MAGICAL Computer Simulation of Mesolithic Foraging

The MAGICAL (Multi-Agent Geographically Informed Computer AnaLysis) software described in this chapter was designed to integrate two of the most important computational methods used by archaeologists during the last decade: Geographical Information Systems (GIS) (e.g., Allen et al. 1990) and multiagent simulation (e.g., Lake 1995; Mithen 1990). At the outset of model development in 1995, it was recognized that GIS provide archaeologists with a sophisticated means of manipulating spatial data, but offer limited support for modeling change through time. Conversely, multiagent simulation models have allowed archaeologists to study change through time, but have either lacked or had simplistic spatial components. Consequently, the research described here aimed to combine the strengths of GIS and multiagent simulation in one software package so as to better facilitate the quantitative study of spatiotemporal variability in the archaeological record. The MAGICAL software was developed within the broader context of the Southern Hebrides Mesolithic Project (SHMP). This project was established in 1988 by Dr. Steven Mithen (University of Reading) to acquire new data from the Scottish Islands of Islay and Colonsay and, by integrating this with existing data, to develop a regional perspective on the early postglacial settlement of Western Scotland (Mithen and Lake 1996). The construction of a computer simulation model was considered a fundamental part of the postexcavation studies of the SHMP (Lake in press). It was hoped that conceptual models which would otherwise remain largely intuitive could be more rigorously explored by formalizing them into mathematical algorithms, translating those algorithms into computer code, and then running simulation experiments. This chapter describes how the MAGICAL software integrates GIS and multiagent simulation. It does so directly in section one and then by example in sections two, three, and four. Section two discusses the conceptual basis of the SHMP simulation model, and section three describes how this was implemented using the MAGICAL software. Section four presents the results of the SHMP simulations. Note that the SHMP simulation model is discussed primarily as a means of demonstrating the capabilities of the MAGICAL software. Those interested in the wider background to this particular modeling endeavor are urged to consult Mithen (ed., in prep).