108 Participants, Three Worlds
What happens when you teach lab safety inside a virtual world instead of a classroom? To find out, 108 university students were randomly assigned to one of three conditions: a fully immersive VR headset, a desktop-based 3D simulation, or a traditional video-and-text control group. Each participant completed the same lab safety curriculum — only the level of immersion changed.
Would more immersion mean more learning?
Simulations Work
The initial results were encouraging. Both the VR and Desktop groups significantly outperformed the control group in procedural knowledge and knowledge transfer. Simulations — regardless of immersion level — proved to be an effective teaching tool for lab safety.
But here's the catch: the VR group did not significantly outperform the Desktop group. Higher immersion didn't translate to higher learning gains. Why not?
But Why?
The original hypothesis centered on presence — the psychological sense of "being there" inside a virtual environment. The theory was elegant: higher immersion induces stronger presence, and stronger presence drives deeper learning.
When we plot every participant's presence score against their learning gains, the data seems to support this. A clear positive trend emerges across all 108 participants. Case closed?
Look Closer
Not quite. A 2026 reanalysis revealed something the original study missed: an ecological fallacy. The overall correlation between presence and learning was real — but it was an artifact of the groups, not the individuals.
When we color each dot by its experimental condition, the picture changes dramatically. Within each group, the relationship between presence and learning virtually disappears. The "trend" was created by the fact that VR participants had both higher presence and higher learning — but one wasn't causing the other.
The Price of Novelty
If presence wasn't the mechanism, what was holding VR back from outperforming desktop? The likely culprit: cognitive load. The VR environment was novel and complex. Participants spent mental energy simply navigating the unfamiliar interface — energy that could have been directed toward learning.
"More immersive" isn't automatically "better for learning" — not until the novelty cost is accounted for.