Categorization and Amnesia - Maintenance and Decline of the Typicality Effect

Amnesia leaves some memory abilities completely intact and others only partially so. Although these dissociations have been interpreted as evidence for separate memory systems, McClelland and Rumelhart (1986) suggest that a simple decrease in the learning rate might account for a dissociation between the retention of category and exemplar information. The present experiment was carried ow to determine whether memory abilities involved in the formation of artificial categories are preserved in amnesia, and whether the results would be compatible with McClelland and Rumelhart's proposal. It turned out that performance was partially preserved and the pattern of results matched what could be predicted from the categorization literature. Although more typical transfer stimuli were classified better by both normal and amnesic subjects, this typicality gradient disappeared within about 24 hours in amnesic patients. However, only normal subjects classified old (previously presented) exemplars better than new exemplars. These results are compatible with McClelland and Rumelhart's suggestion. The continuous degradation of exemplar traces may be responsible for the loss of exemplar information and the decline of the typicality gradient at different rates. The same would be true for the typicality gradient. These results are not easily explained by a multiple memory system, but fit reasonably well into Jacoby's (1991) multi-process approach.