CodeX Speaker Series: To Catch a Thief With and Without Numbers
May 8, 2014 12:45pm - 2:00pm
Evidential reasoning is a notoriously complex task, and mistakes can have severe consequences. For instance, in the Netherlands, nurse Lucia de Berk was convicted to life imprisonment for murdering several patients. She served 6.5 years in prison, before a revision court decided that she was not guilty, now that the investigated deaths were no longer considered murder cases.
In the Lucia de Berk case, statistical reasoning about the number of deaths in a nurse's shifts played a notorious role. What happened in the case illustrates the communicative gap between fact finders and forensic experts. Whereas fact finders, typically jury members and judges, are best acquainted with a qualitative style of reasoning using arguments and scenarios, forensic experts, typically scientists, are focused on quantitative reasoning in terms of probabilities. Put differently, fact finders tend to reason without numbers, while forensic experts reason numerically.
In the presentation, a perspective on evidential reasoning is presented in which quantitative and qualitative techniques can be safely used side by side, choosing whichever is appropriate. It is shown how non-numeric evidential reasoning with arguments and scenarios can be safely integrated with numeric reasoning in terms of probabilities. It is concluded that there is no need to choose a normative framework focusing only on arguments, scenarios or probabilities, and that an integrating perspective is in place.
It is argued that a normative framework integrating arguments, scenarios and probabilities can not only help prevent avoidable mistakes, but also can be the basis for a new generation of rationality support software.
CodeX, Stanford University
Artificial Intelligence, University of Groningen
Bart Verheij is a tenured lecturer and researcher at the Institute of Artificial Intelligence and Cognitive Engineering (ALICE) of the University of Groningen. In 2013-2014, he is resident fellow at the CodeX center for legal informatics, Stanford University. He holds an MSc degree in Mathematics (University of Amsterdam) and obtained his PhD degree at Maastricht University, Faculty of Law, on a dissertation about the formal modeling of legal argumentation. His research focuses on argumentation, law and artificial intelligence. He currently leads a research project on argumentation, narrative and probabilistic approaches to reasoning with evidence, funded by the NWO Forensic Science program (2012-2016). In 2013, he was a visiting lecturer at the Institute of Logic and Cognition, Sun Yat-Sen University (Guangzhou, China). He served as program chair of the Fourteenth International Conference on Artificial Intelligence and Law (ICAIL 2013, Rome) and of the Fourth International Conference on Computational Models of Argument (COMMA 2012) in Vienna. He authored the monograph 'Virtual Arguments. On the Design of Argument Assistants for Lawyers and Other Arguers' (2004) and coedited the volumes 'Legal Evidence and Proof: Statistics, Stories, Logic' (2009) and 'Arguing on the Toulmin Model. New Essays in Argument Analysis and Evaluation' (2006). More information is available at http://www.ai.rug.nl/~verheij/.
Please click here to register.