Economic games are used to elicit a social, conflictual situation in which people have to make decisions weighing self-related and collective interests. Combining these games with task-based fMRI has been shown to be successful in investigating the neural underpinnings of cooperative behaviors. However, it remains elusive to which extent resting-state functional connectivity (RSFC) represents an individual's propensity to prosocial behaviors in the context of economic games. Here, we investigated whether task-free RSFC predicts individual differences in the propensity to trust and reciprocate in a one-round trust game (TG) employing a prediction-analytics framework. Our results demonstrated that individual differences in the propensity to trust and reciprocity could be predicted by individual differences in the RSFC. Different subnetworks of the default-mode network associated with mentalizing exclusively predicted trust and reciprocity. Moreover, reciprocity was further predicted by the frontoparietal and cingulo-opercular networks associated with cognitive control and saliency, respectively. Our results contribute to a better understanding of how complex social behaviors are enrooted in large-scale intrinsic brain dynamics, which may represent neuromarkers for impairment of prosocial behavior in mental health disorders.
Keywords: Machine learning; Multivariate regression analysis; Reciprocity; Resting-state functional connectivity; Trust; Trust game.