Fig. 4: Network eccentricity in various states of consciousness.

a Seven pre-defined functional networks include the visual network (VIS), somatomotor network (SMN), dorsal attention network (DAN), ventral attention/salience network (VAN), limbic network (LIM), frontoparietal network (FPN), and default-mode network (DMN). Schematic illustration of network position in a 3D gradient space. Diffusion map embedding was applied to the group-averaged functional connectivity matrix in conscious baseline and healthy controls (n = 193; pooled across all datasets). Gradient eigenvector loading values (mean ± SD across areas within a given network) are shown for each network. Dashed lines indicated the relative distance from global centroid to each network’s centroid along each axis. b Violin plots (median: solid line; quantiles: dash line) of network eccentricity, measured by the Euclidean distance of a given network’s centroid to the global centroid, across all conditions. Bayesian Paired Samples T tests (two-tailed) were performed for propofol deep sedation (PDS; n = 26), propofol general anesthesia (PGA; n = 23), and ketamine anesthesia (KA; n = 12) against their own baseline conditions (BL; n = 26, n = 23, n = 12, respectively). Bayesian Independent Samples T tests (two-tailed) were performed for unresponsive wakefulness syndrome (UWS; n = 7), schizophrenia (SCHZ; n = 44), bipolar disorder (BD; n = 49), and attention deficit hyperactivity disorder (ADHD; n = 39) against their own healthy control groups (HC; n = 16, n = 116, n = 116, n = 116, respectively). Asterisk indicates BF₁₀ > 10. Detailed statistics are provided in Supplementary Data 1. Source data are provided as a Source Data file.