Cortical activity upon awakening from sleep reveals consistent spatio-temporal gradients across sleep stages in human EEG

How does the brain awaken from sleep? Several studies have suggested that the awakening process occurs asynchronously across brain regions, but the precise nature of these changes and how they are reflected in human electroencephalography (EEG) remains unknown. Here, we recorded 1,073 awakenings and arousals with high-density EEG and mapped brain activity at a second-to-second timescale around movement onset using source modeling. We found that cortical activity upon awakening progressed along highly consistent spatial and frequency gradients. In awakenings and arousals from non-rapid eye movement (NREM) sleep, transient increases in low-frequency power preceded increases in high-frequency power by a few seconds, whereas awakenings from REM sleep were mainly characterized by increases in high-frequency power. Regardless of sleep stage, high-frequency changes were first seen in frontal and last in occipital and inferior-temporal cortical areas, whereas low-frequency changes in NREM sleep started in a centro-parietal "hotspot," progressed frontally, and reached occipital and inferior-temporal regions last. Finally, the presence of these spatio-temporal arousal patterns during sleep, before participants were awakened by sounds, was followed by lower sleepiness ratings upon awakening. These results indicate a consistent spatio-temporal EEG signature of the awakening process that likely reflects the structural organization of arousal systems. Importantly, a transient increase in slow EEG frequencies, which are normally associated with sleep, is inherent to the arousal process and functionally correlates with feeling more awake when awakening from NREM sleep. These findings have important implications for the interpretation of arousal signals and the detection of incomplete awakenings in sleep disorders.