Abstract
There is ample evidence that the amygdala plays an important role in emotion recognition. Bilateral amygdala damage severely impairs the recognition of fear from facial expressions (Adolphs et al., 1994, 1995; Broks et al., 1998; Calder et al., 1996; Young et al., 1995), but the mechanisms behind this impairment remain unknown. Here, we used the Bubbles technique (Gosselin & Schyns, 2001) to shine new light on this issue.
We compared the information used effectively to recognize emotion by SM, an individual with complete, bilateral damage restricted to the amygdala, with the information used effectively by 10 normal controls. Each subject was submitted to 3,072 trials. In a trial, one of four faces (a male and female face each of happiness and fear from Ekman & Friesen's, 1976, set) was decomposed in the Laplacian pyramid space and sparsely revealed by randomly located Gaussian holes (see Schyns, Bonnar & Gosselin, 2002, for details). Subjects were asked to classify either the gender or the emotion of the image.
Statistical analyses on the difference between the classification images of the normal controls and of SM revealed that SM appears to use information about the face normally when classifying gender. By contrast, when classifying emotion, she failed to make normal use specifically of the eye region of the face, an impairment most notable at spatial frequencies between 90 and 45 cycles per face.