“IF WE DIDN’T EAT THEM, THEY WOULDN’T EXIST” The Nonidentity Problem’s Implications for Animals (Including Humans)

If you are a vegan, vegetarian, or even pescatarian, you have likely encountered a seemingly paradoxical objection to your ethics. If a farmed animal—a cow, for instance—has a life worth living, and the animal would never have existed absent the demand for her meat or bodily output, then doesn’t it follow that humans deny that animal a benefit by refusing to eat her or her products? A person who uses animal products accordingly helps rather than harms animals. Absent the omnivore’s demand for these products, the animals in question would not exist at all.¹ Refraining from consuming animals thus deprives those animals of lives worth living. Moreover, the benefits of life that omnivores confer accrue not just to individual animals but to entire species because animals such as domestic pigs, chickens, and cows would not exist were it not for humans breeding, raising, and eating them. Thus, the argument concludes, what may appear to be harm to animals is actually a kind of gift, a gift that vegans, vegetarians, and the like misguidedly refuse to convey to the animals they purport to care about.

One way to challenge what we shall call the animals-benefit objection is to contest its factual premise that farmed animals’ lives are worth living. Although philosophers do not always specify exactly what makes a life worth living,² the concept appears to connote at least some sort of minimally satisfying life. We will not attempt to define the term, because by any reasonable definition, most farmed and otherwise exploited animals do not have lives worth living. The overwhelming majority of egg-laying hens live in confined spaces where they can barely flap their wings and often suffer prolapses from having been bred to lay more eggs than their bodies can handle. Such animals endure sustained misery.³ So do the dairy cows whose babies the farmer repeatedly takes from them at birth so that the milk produced for baby bovine nursing can instead go to human consumers. After experiencing this torment repeatedly, the grieving mother cow must join her children at slaughter when she no longer produces as much milk as she once did.⁴ Similar misfortunes beset the lives of other animals that people raise for food, fiber, and other products.

The foregoing response covers the vast majority of farmed animals. Yet we want to go further. We hope to persuade readers to reject the animals-benefit argument even in the small number of cases in which the farmed animal in question has a life worth living (however defined). We think that consuming animal products even from animals with worthwhile lives is wrong. Accordingly, in this article, we will assume, albeit solely for the sake of argument, that exploited animals can have lives worth living.

Imagine that a farmer raises a litter of piglets and treats them as though they were his pets. Then, when they reach “slaughter weight,” the farmer kills them as humanely as possible. (His would hardly be a profitable enterprise, but we are assuming away such practical concerns.) We would consider the farmer’s pets-to-meat program wrong, even if the pigs had led happy lives up until the day of slaughter and thus even if their lives were, on net, worth living.⁵ Put differently, assume, against the overwhelming weight of the evidence, that defenders of “humane” animal exploitation are right in claiming that animals who become meat enjoy happy lives followed by “one bad day.”⁶ Still, we will argue that the many good days do not justify what happens on the bad day. We make the following core moral claim: if it is wrong to kill an animal for food at the time that one kills the animal (as it is in nearly every circumstance⁷), then it is also wrong to kill that same animal even after months of showing kindness to the animal.⁸

Indeed, we can identify at least two ways in which giving the animal a good life until the “one bad day” might make the slaughter worse rather than better. First, we could regard death as a mercy for a miserable animal who endures nothing but cruelty and torment. In contrast, slaughter deprives a well-cared-for animal of a satisfying life. Second, a farmer who treats an animal with kindness and thus trains her to trust him betrays that animal when he kills her. We think most readers will share this intuition if they imagine a puppy a family raised as a pet. If the family were to slaughter the puppy one morning, they would be engaging in an egregious betrayal.⁹

Whether the element of betrayal makes the killing worse is beside the point, however. Whether killing after cruelty or killing after kindness is worse, neither is morally justified. However kind we might be to an animal (or a human, for that matter), there never comes a point when that kindness entitles us to take the animal’s (or human’s) life for our own gain.

We imagine that at least some of our readers share the intuition that taking an animal’s life is wrong no matter how kind the killer was beforehand. For those who still aren’t sure whether they agree, we offer an example involving humans.

Imagine that Eve has a disease that will require her to have a heart transplant before she turns forty. Her genetic code makes her very unlikely to find a matching organ donor. Accordingly, using reproductive technology, when she is twenty-five Eve conceives, births, and begins to nurture Deborah, who grows into a happy child. Unbeknownst to Deborah, Eve plans to kill her and “harvest” her heart when she is twelve and Eve is thirty-seven. Even though Deborah will have twelve years of a life worth living, it would clearly be monstrous for Eve to carry out her plan. We would judge Eve (and any doctors or others who knowingly assisted her) culpable for doing so. As a general matter, people cannot bank credits for their good deeds (such as being kind to a child) to offset bad ones (such as killing the child). Good deeds do not cancel out bad ones, even if the actor would never have carried out the good deeds absent her plan to carry out the bad.

We can think of no reason to restrict the prior-good-doesn’t-license-later-bad logic to human-on-human morality. If it is ordinarily wrong, as we believe it is, to kill or otherwise harm animals for food, then it is wrong to do so even when those animals existed only because of the plan to use them as food. Case closed, right?

Maybe not. Our argument to this point has demonstrated that if it is wrong to inflict suffering or death on a sentient creature, the farmer acts wrongfully by doing so even if the farmer is also responsible for the creature’s existence and gave the creature a life worth living because of the farmer’s plan to slaughter the creature for sale. Creating a being with a life worth living does not license the farmer to harm or kill that being, even if the harm or killing leaves the being’s life overall in the worth-living category. This statement is true at least in part because we evaluate the rightness or wrongness of each separate act, given that the first does not make the second inevitable: the farmer can decide, after breeding and feeding the animal, not to kill her. From the farmer’s perspective, breeding, on one hand, and killing or harmfully exploiting, on the other hand, are distinct acts, however much he links them in his mind.

From the consumer’s perspective, by contrast, there appears to be only one single act: purchasing the animal products. The consumer of animal products does not send two market signals to the farmer: (1) bring an animal into existence and give that animal a life worth living, and (2) then take the further actions of harmfully exploiting and slaughtering the animal. Buying meat, dairy, or eggs sends a single indivisible market signal to farmers both to breed more animals and to inflict harm on those animals to create the products in question.

If the argument that animals benefit from being farmed fails as a defense for the farmer, might it succeed as a defense for the consumer? We reiterate that in the vast majority of cases, the animals whose flesh and secretions become food for the consumer do not have lives worth living, so the animals-benefit argument does not even get off the ground. But for the small number of animals whose lives we assume are worth living notwithstanding their terrible end, can the consumer justify her consumption habits by drawing on the indivisibility, from her perspective, of the worthwhile life and the sad and premature end?

We can formulate that question as well in a way that makes it relevant to human-on-human morality. Consider a stylized example from the pre–Civil War United States. Suppose that plantation owner Beauregard enslaves a person named Jim. Suppose further that Jim exists only because Beauregard purchased and thereby brought together Jim’s parents. Even if we assume that Jim’s life is worth living, we reject out of hand the claim that Beauregard could ethically continue to enslave Jim. The claim never gets off the ground because the creation of Jim (by bringing his parents together) is severable from the continued enslavement of Jim.

From the perspective of an ethical consumer, however, the moral issue may be less clear. A consumer who purchases clothing made of cotton picked by enslaved persons has performed a single indivisible act. This one act induces both the creation of people (with lives worth living, we are assuming) and the enslavement of those people once they exist. If the enslaved people indeed have lives worth living, then how do we evaluate the morality of a decision to purchase the products of slave labor?

We might try to distinguish the animal case from the human slavery case by observing that consumers of animal products and consumers of slave-labor products, respectively, act as they do with a different mental state. Animal products necessarily come from animal exploitation, whereas generic commodities (such as cotton) that sometimes come from slave labor do not necessarily involve human exploitation. Consumers of animal products therefore knowingly demand injustice to animals, while consumers of cotton picked by enslaved people might just be demanding cotton and holding an attitude of reckless indifference toward sourcing.

Along the dimension of mental state, then, purchasing animal products is arguably more culpable than purchasing generic goods that happen to be the product of slave labor! This observation does not, of course, amount to approval or moral endorsement of the decision to purchase goods resulting from slave labor. Furthermore, despite the possible difference in culpability arising out of the respective mental states, the act of purchasing animal products is sufficiently similar to the act of purchasing the products of slave labor to allow for a useful analogy between the two.

We have a strong moral intuition that purchasing the products of slavery and purchasing animal products are both very wrong. They are wrong, we think, even though, by hypothesis, the enslaved person and the nonhuman animal have lives worth living and would not have existed but for the decision to purchase the products, respectively, of slavery and animal exploitation. Readers may share our intuitions.

We face a challenge, however: defending against the pull of competing moral intuitions. Specifically, when it comes to the moral choices of consumers, how do we defeat the argument that enslaved humans and nonhuman animals benefit from consumption choices that both lead to the existence of the beings in question and demand the infliction of cruelty upon them? We refer to the argument we aim to defeat as the “animals-benefit” claim because we focus in this article on nonhuman animals, but we hope that what we have to say about nonhuman animals sheds light on related moral puzzles involving only humans.

Defeating the animals-benefit claim implicates a family of puzzles that moral philosophers have long struggled to solve. These puzzles go under the name of the nonidentity problem. Here is the problem, in general terms:

• (1) 

  An act or choice is wrong only insofar as it causes harm to some being (what we shall call “no-harm-no-foul”).

• (2) 

  When a single act or choice causes both the existence and the misfortunes of a being, the act can be harmful only if the being has a life that is not worth living. In other words, there can be no harm to a being to whom a single act gave both existence and misfortunes if the being’s life is, on balance, worth living (“net-benefit”).

• (3) 

  Yet some acts that produce net-benefit seem nonetheless to wrong the being created, thereby contradicting no-harm-no-foul (“wrong-anyway”).¹⁰

To answer the animals-benefit argument, we must either demonstrate why the intuition that underwrites wrong-anyway is right, notwithstanding no-harm-no-foul and net-benefit, or show how the nonidentity reasoning goes wrong in some other way. If we succeed, we will have replied to an objection to animal rights and we will also have shed light on an important puzzle in moral philosophy.

Moral philosophical questions may seem abstract, to be sure. Nonetheless, a successful solution to the nonidentity problem would not only answer the animals-benefit claim for vegans and vegetarians. It could potentially address important questions of law and public policy, including whether descendants of enslaved humans are entitled to reparations and whether we have an obligation to preserve the natural environment for our descendants.¹¹

Accordingly, this article uses the animals-benefit claim to offer new insights into the nonidentity problem. Part I specifies the problem in greater detail with some additional stylized examples. Part II conducts a critical review of the leading proposed solutions to the problem. Finding none fully satisfactory, we then articulate alternative approaches. Part III proposes our most substantial innovation; building on but also modifying arguments offered by Elizabeth Harman, Seana Shiffrin, and others, we contend that moral agents are responsible for the harm that results from, but cannot take credit for the benefits conferred by, actions that lead to the creation of a sentient being. Part IV argues, alternatively, that consumers of animal products cannot plausibly describe their actions as only incidentally harming animals; thus, their intentions and actions might be said to take the animal case entirely outside the scope of the nonidentity problem. In other words, part IV contests the claim we made above that the moral choice facing consumers differs in a relevant way from the moral choice facing farmers. Part V builds on the work of Ori Herstein, who conceptualizes historical injustices such as slavery and the Holocaust as involving group harm; although we note that Herstein’s approach does not respond to all variations of the nonidentity problem, it is particularly well suited to addressing the animals-benefit claim.

The nonidentity problem takes its name from the fact that the person who complains of being harmed by someone else’s allegedly immoral choice would not exist in the absence of that choice. In other words, if the complainant had her way, then either somebody else would exist in her place (so that the complained-of act would affect the identity of who came into existence) or nobody would have come into existence. In either event, she would not exist.¹² She is accordingly comparing her state of well-being now (a state that resulted from the choice she condemns) with her well-being in a state of nonexistence if the object of her condemnation had only done precisely what she claims he ought to have done. Consider some stylized examples that we borrow with minor changes from philosophers’ substantial body of work in this area.¹³

Suppose Jane wants to have a child. She suffers from a condition that all but guarantees that her child will be born blind (and otherwise healthy). Scientists have invented a medication that will fully protect any offspring she has against developing blindness. Jane need only take a pill with no side effects one month before conceiving to eliminate the resulting child’s risk of blindness. Under these circumstances, would Jane be doing anything wrong if she conceives a blind child immediately rather than taking the pill and waiting a month to conceive a sighted child?

Some people would say that by failing to take the pill and wait the month, Jane has acted wrongfully. But whom did she wrong? Blindness entails special challenges, in large part because society has organized itself around the assumption that people can see.¹⁴ Nonetheless, Jane has plentiful resources and will love her child. The child, though blind, will enjoy a life worth living. And Jane’s child—the blind one—would not have existed at all if Jane had waited a month to conceive, because a different egg and sperm would have combined to produce a different baby. For the blind child, then, there are two choices only: either to live a life worth living as a blind person or to live no life at all. We have no third option of Jane creating that same baby and endowing the child with the ability to see. Are we prepared to say that creating a blind person with a life worth living, a life that is valuable to the blind person, constitutes a harm to the child? The answer would appear to be no.

Jane’s actions thus satisfy no-harm-no-foul (because they inflict no harm on anyone) and net-benefit (because Jane’s blind child enjoys a life worth living). We might therefore conclude that we cannot criticize Jane for conceiving her child without waiting a month. Yet Jane’s actions still trigger an intuition in many of us that she did something wrong (by failing to take the pill and wait a month to conceive). Her actions thus satisfy the criterion for wrong-anyway because they seem wrong notwithstanding the apparent failure to meet the harmfulness criterion. To solve the nonidentity problem, we must reject one of the three propositions because they are logically inconsistent with one another. Yet each proposition appears sound.

Perhaps the reader is thinking that the scenario we describe fails to satisfy the “no one suffered a setback in his interests” premise of the nonidentity problem. Have we not identified someone who suffers harm because of Jane’s decision? Wouldn’t that be the (sighted) baby that Jane did not conceive? We might regard the sighted child who never comes into the world as having experienced a harm as a result of Jane’s decision to conceive immediately rather than wait a month. Yet the nonexistence of that sighted child might have nothing to do with Jane’s failure to wait a month. Had she waited, any one of millions of possible sighted children might have come into the world, a product of the vast number of sperm cells that might have ultimately fertilized Jane’s egg that month. We accordingly cannot identify one specific child who would be here if only Jane had waited. All but one of those alternative babies would still have never come into existence even if Jane had taken the pill and waited. How, therefore, can we say that having her baby when she did harmed a potential person who almost surely would never have existed anyway?

One could respond that even though almost none of the potential babies would have become actual babies, one of them would have. Perhaps it is that baby, one whose identity we do not and cannot truly know, who suffered the harm of Jane’s refusing to wait a month to conceive.

That response might be convincing if potential-but-never-existing babies were comparable to actual people who might become the victims of a future calamity. Suppose the residents of a town decide in October to hold a lottery in November to identify one resident whom they will stone to death as a sacrifice to the gods. In October, the identity of the unfortunate lottery “winner” is unknown, but we know that someone will be the victim of the awful plan.¹⁵ Thus, we could coherently say that the person who will end up being stoned to death was severely disadvantaged by the October decision to hold the lottery, even though no one knows who that person will turn out to be until November. In November, it would be perfectly understandable for the unlucky “winner” to lament that the town made its fateful decision the prior month. That’s for the obvious reason that the “winner” experiences the harm of stoning in November. By contrast, potential people who never come into existence are not harmed by their not coming into existence because they never experience anything.

Are we sure? Might a reader have the intuition that it is possible to harm someone who does not exist and, as a consequence of the act in question, will never exist?

If one truly believed that the failure to come into existence is a harm, a number of absurd results would follow. Every decision to not reproduce as many times as possible would inflict a harm on all the people who would otherwise have been born with lives worth living. If one thought that Jane’s decision not to wait a month to have a sighted child harmed the nonexistent sighted baby that she could otherwise have had, then one would also be committing oneself to saying that the decision to have two rather than three or more children harms each of the potential children who would have otherwise materialized. Because this idea seems indefensible except perhaps on religious grounds,¹⁶ virtually none of the philosophers who write about the nonidentity problem thinks that we harm potential persons by failing to bring them into existence.¹⁷

So, is it wrong for Jane to decide to conceive now rather than take the pill and wait? Before trying to answer that question, consider another example.

Suppose that a newly discovered fuel source—“contaminatium”—offers some benefits in the present but in a hundred and fifty years will impose catastrophic environmental harm. Using the contaminatium will accordingly lead to an unhealthy and poor population in the future, but because it will radically alter the ensuing course of events through so-called butterfly effects (or what Gregory Kavka called the “precariousness” of existence),¹⁸ it will also bring into existence the very people (and animals) who will experience the degraded Earth. Avoiding use of the contaminatium now would do nothing beneficial for those particular future people (and animals), because they will not be born if we refrain from using the contaminatium now. Thus, so long as those future beings’ lives are even just barely worth living, it appears that we inflict no harm upon those future people and animals when we use a fuel that does catastrophic but delayed damage to the environment in which those future beings will have to live. That conclusion is highly counterintuitive. Can we avoid it?

Just as we can defeat most of the animals-benefit argument by pointing out a fact in the real world—that most farmed animals do not have lives worth living—so too can we invoke facts about the real world to defeat an argument in defense of environmental degradation. Almost every action causing serious environmental damage in the distant future will also harm and indeed is already harming people (and a large number of animals) whose identities are fixed because they were already alive when the harm from pollution commenced.¹⁹ Therefore, the premise of the hypothetical scenario—that the creation of environmental complainants and the creation of environmental harm are inextricably bound to each other—is usually going to be wrong.

We like that kind of response because it is grounded in facts about the world. But it works best if the question is whether humans should care about the environment. The answer is that yes, we should care, regardless of whether or how we solve the nonidentity problem. A sufficient reason to care is that there are living beings who will suffer from a damaged environment and who do not implicate the nonidentity problem because they already exist. An additional reason to care is that there may be particular sorts of environmental problems that realistically look like the hypothetical contaminatium example. A generation ago, one could have conceived of global warming as having had this character. At the time, reasonable people could believe that the world’s extant population would not feel global warming’s effects because those effects would not materialize for over a century. Although that view is no longer reasonable with respect to global warming, there may be other kinds of very-long-latency environmental harms that operate in the way that contaminatium does.

Even if our environmental example is purely hypothetical, we take it seriously for the same reason that we take seriously the highly unrealistic premise that farmed animals have lives worth living. Taking seriously stylized but far-fetched scenarios that cleanly present the nonidentity problem helps us clarify our views about more realistic scenarios in other contexts. Accordingly, we use Jane’s story, the case of distant-future-only environmental harm, the animals-benefit claim, and a variety of other real and imaginary scenarios to test others’ and our own intuitions about the nonidentity problem.

Derek Parfit was not the first philosopher to discuss the nonidentity problem,²⁰ but he addressed it in more detail than just about anyone before or since.²¹ Parfit presented potential solutions to parts of the problem and then explained why some of these solutions are more plausible than others.²² Yet the overall thrust of his argument accentuates rather than resolves the paradox. He demonstrated that plausible solutions to the problem in one context create trouble for versions of the problem in another context. For example, the choice in question might lead to one or another of two potential people, each happy to be alive, coming into existence (as in Jane’s case) so that one commits no wrong against either party, regardless of the decision and notwithstanding one child’s disability. But one would not want to say the same of choosing environmental devastation, even though the choice to pollute the environment will also bring into being an entirely different group of people (all of whom, we would hypothesize, are happy to be alive). Given its complexity, we will not attempt to summarize Parfit’s view, which we shall instead treat as a very powerful restatement of the problem. One comes away from Parfit’s treatment of the subject thinking that the nonidentity problem may well be an unsolvable paradox.

David Boonin agrees with Parfit that there is no solution to the nonidentity problem as conventionally understood.²³ He went on to argue, however, that this means we should abandon the wrong-anyway principle.²⁴ According to Boonin, no-harm-no-foul and net-benefit are sound general claims about moral philosophy, whereas wrong-anyway merely describes a moral intuition that people may have.²⁵ It explains our feeling that Jane acts wrongfully by conceiving a blind child or our feeling that it would be wrong to degrade the natural environment even if doing so harms only those beings who owe their existence to the very act that causes them harm.²⁶

Moral philosophy aims, among other things, to reconcile our moral intuitions.²⁷ But it cannot always make sense of our every pre-reflective moral hunch. In a state of what John Rawls called “reflective equilibrium,”²⁸ we take those moral intuitions that survive critical scrutiny and try to reconcile them in light of one another. Boonin exhaustively examined the many arguments for the common intuition that we should reject no-harm-no-foul and/or net-benefit in the Jane example, and he found them lacking. He accordingly concluded that wrong-anyway must be a mistake.²⁹ Jane does nothing wrong by conceiving right away. And if using the contaminatium now would improve the lives of some presently living beings even a little and adversely affect only future generations who would not have otherwise come into existence and who will have lives worth living, then we do nothing wrong by using contaminatium.

Although we might be willing to give Jane a pass, we find Boonin’s conclusion profoundly unsettling as applied to contaminatium. How can it be morally harmless and victimless behavior to despoil the environment even if the harm will manifest only in the relatively distant future? Boonin’s conclusion as applied to nonhumans would mean that any animal whose life is worth living necessarily benefits from being bred for exploitation and slaughter because the animal in question would not have existed but for the intended terrible fate. Understood in this way, the animals-benefit claim in Boonin’s framework becomes irrefutable—and not only for the farmed animals but for enslaved humans in the corresponding human example as well. We hesitate to accept such counterintuitive results. We therefore consider it worthwhile to examine alternatives before giving up on wrong-anyway. If we must give up on one of the moral propositions we have been discussing, perhaps it ought to be no-harm-no-foul or net-benefit rather than wrong-anyway. We will consider each possibility in turn.

Many philosophers characterize no-harm-no-foul (“an act is wrong only insofar as it causes harm to some being”) as mistaken.³⁰ Surprisingly, one can reject no-harm-no-foul from either of two diametrically opposed positions: strong utilitarianism and strong deontology.

Strong utilitarians have little difficulty rejecting no-harm-no-foul. Utilitarians typically evaluate the rightness of an act or omission by the net overall utility it yields. It can therefore be wrong to do one harmless thing if doing a different harmless thing would have affirmatively increased the amount of pleasure, preference satisfaction, or joy in the world.³¹ If Jane conceives immediately, she will give birth to a blind child whose life is worth living. If Jane instead takes a pill and waits a month, she will give birth to a different (sighted) child whose life is also worth living and whose net utility, all things being equal, is greater than that of the blind child. The world will contain more utility if Jane waits, so the morally correct choice is for her to wait.

Likewise, the world in which the present generation relies on contaminatium for fuel will, in the long run, have less utility in it than it would have had if our generation had used a cleaner fuel. The current use of contaminatium will therefore offer future people and animals less utility than the use of a cleaner fuel would have offered the alternative future beings who would have been born in the absence of contaminatium use.

The utilitarian argument thus succeeds in solving the nonidentity problem and does so by discarding the principle of no-harm-no-foul. However, most people—including us—reject utilitarianism as anything like a comprehensive account of morality. We do so because it appears to license many rights violations and terrible choices. To give just one famous example, a utilitarian doctor could (and perhaps must) kill one healthy person who comes into his office for a checkup to provide organs for five (or even two) people awaiting transplants.³²

To be sure, rule utilitarianism (as opposed to act utilitarianism) may be able to rescue utilitarianism from most of the monstrous results to which it seems to lead. Rule utilitarianism performs this rescue feat by reclassifying what appear to be foundational moral principles as rules of thumb. Such rules, if followed consistently, will maximize utility over the long term.³³ A rule-utilitarian doctor, for instance, would not sacrifice one healthy patient to save several others because doing so would, in the long run, discourage patients from visiting any doctor. Despite the rule-utilitarian “save,” however, we share the skepticism of utilitarianism’s many critics and accordingly question whether such rules really do maximize utility.³⁴ Furthermore, we would reject the violation of a rights-protecting rule even if the rule does not maximize utility.

Notwithstanding our own rejection of utilitarianism, we are not interested in making a comprehensive argument against it or against any version of consequentialism. Indeed, we acknowledge that utilitarianism could be helpful to the achievement of our ultimate goal as we attempt to respond to the nonidentity problem. Responding to the nonidentity problem would, in turn, enable us to reply to the animals-benefit argument. Answers to the nonidentity problem tend to travel together in the human and nonhuman cases.

To understand how utilitarianism responds to the animals-benefit claim in particular, suppose one thinks on utilitarian grounds that Jane acts wrongly by conceiving now rather than taking a pill and waiting a month because she would generate more utility in the world in the latter scenario. If so, then one would also likely think that a consumer acts wrongfully by purchasing animal products, even if they come from animals whose lives are worth living. Why? Because in so purchasing, the consumer drives demand for the creation of low-but-positive-net-welfare animals. If, instead, she and other consumers substitute plant-based foods, farmers will require less arable land for food production. After all, producing just one calorie of animal-based food requires the production of more than ten calories of plant-based feed.³⁵ Producing an animal from plants, in other words, is highly inefficient. Directly eating plant-based foods, on the other hand, uses land much more efficiently. Less land needed for cultivation means, in turn, that more land can go back to nature, where wild animals with their own higher utility can thrive.³⁶ Eating plant foods leads to the existence of different animals from those who exist in our current world of animal consumption. For example, we would see more deer and wild turkeys but fewer cows and chickens.³⁷ Those different animals, the ones who would exist in a vegan world, would have better lives than those who currently exist. Eating plant foods rather than animal foods would thus increase overall utility.

If you are a utilitarian and find the above argument persuasive, we are delighted to hear it, and we welcome you into the vegan movement. Honesty, however, compels us to acknowledge that we cannot wholeheartedly embrace the utilitarian argument ourselves.

We have already noted that we do not endorse utilitarianism as a general matter. It leads to perverse results in the human domain, and there is every reason to expect more or less the same perversity when we apply it to the nonhuman world. The veterinarian who sacrifices one healthy dog to supply organs to five other dogs acts wrongfully in the same way as a medical doctor who acts similarly with respect to human patients does.

In addition, although we believe the world would be a better place if humans replaced the animal products they now consume with plant-based foods (and other products), we do not know with certainty that the change would result in greater aggregate utility. Think about the polluted future world that results from using contaminatium. We introduced that example to illustrate the paradoxical power of the nonidentity problem. Because none of the people in that world would exist but for the use of contaminatium, no-harm-no-foul and net-benefit together imply that our current generation does nothing wrong by using it.

Parfit introduced a related example to offer an even more startling conclusion: not only do we do no wrong by using contaminatium, but using the fuel (and thereby poisoning the earth) could actually be morally preferable on utilitarian grounds to decisions that lead to a seemingly wonderful future. The key concept involves numbers. Parfit explained that one could envision a world in which there are a great many people leading lives just barely worth living but whose total utility exceeds that of a much smaller number of people leading wonderful lives in a world in which resources are abundant (due to population control). Parfit described this utility comparison as leading to “the repugnant conclusion.”³⁸ The resource-depleted world of people whose lives are just marginally worth living seems intuitively much worse than the resource-rich world of extremely happy people, but because of the disparity in their respective numbers, the overpopulated world filled with nearly desperate people can boast a higher aggregate utility.³⁹ A utilitarian who emphasizes aggregate utility seemingly must accept the repugnant conclusion that the depleted world is better and oppose measures to encourage population limits.⁴⁰

We think the repugnant conclusion is equally repugnant when the context is animal agriculture. Modern animal farms use land efficiently in the sense that they crowd an enormous number of animals into a small space. If we accept the (extremely dubious) assumption that all these animals have lives just barely worth living, then the farmed animals might so outnumber the happier wild ones who would replace them⁴¹ if the land returned to nature that a nonhuman version of Parfit’s repugnant conclusion would follow. The world of very large numbers of animals with very low utility per animal would be superior in aggregate utility to the world of smaller numbers of animals who each enjoy a very high level of utility. A utilitarian who includes the utility of animals in the calculus and emphasizes aggregate utility might accordingly resist the abolition of animal agriculture.

We do not here attempt to determine whether the repugnant conclusion applies to intensively farmed animals. Our chief reason is that we believe the utilitarian response to the nonidentity problem focuses on the wrong factors. The correct focus should be on whether an action or inaction inflicts harm on individuals. Bulldozing a stretch of rain forest to plant soybeans to feed to cattle harms the creatures who currently live in the rain forest. But failing to convert back into rain forest an animal farm or land that one is already using to grow soybeans as feed for farmed animals does not harm the creatures who would live in that rain forest if the conversion back occurred. This choice does no harm to those animals because they do not exist and they accordingly trigger the nearly universal view among moral philosophers and others that one cannot harm a potential being by failing to bring it into existence.⁴²

The notion that we cannot inflict harm on nonexistent beings by failing to bring them into existence may represent a variant on the no-harm-no-foul principle. If there is no one to harm (because the putative victims do not exist), then it follows that we cannot harm them because acts can be wrongful only insofar as they inflict harm on somebody who can experience that harm.⁴³ Strong utilitarians reject the no-harm-no-foul principle. For them, an action or inaction may be wrong, even if it does not harm any particular victims, so long as it decreases (or fails to increase) the total utility in the world.⁴⁴ That result can occur, as we have seen, when we limit the population and thereby give rise to much happier individuals who are too few in number to add up to a very high aggregate utility. We reject that view for the reasons we offered in our more general critique of strong utilitarianism. Perhaps one might still find the no-harm-no-foul principle unconvincing. However, to the extent that one rejects strong utilitarianism, as we do, one must find some other reason to condemn acts that bring lives worth living into existence and simultaneously cause harm to those lives.

Many philosophers who reject utilitarianism endorse deontology, which typically treats the rightness or wrongness of an act as at least somewhat independent of its consequences.⁴⁵ For example, Immanuel Kant famously said that it is always wrong to lie.⁴⁶ On Kant’s view, lying is wrong even when the lie harms no one. Indeed, for Kant, lying would be wrong even when telling the truth would be harmful and lying would avert harm. One need not agree with Kant’s absolutist stance against lying to sympathize with the broader notion that an act can be wrong without causing harm to anyone.

Deontology condemns some practices regardless of whether the practice harms anyone. One might accordingly infer that other sorts of actions—including those that change which individuals do and do not come into existence—could also be wrong in the absence of any harm. Jane’s decision to conceive immediately might be immoral even though it doesn’t harm anyone in the same way that a lie could be both wrong and harmless. Likewise, the decision to use contaminatium could be wrong even if it does not harm anyone.

Insofar as no-harm-no-foul states a general claim about morality, even a single counterexample will suffice to disprove it. One may think lying is at least sometimes wrong even when it harms no one. Some believe that graverobbing or breaking a promise to a person who has died is wrong, even in the absence of life after death or anyone learning of the transgression. Psychologist Jonathan Haidt has noted that many people respond to moral dimensions that go beyond harm, including values such as sanctity, loyalty, liberty, and authority.⁴⁷ Along one such non-harm-related dimension, perhaps it is wrong for Jane to conceive a blind child when she could easily have conceived a sighted one instead, even if we think Jane does not harm the blind child (because that child has a life worth living) and does not harm the hypothetical sighted child she does not conceive (because that child does not exist).

As that last example helps illustrate, however, it is not quite fair to deny the validity of the no-harm-no-foul principle simply by declaring that the wrong-anyway principle applies. For this reason, we worry that using provocative counterexamples to reject the no-harm-no-foul principle is similarly unfair. Relying entirely on the wrong-anyway principle as a means of rejecting no-harm-no-foul merely restates the nonidentity problem rather than solving it.

We began this exercise by acknowledging that we have moral intuitions about particular hypothetical cases and that those moral intuitions contradict the implications of both the no-harm-no-foul principle and the net-benefit principle. To invoke one or more of those intuitions does not so much show that no-harm-no-foul is wrong as that the three propositions cannot all be true simultaneously—which is what makes the nonidentity problem a problem in the first place. To show that no-harm-no-foul is a mistaken (or at least incomplete) principle, deontology needs more than a pre-reflective moral intuition that the conduct at issue is wrong anyway. It needs to offer an account of how and why no-harm-no-foul fails.

The demand for an account of what makes no-harm-no-foul fail seems like a steep challenge. Deontological theories typically overlap substantially with rights-based theories. Indeed, many of us tend to think of strong moral rights—which Jeremy Bentham derisively called “nonsense upon stilts”⁴⁸—as defining the difference between deontology and utilitarianism. Yet in Jane’s case, the contaminatium case, and every instance of the nonidentity problem, the nonexistent people, in virtue of their nonexistence, do not experience rights violations. Because they do not exist, it makes no sense to speak of any rights they might have, including a putative right to come into existence. And with no rights, there can be no rights violations.

Can deontologists reject the no-harm-no-foul principle without relying on any notion of rights? Perhaps they (or we) can do so by invoking more diffuse notions of respect. In the same way that lying to a person could be said to show disrespect for her moral agency, even if the lie does not otherwise harm her, maybe using the contaminatium shows disrespect for the natural world and the beings who will come to occupy it.

Yet we worry that the use of the term “disrespect” may covertly rely on notions of harm to existing beings. By comparison, we might say that lying shows disrespect for the person to whom one lies even without any concrete setback to that person’s interest. Still, the lie affects the listener, so we have a plausible object of disrespect. However, just as one cannot tangibly wrong a nonexistent being by failing to bring that being into existence, so it would seem that one cannot show disrespect for a nonexistent being.

While it might be impossible to violate the rights of or show disrespect for the potential beings one will not bring into existence, one can certainly wrong the beings that one does bring into existence. Maybe the wrongness of Jane’s action and of the use of contaminatium pertains to what the actor does to the blind child and to the beings who will live in the polluted world. Perhaps deontology should make us question the net-benefit principle rather than no-harm-no-foul. Couldn’t an action harm someone even though the harmful action is a necessary part of giving rise to that someone’s existence? And couldn’t that same harm be cognizable even if the individual in question, the one harmed, has a life worth living?

Philosopher Elizabeth Harman argued that it can be.⁴⁹ She wrote that “an action harms someone if it causes the person to be in a bad state. Bad states are understood as states that are in themselves bad, not bad because they are worse than the state the person would otherwise have been in.”⁵⁰ Hence, according to this view, the principle of net-benefit is wrong.

Harman’s argument and others like it seem commonsensical enough, but you might worry that her idea of a “bad state” smuggles in a comparative judgment we make between people in different states, notwithstanding her claim to the contrary. To be sure, some states are bad in themselves. Physical pain is an obvious example, although even then, if the pain serves a useful purpose—such as alerting one to an underlying treatable illness—we might consider it beneficial on net. Furthermore, if the pain is moderate and replaces extreme pain, we might believe the lesser pain to be a good thing. Think of a medicine that relieves some but not all pain.

Even so, there is something right about Harman’s argument that some states are bad notwithstanding the good that may come about because of or along with the bad. Seana Shiffrin, whose views are similar to Harman’s in important respects, gives an arresting example. She supposed that a gold bar hits and breaks someone’s arm. The resulting pain and injury constitutes a harm even if the value of the gold bar is far greater than the loss occasioned by the broken arm.⁵¹ This concept of non-comparative harm often works well for sensations such as pain.

Non-comparative harm also goes a long way toward answering the nonidentity problem as an objection to animal rights because much of what goes badly for the animals people eat involves the infliction of suffering: piglets squeal in agony when castrated without anesthesia,⁵² cows suffer from frequent mastitis when exploited for milk and may experience terror when they await slaughter,⁵³ and so forth.

We are reluctant to rely entirely on Shiffrin’s argument, however, because we have doubts about an important piece of it. Shiffrin observed that people do not consent to come into existence and that they then inevitably experience harms; she concluded that this infliction of unconsented harm renders all human procreation “morally problematic” (even when it is not “all-things-considered wrong”).⁵⁴ We are not so sure. All of us experience pain and other forms of non-comparative harm at various points in our lives, but most of us do not think that this fact in and of itself means that our parents harmed us by giving us life. We might consider some amount of inevitable pain and suffering to be consistent with a life that is not only worth living but also not the consequence of morally problematic conduct.

In noting that caveat, we nonetheless want to endorse much of what Shiffrin argued. We also note that while she focused on human procreation, her argument is especially powerful as applied to human control of animal reproduction. Farmed animals lead lives that include more pain and suffering than existence inevitably entails for any sentient being—their lives include considerable surplus and even intentionally inflicted pain and suffering.

Below we will consider an actor’s intent as another possible response to the nonidentity problem as applied to animals, but before doing so, let us turn back to the main branch of the argument. Shiffrin, Harman, and like-minded philosophers helpfully introduce the notion of non-comparative harm, a concept that calculates harm based on impact; the net effect of an action is irrelevant and need not be determined. We agree that this concept goes a long way toward explaining how a farmed animal with a life worth living nonetheless was harmed (and wronged) by the actions of the farmer who brought her into existence and made her endure enormous pain. The harms at issue are mostly harms in an absolute sense that requires no comparative judgment.

Yet at least some of the conduct that we regard as harmful to the animals that humans raise to exploit does seem to call for a comparative judgment. When humans confine hens to cages or crowd them into ostensibly “free-range” sheds,⁵⁵ the hens suffer physical discomfort and pain from the concentrated waste in which they must live⁵⁶ and from the aggressive behavior they display toward one another.⁵⁷ In addition to these deprivations, which humans directly inflict upon the animals, they also miss out on complex social interactions, freedom of movement, flight, and many of the other things that make life worth living for their cousins in the wild, such as the red jungle fowl, the presumed progenitor of modern domesticated chickens.⁵⁸ To say that a being misses out is necessarily to make a comparative judgment about harm. We cannot readily say that captive hens miss out on a life more like that of the red jungle fowl without comparing domesticated hens to the wild birds who would otherwise have existed in the captive hens’ place.

Nonetheless, let us concede for the sake of argument that pain and some other raw experiences are bad in themselves. That concession does not do all the work Harman needs to address the paradigmatic nonidentity cases. The core difficulty is the breadth of her conception of a bad experience that does not reference comparatively better experiences but rather is simply bad in an absolute sense. She gave deafness as an example. Yet how can Harman defend the view that deafness counts as a bad state without comparing the deaf person to someone who can hear?

Many disability rights activists sensibly contend that deafness, blindness, and other conditions that we sometimes call “disabilities” are not objectively bad states.⁵⁹ In so arguing, they emphasize a point we observed above: that often the challenges we associate with a disability arise not from the disability itself but from social conditions and society’s choice to design itself around the “able-bodied”—for example, from the assumption that everyone can see and hear.⁶⁰ We share this view, but we put it aside for purposes of this discussion. We will assume that society is what it is. Regardless of whether deafness is deemed bad in and of itself or bad (if at all) only because society builds itself around people who hear, it remains the case that to characterize deafness as bad implies a tacit comparison to a life in which a person can hear. If we are correct, then it seems mistaken to say that blindness or deafness can be a non-comparative harm.

Consider an analogy. In contrast with bats, most birds, and many insects, humans lack the ability to fly. Life might be more satisfying for humans if we could fly, but no one says that a human exists in a bad state in virtue of being unable to fly. Why not? Because of our baseline. If we were to compare ourselves to crows or to bats, we might well conclude that our flightless-ness is a disadvantage. Instead, we compare humans to other humans. Evaluating states like blindness, deafness, and flightless-ness—which do not necessarily entail subjectively negative experiences—seems to require comparison with some baseline. Harman cast her position as a response to the nonidentity problem in large part by denying the comparative nature of judgments about good versus bad states. Because that denial is implausible for some of the paradigmatic nonidentity cases, her analysis falls short of a general solution to the problem.

Moreover, the notion of non-comparative harm may take us down the wrong path because in some contexts being able to make comparisons is affirmatively useful in assessing whether a state is good, bad, or neutral. Consider farmed animals. Modern “layer hens” pretty clearly find themselves in a bad state—a state in which they suffer calcium depletion and vent prolapses.⁶¹ Their state results from the breeding that gave rise to them, breeding that causes the animals to produce many more eggs than their wild ancestors did.⁶² Perhaps we could say that calcium depletion and vent prolapses are the kinds of states that are bad in an absolute sense, the same way that experiencing pain is bad in an absolute sense (even if it leads to a greater good, like the treatment of illness). But we want to say more than that. We believe that being a hen prone to calcium depletion and vent prolapse is bad even before she experiences these conditions. To make that kind of statement requires us to compare the hen’s life with that of a similar bird not prone to those ailments. We think a layer hen prone to suffering because of her genetic endowment is in a bad state, even if the humans with whom she lives treat her kindly as a family pet so that she has a life worth living. We cannot, however, make this judgment without comparing the hen to a different bird in the state that she would occupy if not for the many generations of selective chicken breeding.

Accordingly, Harman failed to convince us that her argument does or even should succeed in reaching all key badness judgments without making any comparisons. That said, there is much to admire in her argument. Moreover, explaining exactly where it fails may lead us to a more promising line of reasoning. The next part attempts that task.

Harman thinks that it is possible to maintain three propositions simultaneously: an act can be wrong in virtue of harming a being; the same act is necessary for that being’s coming into existence; and the being in question has a life worth living.⁶³ How can these three assertions be mutually consistent? Harman said that “the mere fact that a harming action also benefits—that is, provides positive good things to the harmed person—does not justify the harm.”⁶⁴ In so saying, Harman assumed that conferring existence on a being affirmatively benefits that being. Harman went on to argue, however, that “considerations of benefit are therefore ‘ineligible’ to justify the harm.”⁶⁵

We think Harman’s calculus falters—but only barely. If a being has a life worth living, then by definition the benefit existence confers outweighs the harm associated with whatever act caused the being to exist. So why did Harman say that the benefit is ineligible to justify the harm? Her answer rests on what looks like a rejection of the no-harm-no-foul principle rather than of the net-benefit principle. She said that if a harmful “action also benefits the harmed person, but performing the action is not the only way to provide such benefits—indeed, refraining from performing the action would provide similar benefits to someone—then considerations of benefit simply do not tell in favor of acting as opposed to refraining from acting.”⁶⁶ In other words, if we have two options, each of which would create a being with a life worth living but one of which would bring the same benefits as the other but without the harm, then we can critique the more harmful choice without having to count the (greater) benefits against the harm.

We agree with the bottom line here: that an act that both creates a being and harms that very being can be wrong without the harm having to outweigh the good in that being’s life—indeed, without the good having to be considered at all. We worry, however, that Harman’s undefended ipse dixit makes it all too easy for a skeptic simply to deny a claim of wrong by someone with a life worth living by reasserting no-harm-no-foul and net-benefit. In any event, there is a more straightforward way to get to Harman’s bottom line, and this alternate path better captures what we think Harman meant (or ought to mean!) when she said that considerations of benefit are “ineligible” as a justification for harm.

In our view, the benefit of existence with a life worth living does not count (or in Harman’s terms is “ineligible”) because existence does not confer a benefit on anyone. Existence is instead a precondition for experiencing benefits and harms. Before coming into existence, no being wants or needs existence. No one is “there” waiting to benefit from coming into existence. When you bring a being into existence, you thereby generate wants and needs. You might or might not then satisfy those wants and needs. But creating a being does not itself meet any wants or needs because, by hypothesis, no one was “there” (or anywhere) before coming into existence to need or want anything.

To be sure, some philosophers think that causing a being with a life worth living to exist benefits that being, or that we can at least plausibly describe existence as a benefit to the extent that the being experiences life as one worth living.⁶⁷ We disagree. A wonderful life may be a blessing in the sense that the being enjoying it feels grateful, but we cannot fairly describe it as a benefit to the being. “Benefit” connotes making someone better off than they were before, but before they came into existence there was no being to become better off, so the comparison fails.

One might think existence is better than nonexistence so long as existence involves positive well-being, because positive well-being is better than zero well-being. But we cannot accurately describe those who do not yet and may never exist as having zero well-being. They instead have no well-being at all.⁶⁸ Saying it is better to exist as a happy human than never to have come into existence at all is like saying it is better to exist as a happy human than to exist as a vacuum cleaner; the statement is nonsensical because there is no such thing as experiencing existence as a vacuum cleaner. Likewise, and no less clearly, there is no such thing as existing as a nonexistent being.

Hold on. How can we say that comparing existence and nonexistence is impossible, given that we also claim that nearly all animals exploited by humans have lives not worth living? Doesn’t the very concept of a life worth or not worth living imply a comparison between existence and nonexistence? It does indeed, but this fact does not contradict our core claim that nonexistent beings lack any state of well-being. Our view is asymmetrical. Saying that existence cannot be better than nonexistence does not obligate us to deny that existence can be worse than nonexistence.

Some lives are so miserable that they are never worth living. Other lives that are worth living overall are not worth continuing when they begin to entail more suffering than a being can endure, especially in the absence of compensating benefit.

The practices of suicide and euthanasia capture the idea that for some beings, nonexistence is preferable to (continued) existence. True, many people condemn suicide and euthanasia, but their reasons for doing so may not withstand critical scrutiny. Meanwhile, some people who raise objections to the practices of suicide and euthanasia do so on grounds that do not cast doubt on the possibility of comparing existence to nonexistence. These objectors do not necessarily deny that more life could be a net burden; they might think that they have a religious or other obligation to bear that burden.⁶⁹ Others might worry about the collateral risk to vulnerable populations from legal suicide or euthanasia.⁷⁰ Still others could think that most people who attempt suicide do so impulsively and would be grateful, in the long run, if they failed in the attempt.⁷¹ These are all potentially legitimate objections to suicide and/or euthanasia, but none denies that for at least some people, death is better than continued life.

Now, you might think that we have pulled a fast one by basing an argument on suicide and euthanasia. After all, our claim is not that existence can be worse than death but rather that existence can be worse than nonexistence. Our answer is that death—once one has died—is nonexistence in the relevant sense. Indeed, failing to bring someone into existence is considerably less traumatic and eventful (for the one who does not exist) than is the process of dying by whatever means. To the extent that killing is permissible, it would therefore follow a fortiori that refraining from creating is also permissible in parallel circumstances. We would not claim, of course, that a being who has died is the same in every respect as one who never existed. People and other animals who have lived leave behind others who remember them, and the deeds of departed people and animals continue to affect the world (for good and ill). For purposes of comparing existence and nonexistence, however, we can take lessons from death that apply to the state of never having existed.

Consider an example drawn from popular fiction. In season five of the HBO television series Game of Thrones, Stannis Baratheon condemns Mance Rayder to die by burning at the stake as punishment for failing to accept the former’s authority. Jon Snow, who has come to respect and admire Mance, watches in horror and disgust as his friend screams during the prolonged immolation. Jon leaves the area, quickly returns with his bow and arrow, and shoots Mance in the heart, ending his life.⁷²

It is plain to anyone watching that Jon has done Mance a mercy. But how can that be? Once Mance is dead, he does not experience relief from no longer burning. And up until the instant of death, he continues to suffer the excruciating torment of immolation. So who exactly benefits, and when?

One possible answer is that future Mance, who would have continued to suffer for additional seconds or minutes before dying, benefits from having been spared that additional suffering, but once actual Mance dies of the arrow to the heart, future Mance who lingers and burns is a mere hypothetical person. One might therefore think that we beg the question posed by the nonidentity problem if we count the well-being of future Mance in our calculus. After all, we do not count the well-being of the potential sighted child or the future people who would exist but for the present consumption of contaminatium. So why count the well-being of future Mance? And for that matter, why count the ostensible benefit from euthanasia that we carry out on our beloved companion animals when their illnesses become too painful for them to have to tolerate? Our dogs and cats cease to exist at the very moment we purport to bestow a benefit on them. Rational suicides pose the same dilemma.

The short answer is that Jon’s mercy killing of Mance is obviously in Mance’s interest, so much so that we should reject any theory that fails to acknowledge the intuition. To be sure, we might find it impossible to construct a persuasive moral theory consistent with all our pre-reflective intuitions. Such inconsistencies require us to sacrifice some moral intuitions in favor of others that we hold more deeply. Yet the intuition in Mance’s case and real-world situations like it, including euthanasia and rational suicide, are among those very deeply held intuitions for which we might wish to surrender others that we hold less deeply. As we see it, no plausible moral theory could require us to give up the intuition that Mance is better off if Jon’s arrow puts him out of his misery than he would have been if he had continued to suffer the torment of burning to death while conscious. The fact that we cannot pinpoint a time during Mance’s life when his existence is better because of the arrow does not mean we should give up that extremely compelling moral intuition.

We accordingly affirm with great confidence that a being sometimes benefits when his existence ends so that his suffering too may end. From the fact that nonexistence can be better for a suffering being than existence, we think it follows that coming into existence can be bad for the being who previously did not exist. We would be careful to say, however, that the reason we consider such existence bad in these circumstances is not that existence itself inflicts a harm on the being. The reason is instead that the being in question, after coming into existence, endures a life not worth living, a life full of suffering the predicate for which is existence.

Aha! you might exclaim, if coming into existence and then having a life not worth living harms a being, then why doesn’t coming into existence and then having a life that is worth living benefit a being? And if it is a benefit, then isn’t it still a benefit even if a different being with an even better life could have come into existence? Our answer is that (a) one can experience a life worth living without being better off than one would “be” if one had never existed, but (b) that proposition does not negate the intuition that a person or animal who experiences a life not worth living can be worse off than he would be if he had never existed. We need not, in other words, sacrifice either intuition.

To understand the basis for our asymmetrical view, it may help to think of it in terms of the allegedly “worse off” one in each situation: the suffering being can coherently wish that she had never been born or even wish for nonexistence because life is so filled with pain and suffering. The nonexistent being, by contrast, who would have had a wonderful life if he or she had only existed, cannot (either in fact or conceptually) wish that she had come into existence to have that wonderful life.

We might now be open to a different objection, however. If we may not measure the good in life by comparison to nonexistence, does it follow that we may not condemn as harmful the sudden and painless killing of a happy person? We would answer this question no because of a different asymmetry. We noted in our discussion of suicide and euthanasia that for some purposes we would treat death as equivalent to never being born. But not for all purposes. The so-called endowment effect, which we have developed into a moral concept, not just a psychological disposition,⁷³ marks an important distinction between never existing, on the one hand, and ceasing to exist, on the other.

Researchers have observed that people seek to avoid losses more than they pursue equivalent gains.⁷⁴ They demand more in exchange for giving up an object (regardless of what the object is) than they willingly pay to acquire the same object. Economists regard so-called loss aversion and the endowment effect (overvaluing our current endowments relative to what we might acquire) as irrational, and in some contexts these attitudes indeed make little sense. For example, the stock trader who holds onto a portfolio even when the market is falling because he bought it when the market was still higher will do worse than one who treats each day of trading as a fresh start.⁷⁵

In many contexts, however, we would describe the endowment effect not only as rational but as morally potent as well. The moralized endowment effect recognizes that taking something away from someone who already possesses it is worse than never giving something to someone in the first place, even if what they stand to lose is no more valuable than what they stand to gain. The line we draw here closely tracks the act/omission distinction, which plays a critical role in conventional morality. If a thief steals your bicycle, he harms you; if your friend fails to give you a bicycle, she may disappoint you, but she does not harm you. Our law reflects the moralized endowment effect in numerous ways, including the institution of property rights and the attendant punishments for property crimes.

We understand the difference between ending a being’s life and not bringing a being into existence in the first place as an extension of the endowment effect to a person’s very existence. The loss of life—the loss of existence—is qualitatively different from a failure to come into existence.

Our response to the last objection thus brings us back to an essential point: depriving a being of what was a life worth living harms that being. Putting the point concretely, imagine the best possible scenario for animals destined for slaughter (which is wholly unrealistic): all animals in a fantasy farm have lives worth living. They lead fulfilling, happy lives before they suddenly and painlessly die in the fantasy humane slaughterhouse. The animals all had lives worth living, and the painless slaughter nonetheless harms the animals. Yet because one does not earn “credit” for bringing beings into existence, one cannot offset the harm by invoking the prior life worth living. The nonidentity argument accordingly cannot redeem the creation or purchase of animal products from even the fantasy of a humane farm.

* * *

Maybe our general answer to the nonidentity problem has persuaded you. Or maybe you find compelling our adaptation of Harman’s and Shiffrin’s arguments rejecting the notion that a life worth living on the part of an animal can offset pain and some other raw experiences. Perhaps you agree that we cannot compare the net happy existence we have created with the nonexistence that would otherwise have awaited the creature in question. Perhaps you view this incommensurability as sufficing to answer the nonidentity problem, at least with respect to the skeptic who claims that consuming farmed animals with lives worth living grants a benefit to the animals who would not exist absent the skeptic’s purchase and consumption of animal products. If any of these arguments has persuaded you, we are glad of it.

You may, however, have lingering doubts. Or perhaps you think there might be other solutions to the nonidentity problem. Thus, we consider two more possibilities.

An alternative approach to the nonidentity problem in the animal case would treat purchasers of animal products just like farmers. Recall our claim that a farmer cannot properly invoke the fact that he gave an animal life as an excuse for the suffering and slaughter he then inflicts on the animal. Giving life creates responsibilities, not entitlements. The consumer, by contrast to the farmer, combines the good and the bad into one: his single act of purchasing animal products signals to the farmer a willingness to pay money in the future for such products, so the farmer takes the two discrete steps of breeding the “happy” animal and then unjustifiably exploiting or killing that animal. Early in this article we tried to make sense of the animals-benefit defense of animal use by redescribing it as an instance of the nonidentity logic. The consumer’s purchase, we said, causes both the existence and the exploitation and killing of future animals.

But maybe that framing is too forgiving of consumers. Maybe we can best understand consumers of animal products as demanding exploitation and killing, full stop. When people register their demand for meat, they certainly foresee that animals will be slaughtered to satisfy that demand. We might thus regard the consumer as an accomplice in the slaughter. Indeed, we have good reason to think the consumer demands only the slaughter and not a humane pre-slaughter existence. A thought experiment can help us see why.

Suppose a farmer wants to figure out how many chickens to raise for slaughter and sale at a new grocery store in his community. Eventually, the farmer intends to rely on market signals of demand to determine how many chickens to raise for the new grocery store, but because the store has not yet opened, he decides to conduct a survey. He asks a random sample of people who live nearby whether they plan to shop at the new store and, if so, whether they plan to buy chicken there. The survey asks those who answer yes how much they would be willing to pay for “humanely raised” chicken. Now suppose that another one of the questions in the survey (inserted surreptitiously by the farmer’s daughter, an animal rights activist), asks the following: How much would you be willing to pay for humanely raised chicken if, instead of actually receiving any chicken meat, you were to receive semiannual email reports about the satisfying life the chicken you paid for is living at a sanctuary until the day that chicken dies of natural causes, at which point the farmer buries her corpse in a dignified manner because chickens who die of old age or other natural causes do not meet food-safety standards?

Some small number of people might pay enough to support fulfilling lives for the chickens. That is what supporters of sponsor-an-animal programs at farm sanctuaries do.⁷⁶ But those respondents probably don’t even eat chicken. The vast majority of chicken-eating respondents would find the question bizarre. They are interested in giving chickens good lives, if at all, only on the way to producing chicken meat for them to consume. And that is true regardless of whether they signal demand via a survey or via purchases. They demand a slaughtered chicken. The chicken’s existence as a living being before slaughter (or the cow’s existence apart from her production of milk) is a kind of byproduct of the demand for the meat (or milk or other animal product). Accordingly, we can fairly treat the consumer as complicit in the farmer’s bad acts toward the animals—the farmer is the consumer’s agent in directly carrying out those bad acts.

Now, you might say that consumers of animal products do not demand slaughter, per se. They register a desire for meat, fish, dairy, eggs, leather, and other material goods, however produced. Consumers of animal products are of course aware that creating these products harms animals, but they prefer not to give any thought at all to the sad or unsettling parts of the process.⁷⁷ So, the objection goes, they demand products they know come from the exploitation and killing of animals, but they do not demand the exploitation and killing as such. It just happens that there is no way to obtain these products without exploitation and killing. The consumers of animal products do not so much intend the harm to animals, in this view, as foresee it and proceed anyway.

As Alastair Norcross observed, this counterargument makes a double-effect claim. The doctrine of double effect posits that an action that foreseeably causes even a very serious harm may be morally permissible if the harm is merely a side effect of some act intended to bring about a benefit.⁷⁸ Just as a doctor may foresee but not intend the death of a suffering, terminally ill patient to whom the doctor administers an anesthetic dose sufficient to treat the patient’s pain, so a consumer’s purchase of animal products may foresee but not intend the exploitation and killing of animals. Yet, as Norcross responded, double-effect arguments succeed only where the good of the intended action (such as treating the patient’s terrible pain) outweighs the unintended harm (such as the patient’s hastened death). In the animal context that is plainly not the case. The harm that people inflict upon animals by obtaining animal products outweighs whatever additional gustatory pleasures or fashion statements people can, respectively, enjoy or make by exploiting and killing animals instead of consuming non-animal substitutes.⁷⁹

We agree with Norcross that the double-effect argument fails, but we would go further to say it should not even get off the ground. The terminally ill patient who seeks pain relief that will foreseeably kill her has an aim independent of her own death: to alleviate her pain. By contrast, the purchaser of animal products—a roasted chicken, say—seems to have no aim other than to obtain a dead chicken. Not long ago, a colleague of ours ate what he regarded as a delicious snack, only to discover afterward that it was vegan and thereupon announce that the knowledge retroactively made the snack less tasty. That attitude—reflected in extremely common jokes about supposedly bland or otherwise unappetizing vegan food—is quite common. It shows that for more than a few consumers of animal products, the harm to animals is a feature of what they demand, not a bug.

Wait. You might object that our colleague and other animal-product consumers like him demand the animal products, not the harm to the animals that precedes their appearance on their plates. We grant that the purchaser has an ultimate goal that goes beyond the desire to simply own dead chickens; he wants to satisfy his appetite for chicken flesh. But the existence of ultimate reasons does not convert a primary intention into a mere unintended but foreseen consequence. If Joan hires a hit man to kill her former confederate Albert so Albert won’t testify against Joan at her impending trial for bank robbery, we would hardly say that Albert’s death is the foreseen but unintended consequence of Joan’s action aimed at securing her acquittal. Likewise, to demand dead chickens to eat is to demand the killing of chickens.

In sum, we think we have a powerful argument that purchasing animal products either does not at all or does not fully implicate the nonidentity problem because farmers can refrain from slaughtering the animals they have bred into existence and consumers are complicit in the farmers’ choice not to. Perhaps you disagree. Or perhaps you are persuaded, but you remain interested in responding to the nonidentity problem in other contexts to which it more readily applies. We next consider a potential solution that implicates historical injustices and our current treatment of animals.

In 2014, The Atlantic magazine published an article by Ta-Nehisi Coates calling for reparations to African Americans for the historical injustices of slavery, segregation, and systemic racism.⁸⁰ Since then, various politicians and other public figures have echoed the call.⁸¹ Whether to pay reparations, through what mechanisms, and to whom, are all politically contentious questions with many dimensions that we will not address. We focus here on the ways in which reparations implicate the nonidentity problem.

One of us (Sherry) is the daughter of Holocaust survivors. Sherry’s father, Ben-Zion, lived in hiding while operating a secret rescue effort that ultimately smuggled over a thousand Jews (mostly children) out of Poland.⁸² Her mother, Clara, spent the Second World War in what was then occupied Czechoslovakia under a false identity, narrowly escaping capture on multiple occasions. After the war, Ben-Zion and Clara came to the United States as immigrants. Sherry was born many years later. If the Nazis had not victimized Ben-Zion and Clara and murdered their families, it is virtually certain that Sherry would not exist today, given how little it takes to change the trajectory of one particular sperm cell headed for one particular egg. Likewise, were it not for slavery, most of the African Americans who live in this country now would almost certainly not have been born. If their lives are worth living, are they barred from condemning the perpetrators for wronging them?

The reparations question overlaps only partially with the nonidentity problem. Certainly, individuals who themselves directly experienced historical injustices need not confront a nonidentity problem in making their case for restitution. Until her death in 2009, Clara received reparations from the German government. Pursuant to legislation enacted in 1988, the United States paid $20,000 and apologized to each of the surviving civilians of Japanese ancestry whom the U.S. government imprisoned during World War II.⁸³ Many African Americans who might receive reparations today grew up during the Jim Crow era, faced redlining, and have otherwise suffered as the direct victims of racial injustice.⁸⁴ Even African Americans born more recently suffer from systemic racism.⁸⁵ Reparations to all these direct victims generally raise no nonidentity problem; the injustice for which they would receive compensation did not cause their existence.

Nor do all payments to descendants of victims necessarily implicate the nonidentity problem. Consider the case of stolen property or expropriated labor. Suppose that Sandra’s great-great-grandfather Henry was an enslaved craftsman who built elegant buildings and objects for the people enslaving him. Sandra would not exist were it not for Henry’s enslavement, but we can nonetheless say that Henry should have received payment and that if he had, he would have been able to pass on his wealth as an inheritance to his descendants, including Sandra. Her existence as an individual with a life worth living should not bar her right to an inheritance. Accordingly, George Sher argued persuasively that righting wrongs that result from failure to compensate direct victims escapes the nonidentity objection.⁸⁶

Reparations, however, do implicate the nonidentity problem insofar as they seek to compensate people for current disadvantages resulting wholly from a past injustice in the absence of which they would never have existed. We regard as uncontroversial the proposition that modern society ought to end practices of discrimination and other forms of inequity that directly harm current descendants of the victims of historical injustice. But how can the descendants of direct victims claim, above and beyond the right against discrimination and oppression, a right to reparations for the modern consequences of slavery or the Holocaust or any other historical injustice but for which the descendants would not have existed and led the lives they currently lead, lives worth living? That is a nonidentity question.

We wish to be clear about the nature of the harm we identify here. Many children of Holocaust survivors face challenges that result from the physiological adaptations of their mothers to earlier bouts of starvation.⁸⁷ Slavery and Jim Crow likewise left specific marks on everyone they touched in addition to their general impact in shaping society as a whole. In short, descending from a victim of historical injustice can itself give rise to other disadvantages.

We have an especially strong intuition that descendants of enslaved people, Holocaust survivors, and others who suffered historical injustices are entitled to complain about the harms to them and not just to their ancestors, even though the complainants would not exist but for the earlier injustices. One way to make sense of that intuition is to rely on a general answer to the nonidentity problem—such as the utilitarian rejection of no-harm-no-foul or our own view that one cannot accrue credit for bringing a being into existence because coming into existence does not benefit anyone. However, as we noted above, those answers might not have fully persuaded some readers. Is there a more limited response to the special case of the nonidentity problem involving historical injustices?

Ori Herstein said there is. He explained that because historical injustices targeted groups rather than just particular individuals, we can avoid the nonidentity problem altogether. We do so by observing that Jews and African Americans would exist now even in the absence of the Holocaust and slavery, and those groups would have had greater well-being along many dimensions than we and they have in the wake of these atrocities. Hitler wronged Sherry, even though his conduct was likely a but-for cause of her existence, because he wronged Jews as a group and Sherry is part of that group. Herstein explained that

To be sure, the focus on group harm raises a potentially contentious question: who counts as a member of the relevant group? Is former President Barack Obama among the members of the group of present-day African Americans who suffer constitutive harm as a legacy of slavery? Can that be true even though the ancestors of Obama’s African father never experienced enslavement?⁸⁹ If so, is it because Obama was born and grew up in the United States as what society codes as African American? What about immigrants and the children of immigrants from countries outside of the United States that held the immigrants’ ancestors in slavery? Would they have a claim for reparations against the United States? Did Hitler wrong not only Sherry but also the other author of this article (Michael), who is likewise Jewish but whose great-grandparents came to the United States decades before World War II? Herstein would answer yes to each of these questions because, as he wrote, “[t]here are ways to enter a group besides being born into one or having a familial ancestral lineage made up of past group members.”⁹⁰ Not everyone would agree, of course, and the issue could be divisive should it become relevant to the mathematical formula for dividing up a finite sum of reparations.

Relying on the notion of group harm to solve the nonidentity problem raises another question: what counts as a group? Persecution of a racial, ethnic, or national group is a paradigmatic case, but could we expand the idea to provide a solution to a broader subset of nonidentity problems?

Consider the child who will be born blind if Jane conceives right away. It seems sensible to assume that obstacles specifically confronting blind people result not mainly from intentional discrimination against blind people but instead from architectural and other features of our society that simply assume everyone can see.⁹¹ Jane’s blind child will inevitably suffer so long as society tolerates the structural prejudice that sometimes occurs despite an absence of animus toward the disadvantaged group.

What about a future generation of people who will be born into a world damaged by contaminatium’s effects? They comprise a group in the literal sense that there are many of them, and they all experience the harm of living in a degraded environment. Nonetheless, some readers might think that roughly contemporaneous birth does not suffice to constitute people the kind of group to which Herstein’s argument applies.

Whatever one concludes with respect to the contaminatium example, members of one or more animal species raised for human exploitation do seem to fit Herstein’s group harm paradigm. Many people understandably regard comparisons between the exploitation of animals and slavery and the Holocaust as offensive, but the comparison is more than apt if one attends to the sheer numbers and to what humans in fact do to animals. That is not to say that effective advocates for animal rights should make such comparisons.⁹² It is to say that human treatment of animals tracks the pattern of historical injustices with respect to the essential point: humans deprive cows, pigs, chickens, and other animals of their liberty, impose physical and emotional suffering on them, and ultimately kill them, simply because of the group into which they were born—here, their species.

To be sure, animals do not experience every aspect of oppression that humans experience when falling victim to prejudice and discrimination. An African American man in the Jim Crow era who walked in public with his eyes cast down lest he become a target for a lynch mob claiming he looked at a white woman experienced humiliation as well as fear and economic disadvantage. So did the European Jews who had to wear stars of David under Nazi orders. By contrast, when a farmer takes a cow’s babies, the cow experiences loss but does not feel stigmatized as a cow. So far as we know, only humans and not the animals whom humans exploit can experience group-based stigma.

But so what? When, during the Rwandan genocide, Hutus massacred babies and children too young to know either that they were Tutsi or that Hutus were slaughtering them for that reason,⁹³ those babies and children nonetheless experienced the group-based historical injustice. Stigma often accompanies invidious discrimination, but it does not appear to be an essential feature of the sort of group harm that Herstein identified in his solution to the nonidentity problem. Accordingly, the concept of group harm provides an alternative answer to the animals-benefit claim, even if it does not solve those manifestations of the nonidentity problem that involve only individuals, not groups.

This article took seriously the claim that people who consume animal products confer a benefit on the animals bred, exploited, and slaughtered to satisfy their tastes. We took it seriously by assuming for the sake of argument that such animals have lives worth living. In conclusion we reiterate that this assumption is false in nearly every case.

We nonetheless indulged the arguendo assumption that some exploited animals have lives worth living because doing so enabled us to offer a fresh perspective on the nonidentity problem. As our mostly interchangeable use of examples involving humans and nonhuman animals illustrates, moral questions involving the latter implicate such questions involving the former, and vice versa. Thus, a subsidiary aim of this article is to bring consideration of animal well-being into the discussion of familiar questions of human-human morality. Whether we succeeded in this goal with any particular readers depends on both the strength of our arguments and on those readers’ willingness to look beyond the blinders of species prejudice.