A cortical gradient of distance to criticality governs large-scale resting-state fMRI dynamics

A longstanding puzzle in cortical research is how the cerebral cortex, having largely uniform interconnected architecture, gives rise to such diverse yet highly structured spatiotemporal activity. Here, we propose that local cortical networks’ distance from criticality (DTC) provides a unifying principle related to this conundrum. Analyzing resting-state fMRI BOLD signals and leveraging simple network models of randomly connected recurrent units, we show that DTC robustly explains key dynamical features, in particular, local power spectra and functional connectivity, across the full set of 360 cortical areas. Our analysis shows that a rank-order distribution of DTC values is highly conserved across subjects. Moreover, the empirical analysis of cortical slow dynamics and its fitted network simulations demonstrate similar power-laws across hierarchies of the cortical sheet. These results suggest that recurrent neuronal networks, operating close to criticality, can generate a remarkably rich dynamical repertoire which fit the entire range of experimentally observed cortical dynamics. Our findings underscore the importance of DTC as a powerful, fundamental generator underlying the spectrum of diverse cortical dynamics.