Expected taste intensity affects response to sweet drinks in primary taste cortex - Woods, Lloyd, Künzel et. al. (2011) - Article

Recent studies have shown that expectations about a food or drink’s taste can influence how it is subsequently perceived. The questions is whether the effects of expectation represent a change in taste perception or arise at a later stage of processing, via response bias. In this study, the researchers used MRI to look into these two accounts. The writers of this article hypothesized that if expectation effects are indeed perceptual in origin, they must influence activation within the taste cortex in a top-down manner. If there is no influence of expectancy on the primary taste cortex, then the response bias account is supported.

Scientists know that the primary taste cortex lies within the insula, but the exact location of it is unclear. Many studies have found taste activity in the anterior, anterior-dorsal, anterior-mid, middle and mid-posterior insula. The researchers used a meta-analysis to identify regions responding to taste (eleven studies) to help localize the primary taste cortex in this study. There are only two studies that have examined the effect of expectation on taste cortex. One study found that activation in the secondary taste cortex (area associated with food liking) was influenced by how a certain drink was hedonically labelled. A ‘rich and delicious taste’ label enhanced activation in the medial orbitofrontal cortex, whereas a ‘boiled vegetable water taste’ label did not. In the other study, a similar modulation of activation was observed in the middle and posterior insula after providing a hedonic label. In that last study, participants tasted either a pleasant (sweet) or intense or mild aversive (bitter) stimuli and visual symbols provided cues for the upcoming taste. What the writers of that study found, was that activation to the same intensely bitter stimuli was less when it was cued with a ‘mildly aversive’ expectation rather than a ‘highly aversive’ expectation. The middle-posterior insula region is associated with disgust and it is therefore possible that this represents an expectation of disgust, instead of taste. The researchers of the current article therefore want to know if perceptual expectations also impact upon primary taste cortex activity. In the current study, they tested if taste expectation (sweetness) influences food taste properties rather response bias, within a region that was defined as the primary taste cortex by the meta-analysis. If the response bias was responsible for the effect, then the writers would predict no effect on primary taste cortex.

Method

There were nineteen participants for this research. All participants didn’t smoke and reported no food allergies or cold symptoms. The participants received a drink and had to indicate on five-point scales (from not sweet to very sweet and from neutral to very pleasant) what they thought about the drink. The drinks they received were either 100% commercially available orange juice or a 50% orange juice and 50% water mix. The reason orange juice was used, was because it is recognizable and therefore more likely to promote effects of expectation. Also, there was a neutral tasting artificial saliva mouthwash. Participants tasted 0.75 ml of stimulus and 0.75 ml of mouthwash per trial and there were 40 trials per participant.

On each trial a text cue was presented to the participant to prompt a taste expectation. The text cue was either Very Sweet or Less Sweet. After two seconds, the drink (100% juice or 50% juice) was delivered. On half of the trials, the text cue and taste stimulus were congruent (so when presenting Very Sweet, the 100% juice was delivered and when presenting Less Sweet, the 50% juice was delivered) and on the other half of the trials, the text cue and taste stimulus were incongruent (Very Sweet with 50% juice and Less Sweet 100%). The participants were told that they should move the drink once around their mouth and then to hold their tongue still. When a green cross appeared on a screen, they were allowed to swallow the drink. The participants rated the stimulus for sweetness and liking on the scales. Afterwards, a mouthwash was delivered using the same protocols as for the drink stimulus, except there was no visual text cue. The stimulus order was randomized. MRI data were collected using a MR scanner and this was done for the purpose visualisation of brain activity.

Results

The results showed that the 100% drink was reported sweeter than the 50% drink and that the Very Sweet cue led to drinks being reported sweeter than the Less Sweet cue. However, posthoc tests revealed that Very Sweet (100%) and Less Sweet (50%) drinks did not differ, whereas the sweetness reported in the Very Sweet-50% condition was significantly greater than that in the Less Sweet-100% condition. This means that expectation enhanced the sweetness of the 50% orange drinks, but it had no impact on the 100% drinks. The researchers found that there was an activation in the primary taste cortex and this was in support for the hypothesis that the Very Sweet cue enhances activation above that of the Less Sweet cue for 50% drinks. There was also significantly more activation to Very Sweet cues than to Less Sweet cues and this implies that the expectation of tasting a Very Sweet drink acted to enhance the sweetness of the drink, relative to the expectation of drinking a Less Sweet drink. The data also showed that there was an activation in orbitofrontal cortex, but no activation in medial orbitofrontal cortex.

Discussion

The writers of this article found, with the help of MRI, that the presentation of taste labels (Very Sweet and Less Sweet) influenced activity within the primary taste cortex. This supports the previous study discussed in the beginning in the text, which found that behavioural observations of the apparent perceptual effects of expectations represent underlying changes in food perception, rather than being driven by response bias. The writers found that a drink (50% juice) labelled as Very Sweet led to it being reported to taste sweeter and produced more insula activation than the same drink with a Less Sweet label. The observed insula activation overlapped with that of a meta-analysis of 11 recent studies from the literature. The visual text cues may not have prompted an expectation of sweet or Less Sweet orange juice, but rather a general expectation of Very Sweet and Less Sweet drinks. It kind of seems that expectation would affect congruent and not just incongruent trials.

The activation found in this study was more inferior than that found in the meta-analysis. The writers think that different stimuli might activate different areas of primary taste cortex. The activation in the meta-analysis highlights only activation common across all studies. The findings of this study do not tell us a lot about top-down mechanisms by which primary taste cortex activity is influenced by expectation. The writers did observe activation to taste expectation outside the insula, like the orbitofrontal cortex. Past research has shown that orbitofrontal cortex activity may reflect a top-down modulation of activation through expectations of drink liking. The activation region of this study does not overlap with the medial orbitofrontal cortex activation of that mentioned in the study in the beginning of the text and that study tested explicitly for top-down hedonic effects on taste, but other studies have reported the involvement of lateral and posterior orbitofrontal cortex in reward.

The expectation of perceived intensity can influence neural activation to sweet stimulus within primary taste cortex in the 50% drinks. The behavioural and neural responses to the 100% drink was not affected. The writers think that this happened, because this resulted from too large a discrepancy between what the participants expected and what they tasted. Their findings show that expectation does indeed modulate basic taste perception.

Limitations

Just like most research, this research also has some limitations. This model was formalized using empirical data and for the purpose of generalization it should be tested in other settings, like industries. Also, these findings might not be directly generalized to adult markets (because children were used). There is no high risk of using free Internet applications, because the children don’t lose anything if they don’t like the applications. The children don’t need to evaluate the quality of the product extensively and thus the interpersonal influence might not be very strong in this application. The writers think that their results may be even stronger in different markets in which it is more difficult to judge product quality, because the influence of opinion leaders might increase. Further research should focus on other markets. The writers also suggest that future research should explore the relationship between the innovativeness of a consumer and the importance of the normative influence for this consumer. This study used a simple network structure and network structures are important, because they have a big influence on the results of the model. Future research should use other network structures.