Visuomotor Interactions and Perceptual Judgments in Virtual Reality Simulating Different Levels of Gravity

Virtual reality is used to manipulate sensorimotor interactions in a controlled manner.
A critical issue is represented by the extent to which virtual scenarios must conform to
physical realism to allow ecological human–machine interactions. Among the physical
constraints, Earth gravity is one of the most pervasive and significant for sensorimotor
coordination. However, it is still unclear whether visual perception is sensitive to the
level of gravity acting on target motion displayed in virtual reality, given the poor visual
discrimination of accelerations. To test gravity sensitivity, we asked participants to hit
a virtual ball rolling down an incline and falling in air, and to report whether ball motion
was perceived as natural or unnatural. We manipulated the gravity level independently
for the motion on the incline and for the motion in air. The ball was always visible during
rolling, whereas it was visible or occluded during falling before interception. The scene
included several cues allowing metric calibration of visual space and motion. We found
that the perception rate of natural motion was significantly higher and less variable
when ball kinematics was congruent with Earth gravity during both rolling and falling.
Moreover, the timing of target interception was accurate only in this condition. Neither
naturalness perception nor interception timing depended significantly on whether the
target was visible during free-fall. Even when occluded, free-fall under natural gravity was
correctly extrapolated from the preceding, visible phase of rolling motion. Naturalness
perception depended on motor performance, in addition to the gravity level. In sum,
both motor and perceptual responses were guided by an internal model of Earth
gravity effects. We suggest that, in order to enhance perceptual sensitivity to physical
realism, virtual reality should involve visual backgrounds with metric cues and closed-loop
sensorimotor interactions. This suggestion might be especially relevant for the
design of rehabilitation protocols.