1. Why is Mary walking rather than swimming?
2. Why is Mary walking?
These are distinct questions, of course, and call for distinct answers; the first question calls for a contrastive explanation. (There is a position that holds that all explanations are actually contrastive, but even if that's true, the contrastive explanation for (2) could not be the same as the contrastive explanation for (1), at least in general.) He then goes to argue that there are three presuppositions to question (1): that Mary is walking, that Mary is not swimming, and that it is possible for Mary to be swimming. His argument for the third of these:
If I aim to explain why she is walking rather than swimming, then I presuppose that she is not swimming. But her not swimming is consistent with the possibility of her swimming. Her not swimming is also consistent with the impossibility of her swimming. Nevertheless, if I ask why walking rather than swimming, I presuppose that she might have been swimming. 'Rather than' means 'instead of' (in place of). So if she is walking instead of swimming, and walking is possible because actual, then swimming must also be possible if it is to be something that can be done instead of walking. It might help to consider
3. Why is Mary walking rather than levitating?
4. Why is Mary walking rather than levitating and not levitating at
the same time?
These two questions have presuppositions that are false. (3) presupposes that it is possible that Mary be doing something nomologically impossible, while (4) presupposes that it is possible that Mary being doing something that is narrowly-logically impossible. Questions (3) and (4) are therefore not to be answered but to be rejected -- by rejecting the false presuppositions upon which they rest.
I do not find this argument convincing at all. If someone honestly asks "Why is Mary walking rather than levitating and not levitating at the same time?" it seems to me that the explanation that it's possible for her to walk but not possible for her to levitate and not levitate at the same time is a perfectly good contrastive explanation, and is, in particular, an intelligible answer to an intelligible question.
There seems to me to be an alternative explanation here, which is that questions (1), (3), and (4) do not require us to presuppose that the contrast case is possible; rather, as a practical matter, it would be unusual for anyone to ask such questions if no one thought the contrast case possible. This, however, makes it not a presupposition but an implicature; that is, in most cases it would arbitrarily violate the maxim of quantity. I think the 'thought' is important; it's not necessary for us to be presupposing that the contrast case is really possible, only to be suggesting that someone might think it possible. A pretty good paraphrase of (1) in actual conversation, for instance, would be "Why is Mary walking and not swimming, given that one might think that she would be swimming?"
For both of these reasons, I do not think we have to say that (1), (3), and (4) have a false presupposition, and therefore I don't think the questions can be dismissed on this basis. (Indeed, as one might infer from my comments, I don't think we can reasonably dismiss them at all; it's just that they have very easy answers. But since impossibilities can be pretty subtle, this would not apply to every kind of request for contrastive explanation.)
ADDED LATER: Just thought of a good addendum here. Suppose the question were:
(5) Why is it impossible for Mary to levitate and not levitate at the same time rather than possible for her to do it?
Now, we get potential complications here by the multiplication of Diamond modal operators (the alleged presupposition is that it is possible that it is possible for her to do it). But whether this is an issue will simply depend on which modal system we are presupposing. It seems that we can make perfectly good sense of this question, and of answers to it, without having to dismiss it as having a false presupposition, because we can explain why it is impossible rather than possible, without assuming that the contrast case is actually possible, and we can take it just as something-that-someone-might-think-possible or something-that-someone-might-claim-is-possible.