As we interact with the world around us, we are continually met by a number of distinct and salient smells: the waft of buttered popcorn that strikes us when we first enter the movie theater, the unpleasant smell of milk that has soured after being left out a bit too long, and the cup of freshly roasted coffee that sits at our desk, helping to keep us attentive during the day. Our sense of smell is undoubtedly advantageous from an evolutionary point of view. Being able to perceive and recognize the odors in our environment can provide useful information: Smells can help us to avoid a provide useful information: Smells can help us to avoid a burning building to our right, to consume a nutritious piece of fruit over a rotten one, and to help us select a fitting mate (see Candolin, 2003;Desor & Beauchamp, 1974;Rabin & Cain, 1984;Stevenson, 2009;Vlahos, 2006). y Interestingly, though, our ability to correctly identify substances just on the basis of orthonasal olfactory cues is actually quite poor. When odor cues are presented in isolation (i.e., divorced from any context), we are able f to correctly identify only roughly one third to one half of them (see Cain, 1979;Engen & Ross, 1973; Rabin & of them (see Cain, 1979;Engen & Ross, 1973
CHARLES SPENCE
University of Oxford, Oxford, EnglandIn the present study, we explored the conditions under which color-generated expectations influence participants' identification of flavored drinks. Four experiments were conducted in which the degree of discrepancy between the expected identity of a flavor (derived from the color of a drink) and the actual identity of the flavor (derived from orthonasal olfactory cues) was examined. Using a novel experimental approach that controlled for individual differences in color-flavor associations, we first measured the flavor expectations held by each individual and only then examined whether the same individual's identification responses were influenced by his or her own expectations. Under conditions of low discrepancy, the perceived disparity between the expected and the actual flavor identities was small. When a particular color-identified by participants as one that generated a strong flavor expectation-was added to these drinks (as compared with when no such color was added), a significantly greater proportion of identification responses were consistent with this expectation. This held true even when participants were explicitly told that color would be an uninformative cue and were given as much time as desired to complete the task. By contrast, under conditions of high discrepancy, adding the same colors to the drinks no longer had the same effect on participants' identification responses. Critically, there was a significant difference in the proportion of responses that were consistent with participants' color-based expectations in conditions of low as compared with high discrepancy, indicating that the degree of discrepancy between an individual's actual and expected experience can significantly affect the extent to which color...