Does mere co-occurrence affect evaluation? MPT modelling of relational evaluative conditioning cannot (yet) tell

In relational EC studies, the reversed EC effect induced by contrastive CS-US relations is typically smaller than the regular EC effect induced by assimilative CS-US relations. Dual-process accounts of EC attribute this attenuated reversed EC effect to an effect of CS-US associations that counteract the evaluative implications of CS-US propositions in the contrastive condition. This interpretation is corroborated by two recent studies analyzing evaluative CS classifications by means of a multinomial processing tree model that decomposes evaluations into propositional processes and associative effects of US valence. We argue that these findings are also compatible with a single-process propositional perspective. According to this view, contrastive CS-US relations induce less extreme CS evaluations, and therefore have a lower probability of yielding an evaluative inference during the evaluative classification task. In three simulation studies, we demonstrate that this simple set of assumptions cannot only explain the US valence parameter itself, but also accounts for its correlations with evaluative CS ratings as well as for the somewhat contradictory results of previous parameter validation studies.