主事感(Sense of Agency，SoA)保证了在人机交互过程中人的尊严感，是当 前人机混合智能快速发展背景下研究的热点。意图绑定(Intentional Binding，IB) 范式可以通过时间间隔估计测量主事感，已成为主事感研究的经典行为学范式。?研究在具身认知框架下设计了两个全新的主事感研究范式。在第一视角范 式中，通过与快速序列视觉呈现(Rapid Serial Visual Presentation，RSVP)范式结 合提高了第一视角主事感范式的生态效度;在第三视角范式中将经典第三视角小 球移动任务和意图绑定范式相结合，首次在第三视角范式下客观测量主事感，将 第三视角范式纳入到意图绑定框架中。在此基础上，?研究从问卷、行为学和脑电 表征三个角度，分析第一视角和第三视角实验结果。结果发现，两种视角下均观察 到主动控制条件下问卷得分更高和意图绑定的时间压缩效应。两种视角下，正确 的反馈结果和主动的操作方式以前馈性和反馈性的形式交互提高主事感体验，操 作方式提供对主事感的预期，反馈则通过结果对主事感进行修正，两者共同构建 了主事感。同时在两个实验中发现与主事感相关的脑电成分:N200 和 P300。实验表明 N200 反映了受试在正确反馈条件下产生的视觉注意投入，P300 则反映了在错误反 馈下的不受预期感，并发现 P300 成分与行为学结果呈“U”形曲线关联，揭示了主 事感的动态变化特性;问卷结果发现隐私感、信任感与 N200 波幅有关联。?研究 在关键脑电表征 N200 和 P300 基础上，通过共空间模式(Common Spatial Pattern， CSP)特征提取和支持向量机(Support Vector Machine，SVM)算法对主事感形成 较高准确率的二分类识别，为实现更加和谐的人机交互环境提供基础。?研究从而行为学、脑电、问卷三个维度对主事感进行了整体综合的分析，反 映出主事感在两种视角下的特征、共性与差异。
The sense of agency (SoA) ensures human dignity in human-machine interaction and is a hot topic in the context of the rapid development of human-machine hybrid intelli- gence. The intentional binding (IB) paradigm, which measures SoA by estimating time intervals, has become a classic behavioral paradigm for SoA research.In this study, two new SoA research paradigms were designed under the embod- ied cognition framework. The ecological validity of the first-person perspective SoA paradigm was improved by combining it with the rapid serial visual presentation (RSVP) paradigm. In the third-person perspective paradigm, the classic third-person perspective ball movement task was combined with the IB paradigm to objectively measure SoA in the third-person perspective paradigm and include it in the IB framework for the first time. Based on this, the results of the first-person and third-person experiments were ana- lyzed from three perspectives: questionnaire, behavioral, and EEG data. Results showed that higher questionnaire scores and time compression effects in the IB paradigm were observed under the active control condition in both perspectives. In both perspectives, active operating methods interacted with correct feedback results in a forward and back- ward manner to enhance the sense of agency experience. The operating method provided an expectation for the sense of agency, while the feedback corrected the sense of agency through the results, and the two together constructed the sense of agency.At the same time, N200 and P300 EEG components related to SoA were found in both experiments. The results showed that N200 reflected visual attention investment produced by participants under correct feedback conditions, while P300 reflected unexpectedness under incorrect feedback conditions. The P300 component was found to be related to be- havioral results in a ”U”-shaped curve, revealing the dynamic changing characteristics of SoA. The questionnaire results revealed that there is an association between privacy , trust, and N200 wave amplitude. Based on the key EEG characterization of N200 and P300, this study achieved high accuracy in two-class classification of SoA through common spatial pattern (CSP) feature extraction and support vector machine (SVM) algorithm, providing a foundation for achieving a more harmonious human-machine interaction environment.This study provides a comprehensive analysis of sense of agency through three dimensions of behavioral, electroencephalogram, and questionnaire data, reflecting thecharacteristics, commonalities, and differences of sense of agency in two perspectives.
