Experimental design. Participants took part in two MEG adaptation sleep sessions before the pretraining and post-training sleep sessions. On the third night, brain activity was measured via MEG and PSG (pretraining sleep session). On the fourth night, subjects underwent an initial training of the finger-tapping motor-sequence task that started 2 h before they went to bed. After the initial training, a post-training sleep session with MEG and PSG was conducted at night (post-training sleep session). After post-training sleep, a retest of the task was conducted to test whether performance on the motor task improved after post-training sleep relative to the initial phase of training on the motor task before the post-training sleep session. Anatomical MRI was measured in a separate session.
Localization of ROIs in the right hemisphere of a representative subject. a, Lateral view of the folded format of the brain. b, Medial view of the folded format of the brain. c, Lateral view of the inflated format of the same brain. d, Medial view of the inflated format of the same brain. e, Top view of the inflated format of the same brain. The same hemisphere is shown in a different format of the brain from different viewpoints. Green represents M1; red, PM; yellow, SMA; blue, pre-SMA. Scale bar, 1 cm.
Performance. a, Transition of performance in the initial training (blue) and the retest (red) sessions (mean ± SEM). b, The mean (± SEM) performance of the last two trials in the training session (blue) for the blue-hatched area of a and the mean (± SEM) performance of the retest session (red) for the orange-hatched area of a. These mean values were significantly different. *p < 0.05.
MEG oscillations in the contralateral SMA during slow-wave sleep. a, δ activity in the pretraining (blue) and the post-training (red) sleep sessions. **p < 0.01. b, Fast-σ activity in the pretraining (blue) and the post-training (red) sleep sessions. *p < 0.05. c, δ activity increase (x-axis) versus performance improvement value (y-axis). **p < 0.01. d, Fast-σ activity increase (x-axis) versus performance improvement value (y-axis). *p < 0.05. All the data samples are within 2 SD from average and do not correspond to outliers according to the Grubbs' test (α = 0.05, two-sided).
a, b, Correlation coefficients matrix of δ oscillation in motor areas (M1, SMA, pre-SMA, PM), prefrontal (PFC), and parietal (PAR) cortices during slow-wave sleep, in the pretraining (a) and post-training (b) sleep sessions. The lower and higher triangular parts of the correlation matrix are identical. The color scales indicate the range of z-transformed correlation coefficients. c, p values matrix of post hoc t tests. t tests were conducted on the correlation coefficients in the pretraining (a) and post-training (b) sleep sessions of δ oscillation for each pair of different brain regions during slow-wave sleep. The lower and higher triangular parts of the matrix are identical. An inverted triangle on the color scale shows the significance level (α = 0.05). d, Correlation coefficients between the SMA and prefrontal cortex (SMA-PFC) and the SMA and parietal cortex (SMA-PAR) for δ oscillation in the pretraining (red bars) and post-training (blue bars) sleep sessions. These were calculated from the cells within the white line shown in a (pretraining) and b (post-training). *p < 0.05.
Statistics of post-hoc two-tailed t test of sleep session (pretraining vs post-training) and Pearson product-moment correlation analysis (coefficient and p values) conducted for each frequency band during slow-wave sleepa
↵aThe former analysis corresponds to the first criterion (the activation in a ROI should be significantly different between the post-training vs pretraining sleep sessions), and the latter analysis corresponds to the second criterion (the modified activation in the ROI should be significantly correlated with performance improvement after post-training sleep).
↵*Statistically significant values for the first criterion (p < 0.05).
↵**Statistically significant values for the second criterion (p < 0.05).
↵aData are mean ± SEM. Latency to sleep onset (sleep latency) corresponds to the latency to the first sleep stage 2 from lights off. The percentage distributions of sleep stages were calculated based on total sleep time (total NREM and REM sleep stages). No significant differences were found between pretraining and post-training sleep sessions in any of the sleep variables (paired t test).