Agent-based Modeling and Simulation in Archaeology

The Artificial Anasazi (AA) agent-based simulation model was one of the first agent-based models to address archaeological questions. It grew out of a unique interdisciplinary collaboration between archaeologists/anthropologists (George Gumerman, Jeffrey Dean, Alan Swedlund), computational social scientists (Robert Axtell, Joshua Epstein), and computer modelers (Steve McCarroll, Miles Parker). This chapter describes the kinds of archaeological questions the AA model was designed to address and the ability of the model to answer those questions, as well as an extension of the model that incorporates individual-level age-specific fertility and mortality rates (the Artificial Long House Valley, or ALHV model). We also provide a brief overview of other archaeologically oriented agent-based simulation models and describe how they relate to the AA model. We conclude with a discussion of future directions for the AA and ALHV models as well as suggestions for further research on archaeological questions for which agent-based computer simulations may be especially suitable.

Historical hypotheses are usually very vaguely worded concepts trying to satisfy mankind’s curiosity about its own planet’s past. Although time can never be turned back again in order to make them a proven fact, some of these hypotheses are scientifically rather doubted and some are not as they are confirmed by historical findings. If findings are missing, arguments for the scientific credibility of hypotheses are usually very sparse. One very modern way to at least have the possibility to falsify wrong theses is computer modelling and simulation. Due to exponentially increasing performance of computers, nowadays more and more complex models can be simulated in shorter time with less efforts. Thus so called agent-based models requiring lots of memory and fast computation, are getting more and more popular also among archaeologists. The following chapter explains the basics of agent-based modelling and gives examples showing their scientific value.

This chapter gives an outline of the activities that need to be done in order to make a model reproducible. Starting with a wholistic view of the whole project lifecycle, aspects such as parameter and output definition, documentation as well as verification and validation are examined. Our goal is to take a step back and look at the variety of tasks that can be done to achieve reproducibility, which is not an end in itself but a step towards achieving credibility of a model.

The philosophy of geosimulation is to model spatial processes either as auto-correlative physical geographical processes or as cross-correlative socio-spatial processes. This chapter refers to the latter and stresses that conflating social and spatial phenomena in time needs to consider scale, in all three dimensions, as an explicit distinctive factor to adequately understand the interlinked processes at hand. With respect to residential segregation modeling, it is illustrated that agent-based simulation models mainly focus on the individual, local, and short-term scale in order to detect the generative mechanisms which lead to socially, spatially, and temporarily large-scale structures. The chapter argues that, although leading to thought-provoking results, this bottom-up approach should be complemented by an explicit macro-scale implementation of social and spatial structures. In so doing, social and regional circumstances for ancient societies can be analyzed more comprehensively. Contemporary agent-based modeling and geographic information systems’ software tools are capable of incorporating such complexities.

A current trend of Agent-Based Modeling (ABM) is the constant increase in computing resources used by the simulations. The complexity of the behaviors being explored forces this tendency, as ABMs are computationally expensive pieces of code. Different solutions are being developed to exploit the power of High-Performance Computing (HPC) to solve this bottleneck, and allow for the execution of large scale realistic simulations. These problems seem unrelated to the type of models being developed in archaeological research, focused on small-scale societies, but some of the issues of the discipline are related to this constraint on computing resources. This chapter explores the benefits and issues created by the introduction of High-Performance Computing in archaeological ABMs. It discusses the problems of distributing a model in different computers, as well as potential solutions and pitfalls provided by these systems. The text shows how HPC can contributed no only to an improvement on performance, but also in the task of solving methodological issues that are currently being debated by the community.

Mining structures are among the most complex economic systems in prehistory. Until recently, research into prehistoric production processes has strongly focused on technological reconstruction. The complexity of production processes, their interconnectedness with the surrounding socioeconomic network and issues of quantification have, quite regrettably, been addressed to a much lesser extent. Simulation can contribute important insights into the latter problem areas, which is why we have been applying it to the prehistoric salt mines of Hallstatt (Upper Austria) during the past years. In this chapter, we report on these simulation efforts and show how a multi-level simulation approach can augment archaeological understanding during everyday research work.

This chapter proposes a model of long-term changes in human settlement pattern to identify possible generative processes behind empirically observed fluctuations in their rank-size pattern. The model assumes that per-capita fitness is a combination of the beneficial effect derived from the presence of other individuals in the same group and the finite amount of resource available locally. The former can lead to a positive frequency dependence, whereby an increase in the group size determines an increase in fitness; the latter can have detrimental effects once a group exceeds a certain size and the energetic demand becomes larger than what is available. Given that: (1) individuals have the possibility to modify their condition through group fission and fusion; and (2) the amount of resource is unlikely to be constant over space and time; we can expect fluctuations in the settlement pattern as a function of a non-linear relationship between size and fitness, as well as internally or externally induced variations in the amount of available resources. The simulation model will examine these scenarios and identify how different variables can induce different types of long-term equilibria in settlement systems.

Some of the most significant settlements of late Iron Age Europe were founded in agriculturally marginal landscapes. The specific locations caused their food production potential to be regarded usually as deficit in terms of sustainability. Such notion, however, can be re-examined with help of new methodological tools. In order to capture the dynamics of the agro-pastoral economy processes in recurrent year-to-year cycles, this chapter exploits the objective advantages and limits of coupled GIS environmental and agent-based social modelling approaches. Three consecutive models are presented—the population dynamics in The Population Model, and the sustainability of the land-use strategies in The Crop Production Model and The Field Allocation Model. Models are firmly based on authentic archaeological and environmental record with the region around long-term investigated oppidum of Staré Hradisko (Czech Republic) used as the case study. Results obtained with the simulation demonstrate limits of the sustainable economy practiced by a constantly growing population under particular environmental settings. The immediate or gradual impact of the success rate in food production and its potential influences on the social processes including the oppida abandonment are also addressed.

In the last 40 years, the very idea of ethnicity has evolved from a static and essentialist classification of human groups according to their immutable “nature” to a relational frame of reference used by groups of people to consider themselves “similar” or to be explicitly differentiated by others. Nevertheless, the growing importance of variability analysis of mitochondrial DNA and other biological markers in modern prehistoric studies, with their emphasis on the identification of geographic patterns in genetic and phenotypic diversity of prehistoric populations is going in the opposite direction, as if the existence of genetic variability in the past would be comparable to what is inferred about cultural variation in the present. In this paper we have built a computer simulation of economic processes causing social aggregation, territoriality and ethnogenesis among Patagonian hunter-gatherers. We argue that cultural similarity and the constriction of groups to a restricted geographical area are not necessarily ethnic markers. Our model suggests that the more inter-generational knowledge transmission among socially aggregated individuals in the past, the greater the similarity in the social activity performed by agents in the present, and the same for their territoriality and the way frontiers and social networks were negotiated. Our computer simulation intends to answer the question “Why did human groups modify their traditional residence mobility and dispersal patterns?” In ancient Patagonia, at the extreme south of South America, from 12,000 BC until nineteenth century AD, this social transformation seems to coincide with slow changes in subsistence economy and technology. However, there are historical and archaeological sources that suggest this process was related with increased social complexity: wars and conflicts between different indigenous groups would have preceded this new scenario, even before European contact.

In some regions of the disciplines, highly abstract models such as Sugarscape are now seen as a new kind of whole society model better able to help us understand human societies than their realistic and detailed predecessors. I trace the evolution of archaeological simulation from its realist-generalist roots to its abstract generalist present. Initial models focused on particular archaeological contexts, but their aim was to create general explanations of human social and archaeological phenomena. Highly realist and particularist models soon followed that prioritized detailed understanding of a particular case for its own sake. Models are now emerging that seek to create general explanations of human phenomena and that use completely abstract contexts that are not tied to any particular archaeological case. Whole society modeling was initially seen as requiring very realistic portrayals of particular contexts.