Abstract

Different parts of the visual system respond strongly to particular visual features, such as color (hV4/VO1; Engel et al, 1997; Brewer et al, 2005) or motion (hMT+/TO1/TO2; Tootell et al, 1995; Amano et al, 2009). Computational theories posit that attention is guided by a combination of spatial maps for individual features that are weighted according to task goals (Itti & Koch, 2001; Serences & Yantis, 2006). Consistent with this framework, when a stimulus contains several features, attending to one or another feature results in stronger fMRI responses in regions preferring the attended feature (Beauchamp et al, 1997; McMains et al, 2007; Runeson et al, 2013). We hypothesized that multivariate activation patterns across these feature-responsive regions form spatial ‘feature maps’, which combine to guide attentional priority. One prediction from this hypothesis is that changes to task demands will reweight the spatial representation of objects in the scene within neural priority maps according to the correspondence between the stimulus features, the attended feature, and a region’s preferred visual feature. We tested this prediction by reconstructing spatial priority maps from fMRI activation patterns across retinotopic regions of visual cortex using a 2D spatial inverted encoding model (Sprague & Serences, 2013). Participants viewed a peripheral visual stimulus at a random location on each trial which always contained both visual motion (clockwise/counterclockwise) and color (blue/red). On each trial, participants were precued to report the predominant direction of motion or color of the stimulus. Consistent with previous univariate results, stimulus representations in reconstructed priority maps were selectively enhanced in color-responsive regions when color was attended, and in motion-responsive regions when motion was attended. These results suggest different cortical regions support spatial maps of different visual features, and that these maps can be reweighted based on task demands to guide visual behavior.