Objectively measured facial traits predict in-person evaluations of facial attractiveness and prosociality in speed-dating partners

Many studies have investigated whether facial averageness, masculinity, and similarity are associated with facial attractiveness. These studies have relied on ratings of images of real or digitally morphed faces. It is important to establish whether past findings translate to real-life, face-to-face evaluations of potential partners; lack of effects in this context would cast doubt on the evolutionary relevance of previous findings. Further, previous studies have not considered that, by definition, faces that are more similar to the average face (i.e. higher in averageness) tend to be more similar to raters' faces. Therefore, these image-rating studies, which separately found that averageness and (in some cases) similarity are attractive, are confounded. To address these issues, we conducted a laboratory-based speed-dating study of 682 participants with objectively measured facial traits, where opposite-sex pairs rated each other on facial attractiveness and prosociality. We found that facial attractiveness was predicted separately by averageness and by similarity (to the rater), but with both variables in the same model, neither uniquely predicted attractiveness. Similarity, but not averageness, predicted prosociality ratings. Facial masculinity was positively and negatively associated with facial attractiveness ratings of men and women, respectively. These results confirm, in real-life interactions, some key findings from image-rating studies but raise questions about others, notably the attractiveness of facial averageness.