Perspectives on Mental Representation

Experimental and Theoretical Studies of Cognitive Processes and Capacities

Jacques Mehler, Edward C.T. Walker, Merrill Garrett

• Publisher: Routledge
• Year: 2017
• ISBN:
  9781315521916
• Edition: 1
• Topic: Psychology
• Subtopic: Cognitive Psychology & Cognition
• Index: Psychology

Originally published in 1982, the editors felt that their field was clearly in need of explanatory accounts for many different areas. This volume presents statements of the status of research in several areas by scholars at the forefront of the discipline. It tries at the same time to juxtapose theoretical and experimental perspectives in order to display some of the major lines of tension in the field. Divided into 5 parts it covers: Theoretical Perspectives; Experimental Studies in Processing; Neuropsychological Studies in Processing; Studies in Development; followed by Commentary on some specific chapters.

I	THEORETICAL PERSPECTIVES

1	On the Representation of Form and Function

Noam Chomsky

Massachusetts Institute of Technology

I have departed from the title assigned me, though not really from the topic. It has repeatedly been suggested that there is a rationalist flavor to much recent work in cognitive psychology, specifically, in the study of language. I think this is correct, and not uninteresting. There are, I think, certain leading ideas in what we might reasonably call “rationalist psychology” that can and should be adapted as part of a framework for contemporary study of language and cognition more generally. Among these are the concern for intrinsic determinants of knowledge and perception: innate structures of mind that take a specific form under the effect of experience, and modes of cognition that play a central role in determining what and how we perceive and think. A related notion, emphasized by Descartes and perhaps his most lasting contribution to psychology, is that our interpretation of the world is based in part on representational systems that derive from the structure of the mind itself and do not mirror in any direct way the form of things in the external world. Descartes offers the account of how a blind man comes to know the character of things he touches with a stick; the “resistance and movement of these bodies which is the only cause of the sensations he has of them” is “in no way similar to the ideas that he conceives of them,” the latter forming part of an abstract system of representation. Similarly, in normal vision a sequence of visual sensations evokes in the mind the idea of an object that is “in no way similar” to these sensations or their organization, but rather is based on abstract systems of coding in terms of intrinsic systems of mental representation. We see a triangle, though the mind has been presented with a sequence of curves and angles that are parts of a (distorted) geometrical figure. To take a contemporary example, presented with a sequence of 2-dimensional images the mind might construct the “idea” of a 3-dimensional rigid body that could have given rise to them under motion, perhaps using Shimon Ullman’s rigidity principle for determining structure from motion. The “blind man with a stick” model of perception is a central component in Descartes’s doctrine of innate ideas. Related notions have arisen in contemporary work, and in my view at least, point to some of the directions that should be followed as we try to assimilate the study of cognition to the natural sciences.

I would like to discuss some of the ways in which ideas of this sort have entered research on language in the past several decades. The study of generative grammar in the modern sense—there are, of course, various antecedents, which I will not discuss—was marked by a significant shift in focus in the study of language. To put it briefly, the focus of attention was shifted from “language” to “grammar.” In earlier work, and much contemporary work as well, the object of inquiry is taken to be a particular language, or perhaps general properties of many or all languages. From this point of view, we might think of language, following Aristotle’s familiar phrase, as sound with a meaning. The English language, then, would be regarded as a set of pairs (s, m) where s is a certain real world object, a physical sound, and m is its meaning. The linguist’s theoretical notions are based on these objects. If, for example, the sound s is regarded as a sequence of phonemes, then the phonemes are taken to be bundles of features that literally inhere in successive segments of the physical event, or sets of such segments constructed on the basis of principles of classification. Other linguistic units are constructed by means of various principles of segmentation and classification of parts of s. Morphemes are sequences of phonemes, phrases are classes of sequences of morphemes or words, and so on. The grammar of a language is a description of these elements and their relations, which might be associated with intensional objects (representing meanings) by some sort of compositional rules. (Here and below, I will often ignore obvious type-token distinctions.)

An obvious problem to be faced is that the language, the set of pairs (s, m), is infinite. This is not a logician’s quibble. The set of sentences from which we draw in normal conversation or writing, or that we understand with no difficulty, is so vast that for all practical (let alone theoretical) purposes, we might as well take it to be infinite. And as irrelevant constraints of time and attention are removed, we see at once that there are no bounds to our knowledge of the sound-meaning pairing. Without any change in what we know, we can understand new and more complex linguistic expressions, in principle without limit, as constraints on time, attention, and “computing space” are relaxed. It must be true, then, that our knowledge of language is somehow represented as a finite system of rules (a grammar) that determine the properties of the infinite number of sentences of the language. Furthermore, each person who knows a human language has constructed this finite system of rules with infinite scope on the basis of a finite exposure to data which enormously underdetermines its form. It is obvious in principle, and not difficult to show in practice, that we know many things about sound, meaning, and their interrelation for which we have no adequate inductive evidence or confirmation from experience, in any general sense of these notions; and this knowledge is largely shared within a speech community. What is the nature of this knowledge and how does it develop in the mind? If we are concerned with these questions, we enter into the study of generative grammar.

Pursuing these questions, we shift our focus from the language to the grammar represented in the mind/brain. The language now becomes an epiphenomenon; it is whatever is characterized by the rules of the grammar (perhaps, in conjunction with other systems of mind, or even other factors, depending on how we choose to conceive of the notion “language”). The grammar in a person’s mind/brain is real; it is one of the real things in the world. The language (whatever that may be) is not. From the point of view I am taking, there is no need to suppose that the notion “language” is well-defined at all. One can define it as one likes—as a social phenomenon of some sort, a system of conventions and shared practices, a pairing of linguistic objects and truth conditions, an ideal object of some sort, or whatever. But the grammar, a real object, is what it is: a system of rules that is in fact represented somehow in the mind/brain, more or less in the same way among individuals whom we may choose (from some other point of view) to think of as “speakers of a given language.” We may perfectly well think of the grammar of, say, English, as assigning a structural description to every possible sound. Some will be characterized simply as noise, others as sounds of perhaps some language (but not mine), others as expressions of my language with some figurative interpretation, others as paired with strict “literal interpretations,” and so on. The question “what is the language generated by the grammar? is not well-defined, and it does not seem very important to try to sharpen it. What is important is to determine the rules and principles of the grammars and the kinds of structures they assign to expressions; and further, to discover how these systems arise in the mind and how they interact with other systems in thought, expression, action, and interpretation.

Pursuing this course, we need no longer think of the linguist’s theoretical notions as constructed in some way out of the phonetic or semantic materials. Rather, the linguist is studying mentally-represented rules and the representations they generate. Phonemic representation, for example, is one mode of mental representation, with its specific properties and relations to other modes of representation. It is mapped into sound in some fashion that must be made precise, but there is no reason to suppose that its basic elements will correspond to physically isolable segments of sound or their features. The same is true of other levels of mental representation. At certain levels—syllables or words, for example—elements of mental representation may correspond in a simple way to isolable physical events. At other levels, this appears not to be true in general. This point of view is developed, for example, in my Logical Structure of Linguistic Theory (1955). We are, in short, studying the systems of mental representation that provide our modes of cognition in the case of language, and more deeply, innate principles that select a specific grammar with its highly articulated system of rules and principles, given some interaction with the environment. The rationalist flavor, in the sense of my earlier remarks, is fairly evident, as is the analogy to recent work in other areas of cognitive psychology.

The basic cognitive notion, then, is “knowledge of grammar,” not “knowledge of language”; and if we decompose it in relational terms, the basic cognitive relation holds between a person and a grammar. Knowledge of grammar, like knowledge of mathematics, of biology, of the behavior of objects in the physical world, of one’s place in a social system, etc., is not expressible as some variety of true belief, as in one standard view. Knowledge of language, so understood, involves or perhaps entails particular instances of knowledge-that and belief-that—e.g., knowledge that some expression s means m. But knowledge is not constituted of such elements and no atomistic, element-by-element account of the character or growth of knowledge of this sort is possible. A system of knowledge, in this sense, is an integrated complex represented in the mind, developed in the mind on the basis of some interaction with the environment, but shaped by innate principles of structure and organization and perhaps containing specific innate constituents. There is little reason to believe that all such systems arise in the same way; knowledge of physics and knowledge of language or the behavior of objects appear to arise in radically different ways, for example. We might say that each such cognitive system pertains (in a way that must be made precise) to a specific domain of potential fact, a kind of mentally-constructed world. If there is a domain of actual fact that is not too dissimilar, the cognitive system can be effectively used. The case of language is unusual in that, there being no external standard, the domains of potential and actual fact are identical; X’s grammar is just what X’s mind constructs. But other cognitive systems can be viewed in much the same way. There is some preliminary discussion of these questions, which I think merit much more careful study, in my Reflections on Language (1975) and Rules and Representations (1980).

The study of generative grammar has often been criticized as a departure from the canons of the natural sciences in that it is avowedly “mentalistic”—that is, concerned with the rules and representations of the mind/brain—and does not base itself firmly, so it is argued, on the actual events that are observed, such as particular sounds, speech acts, or whatever. To me it seems that exactly the reverse is true. The shift of focus from language (an obscure and I believe ultimately unimportant notion) to grammar is essential if we are to proceed towards assimilating the study of language to the natural sciences. It is a move from data-collection and organization to the st...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Original Title Page
• Original Copyright Page
• Table of Contents
• Preface
• Part I: Theoretical Perspectives
• Part II: Experimental Studies in Processing
• Part III: Neuropsychological Studies in Processing
• Part IV: Studies in Development
• Part V: Commentary
• Author Index
• Subject Index