Calling a Peach Scent Food or Body Lotion

Imagine smelling a peach-like fragrance after being told it is a food odor, or after being told it is the scent of body lotion. Both descriptions are plausible, but does the brain process the fragrance in the same way?

An fMRI study of 21 women found different responses in the anterior cingulate cortex depending on whether a label described the odor’s source or its use. Yet self-reported pleasantness and intensity did not differ clearly between labels. The study investigated the influence of words beyond situations where they change liking.

Comparing source and use, rather than good and bad names

Some odor names describe what a smell comes from, while others describe where or how it is used. The scent of rose, for example, can be understood as a flower odor or as a cosmetic fragrance.

The main aim was to test whether these two ways of categorizing odors change brain activity during smelling. The researchers also examined whether labeling effects differed between food-like and floral odors. The paper appeared online in 2012 and in a 2014 journal issue.

The same four fragrances, presented with two descriptions

Participants were 21 right-handed women with normal olfaction and a mean age of 24.4 years. The stimuli were formulated fragrances resembling peach, chocolate, linden blossom, and rose, of the kind used in body lotions. Participants were not smelling actual foods or flowers.

Each fragrance was presented under both label conditions, giving eight sessions per participant. In the source condition, peach and chocolate were verbally described as “food,” and linden and rose as “flower.” In the use condition, all four were described as “body lotion.” Labels were given before each session, and session order was randomized.

An apparatus carried the odors to the nose in temperature- and humidity-controlled air. Alternating 21-second odor-on and odor-off periods were repeated while fMRI recorded brain activity. Participants rated pleasantness and intensity after each session.

In a preliminary check, 10 people rated how well each label applied to the four odors. “Food” or “flower” and “body lotion” fit better than the less relevant comparison labels “spice” and “medicine,” with no significant difference in applicability between source and use labels. At least, the design did not pit a plainly implausible description against a plausible one.

Pleasantness and intensity ratings did not differ by label

In the main experiment, neither pleasantness nor intensity differed statistically clearly between source and use labeling. It would therefore be inaccurate to summarize this as an experiment in which calling an odor peach made it appetizing and calling it lotion made it unpleasant.

A separate check using the same presentation procedure outside the scanner measured sniff duration and inspired volume in 14 people. Neither differed significantly by label condition. The authors therefore considered it unlikely that changes in breathing alone explained the brain-activity differences described below.

A direct label comparison found higher anterior cingulate responses for source labels

The researchers first compared odor-on and odor-off periods within each label condition. Source labels were associated with responses in the medial orbitofrontal cortex, anterior cingulate cortex, and insula. Use labels were associated with a response in the medial orbitofrontal cortex.

A direct comparison between the two label conditions then showed a higher anterior cingulate response for source labels than for use labels. No region showing a higher response for use labels was identified in this analysis.

The type of comparison matters. For the insula, a difference from the odorless baseline was detected in one condition but not the other. This is not evidence of the same strength as the direct between-label difference reported for the anterior cingulate cortex.

Analysis was restricted to a prespecified set of brain regions associated with olfaction or linguistic influence. It used an uncorrected threshold of p<.001 together with a cluster-size criterion. These findings from a small, exploratory study provide a basis for further confirmation.

The researchers also examined food-like and floral response patterns

Considering peach and chocolate alone, source labels were associated with responses relative to the odorless baseline in the insula, anterior cingulate cortex, and medial orbitofrontal cortex. Under use labels, a response was found in the medial orbitofrontal cortex.

For linden and rose, source labels were associated with medial orbitofrontal and anterior cingulate responses, whereas no region met the specified criterion under use labels. The authors related the pattern of insular responses particularly under food labels to possible associations between food odors, tastes, and words formed through past experience.

However, the food-like and floral odors also differed in perceived intensity: the food-like odors were weaker. The comparison therefore did not isolate food-versus-flower meaning completely, nor did this experiment directly establish and test learned associations with taste.

Considering the context evoked by words through brain responses

The authors concluded that neural representations of odors depend on the categories supplied by words. Their interpretation was that describing what an odor comes from or where it is used can connect the same stimulus with different experiences and knowledge.

An additional analysis restricted to odor-off periods found no clear brain-activity difference between source and use conditions. The authors viewed this as supporting a contextual effect when an odor is present, rather than a response to words alone.

However, the source condition used two labels, “food” and “flower,” whereas the use condition used only “body lotion.” The authors themselves noted that this asymmetry could contribute to some findings. The sample was also limited to 21 young women and four fragrances. Brain-activity differences do not establish a change in hidden, true preferences or willingness to buy products.

Conclusion

For the same four fragrances, source descriptions such as “food” or “flower” produced a higher anterior cingulate response than the use description “body lotion.” Medial orbitofrontal responses occurred in both conditions, while food-like and floral subgroup analyses showed additional response patterns.

Nevertheless, pleasantness and intensity ratings did not differ clearly by label. Odor context may matter not only when liking scores change, but also in which knowledge and experiences become associated with the same smell. This study explored that possibility through brain imaging.

Coauthor Nicola Phillips was affiliated with Unilever at the time of the study. The samples were manufactured by Firmenich. No separate funding section or conflict-of-interest statement was found in the publisher’s original full text reviewed.

Reference

Bensafi M, Croy I, Phillips N, et al. The effect of verbal context on olfactory neural responses. Human Brain Mapping. 2014;35(3):810–818. doi:10.1002/hbm.22215