Active control of heading and the importance of 3D structure, 2D structure, depth and rotation rates

Purpose. Depth information has been implicated in the perception of direction of heading (van den Berg & Brenner, 1994) and used to account for differences in performance between a ground plane and a dot cloud (van den Berg, 1993). However, no account has been taken of the variation of the 2D layout associated with different 3D environments. An active steering task, that dissociates gaze and heading direction, was/employed to explore the effects of 2D layout and also the incidence and importance of high rotational velocities for heading perception. Methods. Participants underwent simulated (30Hz) locomotion at 2m/s for 8 seconds from an initial trajectory 10° (±20%) to the left or right of a target tower 20m away. The task was to control heading so as to travel straight to the tower. Heading was adjusted with a joystick, simulated gaze was continuously adjusted so as to keep the tower centred on the projected display (80° × 80°). Four conditions were used: a ground plane; a ceiling; a cloud and a 'capped cloud' which was balanced with the ground plane for 2D layout. Environments consisted of 400 dots with a lifetime of 400msec. RMSE heading error (gaze-heading) provided a performance measure. Results. Performance was significantly worse for the cloud than for the ground plane as previously reported. Notably, the 'capped cloud1 did not contain the depth information of the ground plane but supported similar levels of performance. Time-course measures suggest that high rotational velocities may be exploited in the difficult cloud and ceiling conditions. Conclusions. First, an active steering task suggests that 2D layout plays a more important role than 3D layout in the perception and control of heading. Second, high rotational velocities may elucidate rather than hide heading direction.