By Chris Clarke
University of has discovered that we are attracted to people who look like ourselves.
Her experiment saw 682 participants interact with one another across 2,285 speed-dating sessions.
After each 3-minute interaction, participants of the opposite sex rated each other on facial attractiveness, and kindness and understanding. The result? Participants rated partners who had geometrically average faces and faces similar to their own as more attractive.听
Participants also received higher facial attractiveness ratings from partners of the same ethnicity, compared to those from a different ethnicity. And people with similar facial features rated each other as appearing more kind, regardless of ethnicity.
Research News spoke with Ms Zhao about her findings and how they've changed her own perceptions.
You鈥檝e spent the last few years exploring why some people date those who look similar to themselves. Can you tell us why you decided to embark on this work?
Facial attractiveness is an interesting topic and it seems that a lot of people have experienced or observed couples who look similar. For example, Siblings or Dating is a popular Instagram page. But it doesn't seem intuitive as to why people would date someone who looks similar to themselves.
We also wanted to clarify whether past facial attractiveness findings 鈥� which relied on ratings of facial images and computer-generated faces 鈥� could be replicated in a real-life, face-to-face setting (i.e. A speed-dating environment).
How have your findings changed your perception of the dating world?
Yeah it is a cool thing to find. Our measure of facial similarity was mathematically calculated from facial landmarks 鈥� just 28 coordinates on a 2D image. So I find it interesting that this seemingly abstract measure could predict attractiveness ratings that participants give each other after a 3 minute speed-date, especially in such a noisy environment when people are moving, talking, and interacting with each other in 3D space.听聽
How did you conclude that faces that look similar to our own spark a sense of kinship?
In addition to the speed-daters rating each other on facial attractiveness, we also asked them to rate how kind they thought their partner was. We found that the more facially similar a participant and partner were, the higher the kindness ratings, even when we accounted for ethnicity. These findings are in line with an existing theory where a preference for similarity is not driven by the attractiveness of similar faces per se, but that similar faces may signal kinship, and in turn, people find faces that they trust more attractive.
How long did it take you to analyse all of these interactions, and how did you go about categorising them and finding trends?聽
My PhD supervisor Brendan Zietsch has been running a long-term speed-dating study from 2010 until it was interrupted by COVID in 2020. In four of these years, we took facial images of the speed-dating participants.
The analyses itself didn't take very long. The most time consuming aspect was the placing of the facial landmarks, which was done by several of the co-authors and research volunteers.听
Ethnicity is a variable that would play a large role in facial structure, so we wanted to control for this. During the speed-dating survey we asked participants what their ethnicity was, so we didn't need to manually categorise everyone. Another aspect we accounted for was whether each speed-dating interaction consisted of a participant and partner of the same ethnicity. And we did actually find that those who interacted partners of the same ethnicity did on average receive higher attractiveness ratings.
In your opinion, does the old saying 'opposites attract' still ring true, or do you think your research has partly debunked this theory? 聽
I think it partly debunks this theory. There is general evidence that people prefer partners who share similar traits to themselves. So our findings align in that regard. And when it comes to partner choice, couples tend to have similar personality traits, levels of education, age etc. This is called assortative mating.
What is next for you in this area of research?
From our research, we know that people are attracted to those who are facially similar, but despite there being anecdotal evidence for couple similarity, there mixed evidence as to whether couples are facially similar. So we are interested to see whether these preferences translate to actual choices, given the general constraints of the mating market (i.e. you may prefer Brad Pitt, but he might not want you).
Our current study is also limited because we use facial landmarks which only measure facial structure from images. But landmarks ignore all qualitative information such as hair, eye colour, and contrast. We have a follow up study (under review) that uses deep neural networks to extract information from the same photographs. These neural networks are able to analyse the entire image rather than facial structure alone. I recently presented the findings of that study at HBES in Aarhus, Denmark.听
