Reconsidering the (in)sensitivity of evaluative conditioning to reinforcement density and CS–US contingency

Evaluative conditioning (EC) refers to a change in the valence of a stimulus as a result of being paired with an affective stimulus. Based on previous findings, it has been argued that this form of learning does not depend on a statistical relation between the CS and the US. However, the empirical evidence on this issue is still rather limited and based mainly on null-effects. The present series of experiments tested the reliability of EC effects with different types of CS–US relations. The results of Experiment 1 show that unpaired CS presentations during conditioning (i.e., modulations of reinforcement density) do not interfere with the acquisition of EC, whereas the magnitude of the EC effect depended on the total number of contiguous CS–US presentations. Experiment 2 demonstrated that EC does not depend on a predictive relation between the CS and the US (contingency), either. Reliable EC effects were shown even with CSs that predicted the absence of a US (negative contingency). Throughout the experiments, EC was restricted to those contiguities that were memorized by participants. Subjective judgments of the contingency relation accounted for an additional indirect effect on EC. Taken together, the results suggest that EC depends on neither reinforcement density nor contingency, whereas propositions containing contingency information might exert some influence.