Abstract
The visual system is confronted with more information than it can handle at any given moment, and thus must prioritize a subset of this input for processing. The brain may implement this selection by regulating how information flows through the visual system. Due to the challenge of studying interactions between brain regions non-invasively, most research in humans has focused on modulation of evoked activation in individual regions in visual cortex. The purpose of the current study is to examine attentional modulation of the interactions among regions using fMRI. Observers completed a feature-based attention task in which they viewed composite face/scene images and attended to either faces or scenes for entire scanning runs. Stimulus-evoked responses and global noise were removed from the data, resulting in a more pure index of connectivity (i.e., background connectivity) under each type of attention. We then identified regions of interest (ROIs) in retinotopic occipital cortex (V1/V2/V3/V3a/V4), and category-specific ventral temporal cortex (fusiform face area, FFA/parahippocampal place area, PPA). We hypothesized that background connectivity between the PPA and retinotopic ROIs would be stronger when observers attended to scenes than to faces, and that the reverse would hold for the FFA. This pattern of results was borne out in the data, with the greatest effects occurring in late ventral occipital areas, especially V4. While both types of attention altered connectivity, effects were more robust and widespread for scene attention and the PPA. By comparing periods of attention against separate resting data, we can further assess the extent to which task differences reflect enhanced and/or suppressed connectivity with the task-relevant vs. -irrelevant category-specific ROIs. Our results suggest that sustained top-down attention can modulate background functional interactions in ventral visual cortex independent of evoked responses, providing preliminary empirical support in humans for theoretical models of cognitive control and goal-directed attention.
Naseem Al-Aidroos was supported by a Natural Sciences and Engineering Research Council of Canada postdoctoral fellowship.