Probabilistic accounts of conditional reasoning
Conditionals and conditional reasoning are everywhere in ordinary and scientific discourse. Ordinary people, politicians, civil engineers, and scientific consultants all have an interest in the following indicative conditional in natural language:
(1) If global warming continues, then London will be flooded.Both ordinary people and scientists would assert (1) with some degree of confidence short of certainty, and they might well add an explicit epistemic qualifier, applying probably or a related term to (1), expressing some uncertainty. Most ordinary and scientific reasoning takes place in a context of uncertainty, including conditional reasoning (Elqayam & Over, 2013; Oaksford & Chater, 2007). But there have been intense debates about the probability of conditionals (Edgington, 1995; Evans & Over, 2004), and other deep questions about conditionals and conditional reasoning have been disputed for thousands of years (Kneale & Kneale, 1962, pp. 128–138). The unsettled nature of research on conditionals is understandable. Every argument or inference from p to q, relatively strong or weak, can be turned into a conditional, if p then q, in which we have more or less confidence. A complete account of conditional reasoning will be possible only when reasoning itself, from the scientific to the moral, has been fully explained.