The Task-Dependent use of Binocular Disparity and Motion Parallax under Natural Viewing Conditions

Abstract

The use of binocular disparity and motion parallax information was compared in three different psychophysical tasks for which natural viewing and physical stimuli were used. Natural viewing may be an important factor in interpreting experiments which have addressed the ability to use disparity and parallax both separately and in combination (see Frisby et al, 1996 Perception25 129 – 154). The stimuli consisted of configurations of three bright LEDs carefully aligned in the horizontal meridian and presented in darkness. The distance of the middle LED (flashing at 5 Hz) could be adjusted along the midline in accordance with the tasks which included: (i) a depth nulling task, (ii) a depth matching task, and (iii) a shape task—match base/height of triangle. Each task was performed at two viewing distances (1.5 and 3.0 m) and under four different viewing conditions: (i) monocular-static, (ii) monocular-moving, (iii) binocular-static, and (iv) binocular-moving. Note that the different tasks differ in their dependence on viewing distance, and the available cues for viewing distance differ between viewing conditions. Four observers made ten settings in each condition at each distance. Observers, as expected, performed badly (bias and accuracy) in all tasks in the monocular-static condition. Nulling was accurate in the other viewing conditions (no estimate of viewing distance required). Performance was best in the matching task (ratio of viewing distances) but although binocular-static was significantly better than monocular-moving performance in this and in the shape task (absolute distance required), there was no additional improvement in the binocular-moving condition. Results show that observers can recover structure accurately from parallax or disparity information in real-world stimuli.