TESTING THE BOUNDARIES OF MOUSE VISUAL AREAS

The mouse visual cortex is thought to contain a primary area (V1) surrounded by many higher visual areas (HVAs), defined by retinotopic maps measured with widefield imaging. However, these maps vary substantially across individuals. Moreover, the individual HVAs cover only partial regions of the visual field. Is our current understanding of retinotopy limited by the use of widefield imaging? Is it consistent with single-neuron properties? Should HVAs parcellation be reconsidered? We used two-photon mesoscope imaging to record population activity from thousands of excitatory layer 2/3 neurons in V1 and putative HVAs of head-fixed transgenic mice expressing GCaMP6s. Stimuli included a structured noise stimulus (“zebra noise”). Responses were fitted with a nonlinear model based on Gabor wavelets, enabling the estimation of preferred retinotopic position, orientation, size, spatial frequency, phase, and drift. We used these estimates to reconstruct visual field coverage and functional organization at single-cell resolution. The results confirm that each putative HVA covers only a subregion of the visual field. The regions labeled RL, LM, PM, and P collectively tile the visual field. The overlap between the tiles is not superior to the overlap between their corresponding V1 sectors. We further identify systematic biases in receptive-field size, spatial frequency, and phase selectivity as a function of retinotopy within V1. Our findings suggest that a ring of HVAs around V1 might be usefully merged into a single secondary visual area surrounding V1. The difference in visual preference seen across putative HVAs might be explained by differences in retinotopic coverage.