Agent-Based Analysis and Simulation of Meta-Reasoning Processes in Strategic Naval Planning Mark Hoogendoorn 1 , Catholijn M. Jonker 3 , Peter-Paul van Maanen 1,2 , and Jan Treur 1 1 Vrije Universiteit Amsterdam, Dept. of Artificial Intelligence, De Boelelaan 1081, 1081 HV Amsterdam, The Netherlands {mhoogen, treur}@cs.vu.nl 2 TNO Human Factors, Dept. of Human in Command, P.O. Box 23, 3769 ZG Soesterberg, The Netherlands peter-paul.vanmaanen@tno.nl 3 Delft University of Technology, Dept. of Man-Machine Interaction, Mekelweg 4, 2628 CD Delft, The Netherlands C.M.Jonker@tudelft.nl Abstract. This paper presents analysis and simulation of meta-reasoning processes based on an agent-based meta-level architecture for strategic reasoning in naval planning. The architecture was designed as a generic agent model and instantiated with decision knowledge acquired from naval domain experts and was specified as an executable agent-based model which has been used to perform a number of simulations. To evaluate the simulation results, relevant properties for the planning decision were identified and formalized. These properties have been validated for the simulation traces. Keywords: Meta-reasoning, simulation, planning, intelligent agent systems. 1 Introduction The management of naval organizations aims at the maximization of mission success by means of monitoring, planning, and strategic reasoning. In this domain, just as well as in all other domains that are characterized by their resource-boundedness, plan generation and action selection are supported by strategic reasoning; for example, it may help to determine to decide whether a go or no go should be given to a certain possible plan after an incident, whether this should be investigated further, or even whether attention should be shifted to a different plans altogether. An incident is an unexpected event, which results in an unmeant chain of events if left alone. Strategic reasoning in a planning context can occur both in plan generation strategies (cf. [21]) and plan selection strategies (cf. [10, 20]). The above context gives rise to two important questions. Firstly, what possible (candidate) plans are there to be considered? And secondly, what criteria should one use in order to come to a set of possible candidate plans and what criteria should one use to select a certain plan from such sets for execution? In resource-bounded