Facial features in perceived intensity of schematic facial expressions.

Facial expressions may be perceived continuously or categorically, i.e., perception of facial expressions may be explained by feature-based models or holistic models. A linear additive model is proposed, with weightings for facial features to reflect configural spatial relationships or distances of features from each other. Predictions from the model regarding the perception of affect intensities (e.g., of anger and sadness) were tested using schematic faces, with three facial features, eyebrows, eyes, and mouth as predictors. Model predictions, based on multiple regression analyses, corresponded well with rating judgments of anger and sadness in three experiments. The results suggest not only the limit of feature-configural models of facial expression perception but the feature-selective configuration of facial features and continuous perception of facial expressions. Findings also indicate that the salience or weight of each feature varies with the emotion perceived by observers.