Successful reinforcement history suppresses explicit and implicit error corrections.

Abstract

A skilled basketball player expects to make many shots. After a miss, they could make a larger correction because the miss was "surprising." Alternatively, they could make only a small correction since their average past performance was successful. Despite its relevance, past work has not elucidated whether or how a history of reinforcement influences explicit (cognitive) and implicit (involuntary) error corrections. Across three reaching experiments, we compared two competing hypotheses: 1) reward prediction errors ("surprise") modulate explicit and implicit error corrections, or 2) expected value (average reward) modulates explicit and implicit error corrections. Our experimental results and computational modelling support the idea that the expected value of a long-history of reinforcement modulates explicit error corrections. Further, we find that an immediate-history of reinforcement directly modulates implicit error corrections, which could not be explained by an independent task error model. Collectively, we find that a successful reinforcement history suppresses explicit and implicit error corrections.