Associative learning of social interaction alters attention to human faces

This is the authors' abstract. We don't add key points for this paper.

Beyond current goals and physical salience, selection history reflects how past experience automatically biases visual attention. While existing frameworks characterize how we learn to prioritize non-social regularities such as reward and spatial probability, whether selection history can incorporate abstract social structures governing interactions between others remains unknown. Across four experiments ( N = 144), we examined whether observed social configurations produce lasting attentional biases toward individual agents. Participants first viewed pairs of human faces arranged in interacting (face-to-face) or non-interacting (back-to-back) configurations. In a subsequent cueing task, faces previously shown interacting captured attention more strongly than those shown non-interacting. This effect was abolished by face inversion and absent for non-social directional stimuli (arrows, animal faces), ruling out contributions from low-level visual features or general directional cueing. Together, these findings indicate that social interaction history induces lasting changes in the priority map, endowing faces with learned attributes that guide selection in social environments.