Artificial Intelligence and Law 8: 277–281, 2000. © 2000 Kluwer Academic Publishers. Printed in the Netherlands. 277 Book Review Douglas Walton, Appeal to Expert Opinion – Arguments from Authority. Penn State Press, Pennsylvania, 1997. 281 pages. ISBN 0-271-01695-7. 1. Dialectics Dialectics has become quite popular in AI and Law research in recent years. Tom Gordon’ Pleadings Game (1993,1995) was one of the first dialectical models of legal dispute resolution. Since then a number of dialectical models, often in the guise of dialogue games, have been developed (e.g., Hage et al., 1994; Lodder, 1998; Prakken and Sartor, 1998). These models are formal (computational) models of legal dialogues. Because they are formal models, the main focus has been on the formal properties of legal dialogue and as a result they have abstracted from legal reality. This abstraction makes them somewhat unrealistic. They have difficulties in coping with phenomena characteristic of practical legal dialogues, such as the way the burden of proof is handled in the distinct legal procedures. Within AI and Law there are also accounts of actual cases and actual dialogues that shed light on legal dialogue in action (e.g., Rissland and Ashley, 1987; Berman and Hafner, 1995; McCarty, 1995; Leenes, 1998). There appears to be a tension between the formal computational models of dialogue (abstract) on the one hand and the informal models (more ‘natural’ dialogues) on the other. Both formal and informal studies of legal dialogues are needed to improve legal dialogue models. The importance of combining, or at least using, informal and formal accounts is also acknowledged by the more formally inclined, such as Prakken. In his (1999) he uses reported cases in order to refine his notion of burden of proof. A source of informal accounts of (legal) dialogue can be found outside the field of AI and Law, most notably in philosophy and linguistics. The studies in philosophy and linguistics are often more substantive and casuistic than formal. And, although the topics studied in AI and Law and linguistics/philosophy are similar, there seems to be little communication between both communities. As a result, valuable insights are overseen. This book-review bridges one of the gaps between the two worlds. It addresses the relevance of some of the work done in philosophy/linguistics for AI and Law. 278 BOOK REVIEW 2. Argumentum ad Verecundiam A leading scholar in the field of argumentation is Douglas Walton. In a number of books, he has studied phenomena central to argument and dialogue, such as fallacies and commitment. Central to his work is the claim that arguments should be assessed in the context of a dialogue. This makes his work of particular interest to dialectics in AI (and law). Recent books in his series are Appeal to Expert Opinion (1997) and Appeal to Popular Opinion (1999). In this review I discuss Appeal to Expert Opinion because it bears most on legal dialogues. Appeal to Expert Opinion studies the use of arguments from authority in dialogues. Arguments from authority are sometimes clearly reasonable. In an argument about the accomplishments in the field of AI and Law, I may for resort to the opinion of an expert to underpin my claim that ontologies are important. I could for instance say that Pepijn Visser says that André Valente’s functional ontology in his dissertation is an important contribution to the field of AI and Law. This argument is reasonable since Pepijn Visser is an expert in the field of legal ontologies. On other occasions, arguments from authority are fallacious in the sense that they appeal to the feeling of reverence or respect we have of those in authority, especially if the argument itself is a bad one. Examples of the latter use of argument from authority can frequently be seen in commercials. (Someone who looks like) A dentist may promote the use of X because it prevents plaque, or someone else may say that this dentist says to use X because it prevents plaque. This kind of testimonial may be a clear example of a fallacy, but it is not hard to imagine examples in which this is less clear. The trouble with arguments from authority is, of course, that it is often hard to tell if the expert opinion is indeed valid. Is Pepijn Visser really an expert in this field, and is his statement true? The question more generally, is: How can one rationally evaluate an appeal to expert opinion if oneself is not an expert in the given field? This puzzle is essentially the one Walton addresses in Appeal to Expert Opinion. The question is relevant for legal dialectics because expert opinion is often used in court cases. Genuine expert opinion could be a point in a dialogue at which the opponent has no further ways of questioning a proponent’s claim. But now I am getting ahead of myself. Let us first explore Walton’s book in some detail. In logic textbooks, argument from authority is often classified as a fallacy (argumentum ad verecundiam). A reason why it is thought to be a fallacy is because the one who resorts to an argument from authority may reply to an attack by his opponent with the classic put-down: “well, you’re not an expert, are you”. If an argument from authority is used in this way, it quite effectively immunises the argument to criticism. Since not every appeal to expert opinion is fallacious, the question is: When does appeal to authority constitute a fallacy and when is it a genuine argument? Walton answer this question by first giving an overview of the historical 1 See Pepijn Visser’s review of André Valente’s dissertation in AI and Law, Vol 7(4), 1999, pp. 364–375. BOOK REVIEW 279 development of argument from authority. He points out that Locke’s account of Argumentum ad verecundiam (1671) gives a good overall picture of how this type of argument, in principle a reasonable one, can be used as a sophistical tactic to take advantage of the modesty of the opponent. In Locke’s view argument from authority is not fallacious per se, but can be used in a way that makes it fallacious. Later accounts place argument from authority into the corner of fallacies, and hence argument from authority is suspect. The next step Walton takes, is to identify the type of argument in more detail. He divides the argument from authority in a number of subcategories: epistemic authority – position to know –, tradition, celebrity and deontic authority – legal, religious and administrative –. Henceforth Walton uses the expression Argumentum ad verecundiam only to denote the fallacious instances of the argument. Furthermore he focuses on expert opinion: epistemic authority. He proceeds by identifying the general form of the argument from authority. This chapter also contains a somewhat weak expose about expert systems and the way they make use of expert knowledge. The sources Walton uses are heavily outdated (1982–1986) and it is not entirely clear what it adds to his argument. The general form of an appeal seems, not surprisingly, to be: E is a genuine expert in S E asserts that A A is within S A is consistent with what other experts say A is consistent with available objective evidence (if available) Therefore, A can be accepted as a plausible assumption. This general schema hints at the borderline between fallacious and non-fallacious use of appeal to expert opinion. If all requirements are met, the argument is reasonable. The more requirements are not met, the weaker the argument gets, until at some point it becomes fallacious. Argument from authority is used in dialogue, and hence the reasonableness or fallaciousness of the argument depends on dialectical issues. Walton devotes two chapters to the dialectical aspects of the argument. In the first he examines some longer examples of the use of argument from authority in everyday discussion. The second discusses expert testimony as legal evidence. This field is of particular interest because expert testimony is a part of many current legal court cases. Both sides use experts to strengthen their case, which ever so often results in a battle of the experts. Both reasonable and fallacious uses of expert testimony occur. This makes the question how to judge the expert testimony a relevant one. The conclusion Walton draws from these examples is that critical questions play an important role in judging arguments from authority fallacious or non-fallacious. The option of posing critical questions to the alleged expert is of course only possible in a dialogue. This makes the context of the dialogue, and the rules governing the dialogue, important factors. 280 BOOK REVIEW The final two chapters of the book deal with the critical questions and a detailed explanation of the fallacy. These two chapters are, I think, the most relevant to the study of dialectics in AI and law. The critical questions address each of the premises in the general schema for the argument from authority. Examples of these questions are: the expert question: How credible is E as an expert? or the opinion question: What did S assert that implies A? These questions can, of course, be followed by appropriate sub-questions. The critical questions can be seen as argument moves one may use when confronted with an expert opinion. If the dialogue passes the critical questions, then the assumption brought forward by the proponent can be accepted as plausible. This also shows us something about the fallacy. Expert opinion should be treated as defeasible and presumptive. If it is presented as absolutistic and final it becomes fallacious. 3. Relevance to AI and Law One of the central topics in dialectics is the question when a claim adduced by a party in a dialogue is sufficiently justified. A party who makes a claim (the proponent) may be asked to provide justification or proof for his claim (by the opponent). If the opponent keeps asking for further justification, the dialogue would come in an infinite regress. This would also pose an enormous burden on the proponent of the initial claim. One needs a way to terminate the request for more proof at some point. One way the proponent can terminate the infinite regress is by using an argument the opponent is committed to herself. The proponent has no burden to prove statements the opponent is already committed to. Another way to terminate the infinite regress is by showing an argument to be self-evident. Somehow arguments are rooted in generally accepted truths. One may dispute these truths, at the cost of a shift in the burden of proof. A difficult question is of course: what are self-evident or default truths that end a discussion? In legal disputes one can point to the text of a statute and say: ‘This is what the law says’. This claim says nothing about the interpretation of the text, but at least the text itself is beyond doubt.2 Expert opinion may qualify as another source of facts beyond questioning. If so, the capability of handling expert opinion in legal dialogue systems is valuable. This is where Walton’s work comes in. His analysis of appeal to expert opinion shows the general characteristics of this type of argument and it also shows how to deal with it in a critical discussion. It would have been interesting if Walton had addressed the topic in a way more similar to his Commitment in Dialogue (Walton 2 One may, of course, question the source. It may, for instance, be an outdated piece of legislation. BOOK REVIEW 281 and Krabbe, 1995). This latter work contains both informal and more formal accounts of commitment in dialogues, which makes it more directly applicable to AI and Law research than the book under review. Appeal to Expert Opinion is worth reading. It offers valuable insight in a type of argument that is commonly used in legal dialogues. Dialogue systems have to be able to cope with these arguments and hence a thorough insight in their nature is needed. It will be interesting to see what the more formally inclined within AI and Law can do with the material presented in this book. References Berman, D.H. & Hafner, C.D. 1995. Understanding precedents in a temporal context of evolving legal doctrine, in Proceedings of the Fifth International Conference on Artificial Intelligence and Law, ACM Press, New York, pp. 12–20. Gordon, T.F. 1993. The Pleadings Game: An Artificial Intelligence Model of Procedural Justice, Ph.D. Diss., Technische Hochschule Darmstadt. Gordon, T.F. 1995. The Pleadings Game: An Artificial Intelligence Model of Procedural Justice, Kluwer, Dordrecht. Hage, Jaap C., Leenes, Ronald & Lodder, Arno R. 1994. Hard cases: a procedural approach, Artificial Intelligence and Law 2, 113–166. Leenes, R.E. 1998. Hercules of Karneades, Hard Cases in recht en rechtsinformatica. (Hercules or Karneades; Hard Cases in Law in and AI and Law), PhD thesis University of Twente (in Dutch), Twente University Press, Enschede. Lodder, A.R. 1998. DiaLaw: On Legal Justification and Dialogical Models of Argumentation, Kluwer, Dordrecht. McCarty, L.T. 1995. An implementation of Eisner v. Macomber, in Proceedings of the Fifth International Conference on Artificial Intelligence and Law, ACM Press, New York, pp. 276–286. Prakken, Henry 1999. On formalising burden of proof in legal argument, in H. Jaap van den Henrik et al. (eds.), Legal Knowledge Based Systems, Jurix 1999, The Twelfth Conference, GNI, Nijmegen, pp. 85–97. Prakken, H. & Sartor, G. 1998. Modelling reasoning with precedents in a formal dialogue game, Artificial Intelligence and Law 6, 231–287. Rissland, E.L. & Ashley, K.D. 1987. A case-based system for trade secrets law, in Proceedings of the First International Conference on Artificial Intelligence and Law, ACM Press, New York, pp. 60–66. Walton, D.N. & Krabbe, E.C.W. 1995. Commitment in Dialogue: Basic Concepts of Interpersonal Reasoning, State University of New York Press, New York. Walton, Douglas N. 1999. Appeal to Popular Opinion, Penn State Press.Pennsylvania. Faculty of Public Administration and Public Policy University of Twente Enschede The Netherlands E-mail: r.e.leenes@bsk.utwente.nl Ronald Leenes