Highly Connected and Highly Variable: A Core Brain Network during Resting State Supports Propofol-induced Unconsciousness

AbstractLeading theories of consciousness make diverging predictions for where and how neural activity gives rise to subjective experience. The Global Neuronal Workspace theory (GNW) states that consciousness is instantiated through global broadcasting of information across the prefrontal-parietal regions, whereas the integrated information theory (IIT) postulates that consciousness requires the posterior cortex to produce maximally irreducible integrated information. As both theories seem to partially agree on that the neural correlates of consciousness (NCC) require globally integrated brain activity across a network of functionally specialized modules, it is not known yet whether brain regions with such functional configurations would align with the NCC distribution predicted by the GNW or the IIT. We scanned resting-state fMRI data from 21 subjects during wakefulness, propofol-induced sedation and anesthesia. Graph-theoretical analysis were conducted on awake fMRI data to search for the NCC candidates as brain regions that exhibit both high rich-clubness and high modular variability. Another independent dataset of 10 highly-sampled subjects were used to validate the NCC distribution at individual-level. Brain module-based dynamic analysis was conducted to estimate temporal stability of the NCC candidates. Alterations in functional connectivity and modular variability from awake to propofol-induced anesthesia were assessed to test the involvement of the NCC candidates in conscious processing. NCC candidates that are characterized by both high functional interconnectivity and high modular variability were identified to locate in prefrontal and temporoparietal cortices, which covered brain structures predicted by the GNW as well as the IIT. The identified NCC was found to mainly attributed to higher-order cognitive functions, and associated with genes enriched in synaptic transmission. Dynamic analysis revealed two discrete reoccurring brain states, which were characterized by their difference in temporal stability — the state dominated by the NCC candidates appearred to be temporally more stable than the other state predominately composed of primary sensory/motor regions, suggesting that the identified NCC members could sustain conscious contents as metastable network representations. Finally, we showed that the prefrontal GNW regions and posterior IIT regions within the identified NCC was differentially modulated in terms of functional connectedness and modular variability in response to loss of consciousness induced by propofol anesthesia. This work offers a framework to search for neural correlates of consciousness by charting the brain network topology, and provides new insights in understanding the distinct roles of the frontoparietal and posterior network in underpinning human consciousness.HighlightsStudies suggest that there are neural correlates of consciousness (NCC) we experience subjectively everyday. By overlapping regions with both high functional interconnectivity (rich-clubness) and high modular variability, we identified the putative NCC distributed in prefrontal and temporoparietal cortices, attributed to higher-order cognitive functions, and associated with genes enriched in synaptic transmission. We further revealed that the NCC members appeared to sustain conscious contents as metastable network representations in a reoccurring NCC dominant state. The identified NCC architecture was significantly modulated in terms of functional connectedness and modular varibility during propofol anesthesia, demonstrating its critical role in supporting consciousness. These findings testify to the NCC’s abilities in information integration and differentiation, and provide novel insights in reconciling the ongoing discussion of the contribution of anterior versus posterior regions in supporting human consciousness.