Does US expectancy mediate the additive effects of CS-US pairings on contingency instructions?

Verbal instructions are a powerful pathway to learn new fear relations, and an important question has been what fear experience can still add to the effect of such instructions. Therefore, in previous studies, we investigated the effects of pairings between conditioned stimuli (CS) and unconditioned stimuli (US) after CS-US contingency instructions. Although these studies found that CS-US pairings do indeed add to the effects of contingency instructions on subjective, psychophysiological and neural measures of conditioned fear, they also produce increases in US expectancy ratings. In the current report we address whether these enhanced US expectancy ratings can account for the additive effects of CS-US pairings as suggested by expectancy models of fear conditioning. To address this question we made use of pathway models to investigate mediation in within-subjects designs. Our results demonstrate that US expectancy ratings do not mediate the effects of CS-US pairings on fear ratings, the startle reflex or amygdala activation pattern similarity. Additional exploratory analyses, however, revealed that subjective fear ratings do explain the effects of CS-US pairings on the other measures. We discuss how these results relate to expectancy models of fear conditioning and what they implicate for the validity of US expectancy and fear ratings.