Think of someone particularly attractive. This could be your love, a friend—perhaps a famous actor. Now think of telling that person that they look average. Not a good opening line, right? As terms of endearment go, “that’s my wife, isn’t she average?” won’t win many points. But if you have an attractive wife, it would be accurate.
That’s because in general, an average of several individual faces, when combined, is perceived as more attractive than the constituent faces. As far back as the 1880s, Sir Francis Galton used composite portraiture—a technique in which multiple exposures of different faces were composed on the same photographic plate—to create a sort of blurry average of the faces. He and his colleagues were surprised to find that the combination face was, well, hot. Hotter than the individual pictures.
About a century later, this effect would be tested in more depth and at with more rigor using computer composites. Relatively simple programs allowed us to create clear digital averages with many more constituent photos, and test attractiveness in studies with statistically significant numbers. Try it out for yourself here. Across nearly all of the studies, images of faces manipulated to make features closer to average were judged more attractive.
A common and likely inaccurate belief is that symmetrical faces are more attractive. To an extent, we do like symmetrical features, but symmetrical features clearly aren’t everything. If I had perfectly symmetrical green Nike swooshes for eyebrows, it wouldn’t look good. This is actually fairly obvious if you think about it; anything in the extreme, such as eyes so close together they are nearly touching, would strike us as off-putting regardless of the symmetry. The effect of averageness is isolated from symmetry even in less extreme examples. A 2004 study demonstrated that the effect of averageness in determining attractiveness of female faces was independent of any effect of symmetry on the perceived attractiveness of faces. Full PDF here.
As it turns out, averageness really does correlate with attractiveness. A University of Toronto study tested this in an interesting way, using multiple headshots of the same person, with different distances from eyes to mouth, or differences in space between eyes. The person was at her most attractive when the space between her pupils was 46% of the width of her face from ear to ear, and when the distance between her eyes and mouth was 36% of the length of her face from hairline to chin. While these specific numbers may not mean much, perhaps the most important finding is that those ratios are essentially the same as an average female face. For a real-world example, Jessica Alba meets this average criteria.
So what is going on here? As with almost any accurate scientific statement, we can’t be 100% certain. But the strongest explanations tend to cut two ways. The first has to do with evolutionary pressures—essentially, in choosing mates, we tend to favor characteristics close to the mean. This exerts directional selection pressure against more extreme characteristics. Over time, less average characteristics would be selected out, as they are perceived as suboptimal on an individual basis.
The second has to do with cognitive fluency; this theory holds that average faces are closer to what we picture in our mind when we think of a face, so a prototypical member of a category strikes us as more familiar and is effectively easier to register through our visual system. This line of thinking led to the idea that part of it may be the “exposure effect”, in which we trust things we’ve seen more often (think of any multinational brand—Coca Cola, McDonalds, Toyota). At least in this application, exposure effect doesn’t seem to be the key, but that doesn’t rule out cognitive fluency playing a role.
There are neural correlates as well: activity in certain parts of the orbitofrontal cortex increases with increasing attractiveness of faces.
Whatever the explanation, this average-is-attractive idea is consistent. One study suggests that computer-manipulated averageness makes birds, fish, and automobiles more attractive.
There is, however, a wrinkle to this. It turns out that average faces are attractive, but very attractive faces may not be average. In perhaps the strongest study on this topic to date, Perret et al. decided to test the averageness hypothesis by experimenting with three composite groups: (1) an average of 60 faces, (2) an average of the 15 most attractive of those 60 faces, and (3) a hyper-attractive composite, which was a version in which the “attractive” qualities were exaggerated on the digital image. Number 3--the hyper attractive exaggerated face—was judged the most attractive. The implications are fascinating because that face is explicitly not average—it is exaggerated (though it appears slight when you look at the study images). So the takeaway is that average certainly may be attractive, but very attractive may not be average.
Limitations (i.e. that’s cool, let’s complicate it)
I should note a few observations in research in this area. First, the volume of research focused on what we find attractive in women is much larger than that focused on men. Second, the vast majority of both bodies of research tend to take a heterosexual perspective in judging attractiveness. While a female-as-attractive, heteronormative perspective is probably not surprising, it’s still a real limitation. I’ve highlighted that which I could find in the next two paragraphs for men and same-sex relationships, but note that the number of confirming (or contradicting) studies is smaller for these two sections.
Many of the studies have been about women. What about the guys? Is this some weird affirmation of the male-gaze thing? Are average male faces attractive too?
Not quite, kind of yes, and we aren’t sure. Both averageness and symmetry tend to increase attractiveness for everyone, but for male facial attractiveness, sexual dimorphism—or the difference between typically male and female traits—seems also to be key. Interesting results suggest that ovulating heterosexual women prefer faces with masculine traits. These traits are associated with increased exposure to testosterone during key developmental stages, and include a broad forehead, prominent nose and cheekbones, large jaw, and strong chin—all of which may be a departure from average to some extent. This preference was less strong during menses and luteal phases. Thus, not only may women prefer men to have faces that are not average, their preference may also be stronger at different phases during their hormonal cycles. That said, at least one rigorous study suggests that feminized features of a male face were preferred, so to some extent, we aren’t really sure what women prefer when it comes to men’s faces.
Moreover, this preference also seems to vary between countries—masculinity preference increased as general health indicators decreased--suggesting among other things that attraction may be impacted by environment and may be a socio-cultural learned phenomenon to some extent. All that said, this should be read in concert with facial average studies, as those also tend to confirm averageness as attractiveness for men.
Do these findings hold true for people attracted to the same sex?
Again, not quite! In one study—and let’s be clear, as in so many areas, this is understudied for people attracted to the same sex—“homosexual men demonstrated stronger preferences for masculinity in male faces than did all of the other groups. Homosexual women demonstrated stronger preferences for masculinity in female faces than did heterosexual women. These results suggest attractiveness judgments of same-sex faces made by homosexual individuals are not a mirror image of those made by heterosexual individuals of the opposite sex. Our data suggest that face preferences of homosexual individuals reflect a system of biologically and socially guided preferences at least as complex as those found among heterosexual individuals.” So average again may be attractive, but sexual dimorphism and varying preferences in typically masculine and typically feminine traits are relevant.
You’re Extraordinary
So, what do we do if we are—sigh—not average? That’s where the exposure effect mentioned before may actually come in handy. The exposure effect is so powerful that people judge almost everything, from photographs, to songs, to themes in movies, to shapes, to made-up words, to people more positively. This will likely be the subject of a future post—but for the romantics, take heart. Just showing up repeatedly may be enough.
It’s also worth pointing out that these studies are not a simulation of the human mind—they are primarily visual, and don’t account for the rich variety of social, behavioral, and physical interactions that make us love one another. Everything from smell, to shared experiences, to kindness impacts how we feel. As Helena notes in A Midsummer Night’s Dream (Act 1, Scene 1):
Love can transpose to form and dignity.
Love looks not with the eyes but with the mind.
And therefore is winged Cupid painted blind.
-William Shakespeare