Payoff selectivity unlocks collective intelligence in naturalistic human groups

Humans’ abilities for complex social learning and collective adaptation are unparalleled across species. Yet it is largely unknown how people integrate dynamically changing personal and social cues in realistic environments, resulting in collective intelligence or maladaptive behavior. Here, we report an online 3D immersive-reality experiment in which participants (n = 621) searched for and tracked a mobile resource under realistic visual-spatial constraints. Participants completed the task either alone or in groups of five with different types of social information and resource speeds. Contrasting earlier work, participants in groups outperformed solitary individuals, but only when payoff information was available. High-resolution visual field data and movement trajectories revealed that payoff visibility prompted adaptive recalibration of visual information flow, thereby enabling flexible adaptation. Computational models of fine-grained movement decisions and agent-based simulations showed that payoff information let participants dynamically identify and selectively respond to reliable social cues, thus unlocking the group's collective potential.