Abstract
Virtual reality (VR) promises both naturalism and experimental control for vision science, but how valid is the assumption that perception in VR accurately matches perception in the real world? Recently our group found that size and distance perception for real, tangible objects was affected by expectations regarding objects’ familiar sizes. This familiar size effect (FSE) was strongest when participants viewed real objects through a monocular pinhole to minimize vergence and accommodation cues to depth; however, a significant FSE was found even with binocular viewing when oculomotor cues were fully available. Here we repeated the experiment using a commercially available VR system (Oculus Rift) to present typically sized virtual Rubik's cubes (5.7 cm) and dice (1.6 cm), as well as atypically sized virtual Rubik's cubes (1.6 cm) and dice (5.7 cm). Virtual objects were presented in both binocular and monocular pinhole conditions. Participants made manual estimations of the objects’ size and distance. Surprisingly, perception of both size and distance in binocular VR was more similar to monocular pinhole viewing than binocular viewing in reality. Specifically, the FSE during binocular viewing was significantly stronger in VR than reality. A likely explanation is that the vergence-accommodation conflict in VR leads to higher reliance on cognitive factors and less reliance on oculomotor cues. These results introduce an important caveat for the use of VR in the study of perception: even though distance cues like vergence are present in VR, they may not be utilized for perception to the same degree as in reality.