Purpose: The purpose of this study was to determine if the visual search ability of visually impaired (VI) subjects could be improved through computerized training. Methods: Stimulus was 8, 16 or 32 white squares randomly shown on a 6-by-6 square grid. Size of the grid was 10×10, 20×20 or 40×40 deg. In each experimental session, there were 36 stimulus displays that contained one 2×2 deg square (the target) and 7, 15 or 31 1×1 deg squares (distracters), and 9 stimulus displays that contained only distracters. The patient's task was to use two keys to indicate whether a target was present in the display. The computer recorded the reaction time (RT) for each trial (the time between the beginning of the stimulus display and one of the response keys was pressed). Hit rate, false alarm rate, and average response RT for hit were calculated at the end of each session. Each subject performed the same tasks everyday from day 1 and 5, and had the final testing at day 9. There was no training between day 5 and 9. Twenty-four legally blind adults participated in the study. Results: Pairwise comparison between day 1 and day 5 showed that day 5 RT was significantly shorter (avg. 1.65 vs. 1.47 sec, 8 targets at 10 deg) than day 1 RT. There was no significant difference in hit rates between these days. Compared to day 1, day 9 showed significantly shorter RT but similar hit rate. Between day 5 and day 9 mean RT and hit rate remained stable. There was no significant difference in RT or hit rate between 10×10 and 20×20 deg grids, but there was a significant slowing in RT when a 40×40 deg grid was used (1.59 vs. 2.14 sec for 32 targets). Training didn't seem to change this field effect. Conclusions: Visual search training reduced the time VI adults used to locate targets in a cluttered field without surrendering accuracy. The shortening of RT persisted for at least 4 days after training. Whether the shortening of search RT transfers to improvements in obstacle avoidance while walking is under investigation.

VA RR&D