Stereotypes are important simplifying assumptions we usefor navigating the social world, associating traits withsocial categories. These beliefs can be used to infer anindividual’s likely social category from observed traits (adiagnostic inference) or to make inferences about anindividual’s unknown traits based on their putative socialcategory (a predictive inference). We argue that theseinferences rely on the same explanatory logic as other sortsof diagnostic and predictive reasoning tasks, such as causalexplanation. Supporting this conclusion, we demonstratethat stereotype use involves four of the same biases knownto be used in causal explanation: A bias against categoriesmaking unverified predictions (Exp. 1), a bias towardsimple categories (Exp. 2), an asymmetry betweenconfirmed and disconfirmed predictions of potentialcategories (Exp. 3), and a tendency to treat uncertaincategorizations as certainly true or false (Exp. 4).