Nutritional labeling modifies meal composition strategies in a computer-based food selection task

•The study examined the influence of a nutritional labelling on food choice strategies of French students in a computer-based meal composition task.•When considering dishes individually, no significant effects of the labelling could be observed but revealed marked effects on meal composition strategies.•Two types of behaviors seemed to emerge under the influence of the nutritional labelling: homogeneous behaviors, where starters and desserts tended to have similar scores and compensatory behaviors where meals composed were made of foods with contrasted nutritional scores.•When considering the composition of an entire meal, the effect of a nutritional labelling can have complex consequences on food decision that extend beyond the selection of food items taken individually. Nutritional information via simplified labelling on products’ front of pack has become common in retail stores and is now concerning nowcollective catering. While numerous studies have investigated the effects of such information policies on consumers' decisions in shops, few studies have focused on choices made in collective catering. In such settings, consumers must compose a meal by combining dishes to be eaten during the same occasion. Each choice is then dependent of the selection of other foods, yielding a different decision problem as in a store where items are selected independently of one another. The aim of this study was to understand whether a nutritional labelling, (Nutri-Score®), modified the choices of consumers and more precisely modified the meal composition strategies - the associations between dishes made by consumers. A computerized menu composition task was designed, 371 participants were randomly redirected either to an interface displaying the Nutri-Score® of dishes, or to an interface showing the dishes without Nutri-Score®. Bayesian logistic regressions were used to explore dependency relationships between foods in presence or absence of Nutri-Score®. When considering dishes individually, no significant effects of the Nutri-Score® were observed, but significant effects of the Nutri-Score® on composition strategies could be observed. Two types of strategies seemed to emerge: homogeneous behaviors, where selected dishes had similar scores and, compensatory behaviors where selected dishes had contrasted scores. In conclusion, the effect of a nutritional labelling can have complex consequences on food decisions that extend beyond the selection of food items taken indi