Explicit reward stabilizes motor output by attenuating sensory prediction error driven learning

Abstract

Motor adaptation driven by sensory prediction errors (SPEs) is often regarded as an automatic, implicit process that operates independent of reward. However, in most past work, reward and performance outcomes (success / failure) have been intrinsically confounded, making it unclear whether explicitly delivered reward per se influences SPE-driven learning. Here, we used an error-clamp paradigm to dissociate reward from both error and task outcome, enabling us to directly test whether explicit reward modulates implicit adaptation. Participants performed reaching movements under clamped visual feedback that produced a constant SPE, while being instructed to ignore the cursor. In two experiments, reward was delivered when the unseen hand successfully intersected the reach target. In an eight-target task, reward did not alter the overall magnitude or time course of adaptation. However, trial-level analyses revealed that rewarded movements were followed by smaller trial-to-trial updates, reduced variability, and a cumulative suppression of adaptive adjustments. Consistent with this result, individuals who experienced reward more frequently exhibited less asymptotic learning. In our second experiment using a simplified, two-target task, these trial-level effects accumulated to produce robust reductions in both asymptotic adaptation and aftereffects in the rewarded groups. Across both experiments, longer streaks of rewarded trials predicted progressively weaker SPE-driven updating. Collectively, these findings demonstrate that implicit adaptation is not insulated from reward signals. Instead, explicit reward appears to attenuate sensitivity to SPEs, stabilizing motor output particularly when the task structure allows consistent action-reward associations. We conclude that motivational signals can gate the expression of error-driven motor adaptation. SIGNIFICANCE STATEMENTImplicit motor adaptation is often thought to be insulated from motivational signals such as reward. By dissociating explicit reward from task outcome as well as visual error using an error-clamp paradigm, we show that reward reliably attenuates trial-to-trial adaptive corrections and reduces their variability, despite identical visual errors on every trial. These effects accumulate across rewarded movements and under stable task conditions, lead to reduced overall adaptation. Our findings thus demonstrate that reward can influence the expression of sensory error-based learning, and that task structure plays a critical role in revealing interactions between reward and implicit adaptation.