Testosterone-dependent secondary sexual characteristics in males may signal immunological competence and are sexually selected for in several species. In humans, oestrogen-dependent characteristics of the female body correlate with health and reproductive fitness and are found attractive. Enhancing the sexual dimorphism of human faces should raise attractiveness by enhancing sex-hormone-related cues to youth and fertility in females, and to dominance and immunocompetence in males. Here we report the results of asking subjects to choose the most attractive faces from continua that enhanced or diminished differences between the average shape of female and male faces. As predicted, subjects preferred feminized to average shapes of a female face. This preference applied across UK and Japanese populations but was stronger for within-population judgements, which indicates that attractiveness cues are learned. Subjects preferred feminized to average or masculinized shapes of a male face. Enhancing masculine facial characteristics increased both perceived dominance and negative attributions (for example, coldness or dishonesty) relevant to relationships and paternal investment. These results indicate a selection pressure that limits sexual dimorphism and encourages neoteny in humans.

No abstract

Caricaturing the distinctive shape of famous faces can produce an advantage in reaction-time paradigms but the role of distinctive colour and intensity information in recognition of facial identity has not previously been explored. A presentation-time paradigm was developed by which stimuli could be presented for a range of brief display periods. Subjects were required to identify photo-realistic colour representations of famous faces which either were veridical, were caricatured in colour space, or had enhanced colour saturation and intensity contrast (as contrast controls). Recognition accuracy was greater when viewing the caricatured stimuli than either the veridical images or the contrast controls. The removal of colour information to produce grey-scale images also decreased accuracy of face recognition. Both results indicate that colour information aids differentiation of a class of natural stimuli with similar configurations. Thus it is demonstrated that caricaturing faces can be extended to the colour domain and, as with shape caricaturing, enhancement of distinctive information can produce a recognition advantage for famous faces.

Previous forays into the study of recognition have revealed an advantage for line-drawn and photographic shape caricatures of faces in reaction-time paradigms. When a presentation-time technique was used, photographs with enhanced colour intensity and saturation were also found to provide superior recognition accuracy to veridical images. This has provided strong evidence that distinctive information can produce a recognition advantage for famous faces in both colour and shape domains. Such a presentation-time paradigm allows the display of stimuli over a range of brief display periods. Using this paradigm, subjects recognised photorealistic target faces caricatured in shape with greater accuracy than veridical images, consistent with previous findings when reaction time was used as a measure. Subjects were also asked to identify the best likeness for individuals using photorealistic stimuli and an interactive paradigm with shape caricature, colour caricature, and contrast control varied by the user in real-time. The best likeness with shape manipulation was a slight anticaricature, while with colour-caricature and contrast-control images a mildly exaggerated image was selected as the best likeness. Thus, although images caricatured substantially in colour or shape (+40%) induce superior recognition compared to veridical images, such substantial exaggerations are not necessarily seen as best likenesses under prolonged exposure.