The Neurodynamic Computational Model Reveals Abnormal Multistable Spatiotemporal Patterns and Mechanisms in Schizophrenia

In this study, we established a link between data-driven dynamic organizational modeling (including co-activation pattern and hidden Markov model) and theory-driven neurodynamic mechanistic modeling (Wilson-Cowan-type models) of brain based on the concept of attractors. First, we validated, through fMRI signals simulated by neurodynamic models containing multiple attractors, that the co-activation pattern approach is more effective than hidden Markov model in extracting system attractors from functional imaging data. We found that, compared to normal controls, schizophrenia patients exhibited a significantly reduced match between individual- and group-level attractors, with this match correlating strongly with general symptoms. Lastly, we validated that the spatial characteristics of schizophrenia attractors may be influenced by a reduction in global interactions. These findings reveal the multilayered associations between brain global interactions, multistable pattern abnormalities, and general symptoms in schizophrenia, offering new insights into its pathophysiological mechanisms.