Categorical Perception of Morphed Facial Expressions

Abstract

Using computer-generated line-drawings, Etcoff and Magee (1992) found evidence of categorical perception of facial expressions. We report four experiments that replicated and extended Etcoff and Magee's findings with photographic-quality stimuli. Experiments 1 and 2 measured identification of the individual stimuli falling along particular expression continua (e.g. from happiness to sadness) and discrimination of these stimuli with an ABX task in which stimuli A, B, and X were presented sequentially; subjects had to decide whether X was the same as A or B. Our identification data showed that each expression continuum was perceived as two distinct sections separated by a category boundary. From these identification data we were able to predict subjects' performance in the ABX discrimination task and to demonstrate better discrimination of cross-boundary than within-category pairs; that is, two faces identified as different expressions (e.g. happy and sad) were easier to discriminate than two faces of equal physical difference identified as the same expression (e.g. both happy). Experiments 3 and 4 addressed two new issues arising from Etcoff and Magee's (1992) data and the results of our own Experiments 1 and 2: (1) that they might reflect artefacts inherent in the use of single continua ranging between two prototypes- for example, a range effect or an anchor effect, (2) given that the ABX procedure incorporates a short-term memory load, discrimination data obtained with this task might reflect a short-term memory rather than a perceptual phenomenon. We found no support for either of these reinterpretations and further evidence of categorical perception.