Testing the validity of immersive eating environments against laboratory and real life settings

Immersive environments are widely used as a proxy for real life settings for consumer and sensory evaluations. Yet, the external validity of immersive environments is poorly understood. This study investigated whether sense of presence, liking, beverage desire, consumption, and choice are consistent across a real-life setting compared to a sensory laboratory and two different immersive environments. This study followed a randomised balanced cross over design. Participants (n = 80; n = 25 male, n = 55 female, aged 18–65 years) completed four evaluations separated by two to four weeks. Conditions included 1) an individual partitioned testing booth (control), 2) an evoked context showing photos of a café, 3) a virtual reality (VR) session showing a 360○ recording of the same café via a VR headset (360VR), and 4) the same café in real life. After one-minute immersion time, participants rated desire (9-point scale) for hot and cold beverages. Then, participants tasted 25 ml samples of water (control n = 1) and apple juice (n = 5, randomised) varying in sweetness and sourness, and rated liking (9-point hedonic scale) and prospective consumption (5-point scale from ‘none’ to ‘two or more cups’). Participants were then provided with a ∼ 530 ml cup of apple juice and asked to sip, complete the 25 item EsSense Profile (a measure of emotion), and then consume ad libitum. On completion, participants ordered a drink from a menu of eight hot or cold, high or low energy options, and reported sense of presence (7-point Likert scale). Sense of presence increased significantly from the laboratory (estimated means 1.81 ± 0.12), evoked context (3.60 ± 0.15), 360VR (5.25 ± 0.15), and real café (6.40 ± 0.11) conditions, all p values < 0.001. Compared to the laboratory, samples were liked more when evaluated in the evoked context (estimated mean difference from the laboratory 0.22 ± 0.11, p = 0.04) and the real café (estimated mean difference 0.22 ± 0.09, p = 0.01), but not the 360VR condition (estimated mean difference = 0.17 ± 0.10, p = 0.10). There was no significant difference in the frequency of emotion terms selected between conditions (all p values > 0.05), and condition was not significantly associated with apple juice prospective consumption (p = 0.10), intake (grams) (p = 0.16) and time taken for consumption (p = 0.41). Moreover, there was no significant association between the type of drink chosen from the menu and condition (p = 0.76). The 360VR condition was more successful than the picture evoked context in making participants feel they were present in a café. However, liking measures recorded in the 360VR condition were consistent with liking measured in the laboratory. In this case, measurement of apple juice liking is not suitable under controlled laboratory conditions and should be measured under more realistic context, either in real life or realistic conditions that are more immersive than those used in this study. Whereas both the laboratory and immersive environments were valid settings for measuring beverage choice and desire (to some extent), apple juice emotion profile and consumption, suggesting that taking these measurements under immersive or real-life café conditions is unnecessary for apple juice.