Bottom-up and top-down information determinants of naturalistic trait impression updating

Understanding others’ personality traits plays an important role in guiding social interactions. In daily life, people update their impressions of others’ traits using rich verbal and nonverbal information. However, quantifying this rich, naturalistic information is challenging. Here we leverage novel computational tools to characterize multimodal information streams in audiovisual narratives, and examine their contributions to dynamic trait impression updating. Across three studies (N = 817), we find that people update face-based first impressions of most traits after viewing audiovisual narratives. By examining the dynamic process of impression updating, we find that how and when people updated impressions were most strongly predicted by unique information in targets’ voice, followed by head and face movements, and semantic content. Perceivers’ own affect, attitudes, and personality also predicted impression updating. These findings comprehensively characterize the informational drivers of naturalistic trait impression updating.