Go to Semiotic Review of Books Home Page
Go to SRB Highlights
Go to SRB Archives
This review appeared in Volume 1 (3) of The Semiotic Review of Books.
The Computer and the Mind: An Introduction to Cognitive Science, by Philip N. Johnson-Laird, Cambridge, Mass.: Harvard University Press, 1988, 444 p. ISBN 0-674-15615-3.
Cognitive Science, "the mind's new science," the subject of Gardner's valuable 1985 history of the "cognitive revolution", has had a high profile now for some fifteen years. But until recently it has been difficult for the non-specialist to evaluate the consequences of that putative revolution, since even three years ago the only general work on cognitive science was Gardner's survey, already some years out of date. In 1987 however there appeared the MIT introduction by Stillings et al., and the following year brought both the collection of articles edited by McTear and the work under review here.
Johnson-Laird's book was thus a welcome addition to the very small list of general works on Cognitive Science. As a well-known cognitive scientist and psychologist (as witness the superb 1983 Mental Models and the significant 1976 work by Miller and Johnson-Laird, Language and Perception, he was well placed to write an introduction to the new science.
Gardner (1985:6) defines cognitive science as "a contemporary, empirically based effort to answer long-standing epistemological questions--particularly those concerned with the nature of knowledge, its components, its sources, its development, and its deployment." At the same time, he sets forth five characteristics of the cognitive scientist: a belief in mental representations at once distinct from their biological realizations and their cultural functions; faith in the computer as a viable model of how the human mind functions; a strategic de-emphasis of affective, cultural, and contextual factors in cognition; a faith in interdisciplinary studies; and, finally, a focus on epistemological concerns.
Put thus, cognitive science is in direct descent from Plato's Meno and is neither revolutionary nor distinct from cognitive psychology and philosophy of mind. Elsewhere (p. 394), however, Gardner captures what is indeed revolutionary in the field, noting that the first generation of cognitive scientists--for example, Noam Chomsky and George Miller--showed little interest in neurology, believed that mind is disjoint from brain, made the case for an abstract level of mental representation not directly capturable by empirical studies of the sort familiar from experimental psychology, and related human and artificial intelligence via compatible symbolic systems and models of the higher cognitive processes.
Similarly, Johnson-Laird sees the focus of the new convergence of computer science, linguistics, anthropology, neurology, psychology, and philosophy as the attempt "to elucidate the workings of the mind by treating them as computations, not necessarily of the sort carried out by the familiar digital computer, but of a sort that lies within this broader framework of the theory of computation" (p. 9). Hence the title of his book.
The book's first part then, albeit (oddly enough) its second shortest, is "Computation and the Mind", containing the chapters (1) "How should the mind be studied?", its answer to which question naturally being "in terms of models of computational processes"; (2) "Symbols and mental processes," in which Johnson Laird argues that perceptions, ideas, beliefs and the like are all treatable as mental representations or symbols; and (3) "Computability and mental processes," in which he argues, first, that whether the mind itself is computational or not, theories of mind should be; further, that empirical data vastly underdetermine theories of mind; and, finally, that theories of mind should be modelled as computer programs.
From a certain point of view these unexceptionable, even banal, conclusions are troubling, assuming a methodological and conceptual homogeneity not yet achieved in the field, and of questionable value for its future development. There is a wide gap between the high-level theories of the philosopher and the linguist, and the low-level, empirical, work in psychology and artificial intelligence. Philosophy of mind and linguistic theories such as those of Chomsky simply have no testable, empirical consequences, and Chomsky himself, in claiming linguistics to be a branch of psychology, has made it explicit that by "psychology" he does not mean the empirical science psychologists understand by that term.
On the other hand, empirical psychology has discovered little of value to the formation of theories of high-level cognitive processes, and not just because of its anti-mentalistic biases and relative lack of interest in such processes. Just as the abstract linguistic "competence" underlying actual language use ("performance") cannot be directly accessed, and is extraordinarily difficult to get at even through linguistic performance, the actualization of higher-level processes is mediated through lower-level processes and the former are inaccessible to direct observation. Moreover, psychological performance, as much as linguistic, is crucially affected by precisely the sort of factors-cultural, affective, contextual-that Gardner suggested cognitive scientists ignore.
Johnson-Laird's book reflects well the division between low-level, empirical research with no overarching theoretical consequences; and empirically inconsequential high-level, theoretical studies. Despite his title, which would suggest an emphasis on high-level analysis, he leans heavily towards low-level issues, for the simple reason that it was here that the early successes of cognitive science (for example, in the theory of vision) are seen to lie.
In parts II ("Vision"), III ("Learning, Memory and Action"), and IV ("Cognition"), which constitute just over half the book, Johnson-Laird reviews research in three central areas of cognitive science. In keeping with his interest in computational models, his emphasis in these sections is, on the whole, on low-level work close to empirical psychological research and artificial intelligence engineering, and relatively easy to model in a precise way.
It is only in the hundred-odd pages of Parts V ("Communication") and VI ("The Conscious and the Unconscious Mind") that more abstract matter is considered, and here Johnson-Laird, like cognitive science itself, runs into problems. Part VI is short (thirty-odd pages) and largely programmatic, not surprising considering how little is really known about "Self-reflection, free will and intentions" or "Needs and emotions", the subjects of its two constituent chapters, despite centuries of intense philosophical investigation.
Given his emphasis on representation and symbolism, it is surprising that, compared to the MIT introduction, his book barely examines the high-level issues of representation, modularity, and the like, which loom so large in the MIT vision of cognitive science. Unlike Stillings et al. (1987), which has a chapter and a half on representation, Johnson-Laird has none; where they devote four of their twelve chapters to language, he devotes four out of twenty; and he has nothing like their discussion of the philosophical issues.
Johnson-Laird's treatment of linguistics in particular has been strongly criticized in a review article in the pages of the journal Language (Anderson 1989). The thrust of Anderson's critique is that although "linguistics constitutes (apart from the theory of vision and perhaps a few corners of neuropsychology) just about the only cognitive system for which we can say we have something like a formal and explicit theory of its structure, function, and course of development in the organism" (p. 810), Johnson-Laird displays a lack of knowledge of, and interest in, contemporary generative linguistics.
Anderson notes that Chomsky's critique of Skinner's theory of verbal behaviour is made to look like a quibble regarding its capacity to describe sentential syntax, whereas in fast Chomsky's theory of generative capacity would rule out any associationist theory of language, including connectionism. He writes further that Johnson-Laird does not make clear that human languages are fundamentally not the same kind of object as animal communication systems; that he fails to characterize language competence in terms of abstract rule systems, about which linguistics has found out much; and that he exhibits fundamental conceptual confusions, such as his stated belief that it is paradoxical that theories of language should take so much longer to master than language itself.
Anderson concludes (pp. 809-810) that the "lack of penetration of much current Cognitive Science . . by the insights of linguistics" is not an idiosyncrasy on the part of Johnson-Laird; "the field he intended to cover" is simply one in which linguistics does not loom large. Nonetheless, "a good deal of Linguistics is Cognitive Science" and, Anderson claims, should form a major part of that field.
Cognitive science developed originally out of a deep dissatisfaction with the traditional goals and methods of cognitive psychology, along with a desire to exploit advances in a number of fields, but especially in computer science and linguistics. The seemingly innocuous assumption that cognition is the common subject matter of a number of disparate disciplines, along with the apparent intertranslatability afforded by high level abstraction, potentially offered fruitful synergy. That a well-known cognitive scientist could so ignore contemporary linguistics shows the rather different direction the field has actually gone in.
Cognitive scientists have generally imported into the new science the biases of their base fields, and, consequently, the gap between the empirical and low-level, and the theoretical and high-level. For all his claims, Johnson-Laird is simply doing Cognitive Science the way a cognitive psychologist with minimal interest in philosophy of mind could be expected to do it, and from that point of view, linguistics and philosophy alike are at best sideshows. The MIT cognitivists, with very different biases (only three of the seven authors are psychologists) depict a rather different discipline.
Is there any reason to believe that the gap can be bridged, or that it is cognitive science which will do it? Indeed, is there any reason to believe that a distinct science, apart from the traditional, component sciences, can (or is needed to) comprehend cognition? The problems revealed by the volume under review here suggest that only a truly new science, abandoning the commitments of the past (whether these involve empirical research or eschew it), will be able to do so. Given the importance of cognition to human existence, the significance of a unified cognitive science cannot be exaggerated. But, it appears, from The Computer and the Mind we can glean but one half of the rudiments of that nascent science.
Furthermore, on the threshold of the 1990's, the consensus as to what constitutes cognitive science, as delineated in the collection of essays in Posner (1989), has been challenged by a major demarche comprising the work of Fauconnier (1985, 1988), Johnson (1987), and Lakoff (1987, 1988, Lakoff and Johnson 1980).
The theory of meaning and categorization being developed by these three (Danesi 1990) points towards a radically new cognitive science. In what Lakoff calls cognitive linguistics, a meaning is neither an objective given affixed to a linguistic expression nor is it something conveyed by an expression, but rather meanings are mental constructs produced by a cognitive process in which expressions act at most as guides. These processes of contrustion are controlled by mental structures schemata and the like-and are crucially dependent on human mental and somatic experience. This is completely at variance with the view that the functional architecture of the mind is independent at once of both physical brain and physical body.
The theory of metaphor embedded within this cognitive linguistics may have implications beyond linguistics, given the import of metaphor in such realms as visual perception (Kennedy 1982, 1990, Kennedy and Domander 1986), especially as Lakoff and his co-workers reject the modularity of mind (Fodor 1983). Such an expanded horizon would lead to a cognitive science incompatible with much current thinking within both cognitive psychology and generative linguistics.
It is too soon to tell if this theory, or any other new approach currently under exploration, deserves a place in the handbooks. But then, two years after the appearance of Johnson-Laird's, it is uncertain which theories do.
Anderson, S.R. (1989) Review Article: Philip N. Johnson-Laird, "The Computer and the Mind". Language 65(4):800-811.
Danesi, M. (1990) "Language and the Senses: New Directions in Linguistics". The Semiotic Review of Books 1.4-6.
Fauconnier, G. (1985) Mental Spaces. Cambridge, Mass.: MIT Press.
---. (1988) "Quantification, Roles and Domains". In Meaning and Mental Representations. Umberto Eco et al. (eds.). Bloomington: Indiana University Press: 61-80.
Fodor, J.A. (1983) The Modularity of Minds an Essay on Faculty Psychology. Cambridge, Mass.: MIT Press.
Gardner, H. (1985) The Mind's New Sciences: A History of the Cognitive Revolution. New York: Basic Books.
Johnson, M. (1987) The Body in the Minds The Bodily Basis of Reason and Imagination. Chicago: University of Chicago Press.
Johnson-Laird, P.N. (1983) Mental Models Towards a Cognitive Science of Language, Inference, and Consciousness. Cambridge: Cambridge University Press.
Kennedy, J.M. (1982) "Metaphor in Pictures". Perception 11.589-605.
---. (1990) "Metaphor: Its Intellectual Basis". Metaphor and Symbolic Activity 5.115-123.
Kennedy, J.M. and R. Domander. (1986) "Blind People Depicting States and Events in Metaphoric Line Drawings". Metaphor and Symbolic Activity 1.109-226.
Lakoff, G. (1987) Women, Fire, and Other Dangerous Things. Chicago: University of Chicago Press.
---. (1988) "Cognitive Semantics". In Meaning and Mental Representations. Umberto Eco et al. (eds.). Bloomington: Indiana University Press: 119-154.
Lakoff, G. and M. Johnson. (1980) Metaphors We Live By. Chicago: University of Chicago Press.
McTear, M.F. (ed.) (1988) Understanding Cognitive Science. Chichester: Ellis Norwood.
Miller, G.A. and P.N. Johnson-Laird. (1976) Language and Perception. Cambridge: Cambridge University Press.
Posner, M.I. (ed.) (1989) Foundation of Cognitive Science. Cambridge, Mass.: MIT Press.
Stillings, N.A. et al. (1987) Cognitive Science: An Introduction. Cambridge, Mass.: MIT Press.Robert I. Binnick is Professor of Linguistics and Supervisor of the Programmes in Cognitive Science at Scarborough College, University of Toronto. Time and the Verb: A Guide to Tense and Aspect is in press at the Oxford University Press, New York.