Dynamic, Agent-Based, Societal Model for Factional Adversarial Games

Physical Sciences Inc., proposes to develop a novel, dynamic, agent-based, societal model for studying coalition formation and factional competition. Our unique approach uses a detailed representation of human behaviors and environmental elements pertaining to coalition formation, strategic decision-making and factional competition/cooperation: geography, culture, religion, language, individual and collective emotions, history, and resource availability and needs. The proposed model consists of: (1) a global environment (World) and subsequent geographical regions: Spaces, Sites, Districts, and Villages, which are hierarchically arranged; (2) individual Villages that consist of tribes and resources; and (3) human agents endowed with individual and collective learning and decision-making capabilities. Phase I will: (i) develop the proposed societal model; (ii) develop a Java agent-based software simulation tool for the model, using Ethos as societal description and representation framework and RePast as simulation platform; (iii) Develop rule-based and game-tree search factional decision-making algorithms for controlling the brains of the modeled agents; and (iv) simulate a scenario of factional adversarial game based on the Afghan tribal structure and demonstrate both qualitative and quantitative prediction capacities of the proposed model.