Abstract:
We develop an image-driven approach to the question of what makes the shape of a woman's body attractive. We constructed a set of 625 images of female bodies by factorially recombining four independent descriptors of shape derived from a principal components analysis of the variation in natural body shape, and had observers rate these images for attractiveness. We then modelled observers' attractiveness ratings with polynomial multiple regression, using the same shape descriptors as explanatory variables. The resulting model agrees well with existing models based on simple anthropometric indices of shape; however, some interesting new findings emerge. There was considerable variation in the shape of bodies that were judged to be equally attractive. Further experiments confirmed that observers could detect these subtle variations in shape suggesting a dissociation between attractiveness judgement and shape discrimination.

Excepts

Introduction:
What makes a woman's body shape attractive? So far, the majority of attempts to answer this question have focused on exploring the impact of variation in a few simple anthropometric measures of body shape on judgements of attractiveness. Although a clear picture has emerged from these studies, we argue here that the approach has some important limitations, and that a parallel image-driven approach to attractiveness based on an analysis of naturally occurring variation in body shape could yield new insights into what constitutes an attractive body. The use of simple anthropometric measures of body shape in studies of attractiveness is popular, because they provide a crude means of quantifying important variation in body shape, and are easy to obtain for large samples of women. Two of the most commonly used measures are body mass index (BMI), which is calculated as the ratio of a woman's weight (kg) to her height (m) squared, and is a measure of overall fatness, and waist-to-hip ratio (WHR), which is calculated as the ratio of the circumference of the waist to the circumference of the hips, and is a measure of fat distribution or curvaceousness. Many studies have confirmed that for optimum attractiveness a female body should be slim, but not skinny, with a BMI of approximately 19 - 20, and have a small waist relative to the hips with a WHR of approximately 0.7.1999).
  • • •  
Between them BMI and WHR explain the majority of the variance in attractiveness. For example a model by Tovee based on these two measures that explains 76% of the total variance in attractiveness judgements. Theoretical support for BMI and WHR as potential cues to attractiveness has come from the evolutionary psychology literature. Evolutionary theory predicts that males should have been selected to prefer as their mates females with the maximum reproductive potential, and that females should therefore have been selected to honestly signal their reproductive potential to males.
  • • •  
However, despite the appeal of the above story, we believe that the approach of focusing on simple anthropometric measures of body shape has a number of potential limitations associated with it that need to be recognized and addressed if further advances are to be made in the science of body attractiveness.
  • • •  
We build a mathematical description of body shape that captures the more subtle variation in body shape that is missed by the crude anthropometric indices used to date.
  • • •  
In this paper we use the four principal components derived by Tovee et al. in order to construct a new set of stimulus bodies. This is possible, because the principal components are statistically independent shape descriptors that can be linearly recombined to create new shapes, each of which is uniquely specified by its particular combination of principal components.
  • • •  
we estimated the BMI of all 625 images in the stimulus set by measuring BMI_PAR, i.e., the area of the body divided by the perimeter length around the edge of the body. BMI_PAR has been shown to be an accurate proxy for BMI.
  • • •  

The best-fit athropometric model of attractiveness is given by the equation:



Discussion:
  • • •  
Our approach was motivated by three limitations inherent in existing anthropometric models of attractiveness. First, our image-driven model solves the problem of correlated explanatory variables by using statistical descriptors of shape that are by definition independent of one another. Second, in modelling the shape of the entire torso our approach has highlighted subtle variations in body shape missed by the simpler anthropometric models. Finally, our model allows for the first time the possibility of reverse-engineering a body with a given attractiveness rating.
  • • •  
This study has explored variation in torso shape and not variability in leg shape. This is because previous studies have not found the legs to be a significant predictor of attractiveness.   • • •   As leg width, shape, and length do not seem to be significant predictors of attractiveness judgements, we concentrated on torso variability in this study.
  • • •  
Perhaps the most interesting aspect of this study is the fact that there is significant variation in the body shapes that are rated as equally attractive.   • • •   Instead, the results suggest that there is a dissociation between preference for body shape (i.e., what observers rated as attractive) and the ability to discriminate between subtle differences in shape. Even within a subset of bodies, which had all been rated as being of the same level of attractiveness, observers were able to reliably discriminate between them on the basis of their shape.   • • •   What might this dissociation between attractiveness preference and discrimination mean? When observers are in attractiveness preference mode, one possibility is simply that "attractiveness space" has multiple maxima, i.e., different configurations of physical features can produce the same level of attractiveness. This suggests that one or more tradeoffs might exist for different body size and shape cues. For example, a body with a higher BMI may be compensated for by a more curvaceous WHR and WCR, and this configuration might be rated as attractive as a body with a more attractive BMI and less curvaceous WHR and WCR. Thus, there may be many routes to being assigned a particular attractiveness level, a theoretical point that is well accepted in the mate choice literature.   • • •   For example, women have been shown to trade off creative intelligence against wealth when choosing potential partners.   • • •