Developing a novel computational model to quantify predictors of sustainable food choice

This is the authors' abstract. We don't add key points for this paper.

Understanding drivers of food choice is essential for establishing strategies to improve diet sustainability. This online study used a Bayesian multilevel computational approach to model predictors of food choice and quantified how product attributes were associated with choice of a meat-containing versus a fish-containing or non-meat meal. Adults ( N = 523 omnivores) were presented with images of 105 meal pairs formed from 15 different meals that were meat, fish, or non-meat. For each pair, participants were asked to imagine choosing one of two meals to eat for lunch. Ratings of perceived tastiness, healthiness, expected satiation, and perceived macronutrient content were examined as predictors of meal choice. Participants were more likely to choose a meat-based meal over a non-meat meal. Choice of the non-meat option was predicted by higher ratings of taste and health, and lower ratings of expected satiety of the food. Additionally, choice of the non-meat option was predicted by participants being female. Socioeconomic status (SES) moderated the likelihood of picking a non-meat meal. Specifically, high SES participants were more likely to select a non-meat meal if it was rated as healthier or lower in carbohydrate content. Overall, we found that omnivores were more inclined to select non-meat meals when they were regarded as healthier and tastier than meat meals. Our conceptual model suggested that increasing perceived taste may theoretically shift choices towards non-meat options among females, but this requires validation in real-world contexts. Research in real-world contexts is needed to determine whether these findings translate into actual sustainable food choices.