Abstract
In real and virtual spaces, distances are egocentric (from viewer to a target) or exocentric (between two targets). In real spaces, egocentric distances are estimated accurately and exocentric distances are overestimated when compared to geometric accuracy. In head-mounted displayed virtual environments (HMD-VEs), egocentric distance judgments to targets on the ground are underestimated compared to estimates made in real spaces. However, the accuracy of exocentric distance perception in HMD-VEs, compared to estimates in real spaces, is unknown. Here, we investigated judgments of egocentric and exocentric distance perception in HMD-VEs in comparison to estimates made in an analogous real space. In all experiments, participants viewed a distance then turned and walked a distance without vision. They matched the interval by walking the extent between themselves and the targets (egocentric) or the extent between two targets (exocentric). In both environments, participants were allowed to move their heads to view the targets but were not permitted to move their bodies from the home location. In Experiment 1, participants estimated the egocentric extents in depth and exocentric extents in the frontal plane. Like previous research, estimates of egocentric distances in the HMD-VE were underestimated compared to real world estimates. However, estimates of exocentric distances did not differ between spaces. In Experiment 2, we manipulated orientation of displayed exocentric distances. Participants estimated exocentric distances oriented in depth and frontal planes. Again, results showed that exocentric distances oriented in the frontal plane were estimated similarly in both spaces. However, exocentric distances oriented in the depth plane were underestimated in the HMD-VE compared to the real space. The results suggest that underestimation of distance in HMD-VEs does not generalize across depth and frontal planes. Work is underway to confirm results with another measure that does not recruit walking strategies, to control motion parallax, and to test possible mechanisms.
Grant Acknowledgment: NSF 0914488.