On the nature of argument schemes

Since the 1980s, computer science, especially artificial intelligence (AI) has developed formal models of many aspects of argumentation that since the work of Toulmin and Perelman were thought of as belonging to informal logic. Doug Walton is one of the argumentation theorists who has recognised the relevance of this body of work for argumentation theory. One of the concepts on which recent work in AI has shed more light is that of argument schemes (sometimes also called ‘argumentation schemes’), which features prominently in Walton’s work. A study of argument schemes from the perspective of AI is therefore very appropriate for this volume in honour of his work. More precisely, the aim of this paper is to use insights from AI to propose an understanding of the nature of argument schemes as a means to evaluate arguments, and to compare this understanding with Walton’s own account of argument schemes. Walton regards argument schemes as essentially dialogical devices, determining dialectical obligations and burdens of proof. Using this account, a procedure for evaluating arguments should take the form of a set of dialogue rules. I shall instead argue that argument schemes are essentially logical constructs, so that a procedure for evaluating arguments primarily takes the form of a logic. More specifically, I shall argue that most argument schemes are defeasible inference rules and that their critical questions are pointers to counterarguments, so that the logic governing the use of argument schemes should be a logic for nonmonotonic, or defeasible reasoning. The dialogical role of argument schemes can then be modelled by embedding such a logic in a system for dialogue, so that in the end argument evaluation with argument schemes is a combination of logical and dialogical aspects. However, after having developed this account, I shall also argue that not all argument schemes naturally fit the format of defeasible inference rules and that there is an often overlooked distinction between two types of argument schemes. One type fits the model of defeasible inference rules, i.e., elements of a reasoning method, but another type can better be seen as a reasoning method in itself. To model reasoning with such argument schemes, an account is needed of how reasoning methods can be combined, and I shall argue that this goes beyond the usual nonmonotonic logics and dialogue systems.