The 17th ARGDIAP Conference

Reasoned Argumentation. Legal, Computational and Linguistic Perspectives

Fully online

Programme

(All hours are CET)

Wednesday, November 24

Thursday, November 25

KEYNOTE TALKS

Henry Prakken  (Utrecht University)

Explanations and justifications in legal XAI (Wednesday, November 24, 11.30-12.30)

In this talk I will discuss my recent research with Rosa Ratsma on legal explainable (XAI) in a wider legal and AI context. I will first outline our case-based method for explaining predictions of supervised-machine-learning applications, which draws on AI & law research on argumentation with cases. A case-based approach is natural since the training data of supervised machine-learning applications can be seen as cases.

I will then explain that our method strictly speaking does not *explain* but *justify* predictions, and I will discuss the place of justification methods in (legal) XAI. Finally I will put my discussions in the context of legal requirements for explanations, such as those of the GDPR.

Bio:

Senior lecturer in the Intelligent Systems Group of the computer science department at Utrecht University, and professor in Legal Informatics and Legal Argumentation at the Law Faculty of the University of Groningen. He has master degrees in law (1985) and philosophy (1988) from the University of Groningen. Since May 2021 he also has a temporary appointment as a parttime professor at the European University Institute, Fiesole, Florence, Italy.

In 1993 he obtained a PhD degree (cum laude) at the Free University Amsterdam with a thesis titled Logical Tools for Modelling Legal Argument. His main research interests concern artificial intelligence & law and computational models of argumentation. He is a past president of the International Association for AI & Law (IAAIL), of the JURIX Foundation for Legal Knowledge-Based Systems and of the steering committee of the COMMA conferences on Computational Models of Argument. He is on the editorial board of several journals, including Artificial Intelligence (since 2017 as an associate editor).

Jaap Hage (Maastricht University)

What is a good argument? (Wednesday, November 24, 14.00-15.00)

The twentieth century was dominated by a view of argumentation according to which the golden standard for a good argument is a deductively valid argument, preferably formalised in the language of first-order logic. This standard was developed in attempts to provide mathematics with a solid foundation, and the resulting logical theory is well suited to that purpose. However, it is less suitable for the evaluation of arguments in many other fields, and to remedy this shortcoming many alternatives were designed, such as modal logics and non-monotonic logics. Typically, these alternatives were developed in ways that aimed to mimic first-order logic as far as possible, thereby propagating the design mistake from which the theory of argumentation suffered since the beginning of the 20th century. 

This design mistake, as Toulmin pointed out, is to treat the central form of argumentation as an argument of the form Modus Ponendo Ponens, where a statement is derived from two other statements. According to Toulmin, one of the two statements in the analysis should be replaced by a ‘warrant’, a kind of inference rule. This same idea can be found back in the idea to assign a central role in the analysis of argumentation to so-called argumentation schemes.

By giving up first-order logic as the paradigmatic standard for the evaluation of arguments, the necessity arose to devise a new standard. The introduction of warrants or argumentation schemes as such does not suffice, because not all warrants and schemes lead to good arguments. What are good warrants or argumentation schemes, and what tests can we apply to distinguish good ones from bad? 

In this presentation it will be argued that the question what good arguments are cannot be separated from the question what arguments are considered as good in the different practices of argumentation, such as the physical sciences, ethics, law, or economics. This is a naturalistic approach to argumentation, and as all naturalistic approaches to normative or evaluative questions, it is vulnerable to the criticism that the logical gap between Is and Ought, or between Fact and Value is ignored. The concluding part of the presentation will be devoted to arguing, first that these gaps are overestimated if they exist at all, and second how the attractiveness of the Is/Ought- or Fact/Value-criticism can be explained and accommodated without abandoning the naturalist approach to argumentation. A crucial part of the latter argument consists in a distinction between basic, rule-based and constructivist social facts.

Bio:

Jaap Hage holds the chair of Jurisprudence at the Maastricht University Law School. He studied both Law and Philosophy at Leiden University, where he also defended his PhD-thesis on Law and Meta-ethics (1987: ‘Feiten en betekenis’). He worked at the Leiden law school from 1978 until 1989. In 1989 he moved from Leiden to the University of Maastricht, where he first worked at the department of computer science and later (from 1991) at the law school. There he was (or is) responsible for, amongst others, courses that introduce to law, skills courses, legal philosophy and logic.

His research has focused on legal theory in a broad sense, including particularly Law and Logic, the Ontology of Law, Basic Legal Concepts, Philosophy of Legal Science, and Law and the Cognitive Sciences.

His main publications include the books: Reasoning with Rules (Kluwer 1997), Studies in Legal Logic (Springer 2005) and Foundations and Building Blocks of Law (Eleven Publishers 2018). A more extensive overview of his publications, with possibilities for downloading late draft versions, can be found at www.jaaphage.nl.

Presently, Jaap Hage chairs the department Foundations and Methods of Law.

Isabela Fairclough

Frames, speech acts, argument schemes (Thursday, November 25, 11.30-12.30)

I intend to explore the relationship between framing phenomena, as discussed in the political communications literature (D’Angelo & Kuypers 2010), and the concepts of inference, argument schemes and critical questions. I suggest that, in addition to ‘selection and salience’ (Entman 1993), framing theory needs these concepts if it is to explain how framing effects (changes in the audience’s beliefs, values, action) may succeed or fail. I begin by looking at how framing effects can arise via reference and predication in simple assertive speech acts (e.g. via implicit premises that audiences are expected to draw from their linguistic and extra-linguistic knowledge), and at the move from is to ought that seems inherent in any intended framing effect. The arguments that audiences are expected to construct, by putting together, in premises, the explicit ‘frames’ (Fillmore 1982, 2006) they are given and their own background knowledge and values, can be represented in terms of a wide range of argument schemes – from consequence, values, verbal classification, expert opinion, analogy, etc. (Walton et al. 2008). Normative and evaluative premises bridging the move from is to ought may be drawn from knowledge of language (Wierzbicka 1996) or from world knowledge, including knowledge about social norms and laws. As I have suggested elsewhere, since in real-life contexts audiences are usually aware of a wider background of controversy, and of the existence of alternative competing goals and values, of alternative consequences  or definitions of the situation, framing as process can be represented within a deliberative, decision-making framework, where the salient consideration is part of a premise within an argument scheme supporting only one of two contrary conclusions (Fairclough 2016, 2019, Fairclough & Madroane 2020).  

In this talk, I will try to investigate a simpler situation, where one conclusion is intended for acceptance based on a single argument scheme, involving definition and categorization (classification). I will look at an article on migration published in The Guardian, where the journalist makes  a double attempt: to (re)categorize  a group of people  and to (re)define the meaning of migration-related words. He does this via an argument built with the help of several inferences – entailments, presuppositions, implicatures – which he presents as either already shared with the audience or as the product of acceptable reasoning that can be expected from them. However, the comments thread shows a clear reluctance to accommodate the inferences presented as shared and to accept the intended implied conclusions. The structure of the argument made by the journalist is transparent to the audience, but its premises, conclusion and validity are challenged. One such critical challenge is from the legal definitions of migration lexis. I will draw on studies of the way frames function in argumentation (Bigi & Greco Morasso 2012, Greco Morasso 2012) and on a dialogical  view of inferences such as presupposition (Macagno 2016, 2018). I am suggesting a view of media framing  as a ‘dialogue game’ (Reed & Walton 2007), and look at how the audience engage with the asserted and implied/presupposed/entailed content, refusing to play the game that was designed for them.

Bio:

Isabela Fairclough is Senior Lecturer at the University of Central Lancashire (UK). She has published on practical reasoning, deliberation and decision-making, with application to economic, political and environmental debates. Her publications include the monograph Political Discourse Analysis (2012, Routledge). She is working on a new book, The Rhetoric of Framing, commissioned by Cambridge University Press.

Fabio Paglieri, ISTC-CNR Roma

Online arguments in context: what is really wrong with public debate in the digital age, and how to fix it? (Thursday, November 25, 14.00-15.00)

This talk starts with a brief description of the socio-economic forces that shape current online technologies, then combine it with an anecdotal observation: an increasing number of social media users seem to show symptoms of fatigue and dissatisfaction, typically motivated by feeling overwhelmed by the amount, content, and (poor) quality of other people’s contributions, by how often their own posts are misinterpreted and therefore stir trouble, or both. Notably, the vast majority of these flabbergasted users do not quit social media or even reduce their daily exposure, yet they keep complaining about it. Investigating this curious phenomenon requires discussing the collapse of communicative context typical of social media interactions, as well as reminding us of the actual motivations behind the structure of social media platforms (the goals of service providers) and the way people use them (the goals of users). In this scenario, genuine argumentation is shown to happen more like a frequent accident than as a well-designed interaction. This is not meant to suggest that designing argumentative practices to improve public debate on social media is hopeless, but rather to clarify the nature and scope of this noble ambition: the challenge, in a nutshell, is to improve argument quality on platforms that are not designed to reward good arguments, for users that are not primarily motivated to engage argumentatively in the first place. This has important implications on how we should approach this daunting task: in particular, I will argue that (i) facilitating error detection, promoting critical engagement and cultivating digital literacy are crucial mostly, or even solely, as educational tools, whereas, (ii) when it comes to supporting online public debate “in the wild”, the golden rule is to make good arguments more visible than bad ones, without necessarily engaging with the latter at all – an argumentative variation on the old “don’t feed the trolls” mantra. I will conclude by offering some streamlined suggestions to current and future developers of argument technologies on how to turn these insights into (some form of) actionable interventions.

Bio:

The overall assumption of Fabio’s research activity is that cognition is goal-directed, resource-bounded, and adaptively shaped. From this unifying perspective, he focuses his research mainly on two areas of interest: decision-making for action (in particular, temporal preferences, long-term goals and risky choices) and belief dynamics, both individual (belief revision and critical thinking) and social (argumentation).

His work is interdisciplinary, combining theoretical analysis, conceptual modelling, behavioral experiments, and field observations. Although his main background is in psychology, he collaborates regularly with primatologists, economists, philosophers, computer scientists and educators to foster their shared research interests.

President of the Italian Association of Cognitive Sciences (AISC, 2017-2019) and the Chair of the Steering Committee of the European Conference on Argumentation (ECA, 2013-2021). Since 2012, he is Editor-in-Chief of Topoi: An International Review of Philosophy (Springer), and Director of Sistemi Intelligenti (Il Mulino), the leading Italian journal in cognitive science and Artificial Intelligence. He is also in the Editorial Board of the journal Argument & Computation (Taylor & Francis) and of the book series Studies in Logic and Argumentation (College Publications), and a member of the Workgroup on Psychology of the publishing house Il Mulino.