Abstract
We have shown previously that when observers have to find multiple targets in a limited time, immediate feedback following each saccade leads to efficient saccades (Verghese & Ghahghaei, 2013). Here we examine whether feedback is as effective when it occurs at the end of the trial, rather than after each saccade. Observers actively searched a display with six disks that each had an independent probability of being a target, so the number of targets in a trial ranged from 0 to 6. The probability of the target was set to 0.33, 0.5 or 6.7 in separate blocks. The luminance of each disk was drawn randomly from a Gaussian distribution for target and distractor luminance. Due to the overlap of these distributions, it was uncertain whether a luminance value in the overlap zone came from the target or distractor distribution. Because there was insufficient time to inspect all locations, selecting uncertain locations was much more informative than selecting locations most likely to have the target. As an incentive to explore uncertain locations, a saccade to a disk with luminance in the overlap zone generated feedback by switching the disk luminance to the mean value of the "true" distribution. The feedback occurred either immediately after the saccade, or at the end of the 900 ms trial, and lasted 200 ms. Six observers participated in the experiment, with separate sessions of delayed and immediate feedback, in counterbalanced order. Observers made more efficient saccades in sessions where feedback occurred immediately after a saccade, compared to delayed feedback at the end of the trial, particularly when target probability was high and multiple targets were more frequent. Accuracy was also higher in sessions with immediate feedback. These findings indicate that saccade-contingent feedback is very effective in increasing the efficiency of eye movements in visual search.
Meeting abstract presented at VSS 2017