How we think systemically

Systems thinking is essential for understanding and solving problems in complex systems, yet its influence on conceptual understanding and the underlying neural mechanisms remain poorly understood. Here, we developed an AI-driven multi-level systems modeling environment to experimentally manipulate the hierarchy of information accessibility during collaborative learning, exposing dyads to full- (macro–meso–micro), dual- (macro–meso), or single-level (macro) representations of a simulation system—“evolution by natural selection”, respectively. While dyads engaged in collaborative learning of evolution, we recorded their brain activity using fNIRS hyperscanning. Structured access to hierarchical systems modeling systematically shaped how learners interpreted and understood the complex system phenomenon, improving conceptual understanding and collaborative processes. Critically, these effects were supported by interbrain alignment in the dorsolateral prefrontal cortex and by flexible transitions between dynamic interbrain synchrony states. Together, these findings suggest that multi-level systems modeling scaffolds collaborative systems thinking by coordinating cognitive and neural processes across individuals.