Abstract
The human face is a rich source of information for social judgments achieved by other people on the basis of a brief glance. Such inferences have crucial implications on social outcomes such as: mate choice, voting preferences, occupational opportunities, etc. Although several studies have addressed the perceptual aspects of social judgments from faces, very little is known about how facial information is collected and integrated across time to result in a social judgment. Moreover, social interactions occur mostly when confronted with dynamic faces; however, the effect of facial dynamic gestures on social evaluation has not been systematically addressed yet. Here, we use a data-driven psychophysical approach to establish which movements of muscle groups in the face elicit social judgments of trustworthiness, dominance, and attractiveness. A novel computer graphics platform allows us to flexibly synthesize the dynamics of any facial muscle group (Action Unit, AU) using 6 temporal parameters: peak amplitude, peak latency, onset latency, offset latency, acceleration, and deceleration (Yu, Garrod, and Schyns, 2012). Twelve participants judged the trustworthiness, dominance, and attractiveness of random facial animations and rated the perceived intensity of each stimulus. Using reverse correlation, we analyzed the participants' responses to random activation of AUs to estimate the perceptually relevant expression gestures associated with each social judgment. Results showed high correlation across participants, with attractiveness and trustworthiness positively correlated with each other and negatively correlated with dominance.