Classification of dynamic facial expressions of emotion presented briefly

Abstract

A number of studies have shown an impact of speed of a developing facial expression of emotion on its recognition and perceived naturalness. Still, the impact of speed at constant, short presentation times, as normally used in many experiments is unclear. In the present study participants classified faces displaying facial expressions of six basic emotions in static and dynamic presentation modes and three different types of neutral movements. Stimuli were created with computer software that allows fine-grained control over action units and dynamic features. Rise times in dynamic expressions varied between 200 and 900 ms. Results replicated classical findings showing better performance for expressions of happiness, and frequent confusions among morphologically similar expressions, and a general dynamic facilitation for most expressions. Importantly, dynamic presentation as such facilitated a more accurate classification, but variations in speed at the fast range studied here had no noticeable effect for expressions of anger, fear, happiness, and surprise. The main exception was sadness, which was best recognised at slow speed and in static pictures, and disgust, which was most unambiguously categorised at fast to moderate speed.