Task Engagement Gates Interareal Communication Geometry in the Mouse Thalamocortical-Midbrain Visual Circuit

Visual processing occurs within hierarchically arranged brain circuits. Current theories primarily explain alterations in population geometry via localized modifications in gain, firing-rate statistics, or recurrent dynamics. However, they fail to consider how interareal coordination interacts with local population geometry to limit downstream population states.

By analyzing task engagement versus a passive condition, researchers examined this coordination in Neuropixels recordings covering the mouse visual thalamocortical-midbrain circuit. Engagement led to a reduction in network activity, response participation, and dimensionality throughout the hierarchy. To explain this circuit-level organization, the researchers created a theoretical framework where afferent population geometry interacts with local recurrent dynamics to restrict the possible dynamics of downstream populations.

Across the thalamocortical stages, population-wide afferent statistics accurately predicted downstream activity and dimensionality. At the cortex-midbrain junction, engagement reshaped interareal communication geometry. These findings identify interareal input geometry as a crucial constraint on neural population dynamics and reveal a general principle whereby behavioral engagement limits neural state spaces across distributed visual circuits.