Abstract
Face attractiveness confers considerable advantages in social interactions, with preferences likely reflecting psychobiological mechanisms shaped by natural selection. Theories of universal beauty propose that attractive faces comprise features that are closer to the population average while optimizing sexual dimorphism (masculine vs. feminine distinction). However, emerging evidence questions this model as an accurate representation of attractive faces, including representing the diversity of beauty preferences across cultures and their individual members. In this study, we used a data-driven method to model, at the individual and cultural levels, the features of attractive female faces, in two matched groups of young male participants (40 East Asians and 40 Western Europeans). We first generated a broad range of same- and other-ethnicity female faces with naturally varying shapes and complexions that participants rated on attractiveness. Then, we reverse correlated the specific features that modulated the perception of face attractiveness in each individual participant. From these individual models, we reconstructed the representation space of face attractiveness. In contrast to theories of universal beauty, we show that attractive face features are distinct from the average (and from sexual dimorphism) in both cultures. We then disentangle attractive face features into those that are shared across cultures, those that are culture-specific, and those that are specific to the individual participant. Our demonstration reveals that face beauty is grounded in diverse features sensitive to culture and ethnicity. Our results have a direct theoretical and methodological impact for representing diversity in theories of social perception and application for the design of culturally and ethnicity sensitive digital agents.