CATEGORIZATION CREATES FUNCTIONAL FEATURES

Many theories of object recognition and categorization claim that comp lex objects are represented in terms of characteristic features. The o rigin of these features has been neglected in theories of object categ orization. Do they form a fixed and independent set that exists before experience with objects, or are features progressively extracted and developed as an organism categorizes its world? This article maintains that features can be learned flexibly as a consequence of categorizin g and representing objects. All 3 experiments reported in this article used categories of unfamiliar, computer-synthesized 2-dimensional obj ects (''Martian cells''). The results showed that varying the order of category learning induced the creation of different features that cha nged the perceptual appearance and the featural representation of iden tical category exemplars. Network simulations supported a flexible rat her than a fixed-feature interpretation of the data.