For the sake of transparency, here we provide an overview of all the pre-registrations for this set of experiments. Two major deviations from the pre-registrations are: 1. In some experiments (i.e., 3, 4a, 4b, 5), only the untrained items were mentioned as a baseline in the pre-registrations. In the process of conducting experiments, we thought using two baselines (i.e., go and untrained items) could rule out some alternative explanations and serve as a more strict test of the devaluation effect. Therefore in the main text, two baselines were used for each experiment. 2. The exclusion criteria were not explicitly mentioned in some experiments (i.e., 4a, 4b). Including all participants did not change the overall pattern of the results. Below we provide the links to all the pre-registrations and list the differences between the final results and the registered plans for each experiment. ---------- **Experiment 1** No pre-registration. ---------- **Experiment 2** [Click here for pre-registration][2] 1. In pre-registration, we did not specify exclusion criteria. Including all the participants leads to the same result. 2. In pre-registration, repeated-measures ANOVAs with measurement time (before vs. after the training) and food type were planned. For the sake of consistency with other experiments, we calculated the difference score for each type of food and did not use measurement time as a factor in rmANOVA. The two analyses yield the same result. ---------- **Experiment 3** [Click here for pre-registration][1] 1. In pre-registration, only untrained items were mentioned as a baseline. ---------- **Experiment 4a** [Click here for pre-registration][3] 1. In pre-registration, only untrained items were mentioned as a baseline. 2. No exclusion criteria were specified in pre-registration. Including all the participants leads to the same result. ---------- **Experiment 4b** [Click here for pre-registration][4] 1. In pre-registration, only untrained items were mentioned as a baseline. 2. No exclusion criteria were specified in pre-registration. Including all the participants leads to the same result. ---------- **Experiment 5** [Click here for pre-registration 1][5] [Click here for pre-registration 2][6] 1. About two pre-registrations: after the first few participants, the experimenter noticed that for some participants their contingency awareness was below 50%. This could suggest that these participants remembered the association between cues and responses in an opposite way (although no direct evidence existed for this). In order to be more conservative, a second exclusion criterion was added to the second pre-registration (when the experiment was still ongoing and data analysis not started yet). 2. In pre-registration, only untrained items were mentioned as a baseline. ---------- **Experiment 6** [Click here for pre-registration][7] [1]: https://osf.io/xfzd4/?view_only=51e12b7c9e044138a8eb0522b35a9310 "pre-registration exp3" [2]: https://osf.io/7nxk5/?view_only=f4e5a4f8903e495692896c2ec7366d73 "pre-registration exp2" [3]: https://osf.io/wxcpr/?view_only=11f983a072e84997926485a694d8d001 "pre-registration exp4a" [4]: https://osf.io/fz635/?view_only=a8cfd0c37d8043d39f5c8cb5b5e0791e "pre-registration exp4b" [5]: https://osf.io/qw8h2/?view_only=5d356a42b32f4f3594bf6dc8e38a6ce5 "pre-registration exp5 - 1" [6]: https://osf.io/pkmby/?view_only=311f832efeea4763bc8191cab883dfe7 "pre-registration exp5 - 2" [7]: https://osf.io/mwq9z/?view_only=fea7f26e1169442e87603c765d0ac790 "pre-registration exp6"