Fuzzy Concepts and the Perception of Emotion in Facial Expressions

Research on how well one person can recognize the emotion expressed in another person's face has resulted in controversy: accuracy versus inaccuracy; discrete categories versus dimensions versus structural models of emotion. These seemingly disparate conclusions can be reconciled when natural language concepts of emotion (“happiness,” “anger,” “tear,” “sadness,” etc.) are thought of as overlapping and fuzzy, rather than as mutually exclusive and properly defined. The research literature on the perception of emotion in facial expressions is reviewed from this vantage point, and four studies testing predictions from this thesis are reported. When rating the degree to which either posed or spontaneous facial expressions exemplify emotion categories, subjects produced reliably graded responses and indicated that individual (even prototypical) expressions belong to more than one category. The graded “prototypicality” ratings (1) predicted the probability with which the expression was said to be a member of the emotion category or was selected as a referent for the category label, and (2) generated a replica of a circumplex structural model of emotion when subjected to multidimensional scaling. Methodological and conceptual issues about the meaning attributed to facial expressions and about the perception of emotion in general are also discussed.