Figure 1.

Covert spatial attention task and fMRI-based analyses used to link single voxels to population-level measurements. a, Subjects fixated centrally and attended to brief rotations in the pentagon stimulus on the left or right while a flickering checkerboard probe stimulus appeared at 1 of 51 grid locations across the visual field. On control runs, subjects attended to a contrast change at fixation. fMRI data measured during this attention task are used to create visualizable estimates of vRFs and stimulus reconstructions. b, A receptive field model is fit to the responses of each voxel and can be described by its x and y position (center), response baseline, response amplitude, and size (FWHM). c, Given a population of voxels in a retinotopic region, such as V1, we examine two different measures of spatial information in the population. The first, a spatial discriminability metric, scales with the slope of the tuning curve at a given location in space (see Materials and Methods). The second relies on a multivariate IEM for space. By reconstructing images of the mapping stimulus on each test trial, we can measure how population-level spatial information changes with attention. We then can model how changes in individual vRFs affect both of these population measures.