Abstract
Background and Goal. In easy visual search tasks, the target is often detected after a few, quick eye movements toward the most salient display locations, whereas difficult tasks typically lead to more systematic scan paths. Such findings have been obtained almost entirely in 2-dimensional search spaces, disregarding potential effects concerning depth. The few search studies involving depth have typically presented search objects in distinct depth planes with depth serving as a known target feature. Search for conjunctions such as depth and color is extremely efficient, suggesting a special role of depth in guiding attention. Here we studied the dynamics of overt attention in 3-dimensional search spaces under the more natural conditions of a continuous depth (binocular disparity) dimension, with shape and color being the only known target features.
Method. Observers searched stereoscopic displays while their eye movements were recorded. Two search tasks (easy: color-orientation conjunction search; difficult: search for the only mirrored instance among otherwise identical objects) in four display sizes (4, 8, 16, and 32 objects) were employed to manipulate task demands. Results. In the horizontal and vertical dimensions, the easy task was dominated by an eccentricity effect (greater RT with more eccentric target), and the difficult task revealed a reading-direction effect (greater RT with target on the right or at the bottom). In the depth dimension, the only effect was a bias of the initial saccade toward near objects, which occurred under the least difficult task demands (4 objects, easy task). Conclusions. Search dynamics appear to operate almost exclusively in the horizontal-vertical plane, where they systematically depend on task demands. Binocular disparity seems to be virtually disregarded by the mechanisms guiding attention. However, in real-world search, additional depth cues and the need for lens accommodation may give depth a greater impact on search behavior than binocular disparity alone.
Supported by NIH R15 EY017988 to MP and by NIH R01 EY016200 to MC.