Do couples really look alike? What the research says

The short version

• Supported: partners tend to look somewhat more alike than random pairs of people.
• Not supported: couples’ faces converging over decades together. The best-powered test found no convergence.
• Unknown: why partners resemble each other, and whether resemblance predicts attraction between strangers.

1987: the convergence idea

The best-known version of the claim comes from a paper by Robert Zajonc and colleagues, published in Motivation and Emotion in 1987 and titled “Convergence in the physical appearance of spouses.” The researchers compared photographs of couples taken when they were first married with photographs taken 25 years later, and reported that spouses had come to look more alike over that time.

The authors offered an appealing explanation: people who live together for decades share emotions and mimic each other’s expressions, and that shared habit of expression might gradually shape their faces in similar ways. It is a memorable idea, and it spread widely. It also rested on a small sample.

1989: similar from the start?

Two years later Verlin Hinsz published “Facial resemblance in engaged and married couples” in the Journal of Social and Personal Relationships. The study compared real couples with randomly paired men and women and found that real couples resembled each other more than the random pairs did.

That matters because it points to a different story. If couples who have not yet spent decades together already look alike, then at least part of the resemblance is there at the beginning, and does not need years of shared life to explain it.

2020: a large test finds no convergence

In 2020, Pin Pin Tea-makorn and Michal Kosinski published “Spouses’ faces are similar but do not become more similar with time” in Scientific Reports. They set out to test the 1987 claim with a much larger sample: photographs of hundreds of couples, taken at the start of their marriages and again many years later. Similarity was assessed by human judges and also by a facial recognition algorithm.

The title gives the result. Spouses’ faces were more similar than chance would predict, which agrees with the earlier work. But they did not become more similar over time. The convergence effect did not replicate.

So why do partners look alike at all?

This is where the evidence runs out and explanation begins. Researchers have proposed several reasons, and they are not mutually exclusive.

• People pair up with people like themselves. Partners tend to be similar in many ways, including age and background, and some of that shows in a face.
• People meet those who are nearby. Partners often come from the same places and communities, which can mean shared ancestry and therefore some shared features.
• Familiar faces may be appealing. People may be drawn to faces that feel familiar, and few faces are more familiar than your own and your family’s.

The three studies described here show that the resemblance exists. They do not tell us which of these explanations is responsible, or in what proportion.

What the research does not show

It is easy to slide from “couples resemble each other” to “you will be attracted to people who resemble you.” The studies do not support that step.

• They looked at established couples, not at strangers meeting for the first time.
• None of them tested whether resemblance predicts how a first meeting goes.
• A pattern across many couples says little about any one pair. Plenty of happy couples look nothing alike.
• If shared background explains the resemblance, then a similar face is a side effect and not a cause of anything.

So the honest summary is this. Partners tend to look somewhat alike from the outset. Whether that tells you anything useful about who you will enjoy meeting is an open question.

How Magnetic is testing it

Magnetic uses facial-geometry similarity as one ingredient in deciding who meets whom. Members upload a photo after giving explicit consent, and other members see an illustrated avatar, never the photo. Then people simply meet: one or two 20-minute video dates on a Magnetic Night, after which each person privately says whether they would like to see the other again.

We treat the premise as an unproven idea, because it is one. Magnetic is in a free beta in San Diego County and no nights have run yet, so we have no results of our own to report. If resemblance turns out not to predict anything, we will say so. The About page sets out what we know and do not know, and the FAQ covers how photos are handled.

References

1. Zajonc, R. B., Adelmann, P. K., Murphy, S. T., & Niedenthal, P. M. (1987). Convergence in the physical appearance of spouses. Motivation and Emotion, 11, 335–346. https://doi.org/10.1007/BF00992848
2. Hinsz, V. B. (1989). Facial resemblance in engaged and married couples. Journal of Social and Personal Relationships, 6, 223–229. https://doi.org/10.1177/026540758900600205
3. Tea-makorn, P. P., & Kosinski, M. (2020). Spouses’ faces are similar but do not become more similar with time. Scientific Reports, 10, 17001. https://doi.org/10.1038/s41598-020-73971-8