Adaptive proximity to criticality underlies amplification of ultra-slow fluctuations during free recall

Ultra-slow fluctuations are a hallmark of spontaneous cortical activity. We examine the hypothesis that these unique dynamics arise from recurrent neuronal networks operating near a phase transition, a state characterized by ‘critical slowing down’. A further prediction of such dynamics is that a small modulation towards the critical transition should lead to specific amplification of slow fluctuations. Here, we relate this phenomenon to experimental findings using a simulation of a simple random recurrent network. Importantly, the model aligns with direct intracranial electroencephalography recordings from human visual cortex during both rest and visual free-recall, specifically replicating the observed enhancement of slow fluctuations during free recall. These simulations illuminate a simple and powerful mechanism underlying slow spontaneous fluctuations, while enabling the rapid transition between different spontaneous states. They propose that modulation towards criticality might be a universal strategy employed by cortical networks to engage in a spontaneous generative mode.