Integration of Disparity Information across Multiple Fixations

We measured stereo thresholds for observers viewing noisy, dynamic random-dot stereograms (DRDS) containing either a spatial or a temporal step in depth. Stereograms were presented either with or without a small fixation cross at zero disparity. When the fixation marker was present observers were instructed to maintain fixation throughout the presentation.

For spatial steps in depth, the lowest thresholds were found when there was no fixation marker and observers were free to look around the stimulus. Thresholds were higher with a fixation marker in the stimulus. The opposite pattern of results was found for stimuli that contained a temporal step in depth, ie the lowest thresholds were found with a fixation marker. Furthermore, presentation of a single depth region with a fixation marker gave the lowest thresholds of any configuration tested.

The results with temporal changes in depth are reminiscent of the findings of Westheimer (1979 Experimental Brain Research 36 585 – 597). Westheimer argued that sequential presentation in the same location might depress sensitivity to changes in depth, but in DRDS no single dot changes in depth without also changing in retinal location, so this explanation is not applicable.

We conclude that, with a spatial step in depth, it is advantageous to foveate both sides of the step, so multiple fixations help. With a temporal step in depth, the spatial structure of the stimulus is uniform, so multiple fixations may not necessarily help, especially if they result in uncertainty about vergence position.