Cell-Type Clusters in the MICrONS Connectome Reveal Hidden Organizational Principles of the Mouse Visual Cortex and Possible Candidate Substrates for Elementary Perceptual States

Abstract

How neural structure constrains cortical dynamics and perceptual experience remains a fundamental open question. The MICrONS connectomics dataset enables direct examination of this problem by combining dense ultrastructural reconstructions with functional measurements across a mesoscale volume of mouse visual cortex. Here, I test a prediction arising from field-based models of perception: that subsets of cortical neurons should exhibit rare but reproducible structural similarity and spatial clustering. Using exhaustive pairwise NBLAST comparisons across tens of thousands of reconstructed neurons, I quantified morphological similarity among excitatory populations spanning layers 2/3, 5, and 6, including extratelencephalic, intratelencephalic, and near-projecting cells. High similarity scores were exceedingly rare, occurring more than 8-10 standard deviations above population means. Despite this rarity, structurally similar neurons formed discrete, spatially coherent clusters across primary and higher-order visual areas, often aligning across cortical layers. These findings reveal a previously unrecognized level of mesoscale structural organization and provide an empirical foundation for linking cortical microarchitecture to field-based models of visual experience.