Abstract
We make eye movements continuously to obtain an understanding of the 3D world around us, but how does the visual system plan these scanning movements of the environment? Here we study what 3D visual information is used to plan saccades. Cue conflict studies have shown that the visual system combines cues in a statistically optimal way for perception. Recently, it has been shown that 3D surface orientation influences spontaneous saccade directions when viewing a surface (Wexler & Ouarti, 2008). Here we use cue conflict stimuli to study whether cues are combined in an optimal way for eye movements. We investigated the perceived tilt and the direction of saccades made in reaction to a tilted cue conflict stimulus.
Stimuli consisted of slanted planes with monocular (perspective) and binocular (disparity) cues each specifying a different tilt angle. We used both small (0°–45²) and large (180°) tilt differences. The reliability of the disparity signal was varied so as to observe a change in weighting of the two cues for perception and saccade direction.
Observers were asked to view the plane and told that they were free to make eye movements across the plane for 3 s, after which they reported perceived surface tilt using a visual probe controlled by a joystick.
Perceptual and saccade data show similar results. For small cue conflict, a combined estimate of surface depth gradient is used to judge tilt and to plan saccade direction. For large conflicts both oculomotor and perception systems abandon cue integration and base their output on only one of the cues at any one time.
M. Wexler and N. Ouarti. (2008). Depth affects where we look. Curr. Biol., in press.