Intrinsic ignition-based propagation networks reveal hierarchical propagation pathways of spontaneous activity in the human brain

Abstract

Directed communication across brain regions is fundamental to human brain function, yet how local spontaneous activity links to directed whole-brain propagation patterns remains elusive. Prior studies have characterized local propagation events and global wave-like dynamics, but a framework linking local events, interregional propagation pathways, and canonical large-scale propagation patterns remains lacking. Using resting-state functional MRI data from the Human Connectome Project 7T cohort and an intrinsic ignition framework, we quantified cascades of suprathreshold activity events and constructed directed propagation probability networks. We found spatially heterogeneous regional propagation preferences, with high outgoing propagation preferences predominantly localized in somatomotor, visual, and default-mode regions. At the functional system level, propagation modules broadly separated visual, somatomotor, dorsal attention, and ventral attention networks from association systems. Regional stepwise propagation pathways were aligned with the principal functional gradient, and greater hierarchical separation between regions was associated with longer propagation distance. These pathways formed two canonical propagation patterns linking primary and association systems, with somatomotor regions preferentially initiated in bottom-up propagation, default-mode regions in top-down propagation, and attention networks occupying intermediate positions. Notably, the two patterns were not simple mirror images, with top-down propagation showing substantial deviations within somatomotor and visual systems. These findings were replicated in an independent cohort. Moreover, individual propagation architecture predicted cognitive performance and tobacco-use behavior. Collectively, our findings provide cross-scale insights into spontaneous activity propagation in the human brain by bridging local ignition events with hierarchical whole-brain propagation pathways and behaviorally relevant individual differences. Significance Spontaneous brain activity is often quantified by the temporal synchronization between regions, but this approach does not show how local activity propagates across the brain. We used an interpretable ignition framework to link local spontaneous events and their cascades to directed whole-brain propagation patterns. This approach revealed outgoing propagation preferences of primary and default-mode regions, a modular architecture separating primary and attentional systems from association systems, and two canonical propagation patterns aligned with the brain’s functional hierarchy. These two patterns were not simple mirror images, indicating partially distinct bottom-up and top-down routes. Individual propagation architecture also predicted cognitive performance and tobacco-use behavior. These findings provide a path-based view of spontaneous brain dynamics.