Federico Sorcini¹, Jun-Ming Yu¹, Simona Buetti¹; ¹University of Illinois Urbana-Champaign

Repeatedly searching for a target through the same layout of objects should yield some improvement in one’s ability to find a target. Previous work has suggested that this intuition is incorrect and that when performing a repeated visual search, participants prefer to use vision rather than memory to guide their search, even in complex, realistic scenes. However, much of the work that has been done has involved presenting the search stimuli at relatively small eccentricities on a 2-D monitor, both of which ensure that all stimuli are easily accessible by the visual system. Here, we investigate the extent to which memory guides repeated search when visual information is not always easily accessible: in a real 3D space. Participants were eye-tracked as they repeatedly search through sets of real objects placed on a desk in front of them. Four, eight and twelve objects were placed in a wide range of eccentricities (up to 45 degrees to the left and right of the participant), either on the same depth plane as an initial fixation point (horopter fixation condition) or on a different plane (near fixation condition). Participants were instructed to direct their gaze to the target object as quickly as possible. We hypothesized that observers would use memory to a greater extent for objects located at the farthest eccentricities, and perhaps moreso in the near fixation condition because objects would appear initially blurrier than in the horopter fixation condition. Results showed that initial fixation depth, set size, and eccentricity impacted saccadic initiation times and time to foveate the target. Learning was increasingly evident as set size increased and at farther eccentricities, but performance leveled in all conditions. Surprisingly, participants were only about 70% correct at recalling the location of the near and farthest targets, with worse memory for targets at intermediate eccentricities.