Evaluating the Viability of Virtual Reality for Children's Food Choice Research: Comparative Mixed Methods Study.

Background: Virtual reality (VR) systems offer promising potential as a controlled platform to investigate human behaviors specifically related to food choices. Yet, little is known about its viability for conducting food choice studies with children, thus limiting the development of public health research. Objective: This study aimed to investigate the viability of VR technology for understanding children's food choices, focusing specifically on perceptual differences between VR and real-life (RL) settings. We examined how children perceive and interact with food portion sizes and container sizes in VR as compared with an equivalent RL scenario. Methods: A within-subject, mixed methods study was conducted with 437 children aged 5-12 years at a science museum. Participants engaged in a standardized food selection task for a simulated breakfast scenario, choosing portions of cereal and milk in 2 conditions: a head-mounted VR environment and a corresponding RL physical setup. Children's food selection behaviors were quantitatively compared across 3 independent variables: condition (VR vs RL), food healthiness (healthy vs unhealthy options), and container size (small, medium, and large). Qualitative and quantitative data were collected via postsession questionnaires assessing presence, embodiment, and simulator sickness, alongside detailed interaction logs from the VR environment. Data analysis used statistical comparisons and thematic analysis. Results: The findings revealed both behavioral consistency and significant perceptual differences between the VR and RL conditions. A behavioral similarity was identified, as children served significantly larger portions of unhealthy food compared with healthy food in both conditions. Crucially, a difference was observed in size perception: children struggled to accurately match the size of bowls and glasses between the VR and RL conditions. Furthermore, while children reported low feelings of presence and embodiment within the VR scenario, they demonstrated a high degree of control and engagement in the virtual task. Conclusions: Our findings suggest that current state-of-the-art VR technology presents limitations in its viability for conducting food choice studies with children, particularly concerning accurate size and volume perception. Based on the findings, we provide 4 practical recommendations to guide the future development of immersive food environments, thereby supporting more reliable and ecologically valid food choice research with young populations.