Food-level predictors of self-reported liking and hedonic overeating: Putting ultra-processed foods in context

The reward value people assign to foods is determined by their intrinsic (food-level) properties and moderated by individual factors such as traits, states and beliefs. There is a need for more systematic, structured analyses of the food-level characteristics that explain cognitions about food reward such as palatability and their risk for reward-driven overeating. This research, consisting of three studies, aimed to explore the nutritional, sensory and cognitive characteristics and attributes of foods as determinants of food reward-related outcomes. Across three sequential online study designs, 1176 men and 2188 women from the general population rated sub-samples of 436 foods which were sampled from databases and photographed to represent ready-to-eat food and beverage products in the UK. The study outcomes were self-reported food liking and hedonic overeating, while the predictors were the nutritional composition of the foods including ultra-processed food status (UPFs) and carbohydrate-to-fat ratio (CFR); and participants' self-reported beliefs about the nutritional and sensory characteristics of the foods. Correlation and stepwise regression analyses were used to model significant nutritional components followed by hierarchical regression models to examine self-reported food-level attributes, or CFR and UPFs as potential additive models. Across all studies, the nutritional characteristics of foods explained ∼20 % variance in liking and 40–60 % variance in hedonic overeating. Self-reported food-level attributes explained a further 6–33 % variance in liking and 17–38 % variance in hedonic overeating. UPFs explained 0–7 % additional variance and CFR did not add to the nutritional models. This research demonstrates how nutritional characteristics of foods contribute to self-reported liking and hedonic overeating. Considering people's beliefs about nutrient and sensory attributes can explain more than nutrients alone, and there are negligible additive contributions from CFR or UPFs on food reward.