Rights and the health of women and embryos/fetuses

Where might abortion fit into the discussion of health policy? In political discussions, those who favor the right to seek abortion often cast it as simply a part of protecting women’s health and access to health care.Footnote ⁴ However, treating abortion as merely a matter of women’s health care could require taking a contentious view of the means by which such health is achieved. Analogously, providing organs for transplantation would not fairly be cast as merely a matter of health policy, if the organs were to be obtained by killing other people. We shall return to this issue below.

Even putting aside the means used to protect health, questions might still arise about whether abortion access is a health policy. For example, a woman’s desire to have an abortion is only sometimes based on burdens that even an ideal pregnancy or delivery may place on her health. (Pregnancy is instrumentally useful for having offspring, but otherwise it can be considered a negative health condition for the woman herself.) In many cases, someone seeks an abortion for the sake of self-determination with respect to life goals other than health. This is so when pregnancy, birth, or taking care of a child is a threat to one’s quality of life, though not necessarily to one’s physical or mental health. This threat might include putting up for adoption a newborn who grew from an embryo, by which time one might have emotional ties that an abortion could have prevented. Some restrictions on abortions (for example, that they be done only in hospitals) are argued for on the ground that they favor a woman’s health. Even if this were true, there might be a conflict between her health and her other life goals. When maximizing health conflicts with other goals, deciding which to prioritize is not itself a matter of health policy but of general practical reason. Hence, the decision goes beyond health policy expertise.

If restricting abortion poses a great risk to the woman’s health or life and she prioritizes avoiding that risk, then removing such restrictions could be considered at least in part a health policy aiming at her health.Footnote ⁵ As noted above, however, it would not be merely such a policy, because the means by which health was achieved might be morally problematic.

Restrictions on legal abortion may lead women to use alternative means that pose a threat to their life and health, such as illegal abortions performed by untrained personnel. Legally permitting abortions—at least when this does not increase the number of abortions over illegal ones—is a health policy when its justification is preventing the use of these alternative means because they threaten health. However, a policy that eliminates these alternative means or deters women from using them without legalizing abortion could also eliminate the health threat posed by those means. If abortions were morally wrong, it could make sense to eliminate the dangerous-to-women’s-health alternatives rather than to legalize abortion.Footnote ⁶

It seems one might first have to argue that abortion is not morally wrong—or at least that seeking it is excusable or should be a person’s legal right—in order to justify legally and safely providing abortion for the sake of women’s health. Only once such justification of the permissibility of abortion is established can the issue of regulating abortion be addressed as simply a matter of health policy. Analogously, suppose one could show that killing some people to acquire their organs is morally permissible (for example, they volunteer for this when they are about to die). Providing these organs to save others’ lives could then be addressed as simply a matter of health policy.

Is regulating abortion most clearly a health policy insofar as it is concerned with the health of the embryo/fetus?Footnote ⁷ The view that it is in the best health-related interest of the embryo/fetus not to die plays a part in restricting abortion. However, objections to killing the embryo/fetus are often less about concern for its health and more about concern for its (supposed) rights. Similarly, the fact that it would be bad for the health of an adult to die is only part of the reason for concern about his not being murdered. Possibly, the violation of the adult’s right not to be murdered would not be as important if it was not in his health-related interest to go on living, but it is primarily the violation of a right not to be killed unjustly that prompts concern about murder. This is why a police policy to stop murders is not considered a health policy, even if being murdered is unhealthy for people. As noted above, not all policies that affect health are health policies.

Furthermore, it is only if one were first successfully to argue that the embryo/fetus has a right not to be aborted, that one is free to consider what policy on abortion would be good for its health. If it has no right not to be aborted, one may not prohibit abortion on the grounds that it is bad for the embryo/fetus’s health. Analogously, if an aggressor has no right not to be killed in self-defense, one may not prohibit such a killing on the grounds that it is bad for his health. Just as permitting abortion could not be merely a health policy favoring women until it is shown that it is morally permissible to abort the embryo/fetus for her sake, so prohibiting abortion could not be merely a health policy favoring the embryo/fetus until it is shown that it is impermissible to abort the embryo/fetus for the woman’s sake.

Restricting abortion could also restrict doctors’ freedom to practice medicine. Presumably, if restrictions on abortion interfere with a doctor’s ability to provide a woman with health care, that is a reason why the doctor’s freedom to practice medicine could be important. However, doctors being interfered with in providing such care would not be a decisive reason against restrictions, if the embryo/fetus has a right not to be aborted. In addition, some doctors consider the embryo/fetus to be their patient, in which case absence of restrictions on abortion could interfere with their ability to provide the best health care for the embryo/fetus. However, a doctor’s being interfered with in providing medical care for an embryo/fetus would not be sufficient reason to stop a woman from exercising a right (if she has it) to perform her own abortion or to seek another doctor willing to perform the abortion.

I have been arguing that it is not best to treat restricting or not restricting abortions as simply a health policy. This is so, even if the presence or absence of restrictions affects health and how it affects health plays a part in justifying a policy. For both the woman and the embryo/fetus, respectively, the right either to seek an abortion or not to be aborted can be based on factors other than health that ground such rights. Furthermore, the woman’s right to seek an abortion is a presupposition of allowing her health to decide matters and the right of an embryo/fetus not to be aborted is a presupposition of allowing its health to decide matters.

Hence, in what follows, I will assume that considering abortion’s moral permissibility or impermissibility is a precursor to discussing abortion as simply a matter of health policy. I will assume that (i) showing that abortion is morally permissible is necessary to argue that a health policy may pursue women’s health interests while ignoring embryo/fetal health interests and (ii) showing that abortion is morally impermissible is necessary in order to argue that a health policy may pursue embryo/fetal health interests while ignoring (or minimizing) women’s health interests.

Defending abortion even of persons

In standard innocent threat cases, the innocent threats are usually persons rather than, for instance, large birds that present a threat to someone. Presumably, this is because there would be much less if any moral problem with harming a nonperson (such as a bird) to stop the threat it presents to a person. However, most restrictive abortion laws in the United States do not claim that the embryo/fetus is a person from conception onward nor even that it is a person at the point at which these laws set limits on abortion (in case this is later in pregnancy). For example, laws that rule out most abortions once there is a fetal heartbeat do not claim that having human DNA and a heartbeat is sufficient for being a person. It is puzzling that there could be limits on abortions done for the sake of a woman who is a legally recognized person to preserve a type of being that is not a legally recognized person and who is occupying her body. If a bird with a heartbeat were in her body, there would be no such restrictions.Footnote ⁸

In my previous work on abortion,Footnote ⁹ I follow Judith Thomson, who assumes merely for the sake of argument that the embryo/fetus is a person from the moment of conception.Footnote ¹⁰ In what follows I will also assume for the sake of argument that the embryo/fetus is a person. I will first review some aspects of Thomson’s and my previous work defending the permissibility of abortion on this assumption because those arguments might imply that there should be no time limits on abortion, even if the embryo/fetus is declared a person from conception.Footnote ¹¹ Below, I will reconsider this view and also consider time limits on the assumption that the embryo/fetus develops into a person or what has the moral status of a person.

Thomson presents a series of hypothetical cases that suggest moral innocence does not prevent a person from being permissibly killed if this is necessary to stop them from being a threat to another’s life, health, or significant bodily autonomy. Her cases include the following (among others). (a) A woman is in a house with a rapidly expanding baby who threatens to crush her to death. Thomson thinks that if there is no other way to save herself, the woman may kill the baby. She then mentions a variant in which the house they are in belongs to the woman, suggesting that this strengthens the woman’s right to defend herself. (b) A violinist is dying of kidney failure and his friends attach him to another person (assume it is a woman), without her consent or his knowledge, to use her kidneys for nine months by which time he will recover and can safely be disconnected. In this case, neither the woman’s life nor health is at stake. Nevertheless, Thomson thinks that the woman may (i) detach the violinist, foreseeing his death due to lack of life support, or even (ii) directly, intentionally kill the violinistFootnote ¹² in order to remove him. She thinks that this is because he has no right to use her body for nine months to save his life, even when this would not threaten her life or health. In these first two cases, the baby and the violinist are morally innocent threats in that they are not at fault, morally responsible, or even agents when they (to different degrees) threaten the woman. (c) People develop from seeds that come into a woman’s house through a window that she opened. This is so despite her having taken precautions (screens) against the seeds coming in and taking root. Thomson thinks that the woman has not given a person who now develops from the seed and occupies her house any right to be there, despite her having voluntarily opened the window. Therefore, she thinks, the woman may resist their presence, as in cases (a) and (b), even if this will cause their death.

I think Thomson believes that cases (a)–(c) are counterexamples to the claim that an innocent person may not intentionally be killed to stop their imposing on others. Furthermore, I think that she believes the cases also show by analogy that if a woman becomes pregnant involuntarily either by rape (analogous to the violinist case) or through voluntary sex despite contraception (analogous to the seeds-through-screen case), the embryo/fetus has not been given a right to be in her body. She thinks that this implies that even if it is a person, the embryo/fetus may be killed if necessary in order to remove it, at least by the woman herself, to preserve her bodily autonomy even when her life or health is not threatened.

Thomson also argues for the moral permissibility of a third party (such as a doctor) helping the woman to kill the person threatening her in all the hypothetical cases and in pregnancy. She says that it would be an act of a “good Samaritan” to help women who cannot free themselves of such threats and who do not want to be what she calls “splendid Samaritans” by letting others impose on their bodies.Footnote ¹³ Thomson’s view that a doctor’s performing an abortion or detaching the violinist is the act of a good Samaritan implies that there is no professional duty but merely a protected permission to do so at the woman’s request. This could free a doctor who objects to abortion from having to perform one. However, her view on abortion might also be consistent with a doctor having a duty to perform an abortion, given that it is morally permissible and in the woman patient’s health interest.

It is noteworthy that throughout her discussion Thomson focuses first on the right of the person who is threatened to act on her own behalf if she can. This raises the issue of whether doctors should not be called upon to perform abortions when women are themselves physically able to safely abort (for example, by taking a pill), especially if states that ban abortion by doctors would not interfere with or prosecute women who perform their own abortions. Abortion pills work by interfering with an embryo/fetus getting the sustenance it needs and its dying as a consequence rather than through a direct attack on it. This is like the violinist who dies after being detached; the death is a foreseen side effect of the woman exercising her right to bodily autonomy and is not an intentional direct killing.

Here are some concerns about aspects of Thomson’s arguments.Footnote ¹⁴ First, Thomson’s argument for direct killing suggests that whenever someone does not have a right to use something of ours even to save their life, we may directly kill them to stop their use of that item. Call this extrapolation of Thomson’s argument Claim 1. Claim 1 implies that if I need not provide someone with $10,000 to save their life, but they were given it contrary to my wishes and even theirs, I could permissibly directly kill them if that were the only way to get the money back. Indeed, Thomson’s argument suggests that I could do this, even if the money were not being used to save their life, because it is still something that they would have no right to get even to save their life. However, it does not seem permissible to directly kill someone if this were necessary to get back my $10,000 whether they are getting a life-saving benefit from it or not. Hence, while someone might not have a right to something even to save their life, they might have a right not to be directly killed to stop their having it. However, it might still be permissible to retake the $10,000 without direct killing, even if the funds were saving someone’s life and it could be foreseen that he would die as a result of not having the assistance.

Second, Claim 1 does not provide the best justification of killing in the violinist case or in the case of abortion, because in these cases someone is having his life saved by using someone else’s body for nine months, which is (arguably) a more serious imposition than merely possessing another’s money. If the violinist is killed, he loses out on the life that he would not have had without the assistance to which he had no right even to save his life (that is, the use of the other person’s body). His being attached to the woman’s body is necessary (if not sufficient) for his having many future years of life and his attachment is ended by taking from him the period of life he is in the process of receiving from the attachment. Hence, killing this innocent person to stop his being a threat costs him only the loss of the good (life) he could not have without his imposition on the woman. That combined with the particular type of imposition (use of someone’s body) is why he may have no right to continue to impose in order to avoid being killed. This is so, even though he would lose his life if his imposition is ended and the woman would lose only bodily autonomy if the imposition were to be continued. These facts can also be present in the case of abortion and could similarly justify killing the embryo/fetus, even if it is assumed to be a person. I believe that Thomson’s discussion does not sufficiently emphasize that the innocent threat will lose only the benefit made possible by imposing or that this can be crucial to the argument for the permission (especially of a third party) to kill the violinist or the embryo/fetus.Footnote ¹⁵ Arguably, this is morally important in justifying killing the violinist or embryo/fetus when imposition on the woman’s body by itself is much less of a harm to her than being killed is to the violinist or embryo/fetus. (Note that the baby in the house does not benefit from imposing on the woman and, in being killed, does not lose a benefit made possible by imposing on her. Killing it will make it worse off than it would have been if it did not threaten her or occupy what belongs to her.Footnote ¹⁶ It could thus be more difficult to justify killing the baby were it not that it threatens death or some other serious harm to the woman.)

An embryo/fetus derives the benefit of continued existence by using a woman’s body, but its situation may in other respects differ from the violinist’s even if it were also a person. First, unlike the violinist, it does not need to be attached to the woman’s body to begin with to save its life, because it did not exist prior to being in the woman’s body. Hence, the violinist is not overall harmed by being attached to and then removed, because he also faced death before attachment. By contrast, it is not clear whether it is overall not bad for an embryo/fetus to live a short while attached to the woman and then have its life terminated. One way of deciding this is to consider whether we should require significant sacrifices from a woman to prevent a pregnancy that we know will unavoidably end in a natural miscarriage. That is, would we require such sacrifices out of concern for the embryo/fetus being (broadly speaking) “worse off” for living and dying than never having lived at all?Footnote ¹⁷ Even if the answer is “no,” we probably must meet a higher standard in deliberately creating a person (as in voluntary pregnancy), out of respect for persons, than that they will be no worse for living and dying than never having lived. Hence, one question that could bear on the moral permissibility of abortion, if the embryo/fetus were a person, is how much one must do to avoid being in the situation of having created a person (voluntarily or not) who will not get what is involved in meeting that higher standard.Footnote ¹⁸

A second way in which the embryo/fetus differs from the violinist is in having a genetic relation to the woman (at least in nonsurrogate pregnancies). Could this make a difference in what she owes to an embryo/fetus (by contrast to the violinist)? Suppose an embryo/fetus (assuming it is a person) is begun in a laboratory, for example, by a third party using the woman’s egg without her consent. It then needs to be put in the woman’s body for continuing gestation. Would this embryo/fetus have a right to the woman saving its life by putting it in her body? If not, one could not argue that she may not abort it simply because she owes it use of her body to save its life due to their genetic relation.

Third, it might be thought that the “unnatural” factors present in the violinist case—for example, a person’s body is not “designed” to provide support to needy violinists and there is a biological malfunction in the violinist who needs such support—might make conclusions about that case not relevant to pregnancy. This is because it is natural for wombs to harbor embryos/fetuses and it is no malfunction for embryos/fetuses to need to be there. Perhaps a better analogy to pregnancy that eliminates these “unnatural” factors might involve a house with a room designed for occupancy by someone besides the house’s owner when the tenant does not lay claim to the entire house but only to the room. Thinking that the natural-nonnatural distinction (or genetic relation) might affect what is morally permissible to do to the embryo/fetus versus the violinist, seems to concede that more is needed to argue against abortion than that one may not deliberately kill morally innocent persons. However, Thomson herself notes that even if nature turns someone into an unwilling splendid Samaritan, it is permissible to resist being in this position. What is natural (for example, volcanic eruptions that kill people) need not be good or right.

Fourth, Thomson imagines that the violinist requires life support from someone else’s kidneys for the full nine months so that even if he is not directly killed, he would die from lack of support if disconnected early. Would it make a moral difference to the permissibility of ending his attachment if at some point prior to the nine months the violinist did not need further life support from the woman’s body were it possible to safely remove him? This case, unlike Thomson’s, would involve a violinist who, like some embryos/fetuses, was “viable”—that is, could live without dependence on the woman’s body were he safely removed—after a certain point. He would go either to a mechanical life-support device or to living on his own. However, suppose that in this viable violinist case and in the viable embryo/fetus case supererogatory efforts by the woman—that is, efforts she has no duty to make—are needed to get either the viable violinist or viable embryo/fetus safely outside the woman’s body and, if they remain attached to her body, they continue to get life-saving support from her body. Suppose further that either one being killed instead of being separated from her body is much less costly to the woman to whom they are attached than any other supererogatory removal procedure. In this case at least, if the embryo/fetus or the violinist is killed, they would still be losing out on life they could only have had by way of supererogatory efforts the woman would have had to make to have them safely removed and supererogatory efforts she would be making in continuing life support to them while they remain attached to her body.

Time limits and compromises about them

Suppose that some of those who believe that the embryo/fetus is a person (or has the moral status of one) from conception onward do not accept the arguments for the permissibility of aborting innocent threats that we have examined. Suppose they think that the embryo/fetus always being a person rules out abortion from conception onward. It is often argued that a willingness to compromise on public policy is necessary in a liberal democracy where citizens hold many different views. A contributor to The New York Times op-ed page argues that placing a time limit on abortions by only allowing them during the first six weeks of pregnancy was an acceptable compromise between those who favored Roe v. Wade’s wider period of permissible abortions and those who think abortions should be banned from conception because the embryo/fetus is then already a person.Footnote ¹⁹

However, it may not be true that a compromise that allows only early abortions is a way to show respect for the view that abortion should be banned, because embryos/fetuses are persons from conception. This is because a compromise that shortens the time from conception during which abortion is permissible provides an embryo/fetus that is permissibly killed a shorter life than it would have had if it had been killed later in pregnancy. If a person benefits by living a longer rather than a shorter life, then the particular embryo/fetus aborted early would have been better off had it been aborted later. This seems to imply that compromising with the view that abortion is problematic because a person begins at conception should lengthen the period during which abortions are permissible rather than shorten it, and perhaps even require that women who will be having abortions have them later rather than earlier. This would be a different sort of time-limit policy, limiting early abortions. Call this the Puzzling Compromise. It implies that permitting earlier abortions would be a compromise that favors women, not the embryo/fetus, by reducing their time of being pregnant before a desired abortion.

By contrast, suppose that a person develops gradually, not yet existing at conception and existing later in pregnancy. Assuming that it is worse to kill a person than a nonperson, a compromise that shortens the period in which abortion is permissible might respect the views of those (unlike Thomson) who think that being a person makes abortion impermissible. Hence, supporting laws that shorten the period from conception during which abortion is permissible suggests that one does not think that the embryo/fetus is a person from conception.

There are several possible ways to argue against the Puzzling Compromise that permits only late abortions. One way claims that it is only worthwhile for a person to live a longer rather than a shorter life if the life is of good quality. Given that under the Puzzling Compromise an embryo/fetus could be aborted before birth in any case, the quality of its life would only be the quality of life in the womb. If this is sufficiently low, it might not matter that the embryo/fetus dies sooner, and so there is no reason to require that abortions occur later. An indication that quality of life in the womb is sufficiently low is that one would not keep a being alive for long outside the womb if it could only ever have that quality of life. However, those who object to using quality of life to help determine whether it is permissible to kill or let someone die should reject this counterargument.

A second way to argue against the Puzzling Compromise takes account of the number of abortions. If the period in which to have an abortion is shorter, there are likely to be fewer abortions overall than if it were longer. Hence, it could be argued that the correct compromise with those who think that abortion is wrong is to restrict the time from conception for permissible abortions. However, suppose that some of the early abortions women were to have during such a time-restrictive policy would have been delayed under a less time-restrictive policy so that those embryos/fetuses would have lived longer. Interfering with those embryos/fetuses living longer in order to minimize the total number of abortions may be morally problematic if the embryo/fetus has the status of a person. That is, the question becomes whether it is morally permissible to kill a small number of people somewhat earlier than one could have, giving them shorter lives, in order to ensure that a greater number of other people will survive to be born alive. This raises an interesting issue about permissible trade-offs between different people’s lives from the point of view of those who believe that the embryo/fetus is a person from conception. One possible solution to this problem is to have a compromise that requires women to decide very early in pregnancy whether to have an abortion, but also requires the women who meet that condition actually to have the abortion as late as possible.

A third possible counterargument to the Puzzling Compromise is based on the fact that once a pregnancy is far enough along, the embryo/fetus could be removed alive. Given this possibility, why would those opposed to aborting persons agree to a compromise involving late abortions rather than late safe removals? If such a removal that saves the embryo/fetus does not impose on the woman any more than a late abortion would, then insisting on the abortion would make it clear that what the woman wants is that a child not be born rather than not having to go through a pregnancy. If this were a morally legitimate desire, it might support prohibiting safe removal as the alternative to late abortion. A different compromise in the spirit of the Puzzling Compromise that also takes this desire that a child not be born into account might involve permitting only abortions later in pregnancy but not so late that safe removal is possible.

Now consider the issue of time limits on the assumption that a Thomson-like argument is correct. This view that abortion can be permissible, even if the embryo/fetus is a person from conception, may suggest that it would be wrong to have any time limits on abortion. If this were so, the laws that many states have passed placing time limits on abortion except when a woman’s life is at stake would be unjustified. Despite the assumption Thomson makes for the sake of argument that the embryo/fetus is a person from conception, she believes that the embryo/fetus is not a person from conception, but that the embryo/fetus becomes a person at some point before birth.Footnote ²⁰ It might be thought that in this case as well there should be no time limits on abortion. A possible argument for this is as follows. (i) Because killing a nonperson for the sake of a person is justifiable, abortion should be permissible before the embryo/fetus becomes a person. (ii) If abortion is permissible when the embryo/fetus does not have the status of a person and it is permissible even if it is always a person (as claimed by a Thomson-like argument), then (iii) abortion is permissible whether or not the embryo/fetus is a person, and so there should be no time limits on abortion. Call this the No-Time-Limit Argument.

We will consider whether this argument is correct or whether the embryo/fetus becoming a person would make it problematic to defend later abortions by using a Thomson-like argument. That is, we will consider whether (i) and (ii) imply (iii). Contrary to the No-Time-Limit Argument, gradual development may create problems for using an argument that assumes the embryo/fetus is always a person to defend abortion after the point (if there is one) at which it becomes a person. That is, we cannot derive (iii) from (i) and (ii). This is because if there is a period during which the embryo/fetus is not a person and killing a nonperson is morally less serious than killing a person, failing to act during that period might justify limiting one’s right to abortion once the embryo/fetus is a person. This way of thinking about time limits on abortion is not based on thinking that it is always wrong to kill innocent people to stop their using someone else’s body. Rather, it is based on thinking that if one has not made sufficient effort to end a pregnancy before it involves killing a person, one may forfeit the right to end the pregnancy by killing a person through having an abortion. Unlike restrictions on abortion that focus on stages of development of an embryo/fetus in pregnancy per se (like those in Roe v. Wade), the view we are now considering would not be concerned with such stages had someone been unable to have an abortion before a person developed. For the view we are now considering, the moral issue could become how much effort one is morally required to make to end a pregnancy before doing so would involve killing a person.

Here is a possible way to judge the merits of this approach to time limits.Footnote ²¹ Suppose that if a woman does not take a harmless pill every week, the embryo/fetus will drop out of her body and die. If she does not take the pill, she will be letting the embryo/fetus die by refusing it the minimal aid provided by the pill in addition to not housing it in her body. The argument presented above for the permissibility of killing an embryo/fetus that is a person from conception rests on its being deprived of life it gets from use of a body to which it has no right even to save its life and without which it could not go on living. Thus, if the embryo/fetus were a person from conception, according to that argument the woman could permissibly stop taking the pill at any point during pregnancy, even though the pill is only minimal aid to which someone might ordinarily have a right. This is because without the pill the embryo/fetus would still lose only life it gets from use of the body to which it has no right even to save its life.

Suppose now that an embryo/fetus does not begin as a person but develops into one. Would it be impermissible for the woman to let it die after it becomes a person by ceasing to take the pill simply because, had she done so sooner, she would have let it die as a nonperson? Suppose that it would still be permissible to omit taking the pill. This could be accounted for by the embryo/fetus still having no right to the use of her body, despite the delay, rather than by any burden of taking the pill per se. If this is so, it implies that in the case of the delayed abortion, too, the embryo/fetus would still have no right to the use of the woman’s body to save its life. The permissibility of aborting the embryo/fetus that is a person stems from the fact that in being killed, it loses life it is getting from use of a body to which it has no right even to save its life and this use is necessary for its having any further life. If in the case of the delayed abortion the embryo/fetus would continue to lack the right to use the woman’s body, this would help account for the continuing permissibility of an abortion, just as it accounted for the continuing permissibility of not taking the pill.

There are other ways in which time limits on abortion might be justified. In Creation and Abortion, I discuss the possibility of partial external gestation (PEG) devices and total external gestation (TEG) devices that would reduce the burden of pregnancy on women and in that way increase their autonomy.Footnote ²² With PEGs an embryo/fetus could be removed from a woman’s body at a certain point in pregnancy to complete its gestation externally. With TEGs pregnancy could be entirely outside the womb. Suppose that removing the embryo/fetus to a PEG is no more burdensome for a woman than getting an abortion. A woman might then not have a moral right to abort an embryo/fetus if it were a person rather than place it in a PEG.Footnote ²³ A device that could increase women’s autonomy in one way could restrict it in another way by limiting the permissibility of abortion. Now suppose that a TEG could have been used from the beginning of gestation, but a woman chose instead to have the pregnancy in her body. If the embryo/fetus is a person at the time an abortion is desired, the fact that it would have continued to survive in a TEG had the woman not had a pregnancy in her body may also limit her right to abort. This is so, even if it could not be removed to any PEG at the time the abortion would occur. This is because aborting the person would make them seriously worse off than they would have been in the TEG. Hence, in the context of TEGs, the right to abortion could depend on whether a woman has a right to try to bear a child in her body without forfeiting her right to abortion that would have been permissible in the absence of TEGs.

Employing even more extreme science fiction, we could imagine that even without time limits on abortions, there could be less time in which to have an abortion if pregnancy itself could be made much shorter (for example, six weeks). Shorter pregnancy would reduce the burden on women’s bodies. However, like restricting abortion to early in pregnancy, shorter pregnancy would also require women to decide much sooner whether to have an abortion. This is not a significant objection to shorter pregnancy or limiting abortion to early in pregnancy if women could know early enough that they were pregnant and should decide before they become pregnant whether they would have an abortion. Thinking about this decision might start once they began engaging in the sort of activities by which they could become pregnant.

Innocent threats and moral asymmetry

As already noted, a policy that gives priority to health may not always be morally correct, because it would harm or otherwise violate the rights of some people in order to keep others healthy. However, the unifying thesis of this essay is that it can be permissible to interfere with or even harm innocent threats when the goal is to give priority to the health of their potential victims. The questions we deal with in this section are when, how, and why it may be permissible to interfere with or even harm morally innocent threats during a pandemic in order to protect others from the spread of disease.

While many people who infect others with a virus in a pandemic arguably share a trait with embryos/fetuses in an unwanted pregnancy in that they are also (or come close to being) morally innocent threats, they differ from embryos/fetuses in important ways. In the context of a pandemic, I will use “innocent threat” to refer to someone who is not aware that they are spreading disease by virtue of being where they are and breathing. However, because it is typically their own actions that put them where they can transmit a virus to others, compared to a fully innocent embryo/fetus, they are more like “minimally responsible threats.”Footnote ²⁴ Also, unlike the embryo/fetus, someone can acquire knowledge that they do present a threat to others and that there are ways for them to prevent this threat. Hence, the longer a pandemic lasts, the more likely people are to be morally responsible for a failure to acquire such knowledge and use preventive measures. This contributes to there being a time limit on the moral innocence of many threatening people in a pandemic. By contrast, in the case of abortion, the issue was whether there was a time limit on ending the life of a morally innocent threat.

Another difference is that the embryo/fetus usually threatens one other person (the pregnant woman), but an innocent threat in a pandemic might spread a virus to many other people. They also differ in that the embryo/fetus presents a sure-to-occur imposition—at least loss of a woman’s bodily autonomy—whereas someone carrying a virus may only present a risk of harm to others.

Furthermore, even if an innocent threat in a pandemic will definitely harm someone in a large group, each person in the group may have only a small chance of being harmed. This also means that unlike pregnancy, where there is a determinate woman who would be imposed on, the victim of the innocent threat in a pandemic could be indeterminate. In some pandemics, each person may also be an innocent threat to every other person, whereas in pregnancy (prior to trying to evict the embryo/fetus) the woman typically presents no threat to the embryo/fetus. These differences might increase or decrease the loss that could permissibly be imposed on innocent threats in a pandemic to stop the threat they present.

Recall that the discussion of the embryo/fetus focused on its being a category (1) innocent threat. That means it benefits from being a threat and would lose only that benefit (and whatever else that benefit is necessary to bring about) when it is stopped from being a threat by taking away the benefit to it (its life) of its being a threat. By contrast, innocent threats in a pandemic are in category (2); they do not benefit from being threats. By this we mean that it is not causally necessary that they threaten others in order to receive something good for themselves, such as freedom of movement. It is true that when they have the benefit of freedom of movement (rather than being in lockdown), they can become threats to others. However, if there were no others around who could be threatened, they could have just as much freedom of movement as they would under normal circumstances. Hence, being a threat and imposing on others is not causally necessary for their having freedom of movement. By contrast, if there were no woman on whom the embryo/fetus could impose, it would not get the benefits of sustenance that her body provides. Since category (2) innocent threats do not (in this sense) get benefits from presenting a threat, they can lose more than a benefit received from being a threat when they are stopped from being threats. When we take action against such an innocent threat (or he takes action against himself) to prevent his harming others (for example, by reducing his freedom of movement), we cannot say we are only removing a benefit he was caused to have from being a threat. Hence, a category (2) innocent threat could be worse off than he would have been if he had not threatened others when certain things (such as being in lockdown) are done to stop his being a threat.

However, there are some other things that might be done to stop someone from spreading or maintaining a disease in a community that could make him better off than he would otherwise have been because these measures also protect him from disease. These measures include vaccinating him against disease when this helps protect him as well as others to whom he might transmit disease or requiring him to wear a mask that helps protect him as well as others (for example, a properly fitted N95 mask). This goes beyond using measures to stop his being a threat that merely leave him no worse off than he was before being a threat.Footnote ²⁵ The benefit to him is cited not to justify paternalistic action toward him, but to show that there sometimes need be no overall cost to him in preventing him from threatening others.

Nevertheless, means to stop the threat someone presents could be morally required, even if they make him somewhat worse off and even if it would not be permissible to impose them on him merely in order to aid others.Footnote ²⁶ For example, suppose that self-protective masks are unavailable. Wearing a mask that only prevents transmission then seems like a reasonable cost to pay to avoid presenting a threat (or a significant risk of a threat) of a serious virus to others, and so wearing the mask can be a duty. If someone knowingly does not do his duty, he is no longer morally innocent. It can then become permissible for authorized others to impose that cost on him. A virus transmitter who cannot do his duty—for example, he is totally paralyzed and cannot put on a mask—remains a morally innocent threat. However, because he would have a duty to make the effort were he able, it could also be permissible (for some authorized agent) to impose the cost on him. Another way to lose one’s moral innocence is if one becomes a threat through knowingly failing at low cost to prevent one’s becoming a source of infection to others, for example, by not easily preventing oneself from catching the virus.

Having a duty to pay some cost not to harm or present a significant risk of harm to another does not mean that one deserves to pay that cost or that it is in itself good that one pays it. Someone can be liable to pay a cost or have the cost imposed on him, even though it would be better if the cost were not necessary. By contrast, giving someone what he deserves is thought to be intrinsically good. Some hold that one is liable to bear a cost only if one has done something, but an innocent threat who merely breathes may not have “done” something in the morally relevant sense.Footnote ²⁷ Hence, some may think that there is only a “lesser evil” justification for this sort of innocent threat bearing the cost because his breathing would otherwise do much more harm to his victim. This proposal implies that if his potential victim could avoid being threatened at even lower cost, she should bear that cost rather than the threatener bearing the greater cost.Footnote ²⁸

My view is that there is a moral asymmetry between the threatener and the threatened such that the innocent threat has a greater responsibility to avoid harming than the potential victim has to avoid being harmed. Even if we are not morally responsible for being a threatener, we should take responsibility for our body to prevent its harming others (if we can) rather than placing the burden on others to avoid our harming them. For example, an epileptic who has done nothing to have this condition should take some measure to prevent hitting someone else during a seizure rather than expect others to take protective measures at the same or even slightly lower cost to prevent their being hit. I think that this is true even when both people have equal knowledge of when the threat will be presented.Footnote ²⁹

Nevertheless, when someone is a category (2) innocent threat, I think that there is a limit on how much they have a duty to do to protect others from the threat they present. For example, the innocent threat who would kill someone at whom he is hurled may have to swerve into a wall, though he foresees that this will break his leg, but need not swerve if he knows this will kill him.Footnote ³⁰ This is consistent with the permissibility of his potential victim (and even third parties) imposing on the innocent threat greater costs than he has a moral duty to impose on himself. An indication that he lacks the duty is that he could permissibly resist the higher costs being imposed on him, even if successful resistance leads him to continue to threaten his victim.Footnote ³¹

Furthermore, even if there is a moral asymmetry between innocent threat and potential victim(s), there is also moral reason for a potential victim to accept some cost to protect himself (if this is possible) from a category (2) innocent threat rather than impose a great cost on someone to prevent his being a threat. This need not be so if someone were deliberately threatening his victim. For example, a potential victim of an innocent threat might have a duty to wear an N95 mask to protect himself, if an innocent threat would have difficulty breathing were he to wear a mask that prevents transmission.

My aim here is not to defend further the asymmetry view. Rather, my aim in the next section will be to show that some practitioners and researchers of public health policy do not accept this view. I agree that we should not expect nonphilosophers to attend to all the distinctions philosophers may draw. However, it can be important to examine the work of nonphilosophers who believe that they, independently of philosophers, can discuss practical issues that may require philosophical distinctions. Discussions by medical and public health professionals can have great impact on public policy; philosophers considering those discussions (and working with health professionals) can benefit the public.

Rejecting the moral asymmetry of innocent threats and victims

Here is an example of a discussion that does not accept a moral asymmetry between someone who would harm and someone else who would be harmed by them. Nancy Kass says in discussing public health ethics: “Disease reporting is … distributionally unfair, in that the burdens of the program are borne by those with the disease, generally for the benefit of those who do not have the disease. The unevenness … may be justified … when the benefits are important and when there are no less burdensome ways to achieve them.”Footnote ³² Kass views the “unevenness” in this situation as if it involves imposing a burden on a bystander who presents no threat to help someone else. But presumably, when those with the disease are burdened and those without the disease are thereby “benefited” (to use Kass’s term), it is because the burdens prevent those with the disease from transmitting the disease to others. Hence, the “benefit” of which Kass speaks does not improve the potential victim’s condition from what it is or would have been, which is what a benefit does; the burden on the threatener only prevents him from making another’s situation worse than it is or would have been. Hence, a threatener being prevented from harming another should not be referred to as his being made to “benefit” another.Footnote ³³

This is important because some who oppose being required to take preventive measures (such as wearing masks) think that it is morally wrong to be required to provide benefits to others. Hence, their objection could be met by explaining to them that being prevented from making others worse off does not constitute being made to benefit them.Footnote ³⁴ By contrast, a health policy that would require vaccinated people to circulate in the community to provide a protective “cocoon” around those who cannot be vaccinated would require some to benefit others. This is because the cocooned are made better off than they would otherwise have been. However, if a third party (for example, a government agent) prevents someone else from harming you, that agent does benefit you because they make you better off than you would have been without their involvement.

Furthermore, Kass’s view is that, in principle, it is morally arbitrary to distinguish those who would transmit the disease from those to whom they would transmit it. It is only when burdening the person who would transmit the disease is a lesser evil that the burden is not morally arbitrary. This implies that if burdening the potential victim would prevent harm to him at even slightly lower cost than burdening the threatener, the former should bear the cost. If the cost to each were equal, a toss-up would be morally correct. Kass implicitly denies that when someone’s body presents a threat, just in virtue of that he should bear more of the burden of preventing harm than the potential victims.Footnote ³⁵

Similarly, discussion by some economists about the public health policy of lockdowns during COVID-19 is based on (i) the moral equivalence of those who would harm and those who would be harmed and (ii) misuse of the idea of benefiting. For example, while supporting the use of lockdowns to prevent transmission, Olga Yakusheva expresses concern that lockdowns’ associated costs are not shouldered “only by those who benefit from a lockdown,” but also by those who do not benefit (because they are not vulnerable to bad effects of a disease). Similarly, Jay Bhattacharya characterizes lockdowns as efforts to protect the vulnerable that cause “collateral damage” to others less susceptible to the virus.Footnote ³⁶ Yakusheva does not recognize that one does not benefit others simply because one does not harm them by transmitting disease. However, insofar as a lockdown is instituted by governmental agents, it is true that they may benefit those who would otherwise be harmed by transmitters. She also does not recognize that it may not be morally problematic if costs are borne by those who do not benefit when the costs prevent them from causing harm. This could be especially true if they are required to be in lockdown through first failing to pay a much smaller cost (for example, wearing a mask) to prevent their causing harm.

“Collateral damage” is a phrase in just war theory describing unintended harmful effects to nonthreatening civilians of, for example, hitting a military target. Bhattacharya’s applying it to those who are in lockdown is inapt. First, the lockdown of people is not a mere side effect; it is an intended means to an end of protecting the vulnerable from transmission and/or stamping out the virus. This need not mean that the burdens that result from lockdowns are themselves intended. Nevertheless, intending to involve someone in a lockdown while foreseeing that some burden (for example, economic harm) will come to them is still unlike merely foreseeing but not intending both someone’s involvement and harm (as in collateral harm in war).Footnote ³⁷ This could make lockdowns morally more problematic than collateral damage in war were it not that lockdowns are intended to stop those who would threaten others with disease, whereas innocent civilians in war are not threats. Neither Yakusheva nor Bhattacharya, who are health-care professionals and economists, considers the possible moral significance of the harmful causal role (or risk of this) of at least some of those who bear a burden and that the burden prevents them from causing harm to others rather than benefiting those others.

Some clinical and public health researchers use the Trolley Problem (much discussed by philosophers) when searching for analogies to either the threatener or the potential victim in a pandemic.Footnote ³⁸ Again, no moral asymmetry is drawn between these two types of people. For example, in the realm of clinical ethics, Joshua Greene reports (pre-pandemic) that medical doctors have psychological difficulty in isolating patients with infectious diseases.Footnote ³⁹ He says that doctors analogize imposing isolation on such patients to harming an innocent person on a sidetrack by redirecting a threatening trolley toward him and away from hitting five other people. However, this analogy is not correct, because the one person on the sidetrack does not threaten those for whose benefit he is imposed on, whereas an infectious person does threaten those for whose “benefit” he is imposed on. Hence, it seems to require a different type of argument to justify harming the one person in the Trolley Problem from what it takes to justify harming the infectious person to stop him from threatening others.

Margaret Mitchell and Graham Attipoe use (and abuse) the Trolley Problem in public health ethics specifically to understand a pandemic.Footnote ⁴⁰ After identifying the Trolley Problem as originally having been presented by Philippa Foot in 1967, they describe two later versions of the problem (introduced by Judith Thomson): (a) the version in which a person has to be pushed from a bridge and in front of the trolley to stop it (the Bridge Case) and (b) the version in which a bystander (not the trolley driver) must turn the trolley from five people toward killing one other person (the Bystander Case). Mitchell and Attipoe claim that the public health analogy to (a), preceding development of a vaccine, requires medical personnel to risk their own lives (by treating contagious people) to save their patients from the virus. This comparison fails to note, though, that while the person on the bridge has no duty to jump in front of the trolley, medical personnel have a duty to take care of patients. The question of how high the risk of harm to personnel must be before that duty is overridden is something the authors do not discuss. In addition, the analogy fails to note that in the Bridge Case the one person being hit by the trolley is the intended causal means to stopping the trolley from hitting the five other people; if the one person toppled from the bridge is not hit, the five will not be saved. By contrast, while doctors may get sick as a side effect of saving their patients, they could save their patients even if they do not get sick. This is because their being threatened by the virus is not a causal means to saving their patients from the virus.

Regarding the analogy to the Bystander Case in (b), it seems to Mitchell and Attipoe that the virus presents an economic threat to many people, through lockdowns, failed businesses, and so on. If the virus threat is redirected, as the trolley is redirected in the Bystander Case, then instead of economically harming many people, it will kill fewer other people. In this pandemic case, unlike the Bystander Case, not all people face the same threat of death from the threatening entity; the larger group faces economic threats, while the smaller group faces the threat of death. Hence, the authors wonder whether avoiding the economic losses to each of many justifiably overrides the claims of fewer others not to be caused to die.

Above, we discussed medical and public health views that treat morally innocent threatening people as no different, morally speaking, from those they threaten. Mitchell and Attipoe’s use of the Trolley Problem analogy goes even further. It compares those in lockdown (who are potentially threateners) to the five people who are threatened by the trolley. Hence, those in lockdown are put on equal footing with their potential victims as people who could be threatened by the virus (admittedly, less seriously through economic effects of lockdown). It should also be noted that while the authors represent the virus as threatening some with lockdown, it is actually governments that cause lockdowns in response to the virus. In this respect, a better trolley analogy than the Bystander Case would be Foot’s original caseFootnote ⁴¹ in which the trolley driver (like the government) rather than a bystander presents a threat to the larger group of people. This is because the bystander would merely allow a threat to the larger group to continue (if he were to do nothing) rather than present the threat to them.

However, if the threat to the larger group in a pandemic is a lockdown and its economic effects, it is not this threat that would be redirected to the smaller group of people were they to die if the lockdown is lifted. As already noted, this is unlike the Trolley Problem where the very same threat (of death) is redirected. Furthermore, Mitchell and Attipoe fail to notice that how the harm to others would come about in lifting a lockdown is different from how it comes about in redirecting the trolley. That is, the people saved from the trolley do not then threaten the one person on the sidetrack in the way those released from lockdown would then threaten others. In this respect, a better analogy to the problem of lifting lockdowns is one version of what I call the Lazy Susan Case. In this case, the trolley is immovable but the five threatened by it are on a large swivel table that can be turned so that the five are removed from the threat.Footnote ⁴² However, as they are turned away from the threat, they become innocent threats when their bodies bang into and kill another person near the edge of the table. Although those saved from the trolley in this Lazy Susan Case threaten others, as do those released from lockdown, there is still a crucial moral difference between the cases. In neither the Trolley nor Lazy Susan Cases does the threat (of the trolley) to the greater number arise originally because it is a way of stopping the greater number from presenting that threat to others, a threat that they will again present if the threat to them is removed or if they are removed from the threat to them. By contrast, in a pandemic the threat of lockdown to the greater number originally comes about as a way to stop them from transmitting or sustaining a virus. If the lockdown is lifted, they could again be innocent threats to others. They may then also engage in activities while knowingly failing to use measures (such as wearing masks) to mitigate risks and so be noninnocent threats.

Despite not recognizing these disanalogies between the Bystander Case and the choices involved in having or lifting lockdowns, Mitchell and Attipoe do recognize that when people are freed from lockdowns, it is they who could threaten others. The authors’ specific moral concern about this is that the threat is underplayed because the causal connections leading to harm to others are not obvious and the harm is not immediate. They further think that downplaying harm when its causes are not obvious and not immediate is due to reliance on the Doctrine of Double Effect (DDE). They understand the DDE to morally prohibit directly harming—such as pushing someone into harm’s way, as in the Bridge Case—but morally permit indirectly harming that comes about, for example, by flipping a switch to turn the trolley that kills someone or by ending lockdowns and thus opening up an economy with nonobvious causal links between acts and distant harms of viral illness.

This is a misunderstanding of the DDE. The DDE actually distinguishes morally between intended and merely foreseen harm. Hence, the DDE would rule out intentionally toppling the person in front of the trolley, even if that were to come about by flipping a switch so that by a nonobvious and distant causal chain it results in the toppling. Furthermore, the DDE would not rule out using a gas to save five people (to use Foot’s example), even though one were to foresee that the gas would obviously and directly kill an innocent bystander as an immediate but unintended side effect. Mitchell and Attipoe think that the same principle—in their view, the DDE—that would permit opening the economy also permits turning the trolley by the indirect means of flipping a switch in the Bystander Case. Hence, they are concerned that the permissibility of turning the trolley can be used to justify the problematic results of releasing people from lockdown. What speaks against their analysis is that the bystander can know that the trolley he turns will clearly kill the one person on the sidetrack directly and in short order, yet still find it permissible to turn it. By contrast, the authors think that when people are released from lockdown it is indirect, not immediate, and not clear how this would cause others to die and they think that this leads to (mistakenly) thinking that lifting lockdowns is permissible.

I think that the permissibility of turning the trolley, but the possible impermissibility of lifting lockdowns, could be due to other factors (to which I have already pointed). First, lifting lockdowns, by contrast to turning the trolley, involves removing a threat (of lockdown and its effects) that keeps in check those who initially presented and would again present threats (or risk of threats) of disease to others. Second, as Mitchell and Attipoe also emphasize, each person in lockdown stands to suffer less harm than would come to each person made ill by the release of those in lockdown. Consider this second point in more detail. Those constructing public health policies are said to be more consequentialist in not abiding by deontological side constraints when these could interfere with saving more lives.Footnote ⁴³ However, they would be shown to be averse to consequentialist aggregation of all harms and benefits and more prioritarian in their views, if they were to give priority to avoiding great individual harms (for example, deaths) to fewer people rather than avoiding lesser individual harms (for example, economic losses) to more people. By contrast, an additive aggregate of the lesser harms might be large enough to lead a nonprioritarian consequentialist to decide otherwise. Furthermore, their giving priority to lives need not be just a matter of giving priority to the “separate sphere” of health when it conflicts with economic concerns. They might also give priority to preventing fewer deaths rather than a large aggregate of lesser individual health problems, contrary to what a nonprioritarian consequentialist would recommend.Footnote ⁴⁴

However, another difference between trolley and pandemic cases favors ending lockdowns. In the trolley case, it is certain that the one person will die if the trolley threat is redirected. By contrast, the probability of any given person being seriously harmed either by transmission of a current virus variant or continued production of new variants, if lockdowns are eliminated, may be much less. The trolley analogy to this would be diverting the trolley away from killing five people and into a crowd of thousands of people where it will kill only one of them. The lower risk (ex ante) to each one of the thousands does not decrease the number of people who will be harmed (ex post) from that in the traditional trolley case, but it may still be morally significant that the risk to each is low.Footnote ⁴⁵ Thus, public health statistics and policies that focus only on the total number of people who will die of a disease may ignore a morally significant factor—such as the low risk to each of dying—that bears on the moral permissibility of lifting lockdowns.