Bayesian approach to informal argumentation : evidence, uncertainty and argument strength

The work in this thesis contributes towards answering a simple, important and longstanding question: How do people evaluate informal arguments In Chapter 1, I review existing approaches to informal argumentation, and suggest that the Bayesian approach provides the most appropriate way of capturing informal argument strength. The Bayesian approach assumes that arguments are composed of claims and evidence. When people evaluate informal arguments, they make a probabilistic judgment about how convincing it is - that is, how likely the claim is to be true given the available evidence. The Bayesian approach is normative, because it makes predictions about how convincing different arguments should be. In Chapter 2, I examine the Bayesian claim to provide normative guidance for argument evaluation, and conclude that it provides solid normative principles on which to base an account of informal argument strength. The remainder of the thesis comprises experimental work in two distinct but related domains - the evaluation of socio-scientiflc arguments, and the evaluation of slippery slope arguments. Understanding the public response to scientific messages about, for example, climate change, is becoming increasingly important. In Chapter 3, I report the results of four experiments (Experiments 1a, 1b, 1c, & 1d) designed to establish whether there are any differences in the way that people evaluate arguments about scientific topics as opposed to non-scientific topics. The data suggest that both scientific and non- scientific arguments are evaluated in a way that is broadly consistent with the rational predictions of Bayesian theory. In Chapters 4 and 5, I tackle a longstanding philosophical puzzle - when, if ever, is it rational to be persuaded by slippery slope arguments Using Bayesian decision theory, and by identifying a mechanism on which evaluation of these arguments may be predicated, I demonstrate when and why slippery slope arguments are convincing (Experiments 2 - 9). Finally, in Chapter 6, I conclude that the Bayesian approach provides a valuable metric for studying the evaluation of informal arguments, and identify some outstanding questions raised by my research.EThOS - Electronic Theses Online ServiceGBUnited Kingdo