While one can argue that full rationality is the adequate normative framework to evaluate the reasoning of homo scientificus, as science should spare no effort to validate causal hypotheses and to identify correct explanations, I shall argue that ordinary causal reasoning does in fact have other goals – particularly those of cognitive economy and social co-ordination. This translates into a concern with bounded rationality – using heuristics that are simple but smart; and social rationality – using assumptions about conversation which enable speakers and listeners to co-ordinate their efforts around the most relevant explanation. Understanding such concerns can help explain why homo pragmaticus deviates from scientific criteria for causal reasoning while being rational in his or her own terms. It can also help us understand what we need to do to help people reason ‘scientifically’ about causes, where we consider that to be important. Indeed, understanding the pragmatic rationality of human inference processes should help us better understand why they are ‘sticky’, and sometimes resist reformation into ‘fully rational’ modes of thinking which are principally adapted to doing science. I apply this pragmatic analysis to three domains of causal judgement: learning of causal relations; causal hypothesis testing; and discounting vs. backgrounding of causes in explanations.

Introduction

In this chapter I will be primarily interested in how scientific rationality may differ from pragmatic and social rationality with respect to causal thinking.