Shared and individual tuning curves for social vision.

Stimuli with light are clearly visual; stimuli with sound are clearly auditory. But what makes a stimulus "social", and how do socialness judgments differ across people? Here, we characterize group-level and individual thresholds for perceiving the presence and nature of a social interaction. We leverage the fact that humans see social interactions-e.g., chasing, playing, fighting-even in very un-lifelike stimuli like animations of geometric shapes. Unlike prior work using these stimuli, we exploit their most advantageous property: their visual features are fully parameterizable. Using this property, we construct psychophysics-inspired "social tuning curves" for individual subjects. We find not only that simple visual features influence social perception, but also that the exact relationship (shape of the curve) is unique to and stable within each person, thus supporting the view that social information processing begins early in the perceptual hierarchy. Further, individual differences in social tuning curves are related to socio-affective traits. Our approach lays the foundation for a generative account of social perception in single individuals.