Jeongyeol Ahn¹, Lilyanna Gross¹, Elisha Merriam¹; ¹National Institute of Health

Normalization has been proposed as a core computation in early visual cortex and helps explain key response properties such as surround suppression. Recent work suggests that population receptive field measurements reflect both excitatory and suppressive components, and some models have begun to capture these in fMRI data. These approaches show that inhibitory effects can be detected, but they rely on additive suppression and do not offer a generative account based on divisive normalization. A population receptive field framework that explicitly incorporates normalization and explains response changes driven by shifts in neural input has not yet been established. Here we propose a population receptive field model that predicts voxel responses through a generative process based on divisive normalization. We assume that each voxel contains a population of neurons that produces excitatory responses to visual stimulation and another population with similar receptive field structure that contributes suppressive drive through normalization. We simulated the resulting BOLD responses under this framework and found that the model predicted response profiles in two situations. The first involved increases in baseline activity induced by external inputs such as brain stimulation. We compared these predictions with findings from a recent fMRI study that measured population receptive fields after brain stimulation. The model captured increases in baseline activity and surround suppression. The second involved increases in stimulus size that elevated the pooled input to neighboring neurons. To test our model, we presented moving bar apertures of three widths to human participants while acquiring fMRI time series. The observed responses showed systematic changes in suppression across bar widths, and the model successfully captured these patterns. These findings show that a population receptive field model grounded in divisive normalization provides a coherent account of fMRI response changes driven by shifts in neural input in human early visual cortex.

Acknowledgements: This research was supported by the Intramural Research Program of the NIMH (ZIAMH002966) to E.P.M.