Although research has emphasized teams as complex systems that must continuously react to changing contextual demands, construct-based explanations—such as input-output relationships—still dominate and struggle to capture the interaction dynamics of team adaptation. Research often lacks insight into how and why specific construct relationships exist, constraining the design of effective interventions. Consequently, we employ computational process theorizing (CPT) to uncover the generative mechanisms of focal team phenomena and highlight how the layers of theoretical explanation (LTE) provide insights for team cognition in human-AI teams.
Research has linked information-building processes, shared mental representations, and team adaptation, yet most studies only capture static end-states, offering little understanding of their temporal unfolding. Team cognition research also battles contradictory or overlapping constructs that rarely specify what, when, and how knowledge must be shared. Traditional distinctions—such as teamwork versus taskwork information or convergence versus divergence of cognition—are often static and neglect the dynamic interplay between process and outcome. This leads to suboptimal guidance for the development of AI team members whose communicative abilities (e.g., AI transparency) and collaboration outcomes are largely determined by design.
Accordingly, we propose a CPT to formalize how shared understanding emerges, stabilizes, or collapses over time as a function of knowledge types (declarative, procedural, strategic), team composition (human versus AI), and task interdependence. By simulating how knowledge sharing evolves across contingencies, we aim to identify non-linear dynamics (e.g., tipping points) that clarify when shared knowledge benefits adaptation.
This work illustrates how CPT and the LTE can complement traditional research approaches by making underlying process dynamics explicit and clarifying the mechanisms through which phenomena emerge. Situated within the domain of human-AI teaming, we further highlight its potential to inform and test the design of transparent AI agents that better align communication with team needs.