At the population-level, smoking causes lung cancer. But what about 'Harry's smoking caused him to develop lung cancer? Or 'in spite of his heavy smoking, Harry did not develop lung cancer'? What does epistemology have to say about the levels of causation? Are there levels of causation? And if so, how they relate to each other? Traditional defences of twofold causality were motivated by contradictory situations in which preventatives at the population-level become positive causes at the individual-level, as for instance in the celebrated 'squirrel example'. The way out to save probabilistic theories of causality was to distinguish two different levels of causation. However, it can be shown that these defences commit a fallacy in their argumentation. Additionally, this fallacy is the symptom of a more serious problem: to give metaphysical answers to epistemological questions. I will then try to provide a better epistemological understanding of the levels of causation and of the corresponding interpretation of probability.