The Policy Arguer: A computational model of United States foreign policy belief systems, 1949-1960

Abstract

One of the more fruitful systematic approaches to the study of foreign policy views nations as information processors and explains foreign policy as a function of information and processes on that information. Much of the processing of input information depends on the prior beliefs of decision makers, but, for a variety of reasons, these systems of prior beliefs defy "standard" empirical and modeling techniques. Computational modeling, including expert system technology, provides an alternative methodology that allows systematic examination of the impact of complex belief structures on information processing. The Policy Arguer (POLI) is an expert system that represents belief systems in the U.S. foreign policy making environment. In particular, current work focuses on the decade of the 1950s. POLI is validated by comparing its output to the historical trace of U.S. actions in Asia for this period. POLI proves to be remarkably accurate in its predictions of U.S. behavior for this domain, succeeding for 86% of the events in the historical trace. The simulated "policy debate" within POLI is then analyzed in an effort to explain the choices made.