Helping with inquiries: theory and practice in forensic science

Abstract

This thesis investigates the reasoning practices of forensic scientists, with specific focus on the application of the Bayesian form of probabilistic reasoning to forensic science matters. Facilitated in part by the insights of evidence scholarship, Bayes Theorem has been advocated as an essential resource for the interpretation and evaluation of forensic evidence, and has been used to support the production of specific technologies designed to aid forensic scientists in these processes. In the course of this research I have explored the ways in which Bayesian reasoning can be regarded as a socially constructed collection of practices, despite proposals that it is simply a logical way to reason about evidence. My data are drawn from two case studies. In the first, I demonstrate how the Bayesian algorithms used for the interpretation of complex DNA profiles are themselves elaborately constructed devices necessary for the anchoring of scientific practice to forensic contexts. In the second case study, an investigation of a more generalised framework of forensic investigation known as the Case Assessment and Interpretation (CAI) model, I show how the enactment of Bayesian reasoning is dependent on a series of embodied, experiential and intersubjective knowledge-forming activities. Whilst these practices may seem to be largely independent of theoretical representations of Bayesian reasoning, they are nonetheless necessary to bring the latter into being. This is at least partially due to the ambiguities and liminalities encountered in the process of applying Bayesianism to forensic investigation, and also may result from the heavy informational demands placed on the reasoner. I argue that these practices, or 'forms of Bayes', are necessary in order to negotiate areas of ontological uncertainty. The results of this thesis therefore challenge prevailing conceptions of Bayes Theorem as a universal, immutable signifier, able to be put to work unproblematically in any substantive domain, Instead, I have been able to highlight the diverse range of practices required for 'Bayesian' reasoners to negotiate the sociomaterial contingencies exposed in the process of its application.