MIT's McGovern Institute for Brain Research sponsored a lecture on "Interpreting Faces" followed by a discussion of research on facial prototyping and other aspects of higher visual processing.
David Perrett, professor in the School of Psychology at the University of St. Andrews in Scotland , gave the Feb. 11 talk. Research topics in his Perception Laboratory include facial aging, facial attractiveness, perceptual asymmetry and other aspects of how human brain cells respond to images of faces, various body parts and gestures.
Perrett's slide presentation showed, for example, that some brain cells are "tuned" to pick out faces; some select and respond to head direction and others to motion.
"Color is a sluggish signal and not so significant as one might have thought," he said. "Form arrives in the brain at the same time as color, but its effect lasts longer. Coarse and fine details appear to arrive at the cellular level at the same time. Hierarchically speaking, brain cells respond to eyes, then mouth; to frontal views, then side or back."
Perrett's "couch potato study" contrasted images of people lying on sofas with images of athletes who had just stopped moving. Brain cell activity increased only when subjects were shown images of the just-stopped athletes, he said. Also, cells respond more actively to the sight of a body moving toward them rather than away, suggesting responsiveness to intention.
As for hiding, Perrett demonstrated by ducking behind the podium. "If I stand beside this podium or walk away from it, there is no cell action at the sight of the podium. But if I hide behind the podium, 'it'--that is, the podium--'works' on the cells, as shown by their activity," he said.
His appearance suggested a cultural gap between the United States and England for the image of "professor." A slender man reminiscent of pop star Sting, Perrett featured yellow, green and purple hair, a horizontally striped black and green shirt, red pants and a black tank top. Not exactly the prof. in the gray flannel suit.
A version of this article appeared in MIT Tech Talk on February 27, 2002.