Sensory preconditioning of evaluation requires accurate memory of the co-occurrence between the neutral stimuli

Neutral conditioned stimuli (CS) acquire the valence of valent unconditioned stimuli (US) after the stimuli are paired. This effect has often been attributed to a low-level mechanism that occurs automatically, even without awareness of the CS-US pairing. However, recent research has failed to support this assumption. In three experiments (N = 1541), we tested the possibility of attitude formation without awareness of pairing with a US using a sensory preconditioning paradigm. Previous research with that paradigm found attitude formation with no retrospective awareness of contingencies. However, retrospectively reported unawareness might be due to forgetting. Another weakness of retrospective measures is that people might show accurate memory of the CS-US pairing based on their evaluation of the CS. Our procedure circumvented these weaknesses. Participants observed co-occurrences between neutral nonwords and neutral faces, followed by a nonword-face co-occurrence memory test. After that test, participants observed co-occurrences between the faces and affective stimuli, indirectly pairing the nonwords with the affective stimuli. At that time, if participants previously failed to remember with which face a certain nonword co-occurred, then they were unaware of the indirect pairing of that nonword with affective stimuli. We found that indirect pairing influenced the evaluation of the nonwords only when memory for the nonword-face contingency was accurate. Contrary to past research, accurate memory in the present research cannot be explained by an effect of the change in evaluation on memory. Our results suggest that contingency awareness is necessary for the effect of stimulus pairing on evaluation