'The mirror of the soul?' Inferring sadness in the eyes

Abstract

The eyes are widely regarded as the mirror of the soul, providing reliable nonverbal information about drives, feelings, and intentions of others. However, it is unclear how accurate emotion recognition is when only the eyes are visible and whether inferring of emotions is altered across healthy adulthood. To fill this gap, the present piece of research was directed at comparing the ability to infer basic emotions in two groups of typically developing females that differed in age. We set a focus on females seeking group homogeneity. In a face-to-face study, in a two-alternative forced choice paradigm (2AFC), participants had to indicate emotions for faces covered by masks. The outcome reveals that although the recognition pattern is similar in both groups, inferring sadness in the eyes substantially improves with age. Inference of sadness is not only more accurate and less variable in older participants, but also positively correlates with age from early through mid-adulthood. Moreover, reading sadness (and anger) is more challenging in the eyes of male posers. A possible impact of poser gender and cultural background, both in expressing and inferring sadness in the eyes, is highlighted.

Keywords: Basic emotions; COVID-19; Face covering; Poser gender; Reading emotions in the eyes; Sadness.

Figure 1

A female poser expressing six basic emotions. Faces are shown under full-face (top) and covered-by-mask conditions (bottom row). From Carbon, the Creative Commons Attribution [CC BY] license. These images are presented for illustrative purposes only, and have not been used as experimental material in the present study.

Figure 2

Recognition accuracy of facial emotions on the EMF task for Group 1 (apricot, from Pavlova et al.) and Group 2 (olive green; from Moosavi et al.). Vertical bars represent ± SEM. Double asterisk indicates a significant difference (p < 0.05), single asterisks a tendency (0.05 < p < 0.1).

Figure 3

Link between inferring sadness on the EMF task and age. Significant positive non-linear Spearman correlation (p = 0.039) was found between recognition accuracy of sadness and age.

Figure 4

Recognition accuracy of facial emotions on the EMF task for female faces (yellow mustard) and male (olive grey) faces in TD females. Vertical bars represent ± SEM. Double asterisks indicate significant differences (p < 0.05), single asterisk a tendency.

Figure 5

Pronounced differences in expression of sadness within the same cultural background. The images are taken from the MPI FACES database (Ebner et al.; public domain). These images are presented for illustrative purposes only, and have not been used as experimental material in the present study.