Abstract
Purpose: :
To investigate the impact of a central scotoma on visual search behavior.
Methods: :
Feature search behavior was measured on two normal and one legally blind subject. The search task was to find a large square target among smaller square distracters. Set sizes 1, 4, 8 and 32 were tested in separate sessions. An Eyelink II eye tracker was used to fill a square area around the gaze position with the background color at the display monitor’s refresh rate, simulating a central scotoma on normal subjects. The legally blind subject had a well-defined scotoma roughly centered at the fovea of each eye. All subjects performed the search with eye movements monitored.
Results: :
In positive trials (target present), reaction times (RT) of all subjects did not change with search set size. In negative trials (target absent), all subjects exhibited a step RT x Set Size function. Set sizes 1 & 4 had similar RTs that were somewhat higher than that of the corresponding positive trial RTs. Set sizes 8 & 32 had similar RTs but these were at least twice as long as that for set size 1 & 4. The number of saccades x set size functions mirrored the RT x set size functions. Search scan-paths revealed several eye movement patterns. In positive trials, if the target was inside the scotoma at the start of a trial, the eye moved until the target was out of the scotoma. If the target was outside the scotoma at the start, the eye moved toward the target for a closer look. Such behavior usually took only a few saccades and did not change with set size. In negative trials, when the set size was 1 or 4, a few local saccades were made so that all the distracters fall outside of the scotoma. When the set size was 8 or 32, a different strategy was used that covered a larger portion of the search area. The scan-paths were idiosyncratic between subjects but fairly consistent within a subject. Such scan-paths usually left some search items unvisited, but the visited areas were highly oversampled, resulting in many more saccades and much longer RTs. The patient with a real scotoma made more, smaller saccades and searched more slowly than normal subjects, but the general search behaviors were comparable.
Conclusions: :
Even with a central scotoma, feature search terminated when the target was exposed and examined. When there wasn’t a target, search was promptly terminated when there were only a few search items and when they were all outside of the scotoma. A more time-consuming strategy was adopted when the search items were more numerous, indicating that when the subject failed to count the number of search items, they were reluctant to terminate the search because there was always the chance that one or more items might be inside the scotoma.
Keywords: visual search • low vision • eye movements