A deep-learning-based explainable consciousness indicator (ECI) uses EEG responses to transcranial magnetic stimulation and resting-state EEG to separately quantify arousal and awareness. Tested during sleep (n=6), general anesthesia (n=16), and severe brain injury (n=34), ECI distinguishes states such as ketamine-induced anesthesia and rapid eye movement sleep, which combine low arousal with high awareness. Parietal brain regions are most relevant for these measurements. The indicator offers a way to disentangle the two components of consciousness across physiological, pharmacological, and pathological conditions.
During non-rapid eye movement sleep, conscious experiences are linked to reduced phase-locking at low frequencies (<4 Hz) and lower transitivity and clustering coefficient in delta and theta bands compared to unconsciousness, especially over parietal-occipital regions. No significant differences in Granger-causality patterns between frontal and parietal areas were found. These findings suggest that decreased local connectivity at low frequencies in posterior brain regions may indicate consciousness during sleep.