Abstract
Activity in the amygdala during intergroup contact and prejudice has been explained as a conditioned threat response and as a social saliency response. More recently, the theory that both explanations are true has received some empirical support, suggesting that the amygdala subnuclei (basal, lateral, and central) may play distinct roles in the social categorization process. In this study, we used machine learning to decode amygdala subregion activations during simulated encounters with protagonists representing four different stereotypes. In these encounters, a generalized threat was presented: the protagonists initially remained at a distance, then approached the participant, and finally entered their personal space. Using a time-resolved decoding approach, we studied the effect of interpersonal distance on the subnuclei's prejudice response. Moreover, a second classification was used to evaluate how the group relevance modified the perceived generalized threat. Both classifications revealed that all amygdala subregions encoded stereotype content. The likelihood of successful stereotype classification in each subregion was modulated differently by the context. Methodologically, our results show that a time-resolved decoding approach provides tools for studying prejudice within the amygdala's subnuclei. Neuroscientifically, our results support the theorization of different and parallel functions of amygdala subnuclei in prejudice.