Employing Bayesian Probability Theory to Diverse Applications Relevant to Archaeology

Summary

The principle of equifinality describes a system where an end state may be reached from a variety of conditions and in a variety of ways and has proved to be a confounding element in several areas in archaeology. Archaeological data commonly occur in both qualitative and quantitative form and Bayesian modeling, coupled with modern computational routines, permits multiple data types to be incorporated into a single synthetic probability model. The Bayesian approach makes probability statements given observed data, constructing posterior probability statements about unknown model parameters including unknown (unobserved) data.

Here we describe how Bayesian inference offers a solution to several areas relevant to archaeology. We use a Bayesian algorithm to make categorical assignments for unknown archaeological samples in three analytically similar contexts: identifying bone surface modifications, distinguishing between heated and untreated silcrete, and distinguishing the chemical signatures from discrete volcanic eruptions. In each example, we use large samples of observed reference data to train the respective models. Out-of-sample cross validation is then used to assess model performance and predictive ability before analyzing archaeological samples. Monitoring posterior distributions of unobserved data result an assignment of probability associated with individual unknown (archaeological) samples, thereby formally addressing the issue of equifinality.

This Resource is Part of the Following Collections

• Society for American Archaeology 83rd Annual Meeting, Washington, DC (2018)

Keywords

Spatial Coverage

Record Identifiers

Abstract Id(s): 20015