Figure 1. The male face on the left typifies the characteristics women preferred when judging male faces for a short term relationship. The male face to the right shows what they preferred when looking for a longer term relationship. The leftmost of the female face images characterises the preferences of male participants when judging female faces for short term relationships. Characteristics which males preferred for longer term relationships are shown on the right.This page summarises the use of the Q-cgi technique in a recent Proceedings of the Royal Society B paper. Q-cgi brings together a set of novel statistical and computer graphics techniques and in this case is used to investigate variation in people’s perception of facial attractiveness.
Experiments in psychology and other disciplines are typically aimed at testing a specific idea or hypothesis and this manner of testing has also been highly successful in the investigation of facial attractiveness. Elegantly constructed experiments have shown how factors such as 
facial health, masculinity and symmetry influence our perception of how attractive we find a face. However, although this research has shown the effect of these factors in the highly controlled environment of the lab we do not know how important these factors are for real faces. In this experiment we use an alternative, more descriptive approach which allows us to ask whether there is a difference in how groups of individuals perceive the attractiveness of faces and what this difference might be.
Rather than show participants computer generated faces which vary in just one factor that we may be interested in e.g. masculinity or health, we showed them sets of real faces and investigated how their preferences varied using s combination of Q-sort and CGI techniques (Q-cgi). We first investigated whether people had different preferences for long-term and short-term relationships. Participants were shown 30 images of real opposite sex faces and asked to judge the attractiveness of each face initially for a long and then for a short term relationship. Figure 1 shows what the differences in preference look like for male and female participants. As can be seen from the images, for short-term relationships women prefer more masculine looking males whereas for longer-term relationships they prefer men whose facial features suggest kindness and trustworthiness. For short-term relationships (as compared to longer term relationships), men preferred women who look more feminine and youthful.
Figure 2. The left most face image of each pair typifies the characteristics of the faces that less sexually restricted individuals rated as attractive. The images to the right highlight the characteristics which were preferred by the more sexually restricted participants.The same techniques can be used to investigate differences between participants. We compared participants who reported a more sexually restricted lifestyle to those reporting a less sexually restricted lifestyle. The results are shown in figure 2. The less restricted women preferred male faces which appeared more masculine and healthy. Less restricted men preferred female faces which were perceived as being more feminine, healthier, kinder and more trustworthy than those faces preferred by the more sexually restricted men. The preferences of less sexually restricted individuals are in keeping with the idea that cues to good genes and fertility may be particularly important to participants pursuing a less restricted strategy where men with high levels of testosterone and women with high levels of oestrogen appear more attractive.
For further details of this work are published in Burt et al. (2007).
This method is also currently being used to investigate how people interpret facial expressions.
Burt, D. M., Kentridge, R. W., Good, J. M. M., Perrett, D. I., Tiddeman, B. P., & Boothroyd, L.(in press). Q-cgi: new techniques to assess variation in perception applied to facial attractiveness. Proceedings of the Royal Society of London Series B-Biological Sciences. to be published 12/9/07 doi:10.1098/rspb.2007.1037.