Olivia S. Cheung^1,2,3 (), Ayesha Ahmed², Treedom Beiyin Zhang^2,3; ¹United Arab Emirates University, ²New York University Abu Dhabi, ³Center for Brain and Health, NYUAD Research Institute

At a glance of a face, observers make consistent inferences of social traits. Notably, facial attractiveness judgment is thought to be unambiguously based on visual appearance, unlike other traits (e.g., competence). However, it remains underexplored whether first impressions of facial attractiveness might be influenced by higher-level processing. To examine whether such judgments depend primarily on lower-level visual processing of physical appearance, or higher-level processing of social-emotional cues, we presented images of a total of 80 unfamiliar faces briefly (100ms), with 20 faces each for two race (Black/White) and gender (female/male) conditions. First, we tested whether faces may be rated more attractive due to visual (e.g., symmetry) or emotional cues (e.g., genuineness). As emotional expressions lead to a mild degree of facial asymmetry, chimeric faces created from combining either the left or right half of a face with their mirror images could appear to lack emotional genuineness (Kowner, 1996). Participants (N=56) from multicultural backgrounds rated original and chimeric faces. Attractiveness ratings were higher for original than chimeric faces, regardless of face race or gender (p’s<0.02). Second, we investigated whether consistency of attractiveness ratings remains stable among observers with differential implicit social biases for faces from various social groups, as we hypothesized that attractiveness ratings should be consistent when only visual cues are utilized. Ratings for Black and White faces between observers who were either positively biased towards Black or White people, and ratings for female and male faces between observers who were either positively biased towards female or male, were correlated. Attractiveness ratings were consistently correlated for White and Black faces (r=0.72, r=0.70; p=0.43), but were more correlated for female than male faces (r=0.68, r=0.38; p=0.03). Together, these results suggest that higher-level social-emotional information, such as emotional genuineness or implicit biases, is processed rapidly and is crucial for evaluating facial attractiveness.

Acknowledgements: This work was supported by a NYU Abu Dhabi faculty grant (AD174) and by NYUAD Research Institute (CG012) to O.S.C. and a NYU Abu Dhabi capstone grant to A.A.