After decades of promise, virtual reality (VR) is no longer science fiction. With millions of headsets in circulation, tech giants investing heavily, and a new generation of researchers growing up with the technology, the question is no longer whether VR works, but when and how. Bailenson and colleagues highlight five consistent psychological findings from thirty years of VR research in a thorough review. Together, they offer a lesson in nuance—and a call for smart use.

The first finding: the impact of “presence”—that feeling of truly being there—depends on the task. In therapy or training, presence can make a huge difference, like treating phobias or practising medical procedures. But for entertainment or remote meetings? The evidence is much less convincing. Anyone expecting us all to wear headsets for Zoom calls might want to hold off a bit longer.

Second: avatars matter. What you are in VR changes how you act. The so-called Proteus effect shows that people behave differently depending on their avatar’s height, fitness, or friendliness. Perspective-taking through avatars can also support empathy, though effects are often context-dependent and short-lived. Still, we shouldn’t treat representation as neutral when working with VR.

Third: VR works better for doing than for explaining. Practising a skill is more effective for procedural learning than abstract knowledge. That fits with what we know about cognitive load: highly immersive VR can overwhelm working memory. The best VR sessions are short, focused, and hands-on. Don’t expect great results from long headset-based lectures on fractions or grammar.

Fourth: body tracking makes VR unique and raises privacy concerns. Headsets continuously track head and hand movements, and research shows that this motion data alone can identify users with over 90% accuracy. That opens doors for assessment and feedback, but also requires serious conversations about consent and data ethics.

Fifth and final: People underestimate distances in VR. Objects often feel closer than they are, which is a problem in contexts where spatial judgment is key—like simulations or training environments. The tech has improved, but the perceptual challenge remains.

So what does this mean for education? Four key takeaways.

One: Use VR only when it truly adds value. Not because it looks fancy, but because it enables learning experiences that would otherwise be too dangerous, too costly, or simply impossible. A virtual fire drill? Great. Vocabulary revision? Stick to paper.

Two: Keep it short, purposeful, and well-prepared. VR is intense. Too many stimuli or too long in the headset leads to overload. Pre-teaching helps. And not everything belongs in VR—only what really benefits from it.

Three: Avatars matter. Don’t let students use default characters without thought. Who they are in VR shapes what they do. Representation, identity, and agency deserve real attention.

Four: Privacy is not a footnote. VR tracks physical data that can be linked back to individuals. Schools and educators need clear policies about what’s collected, stored, or shared.

Bailenson and colleagues end with sound advice: use VR for things that are dangerous, impossible, counterproductive, or expensive to do in real life. That’s the DICE principle. Training firefighters? Excellent use. Reading a novel in a virtual classroom? Not really.

So, no, VR isn’t a miracle tool. But in the right place, with the right purpose, it might just be the best one. The challenge for teachers, designers, and researchers is to choose wisely. Not everything needs VR, but some things can only happen there.

Abstract of the review-study:

Virtual reality (VR) is an emerging medium used in work, play and learning. We review experimental research in VR spanning three decades of scholarship. Instead of exhaustively representing the landscape, our unique contribution is providing in-depth reviews of canonical psychological findings balanced across various domains within psychology. We focus on five findings: the benefit of being there depends on the activity; self-avatars influence behaviour; procedural training works better than abstract learning; body tracking makes VR unique; and people underestimate distance in VR. These findings are particularly useful to social scientists who are new to VR as a medium, or those who have studied VR but have focused on specific psychological subfields (for example, social, cognitive or perceptual psychology). We discuss the relevance for researchers and media consumers and suggest future areas for human behaviour research.