INTERNAL REPRESENTATIONS OF FAMILIAR ENVIRONMENTS IN VISUAL CORTEX

How do we navigate familiar environments in darkness? We generate internal representations of our surroundings to find our route. It is known that our brain constructs internal models of the world through experience, and that these models are used to generate predictions of sensory inputs. However, whether and how these internal models of the environment are recruited to generate internal representations of our surrounds in the absence of sensory inputs remains largely unexplored.
We hypothesize that primary visual cortex (V1) encodes internal representations of previously experienced visual environments and that these representations are activated when they are behaviourally relevant. We recorded the activity of large populations of V1 neurons in awake mice navigating a well-learned virtual corridor in darkness. We found that V1 neurons reliably encoded the animal’s position and surrounds while they navigate in darkness. We also identified long-range inputs that convey this internal information to V1 neurons. Finally, we probed local circuit mechanisms that differentially support representations driven by sensory input versus internal signals, depending on visual availability. Our findings show that V1 can represent familiar environments in the absence of visual input, suggesting that sensory cortices may contribute to higher-order functions such as imagination and prospective planning by encoding internally generated models of the world.