Dissociating value representation and inhibition of inappropriate affective response during reversal learning in the ventromedial prefrontal cortex

Abstract

Decision-making studies have implicated the ventromedial prefrontal cortex (vmPFC) in tracking the value of rewards and punishments. At the same time, fear-learning studies have pointed to a role of the same area in updating previously learned cue–outcome associations. To disentangle these accounts, we used a reward reversal-learning paradigm in a functional magnetic resonance imaging study in 18 human participants. Participants first learned that one of two colored squares (color A) was associated with monetary reward, whereas the other (color B) was not, and then had to learn that these contingencies reversed. Consistent with value representation, activity of a dorsal region of vmPFC was positively correlated with reward magnitude. Conversely, a more ventral region of vmPFC responded more to color A than to color B after contingency reversal, compatible with a role of inhibiting the previously learned response that was no longer appropriate. Moreover, the response strength was correlated with subjects’ behavioral learning strength. Our findings provide direct evidence for the spatial dissociation of value representation and affective response inhibition in the vmPFC.