Over the last 10 years there has been a marked increase in interest in simulation in the social sciences which appears to be linked to two developments, which are themselves related. The first is a cross-disciplinary interest in the sciences of complexity, which promote a view of the world in which most important phenomena emerge from the interaction of many agents (physical, biological, or social) in systems that are rarely at equilibrium. Moreover, the complexity of agent interactions often gives rise to significant nonlinear behavior in the system of which they are part, making it difficult to understand system-level behavior. This vision in turn promotes a method—agent-based modeling—that provides a computational environment in which the behaviors of such systems can be studied.
Agent-based simulation systems are themselves made possible by advances in object-oriented programming languages (such as Smalltalk, Objective-C, and Java) over the last 15 years. Older languages, such as FORTRAN, promoted a way of thinking about the world as composed of processes. Object-oriented languages, on the other hand, enforce a programming discipline in which it is natural to think in terms of objects that contain both variables and functions (or methods) for interacting with other objects via messages. From this structure it is a relatively short hop to contemporary agent-based modeling libraries such as Swarm, RePast, and Ascape.
Herbert Simon, for example in his 1969 The Sciences of the Artificial; Thomas Schelling in his 1978 Micromotives and Macrobehavior, and Robert Axelrod, in his 1984 The Evolution of Cooperation, can now be recognized as key pioneers in these approaches. Joshua Epstein and Robert Axtell’s 1996 book Growing Artificial Societies: Social Science from the Ground Up provides a clear, more up-to-date argument that agent-based models will lead to interesting applications in the social sciences. Indeed, applications of ABM are now quite common at many general social science conferences. In the last few years several meetings (e.g., a series cosponsored by the University of Chicago and Argonne National Laboratory), organizations, and journals devoted to developing these methods, have appeared. Research using ABM is not confined to the social sciences or to the USA; see, for example, the range of activity in Europe.
Perhaps the most important thing these systems can do for archaeology
is to provide a means for us to discover candidate processes (that we don't
know) that could have generated the patterns we see in the archaeological record.
They also help us move away from an emphasis on variables towards a way of thinking
about the archaeological record that emphasizes interactions among agents. Finally,
to the extent that agents have the possibility of changing their rules for behavior
through time based on the success of those rules, we are able to bring to bear
the advantages of both the selectionist and human behavioral ecological perspectives
on a study of the archaeological record.