How the brain is wired for faces

Faces produce a particular resonance of recognition, even in the youngest infants, who respond to the sight of a face almost from birth. While neurobiologists have known that a particular area of the brain, called the fusiform face area (FFA), lights up with activity when we see a face--and even that the FFA is necessary for us to recognize faces--there is controversy over what kind of processing the area is doing.

Now, Galit Yovel and Nancy Kanwisher have tackled two central questions with one set of experiments: the nature of processing that occurs in the FFA and whether the FFA is "domain specific," that is, exclusively involved in face perception, or whether the area is engaged in more general spatial processing of visual features.

Their conclusions are that the FFA extracts configural information about faces rather than processing spatial information on the parts of faces. Also, their studies indicated that the FFA is exclusively involved in face recognition.

The researchers' experiments combined both functional magnetic resonance imaging (fMRI) and behavioral studies of subjects as they performed recognition tasks. In the widely used technique of fMRI, harmless magnetic fields and radio signals are used to measure brain activity as subjects perform tasks.

In the fMRI studies, volunteer subjects were asked to discriminate differences between faces in which the parts were spaced differently, as well as differences between faces in which some parts were replaced by those of different faces. For the nonface objects, the researchers substituted images of houses, altering the spacing or identity of the windows and doors.

In the behavioral tests, volunteers were asked to match either houses or faces with such differences and their performance was measured. These experiments took advantage of the fact that face recognition shows the unique phenomenon that people find it more difficult to recognize upside-down faces than right-side-up faces.

By sophisticated analysis of their data on the two kinds of experiments, the researchers arrived at new insights into the function and specificity of the FFA.

They wrote that "contrary to a widespread view in the literature, we found no evidence that face processing mechanisms are specifically engaged in the extraction of spatial distance among parts."

Also, they wrote, "we attempted to induce face-like processing of nonface stimuli (houses) by matching discrimination tasks on houses as closely as we could to discrimination tasks on faces. This effort failed dramatically…These data support the domain specificity of face processing."

Galit Yovel and Nancy Kanwisher: "Face Perception: Domain Specific, Not Process Specific"

Publishing in Neuron, Volume 44, Number 5, December 2, 2004, pages 889–898. www.neuron.org

---

---