Can Odor-Discrimination Practice Improve Performance with Untrained Blends?

Does practicing odor discrimination help only with the odors used in practice? A 2026 study found that training people to distinguish the mixing ratios of two compounds improved performance with other mixtures and with the untrained nostril. The improvement was still present two weeks later.

By contrast, practice in distinguishing mirror-image molecules produced effects limited to the trained stimuli and nostril, while improvements from practicing concentration discrimination for a single compound did not last as long. Rather than treating all smell training as one category, the study investigated how the spread and persistence of learning depend on what people practice distinguishing.

Distinguishing three bottles with just one nostril

Ninety-six healthy nonsmokers participated in the research as a whole. Of these, 48 underwent training; the other 48 were separate participants who rated the intensity, pleasantness, and odor similarity of the compounds. Training was divided into three experiments, with 24 participants in the first and 12 in each of the next two. Trainees were selected from people who had answered no more than five of nine trials correctly in the preliminary task.

The basic training task offered three choices. Blindfolded and with one nostril closed, participants sniffed three bottles in sequence. Two held the same sample and one held a different sample; participants selected the odd bottle. Chance performance was one correct answer in three.

They completed 12 trials a day and received feedback after each answer. Training continued until they achieved at least 10 correct answers out of 12 on two consecutive days. Across all trainees, this took an average of 20 days. The time required varied by task and person, ranging from seven to 38 days overall.

Tests without feedback were conducted before training, one hour after the final training session, and one, three, seven, and 14 days later. Both the trained and other stimuli were tested in the trained and opposite nostrils. The other stimuli were not used in daily training, but participants did smell them during the pre- and post-training tests.

Mixture-ratio training transferred to different ingredients

In the first experiment, 24 participants were randomly divided into two groups of 12. One group distinguished odors made by mixing the same two compounds in slightly different proportions. An example was a 7:9 mixture of guaiacol and eugenol compared with a 9:7 mixture. For some participants, the ratios were 9:11 and 11:9.

This group trained with either the guaiacol–eugenol pair or the 2-butanol–2-heptanol pair, while the other pair was used to assess transfer. The question was whether learning remained useful when the set of ingredients changed.

After training, participants improved not only with the mixtures they had practiced but also at distinguishing mixing ratios in the other pair. Improvement was also seen in the opposite nostril. Immediately after training, mean accuracy pooled across nostrils was 70% for trained mixtures and 48% for untrained mixtures. Both exceeded the chance level of 33%, but improvement with different ingredients was smaller than with the practiced ingredients.

The two sets differed in both molecular structure and odor quality. The authors interpreted this as learning to distinguish subtle differences between mixed odors that extended to other ingredients, rather than simply memorizing the smell of one bottle.

With mirror-image molecules, improvement was more restricted

The other 12 participants practiced distinguishing enantiomers: molecules that are mirror images, like a right and left hand. They trained with either a pair of α-pinene enantiomers or a pair of 2-butanol enantiomers.

Here too, performance improved for the trained stimuli presented to the trained nostril. However, no clear improvement was confirmed for the other enantiomer pair or the opposite nostril. The reach of the effect differed from that in the mixture-ratio group.

There was a similarity in persistence. Mixture learning, including transfer to other mixtures and the opposite nostril, remained two weeks later. Enantiomer learning also remained above chance two weeks later, but only for the trained stimuli and nostril. Learning with a narrow scope does not necessarily disappear quickly.

Testing mixtures with greater chemical differences

The molecules in the initial mixtures shared a chemical feature. To examine whether that could explain the transfer of improvement, the next 12 participants used mixtures of acetophenone and 2-octanone, and mixtures of methyl salicylate and isoamyl butyrate. The first pair consists of ketones and the second of esters, two different chemical classes.

In this experiment too, practicing one mixture ratio improved performance with the other mixture and the opposite nostril, and the improvement persisted for two weeks. Immediately after training, mean accuracy for the untrained mixtures was 63%, higher than the 48% in the first experiment.

The authors suggested that structural diversity in training materials might help learning generalize. However, the two experiments used different participants and materials, so these numbers alone do not determine which ingredients would always provide the best training.

Had participants simply learned differences in concentration?

Changing a mixture ratio also changes the concentration of each ingredient. In the third experiment, another 12 participants therefore practiced distinguishing samples containing a single compound at 0.44% and 0.56%. These are concentrations in the bottled solutions, not in the air reaching the nose.

Concentration discrimination also improved for the trained stimuli and nostril, but did not clearly transfer to another compound or the opposite nostril. Performance declined over time after training and no longer significantly exceeded chance two weeks later.

The study also measured adaptation: perceiving a repeatedly smelled stimulus as weaker. In every training task, the trained odor was perceived as weaker through the trained nostril, but this difference diminished over the following days. Mixture and enantiomer discrimination persisted afterward, whereas concentration discrimination declined roughly in parallel with recovery from adaptation.

From this difference, the authors concluded that mixture learning was difficult to explain solely through concentration cues or temporary adaptation. Some participants in the concentration task reported learning strategies that related sniffing order to perceived intensity, suggesting that the cues learned may differ by task.

Where does olfactory learning take place?

The authors used whether effects transferred between nostrils as a clue to the stage of olfactory processing that might have changed. They suggested that mixture learning may involve relatively higher-level processing that integrates information from both sides and judges odor features. They named regions such as the orbitofrontal cortex as candidates, but did not measure brain activity in this study.

They also suggested that active discrimination training with mixtures might prove useful in olfactory rehabilitation or the development of expert skills. The participants here were healthy young adults, with 12 people per group and two weeks of follow-up. Recovery in patients and practical perfumery performance remain to be tested. There was also no direct comparison with passive smelling. The central result is that even with the same three-choice training format, the pattern of improvement differed by task.

Summary

Practice in distinguishing the mixing ratios of two compounds improved performance with untrained mixtures and the opposite nostril, and improvement remained two weeks later. Enantiomer learning persisted but was confined to the trained stimuli and nostril. Learning to distinguish concentrations of a single compound no longer clearly exceeded chance two weeks later.

Olfactory learning can involve becoming more proficient with a particular target, or acquiring a skill that also helps with other targets. This study shows that what people practice distinguishing is an important factor in that difference.

The research was supported by China’s STI2030 Major Projects, the National Natural Science Foundation of China, and the Fundamental Research Funds for the Central Universities. The authors declared no competing interests. This article was prepared by reading the accepted author manuscript made available by the publisher. The version of record was published on February 13, 2026.

Reference

Chang, X., Tan, H., Niu, J., Yuan, K., Chen, R., & Zhou, W. (2026) Mixture discrimination training induces durable and generalizable olfactory learning independent of odorant structure and concentration. eLife, 15, e102999. Original paper / Accepted author manuscript consulted