Predicting Ventral Striatal Activation During Reward Anticipation From Functional Connectivity at Rest.

Abstract

Reward anticipation is essential for directing behavior toward positively valenced stimuli, creating motivational salience. Task-related activation of the ventral striatum (VS) has long been used as a target for understanding reward function. However, some subjects may not be able to perform the respective tasks because of their complexity or subjects’ physical or mental disabilities. Moreover, task implementations may differ, which results in limited comparability. Hence, developing a task-free method for evaluating neural gain circuits is essential. Research has shown that fluctuations in neuronal activity at rest denoted individual differences in the brain functional networks. Here, we proposed novel models to predict the activation of the VS during gain anticipation, using the functional magnetic resonance imaging data of 45 healthy subjects acquired during a monetary incentive delay task and under rest. In-sample validation and held-out data were used to estimate the generalizability of the models. It was possible to predict three measures of reward activation (sensitivity, average, maximum) from resting-state functional connectivity (Pearson’s r = 0.38–0.54 in validation data). Especially high contributions to the models were observed from the default mode network. These findings highlight the potential of using functional connectivity at rest as a task-free alternative for predicting activation in the VS, offering a possibility to estimate reward response in the broader sampling of subject populations.