Philosophy of Cognitive Science

In recent decades cognitive science has revolutionised our understanding of the workings of the human mind. Philosophy has made a major contribution to cognitive science and has itself been hugely influenced by its development. This dynamic book explores the philosophical significance of cognitive science and examines the central debates that have enlivened its history.

In a wide-ranging and comprehensive account of the topic, philosopher M.J. Cain discusses the historical origins of cognitive science and its philosophical underpinnings; the nature and role of representations in cognition; the architecture of the mind and the modularity thesis; the nature of concepts; knowledge of language and its acquisition; perception; and the relationship between the brain and cognition.

Cain draws upon an extensive knowledge of empirical developments and their philosophical interpretation. He argues that although the field has generated some challenging new views in recent years, many of the core ideas that initiated its birth are still to be taken seriously.

Clearly written and incisively argued, The Philosophy of Cognitive Science will appeal to any student or researcher interested in the workings of the mind.

M.J. Cain is a Senior Lecturer in Philosophy at Oxford Brookes University.

Auteur
M.J. Cain is a Senior Lecturer in Philosophy at Oxford Brookes University.

Échantillon de lecture
2 Representation and Computation

1 Introduction

In the previous chapter we saw that, at its inception, cognitive science was committed to the idea that cognition involves the manipulation of representations by means of computation. These representations and representation-manipulating computational processes are implemented or realized in the brain. The upshot of this is that explaining a particular cognitive capacity, such as visual perception, object recognition, high-level reasoning, action planning, language development, understanding the mental states of another person, and so on, involves identifying the representations and computational processes involved whenever that capacity is exercised. But what general form do these representations and computational processes have? Within cognitive science there are two broad answers to this question, reflecting a divide between two competing approaches. The first answer is associated with an approach widely known as classical computationalism. As its name suggests, classical computationalism dominated cognitive science during the early decades of its existence. 1 The second answer is associated with an approach known as connectionism, which, though having its origins in work in the 1950s, 2 came to the fore only in the 1980s with the publication of Rumelhart and McClelland's connectionist 'bible' (Rumelhart et al., 1986; McClelland et al., 1986). Both connectionism and classical computationalism are very much alive today. However, recent years have seen the emergence of a family of approaches offering a radical alternative to both classical computationalism and connectionism, an alternative that questions the role of mental representations in cognition. In this chapter I will examine each of these three perspectives on cognition, beginning with classical computationalism.
2 Classical computationalism

A classical computer is a machine that takes structured language-like symbols as input and produces structured language-like symbols as output. This input-output profile is mediated by the application of rules which collectively constitute the program that the machine runs. The application of the symbol-manipulating rules - the running of the program - is a mechanical process in that it does not require any intelligence or creativity on the part of the machine. 3 In particular, it does not require the machine to understand or appreciate the meaning of the symbols that it manipulates. What the machine is sensitive to are the formal or syntactic properties of symbols, and the symbol-manipulating rules it applies relate to such properties. Hence, a classical computer is, to use Daniel Dennett's (1978c) memorable phrase, a 'syntactic engine'. Nevertheless, the manipulated symbols typically have semantic properties - after all, they are symbols, and it is natural to think that, in order for an item to be a symbol, it must either have some meaning or be apt to have some meaning attributed to it. Moreover, the meaning of the output symbols generated by a classical computer is typically coherently related to the meaning of the input symbols. A simple example of this would be a machine that takes as input pairs of Arabic numerals and produces as output the numeral that is the sum of the input pair. Thus, for example, the machine gives the output '5' in response to the input '2, 3', '10' to the input '3, 7', and so on. Thus, the machine computes the addition function for pairs of numbers despite the fact that it doesn't have any understanding of mathematical concepts such as numbers and addition. Therein lies the beauty of classical computers: despite the fact that they are entirely mechanical and incapable of understanding, they can be built so as to mimic the behaviour of a system that does have understanding and intelligence - for example, a system that has mathemat

Contenu

• Table of Contents

• Chapter One: Cognitive Science and the Philosophy of Cognitive Science

• Chapter Two: Representation and Computation

• Chapter Three: Modularity

• Chapter Four: Concepts

• Chapter Five: Language

• Chapter Six: The Brain and Cognition

• Conclusion

• References

• Notes