Practice makes imperfect: restorative effects of sleep on motor learning.

Emerging evidence suggests that sleep plays a key role in procedural learning, particularly in the continued development of motor skill learning following initial acquisition. We argue that a detailed examination of the time course of performance across sleep on the finger-tapping task, established as the paradigm for studying the effect of sleep on motor learning, will help distinguish a restorative role of sleep in motor skill learning from a proactive one. Healthy subjects rehearsed for 12 trials and, following a night of sleep, were tested. Early training rapidly improved speed as well as accuracy on pre-sleep training. Additional rehearsal caused a marked slow-down in further improvement or partial reversal in performance to observed levels below theoretical upper limits derived on the basis of early pre-sleep rehearsal. This decrement in learning efficacy does not occur always, but if and only if it does, overnight sleep has an effect in fully or partly restoring the efficacy and actual performance to the optimal theoretically achieveable level. Our findings re-interpret the sleep-dependent memory enhancement in motor learning reported in the literature as a restoration of fatigued circuitry specialized for the skill. In providing restitution to the fatigued brain, sleep eliminates the rehearsal-induced synaptic fatigue of the circuitry specialized for the task and restores the benefit of early pre-sleep rehearsal. The present findings lend support to the notion that latent sleep-dependent enhancement of performance is a behavioral expression of the brain's restitution in sleep.