The smoking lesion problem depends entirely on why people with the smoking lesion smoke more often.
If the mechanism does not affect one’s ability to do reasoning, come to logical conclusions, etc., then even though people who smoke usually are more likely to have the smoking lesion, people who smoke because they followed the reasoning telling them to smoke are not more likely to have the smoking lesion and there is no problem.
If the mechanism does affect one’s ability to do reasoning, then it may not even be possible for someone to logically decide what to do. (Consider an edge case where the smoking lesion makes you 100% likely to smoke, but logical reasoning tells you not to.) Or it may only be possible to do so in a subset of cases.
(Of course, “people who did X because of Y” is a simplification and should be something like “the difference in propensities for Y between people who do X and people who don’t.”)
The smoking lesion problem depends entirely on why people with the smoking lesion smoke more often.
If the mechanism does not affect one’s ability to do reasoning, come to logical conclusions, etc., then even though people who smoke usually are more likely to have the smoking lesion, people who smoke because they followed the reasoning telling them to smoke are not more likely to have the smoking lesion and there is no problem.
If the mechanism does affect one’s ability to do reasoning, then it may not even be possible for someone to logically decide what to do. (Consider an edge case where the smoking lesion makes you 100% likely to smoke, but logical reasoning tells you not to.) Or it may only be possible to do so in a subset of cases.
(Of course, “people who did X because of Y” is a simplification and should be something like “the difference in propensities for Y between people who do X and people who don’t.”)