Cognitive Daily

What makes a beautiful body?

How do we know when we see a beautiful body? Is it some social standard such as thinness or proportion? Do we simply think that bodies that are closest to “normal” are also most beautiful? We know that to be the case with faces, where faces that closest to “average” are generally rated as more attractive than others which deviate, and faces that combine the characteristics of several races are rated as more attractive than those typical of a particular race.

We’ve written before on how our perception of faces can be altered. If you look at faces that have been systematically distorted to look bloated or shrunken, eventually you’ll likely believe that those distorted faces are more normal and attractive than undistorted faces. But does the same hold true for body shape?

Christopher Winkler and Gillian Rhodes decided to apply the same methods that Rhodes and her colleagues had used for face perception to perception of bodies. They took 10 photos of identically-clothed women, then systematically stretched and compressed each picture in Photoshop to generate a set of 11 photos ranging from 50 percent wider to 50 percent narrower than the original:


They asked a group of 40 volunteers to rate each of these 110 pictures on a scale of 1 to 9 for attractiveness and normality. Next, the participants were divided into two groups. The first group watched a 5-minute slide show of photos of 10 new women, distorted in the same way as before, with a twist: they were adapted to wide photos, because they were showed only photos ranging from 0 percent to 50 percent wider than normal. The second group was adapted in a similar fashion to narrow photos. Neither group rated these photos. Now both groups repeated the original task of rating the first set of photos, but between each rating, an adaptation (either wide or narrow) photo was shown to maintain the original adaptation for each group.

Ratings for normality showed a similar pattern to what had been found before with faces:


The group that was adapted to wide photos rated bodies that were wider as more normal after the adaptation. The narrow group rated narrower bodies as normal. This makes sense, given the research on faces. The surprising result was for attractiveness:


Both groups rated narrow bodies as most attractive, before and after the adaptation (this is itself a different result from most studies on facial attractiveness). However, while the the narrow group rated even narrower bodies as more attractive after adaptation, the wide group’s attractiveness ratings were statistically indistinguishable, even after adaptation. So after people see a lot of skinny bodies, they begin to perceive skinnier bodies are more attractive, but after viewing a lot of wide bodies, their ratings of attractiveness don’t change.

Winkler and Rhodes suspected that participants might be perceiving the wide bodies as less distorted than the narrow bodies, so they conducted a second experiment. This time, they first had participants rate a much larger set of bodies, ranging from minus 80 percent to plus 80 percent, and they found that bodies 70 percent wider than normal were rated as equally distorted as those 50 percent narrower than normal. They repeated their original experiment, but adapted the “wide” group to these wider photos. Yet the results of this new experiment were the same.

So why don’t perceptions of attractive bodies vary the same way that perceptions of normal bodies do? Winkler and Rhodes are reluctant to suggest that different processes are at work here. They speculate that a more basic perceptual phenomenon may be at work: When we see a lot of shapes that are taller than they are wide, we tend to think that the shapes we view later are wider than they really are. Since even the widest human bodies are still taller than their width, then narrower bodies will look more normal even after viewing wide bodies. That’s why the most normal bodies for the adapted wide group are less different from the original rating than they are for the narrow group.

Regardless of the reason, it seems that our perceptual systems are biased to prefer narrower bodies more readily than wide bodies. We don’t know how this bias interacts with the dominant women’s form on TV–that of the hyper-thin model–but it does suggest that simply having more variety won’t immediately shift an individual’s perception of beauty.

Winkler, C., & Rhodes, G. (2005). Perceptual adaptation affects attractiveness of female bodies. British Journal of Psychology, 96, 141-154.


  1. #1 Gordon Worley
    December 27, 2005

    Interesting, although the study doesn’t rule out the possibility that the participants are sufficiently biased in advance to perceive skinny bodies as normal. I think the results are more a consequence of the participant’s population than anything else. If the study were done with a wider variety of people for many different populations, we’d see the expected results.

  2. #2 Bob
    January 7, 2006

    In the second graph, why is there a difference in attractiveness ratings in the before adaptation conditions? Were people assigned to the groups randomly? Or did their before ratings determine which group they were assigned to?

    Note that in the first graph on normality ratings, the ratings in the before adaptation conditions are very similar.

    The differences in the second graph before adaptation ratings appear to be statistically significant.

  3. #3 Dave Munger
    January 7, 2006


    The researchers didn’t report on whether the groups’ attractive ratings differed significantly before the adaptation. I didn’t include confidence intervals on these graphs, but looking back at the original chart in the report, it looks like it’s a close call. The difference between the narrow group’s before and after ratings is much larger, and clearly significant, even looking only at the graphs.

    The groups were indeed assigned randomly, so that doesn’t explain the difference.

    It’s important to note that the before and after-adaptation groups were the same individuals, whereas the the narrow and wide groups (and attractive and normal groups) were composed of different individuals, so the difference between the groups may be explained by individual preference. The wide group did include proportionally more men (55 percent versus 40 percent in the narrow group), so it’s possible that this might have something to do with the trend you noticed, though Winkler and Rhodes did not report any gender differences in the study.

  4. #4 Jodi
    January 9, 2006

    Well, how could one brief study counteract years of training based on current cultural biases?

New comments have been disabled.