Two women meeting for the first time may decide within minutes whether a friendship is likely to form—and new research suggests that smell can be as influential as anything they see or hear.
The findings come from a Cornell University study published in Scientific Reports, which examined how people form first impressions and decide whether they want future social interaction.
Working with heterosexual women, researchers found that participants’ personal preferences for another person’s everyday scent—captured on a T-shirt—helped predict how much they liked their partner after a four-minute “speed-friending” conversation. The relationship also worked in the other direction: the quality of the face-to-face conversation later shaped how the women judged the T-shirt scent when smelling it again on its own.
“People take a lot in when they’re meeting face to face. But scent—which people are registering at some level, though probably not consciously—forecasts whether you end up liking this person,” said Vivian Zayas, a psychology professor and co-author of the study.
While research on social smell has often focused on romantic attraction, this study looked instead at platonic connection. It also moved away from the idea of a person’s purely “natural” odor—separated from fragrances, pets, and other environmental influences. Instead, the researchers emphasized what they call “diplomatic odor”—a scent shaped by everyday habits and environment, reflecting the unique signature people develop through their choices and surroundings.
In the study, “smell-only” impressions closely mirrored in-person impressions. If a participant rated someone as having strong friend potential based on the scent of the T-shirt, that same person tended to receive a similarly positive rating after the brief conversation.
Additionally, evaluations from the live interaction predicted changes in a second round of scent ratings, suggesting that social experience can change how someone’s smell is perceived.
Zayas noted that what stood out was how consistent each participant’s preferences were—yet how different those preferences could be from one person to another.
“Everybody showed they had a consistent signature of what they liked,” she said. “And the consistency was not that, in the group, one person smelled really bad and one person smelled really good. No, it was idiosyncratic. I might like person A over B over C based on scent, and this pattern predicts who I end up liking in the chat.”
