Beauty is in the eye of the beholder, or so we’re often told. But for the scientific community, this is controversial. Many psychologists have argued that we tend to rate each other quite consistently: if a woman is very attractive to one man, the chances are that she’s very attractive to most men, and vice versa.
This is considered good evidence that beauty is fairly objective – that is, that we all mean the same thing (or at least a very similar thing) when we talk about it, and outliers should generally be discounted.
Others, however, have claimed that this conclusion is a misunderstanding, and that personal preference accounts for at least as much of our taste as any independent sense of beauty.
We’ve taken a look at the strength of both arguments to find an answer.
On the face of it (as it were) this is an easy argument to make. Many of pop culture’s most fêted icons are celebrities at least in part because of how they look. If there were no general agreement on what made someone attractive, it’s hard to imagine that many models, actors, or musicians would achieve the same fame. It’s also unlikely that demand for cosmetic surgery would exist in the way it does today: patients are overwhelmingly female, and typically want smaller, straighter noses, slimmer waists and stomachs, more pronounced lips, breasts, and buttocks, and fewer wrinkles.
That’s all very well, you might say, but isn’t this a terribly white, Western-centric idea of beauty?
Perhaps not. Sociologists have reported significant cross-cultural and cross-ethnic agreement on at least facial attractiveness, and people are reliably able to predict which faces others – even others from very different groups to their own – will find beautiful. The average effect size in the relevant studies is also much larger than usual, which would usually mean that we can put a good deal of confidence in their conclusions. (Get to the end to see if that’s true in this case.)
Both adults and children judged to be more beautiful are universally treated more favourably, even by those who know them well. Attractive adults also tend to be more successful, have higher self-esteem, and exhibit better mental health. Even babies seem to prefer more beautiful faces.
Of course, what’s attractive in a man and what’s attractive in a woman are not always identical. Human males and females tend to have different facial features, and people are generally very good at telling the difference, even in the absence of all hair and make-up. The average adult male has a longer and broader chin, a more developed forehead (meaning smaller, narrower-looking eyes), broader cheekbones, and thinner lips. Correspondingly, the average adult female has a smaller chin, larger eyes, higher and thinner cheekbones, and fuller lips. Facial proportions vary across different ethnicities and geographies, but sex differences remain fairly constant.
This is largely a result of exposure to different ratios of oestrogen and testosterone during adolescence. Both hormones can serve as immunosuppressants, meaning they limit our bodies’ ability to fight off bugs – but according to some researchers, this means that those who survive to adulthood with highly male or female faces (and therefore higher levels of the relevant sex hormones) might actually be displaying a high level of reproductive fitness, because they’ve proved that they can fight off diseases despite constraints.
We know that heterosexual men tend to prefer women to look more feminine, i.e. have exaggeratedly “female” features. It’s no surprise that larger eyes, fuller lips, and higher cheekbones are all effects that make-up, and sometimes surgical intervention, seek to imitate.
But heterosexual women do not straightforwardly prefer men to look facially “male”. There’s some evidence that women prefer more masculine faces, particularly in the most fertile phase of their menstrual cycle (although other evidence contradicts this), but studies have also found effects in favour of feminine male faces or no preference either way.
We do know that men with feminine faces are perceived as more agreeable and more honest than their hyper-masculine counterparts, whereas more masculine faces are perceived as belonging to older and more competent people.
There are, however, some universals. When deliberately trying to create attractive human faces, researchers found something interesting: people are particularly enticed by faces that have been formed by blending many people together. In fact, the more faces included in the blend, the more attractive the merged face is considered to be. Counterintuitively, it seems that what we really want is someone who looks…average.
There are several possible reasons for this. The first is that averageness tends to reflect “good genes”: a normal-looking person is less likely to be carrying any genetic mutations that might harm offspring. They’re also more likely to be heterozygous, which means that they carry one dominant and one recessive allele for various traits, instead of two dominant or two recessive alleles. This is an indicator of health because it means that you’re probably more resistant to pathogens.
There’s also the fact that humans tend to prefer average representations of categories in general. We like average-looking dogs, average-looking birds, and average-looking watches. This is because average examples of objects or creatures are the ones with which we’re most familiar – ask a child to draw a sheep and they’ll rarely produce a Manx Loaghtan – and familiar things are easy for our brains to process.
Another explanation is also possible. Humans consistently rate symmetrical faces as most attractive. This is probably because asymmetries in human faces, like those elsewhere on or in our bodies, are often a result of infection or environmental challenges during our development. These asymmetries (called “ fluctuating asymmetries,” or FAs) are associated with a number of outcomes traditionally considered undesirable:
One paper even found that women with more symmetrical male partners tended to have more orgasms, even controlling for physical attractiveness, relationship length, various emotional components of the relationship, male income, and faking. It has not been replicated.
Several studies have also found that women are able to discriminate between more and less symmetrical (i.e. attractive) men via scent alone, particularly if the women are ovulating.
Without good reason to think that fluctuating asymmetries occur disproportionately on one side of the human body, faces composed of many people’s portraits will by definition tend towards symmetry. This is another plausible reason for the fact that we tend to favour “representative-looking” individuals.
Inevitably, not all the effects discussed above have been independently replicated with large sample sizes. What we can say with certainty is that when it comes to attraction, some near-universal human preferences, such as for facial symmetry or typical appearance, do exist.
But that is not the whole story. Even within exclusively heterosexual attraction, many different archetypes of masculinity and femininity exist, and whilst some (tall, dark, and handsome, petite and blonde, etc.) have tended to dominate Western culture, any group of same-sex friends is likely to pick out quite different-looking individuals as preferred partners.
Why? Well, one answer lies in the fact that we ourselves are very different from each other. And as we know, when it comes to personality we tend to like others who resemble ourselves.
More uncomfortably, the same is true of appearance. Gay couples have often been on the receiving end of criticism for dating people who look exactly like them, because facial similarities are most obvious when couples are the same sex. But the phenomenon extends globally.
It also goes beyond anecdata. Research participants who are shown images of their partner morphed with images of other people almost invariably find partner self-morphs most attractive; that is, they prefer faces that are a mixture between their partner and themselves. To be exact, our favourite faces are our 88% our partner’s and 22% our own. This preference operates at an unconscious level: participants were not informed that their partners’ faces were blended with their own, and nor were they able to detect it.
In fact, people are statistically more likely to be genetically similar to their spouse than to random strangers, even controlling for factors like social sorting.
This throws up an obvious hazard. Inbreeding, or the production of children between closely related family members, results in homozygosity – the opposite of heterozygosity, which we talked about above. Homozygosity risks producing low levels of reproductive fitness: “bad” recessive genes are amplified over successive generations, which can result in susceptibility to disease, disability, or even complete infertility.
Whether or not this happens is largely due to the “good” or “bad” characteristics of the original gene pool; if it contains many “good” genes, the offspring might be disproportionately successful (in a process known as “purifying” or “negative” selection). Non-human animals are frequently inbred on purpose to create offspring that humans consider desirable, and the same phenomenon can occur in people: Cleopatra VII, the famously brilliant and beautiful ruler of Ptolemaic Egypt, had only one set of great-great-great-great grandparents. This makes her statistically more inbred than the Habsburg Charles II of Spain, who was severely physically disabled and died at 38.
Fortunately, humans have a strong anti-incest instinct, and very rarely develop sexual attraction to those they spend considerable time with as young children. But our attraction to people who are otherwise physically and genetically similar to ourselves is a strong argument in favour of beauty’s being, at least in part, in the eye of the beholder.
We also tend to know our own “level” when it comes to attraction. If we’re someone traditionally considered unattractive, but only find ourselves interested in supermodels, we’d have a difficult time in romance. Fortunately, our own attractiveness helps determine whom we select as partners. Unfortunately for the attraction-is-subjective theorists, this doesn’t tend to be because we find them more attractive than other people do; we’re just more willing to accept less-attractive partners.
But even the evidence used to support universal norms in attraction can lend itself to the opposite argument. If the studies that find changes in women’s attraction to masculine or symmetrical faces over the course of their menstrual cycle are right, we have to conclude that women’s preferences aren't even consistent at an individual level, varying with hormone exposure and other environmental factors.
The fact that researchers have so far been unable to determine whether women are more attracted to feminine, masculine, or neutral male faces might also point to an underlying lack of consensus. Sampling different groups appears to produce different results, which is not something you’d expect to find if there were universal agreement on what constitutes “attractive.” Women, at least, appear to be less uniform in their sexual preferences than men.
There’s another problem too. We started out by saying that across cultures and ethnic groups, people tend to agree about which faces are most attractive. But the way that this agreement is measured is not universally accepted. If Alyssa and Brett both look at a face, and Alyssa rates it a 3/10 whilst Brett rates it a 7/10, we’d say they disagreed. The difference between their ratings would be 4 points. More points equal more disagreement.
But if we then added Caitlin, David, and Eric’s ratings, which (let’s say) were 2/10, 5/10, and 8/10 respectively, something different would happen. Even though none of the face judges agreed with each other, the average number of points between all the ratings would have shrunk. In fact the more judges we added, the less difference we’d see between their ratings.
By the time we got to 10 judges, each of whom rated the face a different number out of 10, we’d be able to say that the judges only disagreed with each other by 1 point. This would imply that they agreed with each other by a whopping 90%, when in fact the difference between their scores was (in at least one case) as high as 9 points.
This way of measuring consistency between judges is called Cronbach’s alpha, and it’s the one used in almost all cross-cultural attractiveness studies. This includes the review we linked above, which found agreement between judges of 9-25%.
Or, in other words, not all that much agreement.
On the “objective” side, there may be cross-cultural agreement about which faces are attractive. Men generally prefer women to have stereotypically feminine faces, and everyone finds generic or typical faces to be more attractive than unusual ones. As a species, we’re also drawn to symmetry in the faces of those around us.
On the “subjective” side, the fact that we all tend to prefer partners who look a bit like ourselves, and who share some genes with ourselves, means that over 20% of attraction is likely to be subjective. We also pursue partners who are about as attractive as we are, and women in particular vary in their preferences about male faces. The studies that have found cross-cultural agreement about beauty should probably be reconsidered.
It seems fitting, then, to declare a win for subjectivity. What you find attractive may not be entirely unique to you, but in the absence of some groundbreaking new evidence, your view on whether attraction is objective or not probably will be.