for assistance with data collection; Fabienne Voncken, Simke van Oijen and Sjors van de Schoot for creating the candy images; Xin Gao, Haokui Xu and members of the Food Choice on Impulse lab for helpful comments on earlier versions of the manuscript. The data of Experiments 1 and 2 have been reported in a master thesis by Linda Schmale at the University of Amsterdam. The data of Experiments 5 and 6 have been reported in bachelor theses by Roos Greven and Evert Palm at Radboud University. The data of Experiment 7 have been reported in bachelor theses by Wouter Aarts, Michelle de With and Gerden Ibrahim at Radboud University. Parts of the data reported here have been presented at the 2016 and 2017 annual meetings of Dutch Association of Social Psychology (ASPO; the Netherlands, 2016 and 2017), and the preconference on the self-regulation of health at the 18th General Meeting of the European Association of Social Psychology (Spain, 2017) by the first author.

Evaluations of stimuli can be changed by simple motor responses such that stimuli to which responses are consistently withheld tend to be evaluated less positively than other stimuli. The exact mechanism that underlies this no-go devaluation effect is still unknown. Here we examine whether attention to the stimuli during training contributes to the devaluation effect. Participants received a go/no-go training in which 2 go items or 2 no-go items were simultaneously presented, and attention to 1 of the items was cued before participants executed or withheld a simple motor response (press a key on a keyboard). Next, explicit evaluations of these stimuli and untrained stimuli were assessed. Across 2 experiments we observed a predicted no-go devaluation effect, that is, a decrease in evaluations for items that have not been responded to. Furthermore, as predicted, selectively cueing attention toward stimuli during go/no-go training amplified differences in subsequent evaluations between go and no-go stimuli. Confirmatory analyses showed that the devaluation effect for cued no-go stimuli was not statistically significantly stronger than that for uncued no-go stimuli within each experiment. However, combining the data of both experiments showed moderate evidence (p ϭ .023, BF ϩ0 ϭ 5.88) for stronger devaluation of cued no-go stimuli compared with uncued no-go stimuli. We conclude that attention to stimuli during go/no-go training contributes to revaluation processes of stimuli via motor actions, and that this knowledge is relevant for a better understanding of the underlying mechanism of the training and to optimize go/no-go training for practical use. Public Significance StatementThis study suggests that responding or not responding to food items can respectively increase or decrease liking of these items, particularly when people attend closely to them. This finding is important to optimize applied response training tasks to change people's responses to food items.

Understanding the formation and modification of preferences is important for explaining human behavior across many domains. Here we examined when and how preferences for food items can be changed by linking mere action versus inaction to these items. In 7 preregistered experiments, participants were trained to consistently respond to certain food items (go items) and not respond to other items (no-go items) in a go/no-go training. Next, to assess preferences, they repeatedly chose between go and no-go items for consumption. Decision time during the choice task was manipulated and measured. Immediately after training, participants chose go items more often for consumption when choosing under time pressure, for both high-value and low-value choice pairs. Preferences were reliably changed in favor of go items for choices between unhealthy foods, between healthy foods, and between healthy and unhealthy foods. Furthermore, preference change was still observed one week after training, although the effect size largely decreased. Interestingly, when participants made choices without time pressure, the effect became weaker and statistically non-significant. These results suggest that preference change induced by mere responding versus not responding is constrained to situations where people take little time to make decisions, and the effect is relatively short-lived. By showing the reliability, generalizability and boundary conditions of the effect, these findings advance our understanding of the underlying mechanisms of go/no-go training, provide more insights into how the training can be effectively applied, and raise new theoretical questions on how mere action versus inaction impacts preferences.

Understanding how sustainable preference change can be achieved is of both scientific and practical importance. Recent work shows that merely responding or not responding to objects can change people’s preferences for these objects for at least one week. However, it is unclear whether such lasting preference change is caused directly by the mere (non)responses in the training, or indirectly by the measurement of preferences immediately after training. Two preregistered experiments addressed this question. Participants responded to go food items and withheld responses toward no-go food items in a go/no-go training. Immediately after training, they made consumption choices for half of the items. One week later, participants chose again. Half of the choices had been presented immediately after training (i.e., repeated choices), while the other half had not (i.e., new choices). Participants preferred go items over no-go items both immediately after training and one week later. Furthermore, the effect was observed for both repeated and new choices after one week, revealing a direct effect of mere (non)responses on preference over time. Exploratory analyses revealed that the effect after one week is related to the memory of stimulus-response contingencies immediately after training, and this memory is impaired by making choices. Thus, paradoxically, indicating preferences impairs memory of stimulus-response contingencies, which in turn may reduce training induced preference change. These findings show that mere action versus inaction can directly induce preference change that lasts for at least one week, and memory of stimulus-response contingencies may play a crucial role in this effect.

Understanding how sustainable preference change can be achieved is of both scientific and practical importance. Recent work shows that merely responding or not responding to objects during go/no-go training can influence preferences for these objects right after the training, when people choose with a time limit. Here we examined whether and how such immediate preference change in fast choices can affect choices without time limit one week later. In two preregistered experiments, participants responded to go food items and withheld responses toward no-go food items during a go/no-go training. Immediately after the training, they made consumption choices for half of the items (with a time limit in Experiment 1; without time limit in Experiment 2). One week later, participants chose again (without time limit in both experiments). Half of the choices had been presented immediately after the training (repeated choices), while the other half had not (new choices). Participants preferred go over no-go items both immediately after the training and one week later. Furthermore, the effect was observed for both repeated and new choices after one week, revealing a direct effect of mere (non)responses on preferences one week later. Exploratory analyses revealed that the effect after one week is related to the memory of stimulus-response contingencies immediately after the training, and this memory is impaired by making choices. These findings show mere action versus inaction can directly induce preference change that lasts for at least one week, and memory of stimulus-response contingencies may play a crucial role in this effect.