Whereas experiments 1 and 2 employed pre and post evaluations of the chinese characters and were run locally, this experiment employed only post-IAT evaluations and was run online. We replicate the basic effect on the implicit measure but not the explicit.
Additionally, we examined whether the evaluative learning effect was related to the ability to analogically relate the Chinese characters and valenced images with other stimulus pairs (e.g. same: "hot-hot" vs. opposite: "hot-cold"). However, only a very small proportion of participants passed the analogy test (18/316 participants, c.6%) making results difficult to interpret.
Due to space concerns, this experiment was therefore not reported in the manuscript.