FriSem

Joshua Wilson, Ph.D. Student in Psychology

Title: Variability in population sensory representations

Abstract: People make inferences from neural sensory representations about properties of the world and their certainty of those properties. How observers estimate their certainty of these properties depends on the characteristics of the variability of the representations they are decoding from. For example, a representation in a highly variable population could have been produced by many different stimuli (higher uncertainty), whereas a representation in a less variable population is consistent with fewer stimuli (lower uncertainty).

Here we use fMRI to study the population variability in early visual cortex in response to pRF mapping stimuli. We find that a great deal of response co-variability between similarly tuned voxels can be explained by imperfect modeling. We also find that voxel response variability is relatively unaffected by stimulus drive when accounting for these model failures. These results suggest that fMRI population variability can be well-explained by additive noise and that we can therefore use relatively simple models to decode uncertainty from cortical populations.