Automatic facial expression recognition is an interdisciplinary and frontier field of psychology, computer science and also others. Researchers in the fields of emotional psychology, pattern recognition, and affective computing develop expression recognition-related theories, databases, and algorithms, which greatly promote the progress of automatic facial expression technology. Based on the psychological perspective and combined with our previous related work, the article firstly introduced the theoretical perspectives and practical advances in the psychological basis of automatic facial expression recognition, emotion expressing approaches, facial expression databases development, and emotion annotations; then analyzed and discussed the main problems faced in automatic expression recognition; and finally based on the constructivist view of predictive processing theory, proposed that "understanding" facial expressions in the process of interpersonal interaction is expected to further improve the effectiveness of automatic facial expression recognition, and it is expected that this may be the future direction of automatic facial expression recognition.

自动表情识别是心理学与计算机科学等深度交叉的前沿领域。情绪心理学、模式识别、情感计算等领域的研究者发展表情识别相关的理论、数据库和算法，极大地推动了自动表情识别技术的进步。文章基于心理学视角，结合我们前期开展的相关工作，首先梳理自动表情识别的心理学基础、情绪的面部表达方式、表情数据的演化、表情样本的标注等方面的理论观点与实践进展，然后分析指出自动表情识别面临的主要问题，最后基于预测加工理论的建构观点，提出注重交互过程中的表情“理解”，有望进一步提高自动表情识别的有效性，并预期这可能是自动表情识别研究的未来发展方向。

温州市科技计划项目（G20210027）