In defense of de ® nitions

The arguments of Fodor, Garret, Walker and Parkes [(1980) Against de® nitions, Cognition, 8, 263± 367] are the source of widespread skepticism in cognitive science about lexical semantic structure. Whereas the thesis that lexical items, and the concepts they express, have decompositional structure (i.e. have signi® cant constituents) was at one time a one of those ideas that hardly anybody [in the cognitive sciences] ever considers giving upo (p. 264), most researchers now believe that a [a]ll the evidence suggests that the classical [(decompositional)] view is wrong as a general theory of conceptso [Smith, Medin & Rips (1984) A psychological approach to concepts: comments on Rey, Cognition, 17, 272], and cite Fodor et al. (1980) as a sounding the death knell for decompositional theorieso [MacNamara & Miller (1989) Attributes of theories of meaning, Psychological Bulletin, 106, 360]. I argue that the prevailing skepticism is unmotivated by the arguments in Fodor et al. Fodor et al. misrepresent the form, function and scope of the decompositional hypothesis, and the procedures they employ to test for the psychological reality of de® nitions are ̄ awed. I argue, further, that decompositional explanations of the phenomena they consider are preferable to their primitivist alternatives, and, hence, that there is prima facie reason to accept them as evidence for the existence of decompositional structure. Cognitive scientists would, therefore, do well to revert to their former commitment to the decompositional hypothesis.