Active Control of Heading and Depth

Abstract

The addition of depth information has been reported to improve performance on a locomotor heading judgement task in a cloud of dots. Van den Berg and Brenner [1994 Nature (London) 371 700 – 702] used stereoscopic disparity as a depth cue. Coding dots with disparity provides both cues for depth order and for identity between successive frames (each dot may be disambiguated from its 2-D neighbours by its 3-D position). Here, we attempted to determine the relative importance of depth and identity, by dissociating disparity-depth and disparity-identity.

We employed an active steering task that dissociates gaze and heading direction. Participants underwent simulated (18 Hz) locomotion through a cloud (100 dots, lifetime 440 ms) at 2 m s⁻¹ for 8 s from an initial trajectory 15 deg (±20%) to the left or to the right of a target tower. Heading was adjusted with a joystick and simulated gaze was continuously computer adjusted so as to keep the tower centred on the projected display (20 deg horizontal). Mean unsigned heading error (gaze-heading angle) over the time course provided a performance measure.

Four types of stimuli were used: (i) veridical disparity depth; (ii) non-veridical disparity (disparity and associated motion-in-depth was randomly assigned to each point in the display); (iii) the left half image of the previous condition to both eyes; (iv) synoptic images (no disparity). No marked advantage was observed when the display contained veridical disparity, and non-veridical disparity produced only a minor impairment in performance in a subset of conditions. We suggest that depth information per se may not be utilised in the perception and control of heading.