The study of consciousness faces the challenge of objectively understanding subjective experience. Neuroscientific research explores the relationship between material processes and experience, and how to compare subjective experiences. This study proposes a novel approach using functional connectivity, measured as correlations between blood-oxygen-level-dependent (BOLD) signals in fMRI, as a proxy for the stream of consciousness. Previous research has linked functional networks to mental contents and cognitive states, with network disintegration correlating with unconsciousness. This research assumes that temporal variations in functional network configurations reflect variations in the stream of consciousness, with changes in connectivity patterns corresponding to phenomenological fluctuations. Existing studies often use clustering techniques to identify recurring connectivity patterns across individuals and conditions, which simplifies individual subjectivity. This study addresses this limitation by proposing an approach that highlights the dynamic neural correlates of individual subjective experience within a dynamical systems perspective.

Contemporary neuroscientific theories suggest that functional relationships between neurophysiological events are fundamental to consciousness, creating a unitary experience through information integration and dynamic meta-stable reorganization. A wealth of evidence links functional connectivity patterns to specific mental contents and cognitive states. Studies have shown that typical resting-state functional connectivity networks disintegrate in unconscious or sedated states. This research builds upon this foundation by explicitly adopting a framework that assumes the temporal variation of functional network configuration is directly related to the stream of consciousness. It draws on theoretical concepts from the spatiotemporal theory of consciousness and information integration theory (IIT), focusing on the brain's intrinsic time and space and the cause-effect repertoire of states.

The study uses fMRI data from four conditions: healthy awake controls, moderate propofol sedation, minimally conscious state (MCS), and unresponsive wakefulness syndrome (UWS). A second propofol anaesthesia dataset was used for reproducibility. The methodology involves dividing the data spatially into brain regions and temporally into overlapping time windows (2s). Correlation of timeseries within each window across brain regions generated time-varying connectivity matrices. Similarity between these matrices was calculated to create a Meta-Matrix (MM), representing an individual-specific, intrinsically defined, temporal landscape. Analyses examined the predictability of intrinsic dynamics, proximal temporal complexity (short-term transitions), and distal dynamic complexity (long-term state space). A temporal decay of similarity model (TDSM) was used to assess predictability. Proximal analyses used measures of central tendency, distribution breadth, and temporal complexity (sample entropy). Distal analyses focused on the off-diagonal triangle of the MM, representing similarities between distant connectivity patterns. Diffusion tensor imaging (DTI) data was also used to investigate the relationship between functional connectivity states and structural connectivity. Ordinal logistic regressions were used to analyze the relationship between dynamic characteristics and levels of consciousness.

The study found that in normal conscious states, transitions between connectivity states were faster on average, but displayed more consistent speed over time and were more unpredictable in the short term. Long-term descriptions of connectivity states were more complex. The relationship between functional and structural connectivity was also more complex in conscious states, indicating greater freedom of functional variation. Specifically, unconscious conditions showed greater similarity to the TDSM, indicating more predictable transitions. Short-term transitions were slower and less unpredictable in unconsciousness, and the wider state space was less complex. Analysis of the cortex, subcortex, and cerebellum revealed consciousness-relevant dynamics in all three, with the cortex and subcortex exhibiting strong effects. The complexity of the structure-function relationship also increased with consciousness across all three regions. The subcortical effects were particularly robust, suggesting a significant role for these regions in the dynamics of consciousness.

These findings demonstrate specific temporal network properties associated with awareness. The results align with the spatiotemporal theory of consciousness, showing consciousness-related autocorrelation properties and a complex dynamic repertoire. The approach approximates an empirical measure of the size of the cause-effect repertoire in IIT terms. While increased complexity aligns with the entropic brain hypothesis, the increased breadth of proximal transition distances suggests a functional organization that emerges with consciousness to reduce surprising states. The study expands on prior research by demonstrating that temporally linear, intrinsically-defined, non-clustered spatial connectivity patterns exhibit specific dynamic characteristics in awareness. The results also support a dynamical systems interpretation, suggesting that the underlying dynamic landscape of consciousness differs in awake versus unconscious states.

This study provides a novel approach to understanding the neural dynamics of consciousness. The findings show that consciousness is characterized by specific temporal patterns of connectivity, supporting previous theories and providing a more nuanced view of the neural correlates of awareness. Future research could explore online experience sampling or naturalistic paradigms to link connectivity patterns to specific experiential states, and combine this approach with clustering techniques to provide a more detailed description of dynamic state spaces. This framework is highly promising for a deeper investigation of the mind-brain relationship.

The study relies on resting-state fMRI data, which may not fully capture the complexity of consciousness during active cognitive tasks. The use of propofol sedation as a model of unconsciousness has limitations. Individual differences in brain anatomy and functional connectivity might have influenced the results. The interpretation of the Meta-Matrix as a direct reflection of the stream of consciousness requires further validation.