The relationship between category structure and affect, a core question about the affect-cognition interface, has been largely ignored by both literatures, with the exception of studies on the attractiveness of computer-averaged faces. This article reviews a number of the authors' recent and unpublished studies that demonstrate a robust positive correlation between prototypicality and attractiveness across diverse categories, and that systematically explore several hypotheses about the ultimate, that is, evolutionary, origins of this bias. A tentative dual-origin explanation is offered, in which prototypes of animal categories are preferred as a generalization of a mate-selection adaptation designed for human faces; coincidentally prototypes of artifacts (and possibly natural, nonanimal categories) are preferred by virtue of their subjective familiarity.