Dmitry Todorov

The paper develops a dual-rate model of motor adaptation to determine optimal practice schedules and shows reinforcement learning can recover these schedules under partial observability.

The paper reanalyzes Demarchi et al.'s data and shows that their findings on pre-activation during prediction can be explained by sequence structure bias, concluding no evidence of anticipatory predictive perception.

This study shows that aperiodic broadband power, not just beta oscillations, is linked to Parkinson's disease severity and is increased in the more affected hemisphere.

The paper investigates how error sensitivity changes during motor learning in different environmental conditions and finds it decreases in unstable environments and increases within stable ones.

This study uses machine learning to distinguish Parkinsonian rest tremor from voluntary hand movements by combining subthalamic and cortical activity data.