Consistently not responding to stimuli during go/no-go training leads to lower evaluations of these NoGo stimuli. How this NoGo-devaluation-effect can be explained has remained unclear. Here, we ran three experiments to test the hypothesis that people form stimulus-stop-associations during the training, which predict the strength of the devaluation-effect. In Experiment 1, we tried to simultaneously measure the stimulus-stop-associations and NoGo-devaluation, but we failed to find these effects. In Experiment 2, we measured NoGo-devaluation with established procedures from previous work, and stimulus-stop-associations with a novel separate task. Results revealed a clear NoGo-devaluation-effect, which remained visible across multiple rating blocks. Interestingly, this devaluation-effect disappeared when stimulus-stop-associations were measured before stimulus evaluations, and there was no evidence supporting the formation of the stimulus-stop-associations. In Experiment 3, we found evidence for the acquisition of stimulus-stop-associations using an established task from previous work, but this time we found no subsequent NoGo-devaluation-effect. The present research suggests that the NoGo-devaluation-effect and stimulus-stop-associations can be found with standard established procedures, but that these effects are very sensitive to alterations of the experimental protocol. Furthermore, we failed to find evidence for both effects within the same experimental protocol, which has important theoretical and applied implications.
Keywords: Bayesian multilevel models; Go/no-go training; NoGo-devaluation; preregistration; stimulus-stop association.