Smiling Reduces Masculinity: Principal Component Analysis Applied to Facial Images

Abstract

We examined the effect of facial expression on the assignment of gender to facial images. A computational analysis of the facial images was applied to examine whether physical aspects of the face itself induced this effect. Thirty-six observers rated the degree of masculinity of the faces of 48 men, and the degree of femininity of the faces of 48 women. Half of the faces had a neutral facial expression, and the other half was smiling. Smiling significantly reduced the perceived masculinity of men's faces, especially for male observers, whereas no effect of smiling on femininity ratings was obtained for women's faces. A principal component analysis was conducted on the matrix of pixel luminance values for each facial image x all the images. The third principle component explained a relatively high proportion of the variance of both facial expressions and gender of face. These results suggest that the effect of smiling on the assignment of gender is caused, at least in part, by the physical relationship between facial expression and face gender.

Discrimination of face gender and expression under dual-task conditions

“…Previous studies on the interaction between the gender and expression of faces have shown that gender recognition is influenced by emotional expression. While angry expressions make faces look more masculine and reduce the accuracy with which a face is recognized as female (Aguado et al, 2009;Hess, Adams, Grammer, & Kleck, 2009), the smile makes faces look less masculine and impairs recognition of a face as male (Hess et al, 2009;Kawamura, Komori, & Miyamoto, 2008). The explanation of these effects is that some facial features have common diagnostic value for gender and expression.…”