Zhou Peng, Ma Huimin, Zou Bochao, Zhang Xiaowen, Zhao Shuyan, Lin Yuxin, Wang Yidong, Feng Lei, Wang Gang
School of International Studies, Zhejiang University, Hangzhou, China.
School of Computer and Communication Engineering, University of Science and Technology Beijing, Beijing, China.
Npj Ment Health Res. 2023 Aug 10;2(1):12. doi: 10.1038/s44184-023-00031-0.
To explore the minds of others, which is traditionally referred to as Theory of Mind (ToM), is perhaps the most fundamental ability of humans as social beings. Impairments in ToM could lead to difficulties or even deficits in social interaction. The present study focuses on two core components of ToM, the ability to infer others' beliefs and the ability to infer others' emotions, which we refer to as cognitive and affective ToM respectively. Charting both typical and atypical trajectories underlying the cognitive-affective ToM promises to shed light on the precision identification of mental disorders, such as depressive disorders (DD) and autism spectrum disorder (ASD). However, most prior studies failed to capture the underlying processes involved in the cognitive-affective ToM in a fine-grained manner. To address this problem, we propose an innovative conceptual framework, referred to as visual theory of mind (V-ToM), by constructing visual scenes with emotional and cognitive meanings and by depicting explicitly a four-stage process of how humans make inferences about the beliefs and emotions of others. Through recording individuals' eye movements while looking at the visual scenes, our model enables us to accurately measure each stage involved in the computation of cognitive-affective ToM, thereby allowing us to infer about potential difficulties that might occur in each stage. Our model is based on a large sample size (n > 700) and a novel audio-visual paradigm using visual scenes containing cognitive-emotional meanings. Here we report the obtained differential features among healthy controls, DD and ASD individuals that overcome the subjectivity of conventional questionnaire-based assessment, and therefore could serve as valuable references for mental health applications based on AI-aided digital medicine.
探索他人的心理，传统上称为心理理论（ToM），这可能是人类作为社会存在最基本的能力。心理理论受损可能导致社交互动困难甚至障碍。本研究聚焦于心理理论的两个核心成分，即推断他人信念的能力和推断他人情绪的能力，我们分别将其称为认知性心理理论和情感性心理理论。描绘认知 - 情感性心理理论背后的典型和非典型轨迹，有望为诸如抑郁症（DD）和自闭症谱系障碍（ASD）等精神障碍的精准识别提供线索。然而，大多数先前的研究未能精细地捕捉认知 - 情感性心理理论所涉及的潜在过程。为解决这一问题，我们提出了一个创新的概念框架，称为视觉心理理论（V - ToM），通过构建具有情感和认知意义的视觉场景，并明确描绘人类如何推断他人信念和情绪的四个阶段过程。通过记录个体观看视觉场景时的眼动，我们的模型能够准确测量认知 - 情感性心理理论计算中涉及的每个阶段，从而使我们能够推断每个阶段可能出现的潜在困难。我们的模型基于大样本量（n > 700）以及一种新颖视听范式，该范式使用包含认知 - 情感意义的视觉场景。在此，我们报告了在健康对照组、抑郁症患者和自闭症谱系障碍患者中获得的差异特征，这些特征克服了传统基于问卷评估的主观性，因此可为基于人工智能辅助数字医学的心理健康应用提供有价值的参考。
