Probabilistic flow in brain-wide activity

Abstract

Patterns of low frequency brain-wide activity have drawn attention across multiple disciplines in neuroscience. Brain-wide activity patterns are often described through correlations, which capture concurrent increases and decreases in neural activity. More recently, several groups have described reproducible temporal sequences across the brain, illustrating precise long-distance control over the timing of low frequency activity. Features of correlation and temporal organization both point to a systems-level structure of brain activity consisting of large-scale networks and their mutual interactions. Yet a unified view for understanding large networks and their interactions remains elusive. Here, we propose a framework for computing probabilistic flow in brain-wide activity. We demonstrate how flow probabilities are modulated across rest and task states and show that the probabilistic perspective captures both intra- and inter-network dynamics. Finally, we suggest that a probabilistic framework may prove fruitful in characterizing low frequency brain-wide activity in health and disease.