Interaction Design and Divergent Paths in VR Learning: A Structural Modeling Approach

Immersive virtual reality (VR) has the potential to transform learning, but its success depends not only on technical design but on how learners experience value, agency, and discomfort. This study models how technical efficacy (TE), technology adoption (TAM), and embodied presence (PRES) interact with dizziness (DIZ) and task design to shape performance in VR. Using structural equation modeling (SEM), generalized additive models (GAM), and multivariate regression, we found that TAM fully mediated the impact of TE on PRES (betabeta
ß = 0.542, p betaless than< 0.001), suggesting that technical fluency only enhances immersion when learners perceive the system as meaningful. Presence increased dizziness (betabeta ß = 0.36, p = 0.001), but only TAM—rather than TE—sustained immersion under strain. A nonlinear moderation effect showed that effort (EFFSYS) predicted performance (PERFSYS) more strongly under high DIZ—especially in exploratory designs, where learners retained behavioral flexibility. In contrast, linear paths suppressed this dynamic, flattening performance variance. Despite similar DIZ levels across conditions, only exploratory paths enabled “productive strain,” converting discomfort into engagement. Our findings suggest that immersive learning emerges from the alignment of effort, motivation, and interaction design—not from realism alone. Designing for persistence, not just presence, is critical: discomfort can fuel engagement if learners can act, adapt, and understand why it matters.