Mitochondrial Replacement Techniques, the Non-Identity Problem, and Genetic Parenthood

Abstract

Mitochondrial replacement techniques (MRTs) are designed to allow couples to have children without passing on mitochondrial diseases. Recently, Giulia Cavaliere and César Palacios-González argued that prospective parents have the right to use MRTs to pursue genetic relatedness, such that some same-sex couples and/or polygamous triads could use the process to impart genetic relatedness between a child and more of its caregivers. Although MRTs carry medical risks, Cavaliere and Palacios-González contend that because MRTs are identity-affecting, they do not cause harm to an existing human being, and our intuitions otherwise arise from the non-identity problem. Here, I review several attempts to address the non-identity problem, and propose a solution to the problem. Furthermore, I argue that regardless of one’s stance on whether MRTs are identity-affecting, the use of MRTs to pursue genetic relatedness alone falls outside the scope of the medical profession, as they involve substantive medical risk for no medical benefit.

Introduction

Normally, in cases of human sexual procreation, two genetic parents—one female and one male—contribute roughly equal genetic material, resulting in a zygote containing nuclear DNA (nDNA) in roughly equal parts from both parents, and mitochondrial DNA (mtDNA) from the mother. In a recent article, Cavaliere and Palacios-González (2018) argue that some prospective parents—lesbian couples, some polygamous triads, and perhaps other groups—should be allowed to use mitochondrial replacement techniques (MRTs) to produce zygotes with DNA from three parents—nDNA from a male and female, and mtDNA from a second female. MRTs, they argue, would allow these parents to satisfy their desire to have children genetically related to each of these parents; something that could not be achieved otherwise.1

Some philosophers might appeal to a therapeutic/non-therapeutic distinction to explain why MRTs meant to treat mitochondrial diseases are morally acceptable, while those that are meant for other reasons, like genetic relatedness, are not. According to a therapeutic/non-therapeutic distinction, an intervention is only medically appropriate if it is meant for therapeutic and/or preventative purposes, while non-therapeutic interventions, such as enhancements, cosmetic alternation, and the like would fall outside the scope of medicine and thus be inappropriate for a physician, qua physician, to perform.2 Cavaliere and Palacios-González (2018) argue that MRTs are not therapeutic because they are identity-affecting and thus run afoul of the non-identity problem (Parfit 1976, 1984; Schwarts 1978; Adams 1979). Thus, they argue, a therapeutic/non-therapeutic distinction cannot be used to show that some MRTs are acceptable, while others fall outside the scope of medicine.

However, this argument overlooks the medical rationale for preventative care. Even if we assume therapeutic interventions are identity-dependent, preventative interventions need not be. Some preventative interventions, such as a flu vaccination, are identity-dependent insofar as they’re intended to prevent possible future harm to a specific person; in this case the vaccinated person.

Yet, other preventative interventions, such as the prescription of folic acid to women who may become pregnant in the future, are identity-independent insofar as they’re intended to prevent possible future undesirable medical conditions to one (or more) unspecified members of a set of existing, or possible, persons; in this case, folic acid is meant to prevent birth defects in any fetus that might be conceived in the near future (Grosse et al. 2016). If MRTs are identity-affecting, then MRTs intended to prevent mitochondrial diseases are preventative interventions of this second kind; identity-independent interventions intended to prevent possible future undesirable medical conditions to whatever fetus is created by MRTs. The same cannot be said about MRTs intended to merely create genetic relatedness; identity-independent interventions that risk possible future undesirable medical conditions to whatever fetus is created by MRT with no medical benefit.

Cavaliere and Palacios-González go on to argue that the use of MRTs—including MRTs for mere genetic relatedness—fall under the scope of reproductive freedom. This proposal to use assisted reproductive techniques (ARTs) to help individuals better exercise their reproductive freedom is novel and interesting. Notably, this proposed use of MRTs skirts many problematic concerns about designer children, as parents are aiming for mere genetic relatedness, rather than selecting traits they wish their children to inherit.3

Although one may have a right to pursue genetically related children, it may be morally wrong to do so. Notably, Laura Purdy argues that it would be morally wrong to have a child when that child will likely not live a minimally satisfying life (Purdy 1995a, b). MRTs carry with them some risk, and because the techniques have only recently become available, their health risks are not fully known (Nuffield Council on Bioethics 2012; Reinhardt et al. 2013; Bredenoord and Hyun 2015; Gómez-Tatay et al. 2017). If MRTs are not therapeutic, such risks may mean both MRTs aimed at preventing mitochondrial diseases and those aimed at genetic relatedness may be morally unacceptable and fall outside the scope of medicine, such that even if individuals have a right to pursue MRTs, those in the medical profession cannot morally provide access to them qua physicians.

Here I argue that MRTs may not be identity-affecting, such that MRTs meant to treat mitochondrial diseases may be therapeutic. Then, I defend a solution to the non-identity problem by appealing to parallel cases. Finally, given the risks of MRTs, I argue that using them merely to pursue mere genetic relatedness falls outside the scope of medicine.

Mitochondrial Replacement Techniques

Mitochondria are small structures found within our cells that provide energy for cellular processes. Cavaliere and Palacios-González (2018, 836) note that mitochondria have been described as the “powerhouses” of cells, which “produce the necessary energy for cellular, organ and bodily function.” Mitochondrial DNA (mtDNA) is primarily inherited from the mother, although recent evidence suggests that in extremely rare cases some mtDNA may be inherited from the father (Schwartz and Vissing 2002; Ladoukakis and Eyre-Walker 2004). Mitochondrial diseases can be caused by mutations in both mtDNA and nuclear DNA (nDNA) and can cause serious health problems. MRTs are medical interventions designed to prevent mitochondrial diseases caused by problematic maternal mtDNA by replacing it with healthy donor mtDNA.

Cavaliere and Palacios-González discuss two kinds of MRT—pronuclear transfer (PNT) and maternal spindle transfer (MST). MST involves two unfertilized eggs—one with the desired nDNA and undesired mtDNA (usually from the intending genetic mother) and one with the undesired nDNA and desired mtDNA (from a healthy donor). Physicians remove the nucleus from both eggs and place the desired nucleus in the enucleated egg cell with the desired mtDNA. This egg is then fertilized and transferred to the would-be birth mother.

PNT involves two fertilized eggs—one with the desired nDNA and undesired mtDNA and one with the undesired nDNA and desired mtDNA. As with MST, physicians remove the nucleus from both eggs and place the desired nucleus in the enucleated fertilized egg cell with the desired mtDNA. This fertilized egg cell is then transferred to the would-be birth mother.

Although the goal of MRT is to replace undesired mtDNA with desired mtDNA, both PNT and MST result in a zygote comprised mostly of cellular material from the healthy donor; the desired nucleus is surrounded by the cytoplast from the donor. A cell nucleus is often thought of as the cell’s control center, so one might say performing an MRT on a cell is comparable to performing a lung transplant by transplanting the brain of the patient you’re trying to save into a donor body with the desired lungs (Wrigley et al. 2015).

Palacios-González (2017) contends that the term MRT is misleading, claiming a “better term” would be “nuclear replacement techniques.” However, this begs the question with regard to identity; there are two possibilities—either (1) MRTs are identity-affecting or (2) they aren’t. If (1), then MRTs are best described as destroying two previously existing egg cells and creating a new one (comparable to the story of Frankenstein’s monster4) as for Palacios-González MRTs would result in a new cell constructed from irreparable parts of two previously existing cells.

However, if (2), then either (2a) identity tracks the cytoplast or (2b) identity tracks the nucleus. If (2a), then “nuclear replacement techniques” is appropriate, but if (2b) then “cytoplast replacement techniques” would be more accurate. The problem with such terminology, however, is that it loses sight of the intervention’s goal; to get rid of undesired mitochondria (Newson and Wrigley 2017). Although contemporary MRTs take the cytoplast with the mtDNA, future MRTs need not. Suppose—as if by magic—some physician was able to replace all and only the undesired mtDNA with the desired mtDNA, such a procedure would achieve all medical goals contemporary MRTs are aimed at achieving but intuitively not be identity-affecting.

Are MRTs Therapeutic?

The distinction between therapeutic and non-therapeutic interventions can be helpful in determining the moral permissibility of a medical intervention. One issue here is that many non-therapeutic interventions simply seem to fall outside the scope of medicine. Physicians, qua physicians, are generally understood to have robust, special professional moral obligations to provide medical care. However, if an intervention falls outside the scope of medicine, there may be no obligation to provide such care. For example, a physician’s expertise might prepare them to be excellent crime scene investigators, hair stylists, or tattoo artists; but it would be absurd to require physicians, qua physicians, to perform these tasks, in part, because they fall outside the scope of medicine.

Several philosophers oppose the use of MRTs for any reason, arguing that the techniques are experimental, risky, and may have unforeseen negative effects that may substantively harm their targets and the progeny of their targets (Baylis 2013, 2017; Newman 2013; Morrow 2014; de Melo-Martín 2016, 2017). However, others argue that at least some MRTs may be therapeutic (Caplan 2015; Harris 2016; Johnson 2013; Wrigley et al. 2015; Smith 2014). Notably, Wrigley et al. (2015) draw a distinction between MST and PNT; on their view, MST is not therapeutic because it occurs prior to conception, and therefore there is no patient to benefit or harm, while PNT is therapeutic as it positively affects the existing embryo (Liao 2017, 23). Because of this, they argue PNT is morally preferable to MST. Some critics of Wrigley et al. argue that both MRTs are morally equivalent, neither are therapeutic as both are identity-affecting (Palacios-González 2017; Rulli 2017; Liao 2017).

Perhaps the most interesting of these arguments comes from Palacios-González (2017), who argues that even if we assume Wrigley et al. (2015) are correct about PNT not being identity-affecting, the offer of PNT may be. Recall that PNT involves taking the nucleus of one fertilized egg and placing it in the enucleated cell of another. In practice, these eggs are fertilized outside the womb via an ART and PNT is performed soon afterwards. Palacios-González (2017) argues that this fertilized egg is (probably) numerically distinct from the fertilized egg that would have formed if the parents chose to conceive naturally, as the gametes that fuse during the ART “would most certainly not have fused in the first place if PNT had not been chosen as the course of action.”

This is an interesting distinction as while (by assumption) the process of PNT itself is not identity-affecting, Palacios-González contends the means used to achieve PNT are. However, it is possible to construct a scenario where offering PNT is not identity-affecting:

Mix-up: Due to a mix-up at the lab, Laura’s cryopreserved egg was accidentally fertilized against her wishes. Physicians contact her, and she is furious—after all she possesses a mitochondrial disease she does not want to pass on to. Fortunately, there is another recently fertilized egg that contains healthy mtDNA and would otherwise be discarded, and physicians offer to perform PNT.

Here, the offer of PNT is not identity-affecting. As such, the offer of PNT on accessible fertilized eggs seems to be therapeutic.

In contrast, when the offer of PNT would be identity-affecting, it is identity-affecting in the same way that one’s dining choice, blood alcohol level, sleep schedule, or sexual position might be. It would be absurd for physicians or prospective parents to agonize about the moral significance of the identity-affecting aspects of such decisions. Insofar as the offer of PNT would be identity-affecting in the same way, it would be equally absurd to conclude that it being identity-affecting is morally significant.

Matthew Liao presents a more thorough criticism of PNT, arguing that the intervention is not merely identity-affecting but identity-destroying—it doesn’t merely change who will exist, it actively kills someone that exists and replaces them with someone else (Liao 2017). Liao defends the organism view, the theory that we are essentially organisms (Liao 2006, 2010b). According to him, a zygote is an organism that (Liao 2017, 22):

1. Begins to exist when the capacity to regulate and coordinate the various life processes is there.

2. The zygote persists as long as there is what may be called “organismic continuity,” which is the continuing ability to regulate and coordinate the various life processes.

3. The zygote ceases to exist when the capacity to regulate and coordinate the various life processes is permanently gone.

Liao contends we come into existence at conception, but of course many of the relevant life processes are not present at conception. To best understand this view, consider a similar view—the substance view, the theory that we are essentially rational substances (Beckwith 2007; Friberg-Fernros 2015; Lee 2004; Lee and George 2005, 2007, 2008; Lee and Grisez 2012; George and Tollefsen 2008). Substance view theorists contend that what makes us morally relevant is our capacity for reason, yet like Liao hold that we come into existence at conception, despite the fact that the post-fertilization egg cell cannot engage in reason. Although human fetuses lack a first-order capacity for reason, or the ability to immediately engage in rational thought, substance view theorists contend that (most) human fetuses possess a second-order capacity for reason, or the ability to develop first-order capacity for reason at some point in the future.

To have a first-order capacity to do x is to have the ability to immediately do x. For example, during the summer a normal, a healthy oak tree (with leaves) has the first-order capacity to photosynthesize.

To have a second-order capacity to do x is to have the ability to do x at some point in the future, provided that certain conditions are met. For example, during the winter, after its leaves are fallen, that same tree merely has the second-order capacity to photosynthesize; this is to say that when certain conditions are met (weather warms, it grows new leaves, etc.), then the tree will be able to photosynthesize.

Now, consider a healthy acorn. If acorns are numerically identical to the oak tree that may develop from them, then we might say that acorns have a second-order capacity to photosynthesize; they cannot photosynthesize at the moment, but if certain events occur and certain conditions are met, then they will, at some point in the future, be able to photosynthesize. However, if acorns are not numerically identical to the trees that develop from them, then when that tree comes into existence, the acorn ceases to exist, and thus the acorn never had either the first or second-order capacity to photosynthesize.

Both the substance view and Liao’s organism view contend that fertilized human eggs possess some quality—rationality or the capacity to regulate life processes respectively—that they do not appear to have (yet). (Of course, Liao contends that the single fertilized cell can regulate life processes in some ways, but far different than how developed human organism does.) For both views, gaining or losing this quality is identity-destroying—the death of a human substance or human organism respectively. For the substance view theorist, the first-order/second-order distinction allows them to explain how identity persists despite the zygote not yet demonstrating the relevant identity-preserving quality—in this case rationality.

The problem for Liao’s organism view is that immediately after conception, the zygote does not possess the first-order capacity to regulate the life processes associated with either a single cell or the adult human person it may become; yet Liao believe two apparently contradictory things:

• i)Each of us (1) possesses the capacity to regulate life processes, and (2) is numerically identical to the single-celled zygote that preceded us that lacked the capacity to regulate life processes.
• ii)Whenever an organism or cell gains or loses the capacity to regulate its life processes, it undergoes an identity change.

Fortunately, Liao’s contention c) suggests that he resolves this apparent contradiction in the same way that the substance view theorist does, by adopting a similar first-order/second-order distinction, such that a human organism:

• a*)Begins to exist when it develops the second-order capacity to regulate life processes
• b*)Continues to exist as long as it possesses the second-order capacity to regulate life processes
• c*)Ceases to exist when it loses the second-order capacity to regulate life processes

For the substance view, some biologically human things may not be rational substances; for example, anencephalic humans may lack even the second-order capacity for rationality, where anencephaly is a medical condition in which a fetus develops missing large portions of its skull and brain. It’s not immediately clear what the organism view would say about such beings; though Liao (2010a) notes that most cases of anencephaly are not genetic but rather caused by environmental factors such as folic acid deficiency. Fetuses whose anencephaly is genetic may never have been human organisms on this view, while those whose anencephaly is caused by environmental factors may cease to exist when their condition becomes irreversible.

Liao argues that PNT and MST are both identity-destroying, as they involve the destruction of two cells and the creation of a third, numerically distinct cell. Liao contends that an egg cell is “essentially a cell,” such that it only persists as long as it possesses the second-order capacity to regulate its cellular processes (Smith 2014). Removing a cell’s nucleus as part of MST disrupts the cell’s first-order capacity to regulate its cellular processes; though Liao suggests that doing so does not disrupt the cell’s second-order capacity, saying “If the nucleus of egg X is not put back into the enucleated egg, there would be permanent cellular discontinuity and egg X would cease to exist” (Liao 2017, 23). However, Liao seems to be on the fence on whether the egg can be reconstructed, saying “Moreover, arguably, neither the cytoplasm of egg X nor the nucleus of egg X is egg X itself” (Liao 2017, 23). His rationale is that neither component alone has the capacity to regulate the cell’s life functions. This would mean that even temporary discontinuity is identity-destroying, as when the components are separated, the egg no longer exists. Thus, MST (and PNT) involve the destruction of a single cell (or organism), and the creation of a new cell (or organism) comprised of components from two dead cells (or organisms).

Though neither component alone possesses the first-order capacity, this doesn’t mean these components lack the second-order capacity. Perhaps the removal of the cell’s nDNA or mtDNA is comparable to a deficiency in folic acid or a lack of oxygen; an environmental factor that may arrest a cell’s (or organism’s) first-order capacity; this is to say that if such disruptions are reversible, the cell (or organism) doesn’t cease to exist.

Yet, no one thing can be in two distinct places at once. The question becomes which part is the cell (or organism) numerically identical to—the cell’s nucleus or its cytoplasm; in other words, is MST a cytoplast replacement technique or a nuclear replacement technique?

Liao contends that human beings come into existence at conception, when an egg cell is fertilized by a sperm cell; where fertilization involves the union of the genetic material from the two gametes in the resulting zygote cell’s nucleus. The inciting incident of new life occurs in the cell’s nucleus, and the primary difference between an unfertilized egg and a fertilized egg is found in the cell’s nucleus. Contra Liao; if fertilization involves a change in numerical identity, and the change occurs in the cell’s nucleus, then it makes sense to say that cellular identity tracks the cell’s nucleus.

Liao argues that multicellular organisms have many nuclei, so a single nucleus is not identical to the organism itself. However, this response is problematic, multicellular organisms also have many cells, so by the same reasoning a single cell is not identical to the organism itself, and thus Liao would be forced to reject the view that the (single celled) zygote is numerically identical to the (multicellular) organism it develops into. Don Marquis defends such a view, though for different reasons, arguing that human organisms come into existence only after cells begin to specialize, about two weeks after conception (Marquis 2007).

For Liao, it seems an organism’s identity tracks the second-order capacity to regulate assorted life processes; but the cellular processes relevant to identity in an unfertilized egg are vastly different than the life processes relevant to identity in a fertilized egg, despite the cell’s extra-nuclear cytoplasm being largely unchanged. Thus, if Liao contends that fertilization creates a new, numerically distinct organism with different second-order capacities to the egg that preceded it, it seems these second-order capacities must rest in the part of the cell that has undergone change—the cell’s nucleus; the cell’s nucleus has the second-order capacity to regulate various life processes and under the right environmental factors—one of which is having healthy mtDNA—the single cell might one day possess the first-order capacity to regulate various life processes. Thus, Liao might say that identity tracks the second-order capacity to regulate life processes, which arises in the cell’s nucleus at conception, but after cell division can be said to reside in multiple cell nuclei.

In contrast, if identity tracks the cytoplast, and the cytoplast is largely unchanged following conception, then conception is not identity-affecting; human beings would be numerically identical to the egg cell they fertilized from.5 Both Liao’s organism view and the substance view contend that identity begins at conception, so they seem to be committed to the position that identity cannot track the cytoplast.

Considering this analysis, it seems both substance and Liao’s organism view theorists have good reason to think that neither PNT nor MST are identity-affecting, such that PNT can reasonably be said to be therapeutic for the zygote. In contrast, Don Marquis’s organism view and any view that believes we come into existence sometime after conception would group PNT and MST together as preventative, not unlike taking folic acid to prevent birth defects (Marquis 2007). In either case, MRTs aimed at preventing mitochondrial diseases seem to fall under the therapeutic/preventative scope of medicine.

Non-identity Problem

Consider the following case:

Arial: Arial wishes to exercise her reproductive freedom in new and interesting ways. She uses MST on one of her cryopreserved eggs, discarding a cytoplast with healthy mtDNA and placing the nucleus in a donor egg with mutated mtDNA she expects will lead to developmental problems in whatever child would be produced.

Most of us would agree that Arial’s action here is morally problematic, in part, because of how it affects whatever child develops after fertilization. However, Cavaliere and Palacios-González (2018, 839) argue this line of reasoning is a mistake, contending “neither PNT nor MST leave created children worse off than they would otherwise have been. Such children are not made worse off by MRTs because the only other available ‘option’ for them is not to exist,” noting “this is a classic instance of the Non-identity Problem.” However, the non-identity problem is just that—a problem, as it is inconsistent with our commonsense moral intuitions about such cases. In this section, I will defend two responses to the non-identity problem.

The first response is to avoid the problem entirely; although Cavaliere and Palacios-González believe both PNT and MST are identity-affecting, in the previous section I’ve argued we have good reason to think that neither are identity-affecting; both substance and organism view theorists have good reason to think that the zygote possesses the same second-order capacities after a successful PNT as it did before, such that if a zygote is numerically identical to the person it will become, performing PNT to prevent mitochondrial diseases would be therapeutic, and performing PNT to give the person mitochondrial diseases would harm them.

The relevant difference between PNT and MST for these views is that MST occurs prior to conception, such that there is no existing human being to either help or hurt. However, because PNT need not be identity-affecting, and MST involves the same kind of intervention, then there is no reason to think that MST relevantly affects identity.

To illustrate this, suppose future philosophers use a time-travel technology to tag one of the planks of wood that would be used to construct the Ship of Theseus prior to its construction, so that they might track that plank of wood through time. If such a change does not affect the identity of the plank of wood, it doesn’t make sense to say that such a change would affect the identity of the ship constructed from it, though this would result in a change in properties to the ship, which is now tagged and trackable… at least until that plank of wood is discarded.

On this view, Arial’s action is wrong because it risks avoidable harm to a person who doesn’t yet exist, but that might. In other words, Arial’s action is wrong because it risks identity-independent harm to an unspecified member of the set of possible persons that Arial might bring into existence.

However, suppose one believes MRTs are identity-affecting; a second response involves tackling the non-identity problem in greater detail. The best way to illustrate the non-identity problem is through parallel cases; David Boonin (2008) does as much when he asks us to consider the following cases:

Betty: Betty takes her infant, Bam-Bam, to his doctor for a checkup. The physician explains that Bam-Bam suffers from a condition that, if untreated, will make him blind. Fortunately, if Betty gives him a pill, Bam-Bam will not go blind. Betty decides giving him the pill is too inconvenient, and Bam-Bam goes blind.

Wilma: Wilma is planning to become pregnant and goes to her doctor for a checkup. The physician explains that Wilma has a condition that will cause whatever child she conceives to go blind. Fortunately, if Wilma takes a pill, it will rectify her condition, but in doing so she will expel the current egg in her fallopian tube and cause her to release a different egg and delay conception. If Wilma takes the pill, she will have a numerically distinct child that will not go blind. Wilma decides taking the pill is too inconvenient, and conceives a child—Pebbles—who goes blind. (Had Wilma taken the pill, however, she would have conceived a different child, Rocks, who would not go blind.)

Boonin (2008) contends that most people believe that Betty and Wilma act morally comparably immorally. However, the non-identity problem arises when we realize that Betty’s inaction allows harm to befall Bam-Bam, but Wilma’s inaction doesn’t allow harm to befall anyone. Betty is blameworthy because she had the opportunity to prevent harm to her child but chose not to. However, Wilma never had that option; her choice is between having Pebbles who will go blind, or having Rocks who won’t; as such her action doesn’t harm anyone; to paraphrase Cavaliere and Palacios-González, Pebbles is not made worse off by Wilma not taking the pill, as the only other option available to Wilma is that Pebbles doesn’t exist.

Boonin concludes that despite our intuitions, Wilma is differently morally responsible than Betty. His solution is subtle, but interesting; he contends that our intuitions in non-identity cases come from an error in generalization. Most of us accept something like John Stewart Mill’s (2015) “harm principle,” such that it is prima facie morally wrong to act to harm other people. Betty and Wilma can be said to perform the same kind of action—they both act in such a way that causes a person to have a significant handicap. In most cases, when one acts in such a way, one harms an existing person. Betty’s case, Boonin (2008) notes, is just such a case, and thus Betty violates the harm principle. However, he notes that Wilma’s case is not like this; “Because of facts that are idiosyncratic to that case, the act of causing a person to have a significant handicap in the case of Wilma does not, in fact, cause that person (or any person) to be worse off than they would otherwise have been” (Boonin 2008, 145). Thus, Boonin concludes that Wilma is differently morally responsible than Betty and that our intuitions about Wilma’s case are understandably mistaken.

If Boonin is correct, Arial’s actions are not morally wrong because they do not harm an existing person. However, Arial may still be morally wrong for other reasons. James Rachels (2003) argues we ought to be guided by reason to do what there are the best reasons to do. Even if Arial and Wilma are not harming anyone, there are other reasons that may make their actions deeply immoral; for example, we might argue that both are bad parents, both act to further strain scarce medical resources, or the like. Thus, even if our intuitions in non-identity cases are unreliable, the actions of Wilma and Arial may still be, all things considered, morally wrong, though on this view they’re wrong in fewer ways than Betty is.

Note, however, that Boonin’s cases are parallel cases, cases meant to be identical in every morally relevant way except the feature of the case under moral investigation—here, Boonin (2008) investigates the moral significance of identity-dependent prevention compared to identity-independent prevention. James Rachels (1975, 2001) argues that in parallel cases, if the feature under investigation is morally relevant, our intuitions about the cases ought to be different, but if our intuitions are comparable, then this is reason to think the bare difference under consideration is not morally relevant. For Boonin (2008), our intuitions in these cases are the result of a mistake. The moral principle in question is that of the harm principle. Because Betty and Wilma’s actions are both of a kind that normally violate the harm principle, we mistakenly assume that both do violate the harm principle. But this assumption is false.

Philosophers tend to judge moral principles, like the harm principle, on a variety of criteria—ontological simplicity, coherence with one’s other beliefs, and (perhaps most accessibly) consistency with one’s commonsense moral intuitions. Rather than our intuitions being the result of a generalization error, it’s possible that this formulation of the harm principle in question is an abstraction error.

Suppose that rather than define harm as identity-dependent, we understood harm as identity-independent. An identity-dependent account of harm holds that a harms b if and only if a makes b worse off than b otherwise would have been, where b is a specific being.

In contrast, an identity-independent account of harm would hold that a harms p if and only if a makes p worse off than p otherwise would have been, where p is an indeterminate placeholder for one (or more) members of a set s, comprised of all individuals, existing or possible, that could be affected by a’s action or inaction.

Consider the preventative intervention of giving folic acid to a woman to prevent the identity-independent possible harm of birth defects to any child she conceives in the future. Palacios-González might reasonably be interpreted as believing that the mere offer of folic acid could be identity-effecting. But let’s explore Wilma’s options once again, with this additional variable:

Wilma Redux: Wilma is planning to become pregnant and goes to her doctor for a checkup. Her physician has two concerns—a risk of blindness and a risk of birth defect. Fortunately, if Wilma takes a pill, whatever child she conceives will not be blind. Furthermore, if Wilma fakes folic acid vitamins, whatever child she conceives will not have birth defects.

If Wilma takes both the pill and vitamin, she conceives Rocks, a sighted child with no birth defects. If she takes the pill, but no vitamin, she conceives Shale, a sighted child with birth defects. If she takes the vitamin but turns down the pill, she conceives Pebbles, a blind child with no birth defects. And if she refrains from taking either the pill or the vitamin, she conceives Quartz, a blind child with birth defects.

We can summarize Wilma’s options as follows:

Wilma’s options	Pill	~ Pill
Vitamin	Rocks	Pebbles (blind)
~ Vitamin	Shale (birth defect)	Quartz (blind, birth defect)

According to an identity-dependent account of harm, assuming the offer of folic acid is identity-affecting, folic acid doesn’t prevent harm to anyone anymore than taking the pill to prevent blindness does.

However, on an identity-independent account of harm, we can compare the set of all possible children Wilma might conceive—in this case (by assumption), this is a set of 4 {Pebbles, Rocks, Quartz, and Shale}. On this account, in Wilma Redux, if Wilma were to refrain from taking the pill and/or vitamin she would be risking unnecessary harm to whatever child she ends up conceiving.

This notion of harm is comparable to that in Mill’s harm principle, but with one major exception; philosophers like Boonin interpret the harm normally associated with Mill’s harm principle as a negative change in status to a specific existing person, while an identity-independent account of harm understands harm as a negative, avoidable outcome—an outcome that may affect a person, but not necessarily a previously existing person, nor making a person worse off than that specific person would otherwise be. This account of harm, I think, is more consistent with consequentialist theories like utilitarianism than Boonin’s account of harm, but one need not be a utilitarian to find this conception of harm more intuitive.

This identity-independent account of harm is prima facie consistent with our intuition that Betty and Wilma are morally comparable; both choose to allow harm that could easily be prevented without significant sacrifice. Boonin considers and rejects a similar proposal by Gregory Kavka (1981); according to Kavka’s position, Wilma and Betty might be said to act comparably immorally because both act to bring equally undesirable conditions into the world—in particular, the existence of a blind child. Boonin concludes that if Kavka is correct, “We would have to say that every time a blind person is born, for example, the world becomes a worse place. We would have to say that it would be better, from the moral point of view, that such people never exist. And this is clearly unacceptable” (Boonin 2008, 145). This conclusion is certainly problematic, but it seems like an uncharitable reading of Kavka. An identity-independent account of harm easily avoids this conclusion. In some cases, it can be wrong to risk conceiving a blind child when one could with little effort, reduce this risk; but in other cases, it is not wrong to risk conceiving a blind child, because a blind child is not worse off than other members of the set of possible children and blindness will not prevent the child from living what Purdy calls a minimally satisfying life.

Suppose a couple—the Smiths—wish to procreate as an expression of their love but come to learn that any children they conceive will likely be blind. They may reasonably choose to refrain from procreating, but as being blind doesn’t preclude one from living a minimally satisfying life, philosophers like Purdy wouldn’t conclude they have an obligation to refrain from procreating. This couple faces a choice—(1) refrain from procreating or (2) bring into existence a child that will likely go blind. Given that many people take an interest in having genetically related children, there is good reason for such a couple to choose option (2). While avoiding this harm would be morally acceptable, it is not morally obligatory, as there is no member of their set of possible children that a likely blind child is worse off than.

Contrast this with Wilma in Boonin’s original case, who has three options—(1) refrain from procreating, (2) bring into existence a blind child, or (3) bring into existence a sighted child. If we assume Wilma has taken an interest in having a genetically related child, options (2) and (3) are preferable to (1). The difference between these latter two options is that the child conceived in (2) is worse off than the child conceived in (3), and both belong to the same set of possible children Wilma might conceive. We’re told Wilma concludes that taking the pill would be “too inconvenient,” but surely this conclusion is ridiculous.6

Interpretations of the harm principle that focus on harming a particular individual lead to the non-identity problem as many prima facie immoral actions would end up harming someone who might not exist or doesn’t yet exist. However, an identity-independent account of harm avoids this pitfall by concluding that causing or allowing harm is wrong if and only if that harm can be avoided without unreasonable sacrifice. If this analysis is on the right track, then the question of whether MRTs are identity-affecting is ultimately irrelevant to this discussion because it wouldn’t matter who is harmed, only that the person who comes to exist is less well off than they, or another, would otherwise have been. This would allow us to avoid the controversial questions of when beings like us come into existence and come to have moral significance.

But what of Arial, who wishes to exercise her reproductive freedoms in new and interesting ways? Well, like Wilma, Arial has three choices—(1) refrain from conceiving, (2) conceive an unhealthy child (using MRTs destructively), or (3) conceive a healthy child. I see no reason why (3) cannot be done in new and interesting ways; perhaps she could conceive while SCUBA diving, floating weightless in space, on the top of a newly formed mountain, or as the denouement of a reality TV show of dubious interest. In short, while it’s not clear Arial’s interest in exercising her reproductive freedom in new and interesting ways is all that morally significant, it seems entirely plausible she can do so without risking unnecessary harm.

Genetic Relatedness

Many individuals value having genetically related children, and many couples value having children that are genetically related to each of them. Many people exercise their reproductive freedom in the pursuit of these goods. For example, a healthy heterosexual couple may choose to have children through normal procreative means, and the act itself can have moral significance as an exercise of their love.

However, not all couples can have genetically related children in this way. Many heterosexual couples choose to use expensive and potentially risky ARTs in the pursuit of genetically related children, rather than adopting children or having children that are genetically related to only one of the parents and a donor. Such risky behavior is generally accepted to fall under the scope of an individual’s reproductive freedom for much the same reason that the Smiths are allowed to pursue having a child despite risking conceiving a child that will go blind.

Cavaliere and Palacios-González (2018) contend that the same morally significant interest in having genetically related children should extend to lesbian couples who wish to use MRTs to have a child with nDNA (partially) from one mother, and mtDNA from the other. By the same token, a polygamous triad consisting of two females and a male could use contemporary technology to create a child that is genetically related to all three parents.

Even if we assume that the interest in having genetically related children is morally significant, there are problems with the proposal to use MRTs for this purpose. First, it’s not entirely clear that individuals genuinely value the sort of genetic relatedness that comes with mtDNA. Even if they do, the mtDNA is dwarfed in kind and function by the nDNA, leading to a situation where the mtDNA mother might feel less genetically related to the child than the nDNA mother. A human cell has more nDNA than mtDNA; as such, a child has more DNA in common with a genetically related cousin, uncle, aunt, or the like, than their mtDNA donating genetic parent. Furthermore, normally a child becomes genetically related to their parents at conception; however, in PNT, genetic relatedness is conferred after the fact.

Let us assume that the genetic relatedness of MRTs satisfy an individual’s morally significant interest in having genetically related children. Consider the following cases:

Intervention 1: Adam and Aron fall in love and adopt a child, Bruce. They want to be genetically related to their child, so they request their family physician take blood from Adam and Aron and give it to Bruce.

Intervention 2: Alice and Amy fall in love and adopt a child, Baily. They want to be genetically related to their child, so they request that their family physician remove Baily’s healthy kidneys, and implant replacements, one from Alice and one from Amy, to create a genetic relatedness with Baily.

In each of these cases, the parents ask their physicians to perform medical procedures to imbue their adopted children with their genetic material in the pursuit of genetic relatedness. The first case seems frivolous, but donating blood seems to be relatively safe, so there is little risk involved. Despite the lack of risk, and (by assumption) moral significance of genetic relatedness, this kind of medical intervention is neither therapeutic nor preventative, and thus seems to be prima facie outside the scope of medicine. Generally, parents are said to be able to exercise their parental authority to make medical decisions for their children, but such parental authority is not parental license; their decisions must be aimed at promoting the medical welfare of their child. Even if genetic relatedness can be said to promote the child’s welfare in some sense (perhaps symbolically strengthening the parent–child bond), it does not do so in a medically relevant sense.

The primary difference between the first and second case, however, is the risk involved. Kidney donation comes with risks for donors and recipients, and it is unthinkable that a parent would put their child in such medical danger frivolously, and equally unthinkable that a physician would perform such an unnecessary surgery.

These cases serve to illustrate two important objections the view that one’s right to reproductive freedom allows for potentially risky MRTs. In Intervention 2, it seems clear that Alice and Amy’s choice to pursue genetic relatedness would put their child, Baily, in unnecessary and morally objectionable risk. Parental rights do not allow one to frivolously risk harm to their child, so it would be quite odd if procreative rights allowed such risk. In contrast, Intervention 1 seems to illustrate a more foundational problem with MRTs intended to pursue genetic relatedness; such interventions seem to be outside the scope of medicine as they’re neither aimed at treatment or prevention. Furthermore, such interventions would not medically enhance the subject; just as Bruce is not medically better off for having his parents’ blood literally flowing in his veins, the child of a lesbian couple or polygamous triad resulting from an MRT is not medically better off than a child without an MRT.

In light of this, even if the medical risks associated with MRTs could be addressed, MRTs aimed at genetic relatedness would be medically irrelevant, at best comparable to unnecessary cosmetic surgery. Such an intervention seems to be outside the scope the medical profession.

Conclusion

Current MRTs are risky medical procedures. When therapeutic or preventative, the medical benefits may outweigh the risks. Cavaliere and Palacios-González (2018) argue that using MRTs are not harmful, as they are identity-affecting, and thus don’t harm any existing person. Above, I’ve argued that we have good reason to think MRTs are not identity-affecting. However, even if they are identity-affecting, I’ve argued that this is no excuse to create unnecessary identity-independent harm.

Cavaliere and Palacios-González’s (2018) proposal to use MRTs to help people exert more control over the reproductive process is, at the very least, interesting. However, even if we set aside hard to answer questions about therapy and identity, I’ve argued their proposal to use MRTs to pursue genetic relatedness is outside the scope of medicine. Even if parents value genetic relatedness, cases like Interventions 1 and 2 illustrate parental authority does not give parents the freedom to do whatever they want to their children. Furthermore, these cases serve to illustrate that certain interventions serve no medical purpose, and are thus outside the scope of medicine.