Reactivation of reward representations associated with the reinforcement of behavior

Abstract

Reward-related representations are found distributed throughout many human subcortical and neocortical regions that support different neural processes. These representations get reinstated for different but related tasks. There is rising evidence that representations reactivate during learning to reinforce prior rewarding choices. However, there remains a critical lack of understanding for whether and how reactivations in humans can facilitate behavioral reinforcement. To investigate this, we recorded from the temporal lobe and prefrontal cortex with intracranial electrocorticography while human subjects learnt two-choice decisions across two different scenes. Crucially, subjects were able to straightforwardly reactivate knowledge gained in a prior scene to make decisions in the current scene, but only when the reward contingencies stayed the same between the two scenes. Using a Bayesian learner, we inferred reward expectation from choice behavior, and measured representations of these reward expectations in electrocorticography data. We found that the representations of reward expectation in the medial temporal lobe and orbitofrontal cortex were reactivated, only when the subjects could straightforwardly transfer knowledge between the two scenes. In the anterior temporal lobe, the reactivation of reward representations increased as learning increased, suggesting an area of plasticity during learning. This work presents a novel framework for measuring the reinforcement, or reactivation, of representations from human electrophysiology during the reinforcement of behavior. Significance statementThis study provides a framework for how the brain processes and reinstates reward-related information during the reinforcement of behavior. Using intracranial electrocorticography, the researchers found that reward information is encoded in distant brain regions, the medial temporal lobe and orbitofrontal cortex. These representations are dynamically reinstated across scenes when reward contingencies align and when knowledge can be transferred. Additionally, reward-related representations in the anterior temporal lobe become increasingly reinforced or reinstated as learning progresses, revealing an area of plasticity during learning. These findings offer new insights into the neural mechanisms underlying decision-making and highlight potential avenues for addressing conditions like addiction or depression, where reward processing and decision-making are disrupted.