Cognitive Hierarchy in Day-to-Day Network Flow Dynamics.

The article focuses on developing a day-to-day traffic behavior modeling framework based on cognitive hierarchy (CH) theory to better capture travelers' strategic route choice behaviors over repeated daily trips. Extending two established models—the network tatonnement process (NTP) dynamic for deterministic user equilibrium (DUE) and the logit dynamic for stochastic user equilibrium (SUE)—into this framework, the study accounts for heterogeneous travelers with varying levels of strategic reasoning who form beliefs about others’ decision-making. Calibration with data from a virtual route choice experiment demonstrates that the CH-NTP model significantly improves fit compared to traditional models lacking strategic prediction. The analysis reveals the existence of multiple equilibria, including non-DUE states arising from extensive strategic anticipation, which can either improve or degrade overall system efficiency. Stability analyses show that accurate prediction by higher-level travelers preserves local stability near DUE, while overprediction or underprediction of others’ behaviors affects the size and structure of stable regions differently for both deterministic and stochastic dynamics.