Emotion concepts and facial movements facilitate emotion perception: A computational account

Inferring others' emotions from facial expressions is crucial for social interaction. Previous research has demonstrated that both top-down conceptual knowledge and bottom-up sensory cues from facial movements contribute to emotion perception. However, it remains unclear how these factors jointly influence emotion perception, as they have often been examined in isolation. This study aims to clarify the respective roles of conceptual knowledge and perceived facial movements and to reveal their precise computational mechanisms. We designed two behavioral experiments and applied hierarchical drift diffusion modeling to the data. In Experiment 1, we found that emotion concepts facilitate the recognition of facial expressions, with a stronger effect observed when processing dynamic expressions compared to static ones. Experiment 2 further indicates that facial movements impact emotion perception primarily by facilitating access to conceptual knowledge rather than the perceptual processes. Moreover, drift-diffusion modeling suggests that emotion concepts improve facial expression recognition by increasing the efficiency of evidence accumulation and decreasing the decision threshold, whereas facial movements enhance the efficiency of evidence accumulation when conceptual knowledge is available. Our study disentangles the distinct and interactive roles of top-down knowledge and bottom-up facial movements in emotion perception, providing novel computational evidence in support of constructivist theories of emotion.