Abstract
Purpose: To see if and how focused visual attention improves performance on orientation integration and “pop-out” tasks. Methods: I compared the effect of attentional load on two tasks: estimation of “odd-man-out” orientation (as a function of the number of vertical distractor elements) and judgment of mean or “overall” orientation (as a function of orientation variability). Stimuli consisted of an annulus of Gabor elements preceded and followed by a spatial-frequency matched noise mask. Subjects either reported if the mean or “odd-man-out” orientation was clockwise or anticlockwise of vertical. In high attentional-load conditions subjects also performed a secondary “attentional blink” task: to report the orientation of a centrally-presented white ‘T’ embedded within a sequence of black ‘Ts's rotating at 19 frames/sec. QUEST maintained 83% correct performance on the secondary task by manipulating T-contrast within a mask. In low attentional-load conditions subjects viewed identical stimuli but were instructed to ignore the rotating T sequence. Results: Divided attention elevated thresholds in both tasks (for both naïve and experienced observers) but elevation was greater for the odd-man-out task. An equivalent noise model accounted for data from the mean orientation condition, where divided attention has the effect of modestly reducing subjects’ sampling efficiency (the number of features they can monitor) but does not change estimated internal noise, i.e. their uncertainty about each feature (orientation in this case). Performance on the “pop-out” task was inconsistent with subjects using the mean orientation of all elements. Naive subjects showed substantial learning during the course of this experiment. This had the effect of both reducing internal noise and increasing sampling efficiency.