Decoding adaptive performance

Background:
The plasticity in the sensorimotor system enables the brain to adapt its internal models of movements, according to sensory feedback signals, to refine motor performance. In the electroencephalography (EEG) signal, beta 114band activity (BBA; ~13-30Hz) plausibly reflects sensorimotor processing in the cortex and therefore have been implicated in several aspects of motor learning and motor adaptation. Recent studies have demonstrated an increase in BBA during REM sleep and linked this increase to the sleep-mediated model updating processes. In this study, we elaborate on the role of sleep in the adaptation of fine-motor skills.

Methods:
To disentangle the role of sleep in motor adaptation, we employed a novel fine-motor task, i.e. typing on a mirrored keyboard. To this end, we trai-ned human experts in touch-typing on the regular keyboard, to type on a mirrored keyboard and measured their performance before and after a retention interval of either a full night (~8h) of sleep with polysomnography (PSG) or a similar period of wakefulness. In total, we recruited 33 experts in touch-typing (N =16 for the sleep group and N =17 for the wake group) in a randomized, between-subjects design. Participants had to type eighteen five-letter, German words on a regular as well as a mirrored keyboard as rapidly and accurately as possible. We trained a linear discriminant analysis (LDA) classifier to decode i) correct from incorrect trials as well as ii) Regular from mirrored-typing trials based on pre- as well as post-movement EEG activity.

Results:
We show that brain activity in the 1s pre-movement period predicts typing performance. Specifically, the classifier was able to decode correct from incor-rect typing trials in both regular and inverted typing conditions. Interestingly, following a period of sleep, the decoding accuracy for inverted typing incre-ased significantly, with the change in decoding accuracy correlating with the change in the accuracy of typing on the mirrored keyboard, suggesting a role for sleep in optimizing adaptive behaviour. Confirming such observation, The accuracy of a classifier trained on BBA in the post-movement periods, presu-mably implicated in the process of model updating, increased significantly after a period of sleep but not wakefulness. However, sleep did not influence the decoding accuracy between regular and inverted typing trials.

Discussion:
Our results demonstrate a role for sleep in optimizing motor adaptation processes. Moreover, these findings suggest that post-movement BBA orchestrates memory processes essential for the adaptation of motor behaviour with such processes recommencing during subsequent sleep.