Reward-strengthened task-irrelevant stimulus-response associations are not modulated by working memory load: A diffusion model analysis

Previous studies have established the multifaceted influences of reward on cognitive control. One of the most remarkable findings is that in the conflict tasks, selective reinforcement on congruent or incongruent trials can modulate the conflict effects. However, the mechanism of the selective reward effect remains to be elucidated. Here in the present study, we investigated the underlying mechanism using the extensions of the diffusion model for conflict tasks (DMC). In Experiment 1, we replicated the previous finding, showing that the Simon effect decreased and even reversed when reward selectively followed incongruent trials, compared with selectively reinforcing congruent trials. Our model fitting results showed that, instead of adaptively adjusting the target processing efficiency, participants learned specific irrelevant stimulus-response (S-R) associations through reinforcement, which triggered the starting point bias at the decision stage. Next, to further investigate how working memory load influences the effect of selective reward, we varied the stimulus set size in task-relevant dimension to manipulate working memory load in Experiment 2. Critically, the model fitting results showed no influence of WM load on the starting point bias, suggesting that the task-irrelevant S-R associations acquired through reinforcement learning were not modulated by WM load. Altogether, the present study suggested that individuals can reconfigure cognitive control through reinforcement learning of irrelevant S-R associations, which was not constrained by the memory resources.