Pupil dilation reflects prediction errors in learning about others’ personality traits

Learning about others’ personality traits is crucial for building personal relationships. This learning process can be conceptualized as reductions in prediction errors (PE), i.e., the differences between expectations about other persons and their subsequent feedback. Pupil dilation has been established as an important marker of various non-social learning processes. However, it remains unclear if pupil dilation also relates to PEs in abstract social learning processes. To test this, we conducted a pilot experiment and a pre-registered main experiment where participants were instructed to learn about the traits of other people (i.e., learning profiles) or about a self-profile, a profile that reflected the participant’s own trait ratings. We hypothesized that pupil size reflects PEs during social learning. We applied linear mixed-effects (LME) models to the behavioral data to test whether absolute PEs decreased over time. In the main experiment, we observed a significant reduction in absolute PEs, indicating successful learning. We then used LME models for the pupillometry data to test the relationship between absolute PEs and pupil dilation. In both experiments, larger absolute PEs were associated with larger pupil dilation on a trial-by-trial basis. This effect remained robust after controlling for confounding variables such as signed PEs and feedback. As expected, learning about the self-profile yielded smaller absolute PEs compared to other-profiles, and this was accompanied by reduced pupil dilation. Taken together, these findings suggest that pupil dilation scales with error magnitude and represents a critical physiological index of personality trait learning about others.