We propose a framework to explain the processing of the recognition of angry and happy facial expressions using Item Response Theory (IRT). Without modifying the intensity, natural facial expressions were briefly presented to 72 participants in a categorical judgment experiment. The IRT analysis gave rise to the parameters "difficulty" and "consistency" for the judgment of the expressions. Then the relationship between facial features and these parameters was predicted by multiple regression analysis. The results showed that: (a) the difficulty of the judgments was determined by the "holisticity" of facial features, (b) the consistency of the judgment rose when both the upper and the lower parts of a face contained expressions of anger, and (c) the difficulty and the consistency were mainly affected by the changes of the upper parts of angry faces and lower parts of happy faces.
展开▼
