Toward Adaptive VR Design: Physiological and Cognitive Predictors of Cybersickness in Serious Games

Cybersickness limits the usefulness of virtual reality (VR) for serious games. Multivariate cybersickness prediction has been studied widely, but mostly on tightly controlled stimuli rather than free-play settings. We examined a small set of theory-motivated predictors in a VR fishery-management simulation with 105 participants, recording heart rate variability (HRV), galvanic skin response (GSR), and heart rate (HR), and collecting post-session ratings of cognitive load and session duration. Hierarchical regression on 94 complete cases explained 35.8% of cybersickness variance (R² = 0.358, F(7,86) = 6.84, p < 0.001). Duration dominated (β = 0.87), followed by cognitive load (β = 0.52) and GSR (β = 0.54); HR was marginal (β = 0.45, p = 0.052); HRV correlated bivariately (r = −0.24) but was attenuated in the multivariate model. Duration, cognitive load, and GSR are candidate inputs for adaptive comfort systems in serious-game VR.