Abstract
In our experiments of visual search for a not particularly salient target, the subjects' eye movements were tracked. After a random time interval of search, all 660 visual items, including the target and the distractors, were covered by identical masks, each covering an original search item which consequently became imperceptible even in after images. In about 80% of the trials, the masks were displayed before the subjects had saccaded to the target location. The subjects had to make a forced choice decision, by a button press, regarding whether the target was in the right or left half of the search display. Most of these decisions were made after the mask onset. We observed that the subjects continued to make eye movements on the mask display for a duration up to a few seconds before responding. In about 20% of the trials in which eyes had not reached the target before masking, the post-mask saccades arrived correctly (within 2 degrees in a display of 46×34 degrees of visual angle) on the location of the target, sometimes after more than one post-mask saccades. We measured the target localization performance as a function of post-mask saccade latency, defined as the time difference between the eyes landing onto the target location and the mask onset. We found that the performance is significantly higher than chance for the latencies up to 1000 milliseconds. In comparison, in the trials in which eyes never reached the target location, the performance is at chance level.
Gatsby Charitable Foundation