Fig. 6: EEG exponent tracks the dynamics of sleep architecture.

A Temporally resolved estimates of the spectral exponent, overlaid on top of the sleep staging and spectral plot for the entire night, taken from an example subject. B Epoch-by-epoch exponent values from across the night for each sleep stage. Notably, Wake and N1 show an increase in the exponent as the night progresses, while N3 and REM show a decrease, and N2 shows no significant change. C Stage-specific exponent values split across the quartiles of the night, where each dot represents a subject. The exponent shows no significant differences across the quartiles, except for N1 where it increases and REM sleep where it decreases significantly across the night. D–F Time-resolved estimates of the spectral exponent during sleep stage transitions. Each transition is compared to a control period of adjacent epochs when no change in sleep stage occurred. D Sleep stage transition from N1, showing an increase in the exponent in the transition from N1 to N2 sleep stages, which is significant starting 24 s after transition. E Sleep stage transitions from N2 to either N1 or N3. Post-transition from N2 to N1, the exponent significantly decreased after 18 s. Conversely, transitioning from N2 to N3 led to an increase in the exponent, this change was significantly different from the period of uninterrupted N2 sleep during a short time window (14–30 s). F Transitions from NREM (N2 and N3) to REM sleep, showing a significant decrease in exponent starting 18 s after the transition. All exponent values in this analysis reflect the exponent of the knee model (fit range: 1–45 Hz). We used EEG data from 17 healthy participants during overnight sleep. Vertical dashed lines in (D–F) indicate the transitions between stages. Horizontal lines indicate significant clusters.