Platform is not destiny: Embodied learning effects comparing 2D desktop to 3D virtual reality STEM experiences.

Experiences in immersive 3D virtual reality (VR) are more presence‐inducing, and so it may be tempting to claim that content will be learned better in VR. This randomized control trial study on natural selection challenges that assumption. This study answers the question of whether learning STEM in an immersive 3D VR environment is always superior to learning via a 2D monitor (PC). This is a 2 × 2 × 3 design. The first factor is platform immersivity (low = 2D PC, or high = 3D VR), the second factor is level of embodiment (lower = watching playback video, or higher = using mouse/controller to agentically manipulate content), and the third factor is test time (pretest, posttest, and follow‐up). There was a significant main effect for embodiment, the high embodied and agentic groups learned the most. There was not a main effect for platform, because the participants in the low embodied VR group performed significantly worse than the three other groups. Although, the one high embodied, VR group learned and retained the most knowledge. A path‐analysis revealed that the effect of platform was significantly mediated by presence, agency, and engagement. The smaller learning gain in the low embodied VR condition suggests that participants come to the immersive VR experience with expectations about agency and control of the virtual content, and when those expectations were not met, the disconnect was deleterious for learning. More agentic and interactive control of manipulable virtual content is encouraged. Design is critical, and platform is not destiny. Lay Description: What is already known about this topic?: VR has special affordances like presence and embodied content manipulation that can increase learning, if used correctly.Compared to 2D desktop, being in 3D VR generally induces more presence and more engagement.When students physically reenact, using gesture or embodied learning, greater learning gains are often seen. What this paper adds?: The high embodied, active VR group learned the most and reported the most engagement.Overall, being active and embodied resulted in more learning (regardless of the platform of desktop or VR).Embedded and embodied kinesthetic in‐process assessments during gameplay are valid measures of comprehension.The pathways leading to post‐play and 1 week follow‐up knowledge are complex. Implications for practice and/or policy?: Designers should strive to create embodied and agentic tasks in educational VR games.Watching a playback video on a 2D PC (desktop) platform can be appropriate for learning, but passive viewing videos in immersive VR can attenuate learning.Designers should include more embedded, in‐process data gathering during educational gameplay.Some guidelines are given for in‐process assessments, these assessments can lead to more dynamic, adaptive learning. [ABSTRACT FROM AUTHOR]