Misperception of egocentric distances in virtual environments: More a question of training than a technological issue?

Findings from virtual reality applications in general and driving or flight simulators in particular, are frequently generalized to the study of human behavior. Thus, it is crucial to ensure that the virtuality of the experimental setup has little or no effect on perception of space and motion. Most studies show that observers immersed in virtual environments (VE) perceive virtual space as compressed relative to the real world, resulting in systematic underestimations of egocentric distance. Parallax and stereopsis, known to be important depth cues for distance perception, at least for short distances, are rarely used together in driving simulators, so their interactive role during driving tasks is still not clear. Inter-individual differences in misperception are also referred to, though few studies have explored this. The aim of this study was, first, to determine whether egocentric distance perception in driving simulation depends on two depth cues, binocular disparity and motion parallax, and, second, to examine the effect of inter-individual differences. Several conditions were tested, both with and without stereoscopic vision of the scene and/or motion parallax of the head. We focused first on a range of long distances, 40-80 m (Experiment 1) and subsequently widened the range to distances from 5 to 80 m, thereby including short distances where stereopsis should be more relevant (Experiment 2). The study reveals great inter-individual variability, clearly distinguishing two participant profiles. However, results suggest that such differences do not depend on the availability of motion parallax and stereoscopic vision. The findings also show that an initial familiarization phase, under conditions similar to those of the experiments, can be predictive of participants' perceptual behavior.