Diminished spatial dynamics and maladaptive spatial complexity link resting brain network disruption to cognition in schizophrenia

Resting-state fMRI studies increasingly emphasize the dynamic nature of brain networks. While most approaches examine temporal fluctuations in connectivity, we focus on the spatial dynamics and complexity at voxel level - how networks expand and contract, and change their structural complexity over time. Using dynamic independent component analysis (ICA), we investigate the hierarchical structure of the resulting time-varying spatial networks, from their broad periphery to their most active core. We combine this with fractal dimension (FrD) as a measure of a network’s spatial complexity and analyze temporal changes (dynamic flexibility) in a network and synchronized fluctuations between network pairs (fractal dimension coupling, FrDC). We refer to this approach as “dynamic spatial network complexity and connectivity (dSNCC)”. Using a combined cohort of 508 subjects (315 healthy controls, 193 schizophrenia patients), we found that schizophrenia is associated with higher mean FrD in several networks, suggesting more irregular patterns/boundaries and a disorganized network structure. Critically, patients showed significantly reduced dynamic flexibility, indicating their networks are “stuck” in a less adaptable state. This robust finding is evidenced by a synergistic loss of temporal standard deviations in both network volume and FrD across multiple networks and activity thresholds. This maladaptive complexity was associated with cognitive impairment, with several dSNCC measures showing significant associations with subject scores for processing speed, visual learning, and verbal learning. Higher complexity in these networks and more significantly, their reduced dynamic flexibility as seen in patients, were particularly associated with impaired performance. Furthermore, we found aberrant connectivity (FrDC) in schizophrenia, with certain network pairs exhibiting overly synchronized complexity changes. Our results demonstrate that dSNCC is a powerful tool for characterizing network dynamics and may potentially provide a measurable mechanism for maladaptation in schizophrenia, where the brain’s inability to fluidly change its complexity may contribute to cognitive deficits and symptoms like disorganized thought. These findings highlight the importance of studying the intrinsic spatial dynamic properties to reveal the fundamental principles of brain network organization in health and disease. Our work represents a significant leap in complex systems neuroscience and provides a novel, quantifiable biomarker framework highly relevant for understanding and targeting other complex disorders characterized by network dysfunction, such as Alzheimer’s disease, autism, or other mental health conditions.