Key points are not available for this paper at this time.
SUMMARY The same facial expression will be seen as expressing different types and degrees of emotion, depending on what other faces are seen. A relatively neutral face seems sad when presented alongside a happier face, happy alongside a sadder face. A relatively unhappy face seems sad when presented alongside an angrier face, angry alongside a sadder face. The specific emotion seen in a face can be predicted by combining the idea that human judgment is relative with a model of the scale implicit in the judgment of emotion. In that model, categories and dimensions descriptive of feelings are interrelated in a systematic fashion: Categorical descriptors such as happy, sad, calm, and angry are at specific locations around the periphery of an emotion judgment space defined by degree of pleasure and degree of arousal. Viewing one facial expression (an anchor) shifts the entire scale of judgment, displacing the emotion seen in a subsequent expression (the target) in the opposite direction. Specifically, (a) magnitude of shift in perception of the target is proportional to the distance of the anchor from the center of the emotion space, (b) the angle of shift differs by 180° from the angle from the center to the anchor, and (c) the quantitative pleasure-arousal shift in the target corresponds to predictable shifts in the applicability of categorical descriptors. The geometric formulation of these ideas, although perhaps exaggerating our ability to quantify the message seen in a facial expression, did suggest specific hypotheses, some of which were tested here in six experiments. Supportive results were found with successive and with simultaneous presentation of anchor and target, with various target expressions including expressions of neutral, surprised, and angry feelings, with both posed and spontaneous target expressions, with target expressions posed by the same actor seen in the anchor and by a different actor, with different groups of observers, and with different
Russell et al. (Tue,) studied this question.