This is what is known as the reference class problem. The probabilities we assign to the occurrence of particular events do not seem to be objective features they possess, but instead seem to be determined by our contingent ways of conceptualizing those events on a given occasion. (This may vary depending on the context making it difficult to say what is the probability of the event.) But causation is supposed to be an objective phenomenon concerning the relations between events. Whether one event is the cause of another doesn’t seem to depend on how we think about these events, or whether we think about them at all.