Do Friends Who Click Have Similar Body Odors?

You have only just met, yet somehow you immediately get along. Do friends who experience this kind of connection also have similar body odors?

A 2022 study found that same-sex friends who said they had become close soon after meeting had more similar body-odor profiles than randomly paired people. In an experiment involving silent interactions between strangers, pairs whose body odors were closer according to an electronic nose were also more likely to report mutually that they had “clicked.”

Considering body odor as something friends may share

Ravreby and colleagues studied same-sex friends who were not romantically involved. They asked two questions: whether friends who clicked quickly had similar body odors, and whether that similarity was associated with initial interactions rather than merely resulting from years spent together.

First, they asked 235 people to describe a “click friendship” in their own words and analyzed descriptions from the 225 who said they understood the term. The 20 most frequent characteristics, such as an immediate sense of connection or being on the same wavelength, became criteria for recruiting friends.

Ultimately, 20 same-sex friend pairs, or 40 people, participated, with both members meeting the criteria. There were 10 male pairs and 10 female pairs, and the friendships had lasted an average of about 6.2 years. Thus, at this stage the researchers measured odor in established friends, not at the instant they first met.

Testing body odor collected on shirts with an electronic nose and human noses

Participants washed with a specified unscented soap and wore supplied 100% cotton T-shirts for two nights, for at least six hours each night. They were asked to avoid perfume, deodorant, and similar products, foods likely to influence body odor, and contamination of bedding with other people’s or pets’ odors. The collected shirts were stored frozen.

An electronic nose is a device that quantifies the pattern of responses from multiple gas sensors. Of the instrument’s 10 sensors, five responded to body odor and were used to characterize each person by comparing the air inside jars containing the shirts.

The sensor-response patterns of the 20 actual friend pairs were closer than those of same-sex pairs created by randomly rearranging the same 40 people. The researchers examined the difference from chance pairings using 10,000 rearrangements. The instrument did not identify specific molecules.

Human noses also found similarity, but results differed by task

In the human-smelling tests, 24 people were first given three body odors: those of two friends and one unrelated person. They were asked to select the odd one out. This task did not provide convincing evidence that they could distinguish the friend pairs.

When individual body odors were instead rated on five dimensions, including pleasantness and intensity, the combinations of ratings were closer between friends. Differences in the duration of inhalation while smelling were also smaller within friend pairs.

A separate group of 25 people directly compared two body odors and rated their similarity. Friends were rated as more similar than random pairs. Another 25 people compared photographs and likewise found greater visual similarity between friends. Choosing among three odors and directly comparing two place different demands on memory and comparison.

Importantly, the degree of similarity measured by the electronic nose did not correlate, pair by pair, with the similarity perceived by human noses. Both approaches found similarity among friends, but they were not necessarily detecting the same cues.

Seventeen strangers play a silent game for two minutes

Similarity between friends alone leaves open the possibility that shared diet or lifestyle made them more alike. The researchers therefore collected body odors from 17 people who did not personally know one another and had them perform brief face-to-face tasks with others of the same sex.

The task was a “mirror game”: participants faced one another 50 cm apart and matched each other’s hand movements for two minutes. Without talking, they paired in turn with everyone of the same sex. Seven men and 10 women produced 66 pairwise interactions. These were not 132 independent people; the same 17 people appeared in multiple pairs.

After each interaction, participants reported closeness, willingness to meet again, whether they were on the same wavelength, and similar impressions. Both participants reported clicking in 22 pairs. Electronic-nose body-odor profiles were closer in these 22 pairs than in pairs without mutual agreement that they had clicked.

Greater body-odor similarity was also associated with higher ratings of closeness, willingness to meet again, and related qualities on 10 of 13 measures. The relationships remained in analyses accounting for repeated participation in multiple pairs and examination of multiple measures.

However, body-odor similarity was not associated with movement synchrony derived from video recordings. The clearest associations with body odor in this experiment concerned participants’ reports of interaction quality.

How should the “71% prediction” figure be interpreted?

The researchers also tested a model that used electronic-nose data to classify pairs in which both people clicked versus all other pairs. In cross-validation that left out one of the 66 pairs at a time, accuracy was 71.21% and the area under the curve, or AUC, summarizing classification performance was 0.67.

This was internal validation on pairs formed from the same 17 people. The remaining training data could still contain the same participants, so the result does not mean the model can predict future friendships between entirely new people at that accuracy.

The classes were also imbalanced: 22 pairs mutually clicked and 44 did not. Always choosing the majority class, “no mutual click,” would yield about 66.7% accuracy. It is therefore important to consider the sample composition and AUC rather than accuracy alone.

Odor similarity is one clue to interpersonal connection

The study showed associations with body-odor similarity in two settings: established friendships and first encounters. It did not manipulate people’s odors to test whether relationships changed, however, so shared factors could still have influenced both body odor and closeness.

The relationships studied were same-sex and nonromantic, and the initial-encounter task was a brief, silent game. The researchers did not follow participants to see whether lasting friendships developed. Body odor may play a different role in everyday life, with its richer conversation and shared experiences. Identifying the molecules involved also remains a question for future research.

Summary

Twenty same-sex pairs of friends who had become close quickly had more similar body-odor profiles than random pairs. Direct human similarity ratings showed the same pattern, although the task of choosing an odd one out from three odors did not produce clear discrimination.

In the experiment with 17 strangers, pairs with more similar body odors according to an electronic nose were also more likely to report mutual clicking and to give higher ratings of closeness and related qualities. The study suggests that body-odor similarity may offer one clue to understanding interpersonal connection.

The research was supported by the Israel Science Foundation (ISF) and the Ariane de Rothschild Women’s Doctoral Program. The authors declared no competing interests.

Reference

Ravreby I, Snitz K, Sobel N. There is chemistry in social chemistry. Science Advances. 2022;8(25):eabn0154. doi:10.1126/sciadv.abn0154. Open-access full paper.