Document Type


Publication Date


Publication Title

Argumentation in Artificial Intelligence and Law

First Page


Last Page



Argumentation, Argument Schemes, Critical Questions, Legal Reasoning Support Systems, Artificial Intelligence, Law


Two recent computational models of legal argumentation, by Verheij and Gordon respectively, have interpreted critical questions as premises of arguments that can be defeated using Pollock’s concepts of undercutters and rebuttals. Using the scheme for arguments from expert opinion as an example, this paper evaluates and compares these two models of critical questions from the perspective of argumentation theory and competing legal theories about proof standardsfor defeating presumptions. The applicable proof standard is found to be a legal issue subject to argument. Verheij’smodel is shown to have problems because the proof stan-dards it applies to different kinds of premises are “hard-wired” into the system. Gordon’s model overcomes these problems by allowing different proof standards to be assigned to each issue and by supporting arguments about proof standards within the same framework. These differences are minor however compared to the insight gained from these models jointly about the theory of argument schemes and critical questions. They show how schemes can be used to implement tools for constructing arguments, and not just for classifying arguments ex post facto, and help clarify how critical questions confound declarative knowledge about conditions for using argument schemes with procedural knowledge about how to evaluate and criticize arguments made using these schemes.