State-Dependent Organization of Microscale Functional Circuitry in Visual Cortex

Abstract

How the cortex organizes directed functional interactions among individual neurons and how this microscale circuitry maps onto cell types and the underlying anatomy remains poorly understood. Here, we construct one of the largest single-neuron-resolution directed functional circuit maps in mouse visual cortex from calcium imaging of more than 57,000 active neurons across four visual areas and five cortical layers. We infer directed connections with a time-aware causal-inference method (CITS) that is specifically designed for neural time-series. In the co-registered electron-microscopy connectome, CITS-inferred pairs are 3.6-fold more likely to be synaptically connected than unconnected pairs, which is significantly more than Granger causality (1.1-fold) or lagged correlation (1.6-fold). We found that our inferred functional circuits organization aligns with synaptic connectivity at areal, laminar, and cell-type levels. Among intra-areal connections, anterolateral area (AL) had the highest density, and among inter-areal connections, the AL{leftrightarrow}rostrolateral area (RL) axis formed the strongest pathway. At the cell-type level, layer 6 recurrence dominated the aggregate network in low arousal, whereas layer 6-to-layer 5 output connections dominated in high arousal. Spatial extent expanded with inter-neuron distance in high arousal, for both excitatory-to-excitatory and excitatory-to-inhibitory connections. Finally, in stimulusdriven response prediction, neuron pairs with stronger functional connections exhibited more similar predictive performance in both high-arousal and low-arousal states, with performance varying by layer and cell type. Overall, our findings establish, at single-neuron resolution, the multi-scale organization of directed functional circuitry in mouse visual cortex across brain states.