Can liking be instructed? Evaluations of familiar and unfamiliar food following Go/No-Go instructions and training

Go/No-Go (GNG) training can influence food evaluations, yet the processes underlying these effects remain debated. Associative accounts propose that evaluative change emerges through repeated stimulus-response learning, whereas inferential accounts emphasise the evaluative meaning individuals derive from stimulus-response relations. This research examined whether instructions about stimulus-response contingencies can influence food evaluations before the corresponding responses are performed. Across two preregistered studies, participants received either detailed contingency instructions or general task instructions. In both conditions, go responses were framed as selecting a food and no-go responses as rejecting a food. The detailed instructions additionally specified which food category was associated with each response. Detailed instructions also framed go responses as selecting a food and no-go responses as rejecting a food. Food evaluations were assessed after instruction but before training (T1), and after GNG training (T2). Study 1 (n = 54) used familiar food stimuli, whereas Study 2 (n = 63) used novel, unfamiliar food stimuli. At T1, evaluations of go- and no-go-associated foods did not differ between instruction conditions in either study. Following GNG training, go-associated stimuli were evaluated more positively in both studies, whereas no-go devaluation was observed for familiar foods in Study 1 but not overall for unfamiliar foods in Study 2. Exploratory analyses in Study 2 indicated stronger no-go devaluation among participants higher in trait impulsivity. These findings suggest that providing information about stimulus-response contingencies before training was not reflected in differences in food evaluations. The observed changes in evaluations following training suggest that the evaluative impact of these stimulus-response assignments may depend in part on how responses are experienced and interpreted during task performance, although this possibility requires further investigation.