PROTOTYPE EFFECTS AND PEAK SHIFT IN CATEGORIZATION

People asked to categorize exemplars of 2 categories often respond mor e accurately to the prototypes of those categories than to other exemp lars. The authors suggest that this prototype effect may often have be en confounded with apeak shift as is observed when pigeons are trained to discriminate between two wavelengths (S+ = 550 nm and S- = 560 nm) , and the peak of their postdiscrimination gradient lies at 540 or 530 nm rather than at 550 nm. Three experiments established that a simila r peak shift can occur when people are asked to categorize 2 sets of s timuli, but the authors also provide evidence of a true prototype effe ct uncontaminated by any peak shift. These results appear to pose cons iderable problems for exemplar-based theories of categorization.