The notion that people in long-term relationships start to look more like each other over time has been around for decades, and is in fact rooted in a scientific experiment conducted back in the 1980s. But is this strange phenomenon actually a thing?
The unusual idea entered the world's consciousness in a psychology study published in 1987, which concluded that married couples who had lived together for 25 years began to physically resemble one another as a result of their prolonged cohabitation.
As for how such a thing could be physically possible, the authors, led by the late social psychologist Robert Zajonc, proposed that "convergence in the physical appearance of spouses" could be attributed to the fact that life-long lovers become so in sync with one another, they end up unconsciously imitating each others' expressions, which in time changes the appearance of their faces.
"An implication of the vascular theory of emotional efference is that habitual use of facial musculature may permanently affect the physical features of the face," the researchers suggested.
"The implication holds further that two people who live with each other for a longer period of time, by virtue of repeated empathic mimicry, would grow physically similar in their facial features."
To this day, these ideas have taken hold in both psychological literature and everyday culture, with some going as far as to assert that the hypothesised phenomenon has been "scientifically proven". But has it?
According to a new analysis from researchers at Stanford University, no, it has not.
"A closer look at the literature reveals that while the convergence in physical appearance hypothesis is one of the tenets of current psychological science and has been widely disseminated through textbooks, books, and landmark papers, it has virtually no empirical support," the authors of a new study, led by electrical engineering PhD student Pin Pin Tea-makorn, explain in their paper.
"[The 1987 experiment], while elegantly designed, was based on an extremely small sample of 12 married heterosexual couples. Furthermore, its findings have never been replicated."
To reinvestigate the merits of supposed facial convergence, Tea-makorn and her co-author, computational psychologist Michal Kosinski, conducted a do-over of the 1987 experiment, but with a much larger sample of potential lookalikes, not to mention the capabilities of 21st century technology.
While the 1987 study only examined photos of 12 couples from when they were first married and 25 years later, the new research collected images of 517 married couples from public online sources, comparing their faces shortly after they were married with images taken 20 to 69 years later. (All of the couples were White and heterosexual, with the researchers saying they were unable to secure enough images of homosexual and non-White couples to allow for a meaningful analysis.)
In addition, Tea-makorn and Kosinski had not only human judges to make the comparisons (153 people recruited online), but also a facial recognition algorithm called VGGFace2, which has been previously shown to outperform humans in judging facial similarity.
In other words, this larger and more robust experiment was likely the best chance ever for the physical convergence hypothesis to finally be validated after decades.
Alas, the researchers didn't find anything to suggest that couples start looking more like each other as the years pass, admitting that even they were surprised by the results.
"When we started this project, I was convinced that we will easily find evidence for the convergence in facial appearance," Tea-makorn says. "This is one of those theories that all undergrads learn in their Psych 101."
Nonetheless, based on the similarity rankings provided by both human judges and the all-seeing VGGFace2, the spouses' faces did not become more similar with time, and the human judges even glimpsed a very slight movement the other way (suggesting the couples' faces might actually become less similar, although the observed difference was very small).
However, while the team didn't find any evidence to suggest that couples progressively resemble each other, the results did confirm that people do seem to pick long-term partners that look similar to them, at least compared to other faces picked randomly.
"Spouses' faces are similar but do not converge with time. This brings facial appearance in line with other traits – such as interests, personality, intelligence, attitudes, values, and well-being – which show initial similarity but do not converge over time."
The findings are reported in Scientific Reports.