The content of learning in human evaluative conditioning: Acquired valence is sensitive to US-revaluation

Abstract

Evaluative conditioning refers to the observation that the mere contingent presentation of neutral with (dis)liked stimuli changes the valence of the originally neutral stimuli in a (negative) positive direction. Two theoretical accounts of the representational structure built up during evaluative conditioning are contrasted: intrinsic change versus referential learning. It is argued that previous findings on evaluative conditioning (for example, resistance to extinction) seem to favor intrinsic change explanations, but do not actually allow any definite conclusions. The postconditioning US-revaluation paradigm was used to obtain a more straight-forward result. In Experiment 1, a technique was developed to alter the valence of originally strongly (dis)liked USs (pictures of faces). In Experiment 2, we first presented a contingency between neutral (CS) and (dis)liked pictures of faces (US), and next applied the US revaluation technique. Contrary to expectations based on intrinsic change accounts, we observed that the postconditioning US revaluation did affect the acquired value of the CS; this result was confirmed in a 1-month follow up. Hence, we conclude that the acquired evaluative meaning of the CS is referential, ultimately relying on an association between CS and US representations.