When do we think that X caused Y?
Tadeg Quillien · 2019
When explaining an event, people tend to select a single cause out of the multiple factors that contributed -- for instance, they will say that a forest fire was caused by a lit match, without mentioning the oxygen in the air which helped fuel the fire. Recently scholars have suggested that causal selection is designed to provide explanations that are likely to generalize across a variety of background circumstances. Here, we develop a computational model of causal selection which formalizes this idea. Under minimal assumptions, the model is surprisingly simple: a factor is regarded as a cause of an outcome to the extent that it is, across counterfactual worlds, correlated with that outcome. The model explains why causal selection is influenced by the normality of candidate causes, and outperforms other known computational models when tested against a fine-grained dataset of human graded causal judgments.