Cortical multi-area model with joint excitatory-inhibitory clusters accounts for spiking statistics, inter-area propagation, and variability dynamics

The primate brain uses billions of interacting neurons to produce macroscopic dynamics and behavior, but current methods only allow neuroscientists to investigate a subset of the neural activity. Computational modeling offers an alternative testbed for scientific hypotheses, by allowing full control of the system. Here, we test the hypothesis that local cortical circuits are organized into joint clusters of excitatory and inhibitory neurons by investigating the influence of this organizational principle on cortical resting-state spiking activity, inter-area propagation, and variability dynamics. The model represents all vision-related areas in one hemisphere of the macaque cortex with biologically realistic neuron densities and connectivities, expanding on a previous unclustered model of this system. Each area is represented by a square millimeter microcircuit including the full density of neurons and synapses, avoiding downscaling artifacts and testing cortical dynamics at the natural scale. We find that joint excitatory-inhibitory clustering normalizes spiking activity statistics in terms of firing rate distributions and inter-spike interval variability. A comparison with data from cortical areas V1, V4, FEF, 7a, and DP shows that the clustering enables the resting-state activity of especially higher cortical areas to be better captured. In addition, we find that the clustering supports signal propagation across all areas in both feedforward and feedback directions with reasonable latencies. Finally, we also show that localized stimulation of the clustered model quenches the variability of neural activity, in agreement with experimental observations. We conclude that joint clustering of excitatory and inhibitory neurons is a likely organizational principle of local cortical circuits, supporting resting-state spiking activity statistics, inter-area propagation, and variability dynamics.