Exploring consumer acceptability of leafy greens in earth and space immersive environments using biometrics

Future long-duration missions (e.g., Artemis to the Moon and eventual Mars missions) will require fresh foods from specialized, potentially genetically modified plants to meet nutrition and sensory needs of astronauts. Prior Earth-based studies simulating microgravity suggest decreased aroma perception and increased basic taste intensity, but many have limitations such as small or untrained panels and reliance on univariate statistics. Environments also modulate sensory perception, with immersive and VR settings affecting acceptability and appropriateness compared to traditional booths. This study addresses two core constraints: simulating microgravity and space environments on Earth, and ethical/logistical issues of testing GM-derived foods with humans. The research question is whether immersive space environments and microgravity-like seating positions influence consumers’ acceptability and biometric (physiological and emotional) responses to pick-and-eat leafy greens, generating foundational data for future digital twins to evaluate GM plants for space.

Previous simulated microgravity research (e.g., bed rest, reclining positions) reported reduced olfactory sensitivity and increased taste intensity, though findings vary due to methodological issues (small, untrained panels; limited statistics). More recent work with trained panels using reclining chairs aligns with astronaut reports, showing decreased aroma and increased basic tastes; potential changes in mouthfeel have also been noted. Context effects are well-documented: immersive and VR environments can enhance perceived intensities and situational appropriateness versus standard booths. However, no prior studies jointly examined immersive space environments with microgravity-like positions while capturing emotional and physiological biometrics. Emerging proposals advocate using AI and digital twins to replace or augment human sensory trials, especially important for GM plant foods destined for space.

Design and stimuli: Six leafy greens were grown in three FarmBot systems at The University of Melbourne: Thai basil (Ocimum tenuiflorum ‘Krapao’), sweet basil (Ocimum basilicum), coriander (Coriandrum sativum), kale (Brassica oleracea var. sabellica), mixed cos lettuce (Lactuca sativa L. var. longifolia), and beetroot (Beta vulgaris). Best-quality specimens were harvested the day before testing, washed, stored at 5 °C in sealed containers with damp paper towels, and presented as pick-and-eat samples. Participants: N=51 consumers (37% male, 63% female; 18–51 years), regular leafy green consumers (≥ once/week), Australians or Asians living >2 years in Australia. Power analysis (overall liking) indicated adequate power (1-β=0.97). Ethics approved by UoM HREC (ID: 1953926.6); informed consent obtained. Environments and treatments: Two immersive rooms: (i) space environment (Earth-from-space video on a 180° screen) and (ii) neutral environment (white 180° screen). Two seating positions: (i) normal (90–270°) and (ii) simulated microgravity (0–170°) using Timber Ridge reclining chairs with tablet mounts. Each participant completed four treatments (environment × position), with randomized starting room and balanced order; one participant per room at a time; 40–60 min break between treatments. Procedure: Samples were coded with three-digit random codes and presented all at once; tasting order per treatment was randomized via the Bio-Sensory App (Android). Water and crackers were provided between samples. Measures: Self-reported affective ratings on 15-cm unstructured line scales for aroma, texture, bitterness, sweetness, aftertaste, overall liking, and a FaceScale (emotional valence). A Check-All-That-Apply (CATA) question with emojis captured conscious emotional associations. Biometric data were recorded via the tablet’s frontal camera during tasting and analyzed using custom DAFW-UoM software (Affectiva SDK-based) for facial expressions/emotions (e.g., joy, fear, disgust, sadness, anger, surprise, contempt), dimensions/states (valence, engagement, relaxed), head movements (pitch, yaw, roll), and physiological signals (heart rate; blood pressure systolic/diastolic) via rPPG-based algorithms. Statistical analysis: One-way ANOVAs with Fisher’s LSD (α=0.05) tested level-two interactions: environment × seating position, sample × seating position, and sample × environment. CATA data were analyzed using Cochran’s Q with Sheskin pairwise tests and Correspondence Analysis (CA) to assess associations among samples/treatments and emojis. Principal Components Analysis (PCA) examined relations among self-reports, emotions, biometrics, and treatments/samples. ANOVAs and CAs were performed in XLSTAT 2022.3.2; PCAs in MATLAB R2021a.

Univariate (ANOVA):

• Environment × seating position: No significant differences in self-reported acceptability; significant differences (p<0.05) in head movements. In neutral environment, microgravity position elicited upward pitch (head up) vs normal; normal in space environment also showed increased pitch, likely due to attending to the 180° screen. Microgravity reduced lateral head movement (yaw) vs normal, consistent with a more relaxed, weightless sensation; roll differed with microgravity in space showing right tilt (negative), while others showed left tilt (positive).
• Sample × environment: No significant effects on self-reported or biometric variables.
• Sample × seating position: Significant differences (p<0.05) for all liking attributes; patterns were primarily driven by sample differences rather than position. Product liking patterns:
• Sweet basil: Most liked for aroma.
• Lettuce: Highest liking for texture, bitterness, sweetness, aftertaste, overall, and highest FaceScale (positive emotion).
• Kale: Least liked for aroma and overall; among lowest for texture, sweetness, aftertaste; most negative FaceScale.
• Thai basil: Lowest liking for bitterness; among lowest for sweetness and aftertaste.
• Coriander: Among least liked for texture; beetroot and coriander showed mixed/neutral profiles on other attributes. Multivariate analyses:
• CATA/CA: Normal seating in both environments associated with more positive and some neutral emotions; microgravity in space showed mixed emotions and in neutral environment associated with distinct sets of emotions, indicating novelty/unfamiliarity of position. Little to no environment effect on emotions by sample: the same samples grouped across environments. Lettuce, sweet basil, and coriander clustered with more positive emotions; beetroot associated with mixed emotions; Thai basil and coriander (in space) leaned negative; kale associated with negative emotions, consistent with bitterness rejection.
• PCA (seating × environment): 79.9% variance explained (PC1+PC2). Microgravity (both environments) associated with fear, disgust, and elevated heart rate; also positively related to aroma liking (potentially due to reduced aroma intensity being perceived as more acceptable). Normal/neutral associated with FaceScale, valence, and liking of bitterness, sweetness, aftertaste, texture; normal/space related to engagement, anger, relaxed, contempt, and yaw.
• PCA (sample × environment): 54.3% variance explained. Clear separation of environments along PC2: neutral on positive side; most space-environment samples on negative side (except kale). Coriander, lettuce, beetroot, and sweet basil in neutral grouped with positive emotions and higher sweet/bitter liking and FaceScale. Kale (both envs) and Thai basil (neutral) associated with higher HR, blood pressure, and sadness. In space, Thai basil and coriander associated with engagement, disgust, surprise, fear; beetroot, sweet basil, lettuce associated with aroma, pitch, texture, overall liking.
• PCA (sample × position): 55.4% variance explained. Separation by position along PC2: normal near origin/positive PC2 with engagement, relaxed, smile; microgravity samples farther from origin indicating more intense emotions. Thai basil, kale, coriander, sweet basil in microgravity associated with surprise, increased HR/BP, and disgust/aroma; beetroot and lettuce in microgravity grouped with higher liking across descriptors (except aroma, pitch), aligning with intensified taste perception and sweeter profiles. Overall: Multivariate analyses revealed nuanced effects of immersive environments and microgravity-like positions on self-reported, emotional, and physiological responses not captured by ANOVA alone, supporting the use of biometric-driven models for digital twin development.

The study demonstrates that while univariate analyses find limited environment or position effects on self-reported acceptability, immersive space contexts and microgravity-like positions measurably alter head movements, emotional states, and physiological responses captured via biometrics. Normal seating aligns with more positive, familiar emotional profiles and higher liking of key sensory attributes, whereas microgravity evokes mixed or heightened emotions and elevated autonomic responses, reflecting novelty and altered sensory processing. Product effects dominate liking differences, with lettuce and sweet basil eliciting more positive responses, while kale and Thai basil trigger more negative affect consistent with bitterness and pungency. Multivariate techniques uncover separations between environments and positions and highlight associations between specific biometric/emotional variables and treatments, providing richer insight into causal patterns than ANOVA alone. These findings address the research question by evidencing that simulated microgravity and space-immersive environments modulate consumer perception and biometric responses, validating immersive simulation as a proxy for space conditions and motivating integration of biometrics into digital twin models for predicting acceptability of plant foods in space.

This study pioneers the combined use of immersive space environments, simulated microgravity seating, and non-invasive biometrics to assess consumer acceptability of pick-and-eat leafy greens. Key contributions include: (i) documenting significant head-movement changes with environment × position, (ii) showing sample-driven differences in liking, with lettuce and sweet basil favored and kale/Thai basil less preferred, and (iii) revealing, via CA and PCA, distinct emotional/physiological patterns across environments and positions that are not evident in univariate tests. The results support the feasibility of Earth-based immersive simulations to approximate space influences on sensory perception and provide essential digital information for constructing AI-driven digital twins to evaluate GM plants and derived foods for long-term missions. Future work should: (a) extend to astronaut populations and real microgravity when feasible, (b) incorporate more complex foods (e.g., salads, prepared dishes), (c) refine biometric pipelines and expand multimodal sensing, and (d) train predictive machine learning models linking plant production variables to consumer sensory, emotional, and physiological responses.

Primary limitations include the inability to test with astronauts and in real space conditions; only Earth-based immersive rooms and simulated microgravity positions were used. Findings may be influenced by novelty effects of the microgravity posture and the specific immersive content. The study focused on six leafy greens and pick-and-eat formats; generalizability to other foods or prepared dishes requires further research. Although adequately powered for consumer testing, broader demographic diversity and cross-cultural validation would strengthen generalizability.