Modeling the Effects of International Interventions with Nexus Network Learne

Nexus Network Learner is an intelligent agent based simulation used to study Irregular Warfare (IW) in several major studies at the Department of Defense (DoD). Heterogeneous autonomous agents, each with their own separated inductive learning mechanism, have initial attributes and behaviors in proportion to demographic groups in the simulated population, and learn new behaviors as they serve culturally based goals. Nexus agents create a dynamic role-based network, and learn how to choose partners as well as what behaviors they should have with their network partners. As Nexus agents coevolve, nexus models the emergence of social institutions from individual behaviors, the fundamental social aggregation challenge. Nexus models the formation of learned vicious and virtuous cycles of behavior, some of which have higher average utility for the agents than others, and can be used to test the effects of interventions on the natural motivationbased system. An experiment is presented that uses Nexus to model the vicious cycle of corruption in an African country, from the first Irregular Warfare Analytical baseline at the Office of the Secretary of Defense (Messer 2009). Nexus Network Learner: A Symbolic Interactionist Simulation Nexus Network Learner is part of the author’s research program on Symbolic Interactionist Simulation, that started with the first Cognitive Agent Social Simulation, the Sociological Dynamical System Simulation (SDSS) (Duong 1991, Duong and Reilly 1995). Symbolic Interactionist Simulation addresses the fundamental social aggregation challenge with coevolving autonomous agents that maximize the utility of their actions based on their individual perceptions, acting solely in their own self-interest. From these individual actions emerge institutions, or expected behaviors that are more than the sum of their parts. Institutions in Symbolic Interactionist Simulations are dissipative (dynamic) structures: vicious or virtuous cycles of corresponding behaviors that symbolic interactionist agents deCopyright © 2011, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. velop as they adjust to each other. SDSS concentrates on vicious cycles such as racism, social class, and status symbols. The Symbolic Interactionist Simulation of Trade and Emergent Roles (SISTER) demonstrates Adam Smith’s invisible hand in virtuous cycles such as price, a standard of trade (money), and a role-based division of labor (Smith 1994, Duong 1996, Duong and Grefenstette 2005). Symbolic Interactionist Simulations, as adaptive systems, model not only the emergence of social institutions from individual motivation, but the effect of institutions on individual motivation, a process known as immergence (Andrighetto et al. 2007) (See Figure 1). Individuals Institutions