Dimensions of visual information for categorizing facial expressions of emotion

Using photographs of the human facial expressions as stimuli, we had the subjects judge the facial expressions in terms of the six basic categories of emotion: happiness, surprise, fear, anger, sadness, and disgust. For each stimulus, we measured the displacements of the facial feature points from a neutral face used as a reference. The relationships between the displacements of the facial feature points and the distribution of categorical judgments were examined by a canonical discriminant analysis. We found three major canonical variables, which were similar to those found by Yamada (1993). Thus, we conclude that we human beings use the structural dimensions of the face, such as curvedness/openness and slantedness of the facial elements, as the information for categorizing the facial expressions of emotion.