Cerebral characterization of sensory gating in disconnected dreaming states during REM sleep and propofol sedation using hd-EEG and fMRI

Consciousness, defined as any form of subjective experience, can often detach from the external environment, placing individuals in a state of sensory disconnection with preserved consciousness. Despite its prevalence, the neural mechanisms underlying sensory disconnection remain poorly understood due to the difficulties in behaviorally distinguishing between states of consciousness where sensory awareness is present (connected consciousness [CC]) and those where it is absent (disconnected consciousness [DC]).

Previous studies comparing wakefulness to sleep or anesthesia have been limited by the inherent differences between these physiological states and the intermittent nature of CC and DC within sleep and anesthesia. To address these limitations, this study employed auditory stimulation during REM sleep and propofol sedation, followed by serial awakenings to collect subjective reports on participants' state of consciousness and sound perception before waking. The auditory paradigm used was an oddball paradigm, where a regular sequence of sounds (standard) is interrupted by a sound of a different frequency (deviant or oddball).

Brain activity was recorded using high-density EEG during REM sleep and fMRI during propofol sedation. We compared brain activity of CC and DC participants — different phenomenological states within the same physiological state — to investigate the mechanisms behind sensory disconnection. Additionally, we compared brain activity during CC with that recorded during wakefulness, while participants listened to the same auditory paradigm, to quantify the impact of physiological state on sensory connection, as CC and wakefulness share sensory awareness but differ in the underlying physiology.

EEG analyses during REM sleep revealed distinct scalp-level differences between CC and DC participants. CC participants exhibited more pronounced event-related potential (ERP) components traditionally associated with consciousness, suggesting a more elaborate processing and differentiation of standard and deviant sounds. These ERP differences were accompanied by a decrease in delta power (1-4 Hz) and an increase in beta2 power (18-30 Hz) in the CC state. Dynamic Causal Modeling revealed that these scalp differences likely originated from distinct connectivity patterns between CC and DC states. Specifically, DC exhibited an overall reduction in feedback and feedforward connectivity within a temporo-parietal circuit encompassing the inferior parietal lobule and superior temporal gyrus.

The comparison between CC and wakefulness showed enhanced processing of deviant sounds in wakefulness, marked by increased amplitude of a late ERP component. This scalp difference was attributed to stronger connectivity during wakefulness compared to CC in a fronto-parietal circuit, involving the inferior frontal gyrus and inferior parietal lobule.

During propofol sedation, we observed unexpected activation increases in DC compared to CC in response to all sounds combined, primarily in temporal and pre-/post-central regions. In contrast, CC participants exhibited widespread decreases and focused increases in the precuneus, as well as a distinct network involving prefrontal regions, hippocampal gyrus, and middle occipital gyrus. This paradoxical pattern reversed for deviant sounds, where CC showed widespread activity increases in fronto-parietal and temporal areas, while DC showed minimal changes, potentially suggesting that deviant perception may be exclusive to CC states.

fMRI spectral analysis revealed increased high-frequency oscillations in DC in sensory and attention regions, potentially indicating higher slow-wave activity compared to CC participants in these regions. Interestingly, differences between CC and wakefulness emerged only in general auditory processing, while no differences were observed in deviant sound processing, suggesting similar processing of deviant sounds in CC during sedation and wakefulness.

The current work represents one of the first demonstrations that differential sound processing occurs between self-reported CC and DC within the same physiological state, providing distinct neural signatures of sensory disconnection during REM sleep and propofol anesthesia.