Stability of dynamic functional architecture differs between brain networks and states

Stable representation of information in distributed neural connectivity is critical to function effectively in the world. Despite the dynamic nature of the brain’s functional architecture, characterizing its temporal stability has been largely neglected. Here we characterized stability of functional architecture for each brain voxel by measuring the concordance of dynamic functional connectivity (DFC) over time, and explored how stability was modified by movie watching. High-order association regions, especially the default mode network, demonstrated high stability during resting state scans, while primary sensory-motor cortices revealed relatively lower stability. During movie watching, stability in the primary visual cortex was decreased, which was associated with larger DFC variation with neighboring regions. By contrast, higher-order regions in the ventral and dorsal visual stream demonstrated increased stability. The distribution of functional stability and its modification describes a profile of the brain’s stability property, which may be useful reference for examining distinct mental states and disorders.