Brain-wide pathways for object identity and location are dominated by internal strategies

It is long thought that higher-order sensory processing is divided into two specialized cortical streams that encode in parallel either the identity of an object or its location (i.e., ‘what’ and ‘where’ streams). Here, using the mouse whisker system, we demonstrate an existence of two alternating brain-wide (beyond cortex) subnetworks that are primarily driven by internal strategies, rather than external parameters. We combine simultaneous brain-wide neuronal recordings in male mice trained to identify or locate a certain stimulus. We find that mice deploy either an active search or a passive sensation strategy during task performance. These strategies respectively drive two distinct brain-wide subnetworks, frontal and posterior related to an active and passive strategy respectively. Using a computational framework, we show that the posterior subnetwork encoded additional internal strategic parameters such as trial history, the first task of the day, and training history. When discarding the dominance of the frontal subnetwork, specific areas within the posterior subnetwork encoded meaningful task information regarding object identity and location. Integrated together, these two subnetworks may comprise optimal sensory integration. Neural mechanisms underlying optimal sensory integration are not fully understood. Here authors show that there exist two subnetworks spanning across many cortical and subcortical areas when the mouse performs whisker-dependent ‘what’ and ‘where’ tasks. Furthermore, the recruitment of either frontal or posterior subnetwork is determined by the internal behavioral strategy of the mouse.