Memory-based judgments about multiple categories: A revision and extension of Tajfel's accentuation theory.

Abstract

Accentuation theory states that the classification of stimuli produces encoding biases. Contrast effects enhance intercategory differences; assimilation effects enhance intracategory similarities. Do these biases affect the retrieval of stimuli distributed across many categories? The calendar superimposes arbitrary intermonth boundaries on day-to-day variations in temperature. In Experiment 1, Ss estimated the average temperatures of 48 days. Differences between estimates for 2 days belonging to neighboring months were greater (contrast) and differences between estimates for 2 days belonging to the same month were smaller than actual differences (assimilation). Experiment 2 showed that assimilation accounted for all categorization effects. When modified by assumptions from an exemplar model of category learning, accentuation theory accounts for the results. The relevance of these findings for social categorization is discussed. Merely placing objects into distinct categories affects judgments about those objects. Objects belonging to the same class are more likely to be confused with each other (Stangor, Lynch, Duan, & Glass, 1992; Taylor, Fiske, Etcoff, & Ruderman, 1978) and to be recalled as a cluster (MacNamara, Hardy, & Hirtle, 1989). Often, only the category label is recalled (Park & Rothbart, 1982). People over- or underestimate spatial distances between stimuli depending on whether a category boundary falls between them (Newcombe & Liben, 1982), and they underestimate the degree to which the stimulus distributions of different categories overlap (Krueger & Rothbart, 1990). Categorization effects are intriguing because the act of categorization does not necessarily add information to the stimulus itself, but places a stimulus in a context. Categorization determines which stimuli belong to the same class and which do not. In this study, we examine predictions derived from accentuation theory (Eiser & Stroebe, 1972;Tajfel, 1959,1969). According to this theory, people's perceptions minimize the differences between stimuli falling into the same category (assimilation effect) and maximize the differences between stimuli falling into different categories (contrast effect). The standard design involves the presentation of stimuli that are graded along a quantitative dimension and divided by a categorical boundary falling in the center of that dimension (Tajfel & Wilkes, 1963). That is, accentuation effects have been demonstrated as encoding biases in cases of dichotomous categorization. We first review theory and research on accentuation. Then,