Experiencing nature leads to healthier food choices

The study investigates whether experiencing nature causally leads to healthier food choices, addressing a critical public health question as modern lifestyles involve more time indoors and increasing urbanization. Prior work links nature exposure to psychological, cognitive, physiological, and medical benefits and correlational associations with healthier diets, but causal mechanisms and generalizability to typical eating remain unclear. The authors define natural environments on a continuum from primeval to fully urban, noting that even urban green elements (e.g., parks, tree-lined streets) can confer benefits. Existing evidence on diet-related outcomes is limited, often correlational or focused on delay discounting, niche populations (e.g., dieters), or indirect self-reports rather than consequential choices. There is a need to test whether nature exposure promotes truly healthy, nutritious, unprocessed foods (e.g., fruits, vegetables) versus merely “diet/light” options, whether it affects food quantity, and whether effects are driven by nature exposure increasing healthy choices or urban exposure increasing unhealthy choices. The authors hypothesize that nature exposure increases the weight placed on health in food decisions without altering perceived healthiness per se, does not change total intake quantity, does not increase preference for diet/light foods, and that effects are due to nature (vs. urban or neutral control).

Correlational studies associate greater connectedness to nature with increased fruit and vegetable intake and higher dietary diversity, and proximity to green spaces with lower obesity rates. Experimental work has largely examined nature’s effects on delay discounting and impulsivity, with mixed or indirect evidence on eating (e.g., reduced sugar added to bubble tea among weight-loss motivated participants; family-based programs increasing connectedness rather than direct nature exposure; contradictory findings on stress eating). Field evidence includes: (1) a vending machine study where a nature poster increased healthier snack sales relative to a fair-themed or no poster, but lacked urban-scene controls and analyzed aggregate sales, limiting individual-level causal inference; (2) a restaurant study where a nature ambiance marginally increased vegetarian options vs. the old ambiance and showed no credible difference vs. a fast-food ambiance. Overall, prior work is underpowered, uses indirect measures (desirability scales), focuses on special populations or atypical decisions, and leaves unresolved whether typical eaters make healthier choices after nature exposure across diverse foods and contexts. This paper addresses these gaps with pre-registered, powered experiments using consequential choices and direct behavioral measures, including controls for urban vs. neutral environments and disentangling health vs. dieting motivations.

Design: Five between-subjects experiments tested whether exposure to nature (vs. urban, and vs. neutral control in one study) leads to healthier food choices and probed mechanisms. Online studies were pre-registered with sample sizes determined via power analyses; all studies received IRB approval, informed consent, and adhered to ethical standards.

• Study 1 (field, Paris, France): Participants (N = 39; 20 nature, 19 urban) were randomly assigned to a 20-minute walk along a mapped route in a large park (nature) or nearby city streets (urban). A photography task at route landmarks served as cover story and compliance check. Post-walk, participants had 10 minutes at a snack buffet with four healthy (bananas, apples, dried fruits, mixed nuts) and four unhealthy options (chips, brownies, strawberry/apricot cookies; latter collapsed in coding). Food type and quantities consumed were recorded. Primary outcomes: number of healthy vs. unhealthy servings and total servings.
• Study 2 (online, US; preregistered): Participants (N = 698 after exclusions) imagined a hotel room with a window view showing nature, urban, or closed curtains (control). They chose a lunch consisting of one main, one side, and one beverage from 12 items (4 mains, 4 sides, 4 drinks) pretested on healthiness (2 healthy/2 unhealthy per category). Outcomes: proportion of healthy choices and modeling of choice as a function of item healthiness by condition. Analyses included binary logistic regressions across the 12 items and conditional logistic regressions at the meal-choice set level, with interactions between item healthiness and condition indicators.
• Study 3 (online, US; preregistered): Participants (N = 885) saw the same nature/urban scenes as Study 2 (without window frames) and were randomized to measurement order (healthiness ratings then choices vs. choices then ratings). Each participant rated the healthiness of the 12 menu items used in Study 2 and made the same meal choices. Outcomes: percent healthy choices; conditional logistic regression with participant-specific mean-centered healthiness ratings predicting choice, testing the interaction of ratings with condition and measurement order to assess whether nature exposure increases the weight placed on health in choices without altering perceived healthiness.
• Study 4 (online, US; preregistered; incentive-compatible): Participants (N = 1191) viewed nature or urban photographs taken by the same photographer (attention checks and minimum viewing time enforced). They chose one of three snacks described textually: “a natural, healthy snack,” “a tasty, indulgent snack,” or “a diet, light snack.” Ten participants were randomly selected to receive their chosen snack (financial compensation used for logistics). Outcome: distribution of choices across the three goals to distinguish health vs. dieting vs. taste motivations.
• Study 5 (online, UK; preregistered): Participants (N = 913) viewed nature (waterfront with green cliffs) or urban (modern building) photos from a prior low-powered, inconclusive study. They completed the same lunch choice task as in Studies 2–3. Outcomes: percent healthy choices; conditional logistic regression examining interactions between item healthiness and condition; robustness assessed using externally collected continuous healthiness ratings. Measures and analyses: Healthiness was coded dichotomously (healthy = +½, unhealthy = −½) based on pretests (Studies 1–2, 5) and via participant-provided continuous ratings (Study 3) or external ratings (Study 5). Choice modeling used binary logistic regressions across items and conditional logistic regressions (CLOGIT) clustering by participant to reflect forced-choice sets (1 of 4 per category). Study 1 additionally used mixed-effects models to account for multiple observations per participant. Pre-registered exclusion criteria (attention checks, device eligibility) were applied in online studies.

• Study 1 (field, Paris): Total quantity consumed did not differ by condition (Mnature = 2.58 servings, SD = 1.41; Murban = 2.70, SD = 1.08; F(1,37) = 0.10, p = 0.76, η² = 0.003). However, nature increased healthy intake and reduced unhealthy intake: healthy servings Mnature = 1.80 (SD = 1.28) vs. Murban = 1.05 (SD = 0.85), F(1,37) = 4.56, p = 0.04, η² = 0.110; unhealthy servings Mnature = 0.78 (SD = 0.84) vs. Murban = 1.65 (SD = 1.38), F(1,37) = 5.82, p = 0.02, η² = 0.136. Mixed-effects model showed a significant NATURE × HEALTHINESS interaction (B = 0.48, t = 2.93, p < 0.01).
• Study 2 (online, US): Healthy choices: urban 62%, control 65%, nature 71%; χ²(2,2094) = 13.76, p < 0.01. Logistic regressions showed a significant HEALTHINESS main effect (B = 0.89, p < 0.01). Interaction HEALTHINESS × NATURE vs control significant (B = 0.35, p < 0.01), HEALTHINESS × URBAN vs control not significant (B = −0.19, p = 0.13). Comparing nature vs urban directly: HEALTHINESS × NATURE vs urban significant (B = 0.54, p < 0.001). Conditional logistic regression replicated: HEALTHINESS main effect (B = 0.67, z = 12.56, p < 0.001); HEALTHINESS × NATURE vs control (B = 0.27, z = 2.03, p = 0.04); HEALTHINESS × URBAN vs control not significant (B = −0.15, z = −1.15, p = 0.25). Conclusion: nature exposure drove healthier choices; urban did not drive unhealthier choices relative to control.
• Study 3 (online, US): Nature increased healthy choices overall (Mnature = 70.6% vs. Murban = 63.7%; χ²(1,2655) = 14.15, p < 0.001), both when choices preceded ratings (65.9% vs. 57.5%; χ²(1,1317) = 9.78, p < 0.01) and followed ratings (75.2% vs. 69.8%; χ²(1,1338) = 4.95, p = 0.03). No difference in perceived healthiness between conditions (Mnature = 3.94, SD = 2.25; Murban = 3.98, SD = 2.22; F = 1.11, p = 0.29). Conditional logistic regression: HEALTHRATING main effect (B = 0.25, z = 20.64, p < 0.001) and stronger weight on health in nature (HEALTHRATING × NATURE: B = 0.11, z = 4.58, p < 0.001), with larger coefficient in nature (B = 0.30, SE = 0.02) vs. urban (B = 0.19, SE = 0.02). Measurement order moderated the ratings–choice link (B = 0.11, z = 4.53, p < 0.001) but did not interact with NATURE (three-way interaction ns).
• Study 4 (online, US; incentive-compatible): Nature increased selection of a natural, healthy snack (72.9% nature vs. 33.2% urban; χ²(1,1191) = 188.35, p < 0.001) and decreased preference for diet/light snacks (5.4% vs. 10.3%; χ²(1,1191) = 10.08, p = 0.002) and tasty/indulgent snacks (21.7% vs. 56.5%; χ²(1,1191) = 151.10, p < 0.001). Indicates enhanced health goals rather than dieting or taste goals.
• Study 5 (online, UK; robust replication with prior stimuli): Nature increased healthy choices (Mnature = 64.1% vs. Murban = 54.3%; χ²(1,2739) = 27.27, p < 0.001). Conditional logistic regression: HEALTHINESS main effect (B = 0.38, z = 7.69, p < 0.001) and positive interaction HEALTHINESS × NATURE (B = 0.41, z = 4.17, p < 0.01). Using continuous external health ratings: HEALTHRATING main effect (B = 0.12, z = 9.42, p < 0.01) with positive interaction HEALTHRATING × NATURE (B = 0.10, z = 3.90, p < 0.01). Overall: sufficiently powered designs reliably show nature exposure increases the likelihood of choosing healthier items across countries and contexts.

Across five experiments spanning field behavior and incentive-compatible online choices, exposure to natural environments led to healthier food selections relative to urban or neutral settings. The effect generalizes across countries (France, United States, United Kingdom), contexts (post-walk buffets; room-service style meal choices; abstract snack goals), and time (2016–2023), indicating robustness even amidst societal shifts such as COVID-19. Critically, comparison to a neutral control shows that nature exposure actively promotes healthier choices, rather than urban exposure actively promoting unhealthier ones. Mechanistically, results show that nature exposure increases the weight placed on health considerations in decision-making without altering perceived healthiness of foods, and it shifts preferences away from indulgence and from low-calorie “diet/light” options toward natural, unprocessed healthy options. Supplementary implicit association evidence links nature with health. The findings inform theories of how environmental cues and connectedness to nature shape food-choice processes and suggest that even brief or indirect exposure (e.g., images) can influence consequential dietary decisions.

The paper demonstrates that experiencing nature causally increases healthy food choices without increasing total intake volume, and that it specifically elevates health-oriented goals rather than dieting or indulgence. Effects are consistent across diverse foods, tasks, populations, and stimuli, and are attributable to nature exposure rather than urban exposure. These insights advance understanding of environmental determinants of diet and the decision weights underlying food choice. Practical implications include integrating natural elements into urban design and eating environments (e.g., parks near cafeterias, nature imagery for naturally healthy foods) to nudge healthier choices. Future research should quantify minimal effective doses of nature, test multiple mechanisms (affect, stress, restorativeness, delay discounting, self-perception) concurrently, explore boundary conditions (e.g., role of greenery, seasonal landscapes, awe/vitality), and assess longer-term, longitudinal dietary impacts and heterogeneity across populations.

The studies do not establish how long the effects of nature exposure on food choices persist (short-term, single-occasion focus). Only one eating occasion (snack or lunch) was examined; longitudinal impacts on diet quality and quantity remain unknown. Potential heterogeneity across populations (e.g., urban vs. suburban/rural residents) was not tested. The operationalization of “healthy” focused primarily on item healthiness rather than total caloric intake or portion size; integrating quantity and broader nutrition metrics would add nuance. Access to nature is unequal across demographic groups (e.g., racial/ethnic disparities in urban green space), which may influence generalizability and intervention feasibility. Larger-scale field interventions and dose–response studies are needed.