A novel view of between-categories contrast and within-category assimilation.

This research manipulated the portion of a category distribution that is misclassified by the optimal classifier and investigated the impact on assessments of category attributes. Three separate studies manipulated the direction of overlap, the extent of overlap, and the relative base rate of the comparison category. All 3 studies produced large between-categories contrast and within-category assimilation. As expected, these effects were enhanced in conditions in which the optimal classifier misclassified a larger portion of the target category. Study 4 demonstrated that intercategory overlap in the absence of overt classification does not produce contrast and assimilation. Ironically, optimizing categorization accuracy can produce highly inaccurate beliefs about category attributes.