When we want to remember a fragrance, we sometimes put our impressions into words. But when we encounter the same odor again later, does that description always help?
Yaxiong Cao and colleagues presented a study at an international cognitive science conference in 2024 examining how describing an odor in detail affects recognition: selecting that odor again. They also tested whether the results changed when people answered quickly or were warned not to rely on their verbal memory.
Finding a previously smelled leather odor among eight choices
The participants were 148 university students in China. Eight who did not meet the criteria in a preliminary smell test were excluded, leaving 140 randomly assigned to seven groups of 20.
Everyone smelled a leather odor from an odor-dispensing pen for 5 seconds and tried to remember it. They then spent 20 minutes reading English passages unrelated to odors and answering questions. Some groups next spent 5 minutes writing as detailed a description as possible of the odor they remembered; control groups continued the English task for another 5 minutes.
Finally, participants smelled eight odors, including leather, and selected the one matching the original. The other seven were raspberry, melon, smoked meat, coconut, cinnamon, licorice, and rose. This was not a naming task: they chose among numbered alternatives. They could also answer that the target odor was absent.
The main performance measure was a score from 1 to 6 combining correctness with confidence. More confident errors received lower scores, and more confident correct answers received higher scores. The score therefore cannot simply be read as a percentage correct.
With time to choose, the description group scored lower
Under self-paced conditions, with re-sniffing allowed if needed, the group that had not described the odor averaged 5.70 points, compared with 4.65 in the description group. The difference was statistically significant. At least in this task, a detailed verbal description did not help participants find the same odor later.
In contrast, when participants could smell each odor only once for about 2 seconds and had to answer immediately after all eight, both the description and control groups averaged 3.50 points. There was no difference between the groups attributable to description.
The authors suggest that rapid judgments may reduce verbal interference. However, scores themselves did not improve in the fast condition. The result was not that rushing improved memory, but that the difference between groups disappeared at a lower level of performance. Because permission to re-sniff also changed, this was not a comparison isolating decision speed alone.
Is consciously avoiding verbal memory enough?
The study also included a group instructed to ignore its memory of the written description and rely on odor memory. Another group selected two answers, one based on odor memory and one based on its description. There were still eight alternatives, and the first answer was used to assess performance.
The description groups given these instructions scored lower than the ordinary self-paced control group. The authors interpreted this as evidence that simply being aware of the distinction between verbal and odor memories does not remove interference. However, comparing the two-answer conditions with one another revealed no significant difference between description and no description. Further tests with matched comparison conditions are needed before concluding that such warnings are invariably ineffective.
From the whole odor to verbal features
Why might a description get in the way? The authors emphasize a hypothesis about a change in matching strategy. Participants may initially search using their overall impression of an odor, but after writing a description, start eliminating candidates according to whether they fit the features they selected in words. Cues that could not be verbalized might then be used less.
Indeed, none of the participants who wrote descriptions correctly identified leather as the source. Nevertheless, the self-paced group that did not describe the odor achieved 90% correct recognition. Being unable to name an odor and being unable to recognize it were different problems.
Four measures of the descriptions, word count, source-identification accuracy, elaboration, and variety of expression, were not significantly associated with recognition scores. Participants also borrowed expressions from other senses, including taste, color, and touch. What matters may be less how much detail people write than which aspects of the odor their words capture or miss. The experiment did not, however, directly measure this mechanism.
Conclusion
In this study, the group that described a leather odor in detail performed worse when recognition was self-paced. Words can preserve an experience, but may also change the cues used to remember it later. The authors’ findings raise the possibility that odor memory and verbal memory do not simply add together.
However, participants memorized only one odor, and each group consisted of 20 university students. The study did not examine repeatedly testing perfumes while taking notes or experts using a shared vocabulary. The disappearance of the group difference under speeded conditions, together with the findings on warnings, makes identifying the conditions in which words help or interfere an important next step.
References
Cao, Y., Majid, A., & Vanek, N. (2024). Verbal overshadowing in odor recognition. Proceedings of the 46th Annual Conference of the Cognitive Science Society, 641–649. Article information / Full original paper. © 2024 The Author(s). CC BY 4.0.
