Download as RTF document (49KB)

Battle for the Mind:
Jerry Fodor, Howard Gardner, John Searle

Trevor Pateman

Abstract: An introduction to cognitive science/cognitive psychology presented through a review of the work of J A Fodor, Howard Gardner and John Searle. The writing dates from 1984. Topics treated include the idea of the modularity of mind (as opposed to the idea of general purpose intelligence); Nature vs Nurture controversies; the positions of Chomsky and Piaget in these debates; the importance of naturally occurring deprivation experiments (the worlds of deaf and blind children).

Works cited in this essay can be found listed in the consolidated Bibliography for this and other website essays

Part One

Jerry Fodor, The Modularity of Mind (1983)

Jonathan Miller's B.B.C. TV series States of Mind and the resulting 1983 book of that title introduced a large audience to a new generation of psychologists and philosophers, mainly American, for whom the idea of Mind is not the idea of a ghost in the machine of the Body, to be exorcised by the rituals of behaviourism, but the very thing they are in business to study. At the same time, these students of mind are not the 'humanistic' psychologists for whom psychology is continuous with our everyday ways of talking about people in terms of action, purpose and meaning. Rather, the language of the New Mentalism ('cognitive psychology', 'cognitive science') is continuous with the language of computer science and Artificial Intelligence (A.l.). If you are willing to think of Mind on the analogy of Software and Body as Hardware, you are on the way to getting along with psychologists like Jerry Fodor and philosophers like Daniel Dennett (both interviewed by Jonathan Miller).

On the way, yes, and by some accounts, all of the way but not according to Fodor - and Fodor is influential. For in Fodor's vision of the Mind - one he shares in large measure with his Massachusetts Institute of Technology (M.I.T.) colleague, Noam Chomsky - much of what we are accustomed to think of as mental software, as something we have learnt or been programmed with, is viewed as hardware, something we have been born with. It is something innate, the development of which is not so much shaped by experience as triggered by it, rather as the growth of an acorn is triggered, not shaped, by water and heat. In Fodor's and Chomsky's hands, cognitive psychology changes insensibly into a biology of cognition. At which point, one might well ask three questions:

How can that be the study of Mind?

What could that have to do with computers?

Isn't this just Piaget in new clothes?

The New Mentalism has to do with computers in that, whether learned or innate, the contents of Mind are described in computational terms, in terms of relations to or operations on symbols (or representations - and representations are paradigmatically mental things). So, for example, Fodor reconstructs the idea of having a bellef in terms of being in a computational relation to a representation (a representation of the state of affairs believed to obtain). This doctrine, which Fodor expounds in his collection of essays Representations (1981), is controversial on two counts. First, because it appears to lead straight to the view that there are innate representations, an inference Fodor drew and defended at length in his Language of Thought (1976). Second, because it appears to reduce states of a person (my belief that p) to states of an organism (a relation to a representation p in a mind or brain - and here the slippage from 'mind' to 'brain' is hard to arrest, as philosophers like Charles Taylor have argued).

More interestingly, perhaps, why isn't the Chomsky-Fodor Growth of the Mind doctrine simply Piaget translated into American? The answer to this question became clear in the 1975 encounter between Chomsky and Piaget and their respective allies (including Fodor on Chomsky's side and Seymour Papert on Piaget's) published as Language and Learning, edited by M. Piattelli-Palmarini (Piaffelli-Palmarini 1980). It is that for Chomsky the mind is wholly or partly modular, whereas for Piaget it is not. For Chomsky, language growth (for example) is controlled by a specific modular faculty ('universal grammar' or 'Language Acquisition Device' in Chomsky's terminology; 'bioprogram' in the work of Derek Bickerton to which I shall be referring later on). For Piaget, language development is continuous with other forms of cognitive development and the schemata which govern them. But, then, granted this contrast, the key question becomes this,

What is a Module? - shortly followed by the question,

Can you count and name the modules of the mind?

These are, in fact, the questions Fodor sets out to answer in his short and lively book The Modularity of Mind (1983). Fodor presents a view of the Mind which cuts it three ways. First, there are transducers - roughly the senses which link us to the external world and with the operations of which Fodor is not concerned (partly for reasons spelt out in a paper on 'Methodological Solipsism' in his book Representations). Second and third, there are input systems and central systems, which differ simply in being modular and non-modular systems respectively. Modular systems have three distinguishing characteristics.

First, modules deal with only one kind of cognitive material, they are domain-specific - so the visual input system only computes representations of objects in the visual domain.

Second, in computing a value for what the output of a transducer represents, modules can only draw on a limited range of information encapsulated in the module - each module cannot, or cannot easily, get access to every bit of background information which might determine whether, for instance, this is a dagger I see before me. Informational encapsulation allows modules to operate very fast and in ways inaccessible to our introspection. (Inaccessibility to introspection is generally insisted on as characteristic of the parts of the Mind cognitive psychologists study. Dennett is very clear on this in his book Brainstorms (1979)).

Third, each module is neuroanatomically localized, so that destruction of a modularly specialized part of the brain can result in complete loss of some ability, as in aphasia. In terms of these three characterisitics, Fodor reckons the perceptual systems and language are modular input - systems. Dramatically, and more than half-seriously, Fodor links his account to the doctrines of the founder of phrenology, Franz Joseph Gall; a phrenologically mapped cranium adorns the book's cover.

In contrast and in addition, Fodor argues that in the mind there are also central systems, which are non-modular in respect of each of the three defining characteristics. According to Fodor, central processes of thinking and belief formation are not (contrary to the trend in A.l.) specialized to particular subject matter, but are quite properly regarded as areas of general intelligence. They can draw on anything the mind contains in order to solve a problem or think about a situation - indeed, their strength is in the capacity to use the kind of analogical reasoning so obviously important in science. Finally, they are not localized anywhere in the brain - roughly, there is not a thought centre in the way there is a speech centre.

Fodor believes that we could have and in some cases almost do have quite good scientific accounts of how the input systems work - vision research, for example, has been a prolifically productive field - but that we don't have and possibly can't have any good account of the central systems, which by the way they have been defined must be bafflingly complicated in their operations. This is why attempts in A.I. to modularise the kinds of activities undertaken by central systems fail and are bound to fail. On Fodor's account, A.l. workers have divided intellectual capacities into quite arbitary sub-departments - proving theorems of elementary logic, pushing blocks around (the work of Terry Winograd), ordering hamburgers (the work of Roger Schank) - all of which misses the 'wholism' of central processes. What emerged from A.l. atomisation of central processes was a picture of the mind that looked rather embarrassingly like a Sears catalogue (p.127). This conclusion is meant to and will provoke controversy. So too will Fodor's inclusion of language among the input systems, against a current trend to treat language processing as an activity which employs inferential mechanisms and background knowledge intensively and extensively.

I would like to examine Fodor's arguments on this second controversial issue. Fodor poses himself the question, What representation of an utterance does the language input processor compute? (p.88) and answers that an input system that has access to the appropriate transduced representations of an utterance knows everything about the utterance that it needs to know in order to determine which sentential type it is a token of and, probably, what the logical form of the utterance is up to ambiguity. The problem here is surely that most if not all (and there may be a logical point about under-determination involved here) phonetic representations are ambiguous as to the sentence types and logical forms they instantiate. So does the input system in that case compute all the types and forms compatible with the input representation, or does it give up on ambiguous representations, calling up central systems to take over the computation?

The former idea is hard to reconcile with the fact that some possible structures of an input representation are difficult to hear, and have to be worked out quite conciously; if all possible types and forms are computed in the input system, shouldn't we be able to recover them all with equal facility? If not, why not? If, on the other hand, the input system gives up on ambiguous representations, when does it give up? Consider the following example: I remark that my neighbour was killed when a light aircraft she piloted crashed into a hill, and you comment:

Flying planes can be dangerous.

Now either the language input system computes both grammatical readings, giving as logical forms what is represented in (1) and (2), and then hands over to central processes to decide between them, or else the input system gives up and hands over to central processes the moment it begins to compute the type ambiguity of Flying (Is it a verb or an adjective?):

(1) For all planes, flying them can be dangerous

(2) For all flying planes, they can be dangerous

Neither alternative seems satisfactory. The first seems inefficient. The second, though it may reconcile with our intuition that in context we would compute logical form (1) directly without even considering (2), appears to assign to central systems just the kinds of computational powers Fodor wants to reserve to the language input system. As a third possibility, it could be argued that the input system as it were weakly generates Flying planes can be dangerous without assigning phrase structure (and hence logical form) at all, leaving that to central systems. This would be congenial to many practitioners of linguistic pragmatics, but as far as Chomsky is concerned, would appear to have the grave defect of locating syntax in a system unlikely to recognize 'the autonomy of syntax'. Nor does this third possibility appear to be intended by Fodor. He holds, for example, that the language input system is capable of recognising English interrogative word order (p.90) and surveying his own account remarks, "All this comports with the strong intuition that while there could perhaps be an algorithm for parsing, there surely could not be an algorithm for estimating communicative intentions in anything like their full diversity" (p.90) - and I take it that 'parsing' standardly involves phrase structure assignment.

Finally, in relation to this discussion of the powers and operations of a language input system, one might consider the question of how an account of second language acquisition might be fitted into the input systems/central systems model of mind. Does the second language learner use central systems to analyse input? Does fluency in a second language correspond to the development of a new input system? If so, where is it located? If not, does this mean there can only be bi-lingualism involving the use of an input system for one of the languages and central systems for the other? .....

Fodor's work is fascinating and controversial, like most of the New Mentalism and the New Organology (so called by Oxford neuroscientist John Marshall, on the strength of Chomsky's and Fodor's talk of 'mental organs'). In America, the new approaches are dominant in prestigious institutions and in Britain they have important representatives: Philip Johnson-Laird at Cambridge and Margaret Boden at Sussex, for example; the former author of Mental Models (1983), the latter of the Fontana Modern Master on Piaget (1977b) among other works. In both countries, the new approaches have attracted sustained criticism from two main directions.

First, from the last ditch behaviourists who still think that psychology is all about presenting stimuli, usually in the form of electric shocks, to rats, pigeons and 'subjects' (i.e. people).

Second, from humanistic philosophers, psychologists and sociologists generally influenced by the later philosophy of Wittgenstein. Jeff Coulter belongs to the second group of critics For him, cognitive theory, as he calls it in his 1983 book, fails to recognise the priority of the Social in the interpersonal construction of the Mental, and connectedly, constantly tends to reduce facts about people to facts about bodies. These failings also account for the misplaced preference shown by cognitivists for Nature over Nurture - in these debates, Wittgensteinians always stand on the side of Nurture. Thus, Jerome Bruner has sought to construct a non-Chomskyan theory of language acquisition which stresses the active and co-operative part played by the child's caretaker (see, for example, the papers in C. Snow and C. Ferguson, Talking to Children, 1977), and he recognised in the Wittgensteinian approach of David Hamlyn's Experience and The Growth of Understanding (1978) a philosophical argument running in parallel with his own developmental psychology.

My own view, for what it is worth, is that some at least of the positions in dispute between the cognitivists and the Wittgensteinians are as empirical as you will ever get, and can be assessed better once we understand what would count as a crucial experiment. It is already clear, for example, that the Wolf Children examples used in older Nature versus Nurture debates are irrelevant, for not even the most radical proponent of Nature (for example, Chomsky) believes that there can be cognitive growth without any interactional triggering. Rather, what is at issue between the nativist and the enviromentalist is the closeness of fit between a particular sort of interaction and a particular sort of development. We can get at this kind of problem quasi-experimentally if, for example, we can find situations where children have been deprived of only one sort of interactive input and can then observe what happens developmentally in the relevant domain. So, for instance, deaf children of hearing parents receive no spoken language input, but these children can and do communicate with their parents. How they do so and what conclusions can be drawn from the development of their idiosyncratic signing systems has become a focus of intense research by cognitively-oriented linguists (see, for example, papers in E. Wanner and L. Gleitman, Language Acquisition: the State of the Art.)

Again. the development of new languages in the transition from pidgins to creoles has been studied with a view to isolating the contribution of the organism from that of the environment. In a major study Roots of Language ( 1981), the linguist Derek Bickerton has postulated the existence of a genetic bioprogram for language development as the only way of accounting for observed features of creole development. That the environmentalist Jerome Bruner, co-reviewing Roots of Language in The New York Review of Books (24 June 1982), accepts the essentials of Bickerton's arguments suggests that progress is now possible on the Nature v. Nurture front. This is but one exciting consequence of the growth of approaches to human cognition which in the space of a few years have both imaginatively exploited computer models of human intelligence and provided a rich interpretation of the superficially paradoxical idea of a biology of cognition. No one interested in Mind or Body will want to ignore what is happening within the cognitive paradigm.

Part Two

Howard Gardner, Frames of Mind (1984)

At Harvard they are still thinking about education in the context of human growth and development. This is in part thanks to the Dutch-based Bernard van Leer Foundation which in 1979 commissioned the Harvard Graduate School of Education to assess the state of scientific knowledge concerning human potential and its realization and to summarize the findings in a form that would assist educational policy and practice throughout the world. Four years and six research assistants later, Howard Gardner has produced the first in what promises to be a series of books from the Harvard Project on Human Potential, Frames of Mind (1984). What has he to report?

Above all, that everything points to human intelligence being organised vertically, as a number of faculties, rather than horizontally, as a set of abilities. We are not creatures of general intelligence or general problem solving ability, but creatures capable of uneven achievement in several relatively self-contained or encapsulated domains. As a consequence, l.Q. tests must be held to show a false picture of individual talent. They just do not tap in to the full range of abilities we possess. Gardner identifies these abilities as seven in number occupied by the linguistic, musical, logical-mathematical, spatial, bodily-kinesthetic and two "personal" intelligences. Now the magic number seven should make us a bit sceptical (Why seven? Why not eight hundred and ten?) So what is the evidence for this picture of the mind? Gardner has little time for the a priori philosophical approach of a Paul Hirst and not much more for the superficialities of the psychometricians, even though he allows that psychometry may provide one source of support for his division of the intelligences. But that is as one criterion among eight, and it is two of the other criteria which set the tone of much of the book.

These are the idea, first, that an intelligence will be localised in the brain in such a way that its core properties will be revealed in the effects of localised brain damage or disease.

Second, the idea that idiot savants, prodigies, and in general individuals who show striking development of just one kind of intelligence provide evidence for the existence and relative autonomy of the different kinds of human potential Gardner identifies.

So far so good. Faculty psychology underpinned by the findings of brain researchers like A.R. Luria (The Man with a Shattered World 1972) and Michael Gazzaniga and Roger Sperry (right and left hemisphere specialization) makes very good sense of what we know about human potential. And using the extraordinary cases of brain damage and prodigious talent to illuminate the structure of ordinary competences is a sound research strategy: the disabled and the abled are, for the scientist, naturally occurring experiments. This explains why, for example, psycholinguists are currently so interested in the linguistic development of deaf children - though to see their importance, you have to think of the language faculty as independent of the oral-aural channel. Likewise, as Gardner points out, we can learn a great deal about the nature of spatial intelligence from studying those who are blind, once we realise that spatial intelligence is not the same as visual perception.

However, though Gardner sets out in the right direction - more or less the same direction Jerry Fodor pursues in The Modularity of Mind discussed in the previous section, I think his book stumbles badly in its execution and I cannot agree with Jerome Bruner that it is in many ways a brilliant book. I found it frequently long-winded, vague, evasive and failing to follow through its leading ideas. Part of the problem I am sure lies in the van Leer brief: asked to be Useful to the World, Gardner has chosen to write like a Committee or an Ecumenical Council.

This accounts, I think, for the extensive but scientifically inconclusive use of materials from cultural anthropology, which has more to do with enriching our sense of cultural diversity than plumbing the central issues of faculty psychology. Again, and more importantly, surveying rival theories Gardner writes that, "somewhere between the Chomskyan stress on individuals, with their separate unfolding mental faculties, the Piagetian view of the developing organism passing through a uniform sequence of stages, and the anthropological attention to the formative effects of the cultural environment, it ought to be possible to forge a productive middle ground."

Now politics may be about the middle ground, but science is not. It's about truth, and Gardner is quite clear that he thinks Piaget importantly mistaken in treating logico-mathematical intelligence as the general form of all intelligence. How could it be productive to retain what you think false? As for the contest between Chomsky and the anthropologists, Gardner elsewhere writes,

" l have taken some pains in this book to avoid pitting genetic against cultural factors."

Is that a virtue? I can't see that it is. Rather, it seems to me that the issue between Chomsky and the anthropologists requires us to take seriously Chomsky's distinction between the triggering and shaping of development and seeing what insight it can yield. Gardner surely knows this and frequently uses the word 'trigger' to characterise developmental growth set in motion by a stimulus, but not controlled by it in the way characteristic of learning. But this is one among many lines of critical enquiry he doesn't pursue in enough depth, even though his background in brain science and psychology equips him to do so. One of the more surprising omissions from the book is the absence of any reference to the 1975 face-to-face debate between Chomsky and Piaget and their respective allies, published as Language and Learning (M. Piattelli-Palmarini 1980) in which the controversy between modular and general-purpose approaches to intelligence is debated at length, in relation to language at least.

A main feature of the two hundred pages Gardner devotes to exposition of his six intelligences is the extensive use of introspective reports by great artists and scientists on how they create. I'm sure most readers will have a sense of deja vu as they encounter Einstein on Einstein and all the rest yet again. More importantly, the whole drift of Gardner's argument tells against the value of what artists and scientists say about their creative work in domains which are not wholly or not at all verbal. First, because Gardner accepts the cognitive science idea that introspective access to the cognitive unconscious is not always possible (p.55). Second, because introspection about your extraordinary musical ability (say) is not an exercise of that intelligence, and your powers of introspection (part of personal intelligence) may be no more than ordinary - that is true on Gardner's own argument for the relative autonomy of intelligences. It is surely the performances of extraordinary individuals which are of interest to the student of human potential, not their introspections - or not at the length they take up here, especially when none of them is developed into an incisive case study.

And case studies of extraordinary individuals do not have to rely on introspective evidence, as the studies of Genie (by Sheila Curtiss 1977) and Nadia (by Lorna Selfe 1980) show: Genie was forcibly isolated by her parents in an attic until adolescence and her post-release development has been intensively studied. Nadia showed extraordinary drawing ability though an autistic child.

Gardner's concluding discussion of educational implications of his work, is as he acknowledges, undeveloped. It is not much more than an eclectic survey of a number of practical endeavours to realise human potential: the Suzuki method, the Venezuelan national project to develop intelligence, Puluwat navigation apprenticeships, Paolo Freire's pedagogy, a bit on computers (but no serious discussion of, say, Papert's work). In short, the book fizzles out. If you want 400 pages on current approaches to human potential, Papert's Mindstorms (1981) and Fodor's Modularity of Mind (1983) together offer a livelier and deeper entry into current thinking than does Gardner, even though they approach the questions with which Gardner is concerned on a much narrower front and without his knowledge of the brain sciences.

Part Three

John Searle

Wittgenstein once remarked that a philosopher who doesn't engage in public debate is like a boxer who never enters the ring. By this standard, John Searle, Professor of Philosophy at the University of California, Berkeley, is a true prizefighter. In recent years he has taken on Noam Chomsky, the champion of modern linguistics; Jacques Derrida, the heavyweight of post-structuralism; and endeavoured to deal a knock-out blow to the pretensions of the Artificial Intelligentsia. To his critics, he is punch drunk.

But that can't be true, because Searle is also a constructive philosopher who in his books Speech Acts (1969), Expression and Meaning (1979c), and Intentionality (1983a) has sought to present simply and systematically a vision of language and mind which draws on a rich background in modern philosophy. What is this vision and how does it relate to Searle's contests with Chomsky, Derrida and computer theorists?

Searle's vision is a naturalistic one: we cannot understand mind and language, he says, except on condition that we connect them to human beings as biological organisms, embodied and embedded in the physical world. Our ability to direct ourselves towards the world (our 'intentionality") - to believe things about it, desire things in it and to express those beliefs and desires in language - is something we possess as a causal consequence of having the sorts of bodies (including in our bodies, our brains) that we have. This biological naturalism aligns Searle with the later philosophy of Wittgenstein and, in the most recent work, with Heidegger, as mediated to Searle through the work of his colleague, Hubert Dreyfus. It immediately opposes Searle to the proponents of Artificial Intelligence (A.l.).

For it has been fundamental to A.l. research (at least until very recently) to think that mind can be understood apart from body, just as software (programs) is quite distinct from hardware (computers). The A.l. model of the mind is one which abstracts from and denies the relevance of its neural basis and concentrates instead on its functional or structural properties. Searle seeks to explode the plausibility of A.l. simulations of mental phenomena by means of a single, simple and - by now - notorious thought-experiment, the Chinese Room.

Searle invites us to imagine him locked in a room into which is passed batches of Chinese writing (he doesn't read Chinese) together with rules in English (and he reads English) allowing him to correlate bits of Chinese writing across the different batches. This is sufficient for him to receive material in Chinese and, to outsiders, appear to answer in Chinese questions about it. But does he understand Chinese as a result? Not a bit of it, says Searle. I merely coordinate one bit of Chinese, which I don't understand (say a 'question'), with another bit of Chinese (say an 'answer'), which I don't understand either. But I produce the appearance of understanding. And in this I am just like the computer programs devised by A.l. researchers. I do no more than manipulate uninterpreted (formal, syntactic) symbols. But symbol manipulation is not understanding and, it follows, the computer program therefore neither understands nor simulates understanding. It is the users of the computer program who attribute meaning to its operations. The meaning is not intrinsic, as it is in humans. Searle's argument is meant to show both that computer programs don't literally understand things and that we wouldn't understand things if all we had in our minds were computer programs. We also need bodies, or as Searle puts it, "only a machine could think, and indeed only very special kinds of machines, namely brains and machines that had the same causal powers as brains" (Searle 1980, p.424). For mind is a causal realization of body.

What are we to make of this argument? My own view is that Searle is both right and wrong. He is right to insist that programs do not show understanding, and that perhaps right that they could not. But neither do bodies or brains show understanding. It is to persons that we attribute understanding, beliefs, desires, etc. Now it may well be necessary to being a person that you have a body, but it may equally be necessary that you have a mind in which sub-personal processes of a computational character take place and which eventuate in the understanding, beliefs, desires, etc. of a person. It is then entirely appropriate to seek to explore and model those sub-personal processes through A.l. work. This is, in effect, the argument of Stephen Stich in his Beyond Belief; From Folk Psychology to Cognitive Science (1983).

Searle's naturalism is also involved in his critique of Derrida. Searle makes the familiar point that a great deal of the Western philosophical tradition is a search for secure foundations or 'guarantees' for knowledge, typified by Descartes' search for certainty. The twentieth century claim, to be found in the work of Heidegger and Wittgenstein, that there are no metaphysical foundations to knowledge meets two responses.

First, that this leaves everything as it is - this is Wittgenstein's and Searle's naturalistic response. As Searle puts it, "The only 'foundation' that language has or needs is that people are biologically, psychologically and socially constituted so that they succeed in using it" (Searle 1983b, sec. 4).

Second, the response that in a world without certainty, everything is permitted. In the Nineteenth Century, this thought anguished a writer like Dostoyevsky; today, Derrida proclaims it as the new joyful wisdom. Searle thinks Derrida's position vitiated in the same way as all relativisms; it presupposes what it sets out to deny.

Biological naturalism opposes Searle to A.l. and Derrida, but it does not oppose him to Chomsky, for whom most interesting properties of language and mind are innate, part of our biological inheritance. In fact, Searle's quarrel with Chomsky is that he is too much a biologist and too little a sociologist. For Searle, as for Heidegger and Wittgenstein, mind and language are intrinsically social and cultural, as well as biologically rooted. Chomsky sees only half the picture; his biological naturalism is too restrictive. In particular, he studies language without reference to the use of language in communication and treats syntax as autonomous with respect to semantics. In consequence, Searle argues, when Chomsky explains the growth of syntactic competence in the child in terms of the triggering of innate mental structures, this is not a freely arrived at hypothesis. Rather, it is the position to which Chomsky is driven by his initial restrictive assumptions. If these are relaxed, it is possible to imagine different explanations of syntactic development. For example, Chomsky argues that syntactic features of the interrogative are only explicable by assuming that they are innately specified. In reply, Searle points out that if you are willing to note that the interrogative is standardly used to ask questions, and that success in asking questions depends on being able to identify the referent of the question, etc., it becomes rather easy to explain what looked like esoteric facts of syntax. In short, "an understanding of syntactical facts requires an understanding of their function in communication since communication is what language is all about" (Searle 1974, p.16).

On the same lines, Searle argues against Chomskyan semantics that meaning cannot be understood apart from the use of language. The arguments against Chomskyan semantics are analogous to those deployed in the Chinese Room example. Thus, in one brand of Chomskyan semantics we are simply offered paraphrases: "bachelor" means "unmarried adult male". But this sort of analysis is circular and presupposes the sytem it is supposed to explicate: one bit of English is correlated with another bit of English, but that cannot elucidate what is involved in understanding English. In another brand of semantics (due to J.J. Katz: see Katz 1972), we are told that "bachelor" means (+ MALE, + ADULT, + HUMAN, - MARRIED). But the capitalized words belong to some new language which might as well be Chinese; no account is given of how we are supposed to understand it.

Searle's alternative to such approaches derives from the work of the English philosopher J.L. Austin, whose ideas (Austin 1962) were systematised in Searle's first book Speech Acts (1969). In this alternative, a theory of meaning involves giving rules for the use of expressions in speech acts, rules which enable us to refer to things and predicate things of them. For "speaking a language is a rule-governed form of intentional behaviour" (Searle 1976) and the rules are of the sort which speakers ought to be able to recognise as the rules they follow in asking questions, making promises, giving orders, and so on. Whereas Chomsky thinks of rules of language as non-introspectible mechanisms, for Searle they can be brought to self-consciousness by an act of recall (Plato's anamnesis). Whereas Derrida wants to sever language from intention, for Searle language is inseparable from intentional, communicative behaviour. And whereas cognitive scientists think of rules as in the Mind, Searle thinks of them as between People in Society.


Works cited in this essay can be found listed in the consolidated Bibliography for this and other website essays

Website version 2005, restoring editorial cuts made to the three previously published versions:

Part One was a review of Jerry Fodor's Modularity of Mind published in the Times Educational Supplement, 29 June 1984

Part Two was a review of Howard Gardner's Frames of Mind published in the Times Educational Supplement, 21 December 1984

Part Three was a Profile of John Searle, published in the Times Higher Educational Supplement on 5 October 1984 and just before he delivered the 1984 Reith Lectures (published as Minds, Brains and Science. John Searle was kind enough to OK the general accuracy of my presentation of his views.

Copyright material republished by kind permission of Times Newspapers Ltd