This project arises out of a long-standing collaboration between Philip Dawid and William Twining on issues relating to the logical analysis and interpretation of evidence in courts of law. Probabilistic analyses have been developed to expose and clarify numerous paradoxes and subtleties attending the interpretation of forensic evidence, and given as testimony in court; and formal analyses and decision aids have been developed to solve complex problems of DNA identification. Although focused on legal applications, these analyses embody generic principles for representing and manipulating evidence. The aim of this particular project is to identify such generic aspects, to develop formal methods for expressing and manipulating them, and to explore their applications -- to legal, forensic and intelligence problems, but also more widely to the variety of problem areas being studied across the overall programme on “Evidence, Inference and Enquiry".

Modern technology allows for the collection of vast quantities of data of many different kinds, but the technology for combining, comparing, linking and interpreting all this information --- so turning it from information to evidence --- is almost non-existent.

Different disciplines conceive of and use evidence in different ways, but usually with little intellectual examination, and no conception that there might be an underlying generally applicable rational foundation. Unintelligent use of evidence is widespread and damaging. Even in the face of terrorist threats, training and practice in intelligence analysis largely ignore fundamental principles. In law enforcement there is scant appreciation of the import of missing evidence, while new evidence is sought to try and firm up a currently favoured theory, rather than to discriminate between credible alternatives. In forensic science, distinct types of evidence such as DNA, fingerprints, fibres, etc. are typically handled by different teams using different specialist methods. Similar inadequacies pervade decision-making in politics, medicine, public health, and commerce.

Understanding the nature and impact of evidence is a non-trivial and often counter-intuitive task. Evidence does not speak for itself, but has to be interpreted through the filters of models, assumptions and analyses. Generic attributes of evidence, that have to be accounted for by any comprehensive theory, include such aspects as accuracy, credibility, objectivity, relevance, provenance and weight. One item of evidence may corroborate another, or conflict with it, or explain away its apparent message. Items of evidence and hypotheses can form complex interrelated chains or webs, outstripping unaided human comprehension --- but the relevance and weight of any specific piece evidence can only be assessed in the light of its relation to the other evidence.

Although interpretation of evidence is as fundamental to all human enquiry as Aristotelian logic, and just as ancient, there has been little remarkably little attention paid to its fundamental structures and principles. Our aim is to begin to put this to rights, and to develop a general “substance- and subject-blind” approach to evidential reasoning.

Publications / other docs