Sensory input to cortex encoded on low-dimensional periphery-correlated subspaces

Abstract

AbstractAs information about the world is conveyed from the sensory periphery to central neural circuits, it mixes with complex ongoing cortical activity. How do neural populations keep track of sensory signals, separating them from noisy ongoing activity? Here we show that sensory signals are encoded more reliably, with less noise in certain low-dimensional subspaces. These coding subspaces are defined by correlations between simultaneously recorded neural activity in primary sensory cortex and upstream sensory brain regions; the most correlated dimensions were best for decoding. We analytically predicted and experimentally confirmed that coding subspaces can be further improved when defined based on populations with lower noise correlations between cortex and upstream regions. We show that this principle generalizes across diverse sensory stimuli in the olfactory system and the visual system of awake mice. Our results suggest the cortex may multiplex different functions by executing them in different low dimensional subspaces.Significance statementNeuroscience is a cottage industry; while one study aims to understand how neurons encode function X, a separate study seeks neurons that encode for function Y, etc. However, the brain does multiple things simultaneously. The cottage industry approach leaves unclear how the brain performs functions X and Y simultaneously, sometimes using the same neurons. Here our results suggest that the same neural circuit can perform different functions by performing them in different “subspaces”. We show that the brain may take advantage of correlated interactions between sensory cortex and upstream regions to define these subspaces.