Reinforcement-conflict based control: An integrative model of error detection in anterior cingulate cortex

The concepts of error detection and compensation play a key role in the studies of human brain behavior. In particular, studies of event-related brain potentials have discovered the presence of a component following errors named the error related negativity (ERN). According to the highly used theories of this field, the ERN can occur due to an error detection in reward prediction (the reinforcement learning theory) or conflict detection (the conflict monitoring theory). Since each of these theories has given numerous experimental validations, there is an urge to reconcile them. Although there have been efforts to integrate these two theories, the research in this field needs to be improved. The present study aims at integrating the above-mentioned theories by proposing a model that simulates the role of anterior cingulate cortex in performance monitoring and reinforcement-guided action selection. In this model, an executive control module uses the conflict signal and the reward prediction error to adjust the timing of the control and to regulate the strength of the control over other structures. The simulation results show that the proposed model can successfully produce the expected experimental results in a modified version of the Eriksen flanker task.