Following Yik and Russell (1999) a judgement paradigm was used to examine to what extent differential accuracy of recognition of facial expressions allows evaluation of the well-foundedness of different theoretical views on emotional expression. Observers judged photos showing facial expressions of seven emotions on the basis of: (1) discrete emotion categories; (2) social message types; (3) appraisal results; or (4) action tendencies, and rated their confidence in making choices. Emotion categories and appraisals were judged significantly more accurately and confidently than (...) messages or action tendencies. These results do not support claims of primacy for message or action tendency views of facial expression. Based on a componential model of emotion it is suggested that judges can infer components from categories and vice versa. (shrink)