关键词: 联合行动, 自我-他人整合, 层级任务表征

Abstract: Joint action is an integral part of social life. Its tasks often exhibit a hierarchical structure ranging from concrete (low-level) to abstract (high-level), requiring actors to achieve effective self-other integration across various levels. Previous studies have largely investigated self-other integration in single-level tasks in isolation, focusing more on the representation of a partner's task information while neglecting the integration process following such representation. To this end, adopting the perspective of "hierarchical task representation," this study aims to address the following questions: (1) Whether there exists dissociation and interaction between self-other integration at high and low levels; (2) Whether level-specificity stems from differences in prior information accumulation patterns, and how inter-level interactions are realized through the coupling of posterior and prior distributions; (3) How different interpersonal common ground contexts modulate the aforementioned integration. By constructing a hierarchical nested task paradigm and integrating behavioral experiments, EEG hyperscanning, and cognitive computational modeling, this study systematically investigates the principles of level-specificity and interaction across three dimensions: "phenomenon—mechanism—modulation." The study innovatively proposes a theoretical model of self-other integration grounded in hierarchical Bayesian computational models. This approach overcomes the limitations of previous single-level research and provides a theoretical basis for multi-agent system algorithms and collaborative robots to achieve human-like hierarchical interactions.

Key words: Joint action, Self-other integration, Hierarchical task representation