An 80/20 cortical balance stabilizes information-rich dynamics

Abstract

The cortex maintains a remarkably consistent 4:1 ratio between excitatory and inhibitory neurons, yet the computational advantages of such an architecture remain poorly understood. Here, we demonstrate that this ratio optimally stabilizes a dynamical regime characterized by intermittent, burst-like activity, a state associated with maximal information capacity. Using a balanced spiking network model, we show that near the 80:20 ratio, this intermittent regime emerges robustly across a wide range of parameters and with low energy cost. These findings suggest that the canonical cortical E/I ratio is not arbitrary, but that it is functionally tuned to support efficient and flexible computation. Our results provide a dynamical explanation for a long-standing anatomical observation, bridging structural organization and information processing in neural circuits.