The way I see it, the relationship between causation and normativity makes sense in the Bayesian framework which is generalizable to most domains of cognition. From a memory and statistical learning point of view, it makes sense that the saliency effect and heightened processing in cases of unexpectancy could result in disproportionate causal reasoning. Though I think that the type of reasoning introduced in the paper is applicable in most types of scenarios, I am having a hard time believing this would be the whole story. There are many cases where people are biased in their expectations and that bias leads them to ignore/not process the violations of expectations. For example, a person with trait anxiety might believe that other people do not enjoy their company, and therefore expect that others behave accordingly. If said person gets an invitation to a party, it would be common for them to, instead of process the friendliness behind the invitation in a heightened manner, explain the event away with an explanation that fits their model better. I wonder how these types of biases fit in with the framework of causal reasoning.