So You Think You Can Model? A Guide to Building and Evaluating Archaeological Simulation Models of Dispersals

Abstract

With the current surge of simulation studies in archaeology there is a growing concern for the lack of engagement and feedback between modellers and domain specialists. To facilitate this dialogue I present a compact guide to the simulation modelling process applied to a common research topic and the focus of this special issue of Human Biology-human dispersals. The process of developing a simulation is divided into nine steps grouped in three phases. ThePre-print version. Visit digitalcommons.wayne.edu/humbiol after publication for the final version.conceptual phase consists of identifying research questions (step 1) and finding the most suitable method (step 2), designing the general framework and the resolution of the simulation (step 3) and then by filling in that framework with the modelled entities and the rules of interactions (step 4). This is followed by the technical phase of coding and testing (step 5), parameterising the simulation (step 6) and running it (step 7). In the final phase the results of the simulation are analysed and re-contextualised (step 8) and the findings of the model are disseminated in publications and code repositories (step 9). Each step will be defined and characterised and then illustrated with examples of published human dispersals simulation studies. While not aiming to be a comprehensive textbookstyle guide to simulation, this overview of the process of modelling human dispersals should arm any non-modeller with enough understanding to evaluate the quality, strengths and weaknesses of any particular archaeological simulation and provide a starting point for further exploration of this common scientific tool.All archaeologists start modelling the moment they step out of the excavation trench. We interpret the individual finds (e.g., pots, skeletons, buildings etc.) within certain frameworks (e.g., pottery typologies, bone taxonomies, architectural types etc.) and analyse sets of finds to detect population level patterns (e.g., cultural similarities, age profiles of bone assemblages, urbanPre-print version. Visit digitalcommons.wayne.edu/humbiol after publication for the final version. development, etc.) with strict analytical rigour. However, the interpretations of these patterns in terms of human behaviour and causality are predominantly built using natural language, i.e. they are constructed in the form of conceptual models that hypothesise which causal mechanisms might have led from actions of individual actors (the owners of pots, users of skeletons and inhabitants of buildings) to the detected population-level patterns. These causal relationships are often described as, for example, "culture A influenced culture B," "the dispersal reached area C," or "population D outcompeted population E." Although inferences like these are made on the basis of rigorously collected and analysed data, and are commonly built upon extensive research and a good understanding of multiple strands of evidence, they nevertheless represent a thought experiment and are therefore limited ...

Agent-Based Modeling, Scientific Reproducibility, and Taphonomy: A Successful Model Implementation Case Study

“…The incongruence between the original study and the replication derives from a miscommunication about sampling mechanisms, and speaks to the need for clearer and more thorough documentation of computational models as advocated in calls for reproducibility (e.g. Marwick et al 2017;Romanowska 2015). Nevertheless, attempting the replication and identifying the reasons for discrepancies between the replication and original raised issues with theoretical implications that build on those raised by the original model.…”

Toy Story: Homophily, Transmission and the Use of Simple Simulation Models for Assessing Variability in the Archaeological Record

“…It is worth reminding, as many have done before us (e.g. Gerbault et al 2014;Romanowska 2015), that models are not depictions of reality, but formal thought experiments designed to test and explore specific research questions, a position explicitly acknowledged by our focus on toy models. At the risk of repeating methodological clichés, simplicity in modelling is paramount for-at least-two reasons.…”

Agent-Based Modeling for Archaeologists: Part 1 of 3

“…One class of computational models that has been used increasingly in archaeology over the past two decades is agent-based modeling. Researchers studying archaeological systems worldwide have adopted this formal modeling technique to approach their research questions (e.g., Cegielski and Rogers 2016;Kohler 2012;Linde and Romanowska 2018;Madella et al 2014;Perry et al 2016; Rogers and Cegielski 2017; Romanowska 2015;Wurzer et al 2015). Archaeologists use agent-based modeling to understand archaeological patterns across a range of temporal and spatial settings (e.g., Angourakis et al 2014;Balbo et al 2014;Morrison and Allen 2017;Perrault and Brantingham 2011;Premo 2015;Wren et al 2014).…”