Aesthetic judgments of faces in degraded images

What makes a face beautiful? Facial symmetry and averageness have been proposed as important determinants of attractiveness, and experiments with newborns have investigated the innateness of such preferences. It has been reported that neonates prefer to look at faces judged by adults as more attractive (Walton et al, 1993; Slater et al, 1998). However, considering the limited visual acuity of newborns, it is unclear whether reliable aesthetic judgments can be made given stimuli as degraded as those available to the neonatal perceptual system. Indeed, it has not been established that adults can make reliable aesthetic judgments with similarly degraded visual stimuli. We have addressed this issue by assessing the ability of adults to judge facial attractiveness as a function of image degradation, examining also the relative contribution of featural versus configural information. Subjects were asked to rate the attractiveness of 19 pictures of female Caucasian faces representing a wide range of attractiveness. Likert-scale ratings of the faces, as well as exhaustive pairwise comparisons, were used to establish rank orders. Subjects saw the images at various blur levels, always proceeding from the most to the least blurred conditions. In a second study, only the eyes, nose, and mouth of each face were shown. We evaluated the consistency of the ratings of the blurred faces and of the internal features relative to those of the original images. The results demonstrate the subjects' ability to reliably rate the attractiveness of faces even at a resolution of five cycles across the face, and to provide ratings highly consistent with those by subjects who viewed the corresponding high-resolution images. Our findings suggest that it is possible to make reliable aesthetic judgments of faces with very little image information, and that attractiveness ratings are significantly better correlated with overall facial configuration than with featural details.