Abstract
The human visual system encodes spatial relations along the anatomical scaffold of retinotopic organisation, yet top-down control can flexibly change the encoding of visual space. We investigated this flexible encoding of space by means of population receptive field (pRF) mapping. In separate conditions, participants performed a titrated two-alternative forced choice task on either fixation luminance, or the color or speed of elements in a bar aperture systematically traversing the visual field. Single-voxel pRF parameters were fit for each of the conditions, and their spatial characteristics were compared across conditions and retinotopic regions. Over and above changes in pRF amplitudes, voxels' preferred visual-field position and size changed as a result of stimulus-based attention. Notably, visual-field position changes were restricted to radial movement, i.e. to changes in eccentricity. The sign of changes in eccentricity depended on pRF eccentricity during fixation; central pRFs migrate outward, whereas peripheral pRFs move inward. Changes in size were strongly correlated with these changes in position. That is, outward migration was tied to an increase in size, whereas inwardly migrating pRFs showed decreases in size. We then explored how these pRF dynamics affect stimulus representation at the retinotopic level by decoding the spatial signature of the stimulus based on pRF parameters and BOLD amplitudes. This approach showed that the aggregate effect of spatial attention on a population-level is an increase in the spatial fidelity of the stimulus representation, regardless of the sign of underlying pRF size changes. We show that attention changes pRF positions radially, and that both size increases and decreases may result in increased fidelity of the encoding of spatial stimulus characteristics. These results elucidate how attention changes visual spatial processing.
Meeting abstract presented at VSS 2016