Reframing "paradoxical" excitation: disentangling electroencephalogram complexity and entropy reveals resting-state
Derek Newman1, Charlotte Maschke2, George A Mashour3
1Montreal General Hospital, McGill University Health Centre, Montreal, QC, Canada; Integrated Program in Neuroscience, McGill University, Montreal, QC, Canada.
Background:
Propofol exposure can produce heterogeneous neural responses, from the expected suppression to transient paradoxical excitation. Electroencephalography (EEG) measures of signal complexity and entropy have emerged as reliable markers of consciousness, but different types of complexity and entropy measures are often conflated. We used Type I and II complexity measures on the complexity-entropy causal plane (CECP) to characterise divergent neural trajectories during propofol-induced loss of consciousness. We hypothesised that paradoxical excitation is reflected in both Type I and Type II complexity; that divergent trajectories on the CECP separate paradoxical excitation from suppression; and that baseline EEG complexity is associated with susceptibility to propofol.
Methods:
We analysed EEG data from two independent cohorts of healthy adults receiving propofol: the Chennu dataset (n=20), which included resting-state baseline, mild, and moderate sedation, followed by a recovery period; and the ReCCognition dataset (n=8), which used escalating infusions from baseline to deep sedation. For each participant and sedation level, we extracted Lempel-Ziv complexity (Type I) and statistical complexity (Type II) and projected them onto the CECP. Pearson correlations related baseline statistical complexity to changes in statistical complexity during moderate sedation; behavioural responsiveness; effect-site propofol concentration; and time to loss of consciousness.
Results:
At moderate sedation, participants who remained responsive showed paradoxical increases in Lempel-Ziv Complexity and decreases in statistical complexity, whereas unresponsive participants exhibited the opposite pattern. Baseline statistical complexity correlated negatively with both the change in statistical complexity (r=-0.88) and behavioural responsiveness, indicating that intrinsic brain dynamics influence individual susceptibility to sedation. CECP trajectories revealed a reproducible inflection point demarcating paradoxical excitation from suppression.
Conclusions:
Mapping EEG trajectories on the CECP bridges anaesthetic state transitions with underlying neural dynamics. Baseline EEG complexity indexes individual sensitivity to propofol, determining whether brain dynamics transiently enter excitation or direct suppression.
