THE RED HERRING AND THE PET FISH - WHY CONCEPTS STILL CANT BE PROTOTYPES

There is a Standard Objection to the idea that concepts might be proto types (or exemplars, or stereotypes): Because they are productive, con cepts must be compositional. Prototypes aren't compositional, so conce pts can't be prototypes. However, two recent papers (Osherson and Smit h, 1988; Kamp and Partee, 1995) reconsider this consensus. They sugges t that, although the Standard Objection is probably right in the long run, the cases where prototypes fail to exhibit compositionality are r elatively exotic and involve phenomena which any account of compositio nality is likely to find hard to deal with; for example, the effects o f quantifiers, indexicals, contextual constraints, etc. In this paper, we argue that the Standard Objection to prototype theory was right af ter all: The problems about compositionality are insuperable in even t he most trivial sorts of examples; it is therefore as near to certain as anything in cognitive science ever gets that the structure of conce pts is not statistical. Theories of categorization, concept acquisitio n, lexical meaning and the like, which assume the contrary simply don' t work.