A shared multi-feature population code for sensory reliability across mouse visual cortex

Optimal use of sensory information requires the brain to represent uncertainty alongside sensory variables. Yet, the neural response properties that mediate reliability encoding remain undefined. Using large-scale two-photon calcium imaging of 140,000 neurons in primary visual cortex (V1) and higher visual areas (HVAs) during presentation of random dot kine-matograms, we identified a common coding strategy in which sensory reliability is represented via coordinated changes in response gain, selectivity, and variability. These response properties are expressed at both single neuron and population levels and are consistent across areas, with quantitative differences in magnitude that track retinotopic drive and functional specialization. Together, our results establish reliability encoding as a general cortical computation implemented through a shared population code, providing empirical constraints on probabilistic theories of perception and hierarchical models of visual processing.