When and Why are Social Categories Overused Relative to Individuating Information? A Bayesian Approach to Identifying Biases in Impression Formation Processes

When making inferences about others, people often need to integrate multiple types of information, including information about the person’s social categories (e.g., occupation) as well as other ‘individuating’ information (e.g., their behaviors). The current work re-examines this integration process, to understand when and why it might lead to biases that involve over-relying on category information. To do so, we identify two key challenges in identifying such biases, and develop a novel Bayesian modelling approach to overcome these challenges. As a first step in applying this approach, the current work examined a set of novel predictions based on viewing the Continuum Model of impression formation in light of this Bayesian approach. Specifically, a series of six studies tested whether occupation categories might be overused in general, or especially in conditions thought to reduce effortful processing (i.e., greater cognitive load or information consistency). At baseline, there was no consistent evidence of category overuse in any of these conditions, speaking against the idea that intrinsic differences in how these categories are processed or represented will lead to their overuse. In addition, this work provided the first direct evidence of a case where categories were overused: when contextual factors (i.e., background goals) made them the category especially relevant, while processing resources were limited. More broadly, the current work developed the theoretical and methodological foundations for identifying category overuse that stems from biased inference processes, and demonstrated the power of this approach for understanding when and why these biases occur.