Global and Network-Specific Patterns of Structure-Function Coupling across Structural and Effective Connectomes

The relationship between structural connectivity (SC) and large-scale brain dynamics remains a central question in neuroscience. Because undirected functional connectivity discards directional information, we investigated structure–function coupling by comparing diffusion MRI-derived SC with directed effective connectivity (EC), estimated via sparse Dynamic Causal Modelling of resting-state fMRI. Using data from 142 healthy adults (Human Connectome Project), intra-hemispheric cortical SC and EC matrices were aligned to a common parcellation. Coupling was evaluated globally, node-wise, and at the network level using mean-absolute strength ratios, Spearman rank correlations, and topological overlap. Results demonstrated reliable bidirectional SC–EC coupling that strengthened at more stringent thresholds. While SC→EC coupling exhibited greater mean-strength magnitude, EC→SC coupling revealed a clear spatial hierarchy across the cortex. Unimodal systems exhibited the strongest structural grounding and topological overlap. Conversely, higher-order association networks showed weaker correspondence, and cross-network topological overlap remained consistently low. These findings suggest that structural and effective connectivity are systematically related, but that this relationship varies across cortical systems. Structural constraints are strongest in unimodal cortex and weaker in higher-order networks, where effective dynamics extend beyond the strongest direct structural pathways.