Cognitive Daily

Click on the image below to be taken to a quicktime movie showing 9 different faces. When the movie is finished playing, drag the slider back and forth to pick the face you think is the most attractive.


The faces are composite images—”average” faces made by morphing together 48 different photos. Previous research has shown that people typically perceive average faces as more beautiful than unusual faces (and here we’ve written about how easy it is to change our conception of “average”). But what about people from different racial groups? Would a Caucasian perceive an “average” South Asian composite face as more or less beautiful than a composite from their own race? The example movie above was made from photos used by Gillian Rhodes and a team of colleagues to try to answer that same question. The first frame in the movie is a “Super-Caucasian” created by exaggerating the features that distinguish Caucasian men from Japanese men by 50 percent. The Caucasian-ness is gradually diminished—the middle frame in the movie a 50 percent Caucasian-Japanese blend, and the final frame is a 50 percent “Super-Japanese.” One quarter of the way through the movie, you see a 100-percent Caucasian composite, and three-quarters of the way through, a 100-percent Japanese composite.

You might expect that Caucasians would prefer the average Caucasian face and Japanese would prefer the Japanese face. But the results found by Rhodes’ team were rather different. They presented cards with each of these images (sorted in a random order) to Caucasian college students. They asked the participants to select the the most attractive card from the stack and rate it for attractiveness on a scale of 1 to 10. The process was repeated until all cards were rated. Here’s a summary of the results:


While there was a trend, especially among females, to rate the Caucasian male faces as more attractive, the face rated most attractive of all was the combined average of both Caucasian and Japanese faces. When participants rated female faces, the average faces were favored to an even larger degree. What’s going on here? Why should we prefer combined races over those more similar to us?

Rhodes and her team speculated that the preference for composite race faces may be related to health. People of mixed-race ancestry do appear to have a larger variety of genes, and on the the other end of the scale, when close relatives have children together, they are susceptible to a variety of ailments. In a separate experiment, the team examined not only composite faces, but also faces of Eurasian people (adult children of one Asian parent and one Caucasian parent). Again, participants were asked to rate the faces for attractiveness, but they also rated the faces for how “healthy” they appeared. Here are the results for male faces.


The Eurasian faces achieved a whopping advantage over both Caucasian and Asian faces, for both health and attractiveness ratings. Again, similar results were found for female faces. Rhodes et al. argue that this supports their health hypothesis—that humans select mates based on their perceived health level. However, they are also careful to note some potential problems with their research. By using composite faces, they show us faces no real human actually possesses (a separate portion of their experiment used individual faces, and similar results were found, but individual faces offer problems of their own, because participants may be judging other factors such as perceived wealth). Cultural differences such as a tendency not to smile for photos might also play a role. And it’s entirely possible that “health” ratings may simply be an artifact of “attractiveness” ratings—wishful thinking that a pretty face is also a healthy one.

All those caveats aside, the basic finding—that we tend to find both mixed-race composites and actual people of mixed race more attractive—is surprising and interesting on its own.

Did your own result from the quicktime example above match the research findings? Let us know in the comments.

Rhodes, G., Lee, K., Palermo, R., Weiss, M., Yoshikawa, S., Clissa, P., Williams, T., Peters, M., Winkler, C., & Jeffery, L. (2005). Attractiveness of own-race, other-race, and mixed-race faces. Perception, 34, 319-340.


  1. #1 memer
    September 13, 2005

    Curious. Am I drawn to some (say) sculpture because its beauty implies its creator is in great health? Hm. This is an uncomfortably limited, one-off experiment. Does this experiment actually “prove” anything? I’m just a poor layman like you and I believe I see flaws in the conclusions (nevermind the experiment) already. I’ll bet Google could find more. Oh well, when you’re a prof under the gun to produce papers… you need a thesis, you need a thesis. Dave, is it possible your wife (the pro) can give us more well-rounded critiques?

  2. #2 Dave Munger
    September 13, 2005

    Memer, I’m not quite sure what your question is. The study of aesthetics and appreciation of art is an entirely separate field from studying the factors influencing attraction between humans. If you would like to try to clarify your specific criticisms of this article, we’ll be happy to respond.

    You should take care when searching Google for articles on cognitive psychology. Most peer reviewed journal articles such as the work we reference here are not available on the free Internet.

    I should perhaps add that Greta and I consult on every article in Cognitive Daily. The actual name attached to a particular post often reflects no more than the person who happened to be logged in at the time of writing.

  3. #3 Dana Leighton
    September 13, 2005

    Hi Dave,

    As a psychologist, it is always humbling and comforting that I am subject to the same forces I study and teach about. Yes, interestinglt, I did pick the middle face.

    The relative genetic health of non-homogenous ancestors (as exhibited by facial features) as a moderator of attractiveness is an interesting hypothesis. In my lab at UBC, we looked at factors such as this while using evolutionary theory to derive testable hypotheses. We found that certain people will use a variety of health-related indicators when making decisions about others (and when forming stereotypes and prejudiced behaviors). One such individual difference factor that seems to influence people to use health related indicators is called “Belief In A Dangerous World” (Altemeyer, 1988). For example, people who score highly on BDW also score highly on Anti-Fat Prejudice. Interesting, huh?

    In reply to memer’s comment, experiments rarely actually “prove” anything. Experimental data are gathered to support (or refute) hypotheses derived from theoretical ideas. Only after a long process of hypothesis testing can we ever gather enough evidence in support of a theory to say that it is supported by the evidence. We still never say we have “proved” a theory.

    Especially in psychology, establishing cause-and-effect relationships is a very tricky proposition – human behavior is a moving target, influenced by culture, history, and biological forces. What might be a well-supported theory this century may be quite different in 2 or 3 centuries hence.

    To laypersons, this ambiguity is unbearably frustrating. To psychologists, it’s an irresistable challenge. :^)

    Altemeyer, B. (1988). _Enemies of freedom: Understanding right-wing authoritarianism_. San Francisco, CA: Jossey-Bass/Pfeiffer.

  4. #4 Jon Slinger
    September 14, 2005


    I think the whitey male in the beginning is the pertiest of them all.

  5. #5 memer
    September 14, 2005

    I was theees close to becoming a Cognitive Science major, but instead turned to History and Philosophy of Science instead. I know about the revolutions of scientific paradigms.

    I also know that even Relativity isn’t so much “proven” as a reliable, testable theory. It’s all a work in progress. Hence my quotation marks. But there’s a difference between rank conjecture and a theory that’s been put through the wringer and still stands unbowed (til there comes new data it cannot explain).

    Psychology and Cog Sci in particular has always been labelled one of those “soft” sciences, something that the behavioural dudes have tried hard to shed. Experiments like these are probably why. It is a curious thing, I suppose. I suppose. But it’s the wispy conclusion drawn from it that rankles.

    I suggest that this only carries weight when it’s been repeated with larger, more varied audiences and cross-referenced against competely different experiments that try to

    Right now, it isn’t any more interesting than a logic test manages to trick even trained logicians (and psychologists).

    I rate it (FWIW): meh.

  6. #6 memer
    September 14, 2005

    Ok, next time I proof what I write before hitting “submit.” Whatever. Next time, fellers.

  7. #7 Phil
    September 14, 2005

    1 less than super Japanese, and I’m white caucasian male.
    Mind you, I watch a lot of samurai films so maybe there’s less of the unusualness factor for me when I did it.

  8. #8 Carl Manaster
    September 15, 2005

    What struck me about the first chart is the gender difference. Was it also present when female faces were used, and, if so, in which direction was it split?

  9. #9 Dana Leighton
    September 16, 2005

    Phil, exposure to particular faces does affect the results – see Rhodes, Jeffrey, et al. below.

    Carl, I think the effect might not carry for female faces, as you’ll see in a couple of the citations I give below. You’d need to look up the whole literature to know. I just did a brief lit search.

    Memer wrote:
    “I suggest that this only carries weight when it’s been repeated with larger, more varied audiences and cross-referenced against competely different experiments”

    Actually, this experiment is one in a long chain of such experiments. I did a brief literature search on this, and found a few citations to get you started. You should be able to find these at a university library if you have one near you.

    Rhodes, G., Hickford, C.,Jeffery, L. (2000). Sex-typicality and attractiveness: Are supermale and superfemale faces super-attractive? _British Journal of Psychology, 91_(1), 125-141. (result: male average faces preferred over supermales; no effect for females)

    Fessler, D. M. T., et al. (2005). Cross-Cultural Investigation of the Role of Foot Size in Physical Attractiveness. _Archives of Sexual Behavior, 34_(3), 267-276. (result: small foot size preferred for females; average size for males)

    Rhodes, G., Halberstadt, J., Jeffery, L., Palermo, R. The attrativeness of average faces is not a generalized mere exposure effect. _Social Cognition, 23_(3), 205-217.

    Rhodes, G., Jeffery, L., Watson, T., Clifford, C., Nakayama, K. (2003). Face attractiveness aftereffects: Fitting the mind to the world. _Australian journal of psychology, 55_, 89-89. (result: experience with malformed images made them seem more average).

    Alley, T. R. & Cunningham M. R. (1991). Averaged faces are attractive, but very attractive faces are not average. _Psychological Science, 2_(2), 123-125.

    Langolis, J. H. & Roggman, L. A. (1990). Attractive faces are only average. _Psychological Science, 1(2), 115-121.


  10. #10 Lisa Boucher
    September 18, 2005

    Interesting article. For the record, I chose the third from last picture, and I’m half Asian and half Caucasian.

  11. #11 anon
    September 18, 2005

    Wow. I chose the middle face also. That was humbling and enlightening. thank you.

  12. #12 Tobia
    September 22, 2005

    I suspect that the “average” ratings come out first, because they represent a greater degree of symmetry.
    This is proven in a similar experiment to be the case, because mates prefer the most symmetrical facial features, and when a face is averaged, any assymetry is automaticcaly averaged out, thus producing a face with a higher degree of symmetrical uniformity, which rates high on our biological barometer.
    I think this test is confounding “mate attractivness”, with the biological search for beauty/symmetry.

  13. #13 Dave Munger
    September 22, 2005


    All the faces in the first experiment are averages: what we’re talking about is a composite Caucasian face, and a composite Japanese face. So I’m not sure symmetry can really be an issue in this case.

    Also, in a separate experiment, the researchers studied individual faces—actual Eurasian, Caucasian, and Japanese faces. Again, presumably, since these are not averages, but actual individual faces, the relative levels of symmetry should be about the same—but still, the Eurasian faces were preferred.

  14. #14 Jason Malloy
    September 24, 2005

    We also blogged on this paper in June:

    There is an uploaded PDF there for those who would like to read it.

  15. #15 Tara
    October 1, 2005

    I also chose the middle face (the fifth one). I have a Korean cousin who was adopted as a baby by my Caucasian aunt and uncle. I wonder if a situation like that would affect one’s choice of faces. Has this experiment also been done with composites of people from ethnic backgrounds other than Asian/Caucasian?

  16. #16 Samantha
    October 20, 2005

    Super Japanese… Super Caucasian… hahaha. It looks a bit ridiculous when you first see it, but this is interesting.

  17. #17 Wesley McArthur
    October 26, 2005

    Firstly, interesting site that I discovered entirely by accident, through a link attached to a rave music forum (we were talking about racial backgrounds and someone had included a link to this site).

    Anyway, I don’t have any background in psychology at all, but it still makes interesting reading.

    From the results derived from the data, one hypothesis given is that the ‘combined race’ (CR) face is preferred, as it represents (or is perceived) as a more ‘healthy’ face, and that perceived good health in a potential partner is often the major factor determining their attractiveness to another individual.

    However surely if the ‘super’ faces (i.e. those composite faces deemed to show no ethnic mixing at all), show no signs of ill health (assuming that most social and ethnic groups use the same physical traits as indicators of ill-health such as obesity, assymetry (sic?) etc. etc.) then one cannot say that the CR faces were chosen on this basis alone?

    Possibly there are other (sociologically based) reasons as well. I think it’s possible that the results of the tests show that some individuals are attracted to someone who is slightly out of their ‘ethnic context’, if you will.

    It’s possible that the results show that some individuals express boredom with the ‘usual’ and attraction to the ‘exotic’. Often (though I have no proof for this conjecture, only anecdotal evidence) people find something a little removed from ‘the ordinary’ (in this case our ‘Super Race’ face) more attractive to that which they are typically exposed, in terms of potential partners.

    However it often seems to be the case that they are less likely to be attracted to someone who is completely removed from their ethnic background, hence the low ‘scores’ of the ‘Super’ Japanese face to Caucasians, but the high score for the CR face. The CR face is exotic, but not too far removed from the individual viewing it.

    This would not explain those individuals who are only or primarily attracted to members of other ethnic groups. In such cases there would possibly be other mitigating social factors, such as high exposure to other racial groups, some form of conditioning, or even the old chestnut of rebellion to the norm(!) or even a case such as Phil, who is Caucasion and preferred the ‘super’ Japanese face, but also stated that he watches a high quantity of ‘Samurai’ films).

    It would also be interesting to see results from individuals of mixed race, and / or individuals who live in communities with a high mix of racial intergration, and also ‘contrary’ or cross tabulated test groups (for example results from Black individuals viewing the ‘Super Caucasion to Super Japanese’ faces, etc etc.)

    Anyway, just some food for thought. Apologies for bad grammer and / or spelling.

  18. #18 Goon
    November 7, 2005

    Here are my observations on race and attractiveness, having lived in the San Francisco and Berkeley for 8 year and being a gay Asian myself:

    White Men like Asian Women
    White Women don’t like Asian Men
    Asian Women like White Men more than they like Asian Men
    Asian Men regardless who they like tend to be seen with Asian Women

    Gay Asian Men like White Men much much more than they like other races
    Gay White Men like other White Men more than they like other races

  19. #19 sol_grl
    December 1, 2005

    I think a large part of attractiveness in racial sense has been been built in a socioconstructiveness approach. Surely if All the people find Caucasians or Asians as most attractive, the other races would more or less dwindle in thier offsprings. Education might be one thing resulting in China having the largest population in the world, but if biology plays the larger role in species continuation, then I don’t think races other than caucasian would had survived.

    I won’t biasedly believe that Caucasians are more attractive race and hence superiority although I am well aware that any race would want this advantage.
    Reasons for attractiveness may be varied, perhaps asians were not taught to express their strengths as strongly as caucasian families did and hence in appearance, caucasians are perceived to be more sexy and sociable because they have had more experiences with socializing and are more familiar with the game.
    Be aware that a lot of studies are done in the western culture and interpreted under the light of westernizsed values. Some values are appreciated in a heightened manner and others are in relation, diminished.

  20. #20 Jules
    December 25, 2005

    WOW! This is indeed a breakthrough! I chose the middle one, presumably the Eurasian face due to the factors you said “healthy looking and nicer smiles” (but you didn’d say that did you)
    WOW again!

  21. #21 someoneelse
    March 2, 2008

    Well this was a very interesting experiment. The only downside is, that I feel very depressed now. Among other issues, I do not have a very symetrical face. I therefor think, that my genes are probably not worth passing on.
    I will be perceived as unattractive (or unhealthy as we know now) in a split of a second by every potential mate.
    I really do suffer thinking about this hopeless situation. I cannot turn of the “programs” that force me to try/want to be attractive in order to find a mate to reproduce, therefore this suffering will endlessly continue. With no hope. WOW.

    Well there might be one hope: that a healthy mate emerged from evolution with a broken attractiveness measuring unit, that – unaware of my unhealthiness – elects me for reproduction.
    Immediatly the ethical question arises wheter I should try to be utilitarian or egoistic: should I deny executing necessary routines in order to prevent reproduction and spreading of our faulty genes?

    Should I elucidate the errornous mate about his/her failure in realising my unhealthiness?

    Sorry to write this to you, it’s not your fault, you are just the messengers, but I would really like to read your oppinion…

  22. #22 MattyDee
    August 21, 2009

    I’m thinking that there is a social desirability effect here, meaning that people don’t want to seem (to others and themselves) to be biased towards finding Caucasian faces more attractive, therefore are more likely to vote for the composite face as most attractive.

    Then again, women seemed skewed towards the Caucasian face, so who knows what exactly is going on.

    1. Perhaps men are more able to be socially desirable meaning that they can be emotional detached when rating male faces.

    2. Part of what goes into the heuristics for male attraction in women is dominance and social status. In a Caucasian dominated culture, this will advantage white males obviously. So maybe this is part of the reason we see female ratings lean toward the Caucasian side.

    Very interesting.

  23. #23 MattyDee
    August 21, 2009

    In point 1, I mean that IN THIS PARTICULAR CASE (rating the attractiveness of other males), men are more objective. IT IS NOT a general statement comparing men to women.

  24. #24 MattyDee
    August 21, 2009

    Also, in the second chart, I assume that 50c/50a stands for 50-50 Asian/Caucasian mix. So what’s the difference between this and Eurasian?


  25. #25 Damian
    September 26, 2009

    I think skin color plays a role in these results. The middle pictures (European-Asian mix) are of people with white skin. The “Super Caucasian” has pink skin which is unlike typical caucasian skin color. I think the middle pictures are closer to caucasian skin color and might be the reason why these pictures scored the highest.

New comments have been disabled.