A little finer than Snapchat suggests in these photos. While the male traits seem relevant, the female traits suggested by the application seem far too artificial, don't you think?
Left : Portrait of James, Viscount Maitland (Thomas Gainsborough, 1780). Mid : Snapchat filter applied to the portrait. Right : Gif I created from the « Portrait of Sally Fairchild » : a Snapchat filter applied to the portrait by Sargent (1884-1887). The feminine features change and become more masculine. All photos from Dheera Venkatraman @dheeranet.
The female gender is not limited to long hair and makeup, of course. You can recognize a female face from a male face without this information.
By taking a closer look at the two faces below, we can agree that the one on the left would be male and the one on the right female. The only difference between these two faces is the contrast of each image, which is different: it is decreased on the left, and increased on the right. In order to understand what was highlighted by the contrast, I myself increased the contrast until it was intensified to its maximum.
First two faces by Richard Russell « The illusion of sex » Source : https://www.researchgate.net/figure/The-Illusion-of-Sex-The-left-face-appears-male-while-the-right-face-appears-female-yet_fig4_267802901. The following ones (right and below) I have been modified : I intensified the contrast of the faces. Bottom right: I outlined the contours of the faces. With this delineation the contrast had intensified the contours of the lips and eyes on the face on the right (determined as feminine in the first photo).
Thus, a face with few clues (= uncertainty) can appear sometimes feminine and sometimes masculine by modifying only the contrast, but above all, by not changing the characteristics of the face (chins, palms, etc). Is this a cultural effect of the general use of cosmetics by women? We don't know at this time.
It is this degree of uncertainty, in the perception of faces and voices, that motivated our last study (SBRI, INSERM, France). We did not modified the contrast: we morphed a real female face and a real male face. Ditto for the voice. Using brain imaging, we found that gender perception was coded in several brain regions, notably those dealing with voice and those dealing with faces. And their cortical integration has a bearing on the decision making about the gender of the supposed person. In short, when we don't know, we will look for the most relevant information and it will become important. With several hierarchical levels (= the information processing follows a route of increasingly complex cortical levels). And will influence our perception of gender.
_____
For more details: our article in open-source, waiting for the peer review. Results from several years of work with my colleagues and students: to set up the most relevant imaging protocols, to collect data, to analyze the results with statistical tools that we now master very well (DCM for fMRI data). Fingers crossed ;-) https://www.researchgate.net/project/Face-Voice-Integration