During a long-term research project at the Universities of Regensburg and Rostock (Germany), we have tried to find answers to questions like these. We questioned why some faces seem more attractive than others; and we did experiments on social perception, that is: we tried to find out about the social qualities attributed to faces of specific attractiveness.
We conducted seven large sampling surveys (total n=500) in order to test several hypotheses on human facial attractiveness. These hypotheses are the 'attractiveness is averageness' hypothesis (Langlois & Roggmann, 1990: "average faces are most attractive"), the 'symmetry hypothesis' (Grammer &Thornhill, 1994; Thornhill & Gangestad 1999: "facial symmetry has a positive influence on facial attractiveness ratings") and the theory of 'multidimensional beauty perception' (Cunningham, 1986: "attractive faces show a combination of signs of sexual maturity and babyfaceness"). We furthermore performed experiments on the correlation between attractiveness and attributed social qualities (Dion, Berscheid & Walster, 1972: "what is beautiful is good").
For this purpose, we took standardized digital photographs of 64 female and 32 male faces aged 17-29 years, including eight photo models. In a preliminary test, these faces were randomly presented to test subjects using a self-programmed presentation Software (Authorware 5.0 package). Test subjects rated the attractiveness of the faces on a seven-point Likert scale ranging from 1 (very unattractive) to 7 (very attractive). On the basis of these ratings, the pictures of the faces were ranked for further use according to their average attractiveness values.
Using a morphing software (Morpher 3.0, freeware), new 50:50 percent compound images were generated from each two original faces using a binary tree scheme and following the order of the attractiveness values for the faces. For example, a compound image (w1-2) was generated by combining the least attractive female face (w1) with the second-least attractive female face (w2); in a next step, w3 and w4 were combined to w3-4, and so on. The next generation of pictures was then produced by combining w1-2 and w3-4 to w1-4, then w1-4 with w5-8 to w1-8 and so on, resulting in two single compound images w1-64 for the 64 female and m1-32 for the 32 male faces. For the calculation of each new average face, 500 reference points were defined, resulting in a total of over 75,000 reference points that were defined during the whole study. In this way, we were able to produce pictures of average faces that are almost indistinguishable from 'normal' everyday faces, making the faces created by morphing easily comparable to the original photographs taken. Thus, the applicability of our results is likely to be higher than in previous studies, in which lower-resolution compound images have been produced.
In analogy to the preliminary attractiveness rating of the pictures of original faces, we presented the series of compound images to test subjects, and these were again rated on a 7-point Likert scale. All original and compound faces were furthermore rated by employees of a model agency with regard to their suitability for the model category 'beauty' (third experiment).
In a fourth experiment, we examined the influence of symmetry on the attractiveness of faces. Using the Morpher 3.0 software, we constructed symmetrically optimized images for each of the five most, medium and least attractive original faces of each gender. By morphing the images, rather than just producing mirror images, the symmetrically optimized faces did not look artificially at all. In a paired comparison with test subjects, we tested wether these symmetrically optimized faces were thought to be more attractive than the original images.
For our fifth experiment, we changed the proportions of each three attractive and three unattractive faces in a way so that they resembled 50% of the average face for each gender. Note that surface structures etc. remained unaltered in this experiment, and proportions were the only explanatory variable. Again, a paired comparison with test subjects was conducted, each pair consisting of the original face and its derived image with altered proportions. On the other hand, we showed the test subjects pairs of images that were identical in their proportions but differed in the surface structure of the skin.
Studies on the role of the babyface scheme for human facial attractiveness were conducted in a seventh experimental series. In order to calculate the proportion data for the babyface scheme, we used the morphing software to construct a compound image of four pictures of four to six year-old girls. For six randomly chosen adult original images, we mixed the original faces' proportions in 10% steps with the proportions of the babyface scheme, resulting in a series of images that showed 10%, 20%, 30%, 4-% and 50% babyfaceness. Again, surface morphology of the faces remained unaltered, and only the proportions of the faces were changed. Test subjects chose the most attractive face out of each randomly chosen set of one original face and its five babyface variations.
In order to find out about the relationship between attractiveness and assigned social attributes, 21 of the most, medium and least attractive faces were attributed ten different personality factors on a 7-point Likert scale (seventh experimental series) by test subjects.
The results of our study are quite surprising. Compound (i.e. morphed) faces were, on average, regarded as being more attractive than the original faces (mean attractiveness value 4.3 for female and 4.5 for male faces, respectively). The more original faces one compound face consisted of, the higher its assigned attractiveness value (r=0.57** for female and r=0.64** for male faces). This result confirms on the one hand the 'attractiveness is averageness' hypothesis (Langlois & Roggmann, 1990); on the other hand, the attractiveness of the original faces included in a compound image itself influences the overall attractiveness of the compound face, i.e. the more attractive the original faces, the more attractive the resulting compound face (r=0.75** for female and r=0.68** for male faces). Thus, not simply the number, but also the attractiveness of the original faces influences the average attractiveness rating of compound faces. This result is in contrast to the 'attractiveness is averageness' hypothesis (which states that average faces are always most attractive regardless to their origin). A big surprise is furthermore, that the attractiveness values for compound male faces is higher than for the original ones - a fact that is in sharp contrast to previous studies in which no rise of attractiveness values in morphed male faces could be found. This may be partly due to the low resolution of the images used in these studies.
The questionnaire in the model agency showed that 88% of n=16 faces that had been selected for the 'beauty' category (out of 64 faces presented) had been generated using the morphing software, which means that 14 out of 16 faces chosen by the employees of the model agency did not exist in reality.
The results of our symmetry experiments show a clear but only weak relationship between facial attractiveness and symmetry: very asymmetrical faces are rated unattractive, but unattractive faces don't need to be asymmetrical. Vice versa, very symmetrical faces don't need to be very attractive, and very attractive faces may show remarkable deviations from ideally symmetrical proportions. In summary, symmetry seems only to be a weak factor to explain facial attractiveness.
Rather, it seems that not the proportions but the surface characteristics (e.g. skin texture) of a face decide wether it is regarded attractive or not. In terms of facial proportions, this means that the 'attractiveness is averageness' hypothesis is clearly falsified.
For female faces, it could be shown that babyface attributes - such as large, round eyes, a large domed forehead and small, short nose and chin lead to a rise in attractiveness values. Only very few (9.5%) of the test subjects found the original adult faces most attractive. Most of the test subjects (90.5%) preferred faces with 10%-50% the proportions of the babyface scheme. This means: Even the most attractive female faces can become more attractive when their proportions are altered towards more babyfaceness. It needs to be explicitly stated, however, that not only male, but also female test subjects found babyface pictures more attractive, and we could not observe any inherent preference of babyfaced pictures in our male test subjects. Again, it is surprising that the most attractive faces do not even exist in reality.
By calculating prototypic very attractive vs. unattractive faces for each gender, we were able to show that these faces are remarkably different in their attributes, such as skin texture, proportions etc. Additional surveys showed that attractive female faces are narrower than unattractive ones, and that they possess a brown skin and full, well looked-after lips. The distance between the eyes is larger, eyelids are thinner, there are more, longer and darker eyelashes, darker and narrower eyebrows, higher cheekbones, and the nose is narrower than in less attractive female faces. Surprisingly, more or less the same is the case for attractive male faces: they, too, have a browner skin, a narrower face, fuller lips, thinner eyelids, more and darker eyelashes, darker eyebrows, and higher cheekbones than the less attractive ones. Attractive male faces can furthermore be characterized by a more prominent lower jaw and chin.
Finally, the results of our studies on social perception suggest that there is a well-defined stereotype of attractiveness: People with more attractive faces were assessed to be more successful, contended, pleasant, intelligent, sociable, exciting, creative and diligent than people with less attractive faces. These results particularly show the far-reaching social consequences human facial attractiveness may have. In order to illustrate this, we constructed three-dimensional animated avatars (head models) using original faces that had been given extreme values like "very unintelligent" or "very successful" in previous ratings.
To sum up, our study shows clearly that the most attractive faces do not exist in reality, they are morphs, i.e. computer-created compound images you would never find in everyday live. These virtual faces showed characteristics that are unreachable for average human beings.
Despite this fact, people living in modern post-industrial societies
are exposed to these kinds of artificially created and manipulated, 'perfect'
faces every day, e.g. via TV advertising or fashion magazines. The result
may be that we all may become victims of our self-created, completely unrealistic
ideal of beauty.