Note: Portions of the argument below sound like I accept consensual euthanasia. I don't: intentionally killing juridically innocent persons, whether they consent or not, is wrong. But I don't make use of this belief in the argument, and my argument is in large part aimed at people who do not share this belief of mine on euthanasia.
This post is divided into three sections, the first giving a standard argument against abortion, the second giving a standard response, and the third arguing that the standard response is unsatisfactory, at least if one allows that one can rationally desire to die, a thesis that I am not sure of, but that few people who are not already pro-life will deny.
Part I: A standard argument against abortion: This takes two steps. First, one shows that a typical fetus would have a future of the same sort that we do were she not killed. This may involve metaphysical arguments to establish that the future is indeed the fetus's future in the sense that the fetus is identical with the adult human being. (I argue like this here.) The second step is to show that at least one of the reasons that it is wrong to kill you or me is that it would undeservedly deprive you or me of this sort of future, so that it is wrong to kill a typical fetus for exactly the same reason.
Part II: One standard response: Grant, at least for the sake of discussion, the claim that the fetus would have a future like ours, but deny that the undeserved deprivation of a future life like ours makes the killing wrong. Instead, what makes it wrong to kill someone is that doing so goes against the person's interests, which are defined in terms of the desires that the person has (on a crude version of the response) or would have in ideal mental circumstances (on a less crude version of the response), desires whose fulfillment requires the continuation of life. The reason for going for the "ideal desire" view is some version of the example of the suicidal teenager: it is wrong to kill the suicidal teenager even if the teenager lacks all future-directed desires. But, one argues, in ideal mental circumstances, the teenager would want to live, so a better story is the ideal desire one.
Part III: A response to the response: It is not the case that what makes it wrong to kill x is that x actually or ideally has desires that require the continuation of life. For suppose George does not actually or ideally desire the continuation of life. He is miserable, abandoned by all friends, no longer capable of engaging in any of his past projects, in the grip of a painful terminal illness. It seems not that implausible to suppose that George could rationally desire to die, so that he not only actually but also ideally has the desire to die. We may also suppose that he has no fulfillable ideal desires incompatible with the desire to die. Now some strongly pro-life people will deny the idea that one can rationally desire to die, but I suspect there are very, very few pro-choice philosophers who will dispute this. Moreover, even someone completely opposed to euthanasia can hold that it is rational to have the desire to die as long as one adds that it is wrong to act on that desire (other than maybe by praying for death).
So, to recap, we suppose, and few pro-choicers will deny us this assumption, that George rationally, consistently and ideally desire to die. By the response in Part II, what makes it wrong to kill people is actual or ideal desires that require future existence. But George doesn't have such desires, and hence it follows that is not wrong to kill him. But this is absurd.
A reader might say: "So, you've shown that if you accept Part II, you accept the permissibility of euthanasia. Almost everybody who is pro-choice already accepts the permissibility of euthanasia, so this is no reductio." But that would be a mistake. For I did not say that George consents to being killed. All I said is he desires death. It is one thing to desire something and another to consent to it (see this post). And I suspect that most pro-choice folks will agree that it is wrong to kill a non-consenting innocent adult, even if that adult desires to be killed. For it is the consent rather than the desire that matters here. Why would George not consent? Here, a myriad of possibilities is available, including religious views, views about the sanctity of life, views about the way that killing him would dehumanize the killer, etc.
So it is not enough to establish the lack of actual or ideal desires presupposing future life to show that a killing is permissible. Step II fails.
Could we say that the fetus consents to being killed, so that we could say that it is permissible to kill someone lacking actual or ideal desires presupposing future life if the victim consents to being killed? No: the fetus plainly does not consent. Could we say that the fetus would consent if it could be asked? We have no reason to suppose that. In the typical case, the fetus would be asked to give up a future like ours with no compensating benefit to the fetus, and there is no reason to suppose a positive answer to a deal like that. Alternately, perhaps, we could handle the consent question in the way we handle it in practice: appoint a proxy. But note that a proxy cannot have a potential conflict of interest, and in the case of abortion, the mother does have a potential conflict of interest—if she's considering abortion, this is not unlikely to be the case because she takes the pregnancy to conflict with her interests or the interests of her significant other or of her other children. Rather, the proxy has to be one who considers primarily what is good for the individual that she is a proxy. I suspect that, given "future like ours" considerations, in typical cases such a proxy will not agree to abortion.
Could we instead say that the fetus does not dissent from being killed, and it's permissible to kill someone who (a) lacks actual or ideal desires presupposing future life, and (b) does not dissent from being killed? That requirement seems too weak. It would be wrong to shoot George without positive consent from him (if he's capable of giving it) or presumed consent or proxy consent, I think.
A final option (Frank Beckwith suggested something like this): Perhaps the pro-choice opponent can say that having (actually or ideally) desires, or desires of a certain sort, is part of what makes one a person, and so it is not the case that killing someone who has the desire for life is wrong because it goes against that desire, but what makes it wrong to kill is that it is a person who is non-consensually killed. If one takes this view, then it seems one gets a completely different account of the wrongness of killing from that given in Part II. It is not the having of desires presupposing future life that makes it wrong for one to be killed (at least if innocent), but, simply what makes it wrong to kill x is that x is a non-consenting person. Her desires are beside the point: only the consent matters here. However, this account of what is wrong with killing is inferior to the Marquis account in Part I. For it fails to show how killing someone is different from doing other things that the person does not consent to, such as patting on the shoulder. Patting on the shoulder may be wrong without consent (though probably not always wrong), but is clearly much less of an evil than killing someone. Nor is it that the victim actually or ideally has ideally a stronger desire not to be killed than not to be patted, or that she more strongly dissents from being killed than from not being patted, since neither of these might be true, at least if we assume that one can rationally desire to be dead.
So it seems to me that it is hard to rescue Part II, and hence the Marquis argument's claim that if a fetus has a future like ours then it is wrong to abort the fetus survives.