Seeing the whole picture? An experiment on environmental detail and coherence formation in immersive virtual reality

Immersive virtual reality (IVR) provides rich visual contexts that can enhance or impede learning, depending on how well environmental details align with instructional content. This experiment systematically examined the effects of environmental detail (between-subject factor: situated 360° scene vs. non-situated image) and content coherence (within-subject factor: text coherent vs. incoherent with visual context) on learning, coherence formation, attention, and motivation. A sample of N = 77 university students explored a virtual replica of the Sistine Chapel in a controlled lab setting. Learning outcomes, spatial presence, enjoyment, and cognitive load (as an exploratory measure) were assessed using questionnaires; coherence formation was specifically analysed using eye-tracking data. Contrary to our hypothesis, incoherent content led to better knowledge acquisition than coherent content, regardless of environmental detail. Situated scenes increased enjoyment and drew attention to environmental features but did not improve learning and even impaired it when content was coherent. Notably, we also observed sequencing effects: learning outcomes for the incoherent content were higher when students first engaged with coherent content, suggesting that early exposure to coherent material may help students better process subsequent incoherent information. Our findings challenge the assumption that immersive features inherently support learning and highlight the importance of considering sequencing and coherence in immersive instructional design. Beyond these empirical findings, the study introduces the concept of mental situation models, grounded in established theories of text comprehension and multimedia learning, as a new perspective to understand our findings and the processes of learning and coherence formation in IVR. This perspective provides a promising starting point for advancing theory building and for identifying the boundary conditions of immersive learning.