Spatially distributed and regionally unbound cellular resolution brain-wide processing loops in mice

Until recently, it has been possible to examine activity in the brain globally through regional averaging or locally at cellular resolution. These studies characterized regions as functionally homogeneous entities (e.g. V1 for extracting low-level visual features) or single neurons in a region as heterogeneously tuned (e.g. mixed selectivity). Here, we leverage the unprecedentedly dense IBL electrophysiological recordings in mouse brains during a sensorimotor decision-making task and computationally combine these to generate a global event-aligned high temporal precision cellular-resolution functional map. We find that individual neurons specialize in simple functions, and ordering neurons by functional similarity reveals at cellular and millisecond resolution the temporal arc of computations unfurling over the phases of a trial and across the brain, providing a way to interpret neural variance beyond sensory and motor responses. Functionally similar cells are highly dispersed across regions, and each region contains cells of nearly all response types. Functional tuning is barely predictable by cytoarchitectural boundaries or spatial coordinates, or vice versa. The primary regional distinction is between cortical and subcortical processing, with subcortical neurons carrying most of the internal computations of expert animals, from integration to decision making and movement initiation. At most, the brain exhibits regional structure in a distributional sense: the flatness of an area's distribution of different functional cell types is lowest in cortical regions, but there is no clear correlation between distributional specialization and anatomical measures of cortical hierarchy. Finally, the high temporal resolution of electrophysiological recordings and the inclusion of subcortical areas reveal the existence of a novel set of robust collective and distributed processing networks in the brain, which rapidly emerge and dissolve across phases of a trial. These results point toward the idea of spatially wide-ranging dynamical processing - winding through most brain regions and involving a subset of neurons in each - to carry out basic functions, and contrast with the idea that brain areas are functionally specialized units.