The universal beauty standard that never was

There is a version of this subject that appears everywhere. It runs: science has shown that beauty is universal, that it comes down to symmetry, averageness and specific ratios, that these signal health and fertility, and that cultural variation is superficial decoration on a biological constant.

There is a real research literature underneath that summary. The summary is not what the literature says.

Averageness, which is the strongest finding

Start with the one that holds up best. Composite faces, made by averaging many individual faces together, are rated as more attractive than most of the individual faces that went into them. This has been found repeatedly since it was reported in the psychological literature around 1990, and versions of it were noticed considerably earlier: Francis Galton produced composite photographs by multiple exposure in the 1870s and remarked on their appearance, although he was pursuing an entirely different and considerably less reputable project at the time.

The effect is reasonably robust and has been replicated across several populations. The explanations offered for it, however, range from the evolutionary to the entirely mundane, and the mundane ones are strong. Averaged faces are more prototypical, and prototypical stimuli are processed more fluently, and fluent processing feels pleasant. Averaging also smooths skin texture and removes asymmetries and blemishes as an artefact of the method, which means composite images are not simply average faces; they are average faces with better skin.

So: a real effect, replicated, with competing explanations, at least one of which requires no biology at all.

Symmetry, which is weaker than advertised

The symmetry hypothesis proposes that bilateral symmetry signals developmental stability and therefore genetic quality, and that preference for it is adaptive. It is enormously popular in popular science writing.

The empirical picture is patchier. Effects are typically found but are small, some studies find them only under particular conditions, and manipulating symmetry in images can produce results that are confounded with the averaging artefacts described above, since making a face symmetrical also makes it smoother. Meta analytic work has found associations in the expected direction but has raised questions about effect size and about publication bias, which is a general problem in this literature rather than a specific accusation.

The reasonable statement is that a modest preference for symmetry appears in many studies, that it is smaller than popular accounts imply, and that the adaptive explanation for it is a hypothesis rather than an established mechanism.

The waist to hip ratio, and what happened to it

The best known specific claim in this field is that a waist to hip ratio of around 0.7 in women is preferred across cultures and signals fertility. It was proposed in the early 1990s and became extremely widely cited.

Subsequent work complicated it substantially. Studies conducted with participants from populations outside the industrialised West found different preferences, and work examining the stimuli themselves showed that the original line drawings varied body mass along with the ratio, so that participants may have been responding to overall body size rather than to the ratio in isolation. Research comparing preferences across groups, including work by Martin Tovée and colleagues comparing observers in different countries and in populations experiencing different resource conditions, found that body mass preferences track local circumstances, with heavier bodies preferred where resources are scarce.

That last finding is the important one, and it points at a general principle: preferences correlate with what is locally scarce and locally indicative of security. That is a cross cultural regularity of a kind, but it is a regularity that produces different standards in different places, which is close to the opposite of what universalism claims.

The sampling problem

One methodological point does a great deal of damage to the universalist reading, and it is not controversial within the field.

A very large share of published psychological research, including a large share of the attractiveness literature, has drawn its participants from university students in wealthy Western countries. That sample is unrepresentative of humanity on almost every dimension of interest, and this has been argued forcefully within psychology itself for well over a decade.

A preference found repeatedly in that sample tells you it is robust in that sample. Calling it universal requires evidence from populations that were not sampled, and where such evidence has been gathered it has frequently produced variation.

Finding the same result in twenty studies of the same population is not twenty pieces of evidence about humanity. It is one.

And the wider problem

The attractiveness literature sits within psychology, and psychology has spent more than a decade confronting a serious replication problem: large scale attempts to reproduce published findings have succeeded at rates well below what researchers expected. Evolutionary psychology has been among the areas most scrutinised.

This is not a reason to dismiss the field. Reform has been substantial, with preregistration, larger samples and multi laboratory replication now common, and the newer work is considerably more credible than the older work. It is a reason to treat any confidently stated finding from the 1990s that has not been replicated with modern methods as provisional.

What history adds

The historical record is the strongest single argument against the universalist reading, and it is usually left out because attractiveness research and cultural history do not talk to each other.

This magazine has documented, in other essays, the complete inversion of skin colour preference in Europe within three decades, examined in the tanning essay; the reversal of male facial hair norms twice within a century, in the grooming essay; the shift of an entire culture's ideal complexion under the influence of photographic chemistry; and the manufacture of a bridal aesthetic within a single generation, in the marriage essay.

Each of these is a documented change in a widely shared standard, occurring far too fast for biological explanation and far too broadly to be dismissed as decoration. Any theory of beauty has to account for them, and a theory that treats culture as surface ornament on a constant cannot.

The defensible position

Something like this. There are probably some weak, broadly shared perceptual tendencies, most plausibly towards cues of youth, apparent health and prototypicality, with small effect sizes. On top of that sits an enormous, fast moving, historically contingent structure of standards that varies by place, period, class, technology and medium, and which accounts for most of the variance in what anybody actually finds beautiful.

That is less satisfying than either extreme. It has the advantage of matching both the psychological evidence and the historical record, which is more than can be said for the version that gets repeated.

Where to look

Readers who want to go further should start with the holdings themselves rather than with summaries of them. The PubMed and the Royal Historical Society both publish catalogue and research material relevant to this article. Both links are given for reference only. Neither institution has been assessed by us, neither has any connection to this magazine, and nothing here is an endorsement of any organisation.