Adaptive decision making in agent-based simulation

In the context of adaptive and autonomous decision making, a computational model is needed to formalize and implement a practical goal deliberation mechanism that determines how goals can be evaluated, adopted, or rejected. Such a model is expected to lead to self-explanatory decision making, which is essential for understanding and influencing the behavior of autonomous agents and to pave the way to generate desirable macro-level behavior in self-adaptive, self-organizing systems. This article introduces an adaptive decision making architecture for agent-based simulation by promoting deliberative coherence and its extensions for decision making under uncertainty. To this end, a deliberative coherence-driven agent model is presented, and then is extended with run-time monitoring mechanisms. The architecture enables us to evaluate goals and competing tasks to facilitate the selection of the most coherent tasks with respect to a given goal, whereas the successful completion of those tasks contributes to the achievement of mission objectives for systems in shifting, ill-defined, and uncertain environments.