Resting-state dissection

The brain activity measured by functional magnetic resonance imaging (fMRI) is linked to the cognitive output. This type of activity is considered to be the top layer of processing and subsequently, it is considered to make up the ‘brain states’ that emerge from network interactions (McIntosh, 2000). Using empirical blood-oxygen-level-dependent (BOLD) fMRI data during rest (HCP dataset), this study investigates resting state (RS) dynamics. Our framework follows the hypothesis that there exists Structured Flows on the Manifold (SFM) that govern the dynamics of RS, akin to behavior dynamics (Pillai & Jirsa, 2017).Our analysis uncovers two distinct functional connectivity (FC) states: low coherence state (lcs) and high coherence state (hcs), characterized by low and high levels of fluidity, respectively. Therefore, during lcs the exploration of the RS manifold is limited, but hcs entails extensive exploration.We employ a novel method to probe dynamics at the network level, extracting numerical potential functions to elucidate the underlying energy landscape and probability density functions for both states. Utilizing these observables, we fit a Bayesian hierarchical model, revealing manifestation of bistability at both the network and state levels during hcs, contrasting with monostability during lcs (Figure). These findings deepen our understanding of brain dynamics during rest, highlighting the role of structured flows and dynamics during rest.Figure: Fitted potential (V) and probability density functions (f) for lcs (blue) and hcs (orange) at the networks and states level.