Reverse Evaluative Learning: Paradoxical Contamination of Liked/Disliked Stimuli by Neutral Stimuli

According to De Houwer et al. (1997) the evaluative learning (EL) paradigm deserves the attention of the implicit research community for various reasons, such as its resistance to the general criticism addressed at implicit learning studies on grounds of often failing to meet information and sensitivity criteria, and the strong evidence for EL to be implicit. Nevertheless, EL is controversial because (1) it is unclear what the sufficient and necessary conditions for EL to take place are, and (2) the extent to which EL can take place without awareness of the relevant contingencies is likewise unclear. The present study aimed at conceptually replicating Baeyens et al.'s (e.g., 1988) EL studies to explore the robustness of the EL phenomenon. In a first phase subjects rated the affective value of various faces. During the subsequent acquisition phase, they were shown different contingent pairs of neutral stimuli (NSi) and affective stimuli (ASi), (i.e. liked, disliked or neutral). Next, they were to rate the whole set of faces again, based on their current first impression. Results failed to replicate typical EL findings: Not only did we fail to observe changes in the evaluations of NSi, but ASi, as well as stimuli that had not been presented during the acquisition phase, were evaluated to be more neutral after the acquisition phase than before. We discuss these results in the context of Field et al.'s (1999) alternative accounts of EL phenomena.