Faces do not speak for themselves when it comes to emotion

Chapter 3 endnote 1, from How Emotions are Made: The Secret Life of the Brain by Lisa Feldman Barrett.
Some context is:

(You have just viewed a photograph of a face that seems to be expressing a certain emotion, but when you see the rest of the person's body, the emotion changes.)

The psychologist Hillel Aviezer at the Hebrew University of Jerusalem and his colleagues conducted a clever series of experiments showing that emotions are perceived in faces based on the context, including body posture. In fact, the face is almost always trumped by body postures when it comes to emotion perception. Aviezer grafted together photographs of posed faces and bodies that didn’t belong together. For example, he’d attach a posed scowling (stereotyped angry) face onto a body that’s holding a soiled object, such as a dirty diaper, and ask subjects to identify the emotion the target person was feeling. Subjects nearly always said “disgust,” identifying the emotion appropriate to the body, not the face.

Aviezer and colleagues also used sophisticated machinery to track the subjects’ eye movements and made a key observation: depending on the emotion perceived (anger vs. disgust), subjects scanned the faces differently. When viewing a stereotyped “Anger” facial configuration attached to a “Disgust” body pose, they scanned the face in the manner they’d scan a stereotyped “Disgust” facial configuration with a wrinkled nose. It’s as if the body posture influenced how the face was processed. So, the context brought the emotion concept “Disgust” to mind, leading the subject to see the face differently.

Aviezer and colleagues conducted a similar study using photos of tennis players, grafting faces and bodies of winners and losers in various combinations. Again, subjects assigned more importance to the body than the face. When shown an “extreme pleasure of winning” face on an “agony of losing” body, for example, subjects consistently perceived the face as agony. What’s more, when subjects were asked to pose their own face to match the picture, they looked more distressed than the face in the photo actually did. This discrepancy is evidence of simulation: subjects were adding information about distress that wasn’t present in the photo, using their concept of “Distress,” just like your brain added information about a bee the second time you looked at the blobby picture in chapter 2.