Subjective facial attractiveness is correlated with low-level properties of images G Hayn-Leichsenring1, C Menzel1, O Langner2, C Redies1 |
|---|
1Institute of Anatomy I, FSU Jena, Germany |
| Several properties of faces have been proposed to contribute to subjective ratings of attractiveness. In particular, high-level properties such as symmetry, secondary sexual characteristics and several ratios and distances (e.g., between eyes and mouth) affect attractiveness ratings. The aim of the present study was to investigate whether other (low-level) properties of face images also correlate with attractiveness ratings. We analyzed low-level statistical properties of face images that were rated for attractiveness and found that attractiveness correlated negatively with self-similarity (measured by Fourier transform and PHOG analysis) and positively with complexity and anisotropy. Furthermore, we found positive correlations of self-similarity with the age of the depicted person. In a follow-up experiment, we changed the slope ratio in log-log plot of radially averaged Fourier power (an established measurement for self-similarity) of face images. Participants rated a version of the same face, which was rendered less self-similar, as significantly less attractive. This result can be explained if one assumes that high correlations of self-similarity with age mask the negative correlation with attractiveness. In conclusion, we demonstrated a relation between low-level image properties, such as self-similarity, complexity and anisotropy, on judgments on attractiveness as well as on the age of faces. |
| Up Home |
|---|