Abstract:

Joint action constitutes a core component of social life. Its tasks often exhibit a hierarchical structure ranging from concrete to abstract, requiring co-actors to achieve effective self-other integration at each level. However, existing research has largely been confined to isolated investigations of single levels, treating interpersonal coordination as a static mapping and thus failing to capture the dynamic interactions and information accumulation mechanisms that operate across hierarchical levels. Adopting the perspective of “hierarchical task representation, ” this study explores: (1) the dissociation and interaction of self-other integration at higher and lower levels, at both behavioral and neural levels; (2) whether level specificity arises from differences in the patterns of prior information accumulation, and how cross-level interactions are realized through the dynamic coupling of prior and posterior distributions; and (3) how different interpersonal common ground contexts modulate this process. Methodologically, this study constructs a hierarchically nested task paradigm, integrating behavioral experiments, EEG hyperscanning, and cognitive computational modeling to investigate level specificity and interaction patterns along three dimensions: phenomenon, mechanism, and modulation. Furthermore, this study proposes a Bayesian computation-based theoretical model of self-other integration, enabling a paradigm shift from “static single-level representation” to “dynamic hierarchical prediction.” This research deepens the understanding of the dynamic mechanisms underlying interpersonal integration across higher and lower levels, and provides a computable and verifiable psychological foundation for the development of collaborative multi-agent systems.

Key words: joint action, self-other integration, hierarchical task representation, Bayesian model