Critical scaling of novelty in the cortex

To guide behavior in uncertain environments, the brain must rapidly detect novel or unexpected events. The neocortex, involved with complex perception and decision-making, is thought to contribute to this computation, but underlying mechanisms are poorly understood. Here, we test how a few unanticipated action potentials influence local circuitry in the resting mouse visual cortex. Using targeted holographic stimulation, we evoked sparse "surprise" spikes in single pyramidal neurons and monitored their effects on hundreds of neighboring cells with 2-photon imaging. These novel spikes, distinct from the cortex's ongoing large-scale activity fluctuations, produced strong, transient recruitment, following a power-law with slope 0.2-0.3, indicating that single neurons can mobilize large fractions of the surrounding network. Ongoing activity was dominated by neuronal avalanches, highly variable, scale-invariant spike cascades characteristic of systems near criticality. Yet, the information regarding the origin of our perturbations remained reliably identifiable and distributed across most of the observed network, as shown using machine-learning classifiers. Cortical network simulations confirmed that the measured scaling and distributed information matches predictions for systems operating near criticality. These results demonstrate two hallmarks of criticality, avalanche organization and amplified responses to small perturbations, suggesting that critical dynamics enhance the cortex's ability to detect novel events.