Rational argument, rational inference

Abstract

Reasoning researchers within cognitive psychology have spent decades examining the extent to which human inference measures up to normative standards. Work here has been dominated by logic, but logic has little to say about most everyday, informal arguments. Empirical work on argumentation within psychology and education has studied the development and improvement of argumentation skills, but has been theoretically limited to broad structural characteristics. Using the catalogue of informal reasoning fallacies established over the centuries within the realms of philosophy, Hahn and Oaksford (2007a) recently demonstrated how Bayesian probability can provide a normative standard by which to evaluate quantitatively the strength of a wide range of everyday arguments. This broadens greatly the potential scope of reasoning research beyond the rather narrow set of logical and inductive arguments that have been studied; it also provides a framework for the normative assessment of argument content that has been lacking in argumentation research. The Bayesian framework enables both qualitative and quantitative experimental predictions about what arguments people should consider to be weak and strong, against which people's actual judgements can be compared. This allows the different traditions of reasoning and argumentation research to be brought together both theoretically and in empirical research.

Legal idioms: a framework for evidential reasoning

“…This is particularly pertinent in light of recent movements in cognitive science that apply more sophisticated Bayesian models to cognition (Krynski and Tenenbaum 2007;Chater 2007, 2010;Griffiths, Kemp and Tenenbaum 2008). Of particular relevance here is research that gives a Bayesian account of argumentation (Hahn and Oaksford 2007;Harris and Hahn 2009;Corner et al 2011;Jarvstad and Hahn 2011; see also Hahn, Oaksford and Harris 2012). This latter work dovetails with the claims made in this paper, and potentially provides an overarching framework for argument evaluation in general.…”

Some Artificial Intelligence Tools for Argument Evaluation: An Introduction

“…Bayesian methods are widely used in artificial intelligence. The standard Bayesian method of evaluating arguments (Hahn, Harris and Oaksford, 2013) assigns numerical probability values to the components of an argument and uses Bayesian rules to give as output a numerical probability value for the strength of the argument. These include Bayesian rules defining negation, conjunction, disjunction and conditional probability.…”

The Dialogical Entailment Task

“…These studies were, however, not designed to elicit participants' entailment judgments (as opposed to their acceptance of other types of inferences like, say, inductive inferences or implicatures). In fact, much of the research on argumentation within the New Paradigm has been conducted with the explicit goal of showing how everyday informal arguments that have been set aside by classical logic can nevertheless be captured by rational Bayesian reconstructions (Hahn, Harris, and Oaksford, 2012). In contrast, in Eva & Hartmann (2018) it is argued that even on a Bayesian approach to argumentation, an interest should be taken in valid arguments.…”