Do We Expect Women to Look Happier Than They Are? A Test of Gender-Dependent Perceptual Correction

Rounded, high-browed feminine facial features have been reported to enhance the expressive cues associated with happiness and fear, whereas masculine features like a low brow and a square jaw reinforce those associated with anger (Adams, Hess, & Kleck, 2015; Becker, Kenrick, Neuberg, Blackwell, & Smith, 2007; Hess, Adams, Grammer, & Kleck, 2009; Hess, Adams, & Kleck, 2009a, 2009b). This would make a woman look happier than a man with the same expression intensity as shown, for example, by Hess, Blairy, and Kleck (1997). So, in order to estimate the actual happiness of a woman more accurately, we would have to subtract the effect of these facial features. The question of interest is whether we actually do it. There is enough evidence that facial expression can get processed differently depending on the expresser’s gender (e.g., Bestelmeyer, Jones, Debruine, Little, & Welling, 2010). We hypothesised that our perceptual mechanisms must have evolved or learned to apply the required subtraction for females but not males. We predicted that this subtraction, being woman specific, would cause one to infer the man to be happier than the woman, if they do not differ in the facial features that enhance the expressive cues of happiness. That is, in this case, one would underestimate the woman’s felt emotion intensity. We also predicted that since the expressive cues of sadness are not affected by feminine features, there would be no subtraction and one would not similarly underestimate the woman’s sadness intensity.

Forty student volunteers (M_age = 24.10 years, 20 men and 20 women) participated in the study after giving informed written consent. 20 androgynous faces with happy and 20 with sad expressions were created using the Facegen Modeller software program (Singular Inversions Inc.). The expressions were objectively adjusted at 75% intensity for all the faces. Faces expressing the same emotion were randomly paired such that there were 10 pairs each of happy and sad faces. A trial started by showing the faces in a pair side by side on a monitor. Gender-appropriate hair styles and captions such as ‘Female 22 years’ were used to convey that one is a man and the other a woman. Since the faces were androgynous, they did not differ in other gender-specific features. After 5 seconds, the hair was removed and after another second, the subject was asked ‘Who is happier?’ or ‘Who is sadder?’ depending on the emotion on the faces presented. Note that the question was not ‘Who looks happier?’ or ‘Who looks sadder?’ because we wanted the subjects to estimate the emotion intensities experienced by the expressers and compare them, not specify whose expression intensity looked greater. The question and the faces remained on the screen until the subject pressed ‘m’ for the male-designated face or ‘f’ for the female-designated face. There were 20 trials with the first 10 having faces expressing one emotion and the rest the other. The order of presentation of the two emotions was counterbalanced by showing half the subjects the happy faces first and the others the sad faces first. To counterbalance the gender priming of the faces, if half the subjects were shown a particular face as male, the other half were shown it as female. The left-right positions of the male and female designated faces on the screen were randomised (Figure 1). OPEN IN VIEWER

The binomial responses tested with logistic linear mixed models with random effects of subjects and stimuli revealed a small, yet significant bias towards choosing the male-designated faces for the happiness expression (β = 0.211, z = 2.065, p = .039) but not for the sadness expression (β = −0.09, z = −0.887, p = .375). More model details are provided as supplementary material. The underestimation of the experienced happiness intensity of the female-designated expressers, in spite of matched appearance and expressiveness, indicates the presence of a perceptual subtraction for the female expression of happiness as hypothesised. There was no evidence for such underestimation for sadness, indicating that a similar adjustment may not be present for it. Further exploration may reveal the presence of perceptual corrections for fear expression in females and anger expression in males.

There is a strong consensus that females are emotionally more expressive than males (Chaplin, 2015; Fischer & LaFrance, 2015; Hess et al., 1997). Hence a related question that arises is whether our perceptual mechanisms apply any correction for the enhanced expressiveness of females to arrive at their internal emotion more accurately. Our results did not provide any evidence for this as such a correction should have caused the male sad expressers to be inferred sadder. It may be that we attribute the greater expressiveness of women to a greater experienced emotion and not just to an enhanced facial response to a given experienced emotion intensity. This is consistent with the reported stereotypic belief among people that women feel emotions more strongly than men (Adams et al., 2015).