Connectivity Logic of Dendritic Spines in Cortex: Increased Inputs and Ensemble Formation

Dendritic spines, small protrusions covering the dendrites of most neurons, are fundamental elements of synaptic connectivity, yet their network-level organization remains poorly understood. Here we leverage the large-scale MICrONS volumetric electron microscopy dataset of mouse primary visual cortex to explore the connectivity logic of dendritic spines across multiple spatial scales. Our analysis provides structural support for the ``connectivity and diversity'' hypothesis, showing that, for both excitatory and inhibitory neurons, dendritic spine density correlates with the number and the diversity of presynaptic partners. We further show that the preference for excitatory axons to form synapses on spines increases with distance from the soma. In addition, we find that spines serve as postsynaptic targets for high-output neurons (``out hubs''). We finally uncover an input/output correlation, in which neurons that share their input also preferentially target spines through their axonal output. These correlations persist across network sizes, consistent with a scale-invariant structural organization of cortical connectivity. We hypothesize that dendritic spines provide a structural substrate for network-level cooperation, by enabling distributed cortical activity to propagate through parallel synchronous chains, potentially amplified by nonlinear voltage responses at the spine and cellular level. These findings suggest that dendritic spines are key structural mechanisms for neuronal ensembles, attractor dynamics, and pattern completion.