(a) Top panel: Mean EMG power locked to laser onset shows that EMG power decreases significantly during unilateral TRN stimulation in freely behaving mice (n = 315 trials, 8 sessions, 2 mice). Bottom panel: Mean smoothed motion (6.67 s moving average) detected in video: animals’ motion decreases significantly during optogenetic stimulation (n = 421 trials, 7 mice). (b) Mean change in arousal state during TRN activation: mice spend significantly more time in non-REM sleep and significantly less time in the awake state (n = 560 trials, 3 mice). Stars indicate significant effects at α = 0.05. (c) Individual trial correlation shows that the decrease in EMG power is correlated with the TRN-induced increase in EEG delta power (n = 315 trials, 2 mice). (d) Delta power increases in VGAT-Cre mice expressing ChR2 during TRN stimulation, whether awake or in NREM at time of stimulation. In VGAT-Cre mice expressing halorhodopsin, TRN inhibition has no effect in awake mice, whereas it decreases the delta power that is present in sleeping mice. N = 3 mice expressing ChR2 (160 wake trials; 192 NREM trials), n = 3 mice expressing Halo (459 wake trials; 211 NREM trials), stim. duration = 5 s. All recordings were in freely behaving mice. Dots show mean power +/- std. err; stars indicate a significant effect of the laser on the median power, computed with the Wilcoxon signed-rank test. (e) Cortical recordings in VGAT-ChR2 mice (n = 186 trials, 3 mice). During isoflurane anesthesia, the slow waves appear to be saturated and are not increased by TRN stimulation. Instead, broadband power decreases, suggesting a shift in dynamics that favours the inactivated state.