Practical Ethics

We have seen that there are reasons for holding that the killing of a person is more seriously wrong than the killing of a being who is not a person. This is true whether we accept preference utilitarianism, Tooley's argument about the right to life or the principle of respect for autonomy. Even a hedonistic utilitarian would say that there may be indirect reasons why it is worse to kill a person. So in discussing the wrongness of killing nonhuman animals, it is important to ask if any of them are persons.

It sounds odd to call an animal a person, but this may be no more than a symptom of our habit of keeping our own species sharply separated from others. In any case, we can avoid the linguistic oddness by rephrasing the question in accordance with our definition of ‘person’. What we are really asking is whether any nonhuman animals are rational and self-conscious beings, aware of themselves as distinct entities with a past and a future.

In ancient myths and in contemporary stories and movies, we imagine being able to talk to animals. That dream was at least partially realized in 1967 when two scientists at the University of Nevada, Allen and Beatrice Gardner, guessed that the failure of previous attempts to teach chimpanzees to talk was due to the chimpanzees’ lacking, not the intelligence required for using language, but the vocal equipment needed to reproduce the sounds of human language. The Gardners therefore decided to treat a young chimpanzee as if she were a human baby without vocal chords. They communicated with her, and with one another when in her presence, by using American Sign Language, a language widely used by deaf people.

The technique worked. The chimpanzee, whom they called ‘Washoe,’ learnt to understand about 350 different signs and to use about 250 of them correctly. She put signs together to form simple sentences and, in doing so, provided strong evidence of a sense of self. When shown her own image in a mirror and asked ‘Who is that?’ she replied: ‘Me, Washoe.’ Later Washoe moved to Ellensburg, Washington, where she lived with other chimpanzees under the care of Roger and Deborah Fouts. Here, she adopted an infant chimpanzee and soon began not only signing to him but even deliberately teaching him signs, for example, by moulding his hands into the sign for ‘food’ in an appropriate context. Washoe died in 2007 at the age of forty two.

Gorillas, bonobos and orangutans have also been able to learn sign language, although the extent of their ability is controversial. For more than thirty years, Francine Patterson has been signing and speaking English with Koko, a lowland gorilla. She claims that Koko now has a working vocabulary of more than 500 signs and understands an even larger number of spoken English words. In front of a mirror, Koko will make faces or examine her teeth. Chantek, an orangutan, has been taught sign language by Lyn Miles. When shown a photograph of a gorilla pointing to her nose, Chantek was able to imitate the gorilla by pointing to his own nose. Apes also use signs to refer to past or future events, thus showing a sense of time. The Fouts hold regular festivities for the chimpanzees at Ellensburg. Each year, after Thanksgiving, Roger and Deborah Fouts set up a Christmas tree covered with edible ornaments. The chimpanzees use the sign combination ‘candy tree’ to refer to the Christmas tree. In 1989, when snow began to fall just after Thanksgiving but the tree had not yet appeared, a chimpanzee named Tatu asked: ‘Candy tree?’ The Fouts interpreted this as showing, not only that Tatu remembered the tree, but also that she knew that this was the season for it. Later, Tatu also remembered that the birthday of one of the chimpanzees, Dar, followed closely on that of Deborah Fouts. The chimpanzees got ice cream for their birthdays; and after the festivities for Deborah's birthday were over, Tatu asked: ‘Dar ice cream?’

Suppose that on the basis of such evidence, we accept that the signing apes are self-conscious. Are they exceptional among all the nonhuman animals in this respect precisely because they can use language? Or is it merely that language enables these animals to demonstrate to us a characteristic that they, and other animals, possessed all along?

Some philosophers have argued that thinking requires language: one cannot think without formulating one's thoughts in words. The Oxford philosopher Stuart Hampshire, for example, has written:

The difference here between a human being and an animal lies in the possibility of the human being expressing his intention and putting into words his intention to do so-and-so, for his own benefit or for the benefit of others. The difference is not merely that an animal in fact has no means of communicating, or of recording for itself, its intention, with the effect that no one can ever know what the intention was. It is a stronger difference, which is more correctly expressed as the senselessness of attributing intentions to an animal which has not the means to reflect upon, and to announce to itself or to others, its own future behaviour…It would be senseless to attribute to an animal a memory that distinguished the order of events in the past, and it would be senseless to attribute to it an expectation of an order of events in the future. It does not have the concepts of order, or any concepts at all.

Obviously, Hampshire was wrong to distinguish so crudely between humans and animals; for as we have just seen, the signing apes have shown that they do have ‘an expectation of an order of events in the future’ Hampshire wrote before apes had learned to use sign language, so this lapse may be excusable. Suppose that his argument were to be rephrased so that it referred to animals who have not learned to use a language, rather than all animals. Would it then be sound? If so, no being without language can be a person. This applies, presumably, to young humans as well as to non-signing animals. It might be argued that many species of animals do use language, just not our language. Certainly most social animals have some means of communicating with one another, whether it be the melodious songs of the humpback whales, the buzzes and whistles of dolphins, the alarm calls of vervet monkeys, which vary according to the kind of predator sighted, the howls and barks of dogs, the songs of birds and even the dance performed by honey bees returning to the hive, from which other bees learn the distance and direction of the food source from which the bee has come. Whether these forms of communication amount to language, in the required sense, is doubtful. Because pursuing this issue would take us too far from our topic, I shall assume that they do not, and consider what can be learned from the non-linguistic behaviour of animals.

Hampshire's argument is an example of a pitfall to which philosophers of previous generations were especially prone: reaching conclusions from the armchair on a topic that demands investigation in the real world. There is nothing altogether inconceivable about a being possessing the capacity for conceptual thought without having a language, and there are instances of animal behaviour that are extraordinarily difficult, if not downright impossible, to explain except under the assumption that the animals are thinking conceptually. In one experiment, for example, German researchers presented a chimpanzee named Julia with two series of five closed and transparent containers. At the end of one series was a box with a banana; the box at the end of the other series was empty. The box containing the banana could only be opened with a distinctively shaped key; this was apparent from looking at the box. This key could be seen inside another locked box; and to open that box, Julia needed another distinctive key, which had to be taken out of a third box which could only be opened with its own key, which was inside a fourth locked box. Finally, in front of Julia, were two initial boxes, open and each containing a distinctive key. Julia was able to choose the correct initial key with which she could open the next box in the series that led, eventually, to the box with the banana. To do this, she must have been able to reason backwards from her desire to open the box with the banana to her need to have the key that would open it, to her need for the key that would open that box, and so on. Because Julia had not been taught any form of language, her behaviour proves that beings without language can think in quite complex ways.

Nor is it only in laboratory experiments that the behaviour of animals points to the conclusion that they possess both memory of the past and expectations about the future, that they are self-aware and that they form intentions and act on them. For several years, Frans de Waal and his colleagues watched chimpanzees living in semi-natural conditions in two acres of forest at Arnhem Zoo in the Netherlands. They often observed co-operating activity that requires planning. For example, the chimpanzees liked to climb the trees and break off branches so that they could eat the leaves. To prevent the rapid destruction of the small forest, the zookeepers placed electric fencing around the trunk of the trees. The chimpanzees overcame this by breaking large branches from dead trees (which had no fences around them) and dragging them to the base of a live tree. One chimpanzee then held the dead branch while another climbed up it, over the fence and into the tree. The chimpanzee who got into the tree in this way shared the leaves thus obtained with the one holding the branch.

De Waal also observed deliberately deceptive behaviour that clearly shows both self-awareness and an awareness of the intentions of another. Chimpanzees live in groups in which one male will be dominant and will attack other males who mate with receptive females. Despite this, a good deal of sexual activity goes on when the dominant male is not watching. Male chimpanzees often seek to interest females in sexual activity by sitting with their legs apart, displaying their erect penis. (Human males who expose themselves in a similar way may be continuing a form of primate behaviour that has become socially inappropriate.) On one occasion, a junior male was enticing a female in this manner when the dominant male walked over. The junior male covered his erection with his hands so that the dominant male could not see it.

Not only philosophers like Hampshire, but also some scientists have argued that ‘mental time travel’ – the ability to imagine a future event – is unique to humans. As so often happens with attempts to draw lines between humans and animals, this one had to be stated in a very precise form in order to be at all plausible. Everyone who has a dog as a companion knows that the dog can anticipate going for a walk. The ability unique to humans is therefore said to be that of anticipating the future beyond one's current set of motivations. So this claim is not refuted by a dog who brings her lead and puts it at the feet of her human companion. The dog, it is said, is simply in the grip of her desire to go for a walk and is acting on that desire. Humans, in contrast, can plan on satisfying motivations that they do not presently feel – as when we go shopping to make sure there will be something to eat for dinner, even though we are not hungry now. Many animals will hide food for future use, as squirrels do, but can it be shown that this involves conscious forethought, rather than purely instinctive behaviour?

Jane Goodall has described an incident showing forward planning by Figan, a young wild chimpanzee in the Gombe region of Tanzania. In order to bring the animals closer to her observation post, Goodall had hidden some bananas in a tree:

One day, sometime after the group had been fed, Figan spotted a banana that had been overlooked – but Goliath [an adult male ranking above Figan in the group's hierarchy] was resting directly underneath it. After no more than a quick glance from the fruit to Goliath, Figan moved away and sat on the other side of the tent so that he could no longer see the fruit. Fifteen minutes later, when Goliath got up and left, Figan without a moment's hesitation went over and collected the banana. Quite obviously he had sized up the whole situation: if he had climbed for the fruit earlier, Goliath would almost certainly have snatched it away. If he had remained close to the banana, he would probably have looked at it from time to time. Chimps are very quick to notice and interpret the eye movements of their fellows, and Goliath would possibly, therefore, have seen the fruit himself. And so Figan had not only refrained from instantly gratifying his desire but had also gone away so that he could not ‘give the game away’ by looking at the banana.

For many years, Goodall's observation was dismissed as a mere anecdote. Now, however, similar behaviour has been observed in pigs, both in natural circumstances and in controlled experiments. A pig who knows where to find food will not go there if she is being followed by a heavier pig who does not know where the food is. It seems that she is aware that the heavier pig would push her aside and take the food. Instead, she learns to behave in ways that minimize the chances that the other pig will be able to take her food – for example, she goes to the food only when the heavier pig is out of sight or much further away from the food than she is.

Another example of behaviour that shows an ability to look forward in time comes from Mathias Osvath's sustained observation of Santino, a chimpanzee in a Swedish zoo. Over a decade, Santino has regularly collected and cached stones. He does this in the morning, before visitors are admitted to the zoo. Several hours later, he goes to his stones, which he has placed on the side of his enclosure where visitors appear, and throws them at the visitors. He has even discovered how to detect, by knocking, places where the concrete in his enclosure is thin enough for him to break it into pieces of a suitable size for throwing. He then breaks it in these weak spots and adds the pieces to his cache of natural stones. In winter, when the zoo is closed to visitors, he does not collect stones. Throwing rocks isn't instinctive in chimpanzees, and nor, of course, is breaking up concrete. Santino does these things calmly, when there are no visitors present, so he cannot be gripped by the same motivation he has when he gets excited by their presence.

A still more rigorous demonstration of an animal's ability to anticipate its own future desires comes, remarkably enough, from experiments, not with apes or other primates, but with scrub jays. Scientists have used two characteristics of these birds to design an ingenious experiment. Like us, scrub jays store food for later consumption. Also like us, after gorging on one kind of food, they become satiated with it and prefer something different. Experimenters gave one group of birds pine nuts and then allowed them to store either pine nuts or kibble. Before they had access to their cache, they again got pine nuts. After becoming familiar with this routine, the jays preferred to store kibble. If they were fed kibble on both occasions, however, they preferred to store pine nuts. This could be explained by the fact that at the time of storing the food, the birds were satiated with what they had been eating, and just preferred to store the other kind of food. With another group of jays, however, the experimenters varied the routine. This time they gave the birds one food and allowed them to store it – but before these birds got access to their cache, they were fed the other kind of food from the one that they had been fed before they were able to store food. These birds preferred to store the food that they had just eaten, even though they were satiated with it. It is difficult to see any explanation for the different behaviour of these birds other than their ability to anticipate that, before they could get at their cached food, they would be satiated, not with the food they had just eaten, but with the other kind of food, and so would prefer the one that they did not want now but would want then. If that is the case, scrub jays not only have exactly what Hampshire said creatures without language could not have, ‘an expectation of an order of events in the future’, but more remarkably still, they also have desires based on their awareness that their future desires will be different from their present ones.

I think we should conclude, on the basis of the evidence just summarized, that some nonhuman animals are persons, as we have defined the term. To judge the significance of this, we must set it in the context of our earlier discussion in which I argued that the only defensible version of the doctrine of the sanctity of human life was what we might call the ‘doctrine of the special significance of taking personal life’. I suggested that if most human beings have lives of special significance, or have a special claim for their lives to be protected, this must be tied up with the fact that most human beings are persons. So if some nonhuman animals are persons too, they also have a special claim for their lives to be protected. Whether we base these special moral features of the lives of human persons on preference utilitarianism, on a right to life deriving from their capacity to see themselves as continuing selves, or on respect for autonomy, these arguments must apply to nonhuman persons as well. Only the indirect utilitarian reason for not killing persons – the fear that such acts are likely to arouse in other persons – applies less readily to nonhuman persons, because nonhumans are less likely than humans to learn about killings that take place at a distance from them. This reason does not apply to all killings of human persons either, however, because sometimes it is possible to kill in such a way that no one learns that a person has been killed.

Hence, we should reject the doctrine that killing a member of our species is always more significant than killing a member of another species. Some members of other species are persons; some members of our own species are not. No objective assessment can support the view that it is always worse to kill members of our species who are not persons than it is to kill members of other species who are. On the contrary, as we have seen, there are strong arguments for thinking that to take the lives of persons is, in itself, more serious than taking the lives of those who are not persons. So it seems that killing a chimpanzee is, other things being equal, worse than the killing of a human being who, because of a profound intellectual disability, is not and never can be a person. (Often, of course, other things are not equal: for instance, the attitudes of parents of humans with profound intellectual disability are relevant.)

The great apes may be the clearest cases of nonhuman persons, but as we have seen, there is evidence of future-directed thinking in several other species. Self-awareness is sometimes linked to knowing that when you look in a mirror, you are seeing yourself rather than another being. This has been tested by putting a coloured dye on a part of the animal where it will be seen in the mirror but cannot be seen otherwise – for example, on an ape, the forehead. (The dye is put on when the animal is asleep so that she does not notice.) Then the animal is given a mirror, with which she has previously become familiar. If she looks in the mirror and then touches the dyed spot, this indicates that she knows that the image in the mirror is herself. All the great apes can pass the mirror test, but so too can elephants, dolphins and even magpies. Magpies belong to the crow family, as do scrub jays, which as we have already seen are capable of taking their future desires into account. Alex, an African gray parrot, who was taught a vocabulary of between fifty and one hundred words by Irene Pepperberg, understood concepts like ‘colour’ and ‘shape’ as well as ‘same’ and ‘different’. He did not take the mirror test, but Pepperberg's meticulously recorded account of his abilities and behaviour leaves little doubt that he too was self-aware to some extent. Human children less than one year old typically fail the mirror test, but by the time they are eighteen months old, most can pass it.

Passing the mirror test may show self-awareness, but failing it does not prove that an animal is not self-aware. In contrast to apes, monkeys do not show signs of self-recognition, although some can learn to use mirrors to locate food that they cannot otherwise see. Dogs have not passed the mirror test, but that may be because they rely more on their sense of smell than on sight. Many people who live with dogs and cats are convinced that their animal companions are self-conscious and have a sense of the future. If dogs and cats qualify as persons, the mammals we use for food cannot be far behind. We think of dogs as being more ‘human’ than pigs, but we have already seen that pigs can plan ahead and grasp whether another pig does or does not know the location of food. Are we turning persons into bacon? Additionally, because at least some birds appear to be persons, we should be cautious about excluding chickens, too. In flocks of up to ninety birds, chickens appear to recognize one another as individuals, always knowing whether another bird is above or below them in the pecking order. They also have the capacity for self-control and to envisage at least the near future. In one experiment, chickens were taught that pecking one key would, after two seconds, bring them access to food for three seconds; whereas pecking a different key would, after six seconds, bring them access to food for twenty-two seconds. The hens preferred to wait for the opportunity to feed longer. At a more anecdotal level, many people who keep free range hens and lock them up at night describe them as eager to get outside in the mornings – an attitude that suggests anticipating the future.

Of the animals that regularly appear on our plates, fish may seem the least likely to be persons, but the category is an extremely broad one: there are approximately 28,000 species of fish, more than all the other vertebrates combined. They vary widely in their abilities. In 2003, the journal Fish and Fisheries published a special issue on learning in fish, the introduction of which described fish as ‘steeped in social intelligence, pursuing Machiavellian strategies of manipulation, punishment and reconciliation…and cooperating to inspect predators and catch food.’ Whether any of this involves conscious planning is unclear, but we do know that the popular myth that fish can remember things for only three seconds is quite wrong – experiments have shown that they can remember the location of a hole in a net even if they have not been near the net for eleven months. As for invertebrates, the veined octopus has been observed to pick up coconut half shells discarded by tourists and carry them a considerable distance – which makes movement quite awkward for this small octopus – in order to assemble them later as a kind of protective shelter. Given what we know of the learning abilities of octopuses, it is not too far-fetched to interpret this behaviour as indicating that the octopus is aware of its own future need for shelter and is planning ahead.

It is difficult to establish when another being has a sense of its own self, or of the past and the future. If it is wrong to kill a person when we can avoid doing so, and there is real doubt about whether a being we are thinking of killing is a person, the best thing to do is to give that being the benefit of the doubt. The rule here is the same as that among deer hunters: if you see something moving in the bushes and are not sure if it is a deer or a hunter, don't shoot! (We may think that the hunters shouldn't shoot in either case, but the rule is a sound one within the ethical framework that hunters use.) On these grounds, much killing of nonhuman animals is open to objection. It may be justifiable of course, for overriding reasons, but it is in need of justification.

On the other hand, even for those nonhuman animals who are self-aware, and hence meet our definition of “person” it is still true that they are not likely to be nearly as much focused on the future as normal human beings are. Gary Varner, in his Personhood and Animals in the Two-Level Utilitarianism of R.M. Hare, argues for a more demanding definition of a person than the one I have used. To be a person, in his view, one must have a biographical sense of self. Humans, he points out, typically tell stories about their lives, weaving narratives that bring together where they have come from, where they are now and what they hope for in the future. Only beings with a sophisticated language, Varner suggests, have this kind of biographical sense of their lives, which means that only humans will have it – and not all humans, either, because not all humans are capable of language. Varner believes that this biographical sense of one's life gives a life a special significance that is lacking in the lives of other beings. Some nonhuman animals, in his view, are ‘near-persons’ in that they have some self-awareness but not a biographical sense of self.

Roger Scruton, a British philosopher, has said that the untimely death of a human being is a tragedy because there are likely to be things that she hoped to accomplish but now will not be able to achieve. The premature death of a cow is not a tragedy in this sense, because whether cows live one year or ten, there is nothing that they hope to achieve. Even those great apes who can use sign language do not talk to us about their plans for the distant future. Scrub jays hide food for the next day, but as far as we know, they do not embark on long-term projects that will pay off in the years ahead. (If it could be shown that squirrels and other animals who hide food for the winter are doing this with conscious foresight of their future needs, that would be an impressive counter-example, but this behaviour may be instinctive.)

Accepting these differences between normal mature humans and nonhuman animals, we could see the wrongness of killing, not as a black and white matter, dependent on whether the being killed is or is not a person, but as a matter of degree, dependent on, among other things, whether the being killed was fully a person or was a near-person or had no self-awareness at all, the extent to which, by our best estimate, the being had future-directed desires, and how central those desires were to the being's life. The criminal law can reasonably take a different view on the grounds that public policy is better served by laws that draw sharp boundaries, but the relevant moral considerations suggest a continuum.

Arguments against killing based on the capacity to see oneself as an individual existing over time apply to some nonhuman animals, but presumably there are others who, though conscious, are not persons. Let's assume that there are some animals about whom we can be confident that they are not persons, or even near-persons. The rightness or wrongness of killing these animals then seems to rest on utilitarian considerations, for they are not autonomous and – at least if Tooley's analysis of rights is correct – do not qualify for a right to life.

Before we discuss the utilitarian approach to killing in itself, we should remind ourselves that a wide variety of indirect reasons will figure in the utilitarian's calculations. Many modes of killing used on animals do not inflict an instantaneous death, so there is pain in the process of dying. There is also the effect of the death of one animal on his or her mate or other members of the animal's social group. There are many species of birds, and a few mammals, in which the bond between male and female lasts for a lifetime. The death of one member of this pair presumably causes distress and a sense of loss and sorrow for the survivor. The mother-child relationship in mammals can be a source of intense suffering if either is killed or taken away. (Dairy farmers routinely remove calves from their mothers at an early age so that the milk will be available for humans; anyone who has lived on a dairy farm will know that, for days after the calves have gone, their mothers keep calling for them.) In some species, the death of one animal may be felt by a larger group – as the behaviour of wolves and elephants suggests. All these factors would lead the utilitarian to oppose some killing of animals, whether or not the animals are persons. These factors would not, however, be reasons for opposing killing in itself, apart from the pain and distress it may cause.

Deciding on the correct utilitarian verdict on killing that is painless and causes no loss to others is complicated, because it depends both on how we choose between the two versions of utilitarianism outlined in the previous chapter, that is, the total or the prior existence view, and also on whether we are hedonistic or preference utilitarians. I will begin by supposing that we are hedonistic utilitarians, because this makes the discussion of the differences between the total and the prior existence views more straightforward, and only subsequently will I consider what impact a switch to preference utilitarianism makes.

On the prior existence view, it is wrong to kill any being whose life is likely to contain, or can be brought to contain, more pleasure than pain. This view implies that it is normally wrong to kill animals for food, because usually we could argue that these animals would have had a few pleasant months or even years before they died – and the pleasure we get from eating them would not outweigh this. In contrast, the total view can lead to a different outcome. In Social Rights and Duties, a collection of essays and lectures published in 1896, Leslie Stephen, a British essayist – and the father of the novelist Virginia Woolf – writes:

Of all the arguments for Vegetarianism none is so weak as the argument from humanity. The pig has a stronger interest than anyone in the demand for bacon. If all the world were Jewish, there would be no pigs at all.

Stephen's point is that although meat eaters are responsible for the death of the animal they eat and for the loss of pleasure experienced by that animal, they are also responsible for the creation of more animals, because if no one ate meat there would be no more animals bred for fattening. The loss meat eaters inflict on one animal is thus compensated for by the benefit they confer on the next. The argument is periodically revived by those who seek to defend meat eating – in the twenty-first century, for example, by Michael Pollan in his best-seller The Omnivore's Dilemma, and also by the British chef and food writer Hugh Fearnley-Whittingstall. We may call it ‘the replaceability argument’, for it assumes that if we kill one animal, we can replace it with another as long as that other will lead a life as pleasant as the one killed would have led, if it had been allowed to go on living. Hedonistic utilitarians who accept the total view must agree with this, for that version of utilitarianism regards sentient beings as valuable only insofar as they make possible the existence of intrinsically valuable experiences like pleasure. It is as if sentient beings are receptacles of something valuable, and it does not matter if a receptacle gets broken so long as there is another receptacle to which the contents can be transferred without any getting spilt. (This metaphor should not be taken too seriously, however; unlike precious liquids, pleasure and other experiences cannot exist independently from a conscious being, and so even on the total view, sentient beings cannot properly be thought of merely as receptacles.)

The first point to note about the replaceability argument is that even if it is valid when the animals in question have pleasant lives, it would not justify eating the flesh of animals reared in modern factory farms, where the animals are so crowded together and restricted in their movements that their lives seem to be more of a burden than a benefit to them. Pollan and Fearnley-Whittingstall are aware of this. They unequivocally condemn factory farming and recommend that we avoid its products.

A second point is that if it is good to create happy life, then presumably it is good for there to be as many happy beings on our planet as it can possibly hold. Defenders of meat eating had better hope that they can find a reason why it is better for there to be happy people rather than just the maximum possible number of happy beings, because otherwise the argument implies that we should eliminate almost all human beings in order to make way for the much larger numbers of smaller happy animals that could sustainably replace them. If, however, the defenders of meat eating do come up with a reason for preferring the creation of happy people to, say, happy mice, then their argument will not support meat eating at all. For with the exception of some areas suitable only for pasture, the surface of our globe can support more people if we grow plant foods than if we raise animals.

A third point is that if replaceability holds for animals, it must hold for humans at a similar mental level. Suppose that whenever a child is born, the parents are offered the option of creating a clone of their child to serve as an organ donor for the child later in life. The clones are gestated in artificial wombs and then reared separately from other human beings in order to prevent the parents becoming so attached to them that they will be reluctant to remove the clone's organs. While in embryonic form, the clones are genetically modified so that their mental abilities never develop beyond those of a human infant. Intellectually incapable of understanding their fate, they will lead lives similar to those of happy, well-cared-for infants until the time comes for them to be killed – humanely, of course. Their hearts and other organs are then used to prolong the lives of the children – now usually adults – from whom they were cloned. Those who receive the organs pay for them, and the revenue from these sales makes it possible to rear new clones from the next generation of babies. Suppose that there is one religious group, let's say Buddhism, that objects to this practice, refuses to use clones, and urges us to accept the idea of living a natural lifespan, which Buddhists see as ethically better than using organs from clones to prolong our lives. To this a modern Leslie Stephen might reply: ‘Of all the arguments for a natural lifespan, none is so weak as the argument from humanity. The clone has a stronger interest than anyone in the demand for organs. If all the world were Buddhist, there would be no clones at all.’ Given our earlier rejection of speciesism, it isn't easy to see how we can use the replaceability argument to defend meat eating without also accepting it as a defence of this form of organ banking.

These three points undoubtedly reduce the appeal of the replaceability argument as a defence of meat eating, but they do not go to the heart of the matter. Are some sentient beings really replaceable? The total view and the replaceability argument have been widely criticised, but none of the critics have offered satisfactory solutions to the underlying problems to which these positions offer a consistent, if uncongenial, answer.

Henry Salt, a nineteenth-century English vegetarian and author of a book called Animals’ Rights, thought that the argument rested on a simple philosophical error:

The fallacy lies in the confusion of thought that attempts to compare existence with non-existence. A person who is already in existence may feel that he would rather have lived than not, but he must first have the terra firma of existence to argue from: the moment he begins to argue as if from the abyss of the non-existent, he talks nonsense, by predicating good or evil, happiness or unhappiness, of that of which we can predicate nothing.

Salt claims that the Roman philosopher Lucretius, who lived in the first century before the Christian era, refuted Stephen's ‘vulgar sophism’ in the following passage of De Rerum Natura:

What loss were ours, if we had known not birth?

Let living men to longer life aspire,

While fond affection binds their hearts to earth:

But who never hath tasted life's desire,

Unborn, impersonal, can feel no dearth.

When I wrote the first edition of Animal Liberation, I accepted Salt's view. I thought that it was absurd to talk as if one conferred a favour on a being by bringing it into existence, because at the time one confers this favour, there is no being at all. But I have since changed my mind on this point. As we saw in the preceding chapter, we do seem to do something bad if we knowingly bring a miserable being into existence, and if this is so, it is difficult to explain why we do not do something good when we knowingly bring a happy being into existence.

Derek Parfit has offered a thought experiment that amounts to an even stronger case for the replaceability view. He asks us to imagine that two women are each planning to have a child. The first woman is already three months pregnant when her doctor gives her both bad and good news. The bad news is that the fetus she is carrying has a defect that will significantly diminish the future child's quality of life – although not so adversely as to make the child's life utterly miserable, or not worth living at all. The good news is that this defect is easily treatable. All the woman has to do is take a pill that will have no side effects, and the future child will not have the defect. In this situation, Parfit very plausibly suggests, we would all agree that the woman should take the pill and that she does wrong if she refuses to take it.

The second woman sees her doctor before she is pregnant, when she is about to stop using contraception. She also receives bad and good news. The bad news is that she has a medical condition, the effect of which is that if she conceives a child within the next three months, the child will have the same defect that the first woman's child will have if she does not take the pill. This defect is not treatable, but the good news is that the woman's condition is a temporary one, and if she waits three months before becoming pregnant, her child will not have the defect. Here too, Parfit suggests, we would all agree that the woman should wait before becoming pregnant and that she does wrong if she does not wait.

Suppose that the first woman does not take the pill, and the second woman does not wait before becoming pregnant, and that as a result each has a child with a significant disability. It would seem that they have each done something wrong. Is their wrongdoing of equal magnitude? If we assume that it would have been no more difficult for the second woman to wait three months before becoming pregnant than it would have been for the first woman to take the pill, it would seem that the answer is yes, what they have done is equally wrong. But now consider what this answer implies. The first woman has harmed her child. That child can say to her mother: ‘You should have taken the pill. If you had done so, I would not now have this disability, and my life would be significantly better.’ If the child of the second woman tries to make the same claim, however, her mother can respond: ‘If I had waited three months before becoming pregnant, you would never have existed. I would have produced another child, from a different egg and different sperm. Your life, even with your disability, is worth living. You never had a chance of existing without the disability. So I have not harmed you at all.’ This reply seems a complete defence to the charge of having harmed the child now in existence. If, despite this, we persist in our belief that it was wrong of the woman not to postpone her pregnancy, in what does the wrongness consist? It cannot lie in bringing into existence the child to whom she gave birth, for that child has an adequate quality of life. Could it lie in not bringing a possible being into existence – to be precise, in not bringing into existence the child she would have had if she had waited three months? If we explain the wrongness of the second woman's decision in this way, we are rejecting the prior existence view in favour of the total view, or something closer to it. We are also a step closer to accepting replaceability, for our explanation implies that we should give weight to the interests of beings who would come into existence, if we chose to bring them into existence.

Because some people are unsure what to say about the case of the two women – in particular, whether what they do is equally wrong – I will add one more example, adapting a case that Parfit calls ‘Depletion’ so that it becomes very like the choice that developed nations are facing now on what to do about climate change. We could continue to use the cheapest energy available to give ourselves, our children and perhaps our grandchildren a high standard of living. In discussions of climate policy, this is often referred to as ‘Business As Usual’. If we do this, however, the warming of the planet will mean that sometime in the next century, things will get much worse for future generations and will remain much worse for several centuries – although we shall assume, for the purposes of this discussion, that they will not get so bad that people in these future centuries will not have a life that is not worth living. Alternatively, we could follow a policy we will call ‘Sustainability’: this involves a quick end to the use of fossil fuels, with significantly changed lifestyles, different industries, less travel, less meat and many other changes. We and our children and perhaps our grandchildren would be slightly worse off under Sustainability than under Business As Usual, but more distant future generations would, for many centuries, be much better off. Overall, if we consider the welfare of every generation, including ours, as far as we can foresee, Sustainability has much better consequences than Business As Usual. But imagine that we are selfish, and don't care much about future generations, beyond our own grandchildren, and so we decide to opt for Business As Usual.

Have we done something wrong? Surely we have; but who have we wronged? It may seem that we have wronged the people who will live in later centuries, because they will have less good lives than they would have had if we had opted for Sustainability. But this response overlooks the fact that our choice of policies will have such widespread effects that it will also change who meets whom, and who has children with whom. For example, people will travel less and so will meet different people. New industries will develop in different parts of the country, and people will move there to find employment. Who we are depends on who our parents are – if my parents had never met, I would not exist. Probably my mother and my father would have had other children, with other partners, and none of those children would have been me. So if we choose Business as Usual, we can pre-empt any complaints from twenty-third-century people by leaving them a document explaining that if we had chosen Sustainability, they would not have been better off, but rather they would not have been at all. Moreover, if their lives are not so bad as not to be worth living, they are better off existing than not existing.

What is wrong with this justification of Business as Usual? On the prior existence view, it is difficult to see what could be wrong with it. The prior existence view tells us to do what is best for those who exist, or will exist anyway, and following Business As Usual does that. The people who are made worse off by our continuation of Business As Usual are people who would not have existed if we had chosen Sustainability. The example shows that to focus only on those who exist or will exist anyway leaves out something vital to the ethics of this decision. We can, and should, compare the lives of those who will exist with the lives of those who might have existed, if we had acted differently. Contrary to Salt, we can and should ‘argue as if from the abyss of the non-existent’. We can condemn the decision to continue with Business As Usual only by taking into account the fact that, if we switch to Sustainability, the lives of those who will exist would be much better than the lives of those who will exist under Business As Usual. Granted, the people for whose sake we should switch to Sustainability will remain, in Lucretius's words, ‘unborn, impersonal’ if we do not make that switch. Never having tasted ‘life's desire’, they will ‘feel no dearth’ of life. Yet the quality of the lives they would have led is inescapably relevant to our decision.

If then we should, in making ethical decisions, at least sometimes take account of the impact we could have on the lives of people the existence of whom is, at the time we are making the decision, uncertain, we need to ask: at what stage in the development from people we might bring into existence to people actually in existence does replaceability cease to apply? What characteristic makes the difference?

Here, there is a difference between preference utilitarianism and hedonistic utilitarianism. Preference utilitarians can draw a distinction between self-aware individuals, leading their own lives and wanting to go on living, and those with no future-directed preferences. They would agree with Lucretius that there is a difference between killing living beings who ‘to longer life aspire’ and failing to create a being who, unborn and impersonal, can feel no loss of life. But what of beings who, though alive, cannot aspire to longer life because they lack the conception of themselves as living beings with a future? These being are also, in a sense, ‘impersonal’. We might say that we do them no personal wrong, which a preference utilitarian might understand as meaning that because they have no future-directed preferences, we are not acting contrary to any of their preferences if we kill them instantly and painlessly. So perhaps the capacity to see oneself as existing over time, and thus to aspire to longer life (as well as to have other non-momentary, future-directed interests), is the characteristic that marks out those beings who cannot be considered replaceable.

This conclusion is in harmony with Tooley's views about what it takes to have a right to life. For a preference utilitarian, concerned with the satisfaction of preferences rather than experiences of suffering or happiness, there is a similar fit with the distinction already drawn between killing those who are rational and self-conscious and killing those who are not. Rational, self-conscious beings are individuals, leading lives of their own, and cannot in any sense be regarded merely as receptacles for containing a certain quantity of happiness. Beings that are conscious, but not self-conscious, on the other hand, more nearly approximate the image of receptacles for experiences of pleasure and pain, because their preferences will be of a more immediate sort. Given the evidence we have just reviewed, it is not easy to say with confidence which animals might be conscious but not self-conscious, but it is reasonable to suppose that there are some in this category. They will not have desires that project their images of their own existence into the future. Their conscious states are not internally linked over time. If they become unconscious, for example by falling asleep, then before the loss of consciousness they would have no expectations or desires for anything that might happen subsequently; and if they regain consciousness, they have no awareness of having previously existed. Therefore, if they were killed while unconscious and replaced by a similar number of other members of their species who will be created only if the first group are killed, there would, from the perspective of their awareness, be no difference between that and the same animals losing and regaining consciousness.

For a merely conscious being, death is the cessation of experiences, in much the same way that birth is the beginning of experiences. Death cannot be contrary to an interest in continued life any more than birth could be in accordance with an interest in commencing life. To this extent, with merely conscious beings, birth and death cancel each other out; whereas with self-aware beings, the fact that one may desire to continue living means that death inflicts a loss for which the birth of another is insufficient compensation.

The test of universalizability supports this view. If I imagine myself in turn as a self-conscious being and a merely conscious being, it is only in the former case that I could have forward-looking desires that extend beyond periods of sleep or temporary unconsciousness, for example a desire to complete my studies, a desire to have children, or simply a desire to go on living, in addition to desires for immediate satisfaction or pleasure, or to get out of painful or distressing situations. Hence, it is only in the former case that my death involves a greater loss than just a temporary loss of consciousness, and my death is not adequately compensated for by the creation of a being with similar prospects of pleasurable experiences.

In reviewing the first edition of this book, the late H. L. A. Hart, a major figure in twentieth-century philosophy of law, suggested that for a utilitarian, self-conscious beings must be replaceable in just the same way as non-self-conscious beings are. The type of utilitarianism one holds will, in Hart's view, make no difference here, because:

Preference Utilitarianism is after all a form of maximizing utilitarianism: it requires that the overall satisfaction of different persons’ preferences be maximized just as Classical Utilitarianism requires overall experienced happiness to be maximized…If preferences, even the desire to live, may be outweighed by the preferences of others, why cannot they be outweighed by new preferences created to take their place?