Attractiveness and Intelligence: Is There a Real Correlation?

The correlation between attractiveness and intelligence in adults is close to zero: the largest twin study to measure both independently found r = .03, which is statistically indistinguishable from no relationship. What is large is the belief that good-looking people are smarter, since faces rated as attractive are also rated as intelligent, with correlations as high as .81 between the two judgments.

That gap between perception and reality is the real story. This page covers the meta-analyses that seemed to show a link, the twin study that found none, the problems with the most cited large-sample study, why people see intelligence in attractive faces, and the confounds that can create a correlation where none exists.

Psychologists have tested the idea for almost a century, and the early evidence looked mildly positive. The most cited review is the 2000 meta-analysis by Judith Langlois and colleagues in Psychological Bulletin, which reported 11 separate meta-analyses on how attractiveness relates to the way people are judged and treated and to the way they actually behave.

For actual intelligence, Langlois found two different answers depending on age:

• Children: attractive children scored higher on measures of "intelligence and performance" (14 samples, 3,043 children), with an average effect size of d = .39, roughly a correlation of .19. Many of these measures were grades and school performance rather than IQ tests.

• Adults: across 30 samples (3,853 adults), the effect was d = .07, roughly a correlation of .03. The authors called it a near-zero finding and noted that it matched an earlier meta-analysis by Alan Feingold, which found r = .04.

An effect size ("d") expresses a group difference in standard deviations, the units that also define the IQ scale. A d of .07 would amount to about one IQ point between attractive and unattractive adults if the outcomes had all been IQ tests.

A 1995 meta-analysis by Linda Jackson, John Hunter and Carole Hodge, as summarized by Langlois, found a similar split: a moderate effect in children (d = .41) and an adult effect that fell from d = .24 to .02 once selected studies were removed. Both reviews mixed IQ tests with grades, salary and occupational prestige, which are shaped by many things besides intelligence.

The strongest test so far is a 2015 study by Dorian Mitchem, Brendan Zietsch, Matthew Keller and colleagues in Evolution and Human Behavior. They combined two samples of identical and fraternal twins and their siblings, 1,753 people in total, from Colorado and from Brisbane, Australia. IQ came from standard individual or group tests (the WAIS-III in Colorado, the Multidimensional Aptitude Battery in Brisbane). Attractiveness came from standardized photographs rated by panels of volunteer judges who knew nothing about the participants, and agreement between judges was high (Cronbach's alpha of .86 to .91).

The result was a standardized coefficient of .018 (p = .50) between facial attractiveness and IQ. Within individuals, the correlation in the combined sample was .03, with a 95% confidence interval running from -.02 to .08. Neither the genetic nor the environmental correlation between the traits was statistically significant, there was no difference between men and women, and a test for a curved relationship among less attractive people also came up empty.

The authors also tabulated 41 earlier correlations. The median was r = .09, most studies were small (median sample of 83), and smaller studies tended to report larger effects. That pattern is what you would expect if small studies with striking results were more likely to be published, so the authors suggested that the older literature had overestimated the link.

A smaller Czech study of 40 men and 40 women by Karel Kleisner and colleagues, published in 2014 in PLOS ONE, reached the same conclusion: no relation between measured IQ and rated attractiveness in either sex.

The most widely publicized claim that beautiful people are smarter comes from the evolutionary psychologist Satoshi Kanazawa. A 2004 paper with Jody Kovar argued that, given assortative mating (people pairing with partners similar to themselves in desirable traits) and the heritability of both intelligence and beauty, it follows logically that more intelligent people are more beautiful. His 2011 paper in Intelligence then reported positive associations in two large samples of children and young adults, which were reduced but not eliminated by controls for parental education and income, birth weight, age at puberty and health.

These claims have drawn sustained criticism:

• The logic: the economist Kevin Denny replied in 2008 that the 2004 "theorem" was false and that the evidence did not support it. In a large sample of school children, he found that low intelligence might predict low attractiveness, but that for much of the intelligence distribution there was no significant relationship.

• The ratings: Mitchem and colleagues warned that the 2011 study "must be interpreted with extreme caution" because attractiveness was rated by schoolteachers in one sample and by interviewers who had just measured the participant's intelligence in the other. Raters who know how bright someone is may let that color their judgment of looks, which would manufacture a correlation.

Large samples do not fix a measurement problem. Kanazawa's attractiveness findings have not held up in studies where independent judges rate photographs, such as the twin study above.

The perception is very real even though the underlying correlation is not. Edward Thorndike described the "halo effect" in 1920, the tendency for a rater's overall impression of a person to spill over into judgments of their specific traits. In 1972, Karen Dion, Ellen Berscheid and Elaine Walster showed that people assume attractive strangers have more socially desirable personalities and will lead more successful lives, which they called the "what is beautiful is good" stereotype.

Langlois's meta-analysis found that the stereotype persists even among people who know the person. Attractive children were judged far more academically competent (d = 1.10), and attractive adults were judged more occupationally competent (d = .96).

A 2016 study by Sean Talamas, Kenneth Mavor and David Perrett, published in PLOS ONE, shows the size of the distortion. Separate groups of raters judged 100 photographs of university students on attractiveness, intelligence, conscientiousness and academic performance, and the researchers compared those ratings with the students' actual grades. Attractiveness correlated .81 with perceived intelligence and .03 with actual academic performance. Once attractiveness was controlled, perceived conscientiousness predicted real grades modestly (r = .22), while perceived intelligence did not (r = .06). Judgments became more accurate once the influence of attractiveness was removed, which means the halo was hiding what little real information the faces carried.

Leslie Zebrowitz and colleagues reported in 2002 that people could judge intelligence from faces slightly better than chance at some ages, and that relying on attractiveness as a cue explained much of that accuracy. They found several possible routes linking appearance and IQ, from biology and environment to the influence of each trait on the other, and concluded that neither "good genes" nor self-fulfilling prophecy alone could explain the correlations they saw.

Even a small observed correlation can come from something other than a direct connection between looks and brains.

• Socioeconomic status: family income and education affect nutrition, health, dental care and grooming, and they also relate to IQ scores. Our article on whether IQ reflects socioeconomic status covers how much of that overlap is real. Mitchem's team adjusted for SES, body mass index, acne and age and still found nothing.

• Health and development: poor health or early developmental problems can affect both appearance and cognitive scores. That is one possible reason why Denny, and Zebrowitz and Gillian Rhodes in a 2004 study, found a link mainly at the low end of the distribution.

• Rater knowledge: when the same person rates looks and knows about ability, as with teachers or interviewers, the halo effect can flow in either direction.

• Outcome choice: grades, salary and occupational prestige reward attractiveness directly through how people are treated, so they are poor stand-ins for intelligence.

Height offers a useful contrast. It does show a small, consistent correlation with IQ, which our article on IQ and height describes, and facial attractiveness does not show even that when it is measured carefully.

Not to any meaningful degree in adults. The best twin study found a correlation of about .03, and earlier meta-analyses of adults also found effects near zero.

The halo effect leads people to assume that someone with one desirable trait has others. In one study, attractiveness ratings correlated .81 with ratings of how intelligent the same faces looked.

Only weakly, if at all. Some studies find accuracy slightly above chance for some ages or for men only, and much of that accuracy may come from attractiveness itself rather than from facial shape.

His large-sample findings are disputed because the people who rated attractiveness often knew about the participants' abilities. Studies using independent judges and photographs have not found the link he reported.

Attractive people tend to be judged and treated more favorably, which can affect grades, hiring and pay. Those outcomes reflect treatment as well as ability, so they cannot show that attractive people are more intelligent.

Looks and intelligence are close to independent in adults, and the apparent link in older studies shrinks once attractiveness and IQ are measured separately and small studies are set aside. What remains is a strong bias in how people judge each other. If you want an estimate of your reasoning ability that no one's impression of your face can color, take a full-length online IQ test that scores your answers against a norm group.

1. Denny, K. (2008). Beauty and intelligence may - or may not - be related. Intelligence, 36(6), 616-618. [doi.org/10.1016/j.intell.2008.01.003](. doi.org

2. Dion, K., Berscheid, E., & Walster, E. (1972). What is beautiful is good. Journal of Personality and Social Psychology, 24(3), 285-290. [doi.org/10.1037/h0033731](. doi.org

3. Feingold, A. (1992). Good-looking people are not what we think. Psychological Bulletin, 111(2), 304-341. [doi.org/10.1037/0033-2909.111.2.304](. doi.org

4. Jackson, L. A., Hunter, J. E., & Hodge, C. N. (1995). Physical attractiveness and intellectual competence: A meta-analytic review. Social Psychology Quarterly, 58(2), 108-122. [doi.org/10.2307/2787149](. doi.org

5. Kanazawa, S. (2011). Intelligence and physical attractiveness. Intelligence, 39(1), 7-14. [doi.org/10.1016/j.intell.2010.11.003](. doi.org

6. Kanazawa, S., & Kovar, J. (2004). Why beautiful people are more intelligent. Intelligence, 32(3), 227-243. [doi.org/10.1016/j.intell.2004.03.003](. doi.org

7. Kleisner, K., Chvátalová, V., & Flegr, J. (2014). Perceived intelligence is associated with measured intelligence in men but not women. PLOS ONE, 9(3), e81237. [doi.org/10.1371/journal.pone.0081237](. doi.org

8. Langlois, J. H., Kalakanis, L., Rubenstein, A. J., Larson, A., Hallam, M., & Smoot, M. (2000). Maxims or myths of beauty? A meta-analytic and theoretical review. Psychological Bulletin, 126(3), 390-423. [doi.org/10.1037/0033-2909.126.3.390](. doi.org

9. Mitchem, D. G., Zietsch, B. P., Wright, M. J., Martin, N. G., Hewitt, J. K., & Keller, M. C. (2015). No relationship between intelligence and facial attractiveness in a large, genetically informative sample. Evolution and Human Behavior, 36(3), 240-247. [doi.org/10.1016/j.evolhumbehav.2014.11.009](. doi.org

10. Talamas, S. N., Mavor, K. I., & Perrett, D. I. (2016). Blinded by beauty: Attractiveness bias and accurate perceptions of academic performance. PLOS ONE, 11(2), e0148284. [doi.org/10.1371/journal.pone.0148284](. doi.org

11. Thorndike, E. L. (1920). A constant error in psychological ratings. Journal of Applied Psychology, 4(1), 25-29. [doi.org/10.1037/h0071663](. doi.org

12. Zebrowitz, L. A., Hall, J. A., Murphy, N. A., & Rhodes, G. (2002). Looking smart and looking good: Facial cues to intelligence and their origins. Personality and Social Psychology Bulletin, 28(2), 238-249. [doi.org/10.1177/0146167202282009](. doi.org

13. Zebrowitz, L. A., & Rhodes, G. (2004). Sensitivity to "bad genes" and the anomalous face overgeneralization effect: Cue validity, cue utilization, and accuracy in judging intelligence and health. Journal of Nonverbal Behavior, 28(3), 167-185. [doi.org/10.1023/B:JONB.0000039648.30935.1b](. doi.org

Hero image: Peacock colours, by Bernard Spragg. NZ, CC0. Via Wikimedia Commons.