive concept requires a subject of experience to have thoughts, it is not thoughts about the experience that matter, but thoughts about the world."51 The point is this. For the linguistic philosopher, nonconceptual information is just raw sense data, and is not presented to the conscious mind in such a way that it could be said that we have thoughts about it. This requires an ascription of the experience. When you hold this position, you stay within the bounds of Linguistic Turn philosophy. The problem, as Evans sees it, is that for the Linguistic Turn philosopher, any thought involving nonconceptual information can be only a thought about the experience, not a thought about the world. Moreover, Evans argues that we can have genuine thoughts about the world (i.e., ones that are not thoughts that are merely ascriptions of experiences). The above, in itself, is not an argument in favour of Evans's view, but it does reflect his adherence to the Linguistic U-Turn. There remain two things to say regarding Evans's place in analytic philosophy, given his adherence to the Linguistic U-Turn. First, from one point of view—that of adhering to the Priority Thesis—it might seem that Evans is no longer doing analytic philosophy. Some might even go so far as to say that he is no longer doing philosophy. Perhaps they would say he is doing theoretical psychology, or some such thing.52 From another point of view, however, Evans still can be seen to be within the analytic tradition. It is true that his mode of inquiry seems not to fit with Dummett's definition of analytic philosophy. Recall, this is the view that to be doing analytic philosophy, one must hold that a philosophical account 5 0 Evans (1982), p. 158. 5 1 Evans (1982), p. 158. 52 A similar point will be made with respect to Peacocke being a 'theorist of thought'. 34 of thought can be attained only through a philosophical account of language. However, this definition probably is too strict. Evans utilizes the work of many who are regarded to be solidly in the analytic tradition, especially Frege and Russell. Second, some might think that by taking the Linguistic U-Turn, Evans is simply reverting to the pre-Linguistic Turn, Cartesian style philosophy. Dummett thinks that this is not the case for two reasons. First, Evans, and others, do not focus on knowledge, as such. It is true that their inquiries are not wholly divorced from epistemological considerations, but their starting point is with questions like "What is it to grasp concepts?," or, "What is it to have a thought that corresponds to some sentence?" These are (or are claimed to be) prior to considerations of knowledge, or recognition of truth. Second, Evans's investigations focused on the notion of 'information', which he thought was a more fundamental concept than knowledge. Quoting Dummett: "Information is acquired, for example, without one's necessarily having a grasp of the proposition which embodies it; the flow of information operates at a much more basic level than the acquisition and transmission of knowledge."53 1.7 Peacocke What is involved in suggesting that some have taken the Linguistic U-Turn is that these philosophers have adopted, at some level, a significantly different methodology when coming to do philosophical investigations of, for instance, thoughts. As far as I am aware, Christopher Peacocke was the first to give an account that explicitly indicates a methodological change. For instance, at the beginning of Ch. 8 of Thoughts: An Essay on Content Peacocke asks the question: Dummett (1993), p. 186. 35 Can we give a good account of what it is to have attitudes to thoughts without making use of linguistic notions?54 Normally it is held that thoughts are taken to be the contents of propositional attitudes, so it probably will seem odd to many readers that someone would ask this question, let alone answer it in the affirmative. However, Peacocke does answer it in the affirmative, and although he is considering thoughts in general, he says that "I shall be . . . arguing that no general mention has to be made of linguistic notions in an account of attitudes to thoughts."55 He then becomes, as he says, a "theorist of thought."56 Peacocke notes that with regards to this question there would likely be a "gulf between many psychologists and philosophers. This might leave open to some philosophers to say that a 'theorist of thought' (i.e., someone who claims that we can give a good account of attitudes to thoughts without the use of linguistic notions) just is not doing philosophy. However, I think we should be reluctant to give the activity of 'doing philosophy' such rigid and narrow boundaries. I am not putting forward these ideas with the purpose of defending the conclusions reached. All that I want to do, here, is to show (for the moment) that this methodology is indicative of the Linguistic U-Turn. However, some may object at the outset that really there is no genuine methodological shift. So, in order to show that the shift of the Linguistic U-Turn is genuine, it needs to be shown that this mode of inquiry has a certain legitimacy, i.e., that the Linguistic U-Turn is not simply an illusion. Peacocke identifies and answers three possible challenges with respect to this methodological shift of the Linguistic U-Turn. First, it might be challenged that thoughts are structured entities and that they get this structure from the structure of the sentences used to express them. Peacocke has an answer to this challenge in his "substantive theory of content." He says that "an adequate account of thoughts according 5 4 Peacocke (1986), p. 112. 5 5 Peacocke (1986), p. 113. 5 6 I take it that this might give rise to the philosophy of thought, which is distinct from philosophy of mind or Philosophy of Psychology. 36 to which they are individuated by reference to canonical acceptance conditions will already entail that thoughts are essentially structured."57 A canonical acceptance condition is "a normative condition concerning acceptance of [a particular] content."58 Peacocke adds: Anyone whose judgements manifest that pattern of canonical acceptance conditions is ipso facto judging a structured thought. The thought's structure and constituents are intrinsic to it on this account in the sense that it is impossible for someone to be judging that very content, and not be judging something with the given structure and constituents.59 No doubt, these few passages are insufficient for a full understanding of Peacocke's explanation ofthe structure of thoughts. However, we do not need a full understanding. All that is needed is to put Peacocke's statements concerning his methodology in some sort of context. Second, someone might challenge the theorist of thought to explain how the thoughts he explains through his (essentially) non-linguistic account/theory may still be expressed in language. And further, someone might challenge the theorist to explain the relation between this non-linguistic account and our usual understanding of language. Of course, this is going to require a lot of work and Peacocke's response is too elaborate to give in full, partly because it depends on other parts of his overall account of thoughts. However, I think I am correct in summarizing it in the following way. (1) There is a difference between describing one's thoughts and expressing them. (2) But there is an identity between the conditions whose obtaining would give an agent reason to judge some thought true and those which would give him reason to think that a sentence describing that thought would be true. Further, (3) the agent's intention in producing that sentence is in part explained by his thinking that it is true. Given (1), (2) and (3), (4) structured states (which Peacocke wants to explain) that are systematically caused by the contents of a thinker's attitudes are not necessarily the states by which the thinker expresses (or describes) his thought, Peacocke (1986), p. 114. Peacocke (1986), p. 11. Peacocke uses "content" and "thought" as stylistic variants. Peacocke (1986), p. 114. 37 i.e., the states utilized in an (essentially) linguistic account. It follows, then, that there is a distinction between thoughts and the expression of them in language. But, and this is important, there is a connection, via what supports truth in each case. Third, a challenge may be made that the notion of 'judgement', which is required in this account of thought, cannot be explained without reference to language. Peacocke notes Dummett's comment that judgement "is the interiorization of the external act of assertion,"60 and, that any account of assertion will be linguistic. But Peacocke does not accept Dummett's claim. Again, the response is elaborate, but essentially it involves first showing that there is no necessary connection—like the one Dummett suggests—between judgement and assertion. For Peacocke, judgement is prior to (expression of) belief, and so distinct from assertion. Again, this seems to indicate clearly the Linguistic U-Turn. Peacocke claims to give a non-linguistic account of judgement and thought. Such an account will strike many philosophers as simply being implausible, probably as a result of finding it difficult, or maybe impossible, to step outside the tradition of linguistic analysis. A serious problem arises, then. Peacocke's responses (above) will strike them as intuitively false, or even nonsensical. Part of the challenge, then, is to step beyond traditional methodologies. If such a step can be taken, Peacocke's responses need to be taken seriously. 1.8 Partial Commitments to the Linguistic U-Turn In general, I suspect that there is at least a partial commitment to the basic idea of the Linguistic U-Turn found in a number of streams of contemporary philosophy of mind and language. That is, in addition to these more explicit indications of the Linguistic U-Turn, I find partial acceptance, or rather, something sympathetic in aspects of the work of some writers traditionally thought to be well ensconced in analytic philosophy. The first stream is fed by the group who think that belief de re cannot be reduced to belief 6 0 Peacocke (1986), p. 118. 38 de dicto. In particular, I have in mind those, like Tyler Burge, who reject the idea that belief fife re is just a species of belief de dicto. The second stream is fed by those who, utilizing the distinction between semantic (or linguistic) meaning and speaker's meaning, draw on the tools of Pragmatics to explain certain phenomena, for instance, what has become known as Frege's Puzzle. I will give some representative examples of these two groups. Also, I will give some examples of others who are not so easily categorized. What I present will not amount to conclusive reasons for thinking that each example does come from a partial commitment to the Linguistic U-Turn. Also, I am not claiming that these are all connected, or that the different views are consistent with one another. Burge's arguments in "Belief De Re" provide the foundation for the first group. We can see this, at least in part, in the way in which Burge sets up the distinction between beliefs de re and beliefs de dicto, especially, as he says, "the intuitive epistemic basis for the distinction."61 Note first that Burge is using 'concept' to stand for any mental entity that is a type of representation of an object. He says, then: From a semantic viewpoint, a de dicto belief is a belief in which the believer is related only to a completely expressed proposition (dictum). The epistemic analogue is a belief that is fully conceptualized!32 A de re belief may be characterized negatively, i.e., as a belief that is not fully conceptualized. Alternatively, Burge gives a positive characterization: A de re belief is a belief whose correct ascription places the believer in an appropriate nonconceptual, contextual relation to objects the belief is about The crucial point is that the relation not be merely that of the concepts' being concepts of the object—concepts that denote or apply to it.63 The important thing to take on board, here, is that Burge argued for the primacy of beliefs de re. He argued that beliefs de re are epistemologically prior to beliefs de dicto. That is, he was arguing that Burge (1977), p. 345. Burge (1977), p. 345. Italics in the original. Burge (1977), p. 346. 39 contrary to the widely held view, beliefs de re could not be explained in terms of, or reduced to, beliefs de dicto. Given the above characterizations, then, that beliefs de re cannot be reduced to beliefs de dicto indicates the underlying idea that certain mental states are not explainable solely in terms of descriptive content (belief de dicto). Something will be missed, i.e., that which is drawn from nonconceptual, contextual relations. I am not saying that what Burge is arguing for just is the move of the Linguistic U-Turn, but it certainly looks to be suggestive of it. We get a clearer idea of this when we see what is implied by views which argue, contrary to Burge, that beliefs de re not only can be, but are appropriately, explained in terms of, or reduced to, beliefs de dicto. Holding that all ascriptions of belief de re can be given in terms of ascriptions of belief de dicto seems to be a sure indicator of a strong commitment to Linguistic Turn Philosophy. We can see this, for example, in the early pages of Nathan Salmon's paper "A Millian Heir Rejects the Wages of Sinn."64 He outlines an important thesis maintained in his book Frege's Puzzle. This is that the proposition that is the information content of a declarative sentence (with respect to a given context) is structured in a certain way, and that its structure and constituents mirror, and are in some way readable from, the structure and constituents of the sentence containing that proposition.65 In other, hopefully now more familiar words, Salmon is saying that the informational content of the underlying proposition is expressed (without any loss of content) by an associated sentence. It should be seen, then, that this thesis reflects acceptance of the Assumption of Parallelism. Salmon continues: This thesis . . . yields the consequence that de re belief (or belief of) is simply a special case of de dicto belief (belief that). To believe of an individual x, de re, that (he, she) is F, is to believe de dicto the singular proposition about (containing) x that it (he, she) is F, a proposition that can be expressed using an ordinary proper name forx. Of course, there are complications here coming from interpretations of "belief and "de dicto." If beliefs are interpreted as being items of a purely linguistic representation system, then what Salmon says follows 6 4 Salmon (1990). 6 5 Salmon (1990), p. 216. 40 (almost) by definition. Also, there might be a way of interpreting "believes thaf such that we could accept that belief de re is just a special case of belief de dicto, and reject the Assumption of Parallelism. This interpretation comes from holding that what follows 'believes that' is not a sentence but a thought (containing the informational content) not completely expressible by a sentence. However, this is not what is implied by Salmon's account. Given his implicit acceptance of the Assumption of Parallelism, what follows 'believes that' is some embedded sentence that is purported to give the entire informational content of the proposition. It is from Salmon's acceptance of the Assumption of Parallelism that we get the idea that if you reject this consequence of his thesis, then you reveal at least implicit sympathy to the Linguistic U-Turn. In other words, if you think that ascriptions of belief de re are not captured by (cannot be given in terms of) ascriptions of belief de dicto, then you are, in effect, saying that the informational content of the dicta do not exhaust all content of associated propositions (or thoughts). And of course, this is what underlies the rejection of the Assumption of Parallelism We get a good representation of the ideas of the second group from John Perry's discussion of Howard Wettstein's paper "Has Semantics Rested on a Mistake."66 The general context of the discussion comes from considering attempts to deal with Frege's Puzzle made by those who accept so-called 'New' (or Causal) theories of reference. Here is how Perry characterizes what leads to so-called Frege Puzzles: If there is some aspect of meaning by which utterances u of [sentence] S and u' of S' differ, so that a rational person who understands the meaning of both S and S' might accept u but not u', then a fully adequate theory of linguistic meaning should assign different propositional contents to u and u'.67 Perry points out, correctly, that at face value this formulation of Frege's conditions ofthe individuation of propositions is disastrous for New (or Causal) theories of reference. I do not want to preempt See Perry (1988), and Wettstein (1986). Perry (1988), p. 5. 41 discussions to follow in later chapters, so we will leave aside the details of why this is so. However, Perry thinks that New theories can be saved by holding that the semanticist need not assign different propositional contents to the discriminated utterances, if he can find some other aspect of their meaning that explains the differing effects on the beliefs of a rational, competent, listener. The above formulation is, then, modified to the following: 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 understood both S and S' might accept u but not u', then a fully adequate theory of linguistic meaning should say what this is.68 For Wettstein, this other aspect of meaning comes from Pragmatic considerations. In other words, he claims that the noted deviancies in propositional contents generated by these circumstances must be accounted for by something other than the semantics of the sentences. Now, again, I do not wish to enter into discussions about the virtues of these ideas now. However, for current purposes, the important thing to note is that the original formulation reflects the Assumption of Parallelism, and that the modified formulation indicates a move away from it. We see this by noting that in the second formulation, it is no longer assumed that the content of a sentence is paralleled by the content of the associated thought. As mentioned at the beginning of this section, there appear to be some examples of moves towards the Linguistic U-Turn made by some who are not so easily categorized. For instance, in the introduction to her book, Language, Thought, and other Biological Categories, Ruth Garrett Millikan writes: "I will argue that the meaningfulness of sentences can be described without making reference to the fact that sentences are typically used to express and transmit thought, and that what beliefs and desires and intentions are can be explained without making reference to language."69 It is not altogether clear that this passage secures the idea that Millikan is committed to the Linguistic U-Turn in an unqualified way. 6 8 Perry (1988), p. 5. 6 9 Garrett Millikan (1984), p. 5. 42 However, the passage does, at least, indicate a kind of thinking, and a methodological approach, that accords with the basic idea of the Linguistic U-Turn. The same can be said of some passages in a recent paper by Michael Devitt. In mapping out what he takes to be the appropriate methodology of naturalistic semantics, and commenting on the "Davidsonians"' view, Devitt says that in contrast to the Davidsonians, "I favor a robust mentalism that takes thoughts to be objective states posited independently of language. Thoughts are in an important; respect "prior" to the-language that expresses them and can be used in the explanation of linguistic meaning."70 Again, we see here ideas that appear to accord with the basic ideas of the Linguistic U-Turn as I have described it. In all cases, it seems, the principal motivation for adopting this non-standard metaphysical understanding of the relationship between thoughts and language, and associated methodological strategies, is to find another way of dealing with certain recalcitrant puzzles that arise when attempting to explain thought in terms of language. If the proof is in the pudding, what follows is, I hope, a fair portion. Devitt (1994), p. 556. 43 CHAPTER 2 Thoughts and Experience: Some Preliminaries It should become clear that this dissertation amounts to advocacy of at least the partial independence of the philosophy of thought. As the first chapter indicated, standard investigations of thoughts have proceeded under the assumption that the philosophy of thought is properly subsumed under the philosophy of language. Also proposed was the idea that for some investigators, this assumption is to be rejected. As just stated, however, these suggestions of methodological (and perhaps metaphysical) priority, or its rejection, are too imprecise. That is, they are not applicable to all investigations of language and mind. On one side, some investigations of language are far removed from any ideas about the contents of human minds. Linguists often do not have to trouble themselves with what appear to be the very difficult ideas of thoughts and intentions. Also, many investigators enquiring into the nature of the human mind adopt methodological strategies that pay no heed to this assumed priority of language over thought. They simply have ignored it, or perhaps, have been unaware of it. In most cases, paying no heed to this assumed has not mattered. They are, surely, mostly right in the way they proceed. The cognitive processes under investigation are often far removed from the processes of language. At the more basic levels, investigations of mental processes (e.g., the fundamental operation of the perceptual system) have no need for the philosophy of language. Many of these mental processes we share with organisms from which we are phylogenetically very distant—organisms that have not the slightest sign of linguistic competence. However, at higher levels, and as far as we know only in humans, mental processes clearly are connected to language acquisition and competence. Furthermore, there are many strongly held convictions associated with this connection—for instance, that there is no thought without language, or that we think in a language of thought. The first task for investigators of these higher level cognitive events (i.e., thoughts) is to give thoughts a precise specification. In other words, to address the question 'What is a 44 thought?' Answering this question is even more pressing when the independence ofthe philosophy of thought is advocated. The task of this short chapter, then, is to outline in some fairly general way just what is the target of investigation in the philosophy of thought (at least this aspect I pursue here). 2.1 Subjective and Objective Thoughts In answering the question 'What is a thought?', I want to draw a distinction between kinds of thoughts. In other words, the suggestion is that there is not one answer to this question—or, not simply one target of investigation in the philosophy of thought (or in the philosophy of language or mind, for that matter). The distinction is at the core of the main arguments of this dissertation. Unfortunately, it is difficult to know exactly in what terms to put the distinction because so much rides on how it is presented, which should become clear as I try to elucidate it. The distinction of paramount interest is between kinds of thoughts, where a thought is taken to be a particular kind of mental representation, or as it is sometimes put, mental content.' We will leave aside, for the moment, the issue of whether all these things that we call thoughts are genuinely representational. (A thought is 'genuinely representational' if the thought is grounded directly in its world-involving truth conditions—i.e., what in the world it is about. In other words, we can ask of a genuinely representational thought whether it is directed at an object or state of affairs in the world in such a way that it either does, or does not, actually Represent that object or state of affairs.2) Also, it should be noted that this distinction between thoughts as mental contents, or representations, is imbedded in a broader. 1 We need to be careful not to make any assumptions about the extent of the mental, here. 2 Sometimes, instead of using the explicit term "genuinely representational," I will use the capitalized version "Representational" to indicate when thoughts have this quality of truth determinability, in a non-derivative, object-directed way. For instance, with a subject's thought, linguistically expressed as 'Felix is on the mat' or 'The cat is on the mat', we can ask of it if it is true, i.e., that Felix (or a certain cat) is on the mat. This makes it Representational. However, take 'The cat is on the mat' away from all contexts, and we can make no determination of truth, in this way. It is, then, not Representational. It does, though, represent something. It stands for a linguistic structure, that induces in us a kind of thought. This note foreshadows the discussion that follows. 45 distinction between what I will call mental presentations. These are, reasonably obviously, simply things that are presented in the mind. They may be genuinely representational (e.g., as are thoughts about; particular things), but they may not be (e.g., as are, I will argue, Experiences). Some may use the phrase 'mental content' to pick out what I do by using 'mental presentation'. However, I shy away from this because most seem to use 'content' in a rather strict sense to pick out mental events that are genuine representations—or, at least, representations of some sort. Anyone with even a little familiarity with this area will see already the difficulties faced in setting out these distinctions. I will start with a way of setting up part of the distinction that I hope is fairly uncontroversial. It involves looking at two common uses of the word "thought." Distinguishing these two uses is, I think, philosophically uncontroversial, largely because at this level of drawing the distinction very little is said that would draw in too much of what is at the centre of disputes between competing theories of thought, and certain areas of the philosophy of language and mind. Following this, though, I will try to uncover what various philosophers intend to pick out as thoughts. In other words, when investigating thoughts (their structure, content, etc.), I will try to uncover what they are identifying as the target of their investigations. It is difficult to be certain about this sort of thing, but it seems to me that two different mental presentations, might be picked out as the intended target of investigation—i.e., two kinds of thoughts. It is my contention that due attention to this distinction may shed some light on what appear to be significant disputes in the relevant areas of investigation. This will be directly addressed in the latter chapters of the dissertation (chapters 5 -8 ) . Without looking at philosophical or technical work, there appear to be two conventional uses to which we put the term "thoughts," or cognates like "thinking,"3 that are relevant to the investigation of thoughts. Put another way, intuitively, there are two possible, relevant answers to the question 'What is X thinking?' First, it might be taken that we are asking for a report about what is in X's mind. Second, 31 do not mean the sense of "thinking" that means, roughly, the process of reasoning. 46 it might be taken that we are asking for a report as to which objective state of affairs X directs his thinking. The following example explains this division. If I say "I have hurt my shoulder," there is a sense in which you cannot think the thought I express. If you say "I have hurt my shoulder," you are expressing a different thought in that you are thinking about you, not me. If you say "Michael has hurt his shoulder," you are also not expressing the same thought because your thought does not pick out the possessor of the thought as the subject of the thought (and the hurt shoulder), as does my original thought. However, just because you cannot, in one sense, think the thought I express, it does not follow that there is something I have failed to communicate in saying what I did. You know what I am thinking—what thought (in another sense) I had.4 This directs us to thinking that there is a distinction to be drawn that is reflected in two uses of the word "thought." In the first instance, we might say, rather vaguely, that someone has a thought. It is the product of the cognitive act of thinking, and we say that, at least in part, these thoughts make up the content of consciousness.5 In other words, this use of "thoughts" directs us to a particular kind of 'goings on' in a mind. We say things like "I have a thought," "A penny for your thoughts," or "I can see that are you thinking about something funny." A thought, in this sense, then, is something possessed by creatures with the right kind of cognitive capacity, when that capacity is exercised. To have a (new) thought of this kind is to bring about some change in the internal psychological make-up of the subject doing the thinking. Thoughts, in this sense, would normally be said to belong to the subjective realm, and are produced by us. Such thought contents are such that were God to look inside our heads (minds/brains) that is what he would see there. And, we do not need God. If we look inside our heads, so to speak, thought contents of this kind are what we 'see' there. Because of this, it seems quite appropriate to call 4 The example is adapted from one given by Michael Dummett (1993), p. 140. 5 There is some experimental work that seems to suggests that we may possess thoughts which are not part of our consciousness, e.g., those experiments which investigate so-called 'blind-sightedness'. I will leave this issue aside. It is a topic for future research. 47 them Subjective Thoughts. 6 We could call them 'Cartesian thoughts', and think that this helps characterize them because while they are inaccessible to others, they are thought by some to be completely accessible to the subject of the thought in the Cartesian way.7 This might be helpful, but it also might lead to confusions, so the title 'Subjective Thoughts' is preferable. In the second instance, we say that someone grasps a thought. In this case, the thought does not belong to the content,of consciousness of a subject (or any part of the mind). This other use of "thoughts" directs us to certain kinds of things that are, in some way, mind-independent, i.e., out there in the world. We say things like "That's a hard thought to grasp," "They could not grasp the thought that the Earth was not flat," or "It never occurred to them to think 'forces can operate at a distance''." In these cases, and in everyday talk, the words "concept" or "idea" might do the job equally well. Frege said of these thoughts that they belonged (perhaps a little mysteriously) to what he called the "third realm."8 To grasp a thought, in this sense, is to grasp content that is independent of us. It is, in a way, objective, but not like a tree is objective. And it is not intersubjective—that is, as it is sometimes said of some kinds of claims (e.g., 'peaches taste sweet') that truth is obtained by 'inter-subjective agreement'. Ontological oddities aside, we can think of such thoughts as belonging to an epistemic category, the 'objects' of which can be shared between people, i.e., communicated, or perhaps mutually 'observed'. Frege gives as an exemplar "the Pythagorean Theorem"9—that is, not my theorem, 6 There are, perhaps, some reasons for thinking that Externalist considerations might lead us to think that the division between the subjective and the objective is not as rigid as we normally think. For instance, one way to react to Twin Earth considerations is to hold that the mind (in some modified sense of "mind") is not in the head, or that it extends out beyond the head. John McDowell (see McDowell (1982) and (1994)), reacting to some quite different considerations, argues that there is no distinction between subjective and objective. I'll ignore these claims and take it that at least some division is still recognizable. 7 I am not claiming that the content is completely accessible to the subject. 8 Frege (1977), p. 17. 9 Frege (1977), p. 16. 48 nor your theorem, but the theorem—the contents of which, and the truth of which, are independent of anyone thinking it. In other words, it is not the thought arising from thinking about the theorem, which would be a thought in the first sense—a Subjective Thought. Rather, it just is the theorem. Some call these Fregean Thoughts, but to show the contrast with Subjective Thoughts, I will call them Objective Thoughts. Obviously, already, we are extending this analysis a little beyond what would be necessary for everyday uses of "thoughts." However, we need to extend the analysis even further to cover the possible subtleties of uses of the term "thoughts" in philosophical investigations about the nature of thoughts. I acknowledge that the divisions set up here may strike some readers as being somewhat arbitrary, and perhaps even question begging. There will be some kind of conceptual support, but it most probably will be insufficient to ground the distinction solidly. However, it seems to me that, conceptual matters aside, the best way to determine the value of a distinction is to see its value in subsequent philosophical employment. My request, then, is for readers to reserve judgment about the distinction until it has been put to use. No doubt, though, a judgement of the distinction's value based upon what I present will not be the end of the story either. The history of philosophy has been copiously supplied with both good (i.e., genuine) and bad (i.e., counterfeit) distinctions playing pivotal roles in philosophical investigations. I push on, then, with this caution recommended. Subjective Thoughts, as I have described them, are further divided. This division is what gives us the two members of the distinction mentioned in the Introduction—i.e., Object-Independent Thoughts and Particular Thoughts. These are two kinds of thoughts that are distinguished by the kind of informational content (as an individuating condition) they possess. We can use Twin Earth thought experiment consideration to show these different kinds of informational content. Aside from kinds of Subjective Thoughts, Experiences also appear in discussions of Twin Earth cases, and are distinguished from Thoughts by the kind of information they possess. But, as will probably 49 be understood, it is not so easy to determine what, if any, are their individuating conditions. That is, we may have a fairly clear idea of what makes a state an Experience, but I am not sure that we have a clear idea of how to individuate Experiences. I will not be addressing this problem, but it should not matter. However, the idea of an Experience does feature in many investigations, so I will say a little about Experiences, first, and then move on to characterizing these kinds of Subjective Thoughts. 2.2 Experiences When a mental event is the product of bringing into consciousness perceptual informational, it is an Experience}0 It is an important feature of Experiences that the information they possess is limited by the capacities of the perceptual system. They are, in a qualified way, object/context independent. Also, as I will argue in Chapter 7, they are not Representational. Even though an Experience is a kind of mental presentation, it is not a thought." It might be considered that because Experiences are not thoughts, they have no relevance in the investigation being pursued here. However, it is important to introduce and retain them because they do play important roles in many discussions. For instance, in certain Twin Earth thought experiment cases, what is important in setting up the case is that a subject and her twin are experientially indistinguishable. In other words, it is claimed that they are in type-identical, perception-based mental states. Sometimes, this will be identified as a particular kind of narrow content (i.e., contents, narrowly construed so as to exclude certain external determinants of individuation). This can only mean that in these cases, the subject and her twin are in type-identical experiential states, and that these states are the grounding for the relevant content. It may turn out that we end up rejecting the idea of perception-based, narrow content (as being genuine 1 01 am aware that we can have nonperception-based Experiences (e.g., from dreams or hallucinations), but I will leave this complication aside. I do not think that this affects the discussion. 1 ' Even though, as I say, we do often have thoughts that are based upon Experiences. 50 content) in these kinds of cases, but to do so before the discussion gets going, seems to beg an important question. It should be noted that even though experiences are not thoughts, they may initiate or induce thoughts. Sometimes when they do initiate or induce thoughts, we get Thoughts about Experiences. These are thoughts whose informational content is grounded in an Experience, as opposed to grounded in the world. In certain discussions, we need to be careful to distinguish between an Experience and a thought that concerns an experience, but has no contextual grounding. One way to do this is to look at certain sorts of experiments. Let us say that we get a subject to look at the array of lines used to show the Ponzo Illusion. Ponzo Illusion We ask the subject 'Are the lines of equal length?' This question has two interpretations: (1) Do the lines appear to you to be of equal length?, and (2) Do you think (i.e., judge) that the lines are of equal length? (1) forces the subject into a cognitive state where she has to make a judgement of her experience, not the world. (2) forces her to make a judgement about the state of the world. Another thing that highlights the difference is passivity of experience. This comes from the observation that we cannot help but experience 51 what we do—that is, we have no control over our Experiences12—whereas, we do control what Thoughts about Experiences we have. As C. I. Lewis put it, "we do not invent" the content of an experience.13 However, we do invent the content of a Thought about the Experience. 2.3 Object-Independent Thoughts When thoughts are the products of bringing into consciousness linguistic or conceptual information only,14 they are what I will call Object-Independent Thoughts. The kinds of Object-Independent Thoughts I am concerned with in this dissertation are singular in nature in that they concern a single thing, but importantly, not a particular thing." We come to have these thoughts through the exercise of the faculties (or faculty) that govern linguistic competence and the employment of concepts. For instance, I can induce such a thought in you right now by presenting you with, (1) The man on the street corner is selling newspapers. Sentence (1) induces a thought, but no Particular Thought.16 You do not know who is selling what newspapers on which street corner. Nonetheless, you are thinking something.17 Notice, though, that this sentence must be given (and taken) in the absence of any context. The point of the exercise is simply to 1 2 Of course, we can do things like close our eyes, put on rose coloured glasses, or take some sort of hallucinatory drug. But I do not mean this kind of control. 1 3 Lewis (1946), p. 182. 1 4 The issue of whether there are thoughts that are the products of bringing into consciousness something other than linguistic or conceptual information is left open for the moment. So too is the issue of whether the class of Object-Independent Thoughts is empty or not. 1 5 The relevance of distinguishing between singular and particular follows. 1 6 Of course, (1) could be used as an expression of a Particular Thought. But it is not here. 1 7 I am inclined to think that these kinds of thoughts are parasitic (for their production) on Particular Thoughts (to be explained below). That is, in producing the Object-Independent Thought, linguistically represented as sentence (1), we must imagine a man on a street corner selling newspapers. This seems to require that we pretend that the thought is about a particular state of affairs, or perhaps that we imagine that the thought is about a pretend, particular state of affairs. I set this issue aside, though. 52 show what is induced in us mentally through understanding a sentence in a language, in isolation. Adding context might well spoil the exercise. For instance, two people see a man on a street corner, and one asks the other what he is doing. The other responds with (1). Here the reference is to a particular person, so the thought is not object-independent. Some readers may have a problem with the way in which I present the idea of an Object-Independent Thought. As I said in the Introduction, if I have a proper understanding, what I present potentially suggests a nonstandard use of the term 'Singular Thoughts'. Object-Independent Thoughts concerning single things, for many, just are thoughts about particular things. However, I will be relying heavily upon distinguishing between Object-Independent Thoughts and thoughts about particulars things, so it is important to note this nonstandard use. Even though it is not as likely as was the case with perception-based content, an argument might be brought against the claim that Object-Independent Thoughts are appropriately labelled 'thoughts'. As with Experiences, this is because they are not Representational. Recall, this is because the)' are not directly grounded upon world-involving truth conditions (i.e., because they are context/object-independent). However, if for no other reasons, I think that it is valuable to retain the label 'thoughts', first, for the reason just outlined for Experiences, but also for the additional reason that this satisfies the intuitions that we have which are based upon our phenomenological, introspective experiences. In other words, given the kind of things that Object-Independent Thoughts are, it would seem to be an inappropriate technical restriction not to call them thoughts. We see this, in part, in the examples just given, and I think that we see it even more clearly in the example that follows. This example comes from John Perry. In it we are to suppose that someone receives a postcard that has been damaged by water. The only words that remain legible are "I am having a good time now"—neither the sender's name, nor her return address are legible. Also, presumably, the picture on the front gives no indication of the point of origin. Perry writes: 53 If I am a competent speaker of English, I will understand the meaning of the sentence written on the postcard and hence the truth conditions of the utterance that produced it. It is true, if the person who wrote the postcard was having a good time at the time he or she wrote it. This is a singular proposition, with the event that produced the postcard as a constituent.18 However, even though the reader knows that a singular proposition is being expressed and understands what is required, in general, for it to be true, he does not know which proposition is being expressed. Nonetheless, he knows that some proposition is being expressed, so grasps a thought of some kind. I have more to say about this example later. There is another kind of Object-Independent Thought possible.19 These are thoughts whose informational content is grounded in an Experience, and nothing more.20 These are Thoughts about an Experience, but where there is no individuating knowledge available to the subject. As I have said, we need to distinguish thoughts in this manner to account for all that is going on in certain Twin Earth cases. In setting up certain Twin Earth cases, it is often noted that the subject and his twin are 'experientially indistinguishable'. That is, there is experience-based content that is common to the subject and twin. One way of characterizing the common content, or rather, showing that the content is common, is to note that the two could interchange locations and neither would be in a different experiential state. 2.4 Particular Thoughts When thoughts are the products of bringing into consciousness something other than, or more accurately, something in addition to, perceptual, linguistic or conceptual (but still empirical) informational 1 8 Perry (1988), p. 9. 191 have left this explication to the end of this section, first, to avoid an unnecessarily complicated account of Object-Independent Thoughts; and second, because in most cases, when I talk about Object-Independent Thoughts, it will be those grounded in conceptual or linguistic information to which I am referring. 2 0 It might be thought odd to classify such experience-based thoughts as object-independent, and so not genuinely representational, but the reasons for this are given in Chapter 7. 54 content—let's say Representational information, or, Representational beliefs—they are thoughts about particular things in the world. Representational information is information possessed by a subject that allows the subject to link the thought content to the object thought about. The crucial aspect of the Representational information is that it allows for the unique identification of the object (or substance, property or relation). Just what this information is like depends on what is being Represented, and the circumstances in which the thought is produced. For instance, if what is being Represented is some natural kind, the information may concern the characteristics used in its demonstrative identification of that kind.21 Or if what is being Represented is an object, the information may concern simply its location in space-time, as in the case of a steel ball that is otherwise indistinguishable from other steel balls. Again, it is information that links the thought with what is thought about. These thoughts are not produced by us in the way in which we produce Experiences, or Object-Independent Thoughts, though we can discover them through inquiry of the world. We also decide about their veracity through the faculty of Judgement. They are about objects, substances, properties and relations in the world. Given what they are about, we might call them, perhaps a little loosely, thoughts about the world. But to give them a more precise label, and accord with Evans's use, I will call them Particular Thoughts. The informational content of a Particular Thoughts is limited only by the capacities of scientific inquiry and understanding.22 In sum, then, Particular Thoughts are thoughts about particular things in the world. They are Representational because there is a direct relation between the thought and its truth conditions. Also, importantly, they are context/object-dependent. A Particular Thought is individuated by its mode of 2 1 Note that this need not include specific knowledge of its essential characteristics, e.g., water's hidden microstructure, H 20. A person can still demonstratively identify something that he is ignorant of, or even holds false beliefs about. 2 21 leave aside the idea that there might be a special category of knowledge based on the idea that there are special thoughts (Particular Thoughts) that only gods can grasp. 55 identification. The mode of identification of a particular thing allows for the production of individuating knowledge, which, it will be argued, is the grounding for the Representational property of the thought. It is important to note that Particular Thoughts are not somehow totally independent of Experiences or Object-Independent Thoughts. A Particular Thought is likely to contain both perception-based and linguistic-based information. It is marked off, however, by its possession of this additional objective, or individuating, information. I acknowledge that this account is very sketchy, but, of course, for the moment it has to be. Addressing the question 'What makes a thought Representational?' takes up a substantial part of the rest of the dissertation. 2.5 Object-Independent Thoughts vs. Particular Thoughts For my purposes, the important distinction is between Object-Independent Thoughts and Particular Thoughts. However, the standard methodology of the Priority Thesis allows for these to be conflated. This is because one sentence can be associated with either in the context of certain philosophical investigations. For example, consider: (1) The glass contains water. This sentence is taken to specify a particular content, and via the assumption of parallelism, we are not given the possibility that different thought contents might be associated with the sentence. However, (1) can be associated with the Object-Independent Thought that (somewhere and at sometime) a glass contains water (or some other substance that is called water); and, it can be associated with any number of Particular Thoughts that represent the state of affairs where a particular glass contains some water (i.e., a particular substance). Noticing this gives us a different way of characterizing the distinction. First, consider that there is someone, Ryan, who thinks that a particular glass contains water. Typically, this is linguistically represented as: 56 (2) Ryan thinks that the glass contains water. This kind of expression has given philosophers considerable trouble. But in recognizing this trouble, I think that we can see that there are analogues to the distinction I suggest (from the philosophy of thought), in the philosophy of language and the philosophy of mind. In the philosophy of language, most would feel the pressure to draw the distinction between opaque and transparent readings of (2). The opaque reading, then, corresponds to Ryan's Object-Independent Thought, whereas the transparent reading corresponds to Ryan's Particular Thought. Also, as will probably be seen, this distinction is reflected in a number of other distinctions, though they may not be exactly parallel, and the mapping of the associations might well involve rather complicated qualifications. Of particular importance, from the philosophy of mind, is the narrow content/broad content distinction.23 The parallel between this distinction and the one between Object-Independent Thoughts and Particular Thoughts can be seen by noticing that just as broad content is object/context dependent, so too are Particular Thoughts—that is, you cannot have a Particular Thought without an appropriate connection to the relevant object/context. Object-Independent Thoughts, on the other hand, obviously, are object/context independent, parallel to narrow content.24 We see, then, that Object-Independent Thoughts are specified by the content of a linguistic expression. This puts Object-Independent Thoughts into the class of thoughts for which the Priority Thesis is warranted. That is, I have in principle no objection to explanations of the content of Object-Independent Thoughts that employ the Assumption of Parallelism. Particular Thoughts, on the other hand, are specified by a subject-independent 2 3 The quantity of literature on this distinction is extensive. See Putnam (1975), Burge (1979) and Fodor (1980) for early discussions. 2 4 This needs some qualification. Object-Independent Thoughts are not purely context independent; and neither is narrow content. Superficial changes in a subject's local environment—ones altering the perceptual informational content of her experiences—will, of course, give her different narrow content (and different Object-Independent Thoughts). But as is quite familiar, the Twin Earth thought experiments purposely disallow such superficial changes. The relevant contextual changes are of hidden structures—that is, for example, of essences of objects or substances. 57 state, i.e., some state of the (external) world. It is a central claim of this dissertation that because they have this kind of specification, the Priority Thesis is not warranted for Particular Thoughts. As noted, one thing that marks the difference between Object-Independent Thoughts and Particular Thoughts is that Particular Thoughts are genuinely representational, whereas Object-Independent Thoughts are not. What is required for a Particular Thought to be genuinely representational is that there not only be a direct relation between the thought and the object thought about, but also that the subject possess a certain kind of knowledge of this relation. The subject must know that the relation is the right one. That is, the subject must know that there is a symmetry or harmony between the two aspects of the thought. Knowledge of this symmetry amounts to an understanding that the object thought about (to which she is referring) is the one to which she is attributing the property. This gives us yet another, closely related, way of characterizing the distinction. Possession of a Object-Independent Thought, only, amounts to possession of an attributive thought content.25 To have a Particular Thought—i.e., possess, in addition, a referential thought content—the subject must have established (in consciousness) the right kind of link between the attributive thought content and some particular object. A necessary component of this link is that there be a causal-historical relation between the attributive thought content and the object. But this, alone, is not sufficient for referential thought content. What is also needed is that the subject recognize, or have knowledge of, the causal-historical connection. Again, there must be a symmetry or harmony between the two parts of the thought. If there is no symmetry The terms "attributive" and "referential" (below) come from Keith Donnellan's distinction between attributive and referential uses of definite descriptions first outlined in Donnellan (1966). This spawned an extensive line of literature on the topic. It should be noted, though, that there is wide disagreement about what underlies the distinction. 58 (either the subject has no belief, or the belief is false) there is no Particular Thought.26 But this does not mean that there is no thought at all. There is still the Object-Independent Thought. This is where I take it that Evans overstates the issue. And it is often where his critics have attacked his view. However, I will argue that they are wrong to reject the substance of his view just because of this overstatement. What I have presented above is not meant to count as arguments for the distinction. I merely present some material that should help to characterize it. In most cases (non-aberrant, veridical cases) there is a relative symmetry between the content of an Object-Independent Thought and the content of associated Particular Thoughts. Noting this symmetry, or supposing it, is what is required in forming a belief. Similarly, a piece of practical reasoning aimed at satisfaction of a desire requires noting, or supposing, the symmetry. Without that, acting to satisfy the desire would be irrational. 59 CHAPTER 3 Reference There are many ways in which we refer to things in the world. These ways of referring correspond to ways of identifying these things. We have at our disposal a significant array of referring tools that enable us to pick out, usually for the benefit of communication, ourselves (I, me, myself); other people (Democritus, Auntie Doris, him, she, them, the fool on the hill), individual objects (that big old maple tree beside our house, the house on the hill, Mount Egmont); sets of objects (tigers, humans) or kinds of substance (water, gold); properties (green, hard); relations (bigger than, in front of); times (10 past 12, now, tomorrow, the summer of '69); and places (here, there, behind the bike-shed). Perhaps a little surprisingly, we also have ways of referring to things that are not in the world, i.e., nonexistent or fictitious people (James Bond); individual objects (Vulcan); sets of objects (unicorns); and kinds of substance (phlogiston). However, these ways of referring to nonexistent or fictitious things are special cases, and are probably in some way parasitic on the other, perhaps we would say, genuine ways of referring.1 To narrow the focus of their investigations into referring terms, philosophers typically draw a distinction between singular reference and general reference (though some of what is said of each is applicable to the other). Now, it might be thought that this distinction simply is accounted for by thinking that when a term is used to refer to individual people, objects, etc., reference is singular, i.e., reference is made to some singular thing. And when a term is used to refer to some set of objects or kind of These ways of referring to nonexistent and fictitious objects have caused philosophers a great deal of trouble, and I will set them aside. Also, we might complete the list so to match the list of kinds of existent things, and add, e.g., places (Xanadu); and times (noon on the day between Tuesday and Wednesday); and non-existent properties and relations. But this seems to lead us into deep troubles. For instance, we now know that we must not split space and time when locating things. So, given this, I am not sure that it even makes sense to say that we can refer to some non-existent 'place-time'. 60 substance, etc., reference is general, i.e., reference is made to multiple things, relevantly associated. Noting this aspect of term use is certainly part of what establishes the distinction. To give some largely uncontroversial examples, "Trees are big" is an expression containing general reference, and "That tree is big" is an expression containing singular reference. However, as it is typically drawn, this does not quite capture the distinction. It seems that in many instances, "singular," in the phrase "singular reference," means "particular." Genuine, singular reference, then, requires not only that just one object is pinned down (through satisfying a relevant description), but that a particular one is.2 I will refrain from giving an example to characterize this distinction, here, partly because such an example is likely to be seen as problematic, and partly because I have more to say about the distinction later. Distinguishing between "singular" and "particular" will become especially important when we move from reference to thought. As was suggested in the previous chapter, at a particular level of investigation, I think that we gain something by being able to distinguish between a singular thought and a Particular Thought. For the moment, though, I will continue with the label 'singular term' because this is what is common in the relevant literature. My focus in this chapter is singular reference, as opposed to general reference. However, in developing a theory, or picture, of singular reference, I will draw on some very important work done in analysing certain cases of'general reference'. An enormous amount of work has been done on reference this century in philosophy. I do not propose to supply a thorough survey of this work in this chapter. In any case, such a survey is not necessary because there are many works are thorough.3 The theory of singular reference outlined, here, 2 Singular reference can be to any finite number of particulars as well. 3 See for example, P. A. French et al. (1979b) and (1989); Blackburn (1984), ch. 9; and Recanati (1993). 61 follows closely the theory put forward by Gareth Evans.4 Evans had the work of Gottlob Frege and Bertrand Russell as his principal influences, but seems to have had additional, important influences from the work of P. F. Strawson, Saul Kripke and John McDowell. In outlining Evans's theory I will try to acknowledge the most important influences from these writers when appropriate. Also, although my agreement with Evans is substantial, it is not complete, so I will try to show the deviancies. I say "try to" because some of Evans's ideas are difficult to uncover, so I must acknowledge that sometimes when I think (and say) that I am expressing an idea from Evans, it might be an erroneous interpretation. The theory can be roughly characterized by these statements: (1) All genuinely singular (i.e., particular) reference is indexical in character. (2) Singular terms refer directly—nothing mediates the reference. (3) Fixing the reference requires having (causally related) identifying knowledge of the referent (i.e., requires a Particular Thought). (4) Both speaker and hearer contribute in successful referential communication. Some will find, straightaway, certain tensions arising with these statements. Especially, it is likely that readers will be troubled by the tension between (2) and (3). This tension, of course, will be addressed. All that I will say for the moment, with the hope of perhaps alleviating a little of the discomfort, is that the tension comes from assuming that the only way to have identifying knowledge is to have some definite description (or set or cluster of identifying descriptions) carrying the knowledge. However, Evans rejects this assumption, arguing instead for a view in which an object of thought is identified by linking information (causally derived from the object) with the thought associated with use of the referring term. This information is the product of the information system, which, in a sense, comes before language. This is at least part of what signals the Linguistic U-Turn. We get, then, the idea that singular reference must be explained in terms of Particular Thought (i.e., what it is to be thinking of the particular object), and 4 The principal source is Evans (1982). 62 what brings it about that a subject is thinking of a particular thing is governed, first, by the possession of appropriate information, and second, by having an appropriate cognitive link between the information and the thought. As a way to introduce the relevant ideas about singular reference, I will present and discuss some ideas related to (1) and (2) in sections 3.1 and 3.2 respectively. (3) will be addressed in the section on Evans on reference, section 3.3. Finally, the ideas associated with (4) will be introduced in this chapter, but will be the focus of Chapter 4. 3.1 The Indexical Character of Singular Reference There has been a great deal of work done in the latter part of this century on indexical reference.5 This is probably because of the apparent success, and subsequent rise in popularity, of theories of direct reference, along with the view that the way that ordinary indexicals refer is the paradigm of direct reference. The roots of the idea go back to at least the beginning of the century. Frege gives us a sketch of the idea that singular reference is indexical in character in his essay "Thoughts."6 I will present this sketch, but we need a little background discussion first, so that the point will be clear when it is made. In this essay, Frege introduces a special sense of "thoughts." This sense points to what have come to be known as Frege Thoughts1 For Frege, thoughts are objective in the sense that two people can share the same thought. He is not expressing a belief in telepathy (as telepathy is commonly understood), but rather, the idea that given appropriate circumstances, I can grasp the thought you are thinking by coming to know what it is that you are thinking. Thoughts, in this sense, are out there in the world, so to speak, 51 follow Kaplan in thinking of demonstratives as a subclass of indexicals. So, when reference is made to indexicals, as in this section heading, I mean to include demonstratives. See Kaplan (1989), pp. 489-491. 6 Frege (1977), pp. 1-30. 71 have called them 'Objective Thoughts'. 63 for people to grasp. And it is principally through the medium of language that we can get others to grasp the thoughts we intend them to grasp—although, of course, language is only part of the story, and is not always necessary. However, a sentence does not just express a thought. It has some other communicative elements as well. Frege writes: "An interrogative sentence and an assertoric one contain the same thought; but the assertoric sentence contains something else as well, namely assertion. The interrogative sentence contains something more too, namely a request."8 Further, Frege says that an "assertoric sentence often contains, over and above a thought and an assertion, a third component not covered by the assertion."9 In general, this amounts to the claim that sentences often contain "conversational suggestions" and can undergo surface transformations without affecting the underlying thought.10 From this he concludes that "the content of a sentence often goes beyond the thought expressed by it."" This point is the first to which I want to draw attention. The quotation seems to be an early expression of the idea underlying the division between Semantics and Pragmatics. Of course, noting this division, and understanding it as we now do, we would criticize Frege's expression of the idea. That is, nowadays writers would probably want to avoid saying that an assertoric sentence contains some third component. Some may object, but commonly it is thought that Semantics is the realm appropriate to the investigation of the contents of sentences, whereas Pragmatics is the realm appropriate to the investigation of these 'third components'. In saying this, though, Frege's introduction of the idea of "conversational suggestions" seems quite close to the mark. Frege (1977), p. 7. 9 Frege (1977), p. 8. 1 0 Frege (1977), pp. 8-10. 1 1 Frege (1977), p. 10. 64 Having observed that the content of a sentence may go beyond the thought expressed by it, Frege notes that the reverse is also possible. This point is the on which I want to focus. He writes, "the mere wording, which can be made permanent by writing or the gramophone, does not suffice for the expression of the thought."12 The point is first made with sentences containing indexicals like "yesterday" and "here," and personal pronouns like "I." Frege writes: In all such cases the mere wording, as it can be preserved in writing, is not the complete expression of the thought; the knowledge of certain conditions accompanying the utterance, which are used as means of expressing the thought, is needed for us to grasp the thought correctly.13 Further, Frege suggests that proper names also have this feature. That is, the thought about the person named in a sentence by use of a proper name relies upon, in part, for its full content, something beyond what the sentence, in itself, expresses.14 We see this reliance when we consider a sentence containing a proper name used by two speakers, and where the object named is presented to each speaker in different ways. In this kind of case, different thoughts can be associated with the same sentence. Frege then says that, "Different thoughts thus obtained from the same sentences correspond in truth value . . . Nevertheless the difference must be recognized."15 It must be recognized, Frege thinks, because the sense of a sentence containing proper names varies according to the ways in which the object named is presented. And with this variation comes variation in thoughts. This general point can be put to use beyond what Frege, here, intended. The idea that the sense of a sentence containing a proper name varies according to how the name is presented is also displayed 1 2 Frege (1977), p. 10. 1 3 Frege (1977), p. 10. 1 4 There is a way in which this point may seem to be trivially true given that there could be two sentence tokens of the same type, e.g., 'Aristotle is wealthy', that express different thoughts simply in virtue ofthe fact that each occurrence of "Aristotle" names different persons. But Frege is not alluding to this kind of 'ambiguity' in proper names. 1 5 Frege (1977), p.12. Actually, it is this difference that makes the Linguistic U-Turn possible. 65 in the possibility of an agent having different epistemic attitudes towards two thoughts, each of which is associated with the same sentence. An agent might have two thoughts associated with two tokens of the same sentence containing the same name (naming the same object) but where the name is presented in different ways. For example, consider Ryan's thought (call it thought,) about the person he just met and heard recite a few poems, which we would express as (1) Cyrano is a poet. Consider, also, Ryan's thought (thought^ ) about the author of a book of poetry he read some time ago, which we would also express as (1). However, in this situation Ryan does not know that "Cyrano" names the same person whom he just heard recite the poems. Now, it is clear that without thinking that Ryan is quite irrational, we could imagine that he thinks that Cyrano (as the subject of thought,) is a wonderful poet, but also that he thinks that Cyrano (as the subject of thought^) is an awful poet. This could be so simply by it being the case that the poems recited were good (according to Ryan's poetic judgements, that is), the poems published were bad, and that Ryan does not know that Cyrano-the-reciter and Cyrano-the-author are the same person. This fact about Ryan and his thoughts suggests that different information can be associated with the same name presented in different ways, which further supports the idea that there is no guarantee that a sentence by itself will express any given associated thought. That there is no such guarantee leads us, at least, in the direction of thinking that proper names as operating in the above kind of case have a context-dependent element or character. Part of what completes the picture comes from looking at contemporary theories of direct reference. But looking briefly at some contemporary work that focuses especially on indexicals will be helpful before we proceed. David Kaplan's work on demonstratives is a good example. Kaplan argued against the view, commonly attributed to Frege, that singular terms refer via the mediation of Fregean Sinn, where Sinn is understood as something like descriptive content. In opposition, 66 he defends a view of singular terms whereby such terms are directly referential—that is, they pick out their relevant object without the mediation of descriptive content—or rather, without only the mediation of descriptive content. Such terms, Kaplan says, have "two kinds of meaning."16 The fact that we can mean different things by two uses of the same sentence containing an indexical referring term, leads Kaplan to say that the two uses of the sentence have different content. That is, the sentence can express different propositions, or Fregean thoughts.17 However, Kaplan notes, it will be widely agreed that there is something common between the two uses of the sentence. In other words, that they, in some other sense, mean the same thing. This Kaplan calls the character. He says, "The character of an expression is set by linguistic conventions and, in turn, determines the content of the expression in every context."18 It should be fairly obvious that part of what drives the distinction between character and content is that the distinction allows us to explain how it is that you and I can say the same thing, yet mean different things, e.g., if we both say "I have hurt my shoulder." However, it should be noted that the distinction seems to be at least partly motivated by what led Frege to distinguish between Sinn and Bedeutung—namely, that a subject can hold different epistemic attitudes towards two sentences that, the distinction aside, mean the same thing. What we might call naive direct reference theories have a problem here. Consider the following given in some propositional attitude context, say, 'Ryan thinks that' (1) He is a spy and (2) That man is a spy where (given direct reference) (1) and (2) pick out the same individual, and say of him that he is a spy. Following Kaplan, we can explain how it is that Ryan could think (1) true, but think (2) false, by saying 1 6 Kaplan (1989), p. 500. 1 7 Kaplan (1989), pp. 500-1. 1 8 Kaplan (1989), p. 505. 67 that even though (1) and (2) pick out the same man—i.e., have the same content—they have different character. In Kaplan's terms, (1) and (2) have the same object of thought, but have different cognitive significance (because the thoughts are presented in different ways).19 It is important to be aware, and Kaplan is quick to point out, that he has "not simply generalized Frege's theory."20 As Frege's theory is commonly understood, the mode of presentation associated with a referring term is constant across different uses, within a language community (i.e., the descriptive content of the term is constant within any use in that language community). The theory, then, cannot account for the context sensitivity of indexical expressions. If an indexical term can be used to refer to different things in different contexts, it cannot be the descriptive content that determines to what it refers. This point is important in understanding what Kaplan is saying. However, given what is presented above from Frege, it is neither clear that the common understanding of Frege is correct, nor that Frege always held the position represented by the common understanding. Recall, the passage from "Thoughts," quoted above. Frege notes that "the mere wording, as it can be preserved in writing, is not the complete expression of the thought; the knowledge of certain conditions accompanying the utterance, which are used as means of expressing the thought, is needed for us to grasp the thought correctly."21 What Frege is saying here is that modes of presentation are not (always, at least) constant in the way implied by the anti-Fregean arguments. The quotation seems to allow for a picture to be constructed such that 'knowledge of certain conditions accompanying utterances' provides sufficient flexibility to account for the context sensitivity of indexicals. What exactly do we mean by context sensitivity? Various answers are given to this question based upon the way in which it is thought that the term, the object, and (in some views) the subject are related; 1 9 Kaplan (1989), p. 530. 2 0 Kaplan(1989), p. 530. 2 1 Frege (1977), p. 10. Italics added. 68 where the relation is direct, of course. We will see others as the dissertation proceeds, but here is Kaplan's account. In "Quantifying In," Kaplan argued that a genuinely representational thought about an object requires that the subject of the thought be "en rapport" with the object.22 Two things are required for a subject to be 'en rapport' with an object. First, the subject must be in possession of an appropriate singular term whose "descriptive content" denotes the object. This provides one of the links to the object, and also determines the "vividness" of the term (or thought). Getting a precise idea of 'vividness' is not easy, but roughly, the more detailed the description or conception of the object named, the more vivid the name.23 Second, the subject must be in a suitable causal relation with the object. This second link provides the "genetic character," or lqf-ness\ of the associated singular term, and makes it that the thought is about the particular object, not some other one which the term might also denote. Kaplan explains this with the analogy of the "resemblance" (analogous to 'descriptive content') and "o/-ness" (analogous to 'genetic character') of a picture. That is, two pictures might be 'of two different people (e.g., twins), so 'resemble' one another that either picture could be used to point to either of the twins. The basic point can be applied to singular referring terms. The idea is that the explanation ofthe operation of demonstratives (and other referring terms, e.g., other indexicals, referentially used definite descriptions and genuine proper names) requires that we take account of the fact that these terms are context sensitive. It is true that Frege's theory of reference as it is commonly understood cannot account for indexicals, in addition to any referring term that operates in an indexical like way. However, as I suggest above, it is not clear that Frege was unaware of the indexical-like character of singular terms, including proper names. I must point out, though, that I do not present these views from Frege and Kaplan with the 2 2 See Kaplan (1969), sec. IX. Note: Kaplan later changed his mind about this. See Kaplan (1990), p. 327. 23 Analogously, the clearer and more detailed the picture, the more vivid the picture. 69 intention of endorsing the views. A l l that I want to achieve is to characterize what is meant by the indexical character, or context sensitivity, of singular terms. 3.2 Direct Reference Two kinds of theories of singular reference have competed for acceptance this century. These are description-based theories and causal-historical,24 or direct reference, theories.25 I will not trace the history of the development of these views, but sufficient details of the general nature of the theories will be presented in the discussion to follow. I will give a very brief outline of description-based theories first, so that it is clear just to what the theorists of direct reference reacted. To give a small piece of history in this area, the development of description-based theories traces a path from Frege and Russell, through Strawson, to Searle. The development of theories of direct reference is typically regarded to have its origins in the work of Kripke, Donnellan and Putnam, but Ruth Barcan Marcus's "Modalities and Intensional Languages" seems to preempt some of their ideas, and certain aspects can be traced back to J. S. M i l l in A System of Logic. The dispute between description-based theories of reference and theories of direct reference can be put quite simply. On the description-based theory side we begin with Frege.26 Frege argued that proper names must have sense(s). He gets us to consider two identity statements (i) 'a=a' and (ii) 'a=b' where 'a ' and 'b' name the same thing. Now, it is noted that (i) is trivially true (or analytic), but that (ii) is 2 4 Sometimes I just say 'causal theories' for the sake of convenience, but I always mean 'causal-historical theories'. 2 5 The "or" in "causal-historical or direct reference theories" is genuine, even though the titles "causal-historical reference theory" and "direct reference theory" are nearly always taken to be interchangeable. Probably for very good reasons, the only direct reference theories proposed are causal-historical theories. However, someone could propose that reference is unmediated (i.e., direct), but where the reference is effected by some kind of non-natural relations as opposed to causal-historical relations. Of course, supposing a non-natural relations seems odd to us nowadays, but as pointed out to me by Paul Bartha, the Scotist idea of the relation of the "uniting Medium" may be such a relation (see Duns Scotus (1975), p. 293.) 2 6 See Frege (1960), p. 56ff. 70 (potentially) informative, which, Frege argues, shows us that (i) and (ii) must differ in cognitive value. And given that 'a' and 'b' refer to the same thing, something other than reference must account for the difference in cognitive value, namely, what he called sense (Sinn). Reference, then, must be mediated by sense because it must be the case that we are led to an object named via different paths, so to speak, so that the difference in cognitive value can be generated. Further, crucially for the dispute as it turns out, sense is taken to be characterized in terms of sentences that contain descriptions. Frege said that the sense of an assertoric sentence was a thought.27 Followers of Frege took it that thoughts were only ever characterizable as sentences containing descriptions. As has been noted, it is not clear that Frege accepted something like modem description-based theories of proper names. However, it does seem quite clear that his views led others to such theories, i.e., ones that held that reference was mediated by descriptions. On the direct reference side, we have arguments that seem to show conclusively that proper names, to take an important example, refer directly, i.e., are not in any way mediated by descriptions. Very quickly, to set the scene, names refer directly because descriptions identify attributes of what is named, but it seems true to say that it is not the case that the objects named—these individuals—necessarily have these attributes. But description-based theories do imply that the objects named necessarily have the attributes identified by the description. We will see how the arguments work that show this in what follows. Before we proceed, though, it will be helpful to point out roughly how Evans's picture fits in here. It might be thought that Evans's theory of reference falls just in the same way that theories of mediated reference fall. I will not do anything to support his stance at the moment, but I will make two, brief points. First, Evans objects to the understanding of Fregean sense (Sinn) where it 2 7 There is a complication here. Do non-assertoric sentences (e.g., imperatives) express thoughts? Frege thought not, because they are not the sort of thing to which the question of truth applies. They have Sinn but no Bedeutung. It may help here to add the qualification, 'or the assertoric part of a sentence', which could be explicated by distinguishing between the illocutionary intention behind the act of uttering a sentence and the propositional content. It was something like this distinction that was behind R.M. Hare's distinction between phrastic and neustic parts of imperative sentences—which, by the way, he introduced before Speech Act theory was developed. See Hare (1952), ch.2. 71 is interpreted as the sense of some sentences containing (some set of) definite descriptions, which led some to the idea that it is descriptions that mediate reference. Second, he argues that it is "wrong-headed . . . to consider a Fregean sense as intermediary between thinker and referent.... A way of thinking of an object is no more obliged to get in the way of thinking of an object, or to render thinking of an object indirect, than is a way of dancing liable to get in the way of dancing, or to render dancing somehow indirect."28 This passage is at least suggestive of an explanation of how Evans can hold that reference is direct, while at the same time maintaining the requirement ofthe possession of individuating knowledge.29 In general, description-based theories of reference typically have been thought to have at their centre the idea that associated with a proper name (the focus has traditionally been theories of proper names) is a definite description, or a set or cluster of identifying descriptions. So, for example., if some one was to say "David Lange is the ex-prime minister of New Zealand who campaigned against nuclear weapons," I can understand the reference made by taking it that the named object (person) is picked out by (or the name has as its meaning)30 some uniquely identifying description. This description might be, following our example, 'the ex-prime minister of New Zealand who campaigned against nuclear weapons'. But any uniquely identifying description(s), it is thought, will do—perhaps, 'the prime minister of New Zealand who instigated the right-wing economic reforms during the 1980s'. That is, the object referred to is the one that satisfies some identifying description(s). Or to put it another way, if we were to ask the speaker of the above utterance, 'Who is David Lange?' she could give us, assuming she has a proper understanding of the name, some description (or some set or cluster of descriptions) as expressing bits of 2 8 Evans (1985), p. 302. 29 See section 3.6 for further discussion of Evans's position. 3 0 This division reflects alternative kinds of description-based theories. It is more clearly outlined in discussion of Kripke's arguments against description-based theories presented below. I should say, though, that the view that a name means some uniquely identifying description probably has few, if any, contemporary advocates. However, Bertrand Russell seems to express such a view in, for example, "Lectures on Logical Atomism," in Russell (1956). 72 information that uniquely satisfy the referent, David Lange. It is certainly this kind of view to which causal theorists react, so it is important to see this. It is also typical of the way description-based theorists have argued. Description-based theories, then, have as significant intuitive support the idea that each name has associated with it some uniquely identifying information. This idea gives support because it seems highly plausible to say that if a person is to refer to someone by use of a proper name, she must know to whom this name refers. That is, she must be able to identify which particular person or object she refers to. Further, many, on both sides of the debate, have taken it that this uniquely identifying information is only ever to be found as expressed as definite descriptions. Nearly everyone thinks that having identifying information amounts to having in one's possession a singular (or set or cluster of) identifying definite description that could be communicated i f necessary.3 1 It is by possession and transmission of this information that speakers can refer as they do. Also, it is by remembering this information, and then making a correct association between it and the relevant object referred to, that people can re-identify the referent. This account of re-identification seems to offer further support for at least the most basic idea of description-based theories—i.e., the idea that there is informational content that we use to re-identify an object. Most take it that for description theorists, this informational content is made up of descriptions. Roughly, their idea is that we match descriptions associated with the name with the object named, and if there is a match, we re-identify the object. In sum, it is claimed that it is description-based theories which allow us to explain the phenomenon, or our capacity, o f re-identification. As general points about reference and re-identification, these claims are taken to give description-based theories quite strong intuitive appeal. We will see, below, this idea being contested, which will crucially allow us to hold onto what seems intuitively appealing about the description theorists' view, while admitting what seems most convincing about the direct theorists' arguments against description-based theories of reference. 73 As I have said, what identifies description-based theories as description-based theories is the claim that this uniquely identifying information is made up of a uniquely identifying description, or a set or cluster of descriptions. Theorists were led to this view by their (tacit, or occasionally expressed) acceptance of the Priority Thesis. It was probably not even contemplated that uniquely identifying information could be characterized or possessed in any form other than linguistic expressions, i.e., sentences containing definite descriptions. Unfortunately, this is where they got into trouble. Some powerful arguments have been brought to bear against description-based theories, with the attack focusing on the core claim that proper names have associated with them uniquely identifying descriptions. What resulted from these arguments were, so-called, new theories of reference, i.e., theories of causal or direct reference.32 The Causal Theory of Reference is now well known, but I will give a brief account. Saul Kripke, Keith Donnellan and Hilary Putnam are commonly cited as the founders of the Causal Theory of Reference, but I will present only Kripke's arguments because he is the one that Evans explicitly follows.33 It is Kripke's 'picture' that Evans modifies to produce his theory of singular reference, or more accurately, particular reference. The picture of singular reference that Kripke develops in "Naming and Necessity" has two parts: a general attack on description-based theories, and an account of what he thinks is the correct understanding of how proper names function in language. The general attack can be further divided into modal arguments and arguments based on an investigation into the nature of naming practices (or loosely 3 2 As I mentioned earlier, the causal theory is typically thought to be a kind of 'direct reference' theory and is sometimes called the "new theory of reference," though it is not that new, as notes Keith Donnellan in "Belief and the Identity of Reference" (in French, et al. (1989), p. 275). In fact, if Michael Devitt is correct, it is not new at all. See his "Against Direct Reference" (in French, et al. (1989), pp. 206-212) for a breakdown of the theories and a brief account of the history of their development. Presumably the 'old' theories range from traditional ideational theories to modem description-based (cluster) theories. 3 3 See Saul Kripke (1980). 74 non-modal arguments) where these non-modal arguments lead directly into his account of what he considers is the correct picture. I do not intend to present Kripke's arguments in detail but I do think it will be useful to consider some examples so as to get an idea of what the causal theory rejects and what it advocates. Kripke notes that there are two possible ways in which description-based theories may be viewed. First, it might be considered that a description, or set or cluster of descriptions, "actually gives the meaning of the name."34 Second, it might be considered that "even though the description in some sense does not give the meaning of the name, it is what determines its reference ," 3 5 Kripke ends up arguing against both views. Concerning the first view, for it to be true that descriptions give the meaning of names, it would have to be the case that a proper name had the same meaning as (was identical in sense with) some associated description(s). Kripke's suggestion is that what leads some to think that this view is true, is a conflation of giving the meaning and fixing the reference of a proper name.36 We can see why this is a problem, for in giving the meaning it would be the case that the description(s) would 'name' the same object in all possible worlds37 (i.e., in all counterfactual situations), whereas in fixing the reference the description(s) would merely 'pick-out' whatever object satisfies it (them) in the actual world. Kripke agrees with the general point that when a proper name-using practice is initiated, there is an association between some 'condition(s) of identity' and the corresponding referent. But he does not think that this 3 4 Kripke (1980), p. 258. 3 5 Kripke (1980), p. 258. 3 6 There seem to be intuitive reasons for agreeing with Kripke here. For instance, when we are unsure about someone's use of a name, we do not question 'What do you mean (by that term)?' Rather, we question ' Who do you mean?', which suggests that we are asking 'to whom are you are referring?' 3 7 In contemporary philosophy, reference to a 'possible world' is a common way of referring to a set of possible states of affairs that could have arisen—that is, had things in the actual world been different. In other words, it refers to a counterfactual situation. 75 association leads to determining the meaning of a proper name from the description(s), though it might on a particular occasion fix the reference. So why does he think that this association between the initial conditions of identity and the referent is not what gives the meaning of a proper name? The first thing to note is that his argument hinges upon his particular views of necessity and the notion of identity across possible worlds, which leads to his notion of rigid designation. A large part of Lecture I of Naming and Necessity is given over to discussion of these concepts, but I will not trouble the reader with these details. I think that a sufficiently clear idea of rigid designation (the most important idea for our purposes) should emerge. Given these ideas, then, Kripke's argument (for a particular example) is: "If the name means the same as that description or cluster of descriptions, it will not be a rigid designator. It will not necessarily designate the same object in all possible worlds, since other objects might have had the given properties in other possible worlds."38 Consider this nonstandard example. Someone might say "Cyrano de Bergerac was the famous French poet and duellist with a large nose." If it was the case that the name 'Cyrano de Bergerac' means, for instance, 'the famous French poet and duellist with a large nose', then in some other possible world, where someone else was famous. French, poetical etc., and Cyrano died as a child, that person would have been Cyrano de Bergerac. But that is absurd, which shows that the statement 'Cyrano de Bergerac was the famous French poet and duellist with a large nose' is not a necessary truth—that is, not true in all possible worlds. In other words, the name is not synonymous with the description, or any description (contingently) associated with Cyrano de Bergerac. Kripke's position is simply that proper names have no Fregean sense, or as Mill would have put it, they have no connotation, only denotation.39 Clearly, if Kripke is right, then this first version of description-based theories of naming is a non-starter. J 8 Kripke (1980), p. 276. 3 9 Though Mill thought that it was only proper names that were like this, whereas Kripke thinks that other singular terms, as well as natural kind terms, also have only denotation. 76 What, then, of the second view, that a description, or set or cluster of descriptions, fixes the reference of a proper name? Kripke's manouevre, here, is again to consider the association between a name and some identifying description(s). This time, though, rather than considering the metaphysical possibility of alternative states of affairs and how this affects the association between proper names and descriptions—which gives rise to the modal arguments—he considers the epistemic possibility of our actually (in the actual world) being mistaken in associating a particular name with a particular description. Following Rostand's story, Cyrano, for some time, provides the poetry and romantic letters for Christian, a competitor for Roxane's affections. Roxane is unaware of this deception and might, in response to the question 'Who is Christian?', respond 'The man who writes the most wonderful letters to me'. Now as it happens, Christian is not the man who writes these most wonderful letters. He writes really rather bad letters. As said, Cyrano writes the wonderful ones. Clearly here, the name "Christian" is mistakenly associated with the description, and the description is true of someone else—Cyrano. Given these facts, if it is thought that a description determines the reference, then when someone uses the name "Christian," and only has the associated description 'the man who writes the most wonderful letters to Roxane', she actually refers to Cyrano de Bergerac, because it is Cyrano that fits the description. But, of course, it is absurd to think that Cyrano is the referent of "Christian" in this kind of case.40 Clearly then, it is not necessarily the case that a description even commonly associated with a name determines the reference of that name. What Kripke puts forward as his positive account is what he calls an "intuitive thesis." He suggests that proper names are rigid designators—that is, they designate the same object in any possible We could, taking account of the distinction between speaker's reference and semantic reference, describe a situation where the description did pick out Christian. For instance, when a speaker said "The man who writes the most wonderful letters to Roxane," but thought that this man was Christian; and where the hearer, although knowing that this actually refers to Cyrano, knows further that the speaker thinks that it refers to Christian, so takes the speaker's intended reference not the actual one. [Though some might debate which is the 'actual' one.] 77 world.41 Very briefly, Kripke's intuitive argument is the following: "proper names are rigid designators, for although the man (Nixon) might not have been the president, it is not the case that he might not have been Nixon (though he might not have been called 'Nixon)."42 In general, proper names retain reference across possible worlds, even though it might be the case that the object named changes characteristics such that we associate different identifying descriptions with it. Further, he thinks that "in general our reference depends not just on what we think ourselves, but on other people in the community, the history of how the name reached one, and things like that. It is by following such a history that one gets to the reference."43 In other words, it is through causal-historical chains originating in baptisms that the reference of proper names is determined. Kripke emphasizes that he does not think that he is advancing a theory. Rather, all he purports to do is present a better "picture" of the practice of using proper names to refer to objects than the description-based theories. Briefly, this is the picture. A name is given to a person, or an object, in a baptism. That is, a name-using practice is initiated by someone (or some group of people) directly and explicitly associating a name with an object, i.e., by ostension or description. That specific name-using practice is then passed from speaker to speaker along a causal-historical chain of a particular sort (i.e., each user is a 'link' in the chain, where the interconnection between these links is facilitated by the transfer of that particular name-using practice from one user to another). The interconnection between Not all who advance theories of direct reference accept Kripke's notion of rigid designation as it is explained by Kripke. As we have seen, Kaplan speaks, rather, of terms being 'directly referential'. This characterization is preferred because some terms may just rum out to be 'rigid', but not refer directly. (See Kaplan (1989), p. 492-3). Also, with respect to Kripke's work, Kaplan says that "There are two 'definitions' of 'rigid designator' in Naming and Necessity, pp. 48-49. The first conforms to what seems to me to be the intended concept—same designation in all possible worlds—the second, scarcely a page later, conforms to the more widely held view that a rigid designator need not designate the object, or any object, at worlds in which the object does not exist" (Kaplan (1989), p 493, note 16). 4 2 Kripke (1980), p. 269-270. Kripke (1980), p. 301. 78 each new use of the name is based upon the speaker's communicative intention to use the name to refer to the same object as before, ultimately reaching back to the baptism. The apparent main advantage of this picture is that it seems to explain how it is possible that speakers can use a name correctly to refer to an object even though they may possess no description (or set or cluster of descriptions) that uniquely identifies that object. Or, indeed, that speakers may even associate false descriptions with the name and still (apparently) refer. This picture allows the causal theorists to account for the above examples, ones that are counterexamples to description-based theories. As stated above, I will present and consider advancements of this 'theory' of reference made by Gareth Evans. Evans accepts quite a bit of what is advanced by Kripke, and by association, other causal theorists. However, where he diverges, the differences are significant. 3.3 Evans on Reference Evans begins building up his theory of reference by accepting a fundamental claim of causal reference theories—namely, that reference is direct. He accepts the arguments to the effect that, in reference, contact with the world is not indirect, i.e., mediated by descriptions. However, he does think that Fregean sense (construed in an unconventional way) has a crucial role to play in explaining reference, and in general, how the mind interacts with the world in production of thought. What this role is will be discussed later in more detail when we look at theories of thought (especially in chapters 5 and 6). Evans notes that there are two description-based theories.44 The first is a theory about a speaker's intended reference on a particular occasion of use of a referring term in an utterance. He characterized this as a theory of "speaker's denotation."45 With respect to speaker's denotation, description theorists would say that a name refers to that person or object which satisfies some description (or set or cluster of Evans (1985), p. 2ff. Evans (1985), p. 2. 79 descriptions) which the speaker associates with the referent. The second is a theory about the intended use of a referring term by a particular group of users (perhaps some 'experts'), where some convention is established to the effect that the term names the designated object. Evans characterized this as a theory of "what a name denotes."46 With respect to what a name denotes, description theorists would say that the referring term refers to that person or object which satisfies the description (or set or cluster of descriptions) that this group commonly, and probably conventionally, associates with the referent, perhaps based upon the name-using knowledge of some experts. Evans observes that (early) Kripke failed to distinguish between these two kinds of description theories. However, what underlies the distinction seems to be closely related to the distinction we nowadays employ, which Kripke emphasised in later work, between speaker's reference and semantic reference?1 The connection between the two ideas may well be reflected in the distinction Evans draws between "what the speaker denotes" and "what the name denotes."™ Also, it is, in part, seen in Evans's discussion of the notion of a "speaker's intended referent" in The Varieties of Reference?9 The point is perhaps not obvious in Evans's discussion, but it would seem that it is implicit in the notion of speaker's intended referent, as well as his noting the importance of discussing it, that Evans was alluding to something closely connected with Kripke's speaker's reference and semantic reference distinction. The clearest textual support for this claim is the following. Evans notes that in communicating, "the speaker must use a referring expression conventionally suitable for reference to the intended object." Later, however, he acknowledges that "it is clearly absurd to restrict Evans (1985), p. 2. See Kripke (1977). Evans (1985), p. 1. Evans (1982), p. 316ff. 80 reference to just those cases where referring expressions are used which are conventionally associated, in a context of utterance, with a uniquely identifying referential feature."50 I leave aside the issue of whether Evans was fully aware of the distinction between speaker's reference and semantic reference as Kripke drew it. What is more important is that in the discussions to follow, we be careful to take the distinction into account. That is, it will be crucial to distinguish between to what (or whom) a term refers, and to what (or whom) a speaker refers in referential communication. After pointing out that there are two description-based theories, Evans correctly points out that Kripke's arguments concern only a weak version of the first theory. This version stipulates that a speaker successfully refers when she intends to use a referring term to pick out a certain person or object, and where to have such an intention requires having in her possession an appropriate identifying description (or set or cluster of descriptions). In other words, it is taken that it is necessary that a speaker have this description.51 Evans agrees that this theory is false, but for different reasons from Kripke, as we will see. First, we revisit causal theories. The point of interest is the idea that being in an appropriate causal relation—established through a causal-historical chain fixed by a name-using practice—is sufficient for successful reference. Evans notes the apparent appeal of this view by giving an example with the same relevant characteristics as the following.52 An American tourist, in a fit of adventurousness, enters a small town public house in New Zealand. At the bar, he overhears a conversation. Someone says, "David Lange drives a sports car." Later that evening, having become acquainted with the conversing group, he breaks an awkward silence with "David Lange must like driving fast."53 He does this only with the intention of initiating a new conversation. He has no knowledge of David Lange's automotive desires. 5 0 Evans (1982), p. 319. 5 1 The strong version, in essence, replaces "necessary" with "sufficient." 5 2 Evans (1985), pp.6-7. 5 3 He says this even though he has no idea who David Lange is. 81 Now, it w i l l be agreed that the hearers w i l l understand what he says. That is, not only w i l l they know that this friendly American tourist is claiming that someone must like to drive fast, but they w i l l know to whom he is referring, i.e., by "David Lange." [They may even be impressed by this tourist's apparent knowledge of New Zealand political figures. David Lange, in fact, does (or did) like driving fast cars.] Given this knowledge, it seems that we must say that the tourist successfully refers with his use o f "David Lange" even though he only has the smallest bit of information about the person to which he refers. This is information that would, by itself, be entirely inadequate in identifying David Lange. What this account seems to show is that the tourist's causal contact with the name as it is used by members o f a name-using practice (i.e., we could trace a causal-historical chain back to David Lange) is sufficient for him to refer successfully. But surely there is something troubling about this claim. First, Evans makes two crucial assertions: (1) that contrary to causal theories, appropriate causal-historical connections between the speaker and the person or object named, are not sufficient for reference; and (2) that contrary to causal theories, appropriate causal-historical connections are not even necessary. It is these two claims, often taken together as they pertain to thought, that w i l l be the focus o f attention in the next four chapters. Second, Evans thinks that there is an explanation for this kind o f occurrence that allows us to reject the idea that an appropriate causal relation really is sufficient for successful reference.5 4 We get an early expression o f this idea in "The Causal Theory o f Names." 5 5 Evans says that, like the description Evans notes that Kripke was also suspicious of the sufficiency claim, which is perhaps, in part, reflected in an excerpt previously quoted where Kripke says "In general our reference depends not just on what we think ourselves, but on other people in the community, the history of how the name reached one, and things like that" (Kripke (1980), p. 301). The passage I have italicized presumably implies that Kripke thinks that what we think plays some part, which implies further that the causal relation is not sufficient. 5 5 In Evans (1985), pp. 1-24. Originally published in Aristotelain Society Supplement Al (1973): pp. 187-208. 82 theorist,56 the causal theorist takes insufficient notice of the informational input of the context of utterance. He says: The Causal Theory "ignores the importance of surrounding context, and regards the capacity to denote something as a magic trick which has somehow been passed on, and once passed on cannot be lost."57 In other words, take the utterance out of its original context, and the whole idea of successful reference seems implausible. For instance, consider a later time when the tourist is home and is recounting to his friends some of his adventures down-under. He says "David Lange drives a sports car." In this instance, it is extremely unlikely that he will successfully refer, yet he, qua speaker/referrer, is in no different (intentional) state than when he uttered the same sentence in the pub in New Zealand. What is more, he is in no different causal state. The causal chain of reference that linked his New Zealand use of "David Lange" with David Lange is the same as the one that links his current use of "David Lange." Evans notes that like description theorists, causal theorists take insufficient notice of the importance of the context in which the referential exchange takes place. Failing to take sufficient notice of the context creates a serious problem for causal theorists in trying to deal with certain kinds of ambiguity, for instance, where a name may, on different occasions, refer to different people, objects or places. This point is pushed by the following example given by Evans. Suppose, for example, on a T V quiz programme I am asked to name a capital city and I say 'Kingston is the capital of Jamaica'; I should want to say that I had said something strictly and literally true even though it turns out that the man from whom I had picked up this scrap of information was actually referring to Kingston upon Thames and making a racist observation.58 It seems that Evans might have been led to thinking that description theorists fail to take sufficient notice of context by following Kripke's comments. Searle (1983, p. 331) says that the idea, from Kripke, that description theorists hold that someone can refer simply by associating, in isolation, a name with certain identifying attributes, involves a gross distortion of any description-based theory. 5 7 Evans (1985), p. 8. Evans (1985), p. 9. 83 The reference of "Kingston" is indeterminate between (at least) Kingston, Jamaica and Kingston upon Thames. To what place the speaker is referring, then, is prima facie indeterminate. However, the context of utterance permits a disambiguation on the part of the audience in favour of Kingston, Jamaica. So, the context has given rise to a (strictly) true statement—what the audience understands the speaker to have said—even though the causal origin of the reference, as it is traced back through the speaker, and his acquaintance, is some place that, if it were the referent, would make the statement (strictly) false. Reference is made to the actual capital of Jamaica, but not by it being the case that the speaker's use of the term "Kingston" was causally connected to Kingston, Jamaica; there was no such causal relation. The point, as Evans says, is that causal theories have "unacceptable consequences"59 in that they are unable to account for what is said in cases where context plays a disambiguating role. There is a more general point, too. This observation (and others found in Evans's work), seem to show that not only is being in an appropriate causal relation not sufficient, it is not even necessary for determining what is said. Evans says, "a speaker may excogitate a name and use it to denote some item which bears it without any causal connection whatever with the use by others of that name."60 He suggests that street names in US cities contribute examples. If I understand the point, the following is an example. A companion and 1 are in a city for the first time. We arrive by train and exit the train station not knowing our exact location. We do know, however, that we are within walking distance of the hotel in which we are to stay, and that the hotel is in the exact centre of the city, which we determine, by convention, to be the intersection of Central Avenue and Main Street. Also, we know that street and avenue numbers go up from Central Avenue and Main Street, and that streets run North/South and avenues run East/West. We split up and walk in directions perpendicular to each other (for at least one block) to determine in which direction the numbers decrease. Returning, having each noted also whether we are on a street or 5 9 Evans (1985), p. 9. 6 0 Evans(1985), p. 10. 84 an avenue, say either, "Main Street is in that direction," and, "Central Avenue is in that direction." Our uses of "Main Street" and "Central Avenue," then, refer to the appropriate roads even though there is no causal connection between our uses and the name using practices of the locals. It should be noted that the Kingston, Jamaica example, as presented, is not given as an example of a piece of (successful) communication. In fact, we would be more inclined to think that the speaker has failed to communicate the thought intended. Presumably, naively, and indirectly, he intended to refer to Kingston upon Thames, so from this point of view, referential communication failed. The example suggests that a theory of referential communication must take account of the full communicative context, i.e., including the hearer's background knowledge and suppositions.61 However, it is quite obvious that from another point of view—the uttering of "Kingston" in that particular context—a reference was made. The fact that it diverges from the speaker's thoughts simply further highlights Evans's point that we must distinguish between referring and thinking. Moreover, it highlights that we must incorporate the distinction between speaker's reference and semantic reference in explanations of cases of the kind just given. Here is a more detailed explanation. The contestant utters "Kingston is the capital of Jamaica" in response to the game show host's question. Because of the speaker's wayward background knowledge, it would most probably be thought (by an objective observer) that the speaker's intended referent is Kingston upon Thames. However, because of the context in which the response is given (the game show host expects appropriate, i.e., non-racist, responses, along with the assumption that the contestant is trying to win, etc.), the host determines the referent of the speaker's use of "Kingston" to be the conventional referent of "Kingston," that is, Kingston, Jamaica. To flesh this example out a little, let us suppose that once the answer is given, a computer generated animation of the Earth appears on a screen, which rotates 6 1 In this case, what is especially relevant is the game show host's knowledge that Kingston is the capital of Jamaica, and his supposition that the speaker is trying to win the contest and so intends to give correct geographic information (and not make racist comments). 85 to the Americas, and zooms in on Kingston, Jamaica. The contestant would realize at least that this location was not southern England, and so realize his mistake (but probably keep quiet). In other words, he would realize that he had failed to communicate the thought which he believed to be the answer. However, because of the odd circumstances, a reference was made by what he said. In a sense, the speaker failed to refer, but the term "Kingston," his answer, did not. The point to take from the preceding paragraphs is that a reference was made in the absence of any causal connection between the use of the term on an occasion (by the speaker) and some name using practice in which a path is traceable back to the referent. Before we go on, it is worth pointing out that there is a further kind of case that shows that not only is tracing the causal origin of a referring practice potentially irrelevant, but that it may in fact get us into trouble, referentially speaking. To show this point, Evans introduces examples where there has occurred a 'change of denotation' associated with a name. The (somewhat) famous example is "Madagascar."62 Apparently, "Madagascar" originally picked out a piece of mainland Africa, somewhere on the central east coast. But upon visiting the region, and the island we now call Madagascar,63 Marco Polo began referring to the island by use of "Madagascar." He had misunderstood some reports by local sailors and so brought about the change of denotation. Now, if we were to follow the causal theory, our use of "Madagascar" should lead us back to this mainland region of East Africa. But that would be a mistake. If I went to an airline and asked to be flown to Madagascar, it would be entirely inappropriate of me to complain when I ended my journey on that large island off the east coast of mainland Africa. I am sure that no amount of pontificating about Marco Polo and change of denotation would secure me free passage to the mainland. 6 2 Evans (1985), p. 11. 6 3 There is some dispute over whether Marco Polo actually did venture this far south. But this doesn't matter for the point to hold. 86 We see that all of these considerations begin to give support to the underlying motive of taking the Linguistic U-Turn—that is, of explaining reference in terms of thought. In the above examples we see that not only must we take account of the intentions of the speaker in order to determine to what he is referring, but also we must take full account of the context in which the reference takes place. Abstracting away from its implications, this point is probably not contentious nowadays. However, with a full understanding of its implications, we see that to understand what happens in referential communication, we must understand of which person, or of what object or place, the relevant parties are thinking. So to explain how reference is effected, in this instance, we must first explain how it is that the relevant parties come to have the Particular Thoughts about the person, object or place named. In particular, to explain fully a (successful) communicative interaction, where reference is made to some particular person, place or thing, we must bring into the account the individuating knowledge to which the speaker and hearer have access. In making the above claim, though, we have to keep in mind that Evans does not give up on all kinds of causal theories, and fall back to advancing some description-based theory. He does think that description-based theories are false. Here is his basic point: There is something absurd in supposing that the intended referent of some perfectly ordinary use of a name by a speaker could be some item utterly isolated (causally) from the user's community and culture simply in virtue of the fact that it fits better than anything else the cluster of descriptions he associates with the name.64 Evans is careful to maintain, rightly so I think, Kripke's basic idea that some kind of causal relation between the speaker (and hearer) and the referent is crucial to explaining reference. But he thinks that Kripke, and others who have followed him, I suppose, have "mislocated the causal relation." He continues: "The important causal relation lies between that item's states and doings and the speaker's body of information—not between that item's being dubbed with a name and the speaker's contemporary use Evans (1985), p. 12. 87 of it."65 Having recourse to the former causal relation shows the crucial difference that distinguishes Evans from most who advance causal theories of reference. And it always must be remembered that when Evans indicates that a causal relation is important (perhaps necessary, but not sufficient) in explaining reference, it is this relation between the referent and the subject's information to which he is referring. The elaboration of this causal relation will take up most of the rest of this chapter, and cross over into chapters 5 and 6. The ideas will also draw heavily, for their support, on the theory of communication presented in Chapter 4. To a significant extent, the theories of reference, thought and communication stand or fall together. We will see an example of this interconnection in what follows. 3.4 Referential Communication: Some Preliminaries I said, above, that Evans has an alternative explanation of the tourist in the pub case, i.e., where a speaker apparently successfully refers even though he has no identifying knowledge of the person (or object) to which he is referring. And recall, this explanation is one that allows us to reject the idea that an appropriate causal relation between the speaker and an established name-using practice is sufficient for successful reference. The first thing to take on board is an understanding of how referential communication is effected in this situation. I do not wish to pre-empt Chapter 4 to any great extent, but we need some idea of what we mean by "referential communication" here. Broadly construed, referential communication involves a speaker (assuming linguistic communication) getting a hearer to entertain a thought.66 Now, underlying the causal theorists' claims about the tourist in the pub case, with respect to referential communication, is the idea that getting the M Evans (1985), p. 13. 6 6 For the sake of convenience, sometimes 1 might say something like 'communication involves getting someone to think something'. However, in saying this, I do not mean to suggest that to communicate a thought, a speaker must provide sufficient justification for the hearer to, say, believe what was said. Entertaining a thought (as I mean this expression here) does not require that the subject adopt some pro-attitude towards the thought content. 88 hearers to entertain a thought involves conveying information by use of a linguistic expression. That is, the linguistic expression contains all of the required information, and so long as the hearers are competent in the language spoken, they will grasp what is being conveyed. Further, if they are charitable, or see no reason not to accept what is expressed as true, they will think (or believe, etc.) what the speaker intended. With respect to the specific instance of referring, what follows from the causal theorist's picture of communication is that the speaker's use of "David Lange," say, is sufficient to pick out David Lange. This implication of the causal theory cannot be right. If we go back to the broad construal of 'referential communication', we can build up an alternative picture that does not lead us to this outcome.67 The idea is that communication involves more than simply conveying information by use of a linguistic expression—like a truck conveys logs from a forest to a mill. The picture of referential communication that underlies the causal theorists' account implies that the hearers are passive parties in the process of communication (and in particular, reference). However, there seems to be no reason for thinking that this picture is true. Hearers play an integral role by supplying information not strictly conveyed by the linguistic expression. Indeed, speakers will normally knowingly rely on the hearer's input. In the tourist in the pub case, the hearers understand the communicative intentions of the speaker. These are, roughly, that the speaker intends to refer to the person that they were just talking about (referring to), and say something about that person.68 Further, assuming that they are charitable, while probably not being aware of what they are doing, they will allow for at least some kind The basic ideas expressed here about communication are not particularly new. They stem from Paul Grice's ground-breaking work of the 1950s and 60s. 6 8 The situation is much more complicated than it might appear, but I do not want to go into these complications here. Not only are there other levels of speaker's intentions, but there are multiple levels of background assumptions possessed by both speaker and hearers. For instance, the speaker assumes that there really is this person David Lange. On the other side, the hearers assume that the speaker actually has the intention mentioned in the above text. Also, it is probably likely that the hearers do not take it that the tourist simply has the intention to convey what they already had said about David Lange driving a sports car. Probably, what they will primarily take from his utterance is that he wants the conversation to continue, etc.. 89 of successful communication. When the speaker uses the term "David Lange" they draw on previous instances of use of the term and associate the term with the relevant information, i.e., information that allows the (particular) thought of David Lange. They may or may not believe that the speaker (the tourist) knows who David Lange is. But even if they believe that he does not know who he is, they still allow for some kind of successful communication by supplying the relevant information. This explanation shows how it is that reference is effected eyen though the speaker does not know to whom he is referring. And, it explains why it can be erroneously thought that causal contact with the appropriate name-using practice is thought to be sufficient for successful reference. 3.5 Producers and Consumers We can leave referential communication for the moment and go further into this explanation by looking more closely at the details of the theory of reference that permits this explanation. To give this more detailed account, we must move to The Varieties of Reference where Evans more fully develops his ideas. Here, Evans introduces an important distinction. Evans divides those involved in a name-using practice into producers and consumers. Producers are the "core group of speakers who have been introduced to the practice via their acquaintance with [the person or object named]."69 The paradigm example of a producer is a user who has demonstrative contact with the person (or object) named, and understands the person to be so named. As Evans puts it, producers come to learn "a truth which they then express as 'This is NN' , where 'This' makes a demonstrative reference to" the person named.70 Further, producers have ongoing demonstrative encounters with the person, and as a community of users, maintain the name-using practice. Also, new producers can be generated by being (demonstratively) introduced, formally or informally, into the name-6 9 Evans (1985), p. 376. 7 0 Evans (1985), p. 376. 90 using practice. It will be helpful to modify our tourist in the pub example to best capture what it is to be a producer. Consider this new situation where one of the locals says the following, "Fred Dagg flew his helicopter to the pub last night." To make the details clear, Fred Dagg is well known locally, regularly frequents that very pub, and was introduced to the speaker on the first day of school when they were young boys.71 A distinctive feature of producers is that they have a certain recognitional capacity that is necessary for them to re-identify the person (or object) should they come across the person at a later time. This point needs some qualification and explanation. It is important to note that this recognitional capacity is merely necessary for re-identification. It is not sufficient. A simple modification to our example should show this. Let us suppose that Fred has an identical twin Trevor, and even that occasionally Trevor visits the pub, traveling in his helicopter. In this case, the developed recognitional capacity will not be sufficient to discriminate between Fred and Trevor. The only things that will allow discrimination are: (i) information from some appropriate, complete, causal chain of encounters that leads back to some genuine identification of one or the other of them; or (ii) possession of knowledge that allows for the connection of a previous identification with the current demonstrative contact. If the locals do not have either of these, or, say, the chain is incomplete (and the two really are physically indistinguishable), then at the next encounter they will have to ask "Are you Fred or Trevor?"72 Actually, the clearest example of a producer is the person who originally names the person, i.e., Fred's mother, who once (formally) said, "I hereby name this child 'Fred Dagg'," or something to that effect. Our example is sufficiently clear, though, I think, especially if we take it that the person who introduced the two boys was Fred's mother. 7 2 Street gamblers, playing the game where a small object is hidden under one of three identical cups, exploit both our faith in this recognitional capacity and our belief that we have a complete causal chain of encounters (taking each movement of position as an encounter). They move the cups around such that we believe we have traced the complete path of the cup with the object under it. But by sleight-of-hand they produce a break in our perception of the causal chain. In doing so, they give demonstrative (and, if you are unlucky, lucrative) proof that the recognitional capacity is not sufficient. This example does not quite capture the potential input of the recognitional capacity, but a slightly modified example would. In this case, take it that the cups have distinguishing features, and the street gambler can by sleight-of-hand successfully perform the trick of placing the object under a different cup without your (continued...) 9 1 Consumers, on the other hand, are not directly acquainted with the person or object, but are 'non-demonstratively' introduced into the name-using practice—for instance, by learning to associate the name with some description or set of descriptions. They are 'consumers' "since on the whole they are not able to inject new information into the practice," though as we will see they may 're-inject' information.^ We have, in our little story, a clear example of a consumer, i.e., the tourist who overhears the locals conversing, and comes to form the belief that there is some person who is a local, helicopter flying character, and that he is named "Fred Dagg." The tourist in the David Lange/tourist in the pub example also is a consumer. But I modified the example because, as it stands, it would not be obvious that the locals were not also consumers. An amendment to the producer/consumer distinction is worth noting, I think, before we proceed. It is not crucial, but its importance should become clear. This amendment is made by Gregory McCulloch.7 4 Upon acknowledging the distinction Evans makes, McCulloch distinguishes between active and passive consumers. The active ones are those that, although not directly acquainted with the person (or object) named, do have a kind of indirect demonstrative access to the person through, for example, doing research about what some temporally or spatially remote person did. Active consumers are, then, in a position to maintain the name-using practice. Passive consumers, it seems, merely associate a name with some information according to the way it is given to them. They are, by definition, "those who just parrot sentences containing the relevant names without really knowing what they are talking about."75 It seems to me that this distinction indicates a point where it might be thought that the so-called causal chain (...continued) noticing the switch. Again, the recognitional capacity would not be sufficient for accurate (re-)identification. 7 3 Evans (1985), p. 377. 7 4 McCulloch (1989), p. 28Iff. 7 5 McCulloch (1989), p. 283. 92 ends, or at least where there is a much weaker connection. That is, I do not think that passive consumers are connected causally, in the way Kripke or Evans describe, to the referent of a given name. Moreover, that they are not so causally connected seems to mean that as they use names, they are not using them as rigid designators because they are not, strictly speaking, using them as designators.76 Rather, they are merely associating a name (not a named object) with some information. In the Fred Dagg/tourist in the pub example, the tourist, certainly at the beginning at least, is a passive consumer. In the David Lange/tourist in the pub example, if the locals are consumers, they are probably active consumers.77 Hilary Putnam seems to be drawing attention to roughly the same point as Evans does with his distinction between producers and consumers with his idea of the division of linguistic labour.78 In "The Meaning of 'Meaning'," Putnam argues that the reference of natural kind terms is fixed, ultimately, by the essences of that substance. As it turns out, it is only ever experts that can distinguish between kinds of substance based upon the essence of that substance. So, we rely on experts to determine the extension of certain natural kind terms. Whether this account of the referential function of all natural kind terms is accurate, I leave aside. But it seems that in certain cases, at least (Putnam sites "elm" and "aluminum"), we rely on experts to fix the reference. Putnam writes: "We could hardly use such words as 'elm' and 'aluminum' if no one possessed a way of recognizing elm trees and aluminum metal; but not everyone to whom the distinction is important has to be able to make the distinction."79 It seems to me that the experts are parallel to producers, and the ones to whom the distinction is important but who cannot make the appropriate distinctions are consumers. Notice that I am not saying that they are not rigid designators. Only further details would set this for sure. Putnam (1975), p. 227ff.. Putnam (1975), p. 227. 93 However, although the above account gets us started, it is of course true that most instances of using names are not at baptisms. It does not show us what sort of thought about what is named would normally be ascribed to name users in most instances. To show what sort of thought this is we need to take on board the amendments made by Gareth Evans to this 'new theory of reference'.80 3.6 Reference and Identifying Knowledge At the core of Evans's criticism and subsequent modification of Kripke's picture is the idea that the description theorists get something crucially right. What they get right requires significant qualification (to take account of the problems Kripke's arguments raise) but, in general, it comes from the description theorists' idea that a speaker possesses some uniquely identifying knowledge associated with a name, where this knowledge is appropriately connected to the object named. The full details of the qualifications will unfold with the development of Evans's modifications of Kripke's picture. The crucial point to note as we proceed, however, is that for Evans, the uniquely identifying knowledge associated with use of a referring term is not specified by a description, or cluster of descriptions. Rather, it is information-based. In general, Evans's claim is that the causal theorists fail to take account of the context in which the reference is made. Many who would probably label themselves causal theorists do now explicitly note the need to take full account of the informational content supplied by the surrounding context. For instance, any theorist who suggests pragmatic input into an act of referring (and those who suggest we can have genuine beliefs de re, i.e., ones not explicable in terms of beliefs de dicto) just is noting the input of contextual information. However, I think that such theorists have moved such a significant distance from standard causal theories that we should be cautious about applying the label. Recall, an essential claim of causal theories is that the appropriate causal-historical connection is sufficient to establish reference. Obviously, in claiming that context is important, the sufficiency of causal-historical connections is gone. 8 0 Evans (1982), esp. chapter 11. 9 4 However, Evans also accepts the causal theorists' arguments to the effect that reference is not to be explained in terms of associations between proper names and uniquely identifying descriptions. It will take some time to build up, but Evans's theory marries the idea that reference is direct (even causal, in some qualified sense) with the idea that reference requires possession and transmission of uniquely identifying knowledge grounded in information. Evans notes a now reasonably well known problem with standard causal theories.81 Kripke claimed that a speaker could refer without knowing very much, or indeed anything, about the person or object to whom or to which she was referring. In other words, a speaker can refer without having identifying knowledge.82 For example, someone might have reason to say "Feynman must be a bright guy," where all this person knows about Richard Feynman is that he is a famous physicist. Kripke says that in such a case, the speaker successfully refers to Richard Feynman. In fact, Kripke claimed that a speaker could possess false information and still refer. For example, Kripke envisages a counterfactual situation where Godel did not prove the incompleteness of arithmetic, but stole the idea from his friend Schmidt, and published it as his own. Furthermore, a particular speaker does not know about this fact. Consider, then, a situation in which this speaker has reason to say "Godel must have been a bright guy." What do we say? Do we think that the speaker is really referring to Schmidt, and so agree with her? Kripke would say "No." Given the right causal4iistorical connections, it must be the case that the speaker is referring to Godel, even though the speaker has this false information. Our response should be "No. He was a cheat and a liar." In The Varieties of Reference, Evans presents a more sophisticated argument against causal theories than that which appears in "The Causal Theories of Names." It centres around the idea that causal theorists take it that the relevant causal relation in reference is modeled by the relation between a 81 The problem might be reasonably well known, but Evans's discussion of it is not. 8 2 See Kripke (1980), p. 8Iff. 95 photograph and what is photographed. Evans calls this view the "Photograph model."83 I will look at these arguments in Chapter 5 to 7. These are given, primarily, as arguments against a theory of thought that seems to have grown out of considerations presented by Kripke in his picture of direct reference, and since Kripke never spoke directly about thoughts, it is more appropriate to consider them as they are applied to this theory of thought. In the meantime, we can proceed with our general inquiry. Identifying knowledge grounded in information is that which allows someone who possesses it to pick out some particular object, as that particular object. Now, as we have seen, having this identifying knowledge does not simply involve possession of a recognitional capacity. The identical twin (Fred and Trevor) case shows this point. What else is needed is that there be an understanding of an appropriate causal relation to the object. That is, not only must we be able to recognize the person (or object), say through the visual recognition system (however that works), but we must possess the knowledge that the information comes from the right source, i.e., the particular object to which we are referring. The source might be the person (or object) that the speaker intends to refer to by use of the name. But, as we have seen with certain kinds of examples, it may not always work out this way. Context of utterance has to be taken into consideration when explaining what has happened in referential communication. The Kingston upon Thames/Kingston, Jamaica example shows this point. The speaker may actually intend to refer to Kingston upon Thames, but because of the context (the request for a genuine capital city, the appropriate peoples' background knowledge that Kingston, Jamaica is the capital of Jamaica, that the contestant is trying to answer correctly, and probably some general understanding that in these situations one ought not make racist comments84) Kingston, Jamaica is picked out as the place to which the contestant is referring. However, someone (or some people) appropriately placed must possess identifying knowledge. In this 8 3 Evans (1982), p. 76. 8 4 This last aspect of the context is important for we can imagine a supposed alternative situation where the game show is run by some right-wing extremist group, and precisely the kind of answer they are looking for is the one that would make a racist comment, i.e., "Kingston upon Thames." 96 case, at least with respect to the contest, whoever judges the answers to the questions must have identifying knowledge. Any audience member who is to understand what is going on will also need to have identifying knowledge. Only producers (and appropriately informed active consumers, if this further distinction is legitimate) possess identifying knowledge. A speaker may give the appearance of successfully referring without possessing identifying knowledge. However, this speaker is not a producer (or an appropriately informed active consumer). The speaker may not even be a consumer. As the Kingston upon Thames/Kingston, Jamaica example shows, Kingston, Jamaica is picked out. But the speaker is merely a consumer (perhaps now a producer) with respect to the "Kingston upon Thames" name-using practice, and is (probably) neither a producer nor a consumer with respect to the "Kingston, Jamaica" name-using practice.85 Traditionally, it has been thought that identifying knowledge must amount to possession of descriptions. Kripke, Evans, and others, think that this view of identifying knowledge cannot be right. Evans was probably partly motivated by the early views of the causal theorists, though he had a different picture of how our knowledge of the world is retrieved—or in general, how the mind-language-world complex is structured. However, it is probably worth pointing out that the characterizations of description-based theories to which Kripke, Evans, etc, responded may have been inaccurate in a very important way, at least with respect to what Evans says. In "Proper Names and Intentionality," Searle argues that Kripke's (and others') characterization is inappropriately attributed to contemporary description-based theorists. In this work, we see Searle strip away details to get to what he thinks are the most basic claims of description-based theorists. The principal idea is this: 8 5 After answering the question, and viewing the computer animation in which viewers are shown Kingston, Jamaica, he is likely to grasp the right kind of information that would allow him to be a producer. 97 The descriptivist is committed to the view that in order to account for how a proper name refers to an object we need to show how the object satisfies or fits the "descriptive" Intentional content that is associated with the name in the minds of the speakers; some of this Intentionality will normally be expressed or at least be expressible in words.86 There are two things that need to be noticed in this passage. First is the quotation marks around "descriptive." Second is the use of "some" in "some of this Intentionality." These allude to the idea, expressed earlier by Searle, that in saying that someone possesses an "identifying description" a description theorist might only mean that she has "the ability to recognize the object."87 In other words, what constitutes the recognitional capacity may involve something other than descriptions expressible as sentences of a language. This explication will strike many as an odd thing for a description theorist to claim, and think that if this is really what the description theorists claim, then they are inappropriately named.88 I do not want to go into this naming aspect of the dispute, though we might wonder at how it is that so many people could be so wrong about the view. What is more noteworthy is that what Searle is claiming here, is very close to what Evans claims, and so with respect to this idea of the recognitional capacity, there is no significant disagreement between Searle and Evans, which should become clear as we proceed. In fact, what Searle says helps to characterize what Evans had in mind. Also, my primary purpose in this chapter is not to discuss critically the debate between the description theorists and the causal theorists. I have introduced their ideas simply to provide an account of the framework used by Evans to build his theory. 8 0 Searle (1983), p. 331. 8 7 Searle (1983), p. 331. 8 8 Some might be inclined to think that if this really is what Searle most recently holds, then it constitutes a change in view. They might say that it was from Searle that we got the concept of a 'cluster of descriptions'—though perhaps this concept is flexible enough to allow for this new idea. However, I am sure that many think that it was not by accident, or malicious intent, that descriptivists got their name. 9 8 C H A P T E R 4 Referential Communication My arguments for the acceptance of the theories of the previous chapter (the theory of reference) and the ones that follow (the theory of thought) are connected to the following theory of referential communication. The overall claim is that these theories of reference and thought can be properly understood only if straddled by a theory of referential communication. By 'referential communication' I mean linguistic communication where, in the exchange, reference is made to some particular object, property or relation in the world. Referential communication is to be distinguished from other instances where speakers and hearers still communicate, but where no reference, or no particular reference, to things in the world is made. This distinction is a little vague, but I know of no title under which all other modes of linguistic communication would fit. Within the narrow bounds of considering only communication involving assertoric expressions, referential communication is contrasted with, perhaps we would say, descriptive or attributive communication. But in communication, we use many other kinds of expressions to communicate many kinds of things other than assertions (eg., attitudes, emotions, etc.), so these titles will not cover all modes of linguistic communication. Straightaway, it might be thought that this way of characterizing this kind, or mode, of communication is somewhat circular. That is, in making the distinction between referential and other kinds of communication, e.g., descriptive or attributive communication, I presuppose that reference is, at least sometimes, not descriptive. I acknowledge this presupposition, but see no reason to be troubled by the apparent circularity. For one thing, the presupposition is well grounded in contemporary work. For another, it is not the case that I simply presuppose this distinction in isolation. It is intimately connected 9 9 with the arguments of most of the rest of the dissertation. As I have said, it seems to me that these theories (of reference, thought and communication) stand or fall together. 4.1 The Nature of Communication Evans notes an important point about communication suggested by a metaphor put forward by Hilary Putnam, who says: There are two sorts of tools in the world: there are tools like a hammer or a screwdriver which can be used by one person; and there are tools like a steamship which require the cooperative activity of a number of persons to use. Words have been thought of too much on the model of the first sort of tool.1 Evans thinks that this analogy does not quite capture the right idea. He has in mind that since the traditional position has always been to regard words as instruments for communication, and hence as tools which two people 'use' . . . a better contrast would be one between a steamship or a factory on the one hand and a double handed saw on the other.2 The general point being made is that an adequate explanation of linguistic communication, and in particular, here, referential communication, must take account of the collective, cooperative arrangements of the language (or in particular, referring term) users, i.e., including hearers as well as speakers.3 Here is one way of putting it. It is widely agreed that there are two, general kinds of reference; speaker's reference and semantic reference. Ordinarily it is taken that in both cases reference is to be explained in terms of a relation between the referrer (with respect to speaker's reference) or the referring teim (with respect to semantic reference), and the referent. The crux of the model of referential communication to be presented here is that this explanation misses the important contribution made by the audience, and the 1 Putnam (1975), p. 229. It is from this idea that Putnam gets to his notion of the 'division of linguistic labour'. 2 Evans (1985), p. 375. 3 I'll follow the convention of referring to language users as speakers and hearers even though language users are also writers and readers. 100 linguistic community in general. That is, given this model, reference is not to be explained in terms ofthe above two-way relation, but rather, a three way relation between the use of the referring term (by the speaker), the referent, and the contribution made by the audience.4 This approach provides the foundation for what I take to be the correct model of linguistic, referential communication. The aim of the rest of this present chapter is to set out the characteristics of referential communication that conforms to this model, i.e., the model of a steamship. There is what appears to be a quite common way of thinking about linguistic communication that does not place any importance on the cooperative arrangements of the language users. This way of thinking probably comes from an intuitive understanding of what takes place when we communicate with each other by using linguistic expressions. The idea, roughly, is that in communicating, what a speaker wants to do is get her audience to entertain a thought, say, some particular thought that the speaker herself is thinking. Since it is taken that what is in the speaker's mind is wholly private, what she has to do in order to achieve this goal is put the thought in some public form so that it can be 'transmitted' to the audience. The implication is that this public form (a sentence in a language), captures just what is held privately (the thought). Given this implication, and given appropriate language competence, the audience comes to have the thought the speaker communicated simply by retrieving it from the public form. The above account, then, seems to be the intuitive understanding of linguistic communication. What of philosophical views? A quick survey of the literature in contemporary philosophy of language and mind shows us that outside of work that directly addresses issues in Pragmatics, there is little in the way of explicit discussion of communication.5 It is difficult, then, to have any definite ideas about what Analytic Philosophers regard 4 Perhaps this is really a four-way relation. That is, we ought to split 'use of the referring term' into 'speaker' and 'referring term'. 5 Perhaps it has been thought to be more appropriately the subject matter of Linguistics. 101 as the correct theory of linguistic communication. However, from noting the underlying assumptions, especially the Assumption of Parallelism, it seems that those well entrenched in Linguistic Turn philosophy hold something like the intuitive model outlined above (that assertoric sentences encode information, which is taken to be their semantic function). Given acceptance of the intuitive model, a theory of linguistic communication can be built up in conjunction with this model by holding that in communication, a sentence conveys (literally) information encoded in the sentence uttered, i.e., the information passes from the speaker to the hearer, transported by the sentence. It is, of course, problematic to characterize such things in these highly metaphorical terms, but it seems to capture the underlying intuition. This intuition really seems to be that sentences, or rather, utterances, convey the contents of a speaker's thought roughly in the manner in which a logging truck conveys logs from the forest to the wood mill. Obviously, there is an important difference; a sentence encodes, or symbolizes, the content of the speaker's thought, whereas a logging truck conveys the logs as logs, so the theory outlined here is somewhat more sophisticated than the way in which communication is explained by the intuitive model given in the previous paragraph. However, noting this process of symbolization is irrelevant to the point being made by employing the metaphor of 'conveying'. The fact that there is this symbolization is important in other aspects of the inquiry, though.6 There is, however, an alternative way in which we can think about referential communication, which comes from the steamship model. It can be acknowledged that sentences play an important, perhaps crucial, role in communication. And, that the conveying, or logging truck, model may well accurately reflect what is going on in at least some types of linguistic exchanges. However, the claim is that linguistic, referential communication is not accurately characterized by a view based upon the intuitive model. The essential point of deviation comes from the idea that sentences containing referring terms do 6 If we wanted to incorporate it, we could describe communication as being like the conveying of spoken words via digital signals over telephone lines. Paul Bartha suggested this metaphor. 102 not encode all the informational content of thoughts about things in the world. Given this claim, it becomes inappropriate to say that such sentences convey information in referential communication to the extent that nothing else is going on in referential communication except for the passing of the linguistic information contained in sentences. In other words, the alternative view has it that there is more to referential communication than the conveying of the information that is retrieved only through a competent language user's understanding of the sentence; that is just what you get from understanding the sentence through employing only language competence. Of course, a sentence plays an important role in getting the hearer to have the thought that the speaker intends the hearer to have (hopefully), and indeed, the sentence does convey some information. But the standard picture of communication ignores, among other things, the hearer's role in communication. Hearers are almost never passive recipients of information in the way in which photographic film is a passive recipient of the light that produces the content of the photographic image. In the communication of thoughts about things in the world, hearers must add informational content, quite often, significant amounts of it. 4.2 The Fregean Model of Communication What supports this alternative picture of referential communication? The following is our starting point. Here is how Evans characterizes what he calls the Fregean Model of referential communication. The speaker, S, utters the sentence 'a is F, having in mind the thought whose content is determined by the senses he attaches to the expression 'a' and % is F. This thought will be about the referent of the term V , but he will be thinking of it in the particular way which constitutes the sense he attaches to the name. The hearer, A, will, in understanding the sentence, also entertain a thought, understood to have the same truth value as 5"s utterance. The content of A's thought will likewise be determined by the senses he attaches to the constituent expressions. If we assume, with Frege, that it is an objective property of expressions of the language which S and A speak that they have a definite sense, and further that S and A are competent speakers of their language, then we shall 103 be assured that A will have the very same thought that S was expressing, or making as if he was expressing.7 Although there are important aspects to this model, it is only our starting point because it seems that it is too restrictive. That is, as Evans (and Dummett, as notes Evans) observes, under this model it seems that no linguistic exchange could ever count as an instance of genuine, referential communication. The model seems to express an ideal that is never actualized in linguistic exchanges. As is quite well known, the problem stems from the idea that there seem not to be 'definite senses' associated with referring terms (or rather, expressions containing referring terms) in ordinary language use. In other words, it seems that there can be successful communication—the speaker and hearer come to have the same thought—even in situations where it is quite apparent that the speaker and hearer do not come to think about the object of concern in exactly the same way. As Dummett points out, Frege's solution to this problem began with the assumption that there was something defective in ordinary language. It follows, then, that to get a clear idea of the processes, language needs to be "purified so as to correspond to the ideal."8 But Frege's solution hardly seems to be the right way to approach the problem. The fact remains that we do manage to communicate, so we have to come up with a less restrictive model. To develop such a model, we need a clearer idea of what is involved in genuine, linguistic communication. 4.3 Grice We get a significant start in understanding what is involved in genuine, linguistic communication by taking on board certain insights put forward by Paul Grice in his seminal paper, "Meaning."9 In this paper, Grice had as one objective to establish an idea of speaker's meaning, given in terms of speaker's 7 Evans (1982), p. 20-1. 8 Dummett (1991), p. 106. 9 Grice (1989), pp. 213-233. 104 intentions, with the supposition that semantic meaning could be explained in terms of speaker's meaning. He does not use the terms 'speaker's meaning' and 'semantic meaning', but it is largely due to Grice's work in "Meaning" that we now have these terms. Connected with establishing an idea of speaker's meaning is the objective of setting out the criteria for what is to count as genuine, intentional communication. This second objective amounts to distinguishing between instances of an incidental transfer of information and communication proper. For Grice, this distinction is displayed in an ambiguity we notice with the verb "mean." Consider the following two sentences: (1) Those clouds mean rain and (2) Those three rings on the bell mean that the bus is full. (2) asserts an intention to communicate something, namely that the bus is full, by ringing a bell. 1 0 We understand that 'three rings on the bell' means 'the bus is full.' That is, there is an intention to communicate (communication proper) 'the bus is full' by the intentional act of ringing the bell three times. In (1), however, there is no assertion of an intention to communicate. All we have is an incidental transfer of information. That is, someone notices the clouds, and infers that they mean rain. It does not assert that the clouds intend to communicate that it is going to rain by their very presence. Ultimately, Grice drew a distinction between what he labeled natural and nonnatural senses of "means" (or cognates). Of natural meaning, Grice said that "x meant that p and x means that p entails p." But of nonnatural meaning, "x means that p and x meant that p do not entail p."u With natural meaning, it is not possible that the entailment relation fails to hold. We cannot say, "Those spots meant 1 0 For those unfamiliar with this example, the background context is a double decker bus in London, say, that has a driver and a conductor. The conductor signals to the driver that the bus is full by three rings on a bell. 1 1 Grice (1989), pp. 213-14. 105 [natural] measles, but he hadn't got measles."12 However, with nonnatural meaning (meannn) it is possible—a mistake can be made. That is, we can say that those three rings on the bell mean that the bus is full, "But it isn't in fact full—the conductor made a mistake."13 In this way, then, we can see that nonnatural meaning is to be associated with the intentional. The crux, here, is that with the intentional there is no straightforward causal relation claimed which leads to strict entailment, as is the case with attributions of natural meaning. Grice is aware that some uses of "mean" may not fit easily into the categories of either natural or nonnatural meaning. However, in most cases, what sense of "mean" is meant should be clear enough. Also, he is aware that what underlies the distinction between the senses of "mean" may lead some to distinguish between natural and conventional signs. But this latter distinction is problematic.14 The question that now needs an answer is 'What is meaning,,,,?^ or, 'What is it for an utterance to meannn something?'). Answering this question has turned out to be far more involved than indicated by Grice's answer in "Meaning." Grice does repeat a few times the process of giving a tentative answer, then presenting a counterexample, followed by modifying the answer accordingly. However, there has been a sequence of papers that show that Grice did not go far enough here. I do not wish to get enmeshed in the details of the discussions of this sequence of papers (it gets very complicated), so I will give just part of Grice's answer and then one which is widely regarded as being a good approximation to the right answer, one given by P. F. Strawson. Grice begins the attempt at understanding what it is for an utterance to meannn something by first considering the view that the utterance must have a tendency to produce in an audience some attitude, 1 2 Grice (1989), p. 213. 1 3 Grice (1989), p. 214. 1 4 Grice (1989), p. 215. 106 where this is dependent upon an elaborate process of conditioning in the use of signs in communication. This answer is shown to be inadequate, principally because it fails to take account of the intentions of the speaker that underlie the utterance. Skipping a couple of steps, we get, then, a speaker, S, meant something by an utterance, x, if and only if S intended to induce a belief in the audience—where to say what the belief was would be to say what S meantnn—and that S intended the audience to recognize the (communicative) intention behind the utterance. This answer comes close to being right, but is inadequate because it fails to take account of further speaker's intentions, as well as the requirements of the audience to recognize these intentions. Grice is aware of some of these necessary, further modifications, but I will skip straight to an answer to the original question that incorporates Strawson's modifications. From Strawson we get: S nonnaturally means something by an utterance x if S intends (i,) to produce by uttering x a certain response (r) in an audience A and intends (i2) that A shall recognize <S"s intention (i,) and intends (i3) that this recognition on the part of A of 5"s intention (i,) shall function as^'s reason, or part of his reason, for his response r. [Further] . . . that S should have the further intention (i4) that A should recognize his intention (i2).1 5 As it turns out, it gets even more complicated. The problem in giving a full and accurate account of communication in the required sense comes down to having to find the difference between getting someone to think something and deliberately and openly letting someone know something. The point seems to be that there are all sorts of ways of getting someone to think something that do not count as communication proper. Perhaps the best way to put it is that communication (in this strict sense) requires that there be mutual knowledge of the intentions behind the utterance. This way of putting it might seem a little vague, but once you follow through all of the counterexamples and the modifications, it becomes reasonably clear that this idea of mutual knowledge of communicative intentions does a good job of capturing what Grice had in mind. 1 5 Strawson (1964), pp. 446-7. 107 The connection between these points, with their origins in Grice's work, and the general theory of communication to be advanced in this chapter, is nicely summarized by Dan Sperber and Deirdre Wilson. In the opening chapter of Relevance: Communication and Cognition, they set out the foundations for an alternative model of linguistic communication—i.e., one that is alternative to what might be called an intuitive understanding of communication. The intuitive model is much like the one I characterize above where it is thought that sentences convey all information from speaker to hearer (like a logging truck conveys logs). The alternative is grounded upon Gricean (and Strawsonian) considerations. They say: The description of communication in terms of intentions and inferences is, in a way. commonsensical. We are all speakers and hearers. As speakers, we intend our hearers to recognise our intention to inform them of some state of affairs. As hearers, we try to recognise what it is that the speaker intends to inform us of. Hearers are interested in the meaning of the sentence only insofar as it provides evidence about what the speaker means. Communication is successful not when hearers recognise the linguistic meaning of the utterance, but when they infer the speaker's 'meaning' from it. This is shown by the following easily verifiable observation: when hearers realise that the speaker has misused a word or made a slip of the tongue, they generally discount the wrong meaning. The meaning they discount, however, need not be ill-formed or undecodable; it is 'wrong' only in that it provides misleading evidence about the speaker's intentions.16 The ability of this model of communication to handle cases like the kind of thing described at the end of the quotation, and many others as well, is what many take to be the advantage of Pragmatic explanations of language use. How does this modification help out the Fregean model? Recall, the problem is that if it is taken that thoughts are private, i.e., wholly subjective, it seems that it would be impossible for two people to share the same thought. Communication would always be doomed to fail. Part of the problem for the Fregean model is that sentences of a natural language seem not to have a definite sense. In communication, given the unmodified Fregean position, it is not enough for the speaker and hearer simply to be thinking of the same object. They must be thinking of the object in the same way. In other words, 1 6 Sperber and Wilson (1995), p. 23. 108 for communication, the speaker and hearer must attach the same sense to the sentence containing the referring term. If there are no definite senses attachable to sentences, then it seems puzzling how communication can be achieved. The Gricean picture seems to allow for a way out of this problem. What follows is just a first shot at explaining the advantage of the Gricean picture, but it seems that by acknowledging the cooperative arrangements, and especially the contribution made by the hearer, we can see how communication can be achieved. The logging truck model has the problem just noted (regarding sentences not having definite sense) because total reliance is placed on the sentence expressed to 'transmit' the thought. But in the Gricean picture, the hearer draws on informational content outside the content of the sentence, which allows at least for the possibility of a speaker and a hearer coming to think about the same object in the same way, by the hearer coming to recognize the informational and communicative intentions of the speaker. Of course, communication may fail. We are all well aware the we do sometimes fail to communicate. However, it is also apparent that frequently we are successful, especially when speaker and hearer share the same context. These ideas need to be fleshed out a little, and we can do this by looking at what Evans says about referential communication. One thing before we proceed, though. If the Gricean picture is to be acceptable, it seems that in thinking about successful communication, we must not have too restrictive a notion of 'shared sense'. That is, we must allow a little flexibility in determining whether speaker and hearer are thinking about the object in the same way. We must not cut up thoughts too finely. That this is permissible is supported by discussions elsewhere in the dissertation. It is something we do well to keep in mind in thinking about communication, though. 109 4.4 Evans on Communication I have said above that within the Fregean model, successful referential communication requires not just that the hearer and the speaker think of the same object, but that they think about it in the same way, which was expressed in terms of 'shared sense'. The first thing that we must note in building up Evans's picture of referential communication is that Evans's understanding of'sense'—the one which I follow—-deviates from what is perhaps the common understanding. I do not wish to get into interpretational disputes, here, so I think it is sufficient just to note what Evans has in mind. For Evans, one way in which a subject and a hearer think about an object in the same way is if they employ the same mode of identification of the object—for instance, given appropriate context, if they think of the object demonstratively. Two people in the same physical situation are likely to be thinking of an object, in that situation, in the same way just because they are in (roughly) the same demonstrative relationship with the object. What this amounts to from the perspective of communication, in this simple demonstrative case, is that the speaker and hearer draw on the same information in identifying the object. Referential communication, then, requires that the hearer draw on the same information in identifying the object as does the speaker. For instance, the speaker thinks and expresses something, e.g., 'That tree is big'. By this expression, he intends the hearer to think 'That (some particular local tree) is big'. The hearer understands it only when she associates that expression with the information drawn from her demonstrative identification of that particular tree. And she does this through recognition of the speaker's referential, communicative intentions. The second thing to note is something already mentioned, i.e., not carving up thoughts too finely. I will say more about this later (section 6.1), but to characterize Evans's picture accurately we must note that what is said in the above paragraph is in need of qualification. Evans seems to adhere to a conception of communication such that two people communicate even when there is some difference in the way in which they identify the object. An example he gives is two people employing different recognitional capacities in identifying a person "where one recognizes the person by his face, and the other depends 110 rather more upon the voice."17 Here, it seems, speaker and hearer are not going to have exactly the same identifying information. The crucial point, though, is that they have the knowledge that the information comes from the same object, which allows for successful communication. We get a clearer idea of the motivation for this stance by going back to what seems to be the grounding for Evans's position. Evans quotes this passage from Strawson's "Identifying Reference and Truth-Values." Very often a speaker knows or assumes that a thing of which he has [identifying] knowledge is also a thing of which his audience has [identifying] knowledge. Knowing or assuming this, he may wish to state some particular fact regarding such a thing, for example, that it is thus-and-so; and he will then normally include in this utterance an expression which he regards as adequate, in the circumstances of the utterance, to indicate to the audience which thing it is, in all the things in the scope of the audience's identifying knowledge, that he is declaring to be thus-and-so . . . When an expression . . . is used in this way, I shall say that it is used to invoke identifying knowledge.18 This passage is given to focus attention on a "tremendously important feature of referring expression,"19 and Evans goes on to discuss some important aspects of referring terms that stem from these considerations. However, the passage serves equally well as a way of setting out the foundations for Evans's picture of referential communication. Two points need to be highlighted: (i) the assumption on the part of the speaker of the requirement of the cooperation of the audience; and (ii) invocation of identifying knowledge or information. When a speaker's communicative intentions are referential—that is, when a speaker intends to pick out some particular thing and say something about it, and intends that the hearer recognize this intention, (etc.)—the use of the referring term is information-invoking. As Evans notes, "there is no 1 7 Evans (1982), p. 334. 1 8 Evans (1982), p. 306. From P. F. Strawson, Logico-Linguistic Papers, pp. 77-8. Evans adds the two instances of "identifying" inside the square brackets. 1 9 Evans (1982), p. 306. I l l infallible linguistic guide to when the understanding of a remark requires the invocation of information."20 So, it is probably the case that we as hearers learn through experience when utterances of certain kinds are intended to be information-invoking. With some referring terms, the determination of the intentions of the referring term will be quick and easy because they are just about always intended to be information-invoking—for instance, pure demonstratives like "this" and "that." With other terms, however, the determination may not be so easy. The context of utterance most likely plays a crucial role in these kinds of cases in the determination of the intention for the referring term to be taken as information-invoking. Consider, for instance, referential and attributive uses of definite descriptions. I do not wish to get into a discussion about the semantic status of definite descriptions. But assuming that Kripke and followers are right (i.e., the referential/attributive distinction is not semantically significant, rather it is to be accounted for through Pragmatic considerations),21 referential definite descriptions are intended to be information-invoking, whereas attributive ones are not. The point is that as the discussions of these two uses of definite descriptions show, only a detailed awareness of the context will facilitate the hearer's understanding of which use is intended.22 The idea of there being information-invoking intentions behind the use of referring terms is important. When a hearer comes to recognize these intentions, she is in a position to understand properly what the speaker is saying, assuming she can retrieve, or is already in possession of, the right information. She then comes to have two bits of knowledge that are crucial for communication: (i) the knowledge associated with a correct interpretation of the content of the utterance; and (ii) with respect to z u Evans (1982), p. 309. 2 1 See Kripke (1977). 2 2 Evans discusses the information-invoking uses of definite descriptions in Evans (1982), section 9.3, pp. 320ff. 112 communication, the knowledge to the effect that, as Evans sometimes puts it, 'Ah! That's what the speaker is saying'. That this knowledge is crucial is emphasized by Evans. He writes: It is a fundamental, though insufficiently recognized, point that communication is essentially a mode of the transmission of knowledge . . . . [If a] speaker S has knowledge of x to the effect that it is F, and in consequence utters a sentence in which he refers to x, and says of it that it is F, and if his audience^ hears and understands the utterance, and accepts it as true (and there are no defeating conditions), then A himself thereby comes to know ofx that it is FP Communication becomes, then, "a link in a chain of knowledge-transmission."24 Talk of "transmission" strikes me as an unhappy choice of words, but perhaps only because of the connotations associated with the metaphorical use. However, perhaps Evans is slipping back into thinking in terms of the intuitive model, which would be a mistake. In any case, leaving aside some details,25 we have now a fairly clear idea of Evans's understanding of referential communication. One thing remains for this section. How does this picture fit with the Fregean model of communication given at the beginning? There have been some hints of modifications of the Fregean model along the way, but some more explicit remarks will make it clear how Evans sees the connection. Recall, the Fregean model states that not only must speaker and hearer think of the same object in communicating, but that they must think about the object in the same way. And for Frege, thinking about an object in the same way amounts to associating the same sense with the relevant expression. However, Evans argues that understanding in referential communication requires simply that he think of it in the "right way."26 He says: We recognize the primacy of the referent by recognizing that the hearer always confronts just one question, 'Which object does the speaker mean?'—not two questions, 'Which 2 3 Evans (1982), p. 310-11. 2 4 Evans (1982), p. 311. 2 5 For instance, the many ways in which the speaker gets the hearer to think of the right person or thing. 2 6 Evans (1982), p. 315. 113 object does the speaker mean?' and 'How am I intended to think of it?' The second question is answered in passing; for if he understands the remark, he will know which object is meant; and in the normal course of events . . . he will know which object is meant only if he thinks of it in the particular way intended by the speaker.27 The point is that so long as the hearer is drawn into thinking of the intended object by understanding the speaker's remark, she will be thinking of it in the way intended. A good part of likely disputes about the Fregean model of communication will involve interpretations of the Fregean notion of sense, and how strict we are to be with this notion. For instance, consider the potential for disputes if we follow this through the approach where a thought is taken to be the sense of the sentence that is taken to express it, which in turn is made up out of the senses of the component words. Given this approach, the implication seems to be that a different thought is expressed by "I am in pain" (said by me), and "You are in pain" (said by someone else who is referring to me). But if I intend to communicate that I am in pain by "I am in pain," it is not at all clear that the hearer fails to understand my thought if he understands it by grasping the thought expressed by "You are in pain" (where 'you' refers to me). Put another way, I think that we would not want to say that I have failed to communicate by saying what I did, and the hearer understanding as he did. These issues are discussed elsewhere in the dissertation (especially 6.1), so I will say no more here. All that I wanted to do is characterize the notion of referential communication with which I am working. All that remains is to connect the account of referential communication with discussions in other relevant chapters. 4.5 Connections As I have said, in philosophy of language (and mind) there is a common, but yet, I think, mistaken understanding of the process of referential communication. To recapitulate, associated with this Evans (1982), pp, 315-6. 114 understanding is something like the following picture: When people speak to one another they communicate by use of linguistic expressions, which are the external media through which they transmit thoughts, ideas, beliefs, wishes, commands, etc.. The idea seems to be that linguistic expressions carry or convey the information (or whatever else, e.g., emotions or prescriptions) from speaker to hearer. But this picture can be only part of the story. Some information (or whatever else) is indeed transmitted or conveyed through bits of a language. But a not insignificant amount is not conveyed at all. In successful communication, a hearer may come to have the same thought as the speaker (or the one the speaker intends the hearer to have)—that provides the grounds for successful communication. But it does not follow simply from coming to have the same thought that the thought was transmitted or conveyed. The alternative advocated is that the hearer does a lot of the work herself, i.e., by retrieving information from the world (i.e, local or global context, background knowledge, or from memory), not just from the linguistic expression given by the speaker. The expression may provide a possible 'template', so to speak, but it has informational gaps, and these must be filled by the hearer. The important value of this alternative picture is that it allows for things like conversational implicatures and other Pragmatic phenomena. This value aside, it also accounts for the operation of indexicals, or indeed any referring term with an indexical-like character. In addition, it allows us to account for the everyday occurrence of (initial) failures of communication even when hearers have fully understood the piece of language (semantic meaning) of the speaker's utterance: "I understand what you are saying, but I do not know whom (or what) you mean." How does this alternative picture fit with the distinction between Particular Thoughts and Object-Independent Thoughts? What is to be noted is that the entire content of a Particular Thought is not transmitted via a sentence in referential communication because such thoughts contain informational components that are not expressible in language. On the other hand, Object-Independent Thoughts precisely are what are transmitted in linguistic expressions, which makes sense ofthe idea thai: Object-115 Independent Thoughts are object/context independent because surely it must be a feature of language that its items be at least somewhat general, that is, not particular to contexts. If the items were, language would not function as it does. Even an infinitely rich language would not be sufficient, because we now know that no number of descriptions will replace the content retrieved by demonstrative contact with the objects of the world. We get around this problem by having it that speakers are aware of shared information that does not need to be, or, importantly, cannot be expressed in language. This is the information that moves us from "I understand what you are saying, but I do not know whom you mean" to "Ah! Now I understand whom you mean." 116 C H A P T E R 5 P a r t i c u l a r T h o u g h t O u t l i n e d Roughly speaking, corresponding to the ways in which we refer to things, are ways in which we think of things. However, I think that we should be very cautious in making assumptions about to what this correspondence amounts. Also, just as we can refer to particular things, so too, I and others claim, can we think of particular things. When we are thinking of a particular thing, we grasp a Particular Thought} The clearest example of a Particular Thought is a demonstrative Particular Thought—that is, a thought in which the object thought about is demonstratively presented to the subject. A Particular Thought, if linguistically represented, has associated with it a corresponding singular term that denotes the object thought about. As far as I know, no one doubts that there are genuine, singular referring terms. However, many do, apparently, doubt that there are genuine Particular Thoughts. We will see what this doubt amounts to below. There are some points to note before we proceed. First, because of the insistence on talking of Particular Thoughts, and the reversal of the Priority Thesis with respect to these, when appropriate, to avoid potential misunderstanding, I will replace the terms 'singular reference' and 'singular referring term' with 'particular reference' and 'particular referring term'. Second, the distinctions set out in Chapter 2, between Experiences, Object-Independent Thoughts and Particular Thoughts, play an important role in what follows. Recognizing these distinctions should shed some light on some subtle disputes between advocates of various, competing theories of thought. Of particular interest is the distinction between Object-Independent Thoughts and Particular Thoughts. With respect to thought, there are reasons for wanting sharp distinctions. I am led to this by the kind of example we have already seen from Perry. Recall, it is set up by the following circumstances: 1 What I call a Particular Thought is, apparently, what at least some call a singular thought. 117 Ellsworth goes to Hawaii and sends me a postcard. Unfortunately, it gets a bit wet before I receive it. The post mark, return address, and signature are all illegible. The message stays dry: "I am having a good time now."2 Assume that Perry, as the receiver of the postcard, is a competent speaker of English, and is not delusional (etc.). Without getting into the details of the content of the utterance and associated thoughts, what kind of thought is likely to be produced by understanding of the message? Clearly, it can only be this: there is a person who wrote this postcard and was having a good time when he or she did so. However, even though Perry knows that someone (i.e., some particular person—"I") wrote the postcard, and is saying that he or she was having a good time, he has no idea who, that is, which particular person. Understanding the message as it stands, then, does not allow for the possession of a Particular Thought. In other words, if it is agreed that the message expresses a thought, as it stands (i.e., without additional information) it can only lead to possession of an Object-Independent Thought. Another way of characterizing the distinction as it is drawn here, is to note that possession of the Object-Independent Thought does not allow for the assessment of the thought's truth. Perry, if he understands the sentence, has a pretty good idea of what would have to be the case for the thought to be true, but he cannot judge it to be true. In technical language, he has no grasp of the truth conditions of the thought. Only possession of a Particular Thought would allow this. Put yet another way, which will show the relevance for later discussions, it would seem that Object-Independent Thoughts are not Russellian, whereas Particular Thoughts are.3 It is important to note that Evans focused on Particular Thoughts, but sometimes seems to have been taken to be talking about Object-Independent Thoughts when he was talking about Particular Thoughts. It is, in part, because of this confusion that the distinction between Object-Independent Thoughts and Particular Thoughts is needed. Understanding the nature of the 2 Perry (1988), p. 9. 3 This idea of thoughts being Russellian will be explained in the next section. 118 distinction will show why certain disputes arise, and the distinction itself will give a way of addressing the disputes. 5.1 The Requirements for a Theory of Thought The theory of particular reference outlined towards the end of chapter 3 is grounded upon a specific theory of Particular Thought. It is, of course, an important claim of this dissertation that these theories have this ordering of explanatory priority. However, it is my purpose, here, to show that if we do refer in the manner outlined, then there must be particular ways in which we have thoughts about objects to which we refer. The theory of Particular Thought that will develop in this chapter comes, in essence, from arguing against both the Description-based theory and what I will call the standard Causal Theory of Thought. That is to put it negatively. To put it positively, the proposed theory of Particular Thought is something of a blend of, or mediation between, the two views. In other words, the resulting view has at its foundations important aspects of both views. In order to continue, we have to have at least a rough idea of what a theory of Particular Thought must do. What it must do can be put quite simply. A theory of Particular Thought must give the conditions under which it can be said that we are thinking of a particular object. Of course, there is something of an assumption here. Some seem to think that there can be Particular Thought without what others would take to be necessary requirements of 'particularity'. Others go further and doubt that there are (genuine) Particular Thoughts. These issues are addressed in this, and the following chapters. But for the moment, I think that it is sufficient to note that any statement about what a theory does will contain assumptions of this sort. To express such a statement, we need to have in mind a methodological approach, which in turn presupposes a goal. But as Rorty notes in the introduction to The Linguistic Turn, 119 "To know what method to adopt, one must already have arrived at some metaphysical and some epistemological conclusions."4 My goal in this chapter is far more modest than giving a complete theory. All that 1 hope to achieve is, first, to establish some sort of plausibility for there being genuine Particular Thoughts (of a qualified sort, as it will turn out), and second, to contribute to answering the question 'How is thought related to (particular) objects?'. What I hope will emerge, though, is a relatively clear picture of what it is to be thinking about a particular object, so that we have some understanding of the representational properties of thoughts. Furthermore, given the order of explanatory priority here advanced, this picture should be one that is sufficient to explain particular reference. The discussion can be focused by addressing the following, more narrow question: Can a subject be said to be thinking of an object even when she does not have individuating knowledge of the object? 1 take this question to be at the heart not only of the dispute between the description-based theorists and the causal theorists (of thought), but also at the heart of Evans's picture and some recent criticisms of it. In answering this question and in addressing the criticisms, as well as taking on board some of what these critics say, that we will see that Evans's picture essentially amounts to a mediation between the description-based and causal theories. At the risk of prejudicing readers' acceptance of what I have to say, my feeling is that my interpretation (or modification) of Evans's view comes down slightly closer to the description-based side. In making this statement, though, it should be noted that the final position has no important recourse to descriptions, strange as this may sound to some. It is the general idea of the requirement of individuating knowledge that grounds description-based theories which I advocate. Readers should now be aware that the above question is ambiguous, with the ambiguity having its grounding in the distinction between Object-Independent Thoughts and Particular Thoughts. That is, 4 Rorty (1968), p. 1. 120 there are two ways in which "thinking of an object" can be interpreted. To establish the goal of my arguments, I will just state here that with respect to Object-Iridependent Thoughts, I answer the question with "Yes." A subject can be understood as 'thinking of an object' without possessing individuating knowledge, but only in the sense of possessing an Object-Independent Thought that concerns the object. That is, the connection between the Object-Independent Thought and the object is merely satisfactional, not representational. With respect to Particular Thoughts, I follow Evans in answering the question with "No." There is still something else that we need to understand clearly before we proceed. We need to identify from a different angle the target of this investigation. Consider the following. If I say "Cyrano de Bergerac likes pastries," you might, if you suppose that I am not merely reading a line from the script of a play (or some such thing), report this by saying (1) Michael thinks that Cyrano de Bergerac likes pastries. Now, this investigation targets, primarily, what sort of thought I have of who is named by "Cyrano de Bergerac," that is, what sort of way I think about the man Cyrano de Bergerac. It is important to notice that if it is thought that 'Cyrano de Bergerac likes pastries' is a (whole) thought, then what is the focus, here, is that part of the thought that purports to identify a referent—not the name along with the associated predicative part. The query 'Whom do you mean?' is directed at the part of the thought with which I am concerned. Secondarily, though, this investigation targets what it is to think something about an object—that is, what part is added to the first part to make a (whole) thought. For instance, if someone asks 'What do you mean by "Cyrano de Bergerac likes pastries?'", and then went on to say "He detested them," for example, he would be claiming (wrongly, apparently) something about Cyrano de Bergerac. In combination, then, what he would be claiming is that the (whole) thought is false. 121 5.2 Russell's Principle Evans advanced the view that (information-based)5 Particular Thoughts are "Russellian." This means that Particular Thoughts must satisfy what he called "Russell's Principle"—a somewhat modified form ofthe principle of acquaintance proposed by Russell. In a number of places, Russell expressed the idea that "it is scarcely conceivable that we can make a judgement or entertain a supposition without knowing what it is that we are judging or supposing about."6 Evans admits that it is not at all easy to determine exactly what the principle means, but he interprets it to mean, with respect to Particular Thought, that a subject cannot be thinking of an object unless she has "discriminating knowledge" of the object she is thinking about.7 In general, she must, literally, be able to pick out the relevant particular object from all others, including all of identical appearance. And in particular, discriminating knowledge can be possessed either: (i) when she is in direct perceptual contact with the relevant object (that is the paradigm case of a demonstrative thought); (ii) when she can recognize it; or (iii) when she possesses distinguishing facts (in the form of identifying descriptions). Giving this kind of thought an appropriate title, Evans says that a thought is Russellian if it is of such a kind that it simply could not exist in the absence of the object or objects which it is about.8 I take it that what he means is that the thought would not exist because no particular object (no particular feature of the world) is thought about. If I say to you 'That tree is big' but there is no tree around, there 5 A thought is information-based if the content of the thought ultimately has some grounding in the world. That is, if through various modes of representation, the thought is about some particular feature(s) ofthe world. A thought is not information-based if its content is purely fictional, or if it is purely descriptive (which may amount to the same thing). 6 Russell (1959), p. 58. 7 This is not Evans's understanding of what Russell intended, but rather his version of what is required to make it a substantial principle. 8 Evans (1982), p. 71. 1 2 2 is nothing, according to Evans, that would count as understanding what I have said, so you would grasp no thought. However, the quotation seems to be a little misleading because as (ii) and (iii) imply, the subject does not have to have the object presented to her directly in order to have a Particular Thought. Strictly speaking, then, some kinds of thoughts—ones in which there is something other than demonstrative identification employed—are possible if the object is currently absent. Also, remember that as I am describing things, Evans goes too far in claiming that there is no thought at all in cases where the relevant object fails to exist. In such cases there is an Object-Independent Thought. We will see in Chapter 6 how making this claim gives a way of addressing a dispute between Evans and some of his critics. Here is the example that Evans gives to show what it means for thoughts to be Russellian: Suppose, for example, that on a certain day in the past, a subject briefly observed two indistinguishable steel balls suspended from the same point and rotating about it. He now believes nothing about one ball which he does not believe about the other. This is certainly a situation in which the subject cannot discriminate one of the balls from all other things, since he cannot discriminate it from its fellow.9 The point is that if we imagine that we are in this situation, trying to think of one of the balls, we would find ourselves trying to pick up on some identifying feature so that we could individuate the balls. Furthermore, if we cannot come up with this individuating information, it must be the case that we cannot bring ourselves to be thinking of one of the balls in particular. Circumstance disallows discriminating knowledge. It follows, Evans claims, that there is no Particular Thought if there is no discriminating knowledge. This is shown by modifying the above case. Evans says: Let us tell the story of the steel balls rather differently, so that our subject briefly sees one ball rotating by itself one day, and the other on a later day. And let us suppose that the subject retains no memory of the first episode, because of a localized amnesia produced by a blow to the head. Suppose, finally, that many years later our subject reminisces 9 Evans (1982), p. 90. 123' about 'that shiny ball' he saw many years earlier. If asked which ball he was thinking about, our subject cannot produce any facts which would discriminate between the two.10 In this case, even though the subject may talk as if he possessed a Particular Thought (i.e., say things like "That steel ball really was shiny"), it cannot be the case that he does possess such a thought because he lacks discriminating knowledge. However, as Evans notes, given the amnesia, it would be natural to say that "he was thinking ofthe second ball."" If we witnessed his observations, the blow to his head, and knew that he had amnesia, then we would know of which ball he was thinking. But as we will see, just because it is natural to say this, it does not follow that our subject possesses a Particular Thought. We think that it is natural to say this because if asked we would say that the second ball is the one to which he is referring. But if the central theme of this dissertation is correct—i.e., the Linguistic U-Turn—reference does not explain thought. Explaining how reference is achieved in a specific instance, does not give us what is needed to explain what the subject (the referrer) is thinking. Importantly in this case, in working out the reference, we, as audience members, draw on knowledge that is not available to our subject (to think). We know to which ball he refers because we have discriminating knowledge. The arguments that show the requirement of the audiences input in this case are parallel to those that showed that simply noting an instance of successful referring (in the Tourist in the Pub/David Lange case) was not sufficient to show that the subject knew to whom he was referring by "David Lange." I will say more on this point later. But just as a preliminary move, consider a further modification to the steel balls case. Imagine that our subject observes one of the balls, then the next day, after we have switched them without his being aware of the switch, he observes the other. But this time he does not receive a blow to the head (i.e., does not have amnesia). In this case, he is in the same cognitive state as in the case where he does suffer from amnesia, except for thinking that he saw the same ball on two occasions. But , u Evans (1982), p. 90. 1 1 Evans (1982), p. 90. Original italics. 124 this is not a relevant additional piece of information. Seeing something a second time if you do not notice any additional, potentially distinguishing features will not, in itself, help you identify it. The subject is in no importantly different cognitive state. But we are. As before, our subject may well talk as if he possesses a Particular Thought ("That steel ball really was shiny"). However, in this case we have no idea, and no way to tell, of which ball he is thinking. This modified case shows us that it was our knowledge, not the subject's (discriminating) knowledge'2 that allowed us to say that it was natural for us to suppose that he was thinking of the second ball when he suffered from amnesia. There is another way of showing this point. Imagine that some five star hotel wants its guests to have the services of conveniently located, vending machines, but finds their appearance objectionable. Their solution is to hide them behind attractive facades. The machines are fitted with sensors, so that when you approach one, it informs you of what it dispenses. You approach one, and it says, "I dispense carbonated drinks." The one beside it adds "I dispense chocolate bars." You want a chocolate bar, so you know to which machine to go for your chocolate bar. You know that the machine which said "I dispense chocolate bars" is the one that dispenses chocolate bars. This suggests that the 'uses' of the first-'person' pronouns allow for successful reference. Reference succeeds, but surely we would not want to say that the vending machine possesses the Particular Thought that it dispenses chocolate bars. You might think that this account gets something wrong in thinking that we should attempt to trace the Particular Thought back to the vending machine. Rather, we go back further to the programmer of the vending machines, in which case there is no problem in thinking that the programmer possesses a Particular Thought. However, this move seems to miss an important point. The "I" in "1 dispense chocolate bars" really does refer to the vending machine, not the programmer. Though of course, he didn't actually have discriminating knowledge. 125 Evans notes that the "concept of a thought about an individual is tied to the concept of understanding a statement about an individual."13 Understanding such a statement requires the possession of the right kind of thought, i.e., a Particular Thought—one satisfying Russell's Principle.14 But as we have seen in chapter 3, and now put in Evans's current terms, there is a "divergence between the requirements for understanding and the requirements for saying ...." 1 5 Again, as the Tourist in the Pub example showed, we can have successful referential communication, given appropriate circumstances, without it being the case that the speaker knows anything about to whom or what he is referring, or even knowing to whom or what he is referring, if that is different.16 Some further details about Russell's Principle will be helpful before we proceed. Close to the foundations of much of Russell's thinking in this area is the idea that there are two kinds of knowledge of things (i.e., objects, properties and relations in the world). These two kinds of knowledge are based upon the way in which we come to have the (piece of) knowledge. We can acquire knowledge by being acquainted with the object (etc.), giving us "knowledge by acquaintance"; or we can acquire knowledge by coming to possess an appropriate description, giving us knowledge by description.17 Although this distinction may seem fairly obvious to us at the end of the twentieth century, Russell's account does require a little elaboration. For Russell, we stand in a relationship of acquaintance with an object only if we are "directly aware [of the object], without the intermediary of any process of inference 1 3 Evans (1982), p. 92. 1 4 As Evans notes (note 5, p. 92) we can see a parallel between the requirements of Russell's Principle as they are placed here and Putnam's requirement (in Putnam (1975)) that reference to (and probably thought about) a natural kind is grounded upon an appropriate kind of understanding of the natural kind. That is, at least in most cases, we require 'experts' with the appropriate understanding to identify the natural kind. It seems to me that exactly what experts possess, and pass on to us (by proxy), is discriminating knowledge. 1 5 Evans (1982), p. 92. Original italics. 1 6 Presumably, causal theorists think that it is different. It is, in part, what drives their arguments. ' 7 See for instance Russell (1959), pp. 48ff. 126 or the knowledge of any truths."18 It turns out, then, that the clearest examples of what we know by acquaintance are sense data, i.e., not the objects themselves that are the cause of the sense data. However, even Russell admits that this is too restrictive, so the class of objects of knowledge by acquaintance gets extended to include 'objects' found in memory and introspection, and possibly, in some sense at least, oneself. This view is extraordinarily radical, for it turns out that we are acquainted with very little, and almost none of what we would normally think we are acquainted. Most current philosophers have strong reactions to the radical nature of this view. As a consequence, it is common nowadays to extend the class of things that we know by acquaintance further to include, in a qualified way, physical objects. What motivated Russell not to permit this extension, seems to be reflected in his theory of proper names and its connection to his theory of descriptions. However, I will not go into these theories here. It would require a substantial detour, and, in any case, there is wide agreement that Russell's theory of names and theory of descriptions both have critical flaws. Most think that it has been shown that certain referring terms, like proper names, refer to their objects directly. Moreover, following from adherence to the Priority Thesis, the relationship between a thought and the object thought about is thought to be direct as well. Many think, then, that we have a legitimate extension of the class of things we can be said to be (directly) acquainted with, in thought. So, this class commonly gets extended to include physical objects. Of course, it hardly needs to be said that the foregoing is a very short account of an enormously complicated story. The complications and possible qualifications have kept many writers busy for various parts of most of the twentieth century. As I have said, I have not stopped here to consider them, but some will come out indirectly in what follows. Let us assume, then, that it is permissible to extend the class of objects we can be said to be directly acquainted with to include physical objects. We can alleviate some worries about such an 1 8 Russell (1959), p. 46. 127 assumption by considering three things. First, we should note that Russell's concern was knowledge, whereas our concern is (merely) thought—though of course, for some, there will not be too much to this distinction. For instance, there are those (e.g., Evans) who hold that knowledge is a species of belief, and take it that a belief is a kind of thought. The distinction is still important, but the close connection cannot be ignored. If nothing else, then, we can see that Russell had more specific concerns. Second, the arguments of Kripke, Donnellan, Putnam, Burge, and others, are typically thought to support it. Finally, if what I present as a theory of Particular Thought, based on the ideas of Evans and the other Linguistic U-Turners, is at all convincing, it should be taken to give indirect support to some aspects ofthe extension as well, though it certainly results in different conclusions from those reached by Kripke, Putnam, etc.. For one thing, Kripke et al. reject Russell's Principle even though they accept the idea of direct acquaintance with physical objects. What this rejection amounts to is addressed in the next section. Also, just because it is concluded that a term refers directly, it is not necessarily concluded, in addition, that the direct, causal relation between the use of the term and the referent is sufficient to establish the referential properties of the term. Moreover, even if reference is direct, there is no reason to suppose that thought is direct. 5.3 Explaining Russell's Principle As Evans cautions, "The difficulty with Russell's Principle has always been to explain what it means."19 Of course, the previous section goes some way towards explaining the principle. Also, much of what follows in this dissertation must be seen as contributing to the explanation, especially discussions ofthe representational properties of thoughts, i.e., what makes it that a thought represents what it does. However, before we proceed to these discussions, and to those that focus on arguments which aim to support the rejection of Russell's Principle, we need to try, at least, to establish a clearer idea about what 1 9 Evans (1982), p. 89. 128 it means. We also need to give it some significant degree of plausibility, for as it stands, with a little investigation, it may strike many as being quite implausible. At face value, a principle to the effect that in order for a subject to think of an object, she must know which object she is thinking about, seems reasonably clear. However, as Evans points out, any analysis of the principle using common understandings of'know which' is not going to make a workable principle. Evans makes it clearer, and so, workable, as we have seen, by replacing 'must know which object' with 'must have discriminating knowledge of the object'. Possession of discriminating knowledge means that "the subject must have a capacity to distinguish the object from all other things."20 We can see the motivation for this idea in understanding certain characteristics of genuinely representational thoughts. Such thoughts have the characteristic of being able to be judged as true or false of the world. It would seem, then, that such judgements are possible by a subject only if she can discriminate the object of the thought from all other things. The problem is that the requirement of the possession of discriminating knowledge for thought seems to be too stringent. That is, we might think that we hardly ever have discriminating knowledge. There are three ways to proceed from this point. First, we could just bite the bullet, so to speak, and accept that we possess discriminating knowledge only in special cases, which would make Particular Thoughts rare. Second, we could accept that the requirement of discriminating knowledge is too stringent, and so modify the requirements for Particular Thought. Finally, we could look for a way out, that is, some way of maintaining the requirement of discriminating knowledge that allows for Particular Thoughts to be relatively common. I will look at each of these in turn, in the end supporting this final option. The first approach, of biting the bullet, has some historical support. It is, in essence, what Russell espoused for his view of knowledge by acquaintance. That is, there is not much that we know by acquaintance. However, even Russell seemed to realize that this was an unpalatable conclusion. 2 0 See Evans (1982), p. 89. 129 Nowadays especially, we do not think that we have such a limited hold on the world. This conclusion should be left as a last resort. The second approach—that the requirement of discriminating knowledge is too stringent, and so we need to come up with a weaker 'know which' requirement—has an initial appeal. The appeal is that a weaker requirement allows a much wider range of possible Particular Thoughts, while still retaining some of the theoretical advantages of a 'know which' requirement. Recall, we want our way of individuating thoughts to be appropriately fine-grained to account for our judgements about subjects' epistemic attitudes towards their thoughts. In this case, if the 'know which' requirement is too weak, we will be led to the possibility of judging that a subject has different epistemic attitudes towards the same thought. Obviously, we do not want that. Fortunately, we have a very good example of adoption of this second approach. In his book Thought and Object, Kent Bach argues that Evans's requirement of possession of discriminating knowledge is too stringent, while accepting much of what remains of Evans's picture. The main difference between the two views is the following. Bach agrees with Evans regarding the grounding of thoughts, i.e., either in perception (demonstrative grounding), memory, or the testimony of others in referential communication. He seems to agree also that part of what individuates thoughts is their objects. However, whereas Evans has it that the individuation of Particular Thoughts is completed, so to speak, by the mode of identification of their objects, Bach has it that the individuation of such thoughts is completed by their mode of presentation (i.e., via the more familiar Fregean understanding of the individuation of thoughts). The difference may seem subtle to some, but it is significant. The significance is that individuation by a mode of identification makes a thought 'Russellian'. Such identification requires possession of what is sometimes called a fundamental Idea, or a discriminating conception, of an object. Evans says that for any object, there is what he calls "the fundamental ground of difference" for that 130 object, which is what differentiates that object from all others.21 Further, that "one has a fundamental Idea of an object if one thinks of it as the possessor of the fundamental ground of difference which it in fact possesses."22 This idea has the implication, as we have seen, that if there is no object, then there is no thought—no Particular Thought, that is. If there is no object, there is nothing that could count as the fundamental ground of difference, and so no way of forming a fundamental Idea of the object. Back to the problem. Part of what leads Bach to think that the requirement of discriminating knowledge is too stringent is something of which Evans was aware. Let us call it 'the problem of duplicates'. Here is how Evans portrays it. Surely, it must be said, we can imagine that somewhere is the universe there is an exact duplicate of the solar system and all of its contents, containing doubles of each of us. If such a hypothesis were true, none of us would have a capacity to distinguish the objects of our thought from all other objects by their appearance; yet surely our thoughts would continue to have objects, and the very objects they would have had, if the hypothesis had not been true.23 Bach sees the problem of duplicates as being seriously damaging to the requirement of discriminating knowledge. However, his argument against Evans is somewhat indirect, and I do not see it as being particularly forceful. First, Bach suggests that it is unclear what Evans means with respect to his account of Russell's Principle, especially regarding the 'know which' requirement. His understanding is that "Evans maintains that thinking of an individual requires, in each case, knowing which individual it is."24 He says, "an Idea is a discriminating capacity, but being able to discriminate something does not 2 1 Evans (1982), p. 107. 2 2 Evans (1982), p. 107. 2 3 Evans (1982), p. 278. 2 4 Bach (1987), p. 42. 131 entail knowing which thing it is, at least not in any familiar sense of 'know which'." 2 5 But this response is puzzling. In at least two places, Evans warns against "taking the colloquial use of 'know which' seriously in connection with Russell's Principle."26 In any case, Evans seems to give a clear enough idea of what discriminating knowledge amounts to in the account of the 'fundamental ground of difference' of an object. Bach's main claim against Evans comes from thinking that "one can think of something without knowing which thing it is." 2 7 He says that many of his examples have suggested this criterion for thought, but it is not clear which ones he means. There are some examples in the previous paragraphs, but it is not at all clear to me that they show this point. What he seems to have in mind, in general, is the following: attending to an object in perception is sufficient for thinking of that object, even if after looking away, you look back and cannot identify it amongst look-alikes. But as we will see, Evans has a way of addressing this problem. This is the idea that we have 'descriptive elements' of the form 'that which I saw before', or the like, to fall back on. Evans's response is outlined below, though in giving it he warns us that recourse to descriptive elements limits the ways we have in which we can think of the relevant object. The crucial point in Bach's discussion of Evans is his rejection of Russell's Principle. What does he put in its place? He seems to be claiming that a causal relation is sufficient for a thought to be about an object. For example, Bach says that in a memory-based thought, all that matters is that the object be the cause ofthe percept that resulted in the memory." I am not going to address this claim, here, because much of the remainder of the dissertation is given over to arguing against it. Ultimately, then, I do not see that Bach's criticisms hold up against Evans. Bach (1987), p. 42. Evans (1982), p. 283 Bach (1987), p. 44. 132 The third option was to look for a way of maintaining the requirement of discriminating knowledge while allowing for Particular Thoughts to be relatively common. Evans sees part of the solution to the problem of duplicates in relativizing the object of thought to a particular spatio-temporal location. That is, Evans has never claimed, and thinks that it never should be claimed, that an appearance-based recognitional capacity is sufficient for the production of the right kind of discriminating knowledge. What else is needed for complete recognition is knowledge that, at least for some thoughts, is grounded in spatio-temporal details (i.e., what in some situations will be all we have available to give the fundamental ground of difference). Evans points out that "no one can be regarded as recognizing an individual unless he understands that the current appearance of an object can be only a defeasible basis for its re-identification as something previously encountered."2 8 Re-identification requires recourse to background beliefs of the sort that allow for judgements of possible, current locations of the object—for instance, where it was, where it is now, how fast it can travel, if at all, etc. In sum, a subject can have discriminating knowledge of duplicates so long as they have sufficiently different locations. Having sufficiently different locations is only part of the story, though. The problem of duplicates persists i f we allow for the possibility, which we must, that duplicates be insufficiently dispersed. A s Evans notes, if a subject forms a thought about X , but then soon after notices X', a nearby duplicate, we should not want to say that the subject can no longer think of X . In short, Evans's solution to this problem is to suggest that in addition to all that a complete recognitional capacity provides, we have also at our disposal "descriptive elements" of the kind, 'that which I met before', or 'that which I observed before'. 2 9 It is important to note a qualification here. Evans says that this gives us only a "fall-back" mode of Evans (1982), p. 278-9. Evans (1982), p. 282. 133 identification of the object.30 One way of identifying the object is lost with the introduction of a duplicate into a locale. Presumably, the implication is that the subject now loses the capacity to form new Particular Thoughts of the object (in the absence of coming up with new identifying knowledge). I think that we can adapt an idea from Peter Strawson, in his book Individuals, to help explain this solution to the problem of duplicates. Strawson is aware of the problem of duplicates ("reduplication" as he puts it).31 The first thing to note, or rather, be reminded of, is that when we have available demonstrative identification, there is no problem. That is, even if we are presented with duplicates in the same locale, if we can identify one demonstratively, we can think about that particular one. The problem arises when we lose straightforward, demonstrative identification, and so must resort to some kind of "non-demonstrative identification," to use Strawson's term.32 However, as we have seen, the possibility of there somewhere existing a duplicate of the object purportedly thought about, makes it seem that a Particular Thought of that object is impossible. That is, the subject cannot know that she has discriminated the object from all others. However, what can contribute to discriminating knowledge is that we are linked to a locale, demonstratively. The fact that there might exist, somewhere in the universe, a duplicate of the tree I am now thinking of (and currently not in demonstrative contact with) does not bring it about that I cannot form a Particular Thought of that tree. I have available some knowledge that allows for appropriate discrimination. This knowledge comes, I take it, from what Evans calls "Egocentric Spatial Thinking," a capacity that enables us to locate objects and ourselves in space-time.33 I say a little more about this later. But we can get the gist by understanding that discriminating knowledge is what J U Evans (1982), p. 284. 3 1 Strawson (1959), p. 20ff. 32 Here we should note part of why this is an adaptation, though. For Strawson, non-demonstrative identification is, and only is, descriptive identification. However, a recurrent theme in Evans, one I endorse, is that discriminating knowledge need not be only descriptive. 3 3 See Evans (1982), p 150, and section 6.3. 134 links the thought and what is thought about. I can create such a link, even in face o f the truth o f the hypothesis of duplicates, by arriving at the knowledge that I and the object thought about share (or shared) a spatio-temporal location. It was just this relationship that allowed for the relevant perception (the one that produced the informational state which I now link to the thought so that it becomes part of the thought content). 5.4 T h e Challenge to Russell's Principle: T h e Photograph M o d e l I w i l l consider two challenges to Russell 's Principle. The first is indirect, and involves an extension o f Kr ipke ' s picture o f reference to a theory of thought. I w i l l present and discuss this challenge in this section. The second comes from picking up on a common theme running through several direct attacks on Evans's application of Russell 's Principle. This second challenge w i l l be presented and discussed in Chapter 6. There is much that is common to both challenges, but they have important differences, so it w i l l be illuminating to see both. Commonly, the rejection o f Russell 's Principle is very closely connected to the rejection o f a Russellian (that is, description-based) account of singular reference. It is taken, through adherence to the Priority Thesis, and especially the Assumption of Parallelism, that discriminating knowledge can only ever amount to possession o f an appropriate definite description, or set or cluster o f identifying descriptions. Having discriminating knowledge involves having in one's possession certain identifying (or perhaps recognitional) beliefs. These beliefs are paralleled by sentences—i.e., the appropriate sentences, containing definite descriptions, express the beliefs. Very briefly, because it is thought that the Russellian account of singular reference is to be rejected on independent grounds (Kripke, Putnam, etc.), it is thought to follow that Russell 's Principle must be false. However, now, there is an important extension to note. The move just outlined pertains to reference. But many have taken it that Kr ipke ' s and Putnam's (etc.) considerations are legitimately 1 3 5 extended such that they pertain to thought as well. That is, it is thought that Russell's Principle must be false because the Russellian account of Particular Thought is rejected. Evans put it this way: The abandonment of the principle of identification at the level of saying is a trivial consequence of the distinction between what one says and what thought one intends to express. Its abandonment at the level of belief or thought would be an extremely significant move. What has happened is that the former has been mistaken for the latter.34 Evans considers what he calls the Photograph Model as a model of mental representation that has as a very important element the abandonment of Russell's Principle.35 He argues against the Photograph Model, and it is, in part, his rejection of it that motivates his accounts of particular reference and Particular Thought based on the idea that information states are primitive to belief. There will be more on his positive accounts later. But first, let us consider the Photograph Model. 3 6 The Photograph Model comes from the application of the Donnellan-Kripke-Putnam 'theory' of reference (the Causal Theory of Reference) to thought—Particular Thought. Recall that, roughly speaking, causal theorists hold that proper names are rigid designators—that is, they designate the same object in any possible world—and that it is through causal-historical chains originating in baptisms that the reference of proper names is determined. Recall also that it is thought that the main advantage of this view is that it can explain how it is possible that speakers can use a name correctly to refer to an object even though they may possess no description (or set or cluster of descriptions) that uniquely satisfies that object. Moreover, that speakers may even associate false descriptions with the name, and still refer. Kripke says: "In general our reference depends not just on what we think ourselves, but on other people in the community, the history of how the name reached one, and things like that. It is by following such 3 4 Evans (1982), note 18, p. 76. 3 5 Evans (1982), p. 78. Recall from section 3.6 that the Photograph Model is meant to characterize the view of mental representation that says that mental states represent their objects by being causally related to those objects just in the way that a photograph is causally related to what is photographed. 3 6 See Evans (1982), roughly, pp. 67-79. 136 a history that one gets to the reference."37 When this model is extended, it is supposed that there can be Particular Thought without discriminating knowledge—that is, that Particular Thoughts need not be Russellian.38 Kripke, in particular, thought that a subject could refer by use of a proper name without knowing anything about the object to which she is referring. This claim is a central element of such theories. Not only does it argue against description theories, but it supports the idea that it is through causal-historical chains, alone, connecting current name users with baptisms, that reference is secured. Moreover, it is an easy step, so it is supposed, from this claim to the idea that not only can subjects refer in this manner, but also they can be thinking of the object. Evans points out that Kripke did not explicitly argue against Russell's Principle, but it is easy to see that Kripke's claims could lead some to views that directly oppose the principle. In fact, this has happened, and it led to the Photograph Model. Roughly, the Photograph Model comes about by thinking first that the 'representational properties' of [an] ignorant name-user's utterance depend upon the existence of a causal relation between that utterance and an object, which is extended (controversially) to the representational properties of the ignorant name-user's thought—the thought he uses in expressing the name—depend upon the existence of a causal relation between a psychological state and the object it represents.39 The extension is reflected in the following move. First, it is held that the referential (or representational) properties of a singular term are grounded solely upon a causal relation between the use ofthe term (by a speaker, on an occasion) and some particular person or object given that referring term in a baptism. We have seen how this view of reference has been developed into the Causal Theory of Reference. Kripke (1980), p. 301. What this means is made clear below. Evans (1982), p. 77. The name-user is ignorant in that he has no discriminating knowledge. 137 Second, and correspondingly, it is taken that the representational properties of a Particular Thought are grounded solely upon what amounts to the same causal relation—i.e., replacing 'use of the term' with something like 'thought associated with use of the term' (and where it is held that the thought is paralleled by the sentence used by the speaker). This move gives rise to what I will in later discussions call, to show the connection, the Causal Theory of Thought. For the moment we will follow Evans's title. He calls the result of the second move the Photograph Model of Mental Representation. The 'Photograph Model' is so named because a photograph is of whatever it is that is represented, and there is a straightforward causal relation between the photograph and what is photographed. That is, if I take a photograph of a tree, it is a photograph of that tree in virtue of the causal relation between the tree and the photograph via the camera. With respect to thought, Evans says that "according to the Photograph Model, a mental state can represent a particular object simply in virtue of that object's playing a suitable role in its causal ancestry."40 Lest we be immediately seduced into thinking that because a photograph is of what is photographed, the Photograph Model correctly accounts for the representational properties of thoughts, we should consider a further employment by Evans of the model of a photograph. Evans employs the model of a photograph to help explain the workings of the Information System. He does not say so, but I take it that there must be a connection between the two accounts. The difference, however, is that in the account of the workings of the Information System, we see how Evans will argue against the Photograph Model. Briefly, here is the account. The Information System receives information from things in the world in a way analogous to the way in which a photograph, via a photographic mechanism, does. Just as a photograph possesses informational content, so do our minds. To continue in Evans's words: The mechanism is a mechanism of information storage, because the properties that figure in the content of its output are . . . the properties possessed by the objects which are the Evans (1982), p. 81. 138 input to it. And we can say that the product of such a mechanism is of the objects that were the input to the mechanism when the product was produced.41 However, to say that a photograph is of whatever it is that is represented in the photograph, is not to say that the photograph Represents, in the strict sense, some state of affairs. Evans says that, contrary to a painting, say, "a specification of the content of a photograph should [not] make reference to the object or objects that it is of." This passage is perhaps a little unclear. What I take him to mean, in a somewhat stipulative way, is that in specifying the content of a photograph, we need not make reference to the particular object(s). Continuing, Evans says that we need to distinguish between something that is merely a "representation of a" (some state of affairs), and something that is "an a-representation (i.e.. a species of particular-representation, in a specification of whose content mention of a would figure: something which represents, or misrepresents a)."42 This point is rather subtle as it is characterized here, so it probably will be helpful to say some more before we proceed. Something that may be appropriate is Kaplan's way of characterizing the distinction between representation and resemblance. Kaplan argued that a genuinely representational thought requires that the subject of the thought be "en rapport" with the object.43 There are two kinds of links required for a subject to be 'en rapport' with an object. The first one comes from the subject's being in possession of an appropriate singular term whose "descriptive content" denotes the object. Descriptive content provides one of the links to the object, and is connected to the "vividness" of the term (or thought)—i.e., "the 4 1 Evans (1982), p. 125. 4 2 Evans (1982), p. 125, note 10. 4 3 See Kaplan (1969), sec. IX. 139 descriptive content... determines [the] vividness."44 Getting a precise idea of 'vividness' is not easy, but roughly, the more detailed the description or conception of the object named, the more vivid the name.45 Moreover, the singular term must be sufficiently vivid for it to count as representing the object. Second, the subject must be in a suitable causal relation with the object. This second link provides the "genetic character," or 'of-ness', of the associated singular term, and brings it about that the thought is about the particular object, not some other one that the term might also denote. Kaplan explains this distinction with the analogy of the "resemblance" (analogous to 'descriptive content') and "o/-ness" (analogous to 'genetic character') of a picture. That is, two pictures might be 'of, i.e., represent, two different people (e.g., twins), yet 'resemble' one another such that either picture could be used to point to either of the twins. It may be that what Evans has in mind is that a "representation of a" characterises a relation of resemblance, whereas an "a-representation" is a genuine representation. Noting another distinction also may be helpful. As we have seen, Grice distinguished between what he called Natural and Nonnatural meaning.46 It may not be obvious, but it seems to me that what this distinction points to is something like what Evans has in mind. A photograph, in itself, does not meanNN anything. If I were to take a photograph of a man and a woman meeting, and slap it down on a desk in front of the wife of the man (the woman in the photograph is not the wife), then it would mean something; it would 'carry' intentional content. But the photograph does not, by itself, mean anything. All that it does is picture a woman with a man, who, for all we know, merely resembles the wife's husband. Put yet another way, a photograph is not intentional, whereas a painting or drawing may well be. Notice, also, 'Vividness' is explained with the use of an analogy. Kaplan writes: "In addition to the link with reality provided by the relation of resemblance the descriptive content of a picture determines its vividness. A faded picture showing the back of a man wearing a cloak and lurking in shadow will lack vividness. A clear picture, head on, full length, life size, showing fingerprints, etc. would be counted highly vivid (Kaplan (1969), p. 199.). 4 5 Analogously, the clearer and more detailed the picture, the more vivid the picture. See Grice (1957), pp. 377-388. 140 that in naming the Photograph Model as he does, and given the distinction just mentioned, we get towards Evans's objection to the Photograph Model. Namely, it does not, and cannot, account for the intentional, or representational, properties of thoughts. The important thing to note in considering and assessing theories of thought like those which are grounded upon the Photograph Model is that the causal relation is supposed to be sufficient for the psychological state (the thought) to represent the object.47 Now recall, roughly stated, Russell's Principle is that in order to think about an object (or make a judgment about it), an agent must have discriminating knowledge of the object she is thinking about. Clearly then, the Photograph Model is in direct conflict with this principle because of the claim that the causal relation is sufficient. However, some claim that it is not sufficient, and in addition to the causal relation, discriminating knowledge is required. At this point certain grounding ideas of description-based theories are brought back into the picture. 5.5 The Defence of Russell's Principle: Rejection ofthe Photograph Model Evans's defence of adherence to Russell's Principle can be divided into two general strategies. First, he gives us a fairly explicit statement of the principle, and gives us some examples that, when we consider his statement of the principle, seem to show that the principle has a certain intuitive appeal. The steel ball example was one such example. But he does not rely simply on this example, mostly because there are what appear to be counterexamples to the principle. Second, then, Evans explicitly states that Russell's Principle ought to be accepted only on the basis of appropriate theoretical arguments, and he uses these arguments to undermine the counterexamples. Evans's discussion of these theoretical considerations is thorough and detailed, and I see no point in summarizing all of it. There is, however, one piece that is especially relevant, in part, because it Remember from Chapter 3 that there may be reason to think that Kripke was suspicious of the sufficiency claim. 141 introduces an important technical term. The principle theoretical consideration comes in the form ofthe introduction of a further principle, what Evans calls the Generality Constraint,48 which comes from the general idea that thoughts are "essentially structured."49 Evans notes that this idea may lead some to the notion of a language of thought (perhaps because of thinking that the structure of thoughts is parasitic upon the structure of sentences in the language of thought), but he does not think that we are necessarily led to it. In fact, he is inclined to look for the explanation of structure elsewhere, i.e., in the possession of certain abilities. He prefaces the discussion of the Generality Constraint with this expression of it. We cannot avoid thinking of a thought about an individual object x, to the effect that it is F, as the exercise of two separable capacities; one being the capacity to think of x, which could be equally exercised in thoughts about x to the effect that it is G or H; and the other being a conception of what it is to be F, which could be equally exercised in thoughts about other individuals, to the effect that they are F . 5 0 So, having thoughts of particular kinds (for instance, of the form is F'), involves the exercise of certain sorts of abilities—for instance, an understanding of what it is for some particular thing to be F. These abilities amount to, first, the possession of any relevant concepts (i.e., the concept associated with F), and then, having the ability to apply (or employ) these concepts in any other appropriate case. Adherence to the Generality Constraint, then, requires holding that if a subject has the ability to form the thought 'a is F', then she also has the ability to form the thought that 'b is F'. Or, if she also possesses the concept associated with G, she has the ability to form the thought that 'a is G'. Of course, having this ability does not mean that the subject will ever form any or all of the related thoughts. There may never be any need for her to form the thought that 'a is G' . Alternatively, there may be some independent, unrelated hindrance to her forming the thought—that is, maybe she has not made all of the necessary coherence connections that a is the kind of thing that would be G. Or, maybe it is wildly implausible that a is G. 4 8 Evans (1982), pp. 100-105. 4 9 Evans (1982), p. 102. . 5 0 Evans (1982), p. 75. 142 Why look at the Generality Constraint here? The reason can be seen when we recall what is entailed by a view that rejects adherence to Russell's Principle, in the context of how such a view is typically supported. Recall, in general the supposed support comes from the extension of the arguments that support views which hold that the function of singular referring terms is explained in terms of causal relations between a term's use and an established name-using practice. But, Evans says: If we are to say that the sheer introduction of a subject to a name which has a referent in the community may suffice to enable that subject to have thoughts about the referent. . . in contravention of Russell's Principle, then we are committed to saying that the subject has thereby acquired a capacity to entertain indefinitely many thoughts about the referent; one, in fact, for each simple or complex property of which the subject has a conception.51 This conclusion follows from the Generality Constraint. In part, rejecting the requirement of Russell's Principle for thought entails holding that a subject can grasp a Particular Thought without having discriminating knowledge of the object thought about. But when we see that thoughts are produced and structured in accordance with the Generality Constraint, it follows that if you reject the requirement of Russell's Principle, you must hold that a subject can grasp thoughts about some object even if she knows nothing about the object. At the very least, holding that thoughts can be grasped in this way gives us a very loose idea of 'grasping a thought'. Similarly, it gives us a very loose understanding of the representational properties of thoughts. But surely we want a much more robust idea of 'grasping a thought', and a different understanding of a thought's representational properties. It is part of what I am defending here, and, I think, part of what theorists want in explaining how thoughts relate to the world, that representational properties of thoughts are not so easily established. But why do we need a more robust notion of 'grasping a thought'? Answering this question is, of course, one of the goals of the dissertation as a whole. However, a quick answer can be given with respect to the rejection of Russell's Principle in the face of the requirements of the Generality Constraint. Rejecting Russell's Principle in the face of the Generality Constraint results in being able to ascribe to a 5 1 Evans (1982), p. 75. 143 subject an otherwise implausibly wide range of thoughts about various objects. Just because Laura has the thought 'A is F , it surely does not follow that she has all of what is required to have the thought 'B is F—even if B is F. There are epistemic considerations as well.5 2 Just because Laura is warranted in thinking 'A is F', and has a clear conception of what it is for something to be G, it surely does not follow from these facts alone that she is warranted in thinking 'A is G'—even if A is G. Some readers will not be satisfied. Some will be thinking that Laura indeed does have all of what is required to grasp, respectively, 'B is F' or 'A is G' . However, we need to get a clearer idea about what grasping thoughts like these could mean. It is hard to deny that thoughts are produced in accordance with the Generality Constraint. However, what kinds of thoughts? It seems to me that there are two answers, where the grounding of each answer depends upon which of the two separable capacities is employed in the production of the new thought. First, consider that Ryan possesses the Particular Thought expressed as 'John Walker was a great miler'—he has seen him run, say. Then via the capacity that allows for the attribution of other properties to John Walker, he forms another thought, say, 'John Walker likes breeding horses'. What kind of thought is this? Ryan retains discriminating knowledge of John Walker, so any John Walker-thought he entertains is still about John Walker. That is, any thought will be a Particular Thought. However, there are still the epistemic considerations. Whether Ryan is warranted in assenting to the thought depends on what he knows. If he knows of John Walker's non-running interests—in particular, his penchant for breeding horses—then he will be warranted in assenting to 'John Walker likes breeding horses'. Ryan could be presented with the question 'Does John Walker like breeding horses?', and be warranted in giving his answer. However, if he knows nothing about John Walker—except that he was a great miler—he will not be warranted in assenting to the thought 'John Walker likes breeding horses' formed via his (general) These are not central to the discussion, but they seem to be needed to flesh out the explanations. 144 conception of what it is to like breeding horses. If asked the above question, he would have no way of giving the answer, or rather, no way of being warranted in giving the answer. Alternatively, if Ryan possesses the Particular Thought 'John Walker was a great miler', but this time employs the capacity that allows for the attribution of the same property to a different individual, say, 'Steve Scott was a great miler', he does not, by this move, grasp a Particular Thought. Ryan may come to form the Particular Thought expressed as 'Steve Scott is a great miler' by acquiring discriminating knowledge of Steve Scott. But if, to see the point being made, he does not now have such knowledge, he cannot come to possess this Particular Thought. In going from 'John Walker is a great miler' to 'Steve Scott is a great miler', within the requirements of the generality Constraint, and in the absence of additional discriminating knowledge (of Steve Scott), Ryan's new thought can only be an Object-Independent Thought. Evans's argument against the legitimacy of the Photograph Model is complex, but I think we can get the gist by looking at a case that seems not to fit the model. Evans has in mind various cases where an agent refers (or attempts to refer) to an object by use of a linguistic expression that is based (or is supposedly based) upon an information-based thought, but where something goes wrong. What goes wrong is that the information is not derived from the object to which the agent is referring. He says of such referring expressions (and possible, related information-based thoughts) that they are not "well-grounded."53 In his words: Someone essaying an information-based particular-thought will employ some purported mode of identification of an object: and his attempt at a thought will evidently not be well-grounded unless the object (if any) which that mode of identification would identify is the object (if any) from which the information derives.54 Evans (1982), p. 132. Evans (1982), p. 132. 145 It is important to be aware that things can go wrong in a number of different ways. For example: the mode of identification may not be appropriate for the object being thought of; there may exist no object; there may be too many objects, that is, no unique object is picked out even though the information is derived from one particular object—or, the agent could just have garbled or distorted information. What gets us going in arguing against the counter-examples to Russell's Principle? Evans's starting point seems to be the realization that "not everything which counts as interpreting (making sense of) the linguistic acts of a speaker involves the ascription to him of judgements or beliefs."55 In other words, understanding what thought someone grasps is not necessarily the same thing as understanding what that person is saying. In the steel ball case we have a way of interpreting the speaker's utterance "That steel ball really was shiny." However, the claim is that this understanding will not lead us to what thought the speaker grasps. In fact, in the example, he fails to grasp a Particular Thought, even though, as Evans notes, he "certainly did essay a particular-thought."56 Consider again the modified steel ball example. John sees a steel ball suspended from a hook and rotating around it. The next day he sees another steel ball in exactly the same position, but he is unaware that there are two distinct balls. Now, some time later, John has occasion to say 'That steel ball really was shiny', thereby intending to refer to a particular steel ball. As I have said, and considering what this case shows us about thought, it cannot be the case that John is thinking of a particular ball. What this case shows, speaking generally now, is that it seems that the causal relation that traces a path from an object, through the use of a name in a particular name-using practice to refer to the object, to the object itself, is not sufficient to insure that the thought is about the object. The Photograph Model does not seem to represent what is going on here. However, as Evans claims, an account requiring adherence to Russell's Principle does. 5 5 Evans (1982), p, 131. 5 6 Evans (1982), p. 129. 146 Put generally, Evans claims that what examples like this one show are that information-based thoughts can be not well-grounded. That thoughts can be not well-grounded is important because it allows for things to go wrong, which is important because things do sometimes go wrong. Our thoughts sometimes do fail to represent the world—or indeed, sometimes, w/srepresent the world. The Photograph Model fails because it cannot allow for this 'illgroundedness', i.e., representational failures, or misrepresentations of the world. Evans's proposal is that we must have an appropriate kind of link between the thought and the object. It is a causal link, but not the one suggested by the Standard Causal Theory of Reference. It is one that traces a path through a mode of identification that employs discriminating knowledge which is grounded in information from the object. I will say more on this causal link in chapter 6. But just to close off this section, I think that there may be reasons for thinking that Evans has not gone far enough in distinguishing his view from the Causal Theory of Thought. Recall that in the case of the Photograph Model there are unmediated, causal links between thoughts and objects—objects cause mental states. As mentioned, at least in some of his characterizations, Evans seems to be saying just this, i.e., that the link between a thought and an object is causal. For instance, we see this in some characterizations of what is supposedly the paradigm case of demonstrative thoughts, where the causal, demonstrative contact gives 'aboutness'. He says, for example, "When a person perceives something, he receives . . . information about the world."57 But 1 think that there are reasons for thinking that genuinely representational properties of thoughts are not established by perception-based informational links alone. Evans's exposition, or at least my understanding of it, is a little unclear on this issue. There are times when Evans does seem to be claiming just what I endorse, contrary to the above quotation—for instance, and in general, in claiming that even in the demonstrative case, an information link is not sufficient to Evans (1982), p. 122. The emphasis is mine. He does say later, though, that the information is "of or "from" the object (p. 124). 147 establish that a thought is about an object. 5 8 More specifically, in claiming that (perception-based) informational states are nonconceptual, so belief-independent, the implication is that they are not genuinely representational.59 The next section will explain the nonconceptual/belief independent nature of information states. 5.6 Information-based Thought As just mentioned, in rejecting the Photograph Model, Evans is not rejecting all that is claimed by those advancing the Standard Causal Theory of Thought. As we saw, for one thing, he says that "very often, our thoughts about objects rest upon information which is causally derived from an individual."60 In fact, in his positive account of the relationship between our mental states and the objects with which they are concerned, Evans relies heavily upon there being appropriate causal connections. The paradigm cases for such connections are when we are in demonstrative contact with particular objects. However, the first thing to note is that the causal relation that Evans has in mind here is not the same as the one that traces a path from the object, through the baptism and subsequent use of a name in a particular name-using practice, to the subject of the thought. Rather, it is the one that traces a path back to the source of information, via the processes of what he calls the "informational system," which ultimately generate "information-based particular-thoughts."61 The second thing, again as mentioned before, is that it is not clear what weight, if any, Evans puts on this causal relation in establishing the representational properties 3 8 See Evans (1982), p. 145ff.. 5 9 See Evans (1982), p. 122ff, p.151 ff, and p. 227. An interesting thing to ask, now, is how this affects standard ideas of demonstrative contact. Unfortunately, addressing this directly will have to wait for some other project. 6 0 Evans (1982), pp. 77-8. 6 1 See Evans (1982), ch. 5. 148 of thoughts. What weight is given to this causal relation is addressed to some extent in what follows, but more directly and extensively in chapter 7. Evans gives a thorough account of what he calls information-based thoughts. I will try to be brief in presenting the essential features. One of his goals was to give a theory of thought based on the idea of an Information System, which is characterized in this way: When a person perceives something, he receives (or better, gathers) information about the world. By communicating, he may transmit this information to others. And any piece of information in his possession at a given time may be retained by him until some later time. People are, in short, and among other things, gatherers, transmitters and storers of information. These platitudes locate perception, communication and memory in a system—the information system—which constitutes the substratum of our cognitive lives.62 Evans thinks, then, that it is preferable to take the notion of being in an information state with such and such a content as a primitive notion for philosophy, rather than to characterize it in terms of belief.63 That is, information is taken to be primitive to belief and so expression of belief. States within this information system have the property of, as Evans calls it, "belief-independence." Being belief-independent, we have these informational states independent of, or prior to, having any attitudes towards any associated thoughts about experiences, say, believing that they are true. In general, the idea that states of the information system are belief-independent is supported by the following considerations. The subject's being in an information state is independent of whether or not he believes that the state is veridical. It is a well-known fact about perceptual illusions that it will continue to appear to us as though, say, one line is longer than the other (in the Miiller-Lyer illusion) even when we are quite sure that it is not. Similarly, it may still seem to us as though such and such an episode took place in the past, even though we now believe our apparent experience of it to be hallucinatory. And our being placed in the appropriate Evans (1982). p. 122. To be charitable to Evans, here, and in response to the brief discussion at the end of the last section, I do not read too much into his use of "about" in the first sentence of the quote. I take it that he does not mean that the informational states are genuinely representational. Evans (1982), p. 123. 149 informational state by someone telling us a story does not depend upon our believing the story to be true.64 The Muller-Lyer illusion. I will elaborate just on the perception-based case. Many take it that the fact that we still see, for instance, the Muller-Lyer illusion, even though we can come to know that it is an illusion (say, by taking a ruler and measuring the lines) shows us that the perceptual informational content of the illusion must be of a different kind than that available to the faculty of judgement. What follows from these considerations is that Evans drives a wedge between kinds of content. The content of a belief is associated with judgement and reasons, and is conceptual. These are all connected. But the content of an experience cannot be like the content of a belief. It is nonconceptual. A problem, though, arises concerning how to establish the connection between the nonconceptual and the conceptual because it is clear that we can, if we are pressed, come to make a judgement about the illusion, i.e., a judgement about what we see. Evans does consider what we might call the standard view. This view is that the only way we can establish a link between experience (the nonconceptual content drawn from perceptual information) and Evans (1982), p. 123. 150 the faculty of judgement, is to suppose that an experience is a disposition to make a judgement, or a "prima facie inclination to believe."65 However, Evans thinks that this account gets things the wrong way around. The problem of illusion shows us that with this account we are left with no way of adjudicating between appropriate and inappropriate experiences. There is a distinction to be drawn between having a belief and being entitled to have a belief (or entitled to think that it is true, say). Put another way, the content of the belief is not to be identified with the content of the judgement made about its truth—though there is much in common. This distinction allows misrepresentions. The problem with the standard view is that by having it that experience (merely) gives us a disposition to judge, we do not successfully establish a link between experience and judgement. In John McDowell's words, "In a picture in which all there is behind the judgement is a disposition to make it, the experience itself goes missing."66 The standard view seems to make it very difficult to account for illusion or misrepresentation. If experiences are taken to be reasons for judgements, then there is nothing left to account for illusion. In this context, the only errors that are possible are rational errors—that is, errors in judgement. But in an illusion, say, judgement is what tells us that it is an illusion. It is not the source of the illusion. It is from these kinds of considerations that we get to Evans's important claim that information is more primitive than belief, and hence, judgement. Informational states, then, are nonconceptual. Evans (1982), p. 124. See McDowell (1994), p. 61ff for a discussion of these ideas.. McDowell (1994), p. 61. 151 C H A P T E R 6 Particular Thought Defended There have been a number of quite significant attacks made on Evans's idea of the requirement of discriminating knowledge for Particular Thought. Most are based upon the question asked near the beginning of the previous chapter. Namely: Can a subject be said to be thinking of an object even when she does not have discriminating knowledge of the object? Rejecting Evans's arguments, these opponents claim that even in the absence of discriminating knowledge, a subject can, nonetheless, possess (what they usually call) a singular thought about an object. Typically, it is taken that a failure to have discriminating knowledge of an object is most clearly represented in cases where the putative thought fails to have an object—i.e., where a subject, through illusion or possession of some other kind of misinformation, believes that there is an object, whereas, in fact, there is none. So, they argue, given that there is no object, discriminating knowledge cannot be a necessary requirement for such thoughts. This argument may need a little explanation. Here are two examples of how the rejection of the Russellian status of Particular Thoughts is reflected in answering the above question positively. Peter Carruthers argues against the claim that "most, if not all, singular thoughts are Russellian—that is to say, that in case of the non-existence ofthe object of a putative singular thought, there is no thought."1 Peter Pagin, though not so clearly an opponent of Evans, describes a Russellian thought as "a thought about an object that would not have been available had that object not existed."2 These ways of characterizing Evans's position follow from the requirement of discriminating knowledge for Particular Thought in that if there is no object, then there can be no 1 Carruthers (1987), p. 18. 2 Pagin (1992), p. 27. 152 fundamental ground of difference for that object. And when there is no fundamental ground of difference (nothing that counts as differentiating that object from all others), there is no fundamental Idea of the object (that which leads to discriminating knowledge). It should be noted that this idea that a Particular Thought is not available if the object does not exist characterizes only one way in which a subject can fail to have a Particular Thought. That is, by Evans's lights, failure to have a Particular Thought, does not require the nonexistence of the putative object of thought. It just seems to be that opponents take cases where the object does not exist as paradigm examples. The dispute between Evans and those wishing to reject the requirement of Russell's Principle can be put quite simply, and I will give a general characterization before we go on. However, there have been several different approaches, so it will be necessary to present the details of these. I will focus, first, on the criticisms that come from Simon Blackburn, both because he leads us to some important discussions, and because some think that he is successful in overturning Evans's view.3 Two others are from Peter Carruthers and Marleen Rozemond.4 Rozemond's recent paper "Evans on De Re Thought," gives us a clear way of characterizing the dispute in general. In it she argues in favour of what I call the Causal Theory of Thought, and attacks Evans's criticisms of it. In so doing, she employs arguments typically employed by Causal Theorists. Rozemond sets up her argument by employing Tyler Burge and Saul Kripke against David Kaplan, before moving on to address Evans's arguments. 3 See for instance Carruthers (1987), note 1. He says, "some of [Evans's arguments] have been successfully dealt with by Simon Blackburn." 4 See Blackburn (1984), ch. 9; Carruthers (1987); and Rozemond (1993). See also Segal (1989) for a related discussion, focusing on McDowell's work. 153 Kaplan argued that a genuinely Representational thought, what Rozemond calls a de re thought, requires that the subject of the thought be "en rapport with the object.5 Recall from chapter 5 that three things are required for a subject to be 'en rapport' with an object. The subject must possess an appropriate singular term whose "descriptive content" denotes the object. The singular term must be sufficiently "vivid." In addition, the subject must be in a suitable causal relation with the object. This causal link provides the "genetic character," or 'o/-ness', of the associated singular term, and makes it the case that the thought is about the particular object, not some other one which the term might also denote. Rozemond continues, arguing that in "Belief De Re" Burge is successful in arguing that none of Kaplan's required links is necessary for genuinely representational thoughts. Here is where we see the style of argument to which I am alluding begin to emerge. Burge writes: "On seeing a man coming from a distance in a swirling fog, we may plausibly be said to believe of him, that he is wearing a red cap. But we do not see the man well enough to describe him or imagine him in such a way as to individuate him fully."6 Rozemond argues that what this counterexample shows is that even though our representation of the man in the fog is, literally, not vivid, we nonetheless can form de re thoughts about him (for instance, something like, 'That man is wearing a red cap').7 That is, she argues that we have a genuinely Representational thought even though we lack, or have very poor, descriptive content. How is it thought that we are able to have such thoughts? The argument is not yet made clear, but I think that the explication of the counterexample that follows does show us. Rozemond continues: Another counterexample is provided by Kripke. He points out that many people who use the name Feynman only know that Feynman is an important physicist. Yet they manage to refer to him using that name. Again, the people in question do not have a vivid name 5 See Kaplan (1969), sec. IX. 6 Burge (1977), pp. 351-2. 7 Rozemond (1993), p. 278. 154 of Feynman. It seems clear, however, that they can have de re thoughts about Feynman, by virtue of the causal chain going from their use of the name to the famous physicist.8 Explicitly, the claim is that we manage to form genuinely representational thoughts even in situations where we fail to have an individuating denotation, or descriptive content. It follows, then, that descriptive content must not be necessary for genuinely representational thought. And if that is true, then the causal-historical relation must not only be necessary, but also be sufficient for genuinely representational thought. Rozemond notes that some may claim that (uniquely) individuating descriptive content may not be necessary, but the content, or concepts, must apply to the object in some less than determinate or individuating way. Her response shows again the style of argument I wish to criticize. She says: Imagine the following situation. Mary is reminiscing about what she thinks is an interesting Rumanian movie she saw with Joan. However, she never saw a Rumanian movie with Joan. She did see a Hungarian movie with John, which at the time she did not think was particularly interesting. In fact the concepts constituting the name in question are of the Hungarian movie. The explanation of her misconception may be that after seeing the Hungarian movie she thought about how the movie would have been very interesting if it had been different in certain ways. The result of these speculations had become mixed with the memory of the actual movie. It seems that in this case Mary has de re thoughts about the Hungarian movie she saw, despite the fact that most of her concepts do not apply to it.9 The point of these counterexamples is to show (supposedly) that the requirement of the possession of descriptive or conceptual means of individuation of the object referred to/thought about, is not necessary for genuinely representational thought. Again, the claim supposedly sustained is that the contextually determined, causal-historical relation is not merely necessary, but is sufficient for genuinely representational thought. Rozemond claims that the same kind of observation she points to in the Mary and the movie case was made by Kripke with his famous Godel/Schmidt case. Interestingly, she notes that Kripke was concerned with reference, not thought, but thinks, nonetheless, that Kripke's observation applies to 8 Rozemond (1993), p. 278. 9 Rozemond (1993), pp. 278-9. 155 thought. Conveniently, this example highlights exactly the point I want to make. The assumption that is made in these arguments, the one which is in question, is, in broad terms, that reference explains thought—i.e., that the arguments that support the Causal Theory of Reference are legitimately extended to support the Causal Theory of Thought. We see this extension in Rozemond's argument. For instance, the fact that people nonetheless manage to refer to Feynman through there being a causal-historical chain tracing a path back from their use of the name to Feynman (even though they lack individuating descriptive content) shows that they have thoughts about Feynman. The argument is not explicit in the Mary and the movie case, but the move must be the same because Rozemond talks about the concepts associated with the name of the Hungarian movie. That is, it is taken that the referential properties of the name explain the representational properties of the thought. A comment on terminology is necessary before we proceed. As I have said before, it appears to be common to refer to the thoughts we have regarding individual things by the term "singular thoughts." However, 1 have suggested that we distinguish between Object-Independent Thoughts and Particular Thoughts. It seems to me that often when using "singular thoughts" writers mean "Particular Thoughts," but sometimes they mean "Object-Independent Thoughts." Given this confusion, and given the use of "singular thoughts" in the material I will be discussing, I will italicize "singular thoughts" when appropriate, to indicate what others have meant—i.e., a use outside of the distinction I recommend. 6.1 Blackburn on Evans In the relevant pages of his book Spreading the Word, Simon Blackburn considers the identity conditions of what he calls singular thoughts. The thrust of Blackburn's attack is concentrated upon defending the claim that singular thoughts do not depend upon particular objects for their identity, contrary to Evans (and others, notably John McDowell). I should point out that even though 1 end up doubting the success of Blackburn's criticisms, his discussion does lead us to do two things. First, it requires that we modify, 156 or perhaps simply expand, Evans's account of thought, mostly just to make explicit what it is likely that he already thought. Second, Blackburn's discussion does, at several points, give partial support to the division between Object-Independent Thoughts and Particular Thoughts that I advocate. I will point these out when we come across them. First, a recapitulation of the relevant aspects of Evans's view. As we have seen, Evans had the idea that information-based Particular Thoughts are Russellian, which has the perhaps surprising consequence that when there is no object, there is no thought. In other words, if a speaker was to utter a sentence containing a singular term, in a case where there is no referent (i.e., where the singular term is 'empty'), there is no associated Particular Thought. For instance, consider this case: I say "That cow is big," in a situation in which there is poor light, and what I am perceiving is really a tree stump. The claim is that in this case, I am not in possession of a Particular Thought, even though, as Evans would say, I essay one, as reflected in the expression. This claim strikes many simply as false. If they were to hear my utterance, they would claim to understand exactly what I said, and no doubt claim to be able to identify a corresponding thought in their own minds. That is, all that we need to do, they will say, is look into our own minds and not only see that there a thought there, but also just what the thought is, i.e., its content. The idea seems to be that the thought's content—presumably, that of a cow, somewhere before me, being big—is captured by the understanding of the expression, and this alone. This conclusion is grounded upon the idea, which has strong intuitive appeal, that we have direct access to our own consciousness, and first person authority over what we find there.10 The crux is that it makes no sense to claim that there are object-dependent thoughts (i.e., Particular Thoughts), because thoughts are not individuated by their objects. However, even in the face of this kind of challenge, Evans holds fast to his claim that there are See Blackburn (1984), pp. 323-4 for a brief discussion of this connection between the philosophy of language and mind, and epistemology. See McDowell (1994) for a long discussion. Lecture I gives the gist of his arguments. 157 genuine Particular Thoughts, and that they have a place in a theory of mind (e.g., in explaining action) and language (e.g., explaining reference). Here is a summary of what he says. The central element of Evans's defence comes from his account of the processes involved in the communication of information-based Particular Thoughts, i.e., communication employing use of singular terms. First, we have the claim that in order for a hearer to understand an expression containing a singular term, she must retrieve (assuming she has it) information that is derived from the object to which the speaker refers. The kind of understanding that Evans is talking about requires that the hearer's thought concerning the object be controlled by a conception of the object." The conception of the object is a belief about it that she has because she has received certain information about the world, i.e., that it is thus and so. The belief can be entertained in different ways, though, because we are aware that sometimes we receive w/sinformation from the world. So her belief may be that what thought she entertains is grasped as true, or false, or probable, etc.. But what way she grasps the thought depends on the content of the conception. This is what it means for the conception to be 'controlling'. The thought she grasps, then, is governed by just what information she receives—for instance, that the sky is blue. Of course, as Evans notes, she could entertain the thought (really a speculation) that the sky is green, but in entertaining that thought she would grasp it as false, given the information she has received, i.e., that the sky is blue. We have, then, this idea that in order for a hearer to understand an expression containing a singular term, she must have in her possession information from the right source, where the right source is the object to which the term refers. Evans writes: Thus, if a speaker utters the sentence 'This man is F, making a demonstrative reference to a man in the environment he shares with the hearer, the hearer can understand the remark only if he perceives the man concerned, and, bringing his perceptual information 1 1 See Evans (1982), pp. 12Iff. 158 to bear upon his interpretation of the remark, judges 'This man is F: that's what the speaker is saying'.12 And again: Understanding the kind of use of a referring expression 1 am considering is not a matter of having beliefs with the right sort of content, but rather a matter of having, and using, information from the right source.13 He gives an example of a speaker referring to a bird seen on a hunting trip taken with the hearer some time ago. The speaker says "Do you remember that bird we saw years ago? I wonder whether it was shot." Various things can be said to 'jog' the hearer's memory, but it is not until the information that had its origins in contact with that bird is retrieved by the hearer that reference is successful (i.e., communication is achieved). The important point is that it is this information that leads to the hearer identifying the referent. It is not, by itself, the bare content of the linguistic expression containing the reference to the bird. Nor is it the content of the descriptions used to 'jog' the hearer's memory. Further, here is a point at which it is claimed that a purely linguistic analysis is deficient, so we get an example of the shift of the Linguistic U-Turn. Evans identifies an important point about reference made by Strawson in "Identifying Reference and Truth Values."14 Roughly, the point is that when a speaker refers, there is an assumption of shared (identifying) knowledge, and this knowledge is associated with what is thought to be adequately identifying descriptions. Evans thinks that Strawson's idea suffers from the "general neglect of the concept of information."15 For Strawson, there is the association of knowledge (which is a species of belief) with the descriptions, rather than the idea that uses of referring 1 2 Evans (1982), p. 305. Other than directly perceiving the man, the hearer may recall information from memory, or may receive the appropriate information via the testimony of others (though I think that there is a complication with this case). See the Producer/Consumer distinction as I have applied it to thought in chapter 8. 1 3 Evans (1982), p. 307. Again, the referring terms with which we are concerned are what are being called 'non-descriptive' referring terms. 1 4 Strawson (1971). 1 5 Evans (1982), p. 309. 159 expressions invoke information about an object (i.e., the one being referred to). But there is a problem here, for when it is recognized that the information is incorrect, 'knowledge' must be replaced by 'mere belief. So instead of being able to claim that uses of referring expressions are made with the assumption of shared information (following Evans) we are left with claiming that referring expressions presuppose existential propositions, i.e., supposed identifying beliefs about the object. Evans thinks that this claim is less interesting. That aside, it seems that we do more than presuppose that a hearer has some possibly right identifying belief about an object when we use a referring expression. Following the bird example (section 1.6), when a speaker says "Remember that magnificent bird we saw?," we do not think that the speaker presupposes that the hearer possesses merely an existential (and so object-independent) thought, expressed as something like "There is a magnificent bird we both saw." Possession of such a. thought is not sufficient. The hearer must recall the information he has about the bird, and the speaker must be assuming that he can recall it.16 For Evans, then, understanding a referring term requires possessing information from the object to which the term refers. However, this information, in itself, is not sufficient. A hearer must also have the belief that there is some particular object from which this information comes, and know which object this is, which means that understanding a speaker's utterance just is grasping an information-based Particular Thought. This claim is what rubs many, like Blackburn, the wrong way. Why? The following is Evans's understanding of the position held by his opponents. When it is denied that a singular term is Russellian, it is being maintained that a speaker will have said something by uttering a sentence containing the term, whether or not there is anything to which it refers. And directly following is the summary of the argument that is his response. If this is true, then there is some true proposition such that knowledge of its truth constitutes understanding of the utterance. But when the singular term is information-Perhaps it is even more complicated than this. Probably something analogous to the whole set of Gricean intentions is required. 160 invoking, it seems to me that if there is nothing to which it refers, then we must deny that there is any such proposition. For we have established, by establishing the lemma, that if there is a proposition such that knowledge of its truth constitutes understanding the remark, it is a proposition which commits someone who accepts it as true to the existence of the referent; and ex hypothesi there is no such proposition.17 Blackburn does not accept this conclusion. But let us explicate this argument a little before we see why. First, on one interpretation of "knowledge of its truth," it seems quite clear that given a true proposition, knowledge of its truth could never count as understanding of the utterance that expresses the proposition. Consider this example from Keith Donnellan. I could be assured by a qualified mathematician that a certain mathematical sentence expresses a theorem, without thereby knowing the truth of the theorem expressed, if I am ignorant of what the mathematical sentence means. [And] . . . I can pass on to someone else not only that the sentence in question expresses a truth, but also the truth it expresses.18 Examples like one lead Donnellan to draw a distinction between "knowing that a certain sentence expresses a truth and knowing the truth of what is expressed by the sentence ."' 9 If we mean the former by "knowledge of its truth," then surely, given a true proposition, knowledge of its truth could never count as understanding of the utterance that expresses the proposition. However, as I understand it, Evans means the latter by "knowledge of its truth." We see what Evans means especially in his account of 'understanding', but also in some text that follows the above quotation. In a case where a singular term is information-invoking, but there is nothing to which it refers, Evans says that "the purported 'understanding' is not the knowledge of any truth about what is said."20 The italicized passage indicates that Evans means 'knowledge of the truth of what is expressed' by "knowledge of its truth." 1 7 Evans (1982), p. 330. 1 ft Donnellan (1977), p. 18. Tom Patton directed me to this example. 1 9 Donnellan (1977), p. 18. I have added the italics. 2 0 Evans (1982), p. 331. I have added the italics. 161 Something else needs explanation; Evans grounds his argument in a lemma earlier established. Roughly, the lemma is that in order to understand an utterance containing an information-invoking use of a singular term, one must believe that there is a particular thing to which the speaker is referring. We saw the details of what is involved in the understanding of an information-invoking use of a singular term in Chapter 4, but it may be helpful to see something that helps establish the plausibility of the lemma. Throughout his book, Evans repeatedly reminds us to distinguish between two kinds of reference. The first is what we might call purely descriptive reference. The example most frequently given is an example of what Evans calls a "descriptive name." These are established by a "'reference-fixing' stipulation."21 For instance, recall from the Introduction the 'Boris' example: (1) The hairiest spider in the world is called Boris but where "Boris" operates more like a title than a proper name. (If we were to find a hairier spider than the one currently called Boris, then that (hairier) one would now be called Boris.) Importantly, this use of "Boris" is not information-invoking. The point, then, comes from seeing the difference between understanding an utterance containing a purely descriptive use of a singular term, and an information-invoking use. That is, we can understand (1) without having any belief concerning a particular spider. But this is not the case with an information-invoking use of a singular term. Back to Blackburn. Blackburn rejects Evans's idea that understanding is grasping a thought. Recall, Evans argued that to understand a speaker's expression of a Particular Thought you must believe that there really is an object to which the speaker refers. Blackburn rejects this view of understanding. He gives an example of a speaker referring to an object that the hearer doubts (i.e., does not believe) exists—say, a planet whose existence is surmised by the speaker based on what turns out to be (unbeknownst to both) faulty data. In this case, Blackburn argues, the hearer perfectly well understands what the hearer said. He says, "It is not belief which is necessary for understanding but—well, 2 1 Evans (1982), p. 31. Evans cites Kripke, in Davidson (1972), p. 290-1, as the source of this phrase. 162 understanding: knowing why the speaker is making that remark, and why he is right to choose those words to do so."22 I think that what the above shows is that Blackburn has not fully understood what Evans is saying. It may be true that there is a kind of 'understanding', or perhaps a way in which we use the term, such that it does makes sense to say "I understand what he said when he said 'The planet between Mercury and Venus has a strange orbit'." In essence, this kind of understanding simply involves understanding these English words in this particular grammatical form (in an object-independent way). And given that it does make sense, we have some kind of intuitive support for thinking that this is a legitimate use of "understanding." But this is not the kind of understanding that Evans has in mind. Moreover, the thought (assuming for the moment that it is a thought) that you might associate with 'The planet between Mercury and Venus has a strange orbit' if you did not believe that there was such a planet, is not the kind of thought that Evans is concerned with identifying (even though Evans is not always as clear as he could be here). The thought that you would associate with 'The planet between Mercury and Venus has a strange orbit', if you did not believe that the planet existed, must be something that is not object-dependent. As I have described them, then, it is an Object-Independent Thought, i.e., a form of mental content that is not Representational. Since Object-Independent Thoughts are object-independent, it does not matter that there is no object. But Evans clearly and persistently states that the thoughts for which he is trying to work out the identity conditions, are object-dependent, i.e., Particular Thoughts. Blackburn is probably right to pick up on these other kinds of thoughts (and the type of understanding associated with them, if it is properly labeled 'understanding'), but it is not at all clear that his points affect Evans's argument. I have more to say about the status of these thoughts later. We may be able to see what leads Blackburn astray by noting that he seems to take it that Evans thinks that thoughts are individuated by 'mode of presentation', understood as it commonly is, as the way 2 2 Blackburn (1984), p. 319. 163 in which we think of the object. Now, this is indeed what Frege held (or, at least, is interpreted as having said), and as we have noted, Evans is heavily influenced by Frege. However, again as we have seen, Evans's view either is not purely Fregean in this respect, or it involves an nonstandard interpretation of Frege on this point. First, Evans's view has the influence of Russell's Principle. Second, importantly, Evans has a nonstandard way of understanding Frege on the identity conditions of thoughts. It seems that it is because of these considerations that Evans says that thoughts are to be individuated by 'mode of identification', not 'mode of presentation'. That is, not by the way in which we think of the object, but by the way in which we come to have individuating information (via discriminating knowledge). These are not the same thing given the standard interpretation of'mode of presentation'. Evans says early on in his book, in the chapter on Frege that: We must not discriminate ways of thinking of things so finely that no difference of epistemic attitude can rest upon the discrimination. [And], we must not make our way of discrimination of ways of thinking of objects so coarse that we reckon a subject to be thinking about an object in the same way in two episodes of thinking about it, when it would be perfectly possible for the subject coherently to take different attitudes towards the thoughts thus entertained.23 This point is to be taken in the context of noting that the notion of a way of thinking of an object is "explained in terms of the notion of the account of what makes it the case that a subject's thought is a thought about the object in question."24 As we have seen, Evans goes to a great deal of trouble in arguing for the idea that it is, in essence, discriminating knowledge that makes it true that a thought is about a particular object. Further, and from the point of view of distinguishing between Object-Independent Thoughts and Particular Thoughts, we have to be aware that it is not modes of presentation that individuate thoughts (Particular Thoughts). Rather, it is objective states of the world. It is just that different modes of Evans (1982), p. 20. Evans (1982), p. 20. 164 presentation usually lead one to different objective states. It is this kind of thinking that leads Evans to speak, rather, of modes of identification. Of course, Evans could be right in his interpretation of the Fregean idea of modes of presentation, namely, that they are object-dependent, not Object-Independent as the idea is commonly interpreted. In any case, given Evans's view, and his substantial arguments for this interpretation, critics ought at least to acknowledge that the standard interpretation of 'modes of presentation' is controversial. Blackburn, it seems, gets to this first criticism by not taking account of •Evans's understanding of Frege's notion of a way of thinking of an object. He may respond that Evans is wrong to so interpret Frege, but that would require substantial argument. In any case, this criticism would be beside the point because it is Evans's view, not Evans's understanding of Frege, that is important here. There is another conclusion that can be drawn here, which I think is preferable. As I have just said, it is not clear that Blackburn's arguments count against Evans's position because Evans was focusing on Particular Thoughts (ones individuated by mode of identification). But given Evans's focus, we need not reject all of what Blackburn says that follows from the standard interpretation of the Fregean notion of mode of presentation. What we can do is conclude that things are as Evans says regarding Particular Thoughts. We conclude, further, that there are thoughts which are individuated by mode of presentation, interpreted in the standard way, but that these are Object-Independent Thoughts. In sum, Particular Thoughts are individuated by mode of identification, whereas Object-Independent Thoughts are individuated by mode of presentation. This difference in individuation criteria allows us to accept what seems to be important from both sides. Round two of Blackburn's disagreement with Evans concerns how Evans directly deals with the question: What thought, if any, is communicated by a speaker uttering an expression containing an empty singular term? 165 It is Evans's answer to this question that has troubled some others as well. Both Carruthers and Rozemond defend views (against Evans) that deny the requirement of satisfaction of Russell's Principle for Particular Thought. That is, they deny the requirement of a subject possessing discriminating knowledge of the object thought about. It is just this requirement that would be denied by an advocate of the view that there can be (particular) thinking in an empty world—or, at least, in the absence of an appropriate object. Recall, Evans argues that no thought is communicated by use of an expression containing an empty singular term. Again, attention is focused on the requirements of communication, but this time extended to the requirements for mutual understanding. Blackburn suggests that mutual understanding in communication necessarily requires mutual knowledge of modes of presentation.25 That is, for there to be mutual understanding, it must be the case that both speaker and hearer have, in their minds (so to speak) the same way in which they think of the object with which they are concerned. It is not sufficient that they just be thinking of the same object. They must be thinking of the same object in the same way. Blackburn then says that Evans thinks that just thinking of the same object—i.e., having shared reference26—is sufficient. Further, that Evans thinks that if shared reference is sufficient, then nothing could count as communicating (having mutual knowledge) in cases where there is no object—i.e., there cannot be shared reference since nothing is referred to. That there is no communication in cases where there is no object is indeed roughly what Evans says, though he expresses his ideas somewhat less emphatically than might otherwise be thought by 2 5 Blackburn (1984), p. 320. By "mutual knowledge" I take Blackburn to mean something like what Grice had in mind in setting up the requirements for Nonnatural meaning, i.e.: (i) that the hearer knows that the speaker intends to communicate something; (ii) that the speaker knows that the hearer knows this; (iii) that the hearer knows (ii); etc.. See section 4.3. As we noted, Grice did not go far enough. David Lewis does with his notion of 'common knowledge'. See Lewis (1969). So does Schiffer (1973) in characterizing 'mutual knowledge'. Roughly, A and B have mutual knowledge of X if: (i) both A and B know X; (ii) both A and B know that the other knows X; (iii) both know that (ii); and so on. 2 6 Presumably, given what Evans means by this notion. 166 reading Blackburn's account.27 Also, the kind of communication that Evans argues cannot take place is of a special kind, i.e., "information-invoking, referential communication." He is not arguing that no kind of communication takes place in an empty world. Evans considers a case where two people come to form information-based thoughts of the same object, but where the information is presented to them in different ways—for example, because they see the object from different locations, or one sees it and the other hears it. He argues, then, that because two people could communicate in this case, it must not be the case that mutual knowledge of modes of presentation is necessary. Further, that because this mutual knowledge is not necessary, the only other thing that could be (and so it is sufficient) is mutual knowledge that the information comes from the same source. Evans writes: Communication can surely take place between two people whose identification of a person rests upon their capacity to recognize him, where one recognizes the person by his face, and the other depends rather more upon the voice. But what, apart from the fact that there is a single person they are disposed to identify, could possibly unify these capacities, and show that a communication-allowing 'correspondence' existed between thoughts depending upon them?28 This passage suggests that Evans has a less strict notion of communication than Blackburn, which might, if we note the difference (and have no reason to be troubled by it), permit us to eliminate some ofthe tension between the two views. Each has cases in which it is purported that communication takes place, but where the chosen conditions for communication are satisfied, and the opponent's are not. Of course, these claims depend upon what counts as communication. It is hard to know what to do with these kinds of disputes. But fortunately, I do not think that here is the place to deal with such terminological differences, and the subsequent implications. I move instead to discussing what differences would not be See for example Evans's comments in the final paragraph of section 9.5, and note 37, p. 337. Evans (1982), p. 334. 167 eliminated even if we allowed for differences in conceptions of the notion of communication. This discussion should be sufficient. Blackburn considers, again, the case of the two astronomers speaking to each other (apparently communicating) about a (non-existent) planet. He says that "had their information been veridical, they would have been thinking of the same object."29 His idea is that if the same thing (the same false data, or the same illusion) is the causal origin of the information possessed by both speaker and hearer, then communication takes place if they speak to each other about 'this' (non-existent) planet. This understanding of communication is reflected in the following passage: "In general, if the same illusion or fiction, or cause of hallucination, explains why you and I take ourselves to be talking of the same thing, then we communicate."30 The crux seems to be that it is this 'thing' (false data, illusion) that we would use in explaining the communicative behaviour of both parties. Moreover, if the explanation of their communicative behaviour is the same, and they communicate, then it must be the case that they think the same thoughts, and, importantly, came to think the same thoughts because of the same thing. In other words, cases where there is no object parallel cases where there is an object. In cases where there is an object, "the same object figures in the informational episode which prompts our thinking, and itself provides the explanation of the thinking."31 His idea, then, is that in cases where there is an object, whatever causes the thinking, explains the coordination of thoughts in communication. Because of the parallel, the same goes for cases where there is no object. Again, reference to the cause of the (empty) thought explains the possession of the thought, which, for Blackburn, shows that there is thinking going on in an empty world. Or specifically, in this case, that there is a genuine thought associated with the use of an expression containing an empty singular term. 2 9 Blackburn (1984), p. 320. 3 0 Blackburn (1984), p. 320. 3 1 Blackburn (1984), p. 320. 168 And relating this point back to the topic of this section, the further point is that a subject can be said to be thinking of an object even when he does not have discriminating knowledge of it, because: (i) as Blackburn claims, there is thinking going on; and (ii) there is no discriminating knowledge because there cannot be discriminating knowledge in the absence of a relevant object. The first thing to note is that if Blackburn's really is an example of genuine communication, then it is of an odd sort. Evans might say 'How can it be that the two are referring to the same thing, and by extension, thinking of the same thing, when there is nothing there?' Of course, Blackburn's response probably would be that there is some thing, namely, whatever is the cause of the two astronomers coming to believe, or perhaps more accurately, suppose, that there is a planet at some particular coordinates. Perhaps this thing is some false data from a malfunctioning sensor. This response would be, at the very least, a rather odd thing to say because the false data is not to what we (and they, presumably) would think that they are referring. We would take it, and they would take it, I think, that they are referring to, and thinking about a planet, not some (false) data. The appearance of the (false) data may explain why they come to be in a particular mental state. And with respect to this state, I think that Blackburn is correct.32 But as I have been trying to show, explaining why someone is in a particular mental state (has a certain thought content) is not exactly the same thing as explaining what thought is grasped by a subject. The astronomers example shows this point. Part of the explanation of why the astronomers think that there is a planet in some particular location involves the false data from a malfunctioning sensor, along with some story about associated, background beliefs. However, the thought grasped, even though it has false data as its cause, is not about false data. Of course, strictly speaking, the thought is not about anything, but it has the form of being about a planet. We have to admit, though, that there is something to what Blackburn says here. He gives another version of the astronomers example, but where the two come to have thoughts about 'the planet', but via 32 Others too, most notably Jerry Fodor; and in a different way, Fred Dretske. 169 different causes. In this case there is no single thing that is the cause of their thoughts, and so there is no communication. His idea is that "Communication requires a correspondence between whatever it is which explains my saying what I do, and whatever it is which the hearer takes to explain those sayings."33 In this case, when it was discovered by one of them that there was no planet, and he showed the other why he came to have this false supposition—say, by showing him the malfunctioning sensor and how it caused the false data—the other might still suppose that the planet is where he thinks. The first astronomer, and perhaps the second, would then realize that they were not talking about the same thing. And further, they would realize that they had not communicated as they had thought, which would not be the case with the first version of the example. Once the error was discovered, one of them might still say something like "Well, there is no planet there, but I see that there was the same cause for our thinking that there was." A second thing, which is related to the first, is that it seems that Blackburn is concerned here with a kind of thought with which Evans is not concerned—or put another way, that he means something different by "thought" than what Evans does in the relevant passages. So again, it is not clear how his arguments affect Evans's view. As we have noted, the point to pick up on in Blackburn's argument is that he is looking for what explains the subject's psychology. However, the kind of explanation he is directing our attention towards is an explanation that requires only recourse to certain kinds of thoughts as subjective mental states (i.e., Object-Independent Thoughts). Typically, it is held that it can only be thoughts of this kind that are relevant in this kind of explanation of behaviour to which Blackburn alluding. That is, no external properties can be relevant in explaining behaviour. There are two things to say here. First, as we will see later, there is another way of understanding 'explanation of behaviour' where it is not the case that no external properties are relevant in explaining behaviour. Recourse to particulars becomes necessary, and so explanations of behaviour require recourse to objective states. If nothing else, this alternative view suggests that the dispute is not going to be 3 3 Blackburn (1984), p. 321. 170 resolved by looking at explanations of behaviour. Second, Evans, as we have often noted, is concerned with Particular Thoughts. It is no surprise that Blackburn can argue as he does about Object-Independent Thoughts because for them, by their very nature, the existence of the 'objects' does not matter. Blackburn talks about the common cause of speaker's and hearer's thoughts "cementing our thoughts together."34 So he is acknowledging that object-dependent properties have some role to play in our cognitive lives. However, this is just to the extent where we could say that we understand why it is that they figure in what appear to be corresponding mental states. He also acknowledges the influences that lead Evans to his view of Particular Thought. But he characterizes them in a way that allows his conclusions. He says that he acknowledges a "weak semantic thesis" that allows us to say that we can describe thoughts in different ways, based upon different external features (of subjects' local environments). Everyone acknowledges that there can be different "external relations" between the thoughts two subjects have, and the objects about which they think. However, whereas Evans (and others) would say that these external relations lead us to say that they have (or grasp) different thoughts, Blackburn says, via his weak semantic thesis, that all that this shows is that the same thought (-type) can be described in different ways—according to the different external relations. However, Blackburn's position does not conflict with Evans's. All that it does is restate the dispute in different terms. However, again, what this discussion shows is the failure to recognize the different uses being made of the word "thought." Blackburn means Object-Independent Thoughts, whereas Evans means Particular Thoughts. What Blackburn's arguments do achieve, is to force us to acknowledge, as I have been recommending all along, the existence of a category of thoughts—what he might, in the relevant discussion, call 'Cartesian thoughts'35—that are not admitted by staunch Externalists. However, even in the face of admitting these thoughts, I think that when we understand what 3 4 Blackburn (1984), p. 322. 3 5 Blackburn (1984), p. 324. 171 Evans means by "thoughts" and communication, we are led to thinking that Blackburn's arguments do not affect what Evans argues. 6.2 Carruthers on Evans We see in Carruthers's paper something that seems to lie somewhere near the heart of the dispute between those like Evans, with the Russellian view of Particular Thought, and those who think that such thoughts cannot be Russellian. Carruthers suggests that there are two kinds of interest that we have in thoughts and the descriptions of contents of thoughts. He calls these "the explanatory" and "the communicative (or 'belief-acquisitive') interests."36 He argues that it is the former perspective that is "conceptually tied to the notion of thought-content."37 However, Evans would have held that it is the latter perspective that is 'conceptually tied' to thought-contents because, for Evans, communication is object-dependent. In Carruthers's words, for Evans, "it is knowledge of which object the thought concerns that is crucial."38 We can see the same point in a related way. Evans, as we have seen, argued for a truth conditions theory of the contents of thoughts—that is, that a thought is individuated (at least in part) by what makes it true—which is why (information-based) Particular Thoughts are object-dependent. Thoughts are individuated (in part) by particular objects, and their truth is determined by this 'mode of individuation'. Again, if there is no object, then there is nothing to individuate the thought, so there is no Particular Thought. Carruthers (1987), p. 26. I take it that this division in interests is suggestive of the division I place between Object-Independent Thoughts and Particular Thoughts. That is, it might be thought that Object-Independent Thoughts are what are explanatorily relevant, whereas Particular Thoughts are relevant in determining what is communicated (or grasped). 3 7 Carruthers (1987), p. 26. 3 8 Carruthers (1987), p. 26. Unless I miss something in his account, here, it is odd to see this statement given his earlier endorsement of Blackburn's arguments against Evans. Perhaps he disagrees with Blackburn on this point. Or perhaps he means to attribute the view to Evans. 172 Carruthers defends what he calls 'weak methodological solipsism'. Roughly, this view is that both 'strong methodological solipsism' and 'Russellianism' are false, but a position somewhere in between is true. Strong methodological solipsism is the standard, well known view in which it is taken that thoughts can be explained without reference to anything that is external to the subject's (conscious) mind—including physical states of the subject's body, as well as objects and states of affairs in the world. In Carruthers's words, weak methodological solipsism is just the view "that it is possible to describe the content of any given singular thought, expressing that which makes this singular thought distinct from other thoughts of the same type, without referring to any individual physical object."39 Put in what are now, I hope, familiar terms, it is the view that (given the rejection of strong methodological solipsism, presumably) a thought is still 'contentful' even when the 'object' of the thought does not exist. This view will strike some as trying to bring together incompatible bits from incompatible views. But Carruthers has a defence I will now outline briefly. Carruthers notes that typical Twin Earth considerations are going to be thought to count against his view. Twin Earth thought experiments are set up just so that nothing is different between Earth and Twin Earth except for the identity of the particular objects (or kinds of substances) of concern.40 The considerations that follow from these ideas have led many into thinking that meanings, and by extension, thoughts, are 'not in the head'. In other words, that the contents of thoughts (and meanings) are individuated by particular objects, substance or some other relevant feature of the context to which relevant subjects are indexed. It follows, it is claimed, that because a subject and her twin think about 3 9 Carruthers (1987), p. 29. 4 0 This is the way it is typically put, but it is not quite right. There is a further difference between Earth and Twin Earth, and this is that they occupy different locations in this possible universe where there are both worlds. (See Putnam (1975), p. 223.) This difference will become important later when we look at indexical identification. For the moment, though, it is important to note that Twin Earth is not a possible world, i.e., a counterfactual to actual Earth. Such a situation would resu