This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5).
Game theoretic models provided a simple framework for the analysis of conflict during the cold war when the objectives of the actors (e.g. the United States and the Soviet Union) were relatively well understood. Since the end of the cold war, threats to national security have become unpredictable. In this context traditional game theory provides little insight into predicting the strategies adversaries use. One requirement of game theory that weakens its utility in the present environment of high uncertainty is that the payoffs -- the costs and benefits of particular decisions -- are assumed to be known. This assumption is problematic when little information is available about the specific goals and preferences of actors, or when these goals and preferences change frequently. A second problem is that game theory models are typically loosely based on data and therefore are not very predictive or valuable in suggesting real world interventions. Despite these limitations, no robust, alternative mathematical theory for studying conflict and predicting actor decision-making exists. Given a growing sophistication in data collection and modeling dynamical systems, this project aims to develop an empirically grounded network conflict theory. This includes new mathematical tools and models to describe and predict conflict propagation and effective conflict management.
Two types of tools will be developed in this project. One set of tools, called Inductive Game Theory (IGT), can be used to extract directly from conflict event time-series data (e.g. the sequence of bombings in Iraq and US responses, including location and severity data) the strategies adversaries are playing without having to posit payoffs or to assume stability. The second set of tools, called Conflict Immuno-Kinetics (CIK), takes as input conflict variables (rate of spread, number of actors, structure of actor interaction, etc.) and uses this information to predict which intervention strategies (e.g. pacifying interventions, policing interventions rooted in force, or sanctions) are most likely be effective given properties of the conflict itself. The CIK models that in this project are developed to study conflict at the behavioral level are inspired by the response of the immune system to pathogens.
IGT and CIK tools and models will be developed and tested using model systems of large, heterogeneous, captive primate societies. Non-human primates can provide a model system for the development and testing of these tools. Data from the Yerkes National Primate Reserach Center on the macaque genus will be used because its social organizations have properties of broad interest. These model systems possess the "minimum degree of relevant complexity" thereby ensuring that the methods, and the insights generated, can be generalized to a diverse set of conflicts including those at the human level.