Distinct neural dynamics underlie the encoding of visual speed in stationary and running mice

Abstract

Sensory experiences are often driven by an animals self-motion and locomotion is known to modulate neural responses in the mouse visual system. This modulation is hypothesised to improve the processing of behaviourally relevant visual inputs, which may change rapidly during locomotion. However, little is known about how locomotion modulates the temporal dynamics (time courses) of visually-evoked neural responses. Here, we analysed the temporal dynamics of single neuron and population responses to dot field stimuli moving at a range of visual speeds using the Visual Coding dataset from the Allen Institute for Brain Science1. Single neuron responses had diverse temporal dynamics that varied between stationary and running sessions. Increased dynamic range and more reliable responses in running sessions enabled faster, stronger and more persistent encoding of visual speed. Population activity reflected the temporal dynamics of single neuron responses, including their modulation by locomotor state - neural trajectories of population activity made more direct transitions between baseline and stimulus steady states in running sessions. The structure of population coding also changed with locomotor state - population activity prioritised the encoding of visual speed in running, but not stationary sessions. Our results reveal a profound influence of locomotion on the temporal dynamics of neural responses. We demonstrate that during locomotion, mouse visual areas prioritise the encoding of potentially fast-changing, behaviourally relevant visual features.