Observational conditioning of food valence in humans.

It has been suggested that the observation of a model consuming a food (CS) and facially expressing either to like or to dislike (US') the food, may be a sufficient condition to bring about a change in the valence of the food for the observer. Unfortunately, up to now this hypothesis has not been investigated in a straightforward manner. In this study, during acquisition, children consumed a series of evaluatively neutral colored and flavored drinks, while simultaneously they watched a videotaped model synchronically drinking identical drinks and facially expressing his evaluation (neutral or dislike) of the liquids. In one condition, the presence of a particular flavor in the drinks was designated to function as the CS+ or the CS-, whereas in the other condition it was the color of the drinks which was the critical CS+ or CS-. Next, the children evaluated a series of drinks containing the critical CSs. A clear evaluative learning effect was obtained when the flavor but not when the color of the drinks was systematically paired with the model's facial expression of dislike. Moreover, the flavor conditioning effect was dependent on the presence in the test drinks of the local context cues (c.q. the colors of the drinks) which were used during acquisition. Finally a double dissociation was observed between explicit beliefs and the "evaluative knowledge" expressed in the ratings of the drinks, in that none of the children in the CS = Flavor groups evidenced any explicit knowledge about the crucial CS-US' contingency but showed evaluative conditioning, whereas the majority of the children in CS = Color groups were aware of the CS-US' relation but failed to demonstrate an evaluative CS-/CS-differentiation.