Abstract
Introduction: It is well documented that a vertical bar will pop out amongst many horizontal bars. Since this search task is so natural and automatic, and because orientation selective cells are so widespread in early visual areas, it is commonly assumed that people have access to preattentive mechanisms tuned to vertical bars and not horizontal ones. These mechanisms should enable feature-based attention to vertical versus horizontal bars across different tasks. However, we show here that in a top-down selective attention task, people do not have access to these mechanisms. Methods: Stimuli were random clusters comprising 1-, 2-, 3-, or 4-each of black horizontal, black vertical, white horizontal, and white vertical bars. In different attention conditions, participants strove to mouse-click the centroid of only the (1) black, (2) white, (3) vertical, or (4) horizontal bars in the stimulus display, ignoring the bars of the opposite feature. Results: We found that when there is only one of each type of bar, participants could attend to the feature of interest very well, regardless of what the feature was. However, when faced with two or more of each type of bar, they could only perform the task in the attend-to-black and attend-to-white cases. When attempting to attend to just vertical or just horizontal bars, participants gave nearly as much weight to the distractors as to the targets. We propose that instead of using a mechanism for absolute orientation, a local orientation-contrast mechanism is used. In a follow-up study with 4 targets and 16 distractors in which sparse targets stand out among the more numerous distracters, participant performance improved dramatically. Conclusion: As suggested by Nothdurft (Perception & Psychophysics,52,355-375, 1992), feature-based attention has access only to mechanisms sensitive to spatial orientation-contrast, not to mechanisms selective for absolute bar-orientation
Meeting abstract presented at VSS 2015