Abstract
One of the most intriguing hallmarks of spatial attention is its rhythmic nature, i.e. instead of sustaining a static "spotlight" centered at the attended location, attention rhythmically samples both the attended and the unattended location at ~7 Hz. However, it remains unknown whether such a rhythmic sampling process also occurs in feature-based attention that selects and enhances the representation of visual stimuli sharing the attended feature throughout the visual field. Here, we used human behavioral data to investigate the rhythmic oscillation in feature-based attention. Subjects viewed a stimulus display consisting of two oriented (± ~45 degrees) grating arrays at random locations across the visual field for 300ms. A brief luminance increment (200ms) occurred with all gratings of one array to reset attention to their orientation. After resetting, the gratings remained on the screen for a varying cue-to-target stimulus onset asynchrony (SOA, 50-1050ms). Then they were replaced by a test display comprising of the same number of gratings of one orientation that was approximately either parallel or perpendicular to the cued orientation (50% chance, valid or invalid conditions). Subjects indicated whether the orientation in the test display was clockwise or counterclockwise of its closest orientation in the stimulus display. For both conditions, we measured orientation discrimination accuracy as a function of SOA and derived its power spectrum. We found a common ~7 Hz rhythm in the time courses in both conditions. More importantly, the accuracy time course in the valid condition exhibited a consistent anti-phase relationship with that in the invalid condition at ~7 Hz. These results suggest that the two orientation features are rhythmically sampled in alternation in feature-based attention. Our study is among the first to discover rhythmic attention sampling in feature space and speaks to a more general rhythmic temporal structure that commonly governs spatial and non-spatial attention.
Meeting abstract presented at VSS 2018