Feedback-Feedforward Dynamics Shape De Novo Motor Learning

Abstract

Many real-world motor skills require continuous control under substantial sensorimotor delays, necessitating the coordination of reactive feedback and predictive feedforward control. How these complementary control processes emerge and interact during the acquisition of a novel continuous motor skill remains unclear. We addressed this question by examining how feedback and feedforward control evolve over five days of continuous visuomotor mirror reversal learning. To directly dissociate feedback- and feedforward-related contributions during learning, we combined frequency-domain analyses of continuous tracking with independent behavioral probes of online feedback responses (brief cursor perturbations) and predictive feedforward control (initial movement direction during point-to-point reaches). We found that learning exhibited both temporal and frequency-dependent shifts in the relative contributions of feedback and feedforward control. Early improvements were dominated by rapid online feedback corrections expressed primarily at low movement frequencies, whereas predictive feedforward control emerged more gradually and became increasingly important at higher frequencies, where delayed feedback is insufficient for accurate control. A recurrent neural network model, in which an adaptive cerebellar-like feedback controller provides a teaching signal to a recurrent cortical predictive controller, reproduced these key behavioral signatures, supporting a mechanistic framework in which feedback-driven error signals facilitate the gradual emergence of predictive control. Together, these findings demonstrate that continuous motor learning arises through distinct but interacting feedback and feedforward processes that co-evolve across practice under the constraints imposed by sensorimotor delays.