Sleep Deprivation in Mice: Looking Beyond the Slow Wave Rebound

Abstract

Sleep is a fundamental process by which the brain achieves an optimal computational regime; it has been suggested that criticality is an appropriate theoretical framework in which such a process can be understood. Studying the critical and fractal dynamics of the brain involves modelling the nonlinearity of brain signals (such as EEG or ECoG) yielding, among others, the following metrics: spectral slope, spectral intercept, spectral knee, and normalized spectral entropy. Therefore, the present study investigates the nonlinear and critical dynamics of the brain in relation to sleep deprivation in mice, by comparing the sensitivity of the above-mentioned metrics to classical band- limited spectral indices. Mice were exposed to a 9 day-long sleep deprivation paradigm with baseline, sleep deprivation, and recovery phases. Spectral parameters were computed using the FOOOF algorithm. The results suggest that the classical approach (slow wave activity; 0.75-4.5 Hz) to neural signal processing differentiates between baseline sleep and rebound sleep only during the NREM phase. In contrast, the spectral slope and the spectral intercept both capture sleep deprivation related effects during REM and NREM episodes as well. This is particularly notable considering that the spectral knee is shifted towards higher frequencies, essentially rendering the spectral slope unreflective of slow wave activity - traditionally considered the biomarker of sleep homeostasis. Lastly, normalized spectral entropy fails to differentiate between baseline sleep and sleep following sleep deprivation in mice. These results support the sensitivity of fractal spectral parameters indexing the intricate balance between sleep and wake states.