We use the case of a “whole-system” change program in a national healthcare system to empirically examine the multilevel dynamics underlying organizational adaptation. Our analysis demonstrates how the cognitive distance between agents’ causal representations affects opportunities to cooperate in hierarchical systems. Using complexity theory, we identify a scale-invariant causal pathway that can be applied recursively across many organizational levels. At each level, three coupled feedback loops determine how local agents modify their cognitive representations to include uncovered interdependencies and synchronize their adaptive search across organizational boundaries: a “boundary work” loop, a “small wins” loop, and a “parochialism” loop. Our results also point to the scale-dependency of the strength of dissipative processes across levels. These novel results further develop the theory of organizational change and have practical implications for large multilevel organizations, especially regarding the sustainability of improvements.