In an evaluative conditioning procedure (EC), a conditioned stimulus (CS) is paired with positive or negative unconditioned stimuli (US) and consequently acquires US valence. We introduce a sampling perspective into the EC paradigm by repeatedly giving participants agency over which CSs they want to sample (i.e., for what CSs they want to see conditioning trials). Half of the CSs was paired positively, while the other half was paired negatively. We also introduced a yoked condition, in which participants saw the same information produced by the sampling condition, but did not get to make any decisions. In three experiments, we found that over time a preference for sampling positively paired CSs over negatively paired CSs develops. We replicated standard EC effects and found that a higher number of pairings leads to more positive evaluative shifts, but more so in the sampling than in the yoked condition. This finding shows that evaluative conditioning depends on the learner’s willingness to attend to certain CSs. The results of this set of studies also suggest that agency increases the readiness to process the presented CS-US pairs. This reasoning is supported by memory data that indicate better memory for the number of pairings and paired valence in the sampling condition compared to the yoked condition. Implications for the external validity of the EC paradigm are also discussed.

The research reported in this article was supported by an Emmy Noether grant from the Deutsche Forschungsgemeinschaft (HU 1978/4-1) awarded to Mandy Hütter.