Designing adaptive feedback for hands-on learning in VR: Feedback intervention theory

By providing an immersive and interactive learning environment, virtual reality demonstrates a potential practice before hands-on learning in STEM. However, the cognitive load associated with high immersion, together with the lack of immediate guidance from teachers or peers, can readily lead to learner disorientation, frustration, and even anxiety, thereby impairing performance. Feedback mechanisms are therefore essential; yet many VR courses still employ static feedback that delivers identical error and corrective messages to all tasks carried out by learners, without considering an individual learning state, which may draw attention back to the self and intensify negative affect. This effect is especially pronounced when face-to-face teacher and peer interaction is absent, potentially challenging learners’ motivation. Grounded in Feedback Intervention Theory (FIT), this pilot study explored an adaptive feedback mechanism for VR that dynamically adjusts the feedback focus according to learners’ states, preventing shifts toward self-focus and supporting sustained task engagement. To evaluate the feasibility of this approach, we developed a VR hands-on embedded-systems course and conducted a preliminary quasi-experimental study with 67 undergraduates, randomly assigning participants to an experimental group (FIT-based adaptive feedback) or a control group (static feedback). Both groups used head-mounted displays to complete three sequential units over three weeks. Learning outcomes, learning motivation, and learning anxiety were assessed with pre–post instruments while controlling for prior knowledge and demographic variables. Results showed that, compared with static feedback, adaptive feedback improved learning outcomes and motivation and reduced learning anxiety. Overall, this pilot study demonstrates the feasibility of using FIT as a theoretical basis for VR feedback design and addresses limitations in prior VR research that has often relied on static feedback.These preliminary findings provide a foundation for future large-scale investigations into adaptive VR pedagogical agents.