Proceedings of the Annual Meeting of the Cognitive Science Society

At its best, collaboration allows for humans to overcome individual limitations by pooling the knowledge of many minds. However, realizing the full potential of collaboration is difficult because some team members' perspectives may be systematically excluded based on their social identities. Here, we use a classic marble-urn paradigm to empirically test the underlying causes of and specific conditions that give rise to testimonial injustice. Across two experiments (one preregistered, one pilot; N = 147), we find that men are consistently (a) more confident than women in expressing their testimony, and (b) less likely than women to revise their beliefs after incorporating others' testimony, specifically when the evidence participants have access to is moderately strong (Experiment 1) and highly diagnostic (Experiment 2) but not definitive. Our work provides important insights into when and how collaborations fail, as well as the psychological root causes that underlie persistent gender discrepancies in group decision-making contexts.