Running modulates primate and rodent visual cortex via common mechanism but quantitatively distinct implementation

Abstract

When mice actively locomote, activity in their primary visual cortex (V1) is strongly modulated¹. This observation has fundamentally altered conceptions of a brain region previously assumed to be a passive image processor, and extensive work has followed to dissect the sources, recipients, and functional consequences of running-correlated modulation^2–13. However, it remains unclear whether visual processing in primates might similarly change during active locomotion. We therefore measured V1 activity in a nonhuman primate, the common marmoset (Callithrix jacchus), while they alternated between running and stationary. In contrast to the large increases in mouse V1 during running, responses in marmoset V1 were slightly but reliably decreased during running. Despite this superficially categorical difference, leveraging large-scale recordings to characterize the latent variables driving population activity revealed a common mechanism: trial-to-trial fluctuations of shared gain modulations were present across V1 in mice and marmosets. These gain modulations were larger in mice and were often positively correlated with running; they were smaller and more likely to be negatively correlated with running in marmosets. Thus, population-scale gain fluctuations reflect a common principle of mammalian visual cortical function, but important quantitative differences in their magnitudes and correlations with behavior produce distinct consequences for the relation between vision and action in primates versus rodents.