Universal scaling in neural activity: a renormalization-group perspective

The brain is in a state of perpetual reverberating neural activity, even in the absence of specific tasks or stimuli. Shedding light onto the origin and functional meaning of such an activity is essential to understand how the brain transmits, processes and stores information. An inspiring yet still controversial conjecture proposes that some statistical features of empirically-observed neural activity can be understood by assuming that brain networks operate in a dynamical regime close to the edge of a phase transition, and that the resulting critical behavior, with its concomitant scale invariance, provides neuronal networks with crucial functional advantages. To shed further light on these issues, here we analyze activity recordings of thousands of individual neurons across regions in the mouse brain. We employ a variety of state-of-the-art theoretical approaches, including a recently-proposed phenomenological renormalization group approach, as well as methods that allow us to infer the overall dynamical state of a neural population from measurements of pairwise covariances. By synergetically combining these ideas, we find strong signatures of scale-invariance which turns out to be quite robust or "quasi-universal" across brain regions, providing evidence that these areas may operate ---to a greater or a lesser extent--- at the edge of instability.