Towards a Non-Mechanistic Theory of Meaning

Abstract

In the present paper I consider versions of the mechanistic theory of meaning advanced by Fodor, Dennett, and McGinn. Fodor wants to hold that the nervous system is, or is analogous to, a 'semantic engine'.' Dennett wants to construct a mechanistic alternative to Fodor which does not involve such a 'sub-personal meaning mechanism'.2 But each freely admits the prima facie difficulties of their own mechanistic enterprise.3 In order to counter the threat of incoherence each stresses that their theories are not metaphysical and have lots of empirical constraints. Fodor could be speaking for both of them when he says, 'The program is far from fully clear, but there is no obvious reason to believe that it is fundamentally confused; . . . no obvious reason to deny that it is a program of empirical research.'4 I argue that the mechanistic account of meaning is not a program of empirical research, and that such theories are fundamentally confused. My argument is derived from Wittgenstein, but it is not an argument which is generally associated with him. For this reason, it is useful to begin with McGinn's apology for mechanism in his Wittgenstein on Meaning.5 For McGinn holds that Wittgenstein has no argument for his 'radical' anti-mechanist position.6 McGinn is representative of the commentators in having failed to recognize that (i) in his Tractatus Wittgenstein had self-consciously laid the philosophical foundations of a mechanistic theory of meaning (which includes both the picture theory of meaning and atomistic metaphysics), and (2) in his later philosophy