Abstract for: Structural Feedback and Behavioral Decision Making in Queuing Systems: A Hybrid Simulation Framework

Traditional queuing models often ignore human judgment, yet empirical evidence shows human behavior significantly alter system performance. We propose a hybrid simulation approach to capture the full feedback loop between individual agents and aggregate system behavior. By modeling human responses as feedback control processes, we capture agents' objectives and their perception of system state, including delays and distortions, improving insights into how human decision-making affects system outcomes. Our simulation approach integrates three perspectives: ABMS to capture individual agents’ behavior, DES to track random discrete system events, and SD to model behavioral feedbacks between the system and agents. Implemented in Ventity, the simulation can accurately model the behavioral decision making and realistic processing capabilities of the system in pseudo-continuous time, facilitating consistent design and easy evaluation of a wide range of efficient policies aimed at improving system performance. Forrester and Sterman emphasized the need to capture “macrobehavior from microstructure” for decades, yet traditional SD has limitations. Our hybrid approach accurately represents individual decision-makers (microstructure), offering deeper understanding of system macro-behavior. Agents make decisions based on system state and their intrinsic goals, and the aggregated behavior of all agents in the system determines system performance. This integral approach provides a flexible framework for studying behavioral decision-making in queuing systems. Our hybrid simulation approach improves modeling realism by anchoring decision-making in behavioral decision theory, and accounting for agent heterogeneity. It enables formal structural analysis of emergent behavior and its root causes, helping identify design parameters for system improvement, and facilitates generalizable explanations of behavior and the analysis of complex decision-making dynamics in queuing systems. Future work should extend the application of our method to various empirical contexts.