Vergence eye movements in virtual reality

Abstract

There is increasing interest in incorporating eye tracking in virtual reality (VR) systems to infer gaze in 3 dimensions. Estimation of gaze in depth may be limited by errors of measurement of the small eye vergence angles involved. It may also be limited by true errors of vergence in VR. We found that observers commonly made errors of vergence when viewing in VR, such that the dominant eye was more accurately aligned than the non-dominant eye. Errors of vergence were often larger than Panum’s fusional range. The absence of reported diplopia suggests that some of the visual input from one eye was supressed or the double images escaped attention. Errors of vergence could not be explained by contamination by variations in pupil diameter. Errors of vergence were not reduced when the vergence-accommodation conflict was eliminated by optical correction. These results suggest that the ability to correctly infer gaze distance in VR may be limited by true errors of eye vergence, rather than by measurement errors imposed by limitations of eye tracking.