Causation, Meaning, and Communication

The words we use to describe what happened shape what comes to a listener's mind. How do speakers choose what causal expressions to use? How does that choice impact what listeners imagine? In this paper, we develop a computational model of how people use the causal expressions "caused", "enabled", "affected", and "made no difference". The model first builds a causal representation of what happened. By running counterfactual simulations, the model computes several causal aspects that capture the different ways in which a candidate cause made a difference to the outcome. Logical combinations of these aspects define a semantics for the causal expressions. The model then uses pragmatic inference to decide what word to use in context. We test our model in a series of experiments and compare it to prior psychological accounts. In a set of psycholinguistic studies, we verify the model's semantics and pragmatics. We show that the causal expressions exist on a hierarchy of specificity, and that participants draw informative pragmatic inferences in line with this scale. In the next two studies, we demonstrate that our model quantitatively fits participant behavior in a speaker task and a listener task involving dynamic physical scenarios. We compare our model to two lesioned alternatives, one which removes pragmatic inference, and another which removes semantics and pragmatics. Our full model better accounts for participants' behavior than both alternatives. Taken together, these results suggest a new way forward for modeling the relationship between language and thought in the study of causality.