Conscious wakefulness is characterized by brain dynamics far from thermodynamic equilibrium, while states of reduced consciousness—such as deep sleep and anesthesia induced by propofol, ketamine, or ketamine plus medetomidine—operate closer to equilibrium. This conclusion comes from analyzing electrocorticography data from nonhuman primates and functional magnetic resonance imaging data from humans. Entropy production and the curl of probability flux in phase space reliably distinguished conscious from unconscious states. The findings establish nonequilibrium macroscopic brain dynamics as a robust signature of consciousness and offer a statistical mechanics framework for studying cognition and awareness.
A semi-empirical model combining fMRI data, structural connectivity, and anatomically-informed priors shows that brain states during the wake-sleep cycle are better described by multiple dimensions rather than a single continuum. The best fit used priors based on functionally coherent networks, dividing the cortex into regions with opposite dynamics: frontoparietal regions approached a noise-driven bifurcation from fixed-point dynamics, while sensorimotor regions approached a bifurcation from oscillatory dynamics. Sleep onset involved subcortical deactivation with low correlation, reversed in deeper stages. Periodic forcing simulating external perturbations identified key regions for wakefulness recovery. The model characterizes sleep as having diminished perceptual gating but latent capacity for rapid arousal.
Consciousness depends on brain activity that is far from thermodynamic equilibrium. Analyzing electrocorticography data from non-human primates during sleep and various anesthetics, and fMRI data from humans during deep sleep and propofol anesthesia, all states of reduced consciousness showed dynamics closer to equilibrium than conscious wakefulness. This was measured by entropy production and the curl of probability flux in phase space. Non-equilibrium macroscopic brain dynamics therefore serve as a robust signature of consciousness, offering a statistical mechanics approach to studying cognition and awareness.
The level of consciousness—how conscious someone is—is often measured by how similar their brain activity is to normal wakefulness. However, this approach misses important information about how stable that state is. Using computer models of the whole brain, the authors show that the stability of a conscious state—how easily it can be disrupted—provides additional, complementary information. They propose a new framework that sorts brain states by both their similarity to wakefulness and their stability, which helps distinguish between different types of unconsciousness: natural sleep, anesthesia, and brain injury. This framework offers a more complete way to characterize and differentiate states of consciousness.