One of the major challenges in human behavior modeling for military applications is dealing with all factors that can
influence behavior and performance. In a military context, behavior and performance are influenced by the task at hand,
the internal (cognitive and physiological) and external (climate, terrain, threat, equipment, etc.) state. Modeling the
behavioral effects of all these factors in a centralized manner would lead to a complex rule-base that is difficult to
maintain or expand. To better cope with this complexity we have developed the Capability-based Human-performance
Architecture for Operational Simulation (CHAOS). CHAOS is a multi-agent system for human behavior modeling that is
based on pandemonium theory. Every agent in CHAOS represents a specific part of behavior, such as 'reaction to threat'
or 'performing a patrol task'. These agents are competing over a limited set of resources that represent human
capabilities. By combining the element of competition with multiple limited resources, CHAOS allows us to model
stress, strain and multi-tasking in an intuitive manner. The CHAOS architecture is currently used in firefighter and
dismounted soldier simulations and has shown itself to be suitable for human behavior and performance modeling.