Velocity constancy in virtual reality environments

Purpose. Velocity constancy is thought to be determined by relative scaling (Brown, 1931; Zohary & Sittig, 1993). We wished to determine how the interaction of different depth cues that influence scaling (size, perspective, texture gradient) contribute to velocity constancy. Methods. The point of subjective equality (velocity) of two moving objects (cars) was determined by an adaptive staircase procedure. Both cars were simultaneously displayed on a 180° projection screen (radius 6 m). Initially the cars were displaced to the left and right (40° - 70°) of the fixation point and then moved centripetally. The standard object was at a distance of 20 m and moved with a physical velocity of 3 m/sec. The test stimuli were displayed at 10 m, 20 m or 40 m distance. Subjects were instructed to fixate the center of the display. In further experiments the subjects had to pursue one of the cars while it was moving. We varied the depth cues in the scene (size, perspective, size + perspective, size + perspective + texture gradient). Results. Size had the strongest effect on velocity constancy but the addition of other cues could improve constancy. Relative depth signaled by other cues, without a size cue, resulted in a failure of constancy. Tracking did not have an effect on perceived velocity, except when the tracked car was at a further distance than the standard car. Conclusions. Of the cues we tested relative size is the most important contributor to our ability to judge the physical speed of objects moving at different depths. Signals about eye movements and eye movement errors seem to be an intrinsic part of the computations underlying velocity constancy.