Can Women Identify An Average Face?

People often confuse average with “neither attractive nor unattractive,” but an average face may also be unattractive. How well can people identify the average?

Enjoy DatePsychology? Consider subscribing at Patreon to support the project.

What is attractiveness? Despite an abundance of literature on this topic, very little (if any at all) is dedicated to describing what attractiveness is. Attractiveness is, in practice, how you are rated by others. More specifically, attractiveness is how your objective physical features are perceived by others. At the end of the day, when we want to measure attractiveness, it is a number given to you by other human beings.

In other words, you’re only as attractive as others think that you are. You don’t have attractiveness independent of their perceptions of you. It isn’t an objective quality that exists outside of the minds of other human beings, despite fairly high agreement on what is attractive, and despite attractiveness reflecting objective physical traits.

An implication here is that no one can be “wrong” about how attractive you are — especially not large groups of individuals. If most people rate you low then you’re unattractive (at least to most people). At the end of the day that is all there is to it. You are a rating in the minds of others.

Above is the famous meme chart from the OKCupid dataset showing female ratings of male photos. I have seen countless good minds ruined by this chart. A lot of people see it and think that it’s impossible — how can women rate most men as “below average?”

Some people confuse “average” in a colloquial sense with the mathematical average. We use “average” in common parlance to mean “neither attractive nor unattractive.” An average person is a “mid.” The mathematical average, on the other hand, is the sum of your terms divided by the number of your terms. Most values can be below average: this is actually quite common.

Can most men be unattractive? Yes. There is no rule, neither in statistics nor in the cosmos, that they cannot. Must attractiveness follow a normal distribution? Again, no. There is no rule in math nor the universe that says it must be so. This, too, has broken a lot of minds. The belief that attractiveness must be normally distributed just feels right to some people.

How would you know if attractiveness is normally distributed? What would you do if you wanted to test this? Remember what attractiveness is — the ratings given to you by other people. You would simply take a sample (like the OKCupid dataset) and look at the actual distribution.

Somehow, despite seeing the actual distribution of attractiveness (this is what the chart shows), people continue to assert that attractiveness is normally distributed. “It must be so.” Yet, you can look at this dataset (and many others, as this is quite consistent) to see that it is not.

Another way this chart breaks minds is when the midpoint is labeled “average.” First, if you’re researching attractiveness you should avoid using the term “average” on your scale. “Average” means something, it’s a mathematical term, and the average face can also be an unattractive face (as seen in the OKCupid chart, the average is low in attractiveness).

This has also led people to claim that women “misidentify” what is average. “The women are wrong, we can’t all be unattractive!” However, OKCupid didn’t ask men and women to indicate if faces were close to the mathematical average in attractiveness. OKCupid asked men and women to rate facial attractiveness.

In fact, you probably don’t care if women can accurately identify an average face. You probably care if women find you attractive or not. But we’ll find out if women can identify an average face anyway.

Methodology & Results

I made a selection of seven male faces from the Chicago Face Database (Ma & Wittenbrink, 2015). The average rating of these faces was a 2.64 (the average across White male faces in the CFD was 2.98) on a 7-point Likert scale. At 2.64 on a 7-point scale this was also similar to the mathematically “average” face in the OKCupid dataset.

In other words, these faces represented the “average” male.

I collected a sample of 497 heterosexual raters (16% female) to rate these faces. I instructed participants not to think about if these faces were personally attractive to them, but rather to judge if they were more, less, or about as attractive as the typical man of the same age and ethnicity.

Here are the specific instructions that were given:

Important – try not to think if the face you see is personally attractive to you. Instead, think about how attractive this face is relative to other men of the same age and ethnicity. You can decide if the face is less attractive, about as attractive, or more attractive.

Instructions – looking at this face, is this man more, less, or about as attractive as the average man on the street (of the same age, sex, and ethnicity)?

Here are the results:

What stands out? The assessments made by men and women were quite similar. Despite being average faces, substantial percentages of men and women indicated that the faces were less attractive than the average man. Why?

I believe the answer is that we don’t recognize that the average man is facially unattractive. We operate with the implicit belief that average means “neither attractive nor unattractive.” When we see a face that is unattractive to us we assume that it is below average, rather than making the correct judgment (that the average male face is unattractive).

Is this really that surprising, though? The next time you walk into a room full of people ask yourself how many you find attractive. It’s highly unlikely that you’ll find 50% to be.

Discussion

What’s the takeaway? We’re really bad at identifying what an average face is. We don’t walk around thinking of faces in the context of “this looks like most other faces.” We assess faces as attractive or not. We see faces that we don’t like and call them “below average” even when they are not. And men don’t seem to be any better at categorizing the average face than women are.

5 comments
  1. From an evopsych perspective, why would male faces have evolved to be less attractive than female ones? Does this mean that the selection pressure on male faces to be attractive was weaker than that on female ones? Could this be because women care much less about male faces than males do when it comes to female faces? Conversely, could it be that the same dynamic operates in reverse in sexually dimorphic traits in which males have the advantage, such as size (incl. height and torso size), musculature and pitch of voice?

    1. I read David buss’s evolutionary psychology, in his words, female physical attractiveness not only indicates genetic qualities like male one but also indicates fertility or fertile potential. David listed an example, a tribal man told anthropologists he can tell a woman’s fertility by looking at her face. Moreover, female mates selection standards are not entirely on physical attractiveness, but male mates selection standards are almost on physical attractiveness and age. These evidences and studies are very consistent over decades in evo psych field. Women are avoiding men with less status (eg lower education or income) men are avoiding women who are less attractive than them (eg uglier than them older than them)

  2. As in many things structural, there does seem to be facial elements that are perceived as evocative in a positive way.
    I suspect that you,Alexander, have viewed Qoves’s YouTube videos, attempting to quantify the properties of male and female beauty.

    We’ve all observed in real life, examples of faces that appeal to the majority of viewers.

  3. If Thornhill and Palmer are right, then sadly, women could be more attractive than men on aggregate because early in human evolutionary history, undesirable men raped the women they found attractive.

Leave a Reply

Your email address will not be published. Required fields are marked *

You May Also Like

The Testosterone Blackpill

Enjoy DatePsychology? Consider subscribing at Patreon to support the project. In this article I review 13 recent meta-analyses…