Burden of Proof in Deliberation Dialogs

The literature in argumentation and artificial intelligence has distinguished five types of burden of proof in persuasion dialogs, but there appears to have been no serious investigation so far on how burdens of proof should be modeled in deliberation dialogs. The work in this paper is directed toward filling that gap by extending existing formal models of deliberation dialog to analyze four examples of deliberation dialog where burden of proof is at issue or poses an interesting problem. The examples are used to show (1) that the eight stages in the formal model of Hitchcock, McBurney and Parsons (2007) need to be divided into three more general stages, an opening stage, an argumentation stage and a closing stage, (2) that deliberation dialog shifts to persuasion dialog during the argumentation stage, and (3) that burden of proof is only operative during the argumentation stage. What is shown in general is that deliberation is, in the typical type of case, a mixed dialog in which there is a shift to persuasion dialog in the middle. Both in argumentation studies as an interdisciplinary domain and in artificial intelligence, the type of dialog that has been most intensively studied so far is that of persuasion dialog. In this type of dialog, there is some claim at issue, and the object of the dialog is to prove or disprove that claim. Deliberation has a different kind of goal. It is to solve a problem about what course of action to take. The problem statement is not a proposition, but a question, called a governing question by McBurney, Hitchcock and Parsons (2007). Examples of these are: ‘Where should we go to dinner ?’ and ‘How can we provide all Americans with health care insurance?’. The goal of a deliberation is to find a solution to a common problem. Unlike persuasion dialog, there are no winners and losers. Everyone wins if the dialog is successful. Does burden of proof have a place in this type of dialog? It seems so, because arguments go back and forth in a deliberation dialog, and once an argument is brought forward, like ‘Ricardo’s is the best place to go for dinner, because their food is organic’, it requires evidence to back it up if it is challenged. It appears then that understanding how burden of proof works in it is an important step in the study of deliberation dialog as a form of group decision-making. There is a growing literature on burden of proof in argumentation (Kauffeld, 2003) and in work on formal dialog models in artificial intelligence (Prakken, Reed and Walton, 2005; Prakken and Sartor, 2006, 2007, 2009; Gordon, Prakken and Walton, 2007). Importantly, this work has distinguished several types of burdens in persuasion dialog, as opposed to the widely accepted traditional assumption that there is a single concept of burden of proof. There is also a recent literature on formal models of deliberation dialog (Tang and Parsons, 2006; McBurney, Hitchcock and Parsons, 2007). However,