Figure 3.

Spatial discriminability increases with attention and is mediated by position changes in vRFs. Error bars indicate bootstrapped 95% CIs. a, We formulated a measurement to describe the ability of a population of voxels to make fine spatial discriminations around the attention target. We used the properties of each voxel's spatial tuning curve to make this measurement (see Materials and Methods). Spatial discriminability increased when subjects attended the target, compared with when they ignored the target in the opposite hemifield (resampled p < minimum possible p value 1/1000 for all ROIs). b, The discriminability metric was recomputed for vRFs with a variety of attentional modulations. Abscissa labels from left to right: None, vRF parameters during the neural attend fixation condition; a, amplitude; s, size; p, position. Spatial discriminability increased significantly when we applied position changes measured during the attend L/R task to the vRFs compared with when we applied no parameter changes (solid bar). By contrast, applying size changes did not change spatial discriminability in most ROIs, although it did cause a small increase in IPS0.