ABSTRACT
A logical interpretation of categories in terms of definitions with necessary and sufficient criteria and a prototypical interpretation of categories in terms of clear cases and gradients of membership are normally considered mutually exclusive. So much so that when empirical evidence suggests that both may be operating for a given sample of categories (Gleitman, this volume), that in itself is taken as an argument that we should abandon any search for a general characterization or theory of categorization. In this paper, I would like to situate the discussion of logic versus prototypes in categories in the somewhat broader context of types of reasoning and suggest that there are at least two such types: reasoning using logical structures and reasoning from reference point cases. Use of such a term as reasoning is not meant to imply conscious deliberation but rather refers to any form of inferring, deliberate or automatic, in which we go beyond the information given to form categorizations, judgments, or decisions.
