Someone smells an apple and answers “fruit.” Even if that is incorrect on a test where “apple” is the right answer, it seems unlikely to mean that the person knows nothing about the smell. What information about an odor remains in the words used when its name cannot be recalled correctly?
A study published in 2024 by Thomas Hörberg and colleagues analyzed responses from 2,479 adults in Sweden. Rather than looking only at the proportion of correct answers, the researchers examined the meanings of words used in incorrect responses and how frequently those words are used in everyday language, exploring the connection between odor perception and language.
First, Describe 16 Smells in Your Own Words
The participants, aged approximately 58–100, were taking part in a study following community residents in Stockholm. Of the 2,569 people who completed the odor tasks, 90 were excluded under the study’s criteria, including conditions such as dementia and Parkinson’s disease. The mean age was approximately 72 years.
The study used 16 everyday smells, including apple, banana, coffee, leather, rose, and peppermint. Participants were first asked to say freely what each odor was, without being shown response options. They could give several answers if they could not settle on one.
If the name was incorrect or the participant could not answer, they were asked to choose from four options that included the correct answer. The overall identification score was the number of odors correctly identified either in free responses or in the multiple-choice task. It was not simply the number they could name unaided.
The detailed analysis in this study focused on incorrect answers from the initial free-response task. These were distinguished from omissions, and the analysis centered on single-word responses.
The “Wrong Names” Were Not Random
The researchers used large collections of Swedish blog posts and other texts to quantify how each word was used: whether it was common, whether it tended to appear in discussions of smell, and whether it could describe a broad range of odors. They also examined how close its meaning was to that of the correct name.
For example, “apple” and “pear” are used in similar contexts involving eating and fruit. Semantic closeness between words was estimated from such patterns of use in text. It was not a direct measurement of odorants’ molecular structures or of how similar the odors felt to the nose.
Frequent incorrect responses often used words that were common in everyday language, closely associated with smell or taste, or close in meaning to the correct answer. Broad expressions not limited to a specific object were also common.
For 14 of the 16 odors, the three most frequent incorrect responses included broad category names such as fruit, flower, and spice, or evaluative words expressing good or bad qualities. Even without being prompted to give such answers, people chose somewhat broader descriptions when they could not produce the precise name.
Incorrect Answers Differ in How Close They Are to the Correct One
Furthermore, people with better overall identification scores tended, even when wrong, to use words referring to concrete objects or words semantically closer to the correct answer. They used words that narrowed down the object rather than general terms applicable to a wide range of things.
Answering “pear” for an apple and naming something entirely different may preserve different amounts of information, even though neither reaches the correct answer. At the same time, answers such as “fruit” or “sweet” may also convey some characteristics of the odor.
However, the study demonstrated associations between response characteristics and performance. It did not test a training effect, such as whether making people use more concrete words improves their sense of smell.
Words Used in Incorrect Answers Can Also Reveal How Odors Are Understood
When the words were analyzed together, pleasantness, associations with food, and whether a word refers to a concrete object emerged as major dimensions organizing the range of incorrect responses. This structure resembles that previously found in studies of English odor vocabulary.
Incorrect answers are therefore not merely noise; they can also provide information about the characteristics through which people understand odors. However, some word evaluations used existing data for English translations of Swedish terms, rather than ratings collected directly from the same participants for everything.
The study’s focus on older Swedish speakers and its limited set of 16 odors also requires attention. Descriptions of fragrance vary with culture, experience, and the purpose at hand. This organization of words cannot be treated as the only classification shared by all age groups and languages.
Summary
Failing to name a smell correctly does not necessarily mean that knowledge about it has been lost. Naming something close to the answer, giving a category, or describing a characteristic can each retain information. Even among incorrect responses, there were differences in how closely people had identified the object.
The authors suggest that evaluating “how close the answer is,” rather than only whether it is right or wrong, could provide a more detailed picture of the ability to understand odors. Linguistic properties of incorrect responses may offer clues that complement existing olfactory assessments, but this was not a study showing that an individual’s disease or health status can be determined from such wording alone.
References
Hörberg, T., Kurfalı, M., Larsson, M., Laukka, E. J., Herman, P., & Olofsson, J. K. (2024). A Rose by Another Name? Odor Misnaming is Associated with Linguistic Properties. Cognitive Science, 48(10), e70003. DOI: 10.1111/cogs.70003 (Full original article).
The research was supported by the Swedish Research Council, the Knut and Alice Wallenberg Foundation, and other funders. The authors declare no commercial or financial relationships that could be construed as a conflict of interest. Original article: © 2024 The Author(s), Creative Commons Attribution-NonCommercial (CC BY-NC).
