Agency boosts sensitivity to social interactions through active information seeking

Humans use visual information to detect social interactions not only between others (third-person perspective), but also—and more importantly—between themselves and others (first-person perspective). Previous work in social perception has mostly focused on the former, leaving first-person scenarios underexplored. In this study, we created simple visual scenes with varying levels of evidence for a social interaction between two agents, represented as dots. Participants either actively participated in the scene by controlling one of the dots with their mouse or passively observed the two dots’ movements, then reported whether one dot was chasing the other (social) or the two were moving independently (nonsocial). Active participation biased participants toward reports of socialness and lowered their threshold for detecting an interaction, indicating heightened sensitivity to social information; these perspective differences persisted even when visual input was perfectly matched. Drift‑diffusion modeling on self-paced responses revealed a significantly higher drift rate in the active condition, suggesting that faster evidence accumulation was related to the boost in sensitivity. Further analysis of mouse trajectories showed that participants who engaged in more exploratory, information‑seeking movements (i.e., heading towards the other agent during a possible chase) exhibited lower thresholds for social signal detection in the first-person perspective and larger perspective differences. These findings demonstrate that agency boosts sensitivity to social interaction by enabling active information sampling and accelerating evidence accumulation, highlighting the importance of the first‑person perspective for understanding human social cognition.