Abstract

The neural and computational mechanisms that distinguish cooperative from competitive strategies in risky decision-making remain incompletely understood. In this study, we combine frequency-specific prefrontal inter-brain synchrony (IBS) measured via functional near-infrared spectroscopy (fNIRS) hyperscanning with reinforcement learning modeling to examine how social context shapes dyadic choice. Sixty female dyads performed cooperative or competitive variants of a modified Iowa Gambling Task (IGT). Behaviorally, competitive pairs achieved significantly higher cumulative earnings than cooperative pairs. Reinforcement learning analyses indicated that the Outcome Representation Learning (ORL) model provided the best account of behavior. Cooperative dyads showed increased sensitivity to win frequency (βfre), suggesting a tendency to favor frequent but suboptimal gains. In contrast, competitive dyads adopted more flexible strategies that were less dependent on reward frequency. Neuroimaging results revealed dissociable frequency related patterns. Ultra-low frequency coupling in the dorsolateral prefrontal cortex (DLPFC) within the range of 0.015 to 0.017 Hz was associated with goal directed control and higher earnings. Higher frequency coupling in the frontopolar cortex (FPC) within the range of 0.340 to 0.381 Hz was associated with opponent monitoring and sustained competitive engagement, and was reduced during cooperation, consistent with reduced individual responsibility. These findings support a dual pathway account in which competition engages both control and monitoring processes to facilitate performance, whereas cooperation may incur performance costs through socially shaped learning biases. The results provide mechanistic insight into social decision making and identify candidate neural markers for adaptive behavior in interactive contexts.