Killing and Letting Die
Perhaps, however, the killing—letting die distinction is not itself of huge consequence for the GST. Indeed, Thomson’s own response to the objection based on that difference is to point out that the violinist analogy is just meant to show that the two cases (the violinist case and abortion) can both be assimilated into the Good Samaritan analysis, being in all relevant respects morally alike, notwithstanding that one may be an act of killing and the other not. As she has written:
Now it had not actually escaped my notice that the mother who aborts herself kills the child, whereas a man who refuses to be a Good Samaritan — on the traditional understanding of Good Samaritan — merely does not save. My suggestion was that from a moral point of view these cases should be assimilated: The woman who allows the pregnancy to continue, at great cost to herself, is entitled to praise in the same amount, and, more important, of the same kind, as the man who sets forth, at great cost to himself, to give aid. That is why I proposed we attend to the case of you and the violinist.
In other words, Thomson has gone on to claim that even if abortion does entail killing rather than allowing death, the violinist analogy should show us that the distinction is not meaningful here, since the two cases nevertheless amount to the same thing: a refusal to be a Good Samaritan.
Building on this further claim, David Boonin seeks to defend the GST by arguing that killing a fetus-person is morally permissible even if one accepts that killing is, all things equal, substantially morally worse than letting someone die. Boonin begins by presuming that extraction abortion at least, like abortion by hysterotomy, does not in fact kill the fetus, but only allows it to die. Next, he argues that if a hysterotomy abortion is morally permissible, because a pregnant woman is under no positive duty to rescue a fetus by continuing to gestate it, then killing the fetus is also permissible if abortion by killing is considerably less burdensome or dangerous for the pregnant woman than having it extracted.   In contending thus, Boonin relies on the principle that if one may permissibly let another person die in order to avoid some burden or harm, then one may instead kill that same person if the cost to you of letting die rather than killing is great enough. In order to show that killing the fetus is permissible, then, we need only establish that a pregnant woman would be permitted to allow the fetus to die by having an extraction abortion, and that an abortion which kills the fetus instead will spare her considerable extra burdens. The second condition might be made out if, for instance, partial-birth abortion or dilation and evacuation were a less risky procedure for the woman than a hysterotomy, or require less recovery time.
Boonin’s significant conclusion is that even if killing is substantially morally worse than letting die, abortions which kill the fetus either prior to or in the process of removing it are morally permissible in circumstances where extraction abortion would be. Hence, it is of no consequence to Thomson’s argument if unplugging the violinist is a mere failure to rescue whereas many abortions are not.
Boonin is proceeding here on the footing that extraction abortions are mere failures to rescue, which some will doubt, and that such refusals of aid are morally permissible given the onerous nature of pregnancy (and considering that a pregnant woman has no special responsibility for her fetus). But the crux of Boonin’s argument might come to the defence of the GST even if extraction abortions too are regarded as acts of killing rather than as refusals to rescue. So long as the duty to volunteer gestation would still have been absent, killing so as to avoid additional burdens is acceptable.
An integral premise in Boonin’s reasoning is that a ‘victim’ is not harmed by being killed if it would have been permissible for the killer or her agents to allow him to die. As he explains his argument at one point, ‘if killing is worse than letting die because causing harm is worse than allowing harm, then the doctrine will simply fail to apply in cases where death is not a harmT3 Since one would have been permitted to allow the victim to die, then killing the victim does not amount to harm on Boonin’s view, for it does not change the morally permitted outcome. Boonin thinks it reasonable to hold that one is permitted to spare oneself the relative burdens of letting die if it makes no difference to the outcome for the victim. This, he sees, is no contradiction of the doctrine that killing is more strictly prohibited than allowing death.
The tenor of Boonin’s argument here bears similarities with the philosopher Frances Kamm’s well-known argument that abortion is morally permissible because it renders the fetus ‘no worse off2 than it would have been apart from the pregnant woman’s bodily assistance.44 Kamm also argued that killing a fetus is not necessarily to harm it, since, in being killed, the fetus loses only what it had gained through the pregnant woman’s bodily support—support that she was, moreover, not obligated to volunteer, due to the special burdensomeness of pregnancy. Consequently, aborting a fetus by any method does not harm it relative to its ‘pre-attachment opportunities’, which, Kamm argued, is the correct baseline for assessing what does and does not cause it harm. Kamm argued that in this respect the abortion mirrors the act of detaching Thomson’s violinist. In being unplugged, the violinist is also losing only what he had gained through his host’s support—continued life—a benefit which the host cannot be obligated to bestow on him at the cost of nine months’ connection.
The main strand of Kamm’s argument is what she dubs the ‘Output—Cutoff Principle’. This states that killing is permissible when it eliminates a life sustained by the use of one’s body—a use one is not morally required to offer—in order to end that use. The principle therefore comes into play when the person killed is ‘no worse off’ than he would have been had he never had the benefit of the host’s support. In such a situation, the supporter is justified in cutting off the output of support, terminating the dependant’s life. Because the fetus has no opportunities for life apart from attachment to the pregnant woman, and because (let us presume) the woman has no positive obligation to assist the fetus, killing the fetus rather than letting it die is permissible. It leaves the fetus ‘no worse off than it would be but for assistance to which it is not, in any case, entitled. Again, this is meant to hold even if the fetus is a person. In sum:
The fetus cannot obtain the right to remain in the woman’s body, and she cannot be obligated not to kill it, simply because ending its support would require that we take away the life it has only because it received support ... if letting it die is permissible, then sometimes killing it is permissible as well.45
Kamm’s argument is similar to Boonin’s in that it seeks to move from the absence of an independent positive obligation to gestate a fetus-person to the conclusion that killing the fetus is morally permissible when a few extra conditions are met. Like Boonin, Kamm thinks that there is some difference between what is required to permissibly kill a person rather than to let him die. For her, it is that continued support must meet a higher threshold of burdensomeness before killing, as opposed to letting die, is permitted. You may be justified in attacking the violinist to preserve your physical integrity, but not to spare yourself financial sacrifice. Analogously, killing the fetus is permissible if it is the only way to end bodily support without serious cost to the woman. Kamm concedes that if it were possible to safely extract a viable fetus without much cost, a pregnant woman would not be justified in choosing to have it killed rather than extracted, a concession Boonin also makes at one point.46 But if killing spares the woman additional trouble, the Output-Cutoff Principle renders it permissible.
Both Boonin and Kamm hence reach the conclusion that the killing-letting die distinction is no threat to the GST, even if killing is, in Boonin’s words, ‘substantially morally worse’. As I suggested above, the best version of the argument need not depend on our being convinced that extraction abortions merely fail to rescue the fetus. Even if all abortion is killing, the argument says, it is permissible if the 
45 ibid 193.
burdens entailed by not killing were such that the pregnant woman could not have been obligated to voluntarily assume them, and choosing to kill rather than to extract is better for the woman. ‘No duty to gestate’ and ‘no duty to refrain from killing’ in effect amount to the same thing. This is, it might be said, precisely what Thomson is trying to establish by showing how similar the cases of the violinist and pregnancy really are. But can Thomson’s assimilation of the cases—including where abortion clearly amounts to killing—be sustained using Boonin and Kamm’s reasoning?