Does Deodorant Make Body-Odor Similarity Harder to Detect?

Do long-term partners even smell alike? In a study in which third parties smelled samples from 30 couples, body odors collected on a day without deodorant were rated as more similar within actual couples than within recombined male–female pairs. This difference was not found in samples collected while donors used their usual products.

However, this result came from an analysis treating the 413 raters as the units of analysis. In an analysis using mean ratings for each couple, the overall difference was not statistically clear. This 2019 study explored odor similarity alongside partner choice, fragranced products, hormonal contraception, and relationship satisfaction.

Comparing 30 real couples with recombined pairs

The odor donors were 30 heterosexual couples of European ancestry recruited in the UK, totaling 60 people. They were aged 20–51, had been together for at least six months, and the women were premenopausal. The researchers deliberately recruited 15 couples in which the woman had used hormonal contraception when the relationship began and 15 in which she had not. Methods included oral contraceptives, implants, and injections.

Samples were collected on two consecutive days: first without fragranced products, then with the donor’s usual deodorant or antiperspirant. Cotton pads were worn under the arms for 24 hours each day and frozen after collection. Donors were asked to use unscented soap, restrict strong-smelling foods, alcohol, exercise, and other activities around sampling, and not share a bed during collection.

Of 437 raters who participated, 413 completed the task and were included in the analysis. Each smelled six pairs of samples: the two types of real couple and fake pairs combining a man and a woman from different couples, each with and without products. Presentation order was randomized, and raters were not told the donors’ sex or that the samples came from couples. They rated how similar the two odors were on a nine-point scale.

Higher similarity without products, but no group differences with products

The researchers originally predicted that actual partners would smell less similar than fake couples. The rater-level analysis showed the opposite pattern.

Without products, both groups of real couples were rated more similar than fake pairs, with the highest similarity among couples in which the woman had not used hormonal contraception when the relationship began. Couples with such use fell between that group and fake pairs. With the donors’ usual deodorants or other products, the three groups did not differ significantly. The interaction between couple type and product use was also significant at the rater level.

The authors discussed the possibility that fragranced products change judgments of body-odor similarity. However, these products may introduce antibacterial or antiperspirant ingredients as well as fragrance. The design cannot distinguish masking from changes in odor production. Collection order was also fixed for everyone: no products first, then products.

Averaging by couple calls for a more cautious conclusion

Although 413 people provided ratings, only 30 real couples donated odors. Multiple people rating the same odors is different from demonstrating a result across many different couples.

The paper therefore also analyzed mean ratings for each of the 30 real couples and 15 fake pairs. The means followed a similar pattern, but neither the interaction between product use and couple type (p = .435) nor the comparison between real and fake couples (p = .380) was significant. Some narrower comparisons were significant, but substantial uncertainty remains about whether the overall pattern would generalize to another population of couples.

The researchers also examined whether age differences, cohabitation, or duration of cohabitation could explain the results, but found no clear association with unfragranced odor similarity. This does not eliminate all influences of a shared environment, such as diet. Because odors were not collected before the relationships began, the study cannot directly determine whether people chose similar-smelling partners or became more similar after getting together.

Hormonal contraception and satisfaction: exploratory associations

Hormonal contraception at relationship formation was another central question. In individual rater-level comparisons of unfragranced samples, couples without use at that time were rated more similar than couples with use. However, this difference was also nonsignificant in the overall couple-level analysis.

A further exploratory analysis was restricted to unfragranced real-couple samples from the 28 couples with complete questionnaires. After adjustment for relationship length and both partners’ satisfaction, similarity was higher in couples without hormonal-contraceptive use at relationship formation. This was an observational comparison of past use, not randomized assignment of contraception or follow-up before and after starting it. Some women’s use had changed by the time of sampling.

For relationship satisfaction, after statistically adjusting for the other partner’s satisfaction, greater odor similarity was associated with higher satisfaction in men but lower satisfaction in women. Partial correlation coefficients were 0.499 for men and −0.462 for women. These are associations in a small sample, not evidence that odor causes changes in relationship quality or a scale for assessing individual couples’ compatibility.

Which odor qualities contributed to judgments of similarity?

The study also investigated what a simple similarity score might capture. The same body-odor samples had separately been described in specialist vocabulary by two perfumers and two fragrance assessors. From those data, the researchers compared an axis grouping onion-like, spicy, animalic, and heavy characteristics with another grouping milky and sweet characteristics.

The paper’s main text reports that pairs with smaller differences in spicy/animalic characteristics were also judged more similar by ordinary raters. The relationship with milky/sweet characteristics was not significant. However, real and fake couples did not differ significantly in the differences between their expert spicy/animalic ratings themselves.

The table presenting this comparison contains inconsistencies with the text concerning whether it includes only real couples or also fake pairs, and the explanation of significance symbols. It is therefore best read as a suggestion about the odor characteristics that might guide judgments. The comparison uses a 2018 vocabulary study of the same samples, not independent replication with a different donor population.

Summary

Third-party raters perceived the unfragranced body odors of real partners as more similar than those of recombined pairs, while that group difference was absent when donors used their usual deodorants or other products. However, the overall analysis aggregated by couple was not significant, so confirmation with more donors is needed.

The researchers also explored associations with hormonal contraception at relationship formation, relationship satisfaction, and spicy/animalic odor characteristics. The authors proposed the possibility of choosing partners with similar odors, but did not directly measure genes or actual partner choices. The work can be viewed as an initial study of how people perceive body-odor similarity, including the influence of everyday products.

Research funding and affiliations: Jan Havlíček received Czech Science Foundation funding (18-15168S). Kate Williams was affiliated with Unilever Research at the time. No separate conflict-of-interest declaration was identified.

Reference

Allen C, Havlíček J, Williams K, Roberts SC. Evidence for odour-mediated assortative mating in humans: The impact of hormonal contraception and artificial fragrances. Physiology & Behavior. 2019;210:112541. doi:10.1016/j.physbeh.2019.05.002