Proceedings of the Annual Meeting of the Cognitive Science Society

Humans live in an uncertain world and often rely on socialinformation in order to reduce uncertainty. However, therelationship between uncertainty and social information use is notyet fully understood. In this work we argue that previous studieshave often neglected different degrees of uncertainty that need tobe accounted for when studying social information use. Weintroduce a novel experimental paradigm to measure risky decisionmaking, wherein social information and uncertainty aremanipulated. We also developed a Bayesian model of socialinformation use. We show that across different levels ofuncertainty; social influence follows similar principles. Socialinformation is more impactful when individuals are moreuncertain. Notably, this relationship holds for experimentalmanipulations of uncertainty but also for subjective uncertaintywithin experimental conditions. We conclude with discussing thatsocial influence can be better understood when paying credit tosubjective uncertainties and preferences.