A conditional threat-regulation model: When deliberative processing of ongoing threat reduces intergroup bias.

Ongoing societal threats such as war, civil unrest, and pandemics heighten intergroup bias by inducing cognitive rigidity and suppressing complex reasoning. Yet most bias-reduction strategies require conditions that threat itself undermines, including direct engagement with outgroups. We propose a conditional threat-regulation model that specifies when and for whom threat-induced cognitive rigidity can be regulated in the face of threat. Across three experiments conducted during real societal crises in Israel, Hong Kong SAR, and the United States, we test whether deliberative processing of ongoing threat can enable integrative complexity and thereby reduce intergroup bias toward groups associated with the threat. Two exploratory studies conducted during armed conflict and societal unrest varied writing topic (threat vs. nonthreat) while holding depth of processing constant; deliberative processing of the ongoing threat increased integrative complexity conditionally and in turn was associated with reduced intergroup bias. A preregistered experiment conducted during the COVID-19 pandemic crossed writing topic with depth of processing (deliberative vs. intuitive) to isolate the role of deliberation, and demonstrated that deliberative processing of ongoing threat enabled greater integrative complexity and reduced intergroup bias. Across studies, these effects were conditional on individuals' cognitive preparedness, indexed by multicultural experience, with benefits concentrated among individuals able to translate deliberation into integratively complex reasoning. Integrative complexity served as a structural signature of regulated cognition under threat, mediating the relationship between deliberative processing and intergroup judgment. Together, the findings indicate that regulation of threat-induced cognitive rigidity is possible but conditional on both situational prompts and individual cognitive readiness.