Remember this guy from a few posts ago?!
Nothing is actually wrong with his face. Check out the answer (and another picture) after the break...
Here he is again with a little more telling setup.
According to the website of Lawrence D. Rosenblum and Mike Gordon at UC Riverside:
The point-light technique involves applying small fluorescent yellow cardboard dots to a speaker's face. The dots are affixed to the lips and face with a medical adhesive, and to the teeth and tongue tip with a denture paste. The speaker is videotaped under high-quality fluorescent lighting (seen on left). The resultant videotapes will show only bright points moving against a black background. For this experiment, the speaker's face is filmed within a distractor frame made-up of additional point-lights. The purpose of the frame is to hide structural information about the speaker's identity and articulation. This technique forces observers to rely on the time-varying information available in the displays and makes it difficult for observers to perform accurate judgments with static point-light images.
