In four experiments using a human evaluative flavor–flavor association paradigm, we investigated whether the acquisition of stimulus valence may be subject to Pavlovian modulation or occasion setting. In all studies, subjects (Ss) were exposed to a feature-positive (FP) (XA+/A−) schedule (or a modified version of it), in which flavor A was reinforced by the aversive flavor Tween 20 if and only if it was accompanied by a feature stimulus X. It was expected that Ss would develop an X-dependent dislike for A. In Experiment 1, for half of the Ss a quasi-sequential FP schedule was used, whereby the color of the liquids functioned as X. The other half of the Ss were exposed to a positive-patterning (PP) (XA+/A−/X−) schedule, in which unreinforced color (X−) trials were added to the FP schedule. Both schedules resulted in an unmodulated dislike for flavor A. In Experiment 2, Ss were exposed to a modified FP schedule, in which flavor A was reinforced by Tween 20 when presented with color X, while A was reinforced with positive sugar when presented colorless. Ss again demonstrated an unmodulated dislike for A. Experiment 3, using a sequential FP (X → A+/A−) schedule in which a flavor functioned as X, also resulted in an unmodulated dislike for A. Finally, in Experiment 4, Ss were exposed to the modified FP schedule, in which flavor A was presented with Tween 20 when preceded by flavor X, while A was reinforced with Sugar when not preceded by X. For those Ss liking sugar, no valence shifts were observed, whereas for those not liking sugar, the standard unmodified dislike for A was obtained. Some suggestions are provided with respect to variables which might be related to the absence of modulation.