Abstract
Purpose: An important open issue in cognitive science is the role that 3D scene information plays in object search tasks. Does the visual system employ different strategies for exploring 3D versus 2D scenes? In our work, we have studied this question in the context of a word-search task. Methods: In this series of experiments, we compare search times in two scenes that are closely matched in their 2D characteristics, but differ radically in their perceived 3D characteristics. Subjects search for words in one of two displays. In both displays, words are arranged consistent with a flat 3D surface, but the background enhances or debilitates a 3D percept. In the 3D-consistent display, the background image is a projection of parallel lines on the implied 3D surface extending to a horizon point; in the 3D-inconsistent display, the background consists of the same image, but flipped vertically. Naive subjects consistently report a 3D percept for the first background, but a 2D percept for the second. Results: Two interesting results were found: (1) Subjects who were presented only with a 3D-consistent display were faster in a visual search task, on average, than subjects who were presented only with the 3D-inconsistent display. (2) Subjects who were trained on the task with a 3D-consistent display and then were switched to a 3D-inconsistent display performed faster than subjects who trained on only a 3D-inconsistent display, whereas the converse (3D-inconsistent followed by 3D-consistent) showed no such pattern. Conclusion: These results suggest that a 3D percept of a scene can facilitate visual search. They also suggest that a 3D-inconsistent scene can be “primed” into a 3D-consistent percept by exposure to a 3D-consistent version of that scene, highlighting the influence of top-down processes in visual search tasks.