This paper described a very interesting phenomena. I wonder if the interesting nature of the phenomena actually stems from having a bias to think about causes in context-free terms. Instead, in day-to-day life, it seems that we make causal judgements against certain (unrepresented) background assumptions. When some of these background assumptions are spelt out, they seem what are termed here as norms. I wonder if anyone has attempted a formalization of causality while including the notion of context or (unrepresented) background assumptions.
Taking this into consideration, to me it seems that counterfactuals make sense only against a particular context. If the context is allowed to change arbitrarily, there are no counterfactuals, and thus talking about counterfactual causality in a context-free manner ceases to make sense. A particular example I wonder might be relevant would be causal launching event of ball B by ball A. To say ball A causes the motion of ball B involves, perhaps, an indefinite amount of assumptions. If some of them were represented, they might be: both are free to move, B does not move in a self-propelled manner, both are physical objects and not pixels on a screen, B is a rigid body.
Causality then might better be approached from an embodied perspective.