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

Abstract

There is a growing use of agent-based model (ABM) simulations to reconstruct past human-environment interactions. ABMs are useful in that they offer scientists the opportunity to model processes, phenomena, and study systems that may not be otherwise reproducible or testable. Replication or re-implementation studies of ABMs are, however, infrequently undertaken, and there are few examples within archaeology or other social sciences. This paper documents the process of a successful ABM replication study, as well as two additional modifications to the original model. Results corroborate the findings of the original geoarchaeological model and indicate that episodic geomorphic events significantly affect archaeological deposit formation and the inferences drawn from associated radiocarbon records. One revision of the model further demonstrates that episodic fluvial events can create highly varied radiocarbon distributions. The second modification illustrates that excavation data helps to fill in hiatuses in radiocarbon chronologies on depositional landforms, although there is no effect across landscapes subject to erosion. This successful replication exercise also illustrates the value of open access data and analyses in reproducing, testing, and expanding upon archaeological research and theory building.