Multivariate patterns of food image evaluations distinguish obesity: A key role of highly processed foods.

Subjective evaluations of foods, such as their perceived tastiness and healthiness, influence dietary decisions. Distortions in these evaluative processes are often implicated in the onset and maintenance of obesity. However, it remains uncertain how subjective food evaluations differ between individuals with obesity and lean individuals. In this preregistered study, we demonstrated that evaluation patterns over hundreds of food items modestly but significantly distinguish individuals with obesity from lean individuals. Specifically, we conducted a behavioral study and analyzed data from both lean Japanese participants and Japanese participants with obesity (n = 107 and 93, respectively), who had rated 896 food images for three attributes (i.e., likability, tastiness, and healthiness). Using partial least squares discriminant analysis, we identified latent features of subjective food evaluation that distinguished individuals with obesity from lean individuals. Remarkably, these latent features were closely linked to differential evaluations of highly processed versus minimally processed foods. Together, these findings reveal that subjective evaluations of processed foods show multivariate patterns associated with current obesity status, offering new insights into food evaluation in obesity.