Managing trust under dynamic information

We study how interpersonal trust develops and interacts with social identity in dynamic environments using laboratory trust games with noisy feedback about partners’ trustworthiness. Participants fall into three groups based on their initial trust: those with moderate trust update symmetrically, those with low trust are more responsive to positive feedback, and those with high trust react more strongly to negative feedback. Social identity influences which group participants belong to, but does not alter their updating patterns. In anonymous settings, the balance of high- and low-trust individuals makes aggregate updating appear symmetric and Bayesian. When social identity becomes salient through university affiliation disclosure, distinct behavioral patterns emerge. In-group favoritism increases initial trust levels, shifting more participants into the high-trust category and leading to stronger reactions to negative feedback at the aggregate level. Conversely, outgroup conditions result in greater initial distrust and heightened aggregate responsiveness to positive feedback. Therefore, how people update their beliefs about others’ trustworthiness is driven by their starting beliefs. Social identity helps set these initial beliefs, but does not affect how people update. This explains when trust is fragile (when starting high) or resilient (when starting low), highlighting the interplay between our initial biases and how we react to new information.