| Abstract: | We examined the amenability of abstractions of categories to new and relevant information. In Experiment 1, Ss formed impressions of 2 sets of numbers by periodically estimating the cumulative means of each set. During the 1st half of the procedure, the 2 means were mathematically stable. During the 2nd half of the procedure, the mean of 1 set was modified and the mean of the other set remained unchanged. We predicted and found that the resultant estimates for the modified category changed more when the mean difference between the 2 categories was enhanced than when it was reduced. Experiment 2 suggested that the accentuation effect results from a 2-stage process of category learning (Stage 1) and category change (Stage 2). Experiment 3 replicated the effect with person categories. The relevance of category accentuation is discussed with respect to the modifiability of social beliefs. |
