Return to Carruthers' books 




(Cambridge University Press, 1996.)

ISBN 0-521-48158-9

ISBN 0-521-63999-9 (pbk.)




To what extent, if at all, is language involved in thinking? Do we think in natural language? Or is language but an input and output module for central cognition – a mere communication device, in fact? Peter Carruthers here makes a fresh and provocative contribution to an ancient debate, and claims, in contrast to much recent work in philosophy and cognitive science, that the independence of thought from language is by no means clear-cut. He argues against the communicative conception of language put forward by thinkers such as Fodor and Pinker, and espouses the cognitive conception of language proposed by, for example, Wittgenstein and Dennett; but he refuses to take on board the Whorfian linguistic relativism so often seen as a natural corollary of this position. Instead he argues persuasively for the view that the cognitive conception of language can be deployed in combination with a modularist and nativist view of language and the mind. The book defends a new theory of consciousness which is put to work in the final chapter to argue for the necessary involvement of natural language in human conscious thinking.


Language, Thought and Consciousness will form essential reading for all those interested in the nature and significance of natural language, whether they come from philosophy, psychology, or linguistics.


‘If Carruthers is right, the study of language is – in a very real sense – the study of thought. This has two enormously important consequences. First, it puts linguistics centre stage in cognitive science; second it sanctions the linguistic turn which has so much dominated twentieth-century philosophy. Carruthers’ thesis is thus a very significant one. What impresses most, though, is not the significance of the thesis but its defence. Language, thought and consciousness is a book crammed with penetrating observations and good arguments. Few readers will fail to learn something from it, and none will be disappointed.’

Times Literary Supplement








1          The geography of the issues

            1.1       Conceptual versus natural modality

            1.2       Implications for philosophy and psychology

            1.3       Of wolves and wolf-children

            1.4       Stalnaker's intelligent Martians

            1.5       Anti-realist arguments

            1.6       Realism in mind

            1.7       Innateness and theory of mind

            1.8       Thinking: images or sentences?


2          Which language do we think with?

            2.1       The evidence from scientific psychology

            2.2       The evidence of introspection: images and imaged sentences

            2.3       The scope and strength of the introspective thesis

            2.4       Objections and elucidations

            2.5       Fallible introspection and Fodor

            2.6       Individuating propositional attitudes

            2.7       Animals and infants

            2.8       Language-learning and sub-personal thought


3          Thought-based semantics

            3.1       The argument from foreign believers

            3.2       Grice's thought-based semantics

            3.3       Two objections

            3.4       Searle's version of thought-based semantics

            3.5       A marriage of Searle and Fodor?

            3.6       Causal co-variance theories

            3.7       Misrepresentation, and asymmetric causal dependence

            3.8       The all Ss problem


4          Holism and language

            4.1       From mental realism to Mentalese

            4.2       The demand for scientific vindication

            4.3       The problem of holism

            4.4       Between holism and atomism

            4.5       Arguments for holism

            4.6       The need for a language-based semantics

            4.7       Language-based semantics 1: functional-role semantics

            4.8       Language-based semantics 2: canonical acceptance conditions


5          First steps towards a theory of consciousness

            5.1       Retrospect: the need for a theory of consciousness

            5.2       Conscious versus non-conscious mental states

            5.3       Cartesian consciousness

            5.4       Why Cartesianism won't do

            5.5       What kind of theory do we want?

            5.6       Kirk: presence to central decision making

            5.7       Higher-order discrimination and feel

            5.8       The case for higher-order thought theory


6          Second (-order) steps towards a theory of consciousness

            6.1       Theory 1: actual and conscious

            6.2       Theory 2: actual and non-conscious

            6.3       Theory 3: potential and non-conscious

            6.4       Theory 4: potential and conscious

            6.5       Dennett 1978: availability to print-out

            6.6       Dennett 1991: multiple drafts and probes

            6.7       Time and indeterminacy

            6.8       Dennett on the place of language in thought


7          A reflexive thinking theory of consciousness

            7.1       Reflexive thinking theory

            7.2       Comparisons and advantages

            7.3       Conscious versus non-conscious thinking

            7.4       Objections and elucidations

            7.5       The problem of unity

            7.6       The problem of phenomenal feel

            7.7       A Cartesian Theatre?

            7.8       Animals and infants revisited


8          The involvement of language in conscious thinking

            8.1       An architecture for human thinking

            8.2       An evolutionary story

            8.3       The argument from introspection revisited

            8.4       Working memory and the central executive

            8.5       The thesis of natural necessity (weak)

            8.6       Objections and elucidations

            8.7       The thesis natural necessity (strong)

            8.8       The scope and significance of NN










This book began life as a series of lectures written for the third Sino-British summer-school of philosophy, on philosophy of mind and cognitive science, held in Tianjin, China, between July 27th and August 14th, 1992. I am grateful to the Chairman of the British Committee of the summer-school, Dr Nicholas Bunnin, for extending me the invitation. This provided me with a valuable opportunity to develop, and commit to paper, my ideas on the inter-relationships between language, thought, and consciousness. I am also grateful to our Chinese hosts from the Institute of Philosophy, Chinese Academy of Social Sciences, for making the experience such a pleasant and intellectually fruitful one. I hope that, partly as a result of my efforts, those who attended the summer-school went away fired with enthusiasm for philosophical studies of the nature of human cognition.

            Since then I have received help, of various sorts, from a great many people. I should especially like to thank my students at the University of Sheffield, both undergraduate and postgraduate, who have subjected my ideas to relentless and penetrating criticism over a number of years – mentioning particularly Chris Bowns, Gavin Boyce, Nick Creak, Duncan Cromarty, Kate Distin, Keith Frankish, Gillian Hind, Maxine Holdsworth, Tim Howe, James Kinch, Ewan McEachran, Stuart McWilliam, Simon Tomlinson, Mark Vale, and Clive Witcomb.

            Thanks also go to the following individuals for their advice and/or comments on earlier drafts: David Archard, Judith Ayling, Alan Baddeley, David Bell, George Botterill, Jill Boucher, Jeremy Butterfield, David Chalmers, Jim Edwards, Rosanna Keefe, Bob Kirk, Stephen Laurence, Susan Levi, Bill Lyons, Stephen Makin, Adam Morton, Shaun Nichols, David Owens, Josef Perner, Tom Pink, Gabriel Segal, Barry Smith, Neil Smith, Peter Smith, Bob Stern, Julia Tanney, Ianthi-Maria Tsimpli, and Rosemary Varley.

            I also benefited from the comments of an anonymous referee for Cambridge University Press, and from discussions with colleagues and students at seminars in Manchester, MIT, Oxford, Sheffield, and Cambridge.


What is philosophy? The main character and direction of my project are described in some detail in the Introduction and the opening chapter, and so do not need to be reiterated here. But I should like to say something very briefly about the way in which I understand the nature and methods of philosophy. Besides having some intrinsic interest, these remarks may help to orient the reader with respect to the arguments that follow.

            It is often commented that there are two very different species of philosophy widely practised within the English-speaking world. (I ignore the small minority who attempt to do philosophy in Continental mode – they will not, in any case, be reading this.) One of these conceives of philosophy as broadly continuous with science, takes the goal of philosophy to be truth about a wide variety of subject-matters, and is prepared to make use of a posteriori inferences to the best explanation in pursuit of this goal. In contrast, the other species of philosophy sees itself as sharply distinct from science, takes the distinctive goal of philosophy to be conceptual truth, and insists that the methods of philosophy must be a priori conceptual analysis or argument. Let us label these the substantive and the analytical conceptions of philosophy respectively.

            Substantive philosophy is more common in America, and appears to be heavily indebted to Quine, particularly his (1951) denial of the analytic/synthetic distinction. Analytical philosophy is more common in Britain, stemming partly from the influence of Wittgenstein (1953), who believed that the sole business of the philosopher is to gain a clear view of the inter-relations amongst our concepts. It might then seem that the kernel of the disagreement between the two conceptions concerns the legitimacy of a category of analytic truth. For substantive philosophy, in denying that there is any such category, must deny that there is any sharp distinction between philosophical and scientific truth. Whereas analytical philosophy, in accepting that category, can insist that the class of analytic truths constitutes the proper domain of the philosopher as opposed to the scientist. This way of viewing the disagreement between the two conceptions is mistaken, however, as I shall now try to explain.

            The crucial point is that there is nothing to force a philosopher who accepts the existence of concepts and conceptual connections to believe that philosophy should be concerned only with such connections. Someone can accept the category of analytic truth, and so find a place for a priori conceptual analysis and discovery within philosophy, while maintaining that philosophers can also be concerned with substantive issues of fact, and can legitimately employ a posteriori inferences to the best explanation in attempting to resolve those issues. Indeed, I hold just such a combination of views myself. As will be seen in Chapter 4, I believe that the reasons for rejecting the existence of concepts and conceptual truths are not very powerful. But I also believe that philosophy can, and should, be substantive.

            I think we can distinguish between two distinct strands within analytical philosophy, corresponding to two distinct sources of motivation. For what really pushes people towards a narrowly analytical conception of philosophy, I believe, is not merely the acceptance of the analytic/synthetic distinction, but either, or both, of the following: a certain kind of paranoia about the extent of the domain of philosophy, and a foundationalist position in the theory of knowledge. Let me take each of these sources of motivation in turn.

            In the beginning, of course, philosophy was simply the pursuit of knowledge, and no distinction was drawn between philosophy and science. But with the passage of time the various special sciences – physics, chemistry, biology, and, now, psychology – grew up and became independent of their parent, leaving the subject-matter remaining to philosophy correspondingly diminished. This naturally gave rise to an anxiety, in the minds of some, that the domain of philosophy might one day shrink to zero, as science progressed still further. And it then seemed imperative to define a subject-matter proper to philosophy that would be forever safe from the encroachment of science. This was to be the domain of analytic as opposed to synthetic truth.

            But in fact this anxiety was misplaced. For it should have been plain that there will always be substantive questions for philosophers to answer that are not scientific ones. This must be so, if only because those questions are about science, such as the question whether science is successful in obtaining real knowledge for us. Moreover there are, arguably, many domains of belief which are not, and do not purport to be, scientific, but which nevertheless give rise to philosophical problems. These include ethics, aesthetics, common-sense physics, and many others. One such body of beliefs, in my view, is that of common-sense beliefs about the mind (so called folk-psychology), as will be seen in the chapters that follow. And it should always have been obvious that the question of the relationship between these beliefs and those of science cannot itself be a scientific one.

            The second reason why many have been tempted by a narrowly analytical conception of philosophy relates not so much to the latter's distinctive subject matter, as to its methods. The line of thought goes something like this. We begin with the idea that the business of philosophy is to obtain for us genuine knowledge (or perhaps, on some conceptions of knowledge, to obtain the knowledge that we have such knowledge – see my 1992a, ch. 5). We then add to this a foundationalist conception of the architecture of knowledge, according to which all knowledge must be grounded in beliefs that are certainly true by means of principles of inference that are also certain. And then it follows that only deduction can be legitimate in primary philosophy. If other methods of inference – including inference to the best explanation – are to be employed at all, then they must first be vindicated by means of deductive argument from premises that are certainly true.

            But the mistake in this argument, in my view, lies in the foundationalist premiss. Rather, the best account of the architecture of our knowledge is coherentist. According to this view, a belief comes to be justified by forming part of an explanatory network of such beliefs, which collectively provide the simplest, most coherent, explanation for the course of our experience. From such a perspective, inference to the best explanation can play just as fundamental a role in the construction of appropriately coherent sets of belief as does deduction. There is then no reason why philosophers should not, at least tentatively, avail themselves of such modes of inference in advance of attempting to vindicate them. Indeed, there is some reason to think, contrariwise, that even deduction must rely, tacitly, upon an inference to the best explanation – if only on an inference from the fact that an argument seems valid to us to the conclusion that it is so – in which case there is no philosophy which can be done without employing explanatory inferences.

            In the chapters that follow, then, I shall be concerned with questions that are substantive as well as with those that are more narrowly analytic. Indeed, many of these questions might equally be raised, from a somewhat different perspective, by scientific psychologists. There is a good reason for this. For it is particularly common in periods of scientific revolution (in Kuhn's sense; see his 1962) that scientists should need to consider the sorts of larger 'framework' questions distinctively raised by philosophers. (For example, Einstein engaged in extensive reflection on the nature of space and time, in the period when he was developing the theory of relativity, which might have seemed familiar to Locke, or to Leibniz, or to Kant.)

            Arguably, psychology is in just such a state of revolution and self-analysis. For the basic nature and direction of psychological enquiry are currently topics of intense psychological debate; and one of the strands in this debate is – or should be – the role to be accorded to natural language in our best model of human cognition. Moreover, since the resolution of this latter issue is currently empirically undetermined (as I shall argue in Chapter 2), it is crucial to stake out and explore the implications of the different frameworks within which more detailed investigations may take place. That is what I have tried to do in this book, in connection with the presently-unfashionable hypothesis that much of human thought is conducted in natural language. So my hope is that the ideas defended here will be of interest to psychologists and other cognitive scientists, just as much as to philosophers. But this must be for them to judge.





The topic of this book is the subject of an ancient debate  whether thought is independent of language, or whether our thinking, on the contrary, necessarily requires or involves natural language. I shall be arguing for a version of the latter thesis. My view is that much of human conscious thinking is, necessarily (given the way in which human cognition is structured), conducted in the medium of natural language sentences.

            I do not, however, make any claim to have demonstrated the truth of this view beyond all reasonable doubt. For the arguments that I provide in its support are broadly empirical ones, involving inferences to the best explanation of a range of phenomena. They are therefore vulnerable to counter-attack from those who can provide further recalcitrant data, and may reasonably be rejected by anyone who can provide a better explanation of the phenomena in question. What I do hope that I have shown, however, is that the case for the independence of thought from language is by no means as clear-cut as many philosophers and cognitive scientists have assumed. And I therefore hope to have provided encouragement for further work to be conducted in the alternative, language-involving, paradigm. Roughly speaking, the overall message of the book has the form, 'Hey! Let's have a look over here! There's a view which hasn't been taken as seriously as it should be lately.'

            One question at issue in this debate is our understanding of the nature and function of natural language. If thought is independent of such language, then language itself becomes only a medium for the communication of thoughts. I shall refer to this theory of the role and significance of natural language as the communicative conception of language. According to the communicative conception, the function and purpose of natural language is to facilitate communication and not (except indirectly, by enabling the acquisition of new beliefs) to facilitate thinking. Language thus functions wholly in the public  inter-personal  domain, rather than in the domain of individual cognition. Language will still have to be represented and processed within the cognition of each individual, of course. But such processing will exist only to support the public functions of language  namely, the exchange of information and the inter-personal co-ordination of action  rather than having a direct executive role in the thinking and practical reasoning of the individual subject.

            The communicative conception of language has been widely endorsed in the history of philosophy, by figures such as John Locke (1690), Bertrand Russell (1921), Paul Grice (1957 and 1969) and David Lewis (1969). It is also the standard model for those now working in cognitive science, who view language as an isolable, and largely isolated, module of the mind, which is both innately structured and specialised for the interpretation and construction of natural language sentences. See, for example, Jerry Fodor (1978, 1983 and 1987), Noam Chomsky (1988), Willem Levelt (1989), and Steven Pinker (1994).

            If, on the other hand, natural language is constitutively involved in our conscious thinkings (as I shall argue), then language is itself the primary medium of such thought, and much such thinking is essentially linguistic. I shall refer to this as the cognitive conception of language, since it accords a central place to natural language within our cognition. On this account we often think in language, and the trains of reasoning which lead up to many of our decisions and actions will consist in sequences of natural language sentences. Language thus has an intra-personal cognitive function, as well as having its obvious inter-personal uses. Here the picture of communication through language is quite different. When a speaker utters a sentence, on this view, their utterance expresses a thought by constituting it, not by encoding or signalling it. A hearer who is a competent user of the same language will then understand that utterance in virtue of it constitutively expressing, for them, the very same (or a sufficiently similar) thought.

            The cognitive conception of language has been endorsed by such disparate figures as Ludwig Wittgenstein (1921 and 1953), Lev Vygotsky (1934), Benjamin Lee Whorf (1956), Daniel Dennett (1991), and also sometimes Noam Chomsky (1976 – at least, that is how I interpret Chapter 2 of that work; but Chomsky has since disavowed any such view in personal correspondence). Most often (but not in Chomsky's case) it has been associated with a radical empiricism about the mind, according to which many human concepts and ways of thinking, and indeed much of the very structure of the human mind itself, are acquired by the young child from adults when it learns its native language  these concepts and structures differing widely depending upon the structure and conceptual resources of the natural language in question. This mind-structuring and social-relativist view of language is still dominant in the social sciences, following the writings early in this century of the amateur linguist Benjamin Lee Whorf (many of whose papers have been collected together in his 1956)  indeed, Steven Pinker (1994) refers to it disparagingly as 'the Standard Social Science Model'.

            My own diagnosis of what has happened in the cognitive sciences in recent decades is this. Researchers have become increasingly convinced, by neuropsychological and other evidence, that the mind is more or less modular in structure, built up out of isolable, and largely isolated, components (see Fodor, 1983, Sachs, 1985, and Shallice, 1988). They have also become convinced that the structure and contents of the mind are substantially innate (see Fodor, 1981 and 1983, and Carey, 1985), and that language is one such isolable and largely innate module (see Fodor, 1983, and Chomsky, 1988). There has then been, amongst cognitive scientists, a near-universal reaction against the cognitive conception of language, by running it together with the Whorfian hypothesis. Most researchers have assumed, without argument, that if they were to accept any form of cognitive conception of language, then that would commit them to Whorfian linguistic relativism and radical empiricism, and would hence be inconsistent with their well-founded beliefs in modularity and nativism (see Pinker, 1994).

            It is important to see, however, that someone endorsing the cognitive conception of language does not have to regard language and the mind as cultural constructs, either socially determined or culturally relative. In fact, the cognitive conception of language can equally well be deployed along with a modularist and nativist view of language and mind  or so, at least, it will be one of the main burdens of this book to argue. A large part of my task is to show that there is a position intermediate between the communicative conception of language on the one hand, and Whorfian relativism (the Standard Social Science Model) on the other, which deserves the attention of philosophers and cognitive scientists alike. I hope to show that there is a real possibility here which should be investigated further  one which is nativist as opposed to empiricist about language and much of the structure of the mind, but which nevertheless holds that language is constitutively employed in many of our conscious thoughts. More than this, of course, I hope to convince the reader that this combination of views is not only possible, but plausible.

            In fact the issues before us are ones which have profound methodological implications for both philosophy and psychology, as we shall see in more detail in section 1.2 below. If thought is independent of language, then the philosophy of language has no right to claim the sort of foundational position within philosophy as a whole which it has been accorded through much of the twentieth century. On the contrary, it should be the philosophy of mind – more narrowly, the philosophy of thought – which is more basic. It will also follow that the study of the cognitive mechanisms involved in the acquisition and use of natural language should be accorded no more central position within psychology than the study of any other mental faculty, such as vision or memory. If thought requires or involves natural language, on the other hand, then many philosophical questions will be expressed most appropriately in linguistic mode, and it will follow that psychologists engaged in the study of natural language are examining one of the basic mechanisms of human cognition. It should hardly need saying, therefore, that our main question is an important one.

            Much of this book has the form of an extended debate with Jerry Fodor. While I share many of my premises with him, as will be seen from the latter half of Chapter 1, I disagree in my conclusions, particularly in relation to the role of natural language in cognition. Fodor's view is that language is but an input and output module to central cognition (though perhaps drawing on a centrally-stored lexicon and data-base), not implicated in the central processes of thinking and reasoning themselves (see his 1978 and 1983). These latter processes are held to involve sentence-like structures, to be sure, but these are not sentences of any natural language, but rather of an innate, universal, symbolic system, which Fodor calls 'Mentalese'. I shall begin to discuss Fodor's arguments for this view in the later sections of Chapter 2. Their further elaboration and critique is then distributed over many of the remaining chapters. I shall be arguing that the case for the communicative conception of language is not proven, and that one can share Fodor's nativism while endorsing the cognitive conception instead.

            Plainly, the issues before us are ones that must straddle both the philosophy of mind and the philosophy of language. For one part of what is in question is the best account that can be given of the nature of thought, another being the best account that can be given of the character and semantics of natural language. If some version of the communicative conception of language is correct, then it must be possible to provide a semantics for the expressions of natural language in terms of prior notions of thought and intention. It must also be possible to provide a semantics for thought, in turn, without re-introducing natural language into the story. These possibilities will be considered in Chapter 3, where difficulties will be discovered for each.

            If, on the other hand, any version of the cognitive conception of language is to be defensible, then it must either be possible to provide a semantics for language without presupposing thought, or it must at least be possible to provide a semantics for language which would leave language and thought on a par, without either being prior to the other. This will be investigated in Chapter 4, where two distinct options will be canvassed. I should emphasise that my main aim in these two chapters is not to establish any definite conclusions, but only to argue that options are left open. Their purpose is to show that no conclusive case has yet been made out, in the domain of semantic theory, for the independence of thought from language.

            Almost the whole of the second half of this book is concerned, in one way or another, with the nature of consciousness. This may seem surprising in a book about the relations between language and thought, and requires some elucidation. In fact one of the main messages of the book is that questions to do with the inter-relations of language and thought are, somewhat unexpectedly, closely interwoven with questions about the nature of human consciousness. Part of the explanation of the connection is that, for reasons which will begin to emerge in Chapter 1, my thesis is especially that it is our conscious thoughts which involve natural language, rather than that they all do. If any such thesis is to be made out, then enough has to be said about the nature of consciousness to make it seem plausible. But in the current climate, in order to say enough about consciousness one has to say quite a lot.

            Chapters 5 through 7 are devoted to arguing for a particular version of higher-order thought theory, which I call 'the reflexive thinking theory' of consciousness, contrasting it with relevant alternatives. (Higher-order thoughts are thoughts which are about other mental states. For example, the thought that I have just been thinking about Avignon is a higher-order one.) This theory is then put to work, in Chapter 8, in the form of a proposed architecture for human cognition which links together conscious thinking with the deployment of natural language sentences. The reflexive thinking theory also has, as one of its more surprising and controversial consequences, that the thoughts and experiences of most non-human animals and (probably) of human infants are non-conscious ones. But this is no mere accidental spin-off from the theory. On the contrary, it is a crucial component of the argument against Fodor, as will begin to emerge in the opening chapter.

            Another reason for the extensive discussion of consciousness in the latter half of this book, is that the case for saying that conscious thinking involves natural language is partly grounded in introspection, as I explain in the early sections of Chapter 2. This introspective data will need to be explained, or explained away, by anyone who wishes to insist that all thinking takes place in Mentalese. To assess the prospects for providing such an explanation, some plausible candidates for the nature of introspective knowledge have to be laid on the table, as I do in Chapters 6 and 7. In fact, I shall go on to argue in Chapter 8 that it is a deep presupposition of our belief that we do sometimes entertain (propositional, as opposed to imagistic) thoughts which are conscious, that natural language sentences are constitutive of those thoughts.

            The minimal (because a-modal) thesis that I want to defend in Chapter 8, relates only to some token thoughts of ours (that is, to thoughts as particular occurrences, or mental events, such as me thinking to myself right now that grass is green), rather than to any of our thoughts considered as types (such as the thought that grass is green, which may be entertained on many different occasions, and by different people). The thesis is, that many human thought-tokens (specifically, those conscious human thoughts which are, in fact, tokened in the form of inner speech) are constituted by natural language sentences. So there are at least some token thoughts of ours (specifically, conscious verbalised thoughts) which do, as a matter of fact, constitutively involve natural language. This is already enough to vindicate the cognitive conception of language, and to falsify the communicative conception.

            Building on this, the stronger thesis to be defended in Chapter 8 is that there are some thought-types of ours (specifically, those thought-types tokens of which are, as a matter of fact, constituted by language in inner speech) which can only be tokened consciously, in us, in language. (This is the thesis I shall call 'NNw'  a weak form of natural necessity thesis, establishing a modal version of the cognitive conception of language.) According to this thesis, many thought-types are such that their conscious tokenings in us are necessarily language-involving, even if thoughts of those types can also be tokened non-consciously without language. So human beings who lacked a language would be incapable of thinking thoughts of these types consciously.

            The strongest thesis I want to defend in Chapter 8, is then that many (though perhaps not all) of these conscious verbalised thoughts belong to types which can only be entertained at all, by us, (whether consciously or non-consciously) in virtue of their expression in language. So there are some thought-types which, for us at least, constitutively involve natural language. (This is the thesis I shall call 'NNs'  a strong form of natural necessity thesis, establishing a version of the cognitive conception of language which is modal in relation to certain thought-types.) On this view, then, there would be many thought-types which could not be entertained at all by human beings who lacked a natural language.

            In fact this last thesis will be left crucially vague, since I shall make little attempt to specify exactly which kinds of thought-types constitutively involve natural language, and which do not. Such vagueness is unavoidable, however, since I have neither the space nor the expertise to review and gather the vast extent of empirical data necessary to make it more precise. This would include a range of developmental evidence relating to the acquisition of language and the increasing sophistication of children's thought, including the evidence from a variety of language-related developmental disorders such as Williams' syndrome and autism. It would also include the evidence of thought of various kinds in the case of people who are profoundly deaf, but have not yet learned to sign properly. And of course there is also the evidence from cases of global aphasia following brain damage, relating to the kinds of thought which may still be possible for those who have lost their capacity to use and to understand language.

            Note that a vague conclusion can still be an interesting one, however. If I tell you that the British Prime Minister will resign sometime next year, for example, then this is certainly vague in relation to the precise date on which a letter of resignation will be handed to the Monarch; but it may still be well worth knowing. And often a vaguely formulated item of knowledge can be a crucial step in the search for a precise solution to a problem. It would be highly relevant to the police engaged in a murder-hunt to learn that the murderer resides in a particular town, or a particular street, for example. Similarly, then, in respect of thesis NNs: if I am right, then cognitive scientists should now begin investigating a particular – vaguely circumscribed – region of logical space, hitherto largely ignored.

            Although I am constitutionally inclined towards megalomania, I have tried to resist the impulse to announce that I have solved all of the outstanding problems of cognitive science in the space of just one book. In fact, all that I can really claim to have done, is to have provided some non-conclusive reasons for taking seriously a particular, vaguely formulated, view of the relations that obtain between natural language, human thought, and human consciousness. But that, surely, is enough.