This research examines the contribution of the Theory of Relational Frames (TRF) to the study of attitude change of the consumer facing some brand. In this experiment, using a matching-to-sample procedure, subjects were trained to from three member equivalence classes (A1, B1, C1; A1, A2, A3; C1, C2, C3) using emotional icons as stimuli. The three stimuli of the class A1, A2, A3, was the name of a beer brand (the favorite brand, the neutral brand and the non favorite brand); The three stimuli of the class B1, B2, B3, were emotional icons (happy face, neutral face, and sad face), these stimuli were trained as conditioned stimuli and last three stimuli of the class C1, C2, C3, were nonsense graphics, that function as transfer stimuli. This study represented our attempt to investigate on attitude change to a brand without experience it directly. Results from the experiment indicated that subjects’ behavior was in accordance with the expected equivalence relations but the level of preference to a brand had a moderated change, because conditioned stimuli have not enough strong relation with the determinant variables of the preference to a brand. The new hypothesis is that conditional stimulus should be positive relational events with respect to the subjects’ learning history.