NOT HOW THE NETWORKS ARE, BUT HOW THE NETWORKS DANCE: FUNCTIONAL CONNECTIVITY DYNAMICS
Functional connectivity (FC) analyses shed light on interactions between different brain regions giving rise to cognition. However –exactly as cognition unrolls in time (the mind “flows”)– FC networks are not frozen but are continually reconfigured, in a way possibly reflecting the underlying complex dynamics of brain circuits. A variety of approaches to track such Functional Connectivity Dynamics (FCD) have been introduced, bearing promise to serve as better markers of resting state neural activity and of its healthy or pathological variations. Here we describe resting state FCD, extracted from fMRI imaging, as a stochastic sampling, a random walk exploration of the space of possible FC networks. We show that is exploration does not occur in an homogeneous manner but that the rate at which rs FC is being updated fluctuates in time, giving rise to episodes in which FC transiently stabilize (“knots”) and epochs in which it is wildly remodelled (“leaps”). We consider here two applications of our FCD flow metrics. We first consider how FCD is modified along healthy aging, showing that the flow of network reconfiguration gets slower and more “viscous” with aging. We then consider sleep deprivation, used as a cognitive challenge model for amnesic Mild Cognitive Impairment. We show that Sleep Deprivation can slow down FCD at the global brain level in a way that predicts subject-specific variation of motor reaction time. Furthermore the selective slowing down of specific sub-systems can predict performance variations in a sustained attention task. Such remodeling of FCD discloses qualitatively novel effects that could not be captured by conventional analyses of static FC.
