Flavor evaluative conditioning and contingency awareness

SUSAN G. WARDLE 0 1 CHRIS J. MITCHELL 0 1 PETER F. LOVIBOND 0 1 0 University of New South Wales , Sydney , New South Wales, Australia 1 (Manuscript received March 23, 2007; revision accepted for publication July 25 , 2007.) The relationship between flavor evaluative conditioning and contingency awareness was examined in two experiments using flavored drinks. In Experiment 1, one flavor was always paired with sugar and the other with bitter tween ( polysorbate20) during conditioning. In a subsequent test phase, participants tasted the two flavors, and their evaluative ratings indicated an overall preference for the sugar-paired flavor. Moreover, participants were generally able to report which flavor had been paired with sugar and which with tween. This finding was replicated and confirmed in Experiment 2A. Furthermore, in both experiments, evaluative conditioning was seen only in those participants who were aware of the contingencies. Experiment 2B demonstrated that evaluative conditioning does not occur to colors, although participants are contingency aware. The differences between the present findings and prior studies, in which apparently unaware flavor conditioning has been found, are discussed. - ness use insensitive measures of participants contingency knowledge. However, Lovibond and Shanks identified experiments which use flavors as the conditioned stimuli as providing some suggestive evidence for unaware evaluative conditioning. Flavor conditioning is a variant of the evaluative conditioning procedure that has received surprisingly little attention. In the original demonstration (Zellner, Rozin, Aron, & Kulish, 1983), participants drank a series of flavored teas. During conditioning, one of the tea flavors (CSpos) was always mixed with sugar, and the other was presented unsweetened (CSneg). Participants then preferred the tea flavor that had been paired with sugar during conditioning, even when it was presented in the absence of the US (i.e., unsweetened) on test. Baeyens, Eelen, Van den Bergh, and Crombez (1990) were the first to explore the relationship between conditioning and awareness with flavors. The CSs were either fruit flavors or colors. Sugar was used as the positive US and tween (polysorbate 20, which has an unpleasant bitter taste) was the negative US. Baeyens, Eelen, et al. reported an interesting dissociation between conditioning and awareness. Participants exhibited flavorflavor evaluative conditioning and rated the flavor that had been paired with sugar as more liked than the flavor that had been paired with tween. However, they showed no awareness of the CSUS contingencies. In contrast, participants did not show any evidence of colorflavor evaluative conditioning, yet they could reliably identify which colors had been paired with sugar and which with tween earlier in the experiment. The demonstration of unaware flavor conditioning has some weight because the awareness test was sensitive enough to detect awareness of the colorflavor contingencies. Evaluative conditioning occurred with flavor CSs, yet participants could not accurately report the stimulus contingencies. In the Baeyens, Eelen, et al.s (1990) measure of awareness, participants were explicitly informed in the test phase that some of them had received sugar in their drinks, others tween, and others just the fruit flavors. On a response sheet, participants were required to indicate what flavor (or color) of CS the sugar (or tween) US was presented with most of the time, and how certain they were of their answer. The flavor and color stimuli were placed in front of participants, and they were permitted to smell and taste the test stimuli as many times as they wished before answering the questions. The names of the test stimuli were written underneath each cup; once participants had made their selections on the basis of the sensory characteristics of the stimulus, they confirmed the name of the CS by consulting the label in order to answer the questions. A weakness of this awareness measure is that the awareness test procedure was not identical to the test of evaluative conditioning. During the evaluative test, and also in the conditioning phase, participants tasted the drinks in a precise timed sequence, whereas in the awareness test they could taste each drink as little or as often as they wished. Lovibond and Shanks (2002) stressed that the conditions experienced in the learning and testing phases should be as similar as possible in order to increase the sensitivity of the test. One might, of course, argue that Baeyens, Eelen, et al.s (1990) test would be more sensitive because multiple tastings were allowed. However, the possibility that this test procedure might have confused participants is enough to cast doubt on the claim that the participants were appropriately classified as aware and unaware in that experiment; what is needed is a contingency awareness test as similar as possible to the test of evaluative conditioning. An additional concern with the Baeyens, Eelen, et al. (1990) measure is that the order in which the evaluative and awareness tests were completed was not counterbalanced. As the awareness test was always completed second, it is possible that the evaluative ratings were affecting the awareness measure or that the lack of knowledge of flavor CS contingencies was simply due to forgetting. However, in spite of these shortcomings the test was still sensitive enough to detect participants knowledge of the color contingencies, and this is the main strength of Baeyens, Eelen, et al.s results. Dickinson and Brown (2007) have provided the only replication of unaware flavor conditioning that we are aware of; but see Stevenson, Boakes, and Prescott (1998) for related studies. Dickinson and Browns focus was not contingency awareness but blocking of flavor evaluative conditioning by colortaste pairings, for which they found no evidence. Nevertheless, their evidence for unaware flavor conditioning is the best in the literature. Dickinson and Brown used a somewhat improved experimental design partially based on parametric experiments by Baeyens, Crombez, Hendrickx, and Eelen (1995) and suggestions by Lovibond and Shanks (2002) on what constitutes an appropriate and sensitive measure of contingency awareness. Dickinson and Browns awareness measure was superior to that of Baeyens, Eelen, et al. (1990), in that Dickinson and Brown used a timed tasting procedure identical to that used in conditioning, and the same continuous rating scales for both the evaluative and the contingency tests. For each US (sugar and tween), participants were asked to rate their level of certainty that it had been presented with each flavor. However, as in Baeyens, Eelen, et al.s (1990) study, the order in which the evaluative and contingency tests were completed was not counterbalanced. In agreement with Baeyens, Eelen, et al. (1990), Dickinson and Brown reported evaluative conditioning with (...truncated)