Fig. 1: The framework designed for examining the complexity of the psychotic brain.

The data-driven references for brain networks were derived from a large cohort of subjects using group-level spatial independent component analysis (GICA), as illustrated in A. Spatial networks and their corresponding time courses were constructed for subjects with NC, SZ, and BP using spatially constrained ICA, based on previously established group-level intrinsic connectivity networks, as shown in B. To assess spatiotemporal complexity in the psychotic brain, complexity measures were applied to its activity, aiming to gain a better understanding of its specific spatiotemporal complexity patterns and preferences. Additionally, information interactions within brain networks are not limited to isolated (order = 0) or pairwise (order = 1) relationships. Instead, high-order interactions (order≥2) also play a significant role, highlighting the complexity of brain network dynamics, as illustrated in C. In this study, we employed integrated information decomposition, which is based on partial information decomposition (with the information distribution lattices for the triplet case presented to demonstrate the methodology for decomposing mixed information into entropy, redundant, synergistic, and total correlation components), to assess these high-order interactions within the brain.