Abstract
Research in reward-based decision-making showed that humans and animals dynamically modulate learning rate according to their belief about environmental change (volatility) and surprise about observation. Recent evidence also suggests that neuromodulator noradrenaline (NA) signals volatility and surprise. Despite the rich anatomical evidence suggesting the potential influence of NA on the motor system, it is still elusive how NA and volatility/surprise affect human motor learning. To address this issue, we ran a series of experiments in which we simultaneously tracked the pupil diameter, a non-invasive proxy for the central NA/arousal activity, during a short-term force-field reach adaptation paradigm. A sudden increase in error due to the force-field resulted in increased pupil dilation during movement followed by an elevated baseline diameter in the following trials. These online and offline pupil responses showed a consistent pattern with surprise and volatility simulated by a recent computational model which dynamically adjusts learning rate according to volatility estimated from experienced error (surprise). However, unlike the model’s prediction, when participants experienced frequent reversals in force-field, the size of pupil responses rapidly diminished regardless of large errors induced by reversals. We further confirmed that the causal manipulation of participants’ arousal by task-irrelevant auditory stimuli modulated the single-trial motor learning rate. Collectively, these results provide a compelling evidence that NA/arousal system acts as a common modulator of learning rate in both cognitive and motor domains. Rapid reduction in pupil responses at reversals suggests that error sensitivity for computing current environmental uncertainty and surprise is also highly dynamic.
Highlights
We assessed the dynamics of pupil-linked arousal system during short-term force-field adaptation paradigm in humans.
Pupil dilated in both online and offline manner to movement error induced by force perturbation.
The pupil responses showed a habituation-like reduction in the sensitivity to error size when faced with multiple reversals in perturbation.
Arousal manipulation modulated single-trial motor learning in a baseline-dependent manner.
Competing Interest Statement
The authors have declared no competing interest.