Moral inference under stress

Adaptively navigating an uncertain social world often requires individuals to infer other people’s moral character based on their observed behavior. Recently, this inferential process has been elucidated through the lens of Bayesian learning, wherein beliefs about others’ character are dynamically updated in proportion to the uncertainty of those beliefs. This theoretical framework suggests that stress, a psychophysiological state intricately tied to uncertainty, may influence moral inference. Specifically, we hypothesized that acute stress would increase belief uncertainty and, consequently, volatility. To investigate this link, participants were randomly assigned to either a psychosocial stress or control condition, before engaging in predicting and observing moral decisions of two distinct agents with differing moral characters. Consistent with prior research, Bayesian computational modeling revealed that beliefs concerning the morality of a perceived “bad” agent were more uncertain and therefore more volatile compared to beliefs about “good” agents. Remarkably, our findings show that acute stress increased belief uncertainty and volatility for good agents, provided the good agent was observed first (i.e., temporally closer to the stressor). Furthermore, this effect was found to be associated with elevated salivary cortisol levels, suggesting a potential mediation through glucocorticoid action. Our study sheds light on the intricate interplay between acute stress and moral inference. The observed increase in belief uncertainty and volatility for putatively good agents may reflect an adaptive mechanism in an uncertain environment. These insights contribute to our understanding of the nuanced ways in which stress can shape social cognition and moral judgments.