Xue Teng^1,2 (), Laurie Wilcox², Robert Allison^1,2; ¹Department of Electrical Engineering & Computer Science, York University, ²Centre for Vision Research, York University

Humans rely on both visual and kinesthetic cues to register self-motion and navigate through the world. Normally, these sources of information are consistent with the perceived motion through the environment. Virtual Reality (VR) often introduces visual/kinesthetic inconsistency due to scale differences between the physical space and the simulated virtual space (motion gain). Large amounts of left-right self-motion gain has been found to compress apparent distance and monocular depth. In the present study, we asked whether observers adapt to exposure to extended periods of high or low motion gain. In the adaptation phase, observers played a VR game. They moved laterally to intersect targets with their body; their virtual motion was scaled to be either 0.67, 1 or 2 times their physical motion. These three adaptation conditions were presented in separate sessions, each starting with 5 minutes of initial adaptation, followed by testing interleaved with three 2-minute top-up adaptation periods. During the test phase, observers swayed left-right over 20 cm to the beat of a 0.5 Hz metronome and indicated if the virtual environment moved ‘more or less than they did’. Using a method of constant stimuli, we measured the PSE for each gain in separate blocks. Results from 18 observers showed that there were no consistent differences between the PSEs obtained in the three gain conditions. Neither increasing the adaptation duration nor testing monocularly affected this pattern of results. These data suggest that while observers are sensitive to differences between their intended movement and that rendered in VR, they do not appear to adapt to a constant mismatch in the current setup. Ongoing experiments are evaluating if this is also true for forward-backward motion. Lack of adaptation to mismatches between self and world motion may be an important evolutionary strategy as such distortions could signal a hazardous situation (e.g. poisoning).

Acknowledgements: Thanks to NSERC Canada for support under an Alliance grant, ALLRP 570802 - 21 in partnership with Qualcomm Canada.