Effects of motion parall\x and reference objects on heading perception during translation and rotation

Abstract

Purpose. Last year we reported that mean judgmenls of heading during simulated translation and rotation of the observe were accurate with displays of a complex texture-mapped scene (Warren et al., ARVO, 1996). Here ue investigate the information used in these judgments by nanipulating edge parallax, motion parallax, and the presence of reference objects. Methods. Subjects viewed live types of displays presenting observer translation wer a ground plane al 2 m/s: ( 1 ) Occluding tombstones on a textured ground plare; (2) Non-occluding posts on a tcxturcd ground plane; (3) Textured ground pkne alone; (4) Occluding tombsloncs alone; and (5) Non-occluding posts alone. In the simulated rotation condition, a fixation point at eye level remained stationary on the screen, and displays simulated optic flou at an eye rotation rate of 0 lo ±7 deg/;. In the actual rotation condition, subjects tracked a moving fixation point with the same rotation rate. They adjusted a probe at 10 m to indicate perceived heading. Ri suits, Mean heading error was small at all rotation rates when the densely text ired ground plane was present (providing differential motion and radial foreground flow), with or without added lombstones or posts. However, with tombstones ahme (providing edge parallax and reference objects) or posts alone (providing im lion parallax and reference objects), crroi increased substantially with rotation rate. Conclusions. A densely tcxturcd ground plane containing differential motion aid radial foreground flow is sufficient lor perceiving heading during simulated rotation. The pattern of results is reproduced in simulations of a heading model based on differential motion (Rieger & Lawton, 1985).