Neurocomputational mechanisms underlying the distinct motivational influences of reward and punishment on cognitive control.

Human motivation is fundamentally shaped by one's expectations of the reward they could earn for good performance or the penalty they would avoid for poor performance. However, the extent to which distinct brain regions are selectively associated with specific incentives and/or their corresponding influence on control strategy remains unclear. Using model-based fMRI and a novel multi-incentive control task, we observed distinct neural patterns by incentive valence, with ventral striatum and caudal subregion of dorsal anterior cingulate cortex showing greater sensitivity to rewards whereas inferior frontal gyrus and rostral subregion of dorsal anterior cingulate cortex showing greater sensitivity to penalties. Reward-sensitive regions were associated with increased efficiency (e.g., faster responding with moderate decreases in accuracy) whereas penalty-sensitive regions were associated with increased caution (e.g., slower responding with increased accuracy). We disentangled the global and selective influences of motivation on control processes and subjective experience, providing novel insight into the neurocomputational mechanisms of how effort is determined by expected reward and punishment.